Browse the corpus

Introduction Atherosclerosis was initially considered a disease exclusively caused by cholesterol deposition in the arterial wall. Significant evidence demonstrated that atherosclerotic plaques are not passive collections of lipids, but sites of active interactions between cells of the immune system and vascular cells that influence the fate of atheromas.1 As evidenced in several studies in animal models and patients, both innate and adaptive immune cells participate in this process and have significant effects on the initiation and progression of atherosclerotic lesions.2,3 Interactions between immune and vascular cells trigger a self-perpetuating inflammatory cycle that generates a chronic inflammatory milieu that promotes atherosclerotic plaque growth and rupture.1 The hypothesis that inflammation is a significant driving factor in atherosclerosis is now the focus of phase III clinical trials that test strategies to reduce inflammatory mediators.4 On-going phase III trials target the innate inflammatory cytokines interleukin-1β (IL-1β; targeted with Canakinumab, a human monoclonal anti-IL-1β antibody in the CANTOS trial), IL-6, and tumour necrosis factor-α (TNF-α; targeted with low-dose methotrexate in the CIRT trial). The immune response in atherosclerosis is multifaceted and, in addition to innate cells, adaptive immune responses have significant contributions. Indeed, adaptive immune cells such as T lymphocytes, B lymphocytes, and antibodies have been identified in atherosclerotic plaques and circulation of patients with atherosclerosis.5 Moreover, substantial evidence from animal studies clearly demonstrates that specialized subsets of T and B lymphocytes exert either protective or promoting effects on atherogenesis.2 These data suggest that more targeted approaches may be required to successfully retune the immune system and modulate the complex inflammatory response in atherosclerosis.

animal studies clearly demonstrates that specialized subsets of T and B lymphocytes exert either protective or promoting effects on atherogenesis.2 These data suggest that more targeted approaches may be required to successfully retune the immune system and modulate the complex inflammatory response in atherosclerosis. In the following sections, we will summarize recent advances on the role of CD4+ T lymphocytes in atherosclerosis, focusing on two subsets [CD4+CD28null and regulatory T (Treg) cells] that are at opposing poles of the inflammatory process that underlies atherosclerosis and harbour therapeutic potential. CD4+ T cells in atherosclerosis T lymphocytes represent the second largest immune cell population in atherosclerotic plaques after macrophages. Both helper (CD4+) and cytotoxic (CD8+) T cells have been identified in atheromas. However, the inflammatory response in atherosclerosis is dominated by Th1 lymphocytes, the most abundant T-cell subset in plaques. Characteristically, Th1 cells are identified by the production of interferon-γ (IFN-γ), a cytokine that promotes atherosclerosis and plaque rupture through effects on several cells in atherosclerotic lesions.1 While the pro-atherogenic role of Th1 is well documented, the precise contribution of other CD4+ T-cell subsets (e.g. Th2 and Th17) remains debatable due to conflicting reports.2

terferon-γ (IFN-γ), a cytokine that promotes atherosclerosis and plaque rupture through effects on several cells in atherosclerotic lesions.1 While the pro-atherogenic role of Th1 is well documented, the precise contribution of other CD4+ T-cell subsets (e.g. Th2 and Th17) remains debatable due to conflicting reports.2 CD4+CD28null T cells Most naive CD4+ T cells constitutively express the co-stimulatory receptor CD28, which delivers vital signals that sustain the proliferation and survival of T cells upon antigen recognition.6 A subset of CD4+ T cells that lack CD28—known as CD4+CD28null (CD28null) T cells—has been identified and implicated in several chronic inflammatory diseases, including atherosclerosis.7,8 These cells share features with Th1 cells, but also diverge phenotypically and functionally from conventional CD4+CD28+ (CD28+) Th1 lymphocytes (Figure 1).7,9 Noteworthy, CD28null T cells are more adept at secreting pro-inflammatory cytokines IFN-γ and TNF-α than conventional CD28+ T cells, both in the resting state and following activation.10 We have recently demonstrated that, in patients with acute coronary syndrome (ACS), CD28null T-cell cytokine production is regulated by the alternative co-stimulatory receptors OX40 and 4-1BB6 that were significantly up-regulated on circulating CD28null compared with CD28+ T cells.10 In addition, CD28null T cells express and release cytotoxic molecules perforin and granzyme B.10,11 These molecules, usually restricted to cytotoxic CD8+ T cells and natural killer cells, have been suggested to enable CD28null T cells to lyse endothelial cells in vitro.12 We found that, in ACS, OX40 and 4-1BB regulate not only the cytokine production, but also the release of perforin from CD28null T cells.10 Another distinguishing feature from conventional CD28+ T lymphocytes is the reduced sensitivity of CD28null T cells to apoptosis induction. In rheumatoid arthritis (RA), this has been attributed to an increase in anti-apoptotic proteins (e.g. Bcl-2),13 while in ACS we found a reduction in pro-apoptotic proteins (e.g. Fas, Bim, and Bax).14 Other characteristics of CD28null T cells have been described elsewhere.9 Figure 1 Overview of cardinal features and potential strategies to target CD28null and regulatory T-cell lymphocytes in atherosclerosis. CD28null T cells expand in the circulation and atherosclerotic plaques in patients with acute coronary syndrome.

er characteristics of CD28null T cells have been described elsewhere.9 Figure 1 Overview of cardinal features and potential strategies to target CD28null and regulatory T-cell lymphocytes in atherosclerosis. CD28null T cells expand in the circulation and atherosclerotic plaques in patients with acute coronary syndrome. These cells produce high levels of inflammatory cytokines (interferon-γ and tumour necrosis factor-α) and express and release cytotoxic molecules (perforin and granzyme B). The production of cytokines and cytotoxic molecules is modulated by co-stimulatory receptors OX40 and 4-1BB, which are up-regulated on CD28null T cells. Moreover, CD28null T cells are resistant to apoptosis due to reduction in apoptotic molecules Fas, Bim, and Bax, which underlies the expansion of this cell subset in acute coronary syndrome patients. Several strategies have been proposed to reduce the number and/or function of CD28null T cells such as tumour necrosis factor-α inhibition, statins, sensitization to apoptosis, and inhibition of co-stimulation (CTLA4-Ig, anti-OX40, and anti-4-1BB antibodies). Regulatory T cells have crucial roles in counteracting the actions of inflammatory T lymphocytes such as CD28null T cells. Redundant mechanisms enable regulatory T cells to suppress inflammation and T-cell actions among which are the production of anti-inflammatory cytokines (interleukin-10 and transforming growth factor-β), CTLA-4-mediated suppression, and interference with metabolic pathways (conversion of ATP into adenosine, which has anti-inflammatory properties). On-going experimental and clinical work is testing various strategies to boost the number and/or the function of circulating or intraplaque regulatory T cells such as adoptive transfer of regulatory T cells expanded in vitro, vaccination protocols that induce expansion of naturally occurring regulatory T cells or conversion of conventional CD4+ T cells into regulatory T cells, and blockade of inflammatory cytokines (tumour necrosis factor-α and interleukin-6).

latory T cells such as adoptive transfer of regulatory T cells expanded in vitro, vaccination protocols that induce expansion of naturally occurring regulatory T cells or conversion of conventional CD4+ T cells into regulatory T cells, and blockade of inflammatory cytokines (tumour necrosis factor-α and interleukin-6). Whereas circulating CD28null T cells are nearly undetectable in healthy individuals, this subset expands in patients with ACS,10,15 chronic inflammatory disorders such as autoimmunity (extensively characterized in RA), chronic infections, and chronic or end-stage kidney disease.9 Acute coronary syndrome patients show significantly higher frequencies of circulating CD28null T cells than healthy controls or stable angina (SA) patients.10,15 A 4-year follow-up study found that ACS patients with recurrent myocardial infarction had four-fold higher circulating CD28null T-cell frequencies compared with patients with only one acute event and that the expansion of CD28null T lymphocytes associated with ACS severity and poor outcome, indicating a pathogenic role for these cells.16 This is also supported by identification of CD28null T cells not only in the peripheral circulation but also in atherosclerotic plaques, with preferential accumulation in unstable lesions.17 As CD28null T cells secrete IFN-γ10,18 and lyse endothelial cells,12 these effector functions may promote local inflammation and trigger plaque destabilization.

ification of CD28null T cells not only in the peripheral circulation but also in atherosclerotic plaques, with preferential accumulation in unstable lesions.17 As CD28null T cells secrete IFN-γ10,18 and lyse endothelial cells,12 these effector functions may promote local inflammation and trigger plaque destabilization. Importantly, RA and other inflammatory diseases associate with increased ACS incidence and poorer outcome.19 It is tempting to speculate that CD28null T cells are likely candidates for driving the accelerated atherosclerosis process observed in these disorders. Indeed, RA patients with high frequencies of CD28null T lymphocytes have increased carotid artery intima-media thickness (IMT) and decreased flow-mediated vasodilatation than those in whom this subset does not expand.20 Similarly, the presence of CD28null T cells in end-stage kidney disease patients has been associated with increased C-reactive protein and IMT.21 Moreover, CD28null T-cell expansion in patients with type II diabetes correlates with ACS occurrence and poor outcome.18

ion than those in whom this subset does not expand.20 Similarly, the presence of CD28null T cells in end-stage kidney disease patients has been associated with increased C-reactive protein and IMT.21 Moreover, CD28null T-cell expansion in patients with type II diabetes correlates with ACS occurrence and poor outcome.18 The mechanisms involved in CD28null T-cell expansion in atherosclerosis and inflammatory disorders remain poorly understood. It has been suggested that CD28 downregulation is antigen-driven.17 Several exogenous and endogenous antigens have been proposed including cytomegalovirus and heat shock proteins (HSPs).22,23 Strikingly, CD28null T cells that react to oxidized LDL (ox-LDL), one of the antigens frequently implicated in atherosclerosis, have not been described.23 An alternative hypothesis implicates inflammatory cytokines in CD28null T-cell expansion. CD28+ T-cell clones from RA patients downregulated CD28 transcription following TNF-α treatment in vitro.24 Conversely, infliximab triggered CD28 re-expression in cells from patients with RA or unstable angina (UA) in vitro;25,26 however, a different study found that etanercept and infliximab did not affect the percentage of circulating CD28null T cells in RA patients.27 Another mechanism that may underlie CD28null T-cell expansion is resistance to apoptosis, due to defects in molecules regulating apoptosis entry.13,14

ngina (UA) in vitro;25,26 however, a different study found that etanercept and infliximab did not affect the percentage of circulating CD28null T cells in RA patients.27 Another mechanism that may underlie CD28null T-cell expansion is resistance to apoptosis, due to defects in molecules regulating apoptosis entry.13,14 It has been suggested that CD28null T cells may link innate and adaptive immune responses. In line with this hypothesis, CD28null T lymphocytes from patients with RA, psoriatic arthritis, or ankylosing spondylitis (AS) were found to express Toll-like receptors TLR4 and TLR2 (the latter identified only in CD28null T cells in AS).28 The frequency of CD28null T-cell-expressing TLRs varied, with a median of roughly 25% for CD28nullTLR4+ T cells and <5% for CD28nullTLR2+ T cells. In vitro treatment with TNF-α up-regulated TLR4 and TLR2 expression on CD28null T cells from AS patients. Contrastingly, TNF-α neutralization in AS patients decreased expression of these TLRs on circulating CD28null T cells analysed ex vivo.28 These receptors were functional as demonstrated by increased perforin expression in CD28null T cells following stimulation with the TLR4 and TLR2 agonist lipopolysaccharide.28 Of note, both exogenous (microbial) and endogenous (e.g. HSP60) TLR agonists have been described in atherosclerotic plaques, and TLR2/4-mediated signals have been implicated in atherosclerosis.29 Whether TLR agonists contribute to CD28null T-cell activation in situ in atherosclerotic lesions warrants further investigation. Overall, CD28null T cells produce high levels of inflammatory cytokines, release cytotoxic molecules, and infiltrate atherosclerotic lesions, wherein these features may allow them to contribute to the on-going inflammatory response and plaque destabilization.

u in atherosclerotic lesions warrants further investigation. Overall, CD28null T cells produce high levels of inflammatory cytokines, release cytotoxic molecules, and infiltrate atherosclerotic lesions, wherein these features may allow them to contribute to the on-going inflammatory response and plaque destabilization. Regulatory CD4+ T cells The actions of pro-inflammatory T lymphocytes are normally restrained by Treg cells. This specialized subset has critical roles in immune homeostasis and preventing excessive immune responses.30,31 The most numerous and best-characterized are thymus-derived (naturally occurring) Treg (identified as CD4+CD25highCD127lowFOXP3+ T cells), as opposed to peripherally derived (induced) Treg, which originate from naive conventional T cells.31 Regulatory T cells comprise around 5% of CD4+ T cells in the peripheral blood in humans, and are characterized by the expression of the Forkhead box P3 transcription factor (FOXP3), high CD25 levels, and low/no CD127 expression.30,31 Forkhead box P3 transcription factor is essential for Treg development and suppressive function.32 Regulatory T cells employ several mechanisms to suppress effector cells, among which are inhibitory cell–cell interactions, release of anti-inflammatory cytokines (IL-10 and transforming growth factor-β, TGF-β), and disruption of metabolic pathways (Figure 1).30,31 Regulatory T-cell impairment through numerical and/or functional defects has been implicated in autoimmune diseases including type I diabetes, systemic lupus erythematosus, RA, multiple sclerosis, and inflammatory bowel disease.31

ansforming growth factor-β, TGF-β), and disruption of metabolic pathways (Figure 1).30,31 Regulatory T-cell impairment through numerical and/or functional defects has been implicated in autoimmune diseases including type I diabetes, systemic lupus erythematosus, RA, multiple sclerosis, and inflammatory bowel disease.31 Evidence from animal models and patients with atherosclerosis suggests an overall protective role for Treg. Depletion of Treg in Apoe−/− or Ldlr−/− murine models aggravated atherosclerosis, with Treg suggested to limit plaque inflammation and disease progression, although the mechanisms responsible remain poorly defined.33–35 Data on Treg in human atherosclerosis are scant and fraught with contradictory findings. Initial studies suggested that the percentage of circulating CD4+CD25+ Treg is reduced in ACS compared with SA patients and healthy individuals.36,37 Recent studies failed to identify a consistent correlation between the percentage of circulating CD4+CD25highCD127low Treg and the severity of coronary artery disease,38 whereas other authors suggested that CD4+FOXP3+ Treg reduction associates with an increased risk for myocardial infarction.39 A likely explanation for these contradictory findings is the inconsistent use of accurate markers to identify Treg, particularly FOXP3, which remains the most specific marker for delineating Treg from other T cells in humans.32 A recent study quantified Treg by assessing the demethylation of a conserved non-coding sequence in the FOXP3 locus (the Treg cell-specific demethylated region), a feature essential for Treg suppressive function.30,31 Regulatory T cells identified by this method were reduced in ACS patients compared with controls, and their reduction correlated with ACS severity.40 Even less information is available on the suppressive function of Treg in patients with atherosclerosis. A report published in 2006 suggested a reduced suppressive function of circulating CD4+CD25high Treg in ACS patients,36 but the study was insufficiently powered and did not employ a robust suppression assay.

rity.40 Even less information is available on the suppressive function of Treg in patients with atherosclerosis. A report published in 2006 suggested a reduced suppressive function of circulating CD4+CD25high Treg in ACS patients,36 but the study was insufficiently powered and did not employ a robust suppression assay. Compared with other inflammed tissues, relatively low levels of FOXP3+ Tregs were observed in human atherosclerotic plaques (0.5–5% of CD3+ T cells), which may explain persistent inflammation in these lesions.41 Moreover, fewer FOXP3+ Treg were present in vulnerable rather than stable plaques.42 Impaired Treg survival has been suggested to have a role in this process, and in vitro studies indicate that ox-LDL may trigger Treg apoptosis.43 Recent data in ACS patients suggest that circulating CD4+ T cells may have impaired ability to differentiate into Treg due to increased expression of protein tyrosine phosphatase PTPN22.44 A different study suggested that CD4+CD25highCD127low Treg are enriched in coronary thrombi adjacent to culprit lesions compared with peripheral blood in ACS patients and that Treg from thrombi express a restricted repertoire of antigen receptors compared with circulating Treg.45 This suggests that circulating Treg may migrate into atherosclerotic lesions to control the inflammatory response, although further work is warranted to clarify the contribution of circulating and plaque-resident Treg in human atherosclerosis.

restricted repertoire of antigen receptors compared with circulating Treg.45 This suggests that circulating Treg may migrate into atherosclerotic lesions to control the inflammatory response, although further work is warranted to clarify the contribution of circulating and plaque-resident Treg in human atherosclerosis. Potential strategies to target CD28null T cells Several attempts have been made to identify strategies to target CD28null T cell (Figure 1). Initial studies on small numbers of patients suggested that TNF-α blockade decreases circulating CD28null T-cell number in RA and UA.25,26 However, recent studies failed to show consistent depletion of this cell subset in RA patients treated with infliximab or etanercept for 1 year.27 Whether TNF-α inhibitors have beneficial effects in patients with coronary atherosclerosis remains to be established. Statins have also been suggested to reduce CD28null T cells in UA, although the effect was modest (from 3 to 2.3% CD28null T cells, P= 0.022).46 Moreover, in a small study on patients with myocardial infarction, Rosuvastatin treatment was linked to apoptosis of CD28null T cells analysed ex vivo.47 We recently demonstrated that Atorvastatin or Rosuvastatin failed to induce apoptosis in CD28null T cells isolated from ACS patients.14 Our in vitro findings are in line with previous reports that did not identify changes in CD28null T-cell frequency after the acute coronary event in a 2-year follow-up study of ACS patients,13 indicating that statins do not have major effects on CD28null T cells.

ptosis in CD28null T cells isolated from ACS patients.14 Our in vitro findings are in line with previous reports that did not identify changes in CD28null T-cell frequency after the acute coronary event in a 2-year follow-up study of ACS patients,13 indicating that statins do not have major effects on CD28null T cells. Protocols that modulate the inflammatory immune response by blocking T-cell co-stimulation are being developed in autoimmunity and other inflammatory disorders. Treatment with a CTLA-4Ig fusion protein (Abatacept) that blocks CD28 ligation on T cells is used in RA. This drug was found to reduce CD8+CD28null T cells, but did not influence significantly CD4+CD28null T cells in RA patients.48 Interestingly, in ACS, we found similar CTLA-4 levels on CD4+CD28null and conventional CD4+CD28+ T lymphocytes, while alternative co-stimulatory receptors OX40 and 4-1BB were markedly up-regulated on CD4+CD28null T cells.10 This may explain why Abatacept had minor effects on CD4+CD28null T cells in RA, and suggest OX40 and 4-1BB blockade as a more rational approach. Importantly, OX40 and 4-1BB are selectively expressed on activated/effector T cells, and are absent from naive/resting lymphocytes. Thus, blockade of OX40 and/or 4-1BB may allow specific modulation of effector T cells that mediate tissue damage, while preserving the ability of naive T lymphocytes to respond to exogenous antigens. Tools to block OX40 and 4-1BB are being developed for RA, multiple sclerosis, inflammatory bowel disease, asthma, transplantation, and graft vs. host disease,49 which should facilitate their use in atherosclerosis.

iate tissue damage, while preserving the ability of naive T lymphocytes to respond to exogenous antigens. Tools to block OX40 and 4-1BB are being developed for RA, multiple sclerosis, inflammatory bowel disease, asthma, transplantation, and graft vs. host disease,49 which should facilitate their use in atherosclerosis. Recently, we have proposed another strategy for targeted modulation of CD28null T cells that exploits molecules that regulate apoptosis. We demonstrated that the pro-apoptotic mitochondrial protein Bim, which has central roles in controlling apoptosis induction, was reduced in CD28null T cells from ACS patients and this associated with apoptosis resistance of these cells.14 Moreover, we identified the proteasome, a protein degradation system, as a key molecular switch that controls apoptosis of CD28null T cells by degrading Bim and that, when targeted by proteasome-inhibiting drugs, can restore apoptosis sensitivity of CD28null T cells. Encouragingly, proteasome inhibitors preferentially sensitized CD28null T cells to apoptosis, indicating that the proteasome may be an attractive target to enable selective elimination of CD28null T cells, while sparing conventional CD28+ T lymphocytes and avoiding bystander immunosuppression.

vity of CD28null T cells. Encouragingly, proteasome inhibitors preferentially sensitized CD28null T cells to apoptosis, indicating that the proteasome may be an attractive target to enable selective elimination of CD28null T cells, while sparing conventional CD28+ T lymphocytes and avoiding bystander immunosuppression. Recent studies implicate endogenous microRNAs (miRs) in the regulation of T-cell development, differentiation, and function, and on-going research is trying to harness miRs to target inflammatory responses in atherosclerosis.50 Moreover, miRs have important roles in the pathophysiology of cardiovascular diseases. Cardiovascular patients have altered patterns of circulating miRs, and potential diagnostic and therapeutic applications are under investigation.51 Of particular relevance to T cells is miR-29 that specifically inhibits IFN-γ production from CD4+ T cells by targeting the transcription factors T-bet and Eomes.52 MicroRNA-155 has also been implicated in the generation of Th1 cells, as T lymphocytes from miR-155−/− mice showed skewing towards Th2 with predominant production of IL-4 and IL-10 and deficient IFN-γ secretion.53 Interestingly, miR-155 was also implicated in Treg survival.54 MicroRNA-146a is upregulated in ACS and promotes Th1 differentiation through T-bet induction.55 The precise contribution of these miRs to the generation of CD28null T lymphocytes and IFN-γ production is currently unknown. A better characterization of the precise mechanisms that drive the inflammatory/cytotoxic functions and expansion of CD28null T cells may unveil additional strategies to modulate this CD4+ T-cell subset in ACS patients.

of these miRs to the generation of CD28null T lymphocytes and IFN-γ production is currently unknown. A better characterization of the precise mechanisms that drive the inflammatory/cytotoxic functions and expansion of CD28null T cells may unveil additional strategies to modulate this CD4+ T-cell subset in ACS patients. Regulatory T cell-based therapies for atherosclerosis Given their pivotal roles in immune homeostasis and prevention of excessive/harmful immune responses, substantial research efforts are focused on developing Treg-based therapies to reset dysfunctional immune responses in inflammatory diseases. Clinical trials employing Treg are on-going in solid-organ transplantation, type I diabetes, and graft vs. host disease.56 These trials primarily use purification of naturally occurring FOXP3+ Treg from patients, followed by in vitro expansion and reinfusion. Protocols involving in vivo manipulation of Treg subsets (expansion of naturally occurring Treg or conversion of antigen-specific conventional T cells into Treg) are also explored (Figure 1). Of note, adoptive transfer of Treg in animal models markedly reduced atherosclerosis, suggesting that a similar strategy may be beneficial in patients.33 Moreover, several murine studies successfully demonstrate that in vivo induction of polyclonal or antigen-specific Treg (e.g. ox-LDL, ApoB100, and HSP60) reduced atherosclerosis development and/or progression.57 One of the most promising antigens is apolipoprotein B100 (ApoB100), with several studies showing that ApoB100 peptide-based vaccines inhibited atherosclerosis in mice through Treg induction.58,59 Vaccination protocols using ApoB100 are being developed for first-in-human clinical trials.

velopment and/or progression.57 One of the most promising antigens is apolipoprotein B100 (ApoB100), with several studies showing that ApoB100 peptide-based vaccines inhibited atherosclerosis in mice through Treg induction.58,59 Vaccination protocols using ApoB100 are being developed for first-in-human clinical trials. Immunosuppressive drug treatment has been suggested to affect Treg in atherosclerotic plaques. A randomized control trial on a small group of patients with atherosclerotic carotid artery stenosis found that mycophenolate mofetil, an immunosuppressive drug used to prevent allograft rejection due to its ability to inhibit activated T-cell proliferation, caused not only a reduction in activated (CD3+CD69+) T cells, but also an increase in CD3+Foxp3+ in carotid atherosclerotic lesions.60

tic carotid artery stenosis found that mycophenolate mofetil, an immunosuppressive drug used to prevent allograft rejection due to its ability to inhibit activated T-cell proliferation, caused not only a reduction in activated (CD3+CD69+) T cells, but also an increase in CD3+Foxp3+ in carotid atherosclerotic lesions.60 The safety and efficacy of Treg-based therapies in human atherosclerosis has not yet been investigated and important issues remain to be addressed for optimal design of protocols utilizing adoptive transfer of Treg or vaccines that induce Treg in vivo. An important challenge for Treg-based immunotherapy remains the stability of Treg during in vitro manipulation and following reinfusion in patients with on-going inflammation. Several studies suggest that Treg may lose FOXP3 expression during in vitro culture or following reinfusion into hosts, which associates with decreased suppression, and possibly Treg conversion into pathogenic T cells.32 Moreover, to efficiently break the chronic inflammation cycle, Treg-based therapies may not be sufficient and may need to be combined with strategies that deplete/modulate pro-inflammatory T cells.

einfusion into hosts, which associates with decreased suppression, and possibly Treg conversion into pathogenic T cells.32 Moreover, to efficiently break the chronic inflammation cycle, Treg-based therapies may not be sufficient and may need to be combined with strategies that deplete/modulate pro-inflammatory T cells. Future directions On-going clinical trials testing anti-inflammatory therapies in patients with coronary atherosclerosis target inflammatory cytokines (IL-1β, IL-6, and TNF-α).4 Although these trials will bring valuable information on the role of inflammation in atherosclerosis, it is clear that both pro-atherogenic and atheroprotective immune networks operate in this disease. Moreover, considering the complexity and heterogeneity of atherosclerosis in humans, a more targeted immune-modulatory approach, adapted to the predominant pathogenic mechanisms, may be required. Current long-term immunomodulation protocols harbour considerable side effects (infections, immunosuppression, and malignancy). Therefore, careful assessment of benefit/risk profile and development of safer, more targeted immunomodulation therapies are required in atherosclerosis, wherein the inflammatory process is often more discreet than in other chronic inflammatory disorders. Although translation of recent advances on the role of T lymphocytes and other immune cells in atherosclerosis is complex and fraught with challenges, targeted immunomodulation harbours high potential to complement and synergise with drugs/interventions currently used in atherosclerosis patients. Encouragingly, antibodies and other biological or pharmacological reagents that modulate T and B lymphocytes or molecules that regulate their function have been incorporated with promising results in the clinical armamentarium in cancer, autoimmunity, and transplantation, which lends hope to their future applications in atherosclerosis.

ibodies and other biological or pharmacological reagents that modulate T and B lymphocytes or molecules that regulate their function have been incorporated with promising results in the clinical armamentarium in cancer, autoimmunity, and transplantation, which lends hope to their future applications in atherosclerosis. Funding Work in the Cardiovascular Immunology Laboratory at St George's, University of London is funded by the British Heart Foundation (grant nos PG/10/50/28434, PG/13/24/30115, and PG/14/18/30724) and St George's Hospital Charity, London, UK. Funding to pay the Open Access publication charges for this article was provided by the Charity Open Access Fund provided to St George's, University of London. Conflict of interest: none declared. Acknowledgements Several seminal contributions could not be cited in this work due to space constraints.

Introduction Acute coronary syndromes (ACSs) represent a life-threatening range of clinical conditions that are almost always associated with the rupture of an atherosclerotic plaque and partial or complete thrombosis of the infarct-related artery. Platelet aggregation, induced by plaque rupture, is an important contributor to the generation of atherothrombotic events.1 Dual antiplatelet therapy (DAPT) consisting of aspirin and a P2Y12 receptor inhibitor is the recommended treatment following an ACS. Guidelines include maintaining a long-term DAPT course for one year after an ACS event.2–4 Early discontinuation of DAPT has been associated with adverse outcomes. A retrospective observational cohort study from the UK assessing clopidogrel therapy persistence in a population of 4650 patients discharged after acute myocardial infarction (MI) found that premature discontinuation of clopidogrel during the first 12 months of treatment was associated with a significant increase in the risk of death or recurrent infarction.5

y from the UK assessing clopidogrel therapy persistence in a population of 4650 patients discharged after acute myocardial infarction (MI) found that premature discontinuation of clopidogrel during the first 12 months of treatment was associated with a significant increase in the risk of death or recurrent infarction.5 In Belgium, a full year of DAPT (either clopidogrel, prasugrel, or ticagrelor) is universally reimbursed following an ACS. To date, however, there is little information on oral antiplatelets (OAPs) treatment persistence after an ACS among Belgian patients. The current study (REal World insights on the INitiation and treatment Duration of ticagrEloR and other OAPs in patients with ACS in Belgium [REWINDER]) aimed at retrospectively collecting data on the treatment persistence with OAPs after an ACS, and the reasons and deciders of OAP treatment switch, discontinuation, or re-initiation. Methods Study design The study was a real world, multicentre, noninterventional, retrospective study carried out in 18 medical centres in Belgium between 4 September 2014 and 30 January 2015, among patients treated with OAPs after a hospitalization for an ACS. All data were collected from medical records, for a period of 1 year after discharge from the hospital following the event. If the follow-up period was not fully covered by the available files, the cardiologists were required to contact the general practitioner (GP) or to consult hospital records in order to complete the case report files.

ed from medical records, for a period of 1 year after discharge from the hospital following the event. If the follow-up period was not fully covered by the available files, the cardiologists were required to contact the general practitioner (GP) or to consult hospital records in order to complete the case report files. The OAPs targeted by the study were clopidogrel (Plavix™), prasugrel (Efient™), or ticagrelor (Brilique™). Currently, clopidogrel, and ticagrelor are reimbursed by the social security in their registered indication in Belgium. Clopidogrel is indicated in adults suffering with MI, ischaemic stroke, established peripheral arterial disease, ACS, ST-segment elevation myocardial infarction (STEMI), or non-STEMI (in combination with acetylsalicylic acid). Ticagrelor co-administered with acetylsalicylic acid is indicated for the prevention of atherothrombotic events in adults with ACS or a history of MI and a high risk of developing an atherothrombotic event. Prasugrel is only reimbursed in STEMI patients undergoing percutaneous coronary intervention (PCI), diabetic NSTEMI patients, and patients with a stent thrombosis. All drugs are explicitly reimbursed for a period of 1 year after an ACS event. Prior to data collection, all patients were informed of the study through an information letter. The study was designed and conducted according to the Declaration of Helsinki, Good Clinical Practice and local regulations and is registered at www.clinicaltrials.gov (NCT02190123). The study protocol was approved by local ethics committees.

Prior to data collection, all patients were informed of the study through an information letter. The study was designed and conducted according to the Declaration of Helsinki, Good Clinical Practice and local regulations and is registered at www.clinicaltrials.gov (NCT02190123). The study protocol was approved by local ethics committees. Study population The patients enrolled in the study were men and women over 18 years, discharged alive from the hospital following an ACS event diagnosed with STEMI, NSTEMI, or unstable angina (UA) that had occurred between 1 July 2012 and 1 June 2013. The ACS event was required to have been either UA or MI of type 1 (spontaneous MI related to ischaemia due to a primary coronary event such as plaque erosion and/or rupture, fissuring, or dissection) in order to be included in the study. Eligible patients were receiving one of the 3 OAPs at discharge: clopidogrel, ticagrelor, or prasugrel. Only patients with documented use of OAPs and/or exact discontinuation dates were analysed, regardless of the source of the information (medical records from cardiologist or GP). The patients were considered ineligible if they had participated in any interventional clinical study during the REWINDER observation period, if the ACS event occurred during a stay in the hospital or if the ACS event was precipitated by or arose as a complication of surgery, trauma, gastrointestinal bleeding or PCI.

Study population The patients enrolled in the study were men and women over 18 years, discharged alive from the hospital following an ACS event diagnosed with STEMI, NSTEMI, or unstable angina (UA) that had occurred between 1 July 2012 and 1 June 2013. The ACS event was required to have been either UA or MI of type 1 (spontaneous MI related to ischaemia due to a primary coronary event such as plaque erosion and/or rupture, fissuring, or dissection) in order to be included in the study. Eligible patients were receiving one of the 3 OAPs at discharge: clopidogrel, ticagrelor, or prasugrel. Only patients with documented use of OAPs and/or exact discontinuation dates were analysed, regardless of the source of the information (medical records from cardiologist or GP). The patients were considered ineligible if they had participated in any interventional clinical study during the REWINDER observation period, if the ACS event occurred during a stay in the hospital or if the ACS event was precipitated by or arose as a complication of surgery, trauma, gastrointestinal bleeding or PCI. Study objectives The primary objective of the study was to evaluate the actual treatment persistence with OAPs up to 1 year after an ACS in the clinical practice of Belgium. Secondary objectives were: (i) to describe the most frequent reasons for OAP treatment switch, discontinuation, or re-initiation; (ii) to identify the persons (patient, interventional, or non-interventional cardiologist, GP, other) who asked for or decided on the OAP treatment switch, discontinuation, or re-initiation; and (iii) to assess distinct patient profiles associated with premature treatment discontinuation. Treatment switch was defined as a change from the index OAP to another OAP targeted by the study.

ventional cardiologist, GP, other) who asked for or decided on the OAP treatment switch, discontinuation, or re-initiation; and (iii) to assess distinct patient profiles associated with premature treatment discontinuation. Treatment switch was defined as a change from the index OAP to another OAP targeted by the study. Owing to the non-interventional character of this study, no pro-active safety data collection took place. Statistical analysis Demographics and baseline characteristics analyses were performed on the study population containing all evaluable patients, i.e. patients for whom follow-up data on OAP treatment were available at least 11 months after discharge either through hospital or GP records. Only patients for whom the status was clear at 360 days and for whom the treatment stop date was known were included in the treatment persistence analysis. Analyses consisting of simple frequencies and descriptive statistics of all variables were carried out. Cross-tabulations of variables were performed when relevant. Patients’ demographics and baseline characteristics were tabulated overall, by gender and stratified by 2 age categories: <65 years and ≥65 years. The relationship between overall OAP treatment persistence rate and multiple factors was analysed using multivariate logistic regression models, with age, gender, hospital type, ACS diagnosis and management, cardiovascular (CV) history, CV risk factor, OAP treatment, and concomitant medication at discharge included as independent variables. Adjusted P-values were computed for each analysis.

ple factors was analysed using multivariate logistic regression models, with age, gender, hospital type, ACS diagnosis and management, cardiovascular (CV) history, CV risk factor, OAP treatment, and concomitant medication at discharge included as independent variables. Adjusted P-values were computed for each analysis. Statistical significance was defined as P < 0.05 and all tests were 2-sided. Univariate analyses were carried out using the Statistical Package for the Social Sciences (SPSS version 13.0, IBM). Multivariate analyses were performed using the Statistical Analysis Software (SAS). The non-randomized, retrospective design of this study cannot support any formal cross-therapy comparisons with respect to the endpoint variables. Hence, all analyses are considered exploratory and the P-values will be interpreted descriptively. Results Demographics Eighteen PCI and non-PCI hospitals from Belgium were included in the study: 39% of the participating sites were academic PCI centres, 33% were non-academic PCI centres, and 28% were non-academic non-PCI centres. A total of 671 case report files were screened and 314 were finally included in this analysis (total study population). The reasons for exclusion are presented in Figure 1. An additional 19 patients were excluded from the treatment persistence analysis: 3 patients with an exact date for the initial OAP treatment stop, but with unclear status at 12 months, after OAP re-initiation and 16 patients with a follow-up of more than 11 months, but without an exact date of treatment stop (Figure 1).

n additional 19 patients were excluded from the treatment persistence analysis: 3 patients with an exact date for the initial OAP treatment stop, but with unclear status at 12 months, after OAP re-initiation and 16 patients with a follow-up of more than 11 months, but without an exact date of treatment stop (Figure 1). Figure 1 Patients’ flow chart. CRF, case report form; n, number of patients. Asterisk indicates no information available at 12 months, therefore the CRFs were excluded from the analysis. Demographic characteristics for the total study population (n = 314) are given in Table 1. Approximately one-half of the patients belonged to the <65 years age group. The majority of patients were men, with an average age of 63.4 ± 11.7 years (mean age ± standard deviation) at the time the ACS occurred, while female patients were older (67.5 ± 13.3). The average age among all patients at the onset of the ACS event was 64.4 ± 12.2 years. The cohort for treatment persistence was similar in terms of demographics, cardiovascular history, final diagnosis, and established course of treatment to an analysed study population including 560 case report forms with OAP treatment status recorded according to a less stringent certainty criterion (i.e. lacking exact information at 12 months; Figure 1). Table 1 Baseline characteristics for the study population (n =  314)

iagnosis, and established course of treatment to an analysed study population including 560 case report forms with OAP treatment status recorded according to a less stringent certainty criterion (i.e. lacking exact information at 12 months; Figure 1). Table 1 Baseline characteristics for the study population (n =  314) Count % Age category <65 years 164 52 ≥65 years 150 48 Gender Male 240 76 Female 74 24 Type of hospital Academic PCI 127 40 Non-academic PCI 91 29 Non-academic non-PCI 96 31 ACS type Unstable angina 31 10 STEMI 160 51 NSTEMI 123 39 ACS management PCI 286 91 CABG 6 2 PCI+CABG 4 1 Angiography without coronary intervention 7 2 Non-invasively managed (medically managed) 11 4 OAP Clopidogrel 99 32 Prasugrel 60 19 Ticagrelor 155 49 n, number of patients in the study population; PCI, percutaneous coronary intervention; ACS, acute coronary syndrome; STEMI, ST-segment elevation myocardial infarction; NSTEMI, non-ST-segment elevation myocardial infarction; CABG, coronary artery by-pass grafting; OAP, oral antiplatelet. Diagnosis, treatment, and CV history and risk factors Half of the total patients (160/314; 51%) were diagnosed with STEMI, while only a minority of patients (31/314; 10%) were diagnosed with UA (Table 1). The mean age of patients at the time of the ACS event was 63.3 ± 12.4 years for patients diagnosed with STEMI and 66.3 ± 12.1 years for patients diagnosed with NSTEMI.

s Half of the total patients (160/314; 51%) were diagnosed with STEMI, while only a minority of patients (31/314; 10%) were diagnosed with UA (Table 1). The mean age of patients at the time of the ACS event was 63.3 ± 12.4 years for patients diagnosed with STEMI and 66.3 ± 12.1 years for patients diagnosed with NSTEMI. The prescribed OAP at hospital discharge was clopidogrel for 32% of the patients, prasugrel for 19% and ticagrelor for 49% of the study population. Patients on concomitant oral anticoagulation treatment at discharge (24/314; 8%) received more frequently clopidogrel (11/24; 46%) and ticagrelor (10/24; 42%) than prasugrel (3/24; 8%). The CV history and risk factors among the patients are shown in Table 2. More than half of the patients (59%) reported no past CV diseases, while prior coronary artery diseases were documented in 193 instances. Only 8% of the study population had a prior history of peripheral vascular disease. The most prevalent CV risk factor was dyslipidaemia or treatment with cholesterol lowering drugs (225/314; 72%). Table 2 CV history and risk factors for the study population (n = 314)

ry artery diseases were documented in 193 instances. Only 8% of the study population had a prior history of peripheral vascular disease. The most prevalent CV risk factor was dyslipidaemia or treatment with cholesterol lowering drugs (225/314; 72%). Table 2 CV history and risk factors for the study population (n = 314) Count %a CV history None 184 59 Prior CADb 193 61 Cardiac heart failure 12 4 Atrial fibrillation 19 6 Stroke/transient ischaemic attack 16 5 Peripheral vascular disease 24 8 Major bleeding events 2 1 Other CV history 8 3 CV risk factors None 19 6 Arterial hypertension or antihypertensive drugs 198 63 Dyslipidemia or cholesterol lowering drugs 225 72 Diabetes or hypoglycaemic therapy 72 23 Active smoker 126 40 Familial history of coronary artery disease 95 30 Obesity (BMI >30) 55 18 Chronic kidney disease (GFR <60 mg/dl/min) 29 9 Other CV risk factors 2 1 n, number of patients; CV, cardiovascular; CAD, coronary artery disease; BMI, body mass index; GFR, glomerular filtration rate. a Total percentages are higher than 100 due to multiple responses possible per patient. b Prior CAD includes prior myocardial infarction, prior percutaneous coronary intervention, prior coronary artery bypass graft, coronary artery disease and stenosis >50%.

Count %a CV history None 184 59 Prior CADb 193 61 Cardiac heart failure 12 4 Atrial fibrillation 19 6 Stroke/transient ischaemic attack 16 5 Peripheral vascular disease 24 8 Major bleeding events 2 1 Other CV history 8 3 CV risk factors None 19 6 Arterial hypertension or antihypertensive drugs 198 63 Dyslipidemia or cholesterol lowering drugs 225 72 Diabetes or hypoglycaemic therapy 72 23 Active smoker 126 40 Familial history of coronary artery disease 95 30 Obesity (BMI >30) 55 18 Chronic kidney disease (GFR <60 mg/dl/min) 29 9 Other CV risk factors 2 1 n, number of patients; CV, cardiovascular; CAD, coronary artery disease; BMI, body mass index; GFR, glomerular filtration rate. a Total percentages are higher than 100 due to multiple responses possible per patient. b Prior CAD includes prior myocardial infarction, prior percutaneous coronary intervention, prior coronary artery bypass graft, coronary artery disease and stenosis >50%. Follow-up visits On average, the patients were visiting their cardiologist almost 4 times (3.7 ± 1.6) during the year following discharge. Half of the patients 163/314 (52%) had between 1 and 3 visits, 129/314 (41%) had between 4 and 6 visits and only 2/314 (1%) had up to 10–12 visits to the hospital or cardiologist during the follow-up period. The mean time interval between discharge and the last visit to hospital was 11.0 ± 2.7 months after discharge and the interval between hospital discharge and visit 1 was 1.1 ± 1.6 days (calculated for n = 308 patients).

only 2/314 (1%) had up to 10–12 visits to the hospital or cardiologist during the follow-up period. The mean time interval between discharge and the last visit to hospital was 11.0 ± 2.7 months after discharge and the interval between hospital discharge and visit 1 was 1.1 ± 1.6 days (calculated for n = 308 patients). Treatment persistence The treatment persistence analysis included 295 patients out of the 314 included in the study population (Figure 1). The measured treatment persistence with OAP after an ACS in the clinical practice of Belgium at 360 days from hospital discharge was 216/295 (73%) (Figure 2). For the 79 patients who stopped treatment before 360 days, the mean treatment duration was 197.0 ± 125.18 days (median 216 days). Figure 2 Persistence of medication during the treatment course for the analysed population (n = 295), by OAP type and overall. n, number of patients; OAP, oral antiplatelet. At 90, 180, and 270 days, respectively, the proportion of patients who were still using OAPs was 92, 89, and 83%. Treatment persistence for clopidogrel was lower compared with prasugrel or ticagrelor (Figure 2). As a consequence, the proportion of patients still treated at 360 days was higher among patients receiving at discharge prasugrel (45/56; 80%) and ticagrelor (117/147; 80%) compared with those receiving clopidogrel (54/92; 59%). No statistically significant difference was observed between patients receiving prasugrel or ticagrelor.

nce, the proportion of patients still treated at 360 days was higher among patients receiving at discharge prasugrel (45/56; 80%) and ticagrelor (117/147; 80%) compared with those receiving clopidogrel (54/92; 59%). No statistically significant difference was observed between patients receiving prasugrel or ticagrelor. At 360 days, 119/155 (77%) patients under 65 years were still on OAP treatment, compared with 97/140 (69%) patients older than 65 years. The percentage of men (166/224; 74%) and women (50/71; 70%) still treated at 360 days was similar (Table 3). Table 3 Treatment persistence at 360 days among the persistence population (n = 295)

s, 119/155 (77%) patients under 65 years were still on OAP treatment, compared with 97/140 (69%) patients older than 65 years. The percentage of men (166/224; 74%) and women (50/71; 70%) still treated at 360 days was similar (Table 3). Table 3 Treatment persistence at 360 days among the persistence population (n = 295) Count at treatment start Count at 360 days Persistence at 360 days (%) Age Less than 65 years 155 119 77 More or equal to 65 years 140 97 69 Gender Male 224 166 74 Female 71 50 70 ACS type Unstable angina 28 21 75 STEMI 149 117 79 NSTEMI 118 78 66 ACS management PCI 267 202 76 CABG 6 3 50 PCI + CAGB 4 2 50 Angiography without coronary intervention 7 5 71 Non-invasively managed 11 4 36 OAP treatment at discharge Clopidogrel 92 54 59 Prasugrel 56 45 80 Ticagrelor 147 117 80 CV history Prior CAD 178 126 71 Cardiac heart failure 12 5 42 Atrial fibrillation 17 13 76 Stroke/transient ischaemic attack 16 11 69 Peripheral vascular disease 21 14 67 Major bleeding events 2 1 50 CV risk factors Arterial hypertension or antihypertensive drugs 186 131 70 Dyslipidaemia or cholesterol lowering drugs 209 160 77 Diabetes or hypoglycaemic therapy 70 57 81 Active smoker 112 82 73 Familial history of coronary artery disease 90 68 76 Obesity (BMI >30) 53 40 75 Chronic kidney disease (GFR <60 mg/dl/min) 24 19 79 Concomitant medication Acetylsalicylic acid 285 210 74 Vitamin K antagonist 10 6 60 LWMH 7 5 71 Dabigatran 1 1 100 Rivaroxaban 4 2 50 None 5 3 60 n, number of patients; ACS, acute coronary syndrome; STEMI, ST-segment elevation myocardial infarction; NSTEMI, non-ST-segment elevation myocardial infarction; PCI, percutaneous coronary intervention; CABG, coronary artery by-pass grafting; OAP, oral antiplatelet; CV, cardiovascular; CAD, coronary artery disease; BMI, body mass index; GFR, glomerular filtration rate; LWMH, low molecular weight heparin.

ardial infarction; NSTEMI, non-ST-segment elevation myocardial infarction; PCI, percutaneous coronary intervention; CABG, coronary artery by-pass grafting; OAP, oral antiplatelet; CV, cardiovascular; CAD, coronary artery disease; BMI, body mass index; GFR, glomerular filtration rate; LWMH, low molecular weight heparin. At 360 days, 21/28 (75%) patients diagnosed with UA, 117/149 (79%) patients diagnosed with STEMI and 78/118 (66%) patients diagnosed with NSTEMI were still taking the OAP treatment prescribed at discharge. Most of the patients managed with PCI (202/267; 76%) and angiography without coronary intervention (5/7; 71%) vs. only half of those managed with coronary artery by-pass grafting (CABG) (3/6; 50%) or PCI and CABG (2/4; 50%) were still on OAP treatment at 360 days. Less than half of the non-invasively managed patients (4/11; 36%) continued OAP treatment at 360 days after discharge from the hospital (Table 3). Of the 285 patients taking acetylsalicylic acid at discharge from hospital, 210 (74%) were still using the prescribed OAP treatment at the end of the follow-up period (Table 3). The majority of patients in the persistence population receiving anticoagulants at discharge (64%; 14/22) continued OAP treatment after 360 days.

Of the 285 patients taking acetylsalicylic acid at discharge from hospital, 210 (74%) were still using the prescribed OAP treatment at the end of the follow-up period (Table 3). The majority of patients in the persistence population receiving anticoagulants at discharge (64%; 14/22) continued OAP treatment after 360 days. Changes in OAP treatment: reasons and deciders In the 11-month interval from discharge, 217/295 (74%) of the patients did not change the treatment. Sixty-one out of 295 (21%) patients stopped any OAP treatment before 11 months, 8 (3%) patients stopped and subsequently reinitiated and continued treatment with the same OAP, and 9 (3%) patients continued the treatment with another OAP. The main specific reasons for stopping the treatment were minor surgery (25%) and perceived high-bleeding risk (19%) (Figure 3). Lack of treatment reimbursement was reported as a reason for OAP discontinuation in only 4% of cases, and 2 (4%) of patients discontinued the treatment of their own accord; however, the reasons for this choice were not recorded in the case report forms. In total, patient death accounted for 15% of treatment cessation, treatment disruption related to the patient (due to non-compliance or bleeding) was recorded in 10% of cases, planned treatment interruption due to an upcoming invasive procedure occurred for 28% of instances, and in 47% of cases the discontinuation was recommended by the physician.

nted for 15% of treatment cessation, treatment disruption related to the patient (due to non-compliance or bleeding) was recorded in 10% of cases, planned treatment interruption due to an upcoming invasive procedure occurred for 28% of instances, and in 47% of cases the discontinuation was recommended by the physician. Figure 3 Reasons for stopping OAP treatment before 11 months. OAP, oral antiplatelet; CAGB, coronary artery bypass graft; FXA, coagulation factor X; AF, atrial fibrillation. Note: Other reasons were: standard practice; palliative care; no objective reasons/reasons related to patient. Twelve instances in which the OAP was changed before 11 months were recorded (in 3 instances the treatment was then stopped); 8 (67%) of these treatment switches were triggered by side-effects (non-bleeding complications). Other incidental reasons for OAP switch included cost of treatment (1/12; 8%), high-bleeding risk (1/12; 8%), a new CV event or revascularization (1/12; 8%), or unknown reason (1/12; 8%). Fifteen OAP treatment re-initiations before 11 months were observed (treatment was stopped again prematurely in 4 instances). About half of the re-initiations occurred after a new CV event (7/15; 47%) and 5 (33%) following surgery. Two patients were re-challenged after side-effects and 2/15 for other reasons.

Twelve instances in which the OAP was changed before 11 months were recorded (in 3 instances the treatment was then stopped); 8 (67%) of these treatment switches were triggered by side-effects (non-bleeding complications). Other incidental reasons for OAP switch included cost of treatment (1/12; 8%), high-bleeding risk (1/12; 8%), a new CV event or revascularization (1/12; 8%), or unknown reason (1/12; 8%). Fifteen OAP treatment re-initiations before 11 months were observed (treatment was stopped again prematurely in 4 instances). About half of the re-initiations occurred after a new CV event (7/15; 47%) and 5 (33%) following surgery. Two patients were re-challenged after side-effects and 2/15 for other reasons. The decision taker for either treatment stopping or switching to another OAP were in most instances an interventional cardiologist (19/61; 31% and 5/12; 42%, respectively) followed by a non-interventional cardiologist (18/61; 30% and 4/12; 33%, respectively), whereas the decision to re-initiate was in almost half of the instances made by a surgeon (5/11; 45%).

opping or switching to another OAP were in most instances an interventional cardiologist (19/61; 31% and 5/12; 42%, respectively) followed by a non-interventional cardiologist (18/61; 30% and 4/12; 33%, respectively), whereas the decision to re-initiate was in almost half of the instances made by a surgeon (5/11; 45%). Multivariate analysis In a multivariate analysis, ACS management, choice of OAP at discharge, CV history and CV risk were significantly associated with OAP treatment persistence (Table 4). The type of ACS only showed a trend of association (P = 0.0714). Patients treated by PCI were more likely to persist on DAPT than those non-invasively managed (odds ratio = 8.084; P = 0.00756). Conversely, patients discharged on clopidogrel were less likely to be adherent to the treatment up to 1 year, when compared with those using ticagrelor (calculated odds ratio = 0.250, P = 0.00046). Table 4 Summary of logistical regression analysis for variables predicting OAP treatment persistence with age as continuous variable

, patients discharged on clopidogrel were less likely to be adherent to the treatment up to 1 year, when compared with those using ticagrelor (calculated odds ratio = 0.250, P = 0.00046). Table 4 Summary of logistical regression analysis for variables predicting OAP treatment persistence with age as continuous variable Variables Effects OR 95% CI P-value Age (years) – 0.975 0.944–1.006 0.11803 Gender Male vs. female 0.811 0.372–1.770 0.59904 Hospital type Academic vs. non-academic PCI 2.019 0.872–4.674 0.10095 Non-academic non-PCI vs. non-academic PCI 1.816 0.751–4.391 0.18521 ACS diagnosis Unstable angina vs. NSTEMI 3.728 0.914–15.205 0.06660 STEMI vs. NSTEMI 1.996 0.942–4.233 0.07140 ACS management PCI vs. non-invasively managed 8.084 1.744–37.471 0.00756 CABG vs. non-invasively managed 4.747 0.413–54.574 0.21130 PCI+CABG vs. non-invasively managed 0.862 0.044–17.008 0.92204 Angiography without coronary intervention vs. non-invasively managed 8.949 0.651–123.099 0.10132 OAP at discharge Clopidogrel vs. ticagrelor 0.250 0.115–0.543 0.00046 Prasugrel vs. ticagrelor 0.916 0.322–2.608 0.86930 CV historya Prior CABG 7.317 1.208–44.314 0.03033 Congestive heart failure 0.136 0.023–0.790 0.02624 CV risk factorsb Arterial hypertension or antihypertensive drugs 0.409 0.192–0.869 0.02019 Dyslipidemia or cholesterol lowering drugs 2.416 1.129–5.174 0.02313 Diabetes or hypoglycaemic therapy 4.185 1.653–10.595 0.00252 Bolded values indicate significant associations with OAP treatment persistence (P < 0.05).

0.02624 CV risk factorsb Arterial hypertension or antihypertensive drugs 0.409 0.192–0.869 0.02019 Dyslipidemia or cholesterol lowering drugs 2.416 1.129–5.174 0.02313 Diabetes or hypoglycaemic therapy 4.185 1.653–10.595 0.00252 Bolded values indicate significant associations with OAP treatment persistence (P < 0.05). OR, odds ratio; CI, confidence interval; PCI, percutaneous coronary intervention; ACS, acute coronary syndrome; NSTEMI, non-ST-segment elevation myocardial infarction; STEMI, ST-segment elevation myocardial infarction; CABG, coronary artery by-pass grafting; OAP, oral antiplatelet; CV, cardiovascular; MI, myocardial infarction; PCI, percutaneous coronary intervention.. a CV history included 9 categories: prior MI, prior PCI, prior CABG, coronary artery disease and stenosis >50%, congestive heart failure, atrial fibrillation, stroke/transient ischemic attack, peripheral vascular disease and major bleeding events. Only those with significant association are presented. b CV risk factors included 7 categories: arterial hypertension or antihypertensive drugs, dyslipidaemia or cholesterol lowering drugs, diabetes or anti-diabetes drugs, active smoker, familial history of coronary artery disease, obesity (body mass index >30) and chronic kidney disease. Only those with significant association are presented.

cluded 7 categories: arterial hypertension or antihypertensive drugs, dyslipidaemia or cholesterol lowering drugs, diabetes or anti-diabetes drugs, active smoker, familial history of coronary artery disease, obesity (body mass index >30) and chronic kidney disease. Only those with significant association are presented. NSTEMI, clopidogrel treatment, congestive heart failure, arterial hypertension, or antihypertensive drugs were all associated with shorter OAP treatment persistence while prior CABG, dyslipidaemia, or cholesterol lowering drugs, diabetes or anti-diabetes drugs were associated with longer OAP treatment persistence (Table 4). Discussion The main finding of this retrospective study is that the majority of Belgian ACS patients remain adherent to the recommended course of OAP treatment, in line with the European Society of Cardiology guidelines.2 Premature OAP treatment discontinuation was observed mainly in the last 3 months of the year following the index event. The discontinuation was more prevalent in patients treated with clopidogrel and in most cases, it was decided by the cardiologist. The most frequent reasons for premature treatment cessation were minor surgery and high bleeding risk.

uation was observed mainly in the last 3 months of the year following the index event. The discontinuation was more prevalent in patients treated with clopidogrel and in most cases, it was decided by the cardiologist. The most frequent reasons for premature treatment cessation were minor surgery and high bleeding risk. A systematic review6 assessing adherence to DAPT after coronary stenting emphasized an increasingly persistent DAPT use at 12 months starting with the year 2007, when ACC/AHA guidelines were changed to recommend a 12-month duration of DAPT. The use of OAPs at 12 months after drug-eluting stent implementation increased from 63.8% for studies ending between 2004 and 2006 to 78.1% for those ending between 2007 and 2009. A decline in OAP adherence was evidenced by Month 6 of the treatment period.6 These trends mirror those of our study: the overall assessed persistence to OAP treatment showed a decrease in time, with more than 90% of the patients still using OAPs at Month 3, to 73% by the end of 1 year of treatment. The lowest persistence at the end of the follow-up period was found for patients receiving clopidogrel (54%), probably due to large extent to their comorbidities, including the more frequent need for oral anticoagulation. The difference between clopidogrel and prasugrel/ticagrelor treatment persistence was not fully explained by data available from the current study, but further investigation might prove worthwhile. The percentages observed at every 90-day interval are similar to those found in a previous study on DAPT medication adherence for clopidogrel and aspirin.7 Better medication persistence was found in our study for both prasugrel and ticagrelor at each time point, accompanied by fewer discontinuations between assessments. Similarly to our observations, a recent study in the USA found no difference between adherences to prasugrel and ticagrelor in patients with ACS treated with PCI, while mean persistence was marginally longer with prasugrel than with ticagrelor.8

ach time point, accompanied by fewer discontinuations between assessments. Similarly to our observations, a recent study in the USA found no difference between adherences to prasugrel and ticagrelor in patients with ACS treated with PCI, while mean persistence was marginally longer with prasugrel than with ticagrelor.8 OAP adherence may be influenced by the variety of treatment options, comorbidities, and prior use of the medication, as well as reimbursement restrictions. A study on clopidogrel use after discharge following PCI in the US shows that premature discontinuation was more likely in patients younger than 55 years, in patients having been hospitalized previously or undergoing PCI without stenting, or in patients with chronic obstructive pulmonary diseases and diabetes.9 Prior clopidogrel use was also significantly associated with lower adherence, while prior use of β-blockers and statins was associated with longer medication adherence.9 In a study assessing adherence to prasugrel treatment following an ACS event in US patients, prior PCI, depression and bleeding were found to be indicative of lower adherence, while baseline statin use predicted better adherence to treatment.10 In our study, prior PCI did not significantly impact adherence to treatment, but prior CABG, dyslipidaemia, and diabetes or associated treatment were identified as predictors for OAP treatment persistence, while arterial hypertension or use of antihypertensive drugs were more likely to lead to lower persistence. The type of invasive management was also found to be of importance, as patients with PCI are more likely to display higher medication persistence than those for which the ACS was managed non-invasively. It is likely that a medical, non-invasive approach might lead to patients (or physicians) not realizing the importance of the ACS event. Similar to a previous study on medication persistence among NSTEMI patients,11 no significant association was identified between the hospital type (academic vs. non-academic) and medication persistence, regardless of the type of OAP. Although patients treated at academic sites were believed to be more likely to receive medication treatment,12 there is no conclusive study showing an improvement in medical adherence.

ciation was identified between the hospital type (academic vs. non-academic) and medication persistence, regardless of the type of OAP. Although patients treated at academic sites were believed to be more likely to receive medication treatment,12 there is no conclusive study showing an improvement in medical adherence. DAPT discontinuation after PCI was analysed in the PARIS observational study in clinical hospitals in Europe and USA.13 The study determined that the main reason for treatment cessation is physician guidance, while bleeding and patient non-compliance was recorded for only 10% of the patients. In line with these observations, in our study, patient-related treatment disruption was identified in 10% of cases and physician-recommended discontinuation accounted for 47% of cases. Physician discretion was also recently shown to account for approximately one-half of reasons for premature cessation of treatment with P2Y12 receptor inhibitors following MI, while only in ∼8% of instances a switch to another drug class was recommended.14 Our study also identified the cardiologist (either interventional or non-interventional) as the main decision taker for treatment discontinuation, as opposed to the GP or the patient.

treatment with P2Y12 receptor inhibitors following MI, while only in ∼8% of instances a switch to another drug class was recommended.14 Our study also identified the cardiologist (either interventional or non-interventional) as the main decision taker for treatment discontinuation, as opposed to the GP or the patient. The majority of reasons for OAP treatment cessation in this study were objective. Surgery and perceived high-bleeding risk were frequently cited as the driving factor for a shorter DAPT course, in line with contemporary guidelines.2 Cost-related reasons for treatment discontinuation following an ACS event have been frequently reported before.14,15 Due to the particularities of the Belgian healthcare system (i.e. compulsory state-based health insurance and universal reimbursement for post-ACS DAPT), cost was found to be of minor importance here, with only 4% of the patients reporting discontinuation of the treatment for lack of reimbursement. The patients included in the current study had comparable baseline characteristics to other Western European patients included in the EURHOBOP study.16 Therefore, the study results can be extended to the Belgian and European population, within the limitations of observational analysis.

The majority of reasons for OAP treatment cessation in this study were objective. Surgery and perceived high-bleeding risk were frequently cited as the driving factor for a shorter DAPT course, in line with contemporary guidelines.2 Cost-related reasons for treatment discontinuation following an ACS event have been frequently reported before.14,15 Due to the particularities of the Belgian healthcare system (i.e. compulsory state-based health insurance and universal reimbursement for post-ACS DAPT), cost was found to be of minor importance here, with only 4% of the patients reporting discontinuation of the treatment for lack of reimbursement. The patients included in the current study had comparable baseline characteristics to other Western European patients included in the EURHOBOP study.16 Therefore, the study results can be extended to the Belgian and European population, within the limitations of observational analysis. The results of this study will hopefully improve the awareness of Belgian physicians and patients on the actual, real-world OAP treatment persistence. Although already relatively high, OAP persistence might still be improved, as premature DAPT treatment discontinuation could be associated with significant risk. Providing the patient with a list of instructions, explaining the reasons and side-effect for each medication and a consistent follow-up planned before discharge have been associated with higher medication persistence after an MI event.14

emature DAPT treatment discontinuation could be associated with significant risk. Providing the patient with a list of instructions, explaining the reasons and side-effect for each medication and a consistent follow-up planned before discharge have been associated with higher medication persistence after an MI event.14 The study has some potential limitations. First, the number of patients with complete dataset included in the analysis was lower compared with the estimated number of patients. Moreover, the patients with complete data are most likely to be those with better adherence to medication and follow-up plan, either induced by the physician or self-motivated, while patients not willing to adhere to the follow-up plan are conceivably less likely to continue treatment as well. The retrospective design of the study was probably the reason for the relatively high rate of missing data,17 but at the same time it allowed the collection of accurate “real-life” data on OAP treatment and/or its duration. On the contrary, a prospective follow-up registry has the disadvantage of a positive bias towards a better treatment persistence as both physician and patients are more engaged in the project, and will be probably more aware about the benefit of treatment persistence.

al-life” data on OAP treatment and/or its duration. On the contrary, a prospective follow-up registry has the disadvantage of a positive bias towards a better treatment persistence as both physician and patients are more engaged in the project, and will be probably more aware about the benefit of treatment persistence. Taken together, a high OAP treatment persistence was observed in Belgian patients throughout the year following an ACS. OAP treatment was prematurely discontinued mainly in the last 3 months of the year following the index event. Discontinuation was observed especially in patients treated with clopidogrel and was most often initiated by the patient’s cardiologist.

ce was observed in Belgian patients throughout the year following an ACS. OAP treatment was prematurely discontinued mainly in the last 3 months of the year following the index event. Discontinuation was observed especially in patients treated with clopidogrel and was most often initiated by the patient’s cardiologist. Acknowledgements The authors would like to thank all the investigators of the Rewinder Study Group for the collection and cleaning of data (in alphabetical order): Frank Cools (General Hospital Klina, Brasschaat), Yves Dascotte (Grand Hôpital de Charleroi, Charleroi), Filip De Man (Hospital St Elisabeth, Brussels), Philippe Decroly (Centre Hospitalier Epicura, Baudour), Lucien Finianos (Clinique de l'Espérance, Liège), Sofie Gevaert (University Hospital, Gent), Alex Heyse (General Hospital, Glorieux), Luc Janssens (Imelda Hospital, Bonheiden), Luc Missault (General Hospital Sint-Jan, Brugge), Jan Nimmegeers (General Hospital Sint-Lucas, Gent), Nicolas Preumont (Hospital Erasmus, Brussels), Paul-Gaël Silance (University Hospital Saint-Pièrre, Brussels), Jeroen Sonck (University Hospital, Jette) and Carlos Van Mieghem (Hospital Onze-Lieve-Vrouw, Aalst). The authors would like to acknowledge Virginie Durbecq (XPE Pharma & Science, on behalf of AstraZeneca) for project management, Wouter Houthoofd (XPE Pharma & Science, on behalf of AstraZeneca) for manuscript coordination, and Iudit-Hajnal Filip and Petronela M. Petrar (XPE Pharma & Science) for medical writing services and editorial support in preparing this manuscript.

q (XPE Pharma & Science, on behalf of AstraZeneca) for project management, Wouter Houthoofd (XPE Pharma & Science, on behalf of AstraZeneca) for manuscript coordination, and Iudit-Hajnal Filip and Petronela M. Petrar (XPE Pharma & Science) for medical writing services and editorial support in preparing this manuscript. Funding This work was supported by AstraZeneca, who also covered all costs associated with the development and publishing of the manuscript. Conflict of interest: M.C. has received honoraria and/or advisory board fee from AstraZeneca, Eli Lilly, and Bayer. S.P. and C.B. have received financial support from AstraZeneca for the conduct of this study. P.S. is a Clinical Investigator for the Fund for Scientific Research, Flanders and received institutional honoraria from AstraZeneca, Eli Lilly, Daiichi Sankyo, BMS and Sanofi, and has received modest institutional grants from Astra Zeneca and Daiichi Sankyo.

Low-dose aspirin is used worldwide for preventing thromboembolic disorders. Its use, however, is often associated with gastrointestinal bleeding, mostly due to direct irritation of the gastric mucosa. Here we provide evidence for a novel sublingual formulation of aspirin micronized and co-grinded with collagen proven to be as effective as oral standard formulation in inhibiting platelet aggregation but with attenuated gastric irritation. This represents a new option for better aspirin treatment in the prevention of myocardial infarction and stroke. Orally given low-dose aspirin has been used for decades due to its anti-inflammatory and antithrombotic properties.1 Although standard oral formulation of aspirin allows rapid and complete absorption from the GI tract, new formulations have been developed and marketed, (e.g. dry granules, effervescent solution, and chewable tablets)2,3 with the aim to achieve faster dissolution and faster absorption4,5 as well as to reduce direct aspirin-induced gastric lesions. However, the occurrence of gastrointestinal bleeding still remains a significant problem of chronic aspirin administration and, sometimes, limits the use of aspirin in primary prevention.6,7 On the other hand, co-administration of proton pump inhibitors is currently used to counteract aspirin-induced gastric lesions,8,9 thereby representing a pharmaco-economic issue in the area of health care sustainability.

chronic aspirin administration and, sometimes, limits the use of aspirin in primary prevention.6,7 On the other hand, co-administration of proton pump inhibitors is currently used to counteract aspirin-induced gastric lesions,8,9 thereby representing a pharmaco-economic issue in the area of health care sustainability. Recently, we developed and patented (N. 102015000079955) a new formulation of aspirin which leads to faster absorption and activity but devoid of direct gastrointestinal lesioning effect. In addition, this formulation allows sublingual administration of the drug which, by passing the liver metabolism, leads to faster serum peak concentration with more prominent and rapid inhibition of cycloxygenase (COX), the major target of antiplatelet and antinflammatory action of aspirin. Methods Cristalline aspirin was amorphized via micronization and co-grinding with collagen (see Supplementary material online, Materials & Methods). Collagen was chosen due to its gastro-protective action.10 The occurrence of aspirin amorphization was demonstrated via measurement of spectra collected with Raman spectroscopy (Figure 1).

ne aspirin was amorphized via micronization and co-grinding with collagen (see Supplementary material online, Materials & Methods). Collagen was chosen due to its gastro-protective action.10 The occurrence of aspirin amorphization was demonstrated via measurement of spectra collected with Raman spectroscopy (Figure 1). Figure 1 Micro-Raman spectra were excited by a 514 nm laser line through a 50X objective with a laser power of 10 mW at the sample level. The samples were deposited as powder on a calcium fluoride slide, and measurements were performed with an accumulation time of 30 s, in the range from 500 up to 4000 cm−1. In (A) the image of the chemical compost aspirin in crystalline form. Raman spectra in (B) relative to the analysis of the compost. The measures yield a spectrum with very sharp, intense Raman peaks. In (C) the optical image of the dried mixture composed of the amorphous material and the crystalline aspirin. The amorphous form is due to the presence of collagen. RAMAN measurements (D), collected in the marked points of the optical image, show broader less intense Raman peaks (indicated by the black square) in the presence of the amorphous species and sharp peaks (red squares) related to aspirin molecules.

stalline aspirin. The amorphous form is due to the presence of collagen. RAMAN measurements (D), collected in the marked points of the optical image, show broader less intense Raman peaks (indicated by the black square) in the presence of the amorphous species and sharp peaks (red squares) related to aspirin molecules. Micronized collagen co-grinded aspirin formulations were developed for both oral and sublingual administration (see Supplementary material online, Materials & Methods) and used in healthy volunteers, in a phase 1, randomized, double blind, placebo-controlled study (study EudraCT N. 2013-002980-24) to verify their pharmacokinetic profile compared to standard crystalline formulations of aspirin (see Supplementary material online, Clinical Study Protocol and healthy volunteer demographics). Measurement of serum thromboxane B2 (TXB2), and its urinary metabolite 11-dehydro-TXB2 after both oral and sublingual standard as well as micronized aspirin were carried out to assess the inhibitory activity at the COX level.

ee Supplementary material online, Clinical Study Protocol and healthy volunteer demographics). Measurement of serum thromboxane B2 (TXB2), and its urinary metabolite 11-dehydro-TXB2 after both oral and sublingual standard as well as micronized aspirin were carried out to assess the inhibitory activity at the COX level. Results Oral administration of 100 mg of standard crystalline aspirin (n = 10 healthy volunteers) for 7 consecutive days produced a rapid rise of acetylsalicylic acid serum concentration which peaked at 2–4 h after the administration and declined 8–12 h later (Figure 2–F). This effect was accompanied by a decrease of TXB2 serum concentration at 6 h which declined 16–20 h after the administration. (Figure 2G). A similar response was seen in healthy volunteers (n = 10) receiving 50 mg of standard aspirin orally, with lower effect compared to 100 mg, (Figure 2A–F). No significant differences were seen when 50 and 100 mg of micronized and collagen co-grinded aspirin were given orally to healthy volunteers (n = 10 for each dose; Figure 2A–F)

ponse was seen in healthy volunteers (n = 10) receiving 50 mg of standard aspirin orally, with lower effect compared to 100 mg, (Figure 2A–F). No significant differences were seen when 50 and 100 mg of micronized and collagen co-grinded aspirin were given orally to healthy volunteers (n = 10 for each dose; Figure 2A–F) Figure 2 Plasma concentration vs. time profile of acetylsalicylic acid after oral and sublingual administration of aspirin either standard crystalline and micronized and co-grinded with collagen (MC + coll) (detected via LC-MS and expressed as ng/mL−1) of healthy subjects (n = 10 for each treatment) after dosing with 50 and 100 mg/daily of both formulations (Graphs A–D). Curves display changes of levels of acetylsalicylic acid after administration of the first and seventh dose of aspirin of both formulations. Graphs E and F show the area under the curve (AUC calculated as ng/dL/h) and Cmax (expressed as ng/dL), respectively, of acetylsalicylic acid serum concentrations after oral or sublingual administration of both aspirin formulations (standard or MC + coll) given in healthy volunteers (n = 10 for each treatment). In G are shown the changes of TXB2 serum levels detected via ELISA immunoassay in subjects treated with oral standard aspirin vs. sublingual administration of MC + coll formulation. Data show that the decrease of serum TXB2 occurs earlier in subjects treated sublingually with aspirin MC + coll compared to the standard oral administration of 50 and 100 mg, respectively. Similar effect was seen in urinary 11-dehydro-TXB2, the metabolite of TXB2, thus reflecting a better effect of sublingual aspirin MC + coll formulation on platelet COX enzyme.

urs earlier in subjects treated sublingually with aspirin MC + coll compared to the standard oral administration of 50 and 100 mg, respectively. Similar effect was seen in urinary 11-dehydro-TXB2, the metabolite of TXB2, thus reflecting a better effect of sublingual aspirin MC + coll formulation on platelet COX enzyme. In contrast, sublingual administration of 50 and 100 mg (n = 10 for each dose) of aspirin micronized and co-grinded with collagen, produced a dose-related peak of serum concentration of acetylsalicylic acid which occurred earlier compared to both sublingual and oral administration of aspirin (1 h) with a decline of serum concentration occurring 4–5 h after the administration (n = 10 for each dose; Figure 2A–F). This effect was accompanied by early inhibition of TXB2 levels which was observed after 2 h and lasted 10 h after aspirin administration. Furthermore, the effect of aspirin formulation in TXB2 was confirmed by detection of its urinary metabolite11-dehydro-TXB2 after Day 7 of the study (Figure 2H), thus suggesting that sublingual administration of micronized co-grinded aspirin displays a non-inferiority response on COX enzyme compared to crystalline standard formulation. Determination of serum TXB2 serum levels and of urinary 11-dehydro-TXB2 showed no changes before and after treatment in healthy volunteers receiving placebo (not shown).

ingual administration of micronized co-grinded aspirin displays a non-inferiority response on COX enzyme compared to crystalline standard formulation. Determination of serum TXB2 serum levels and of urinary 11-dehydro-TXB2 showed no changes before and after treatment in healthy volunteers receiving placebo (not shown). Neither changes on routine blood analytical biomarkers nor side effects or adverse drug reactions were noted in any of the groups after administration of oral or sublingual aspirin. Pill count adherence was 100% and no enrolled subject was excluded from the study (see Supplementary material online).

ingual administration of micronized co-grinded aspirin displays a non-inferiority response on COX enzyme compared to crystalline standard formulation. Determination of serum TXB2 serum levels and of urinary 11-dehydro-TXB2 showed no changes before and after treatment in healthy volunteers receiving placebo (not shown). Neither changes on routine blood analytical biomarkers nor side effects or adverse drug reactions were noted in any of the groups after administration of oral or sublingual aspirin. Pill count adherence was 100% and no enrolled subject was excluded from the study (see Supplementary material online). Finally, to verify the attenuated impact on gastric mucosa of micronized collagen co-grinded aspirin compared to standard oral formulation, experiments have been carried out in rats receiving doses of aspirin proven to produce gastric lesions (see Supplementary material online, Materials & Methods). In particular, acute oral administration of aspirin (400 mg/Kg) both crystalline or micronized and co-grinded with collagen produced gastric lesion with an elevated ulcer score index.11 In particular 83% of the stomachs in the group of rats treated with standard aspirin contained one or more lesions and mean lesion score was 23.7 ± 3–5. The severity of ulceration was however reduced by 73 ± 6% in the gastric tissues when aspirin was given micronized and co-grinded with collagen. This was confirmed by evaluating microphotographs of gastric mucosa stained. In particular, we have found that standard aspirin formulation leads to production of severe erosions marked by the presence of heterogeneous mixture of tissues retracting from the mucosal surface. A variety of surface epithelial changes in shape, size, and orientation, accompanied by marked loss of surface mucus epithelial cell were found. Marked disorganization and atrophy of glands were invariably noted. This seems to be prevented when aspirin is micronized and co-grinded with collagen, which maintains the adherent mucus lining resisting the erosion of glandular cells. No significant gastric mucosal lesion was observed in control group of rats.

found. Marked disorganization and atrophy of glands were invariably noted. This seems to be prevented when aspirin is micronized and co-grinded with collagen, which maintains the adherent mucus lining resisting the erosion of glandular cells. No significant gastric mucosal lesion was observed in control group of rats. Conclusion Our data show that sublingual formulation of aspirin micronized and co-grinded with collagen displays a better pharmakinetic profile compared with standard crystalline aspirin. This effect was accompanied by attenuated direct gastric ulcerogenic effect of the new formulation and by non-inferiority profile on TXB2 serum and urinary levels after a 7-day treatment compared to the standard formulation of aspirin. The potential remaining warning on gastrointestinal bleeding due to inhibition on prostaglandin-related production of protective gastric mucus of the new formulation is to be better clarified. This represents a new option for better aspirin treatment in the prevention of thromboembolic disorders. Supplementary material Supplementary material is available at European Heart Journal—Cardiovascular Pharmacotherapy online. Conflict of interest: none declared. Supplementary Material Supplementary Material Click here for additional data file.

Introduction Since 2010, direct oral anticoagulants (DOACs) have been available for clinical use in the US for the prevention of thromboembolic stroke and related disorders. Based on large, randomized controlled trials (RCTs) comparing DOACs with warfarin in patients with non-valvular atrial fibrillation (NVAF),1–4 current treatment guidelines recommend them as safe and efficacious alternatives to oral anticoagulation with vitamin K antagonists/warfarin.5 Despite this, concerns remained about the safety of these agents when applied to a broader clinical practice population, typically with multiple morbidities. Large-scale ‘real-world’ studies comparing DOACs with warfarin have served to reassure healthcare providers that the results of RCTs generally translated to routine clinical care.6,7

ns remained about the safety of these agents when applied to a broader clinical practice population, typically with multiple morbidities. Large-scale ‘real-world’ studies comparing DOACs with warfarin have served to reassure healthcare providers that the results of RCTs generally translated to routine clinical care.6,7 A similar approach has been used to compare DOACs, again because of a lack of ‘head-to-head’ RCTs.8,9 For example, Medicare data were used by Graham et al.10 to study outcomes in elderly, propensity score matched (PSM) patients with NVAF who initiated treatment with dabigatran or rivaroxaban. The investigators concluded that treatment with standard-dose rivaroxaban was associated with statistically significant increases in intracranial haemorrhage and major extracranial bleeding, including major gastrointestinal (GI) bleeding, compared with standard-dose dabigatran. However, the mean follow-up in this study was less than 4 months (108 days and 111 days for the dabigatran and rivaroxaban groups, respectively). Additional studies with a longer follow-up, and involving a greater diversity of patients in real-world clinical settings, would add to the available body of evidence. The purpose of the present study was to compare the safety and effectiveness of dabigatran, rivaroxaban, and apixaban in a large-scale, real-world cohort of patients with NVAF, who were newly initiated on standard doses.

A similar approach has been used to compare DOACs, again because of a lack of ‘head-to-head’ RCTs.8,9 For example, Medicare data were used by Graham et al.10 to study outcomes in elderly, propensity score matched (PSM) patients with NVAF who initiated treatment with dabigatran or rivaroxaban. The investigators concluded that treatment with standard-dose rivaroxaban was associated with statistically significant increases in intracranial haemorrhage and major extracranial bleeding, including major gastrointestinal (GI) bleeding, compared with standard-dose dabigatran. However, the mean follow-up in this study was less than 4 months (108 days and 111 days for the dabigatran and rivaroxaban groups, respectively). Additional studies with a longer follow-up, and involving a greater diversity of patients in real-world clinical settings, would add to the available body of evidence. The purpose of the present study was to compare the safety and effectiveness of dabigatran, rivaroxaban, and apixaban in a large-scale, real-world cohort of patients with NVAF, who were newly initiated on standard doses. Methods Data source The Department of Defense (DoD) Military Health System is a large-scale, comprehensive system with more than 9 million beneficiaries who generally have long-term coverage and extended treatment follow-up in comparison with most patients in commercial insurance plans.11,12 The DoD claims database has previously been used to study oral anticoagulation with dabigatran or warfarin in PSM patients (n = 12 793 for both groups). Dabigatran-treated patients had lower rates of stroke, major intracranial bleeding, urogenital bleeding, and other bleeding, as well as fewer myocardial infarctions (MIs) and deaths than warfarin-treated patients. While rates of major bleeding and major GI bleeding were similar in both groups, major lower GI bleeding events were more frequent in the dabigatran-treated patients.7

intracranial bleeding, urogenital bleeding, and other bleeding, as well as fewer myocardial infarctions (MIs) and deaths than warfarin-treated patients. While rates of major bleeding and major GI bleeding were similar in both groups, major lower GI bleeding events were more frequent in the dabigatran-treated patients.7 Study design This retrospective study compared outcomes in two cohorts of DoD patients with NVAF who were newly initiating DOAC therapy: a dabigatran vs. rivaroxaban cohort and a dabigatran vs. apixaban cohort. To improve comparability and minimize potential bias, comparisons were only made following the approval dates of both compared medications (1 July 2011 for the dabigatran vs. rivaroxaban cohort, 28 December 2011 for the dabigatran vs. apixaban cohort). Although rivaroxaban was approved for stroke prevention in atrial fibrillation (SPAF) on 4 November 2011, the venous thromboembolism approval date for rivaroxaban was used for this analysis. It was confirmed that no one receiving rivaroxaban prior to SPAF was included in the study based on stringent inclusion/exclusion criteria of no other DOAC alternative indication usage allowed. At the time of the analysis, data were available to 30 June 2016.

sm approval date for rivaroxaban was used for this analysis. It was confirmed that no one receiving rivaroxaban prior to SPAF was included in the study based on stringent inclusion/exclusion criteria of no other DOAC alternative indication usage allowed. At the time of the analysis, data were available to 30 June 2016. The index date (baseline) for each patient was defined as the date of their first claim of a DOAC prescription. The pre-index period (lookback period) was the 12 months before the first DOAC prescription claim, during which all patients were to have a NVAF diagnosis and be oral anticoagulation treatment-naïve (defined as having no claim for any oral anticoagulant in the pre-index period). Including the pre-index period, the time frames were 1 July 2010 to 30 June 2016 for the dabigatran vs. rivaroxaban cohort, and 28 December 2011 to 30 June 2016 for the dabigatran vs. apixaban cohort.

e oral anticoagulation treatment-naïve (defined as having no claim for any oral anticoagulant in the pre-index period). Including the pre-index period, the time frames were 1 July 2010 to 30 June 2016 for the dabigatran vs. rivaroxaban cohort, and 28 December 2011 to 30 June 2016 for the dabigatran vs. apixaban cohort. Patient follow-up began the day after the DOAC index date, and ended on the earliest occurrence of either (i) discontinuation of the index DOAC exposure (index exposure was considered discontinued if there was a treatment gap longer than the 30-day allowable gap specified from the end of the calculated days supplied), (ii) switching to a different anticoagulant, (iii) a change in index DOAC dosing, (iv) disenrolment, or (v) death. If a patient discontinued the index DOAC or switched to a different anticoagulation therapy, the outcomes assessment did not continue beyond the date of discontinuation or the switch (i.e. there was no latency period). All outcomes were studied using on-treatment analyses rather than initial treatment carried forward analyses.

ient discontinued the index DOAC or switched to a different anticoagulation therapy, the outcomes assessment did not continue beyond the date of discontinuation or the switch (i.e. there was no latency period). All outcomes were studied using on-treatment analyses rather than initial treatment carried forward analyses. Protection of human subjects This study was reviewed by the Naval Medical Center Portsmouth Institutional Review Board, and conducted in compliance with applicable federal and state laws, including the Health Insurance Portability and Accountability Act of 1996 (HIPAA). Informed consent was waived due to the retrospective nature of the study. All patient data were fully de-identified in compliance with HIPAA regulations, to ensure adherence to the Privacy Rule and to safeguard patient confidentiality. Each patient was assigned a unique, random 15-digit identifier used to link data collected through the retrospective query of the Military Health System Data Repository. A DoD clinical epidemiologist certified the data de-identification, and the study Project Manager maintained a master log.

rd patient confidentiality. Each patient was assigned a unique, random 15-digit identifier used to link data collected through the retrospective query of the Military Health System Data Repository. A DoD clinical epidemiologist certified the data de-identification, and the study Project Manager maintained a master log. Patients Study subjects were treatment-naïve patients with NVAF, who then had ≥1 prescription claim for dabigatran, rivaroxaban, or apixaban during the patient identification period. Patients were treated according to the recommended standard dosing regimen for each DOAC (apixaban, 5 mg b.i.d.; dabigatran, 150 mg b.i.d.; rivaroxaban, 20 mg QD). Patients receiving edoxaban were not included in this analysis as edoxaban represents a very small fraction (∼0.5%) of the total market for oral anticoagulants in the US.13

eated according to the recommended standard dosing regimen for each DOAC (apixaban, 5 mg b.i.d.; dabigatran, 150 mg b.i.d.; rivaroxaban, 20 mg QD). Patients receiving edoxaban were not included in this analysis as edoxaban represents a very small fraction (∼0.5%) of the total market for oral anticoagulants in the US.13 Included patients were aged ≥18 years on the index date, had ≥12 months of continuous eligibility prior to the index date, and had ≥1 diagnosis code of atrial fibrillation (AF), defined as ICD-9-CM diagnosis of 427.31 or ICD-10-CM diagnosis of I48.0, I48.1, I48.2, and I48.91 on the index date or during the pre-index period. Key exclusion criteria included any claim suggesting transient AF in the 3-month pre-index period, any claim suggesting that the patient had ‘valvular’ AF in the pre-index period, or any instance of cardiac surgery, pericarditis, or myocarditis. A complete list of codes for exclusion diagnoses and procedures are given in Supplementary data online, Tables S1 and S2. Table 1 Baseline characteristics for the two cohorts, after propensity score matching

. The risk of major bleeding was similar for dabigatran vs. apixaban (n = 4407, HR 0.99, 95% CI 0.64–1.53).14 Another recent study looking at a 5% random sample of Medicare claims between 2013 and 2014 had even more limited sample sizes (1415, 2358, and 5139 dabigatran, apixaban, and rivaroxaban users, respectively).16 Overall, the effectiveness and safety findings for the novel DOACs in our large patient population are similar to the results reported in retrospective cohort comparisons performed in other real-world populations. Comparisons with apixaban are hampered by the limited sample sizes available. In general, the findings common to these studies are that the rates of stroke and systemic embolism were not significantly different between patients on dabigatran, rivaroxaban, or apixaban, albeit as yet there is no truly adequately powered analysis with apixaban. However, some differences were noted in the risk of major bleeding, with patients treated with dabigatran and apixaban being generally at lower risk than patients on rivaroxaban. A recent meta-analysis of observational studies by Li et al.17 also found that both apixaban and dabigatran were associated with a similar risk of stroke to rivaroxaban, but with a lower risk of major bleeding events. The analysis also found that the risk of stroke was similar for both dabigatran and apixaban, but that dabigatran was associated with a higher risk of major bleeding. However, the authors noted significant heterogeneity in the risk of major bleeding between studies, with only one of the six individual studies included in the analysis concluding a significant difference in major bleeding events between the two treatments, and this study had included both treatment-naïve and treatment-experienced patients across diverging periods of drug availability.18

ent had ‘valvular’ AF in the pre-index period, or any instance of cardiac surgery, pericarditis, or myocarditis. A complete list of codes for exclusion diagnoses and procedures are given in Supplementary data online, Tables S1 and S2. Table 1 Baseline characteristics for the two cohorts, after propensity score matching Dabigatran vs. rivaroxaban Dabigatran vs. apixaban Dabigatran (n = 12 763) Rivaroxaban (n = 12 763) Dabigatran (n = 4802) Apixaban (n = 4802) Age (years) Mean (SD) 70.9 (10.0) 70.9 (10.1) 70.2 (10.2) 70.2 (10.0) Median (range) 72 (19–85) 72 (18–85) 71 (19–85) 71 (19–85) Age category (years) <65 2864 (22.4) 2722 (21.3) 1171 (24.4) 1137 (23.7) 65–74 4710 (36.9) 4837 (37.9) 1846 (38.4) 1888 (39.3) 75–84 4310 (33.8) 4429 (34.7) 1501 (31.3) 1509 (31.4) ≥85 879 (6.9) 775 (6.1) 284 (5.9) 268 (5.6) Gender Male 7902 (61.9) 7839 (61.4) 3028 (63.1) 3039 (63.3) Female 4861 (38.1) 4924 (38.6) 1774 (36.9) 1763 (36.7) Medical history Hypertension 9566 (75.0) 9577 (75.0) 3508 (73.1) 3495 (72.8) Diabetes 3555 (27.9) 3523 (27.6) 1344 (28.0) 1333 (27.8) Prior stroke (all types) 989 (7.7) 979 (7.7) 338 (7.0) 334 (7.0) Transient ischaemic attack 618 (4.8) 595 (4.7) 209 (4.4) 197 (4.1) Congestive heart failure 1802 (14.1) 1797 (14.1) 669 (13.9) 673 (14.0) Renal disease 1853 (14.5) 1807 (14.2) 733 (15.3) 724 (15.1) Risk scores, mean (SD) Charlson comorbidity index 4.27 (2.41) 4.26 (2.40) 4.17 (2.45) 4.18 (2.44) CHADS2 stroke risk score 1.77 (1.22) 1.77 (1.23) 1.70 (1.21) 1.69 (1.20) CHA2DS2-VASc stroke risk score 3.10 (1.66) 3.10 (1.63) 2.98 (1.65) 2.97 (1.61) HAS-BLED bleed risk scorea 2.3 (1.2) 2.3 (1.2) 2.3 (1.2) 2.3 (1.2) Values are expressed as n (%) except where indicated.

4.27 (2.41) 4.26 (2.40) 4.17 (2.45) 4.18 (2.44) CHADS2 stroke risk score 1.77 (1.22) 1.77 (1.23) 1.70 (1.21) 1.69 (1.20) CHA2DS2-VASc stroke risk score 3.10 (1.66) 3.10 (1.63) 2.98 (1.65) 2.97 (1.61) HAS-BLED bleed risk scorea 2.3 (1.2) 2.3 (1.2) 2.3 (1.2) 2.3 (1.2) Values are expressed as n (%) except where indicated. INR, international normalized ratio; SD, standard deviation. a Based on the modified HAS-BLED risk score with a maximum score of 8 because INR data/information were not available for all patients in DoD data. Table 2 Primary outcome event rates and hazard ratios for the two cohorts, after propensity score matching (on-treatment analysis) Dabigatran vs. rivaroxaban Dabigatran vs. apixaban Dabigatran (n = 12 763) Rivaroxaban (n = 12 763) P-value Dabigatran (n = 4802) Apixaban (n = 4802) P-value Stroke (overall)a Patients with event, n (%) 77 (0.60) 100 (0.78) 21 (0.44) 17 (0.35) Event rate per 100 person-years (95% CI) 0.52 (0.41–0.66) 0.69 (0.56–0.84) 0.46 (0.28–0.70) 0.36 (0.21–0.58) Hazard ratio (95% CI) 0.77 (0.57–1.04) 0.084 1.26 (0.66–2.39) 0.489 Major bleeding (overall)b Patients with event n (%) 266 (2.08) 323 (2.53) 77 (1.60) 58 (1.21) Event rate per 100 person-years (95% CI) 1.82 (1.60–2.05) 2.24 (2.00–2.49) 1.69 (1.33–2.11) 1.24 (0.94–1.60) Hazard ratio (95% CI) 0.82 (0.70–0.97) 0.018 1.37 (0.97–1.94) 0.070 CI, confidence interval. a Stroke includes ischaemic and haemorrhagic stroke. b Major bleeding includes haemorrhagic stroke, major intracranial bleeding, and major extracranial bleeding.

Dabigatran vs. rivaroxaban Dabigatran vs. apixaban Dabigatran (n = 12 763) Rivaroxaban (n = 12 763) P-value Dabigatran (n = 4802) Apixaban (n = 4802) P-value Stroke (overall)a Patients with event, n (%) 77 (0.60) 100 (0.78) 21 (0.44) 17 (0.35) Event rate per 100 person-years (95% CI) 0.52 (0.41–0.66) 0.69 (0.56–0.84) 0.46 (0.28–0.70) 0.36 (0.21–0.58) Hazard ratio (95% CI) 0.77 (0.57–1.04) 0.084 1.26 (0.66–2.39) 0.489 Major bleeding (overall)b Patients with event n (%) 266 (2.08) 323 (2.53) 77 (1.60) 58 (1.21) Event rate per 100 person-years (95% CI) 1.82 (1.60–2.05) 2.24 (2.00–2.49) 1.69 (1.33–2.11) 1.24 (0.94–1.60) Hazard ratio (95% CI) 0.82 (0.70–0.97) 0.018 1.37 (0.97–1.94) 0.070 CI, confidence interval. a Stroke includes ischaemic and haemorrhagic stroke. b Major bleeding includes haemorrhagic stroke, major intracranial bleeding, and major extracranial bleeding. Outcomes For the purposes of the analysis we defined primary, secondary, and additional outcomes. The primary efficacy outcome was stroke (including haemorrhagic or ischaemic) and the primary safety outcome was major bleeding (including haemorrhagic stroke, major intracranial bleeding, or major extracranial bleeding). Secondary outcomes included type of major bleeding (intracranial, extracranial, GI, or other), type of stroke (ischaemic or haemorrhagic), transient ischaemic attack, and all-cause mortality. Additional outcomes were MIs, and venous thromboembolic events, presenting as either deep vein thrombosis or pulmonary embolism. All outcome ICD-9 and corresponding ICD-10 codes are listed in Supplementary data online, Table S3.

of stroke (ischaemic or haemorrhagic), transient ischaemic attack, and all-cause mortality. Additional outcomes were MIs, and venous thromboembolic events, presenting as either deep vein thrombosis or pulmonary embolism. All outcome ICD-9 and corresponding ICD-10 codes are listed in Supplementary data online, Table S3. Study size Formal sample size calculations were not undertaken in the study protocol, but the minimum effect size that could be detected with sufficient power was estimated from the expected number of patients in the DoD database. The primary safety outcome of major bleeding was used for power assessments, assuming an annual event rate of 3.1% for the dabigatran patients and mean follow-up duration of 0.82 years for both groups, based on previously published data.7 Based on these hypotheses, 11 682 PSM patients per group would be sufficient to detect a relative difference in the hazard of major bleeding of 22% with 80% power. For 12 763 and 4802 PSM patients per group, the power to detect a 22% difference would be 85% and 30%, respectively.

s, based on previously published data.7 Based on these hypotheses, 11 682 PSM patients per group would be sufficient to detect a relative difference in the hazard of major bleeding of 22% with 80% power. For 12 763 and 4802 PSM patients per group, the power to detect a 22% difference would be 85% and 30%, respectively. Statistical methods All safety and effectiveness outcomes were assessed separately in each of the two cohorts (dabigatran vs. rivaroxaban and dabigatran vs. apixaban). Baseline characteristics for the two cohorts of DOAC-treated patients were summarized using standard descriptive statistics. Logistic regression analysis was performed to derive propensity scores, reflecting the estimated likelihood of each patient being dispensed dabigatran based on baseline demographics and clinical characteristics. The baseline variables included in the models were proposed a priori based on medical knowledge (variables are listed in Supplementary data online, Tables S4 and S5); the index year was not included in the propensity score model. We used nearest neighbour 1:1 matching of dabigatran to rivaroxaban and dabigatran to apixaban patients within a caliper of 0.20 of the standard deviation of the propensity scores.

ledge (variables are listed in Supplementary data online, Tables S4 and S5); the index year was not included in the propensity score model. We used nearest neighbour 1:1 matching of dabigatran to rivaroxaban and dabigatran to apixaban patients within a caliper of 0.20 of the standard deviation of the propensity scores. To examine the effectiveness of propensity score matching in balancing baseline characteristics within the matched cohorts, standardized differences (STD) were calculated for the variables included in the propensity score model. The matched cohorts were considered balanced if the absolute value of the STD was ≤10%. A full listing of all the variables included in the models can be found in Supplementary data online, Tables S4 and S5. Outcome event incidence rates and 95% confidence intervals (CI) were calculated using a person-time approach for each outcome among each treatment group. Incidence rates were based on the total number of patients in each treatment group who had the outcome during follow-up divided by the total person-years at risk in the cohort. Kaplan–Meier cumulative incidence plots were generated to characterize risk over time. Cox proportional-hazards regression was used to evaluate the association between DOAC treatment and time-to-event. Statistical significance was assessed at the two-sided alpha level of 0.05. All statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA).

ted to characterize risk over time. Cox proportional-hazards regression was used to evaluate the association between DOAC treatment and time-to-event. Statistical significance was assessed at the two-sided alpha level of 0.05. All statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA). Results Dabigatran vs. rivaroxaban For the dabigatran vs. rivaroxaban cohort, 116 900 patients had ≥1 prescription for dabigatran or rivaroxaban during the study period, with 94 870 receiving the standard dose (Figure 1). After applying inclusion/exclusion criteria, 12 763 patients on dabigatran and 17 177 patients on rivaroxaban were identified (Figure 1). Following 1:1 propensity score matching, the final dabigatran vs. rivaroxaban cohort included 12 763 patients on dabigatran and 12 763 patients on rivaroxaban. In this cohort, the mean on-treatment follow-up time for the dabigatran group was 422.1 days (range 2–1827 days), and for the rivaroxaban group it was 417.0 days (range 2–1664 days). Reasons for discontinuation before the study end date are shown in Supplementary data online, Table S6. Figure 1 Patient selection and attrition. Dabigatran vs. rivaroxaban (1 July 2011–30 June 2016); dabigatran vs. apixaban (28 December 2012–30 June 2016). aPatients needed ≥2 days of exposure to the index DOAC to ensure they had ≥1 day of index DOAC exposure in the post-index follow-up period. DOAC, direct oral anticoagulant; PSM, propensity score matched.

ion. Dabigatran vs. rivaroxaban (1 July 2011–30 June 2016); dabigatran vs. apixaban (28 December 2012–30 June 2016). aPatients needed ≥2 days of exposure to the index DOAC to ensure they had ≥1 day of index DOAC exposure in the post-index follow-up period. DOAC, direct oral anticoagulant; PSM, propensity score matched. Prior to propensity score matching, the dabigatran and rivaroxaban groups had similar baseline characteristics (Supplementary data online, Table S4). Dabigatran-treated patients were slightly younger (mean age 70.9 ± 10.0 years for dabigatran vs. 71.3 ± 9.7 years for rivaroxaban) and 62% of the dabigatran group were men compared with 61% of the rivaroxaban group. After propensity score matching, balance between the groups was further improved (Table 1). The mean age was 70.9 ± 10.0/10.1 years in the dabigatran and rivaroxaban groups and the proportion of males was 61.9% and 61.4% for the two groups, respectively. Risk factor scores were also well balanced between the groups, with a mean value for CHADS2 score of 1.77 for both groups, CHA2DS2-VASc score of 3.10 for both groups, modified HAS-BLED score of 2.3 for both groups, and Charlson comorbidity index (CCI) of 4.3 for both groups (Table 1).

for the two groups, respectively. Risk factor scores were also well balanced between the groups, with a mean value for CHADS2 score of 1.77 for both groups, CHA2DS2-VASc score of 3.10 for both groups, modified HAS-BLED score of 2.3 for both groups, and Charlson comorbidity index (CCI) of 4.3 for both groups (Table 1). During follow-up, 77 of 12 763 dabigatran patients (0.60%) and 100 of 12 763 rivaroxaban patients (0.78%) had a stroke, giving incidence rates per 100 person-years (95% CI) of 0.52 (0.41–0.66) with dabigatran and 0.69 (0.56–0.84) with rivaroxaban (Table 2). As shown in Table 2 and Figure 2A, no significant difference in the risk of stroke was observed between the PSM patients receiving dabigatran or rivaroxaban [hazard ratio (HR) 0.77, 95% CI 0.57–1.04; P = 0.084]. For major bleeding events (Table 2 and Figure 2B), dabigatran was associated with a lower risk compared with rivaroxaban (HR 0.82, 95% CI 0.70–0.97; P = 0.018). Figure 2 Time from index DOAC to first event (on-treatment analysis): PSM dabigatran vs. rivaroxaban cohort. (A) First stroke. (B) First major bleeding event. Kaplan–Meier curves are shown. Hazard ratios are based on Cox regression analyses. CI, confidence interval; DOAC, direct oral anticoagulant; HR, hazard ratio; PSM, propensity score matched.

C to first event (on-treatment analysis): PSM dabigatran vs. rivaroxaban cohort. (A) First stroke. (B) First major bleeding event. Kaplan–Meier curves are shown. Hazard ratios are based on Cox regression analyses. CI, confidence interval; DOAC, direct oral anticoagulant; HR, hazard ratio; PSM, propensity score matched. Figure 3 shows the secondary and additional outcomes in the dabigatran vs. rivaroxaban cohort. Among reported strokes, patients on dabigatran had a similar event rate of ischaemic stroke (HR 0.92, 95% CI 0.67–1.28; P = 0.631), but a lower risk of haemorrhagic stroke (HR 0.22, 95% CI 0.09–0.59; P = 0.002) than patients on rivaroxaban. For the individual component sites of major bleeding events, dabigatran treatment was associated with a lower risk of major intracranial bleeding vs. rivaroxaban (HR 0.65, 95% CI 0.44–0.98; P = 0.041). However, no statistically significant differences were observed in major extracranial bleeding (HR 0.86, 95% CI 0.72–1.03; P = 0.090), nor in all-cause mortality (HR 1.01, 95% CI 0.83–1.23; P = 0.936) between the dabigatran vs. rivaroxaban groups. Figure 3 Hazard ratios for first outcome event (on-treatment analysis): PSM dabigatran vs. rivaroxaban cohort. Note: numbers of events across subtypes of an outcome may exceed the total number of events for that outcome as a patient can be diagnosed with >1 condition. Cox regression analyses. Hazard ratios were not calculated for major urogenital bleeding. CI, confidence interval; PSM, propensity score matched.

aroxaban cohort. Note: numbers of events across subtypes of an outcome may exceed the total number of events for that outcome as a patient can be diagnosed with >1 condition. Cox regression analyses. Hazard ratios were not calculated for major urogenital bleeding. CI, confidence interval; PSM, propensity score matched. Dabigatran vs. apixaban In the dabigatran vs. apixaban cohort, 89 621 patients had ≥1 prescription for dabigatran or apixaban during the study period (between 28 December 2012 and 30 June 2016), with 70 982 receiving the standard dose. After applying inclusion/exclusion criteria, 4802 patients taking dabigatran and 12 594 taking apixaban were identified (Figure 1). Following 1:1 propensity score matching, the final cohort included 4802 patients on dabigatran and 4802 on apixaban. In this cohort, mean follow-up for the dabigatran-treated patients was 349.5 days (range 2–1280 days) vs. 357.7 days (range 2–1234 days) for the apixaban-treated patients. Prior to propensity score matching, dabigatran-treated patients were younger (mean age 70.2 ± 10.2 years vs. 72.4 ± 10.0 years for apixaban-treated patients) (Supplementary data online, Table S5). Dabigatran-treated patients had lower mean CCI, CHADS2, CHA2DS2-VASc, and modified HAS-BLED scores.

Dabigatran vs. apixaban In the dabigatran vs. apixaban cohort, 89 621 patients had ≥1 prescription for dabigatran or apixaban during the study period (between 28 December 2012 and 30 June 2016), with 70 982 receiving the standard dose. After applying inclusion/exclusion criteria, 4802 patients taking dabigatran and 12 594 taking apixaban were identified (Figure 1). Following 1:1 propensity score matching, the final cohort included 4802 patients on dabigatran and 4802 on apixaban. In this cohort, mean follow-up for the dabigatran-treated patients was 349.5 days (range 2–1280 days) vs. 357.7 days (range 2–1234 days) for the apixaban-treated patients. Prior to propensity score matching, dabigatran-treated patients were younger (mean age 70.2 ± 10.2 years vs. 72.4 ± 10.0 years for apixaban-treated patients) (Supplementary data online, Table S5). Dabigatran-treated patients had lower mean CCI, CHADS2, CHA2DS2-VASc, and modified HAS-BLED scores. After propensity score matching, patients in the dabigatran and apixaban groups had very similar baseline characteristics (Table 1). The mean age for each group was 70.2 years, with 76% being ≥65 years of age. The two groups also had essentially similar mean scores for CHADS2 (1.70 for dabigatran vs. 1.69 for apixaban), CHA2DS2-VASc (2.98 for dabigatran vs. 2.97 for apixaban), modified HAS-BLED (2.3 for both groups), and CCI (4.2 for both groups).

. The mean age for each group was 70.2 years, with 76% being ≥65 years of age. The two groups also had essentially similar mean scores for CHADS2 (1.70 for dabigatran vs. 1.69 for apixaban), CHA2DS2-VASc (2.98 for dabigatran vs. 2.97 for apixaban), modified HAS-BLED (2.3 for both groups), and CCI (4.2 for both groups). The dabigatran vs. apixaban cohort was underpowered, as the minimum sufficient sample size to detect a difference, were a difference to exist, was estimated to be 11 682 dabigatran patients. Therefore, the results that follow should be interpreted with caution. During follow-up, 21 of 4802 dabigatran patients (0.44%) and 17 of 4802 apixaban patients (0.35%) had a stroke, giving incidence rates per 100 person-years (95% CI) of 0.46 (0.28–0.70) with dabigatran and 0.36 (0.21–0.58) with apixaban (Table 2). In the dabigatran vs. apixaban PSM cohort, the stroke HR was 1.26 (95% CI 0.66–2.39; P = 0.489; Table 2 and Figure 4). For major bleeding, the HR for dabigatran and apixaban users was 1.37 (95% CI 0.97–1.94; P = 0.070; Table 2 and Figure 5). Figure 4 Time from index DOAC to first event on treatment: PSM dabigatran vs. apixaban cohort. (A) First stroke. (B) First major bleeding event. Kaplan–Meier curves are shown. HRs are based on Cox regression analyses. CI, confidence interval; DOAC, direct oral anticoagulant; HR, hazard ratio; PSM, propensity score matched.

re 4 Time from index DOAC to first event on treatment: PSM dabigatran vs. apixaban cohort. (A) First stroke. (B) First major bleeding event. Kaplan–Meier curves are shown. HRs are based on Cox regression analyses. CI, confidence interval; DOAC, direct oral anticoagulant; HR, hazard ratio; PSM, propensity score matched. Figure 5 Hazard ratios for first outcome event (on-treatment analysis): PSM dabigatran vs. apixaban cohort. Note: numbers of events across subtypes of an outcome may exceed the total number of events for that outcome as one patient can be diagnosed with >1 condition. Cox regression analyses. Hazard ratios were not calculated for haemorrhagic stroke or major urogenital bleeding. CI, confidence interval; PSM, propensity score matched.

mbers of events across subtypes of an outcome may exceed the total number of events for that outcome as one patient can be diagnosed with >1 condition. Cox regression analyses. Hazard ratios were not calculated for haemorrhagic stroke or major urogenital bleeding. CI, confidence interval; PSM, propensity score matched. In this cohort, the two groups appeared to have similar risk of ischaemic stroke, (HR 1.05, 95% CI 0.54–2.06; P = 0.878) (Figure 5). For haemorrhagic stroke, the number of events was low (n = 3) and the HR was not estimated. With regard to the individual component sites of major bleeding events, overall no difference was found in the dabigatran vs. apixaban cohort for intracranial bleeding (HR 1.11, 95% CI 0.47–2.63; P = 0.812). There was also no difference in the rates of major extracranial bleeding (HR 1.43, 95% CI 0.98–2.08; P = 0.062). However, there were fewer major GI bleeding events in the apixaban group (HR 1.50, 95% CI 1.02–2.23; P = 0.042). For MI, the HR was 2.72 (95% CI 1.19–6.18; P = 0.017). Looking at all-cause mortality, there were no differences in risk identified between the two groups (HR 1.02, 95% CI 0.72–1.47; P = 0.895).

P = 0.062). However, there were fewer major GI bleeding events in the apixaban group (HR 1.50, 95% CI 1.02–2.23; P = 0.042). For MI, the HR was 2.72 (95% CI 1.19–6.18; P = 0.017). Looking at all-cause mortality, there were no differences in risk identified between the two groups (HR 1.02, 95% CI 0.72–1.47; P = 0.895). Discussion This observational analysis of the DoD Military Health System clinical database explored the comparative safety and effectiveness of dabigatran vs. rivaroxaban or apixaban in patients with NVAF treated in routine clinical practice. In the dabigatran vs. rivaroxaban cohort, after propensity score matching, there was no significant difference in the risk of stroke for either group. However, we observed a significantly lower risk of major bleeding with dabigatran compared with rivaroxaban; specifically, dabigatran was associated with a significantly lower incidence of intracranial haemorrhage. No differences between groups were identified for all-cause mortality. In the dabigatran vs. apixaban cohort, the number of patients identified was much lower and, based on our power calculations, the results of the statistical tests for this cohort should be interpreted with caution, and no definitive conclusions can be drawn.

Discussion This observational analysis of the DoD Military Health System clinical database explored the comparative safety and effectiveness of dabigatran vs. rivaroxaban or apixaban in patients with NVAF treated in routine clinical practice. In the dabigatran vs. rivaroxaban cohort, after propensity score matching, there was no significant difference in the risk of stroke for either group. However, we observed a significantly lower risk of major bleeding with dabigatran compared with rivaroxaban; specifically, dabigatran was associated with a significantly lower incidence of intracranial haemorrhage. No differences between groups were identified for all-cause mortality. In the dabigatran vs. apixaban cohort, the number of patients identified was much lower and, based on our power calculations, the results of the statistical tests for this cohort should be interpreted with caution, and no definitive conclusions can be drawn. The results from this study are consistent with those of several other recent studies that compared dabigatran with rivaroxaban cohorts.10,14 As mentioned above, a retrospective analysis of PSM patients with NVAF who were ≥65 years of age and newly initiated on either dabigatran or rivaroxaban, found no significant difference in thromboembolic stroke (HR 0.81, 95% CI 0.65–1.01; P = 0.07), and a significant increase in intracranial haemorrhage (HR 1.65, 95% CI 1.20–2.26; P = 0.002) and major extracranial bleeding (HR 1.48, 95% CI 1.32–1.67; P < 0.001) with rivaroxaban compared with dabigatran.10 It is important to keep power in mind when interpreting results, as a small sample size not only results in the inability to distinguish between clinically important and unimportant outcomes (wide CIs) and reduces the likelihood to identify moderate differences, but also results in an increased risk of chance results. The probability that a significant result is true depends on the statistical power of the study.15 Considering that both low sample size and multiple testing increase the potential for false-positive results, and taking into account the wide CIs, the meaningfulness and robustness of the significant differences observed do not allow for definitive conclusions in the dabigatran vs. apixaban cohort.

wer of the study.15 Considering that both low sample size and multiple testing increase the potential for false-positive results, and taking into account the wide CIs, the meaningfulness and robustness of the significant differences observed do not allow for definitive conclusions in the dabigatran vs. apixaban cohort. The risk for a major bleeding event (requiring hospitalization) among patients with NVAF who were newly initiated on either warfarin, apixaban, dabigatran, or rivaroxaban in clinical practice settings was assessed in an analysis of the Truven MarketScan® Commercial and Medicare supplemental claims database.14 Adult patients with NVAF, newly initiating oral anticoagulation after ≥1-year baseline period were identified, and propensity score matching was used to balance patient variables including age, sex, region, baseline comorbidities, and concomitant medications. Of 45 361 newly anticoagulated patients, 34.1% (n = 15 461) initiated warfarin, 16.4% (n = 7438) initiated apixaban, 39.2% (n = 17 801) initiated rivaroxaban, and 10.3% (n = 4661) initiated dabigatran.14

s used to balance patient variables including age, sex, region, baseline comorbidities, and concomitant medications. Of 45 361 newly anticoagulated patients, 34.1% (n = 15 461) initiated warfarin, 16.4% (n = 7438) initiated apixaban, 39.2% (n = 17 801) initiated rivaroxaban, and 10.3% (n = 4661) initiated dabigatran.14 Among previously published studies, this analysis has the largest number of apixaban patients. However, as with our analysis, limited sample size—and therefore power—also warrants caution in the interpretation of the comparisons to dabigatran. When compared with PSM patients initiating warfarin, significantly lower risks for major bleeding events were reported for those initiating apixaban (n = 6964 per group, HR 0.53, 95% CI 0.39–0.71) or dabigatran (n = 4515, HR 0.69, 95% CI 0.50–0.96). Patients initiating rivaroxaban demonstrated no significant difference in their risk of major bleeding compared with PSM warfarin patients (n = 12 625, HR 0.98, 95% CI 0.83–1.17). Between standard-dose DOACs, PSM patients on rivaroxaban had a significantly higher risk of major bleeding vs. patients on apixaban (n = 7399, HR 1.77, 95% CI 1.29–2.45) or dabigatran (n = 4657, HR 1.65, 95% CI 1.15–2.36). The risk of major bleeding was similar for dabigatran vs. apixaban (n = 4407, HR 0.99, 95% CI 0.64–1.53).14 Another recent study looking at a 5% random sample of Medicare claims between 2013 and 2014 had even more limited sample sizes (1415, 2358, and 5139 dabigatran, apixaban, and rivaroxaban users, respectively).16

jor bleeding between studies, with only one of the six individual studies included in the analysis concluding a significant difference in major bleeding events between the two treatments, and this study had included both treatment-naïve and treatment-experienced patients across diverging periods of drug availability.18 As with all real-world analyses, our analysis has particular strengths in terms of translation to clinical practice. Due to the wide representation of patients in the DoD database, in terms of both demography and geography, the results of this study are expected to have broad external validity to the US population.11,12 Careful calculation of propensity scores for patients on baseline characteristics, comorbidities, and concomitant medications identified well-matched cohorts. Our study also provided a long on-treatment follow-up duration (mean 422 days for dabigatran vs. rivaroxaban and 350 days for dabigatran vs. apixaban), which was probably facilitated at least in part to the unique stability of enrolment of active duty personnel, retirees, and their families.

ified well-matched cohorts. Our study also provided a long on-treatment follow-up duration (mean 422 days for dabigatran vs. rivaroxaban and 350 days for dabigatran vs. apixaban), which was probably facilitated at least in part to the unique stability of enrolment of active duty personnel, retirees, and their families. Similar to other retrospective database analyses, this study is subject to several inherent limitations including the possibility of coding errors of omission and commission, a lack of central adjudication for events, with use of ICD-9 and ICD-10 codes and claims to identify baseline medical conditions and medications, and incomplete claims.19,20 Despite the standardization of the ICD codes system, there is the potential for incomplete or inaccurate event accounting related to the use of ICD codes to identify events. ICD coding for stroke has been reported to be equally good when using ICD-9 codes (90% positive predictive value, 95% CI 86–93), and ICD-10 codes (92% positive predictive value, 95% CI 88–95).21 Similarly, an analysis of algorithms based on ICD-9 codes, which were used to define treatment outcomes (including intracranial haemorrhage, major extracranial bleeding, and major GI bleeding), reported positive predictive values ranging from 86% to 97%.10

d ICD-10 codes (92% positive predictive value, 95% CI 88–95).21 Similarly, an analysis of algorithms based on ICD-9 codes, which were used to define treatment outcomes (including intracranial haemorrhage, major extracranial bleeding, and major GI bleeding), reported positive predictive values ranging from 86% to 97%.10 Furthermore, as with all non-randomized studies, imbalance in unmeasured prognostic factors could bias the results (residual confounding). Our study design minimized this risk by restricting inclusion to treatment-naïve patients initiating treatment in the period of common treatment availability following the approval date of both compared medications, and the use of propensity score methods, a very powerful method for controlling for confounding if proper variables and data are utilized in the analysis.22 Conclusions Overall, this analysis showed that patients in the DoD Military Health System Data Repository who were newly initiated on dabigatran had a statistically significantly lower risk of major bleeding than patients newly starting rivaroxaban, while the risk of stroke was not significantly different. For the cohort comparing dabigatran vs. apixaban, the reduced sample size limits the ability to draw definitive conclusions. In the absence of head-to-head clinical trials comparing available DOACs, this practice-based analysis of direct comparisons of patient outcomes data may help inform clinical decision-making. Supplementary Material Supplementary Information Click here for additional data file.

Conclusions Overall, this analysis showed that patients in the DoD Military Health System Data Repository who were newly initiated on dabigatran had a statistically significantly lower risk of major bleeding than patients newly starting rivaroxaban, while the risk of stroke was not significantly different. For the cohort comparing dabigatran vs. apixaban, the reduced sample size limits the ability to draw definitive conclusions. In the absence of head-to-head clinical trials comparing available DOACs, this practice-based analysis of direct comparisons of patient outcomes data may help inform clinical decision-making. Supplementary Material Supplementary Information Click here for additional data file. Acknowledgements The authors meet criteria for authorship as recommended by the International Committee of Medical Journal Editors (ICMJE) and were fully responsible for all content and editorial decisions, and were involved at all stages of manuscript development. The authors received no direct compensation related to the development of the manuscript. Writing assistance was provided by José L. Walewski, PhD, and Geraldine Thompson of Envision Scientific Solutions, which was contracted and compensated by BIPI for these services. Funding This work was supported by Boehringer Ingelheim Pharmaceuticals, Inc. (BIPI).

occurred with an event rate of 1.84/100 person-years compared with 2.21/100 person-years in the rivaroxaban group [HR 0.88; 95% confidence interval (CI) 0.76–1.02]. A major bleeding event occurred at a rate of 1.40/100 person-years in the dabigatran group, and 1.93 in the rivaroxaban group (HR 0.75; 95% CI 0.64–0.88). Dabigatran–apixaban-matched cohort The median follow-up time was 18.2 months for dabigatran users and 12.2 months for apixaban users. Among dabigatran users, stroke/SE occurred at a rate of 1.83/100 person-years, while the event rate was 2.62/100 person-years for apixaban users (HR 0.88; 95% CI 0.75–1.02). Major bleeding occurred at an event rate of 1.38/100 person-years in the dabigatran group vs. 1.54/100 person-years in the apixaban group (HR 1.03 95% CI 0.85–1.24). The risk of GI bleeding was significantly higher for dabigatran with event rates of 3.22/100 person-years vs. 2.17/100 person-years in the apixaban group (HR 1.48; 95% CI 1.28–1.70). Apixaban–rivaroxaban-matched cohort The median follow-up time was 18.1 months in the rivaroxaban group, and 12.5 months in the apixaban group. The event rate of stroke/SE was 2.65/100 person-years for the apixaban group vs. 2.31/100 person-years for the rivaroxaban group (HR 1.00; 95% CI 0.89–1.14). The event rates of major bleeding were 1.76/100 person-years vs. 2.10/100 person-years in the apixaban- and rivaroxaban groups, respectively (HR 0.79; 95% CI 0.68–0.91).

Acknowledgements The authors meet criteria for authorship as recommended by the International Committee of Medical Journal Editors (ICMJE) and were fully responsible for all content and editorial decisions, and were involved at all stages of manuscript development. The authors received no direct compensation related to the development of the manuscript. Writing assistance was provided by José L. Walewski, PhD, and Geraldine Thompson of Envision Scientific Solutions, which was contracted and compensated by BIPI for these services. Funding This work was supported by Boehringer Ingelheim Pharmaceuticals, Inc. (BIPI). Disclaimer The identification of specific products or scientific instrumentation is considered an integral part of the scientific endeavour and does not constitute endorsement or implied endorsement on the part of the author, DoD, or any component agent. The views expressed in this manuscript reflect the results of research conducted by the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, or the United States Government. CAPT Eric Schwartzman and CAPT Todd Villines are military service members. This work was prepared as part of their official duties. Title 17 U.S.C. 105 provides that ‘Copyright protection under this title is not available for any work of the United States Government’. Title 17 U.S.C. 101 defines a United States Government work as a work prepared by a military service member or employee of the United States Government as part of that person’s official duties. Research data derived from an approved Naval Medical Center, Portsmouth, Virginia IRB, protocol; number NMCP.2016.0094.

ment’. Title 17 U.S.C. 101 defines a United States Government work as a work prepared by a military service member or employee of the United States Government as part of that person’s official duties. Research data derived from an approved Naval Medical Center, Portsmouth, Virginia IRB, protocol; number NMCP.2016.0094. Conflict of interest: T.C.V. has served as a consultant and on the speakers’ bureau for Boehringer Ingelheim; he received no compensation related to this study. A.E. and T.R. are employees of Health ResearchTx LLC, which was paid by BIPI to provide consulting and analytical services in support of this study. D.T. and K.O. are employees of Syneos Health, which was paid by BIPI to provide consulting and analytical services in support of this study. A.A., M.P., and W.T. are employees of Boehringer Ingelheim. E.S. reports no financial interests. Prior publication Analyses were presented, in part, at the International Stroke Conference (ISC), 24–26 January 2018; Los Angeles, CA. Footnotes The link for the video is: http://usscicomms.com/cardiovascular/villines/dod-study/

Introduction In acute coronary syndrome with or without ST-segment elevation, European guidelines recommend dual antiplatelet treatment for at least the first year.1,2 Notably, many stable patients with a history of myocardial infarction (MI) remain at high risk after this period.3,4 In PEGASUS-TIMI 54, ticagrelor, at doses of either 90 mg b.i.d. or 60 mg b.i.d., significantly reduced the risk of the composite of major adverse cardiovascular events [MACE; cardiovascular (CV) death, MI, or stroke] by 15–16% in stable patients at high risk with a prior MI 1–3 years earlier.5 The benefit of ticagrelor appeared more marked in patients continuing on or restarting after only a brief interruption of adenosine diphosphate (ADP) receptor inhibition and in those closer to their qualifying MI.6 Accordingly, the CHMP-EMA approved European (EU) label recommends that, after the initial 1-year treatment with ticagrelor 90 mg b.i.d. (or other ADP receptor inhibitor) in high-risk MI patients, treatment with ticagrelor 60 mg b.i.d. may be started without interruption as continuation therapy.7 Treatment with ticagrelor 60 mg b.i.d. can also be initiated up to 2 years from the MI, or within 1 year after stopping previous ADP receptor inhibitor treatment. While the PEGASUS-TIMI 54 trial had wider inclusion criteria, the present analysis aimed to describe the effects of extended treatment with ticagrelor 60 mg b.i.d. in a clinically relevant subset of patients, treated according to the approved label. We, therefore, report the efficacy and safety in the PEGASUS-TIMI 54 subpopulation recommended for treatment in the EU label.

teria, the present analysis aimed to describe the effects of extended treatment with ticagrelor 60 mg b.i.d. in a clinically relevant subset of patients, treated according to the approved label. We, therefore, report the efficacy and safety in the PEGASUS-TIMI 54 subpopulation recommended for treatment in the EU label. Methods Study population PEGASUS-TIMI 54 randomized patients with prior MI to ticagrelor 60 mg b.i.d., ticagrelor 90 mg b.i.d., or placebo, all on a background of low-dose aspirin. The protocol was approved by the relevant ethics committee at each participating site. Written informed consent was obtained from all the patients. The design8 and primary results of the trial have been published.5 In brief, patients aged at least 50 years were included with a spontaneous MI occurring 1–3 years prior to enrolment and at least one of the following additional high-risk features: age of 65 years or older, diabetes mellitus requiring medication, a second prior spontaneous MI, multivessel coronary artery disease, or chronic renal dysfunction, defined as a creatinine clearance <60 mL/min as estimated by the Cockroft–Gault equation. Patients were ineligible if there was planned use of a P2Y12 receptor antagonist or anticoagulant therapy during the study period; if they had a bleeding disorder, a history of intracranial bleeding, a central nervous system tumour, or an intracranial vascular abnormality; or if they had had gastrointestinal bleeding within the previous 6 months or major surgery within the previous month.

ntagonist or anticoagulant therapy during the study period; if they had a bleeding disorder, a history of intracranial bleeding, a central nervous system tumour, or an intracranial vascular abnormality; or if they had had gastrointestinal bleeding within the previous 6 months or major surgery within the previous month. The present analysis focuses on data from 10 779 patients that were randomized ≤2 years from qualifying MI or ≤1year from prior stopping ADP receptor inhibitor treatment, 5388 in the ticagrelor 60 mg and 5391 in the placebo group (EU label group). Patients randomized to ticagrelor 60 mg or placebo who did not qualify per the EU label are termed the non-EU label group. Data on patients randomized to ticagrelor 90 mg who would have qualified for the EU label (n = 5374) are also presented in the Supplementary material online for the sake of completeness.

el group). Patients randomized to ticagrelor 60 mg or placebo who did not qualify per the EU label are termed the non-EU label group. Data on patients randomized to ticagrelor 90 mg who would have qualified for the EU label (n = 5374) are also presented in the Supplementary material online for the sake of completeness. Endpoints The primary efficacy endpoint for PEGASUS-TIMI 54 was the composite of CV death, MI, or stroke (MACE). Additional efficacy endpoints included the individual components of the composite as well as coronary heart disease-related death and all-cause mortality. The primary safety endpoint was Thrombolysis in myocardial infarction (TIMI) major bleeding. Other safety endpoints included TIMI minor bleeding, intracranial haemorrhage, and fatal bleeding. All potential events were adjudicated by the TIMI clinical events committee, which was blinded to treatment allocation. Net clinical benefit was calculated as the number of events prevented (CV death, MI, stroke, or the composite of these) vs. events caused (TIMI major bleeding, intracranial haemorrhage, or fatal bleeding) per 1000 patients treated for 3 years with ticagrelor. Within these events we also examined the irreversible hard outcomes which included all of the aforementioned outcomes except TIMI major bleeding.9

troke, or the composite of these) vs. events caused (TIMI major bleeding, intracranial haemorrhage, or fatal bleeding) per 1000 patients treated for 3 years with ticagrelor. Within these events we also examined the irreversible hard outcomes which included all of the aforementioned outcomes except TIMI major bleeding.9 Statistical considerations Cumulative event rates at 3 years were calculated by the complement of the Kaplan–Meier (KM) survival estimates. Hazard ratios (HRs) and 95% confidence intervals (CIs) were generated with the use of a Cox proportional-hazards model, and all reported P-values are two-sided. Interactions between the agreed EU label and the treatment group were also examined by Cox proportional hazards model. The assumption of proportional hazards was tested by including time dependent covariates in the model and examined by scaled Schoenfeld residual plots. Number needed to treat (NNT) was calculated by the reciprocal of the absolute risk difference based on 3 year KM estimates. The number of events prevented and caused per 1000 patients were based on the difference between 3 years incidence rates/person years in the treatment and placebo arm, with negative difference being ‘prevented’ events and positive difference being events ‘caused’ by treatment arm. This difference was multiplied by 1000 to aid the clinical interpretation. Efficacy analyses were performed on an intention-to-treat basis. Safety analyses included all the patients who received at least one dose of study drug and included all the events occurring after receipt of the first dose and within 7 days of the last dose of study drug. The bleeding analysis is on-treatment. Results for the 90 mg b.i.d. dose are presented in the Supplementary material online.

yses included all the patients who received at least one dose of study drug and included all the events occurring after receipt of the first dose and within 7 days of the last dose of study drug. The bleeding analysis is on-treatment. Results for the 90 mg b.i.d. dose are presented in the Supplementary material online. All analyses were performed according to the intention-to-treat principle, utilizing SAS version 9.4. The statistical significance was set at an α-level of 0.05 significance. Results A total of 14 112 patients were randomized to ticagrelor 60 mg bid (the EU dose approved for long-term therapy) or placebo. Of this group, 10 799 patients were within 2 years from qualifying MI or within 1 year from prior ADP receptor inhibitor treatment and were randomized to ticagrelor 60 mg bid or placebo. As expected, there were no differences in baseline characteristics by randomized treatment arm (Table 1). An additional 3333 patients were in the ticagrelor 60 mg or placebo arms but fell outside the EU label parameters. The median time from MI was 1.5 years vs. 2.5 years and median time from P2Y12-treatment discontinuation was 34 days vs. 588 days for the EU label vs. non-EU label patients, respectively. Compared with non-EU label patients, EU label patients were more likely to have had a history of multivessel coronary artery disease (61.4% vs. 53.7%, P < 0.001) and a history of percutaneous coronary intervention (PCI) (84.9% vs. 77.1%, P < 0.001) (Supplementary material online, Table S1).

abel patients, respectively. Compared with non-EU label patients, EU label patients were more likely to have had a history of multivessel coronary artery disease (61.4% vs. 53.7%, P < 0.001) and a history of percutaneous coronary intervention (PCI) (84.9% vs. 77.1%, P < 0.001) (Supplementary material online, Table S1). Table 1 Baseline characteristics ticagrelor 60 mg and placebo, European label and non-European label population

abel patients, respectively. Compared with non-EU label patients, EU label patients were more likely to have had a history of multivessel coronary artery disease (61.4% vs. 53.7%, P < 0.001) and a history of percutaneous coronary intervention (PCI) (84.9% vs. 77.1%, P < 0.001) (Supplementary material online, Table S1). Table 1 Baseline characteristics ticagrelor 60 mg and placebo, European label and non-European label population Characteristics EU label population Non-EU label population Ticagrelor 60 mg bid (N = 5388) Placebo (N = 5391) Ticagrelor 60 mg bid (N = 1657) Placebo (N = 1676) Age (years), mean (SD) 65.1 (8.5) 65.3 (8.3) 65.5 (8.1) 65.6 (8.2) Female 1267 (23.52%) 1314 (24.37%) 394 (23.78%) 403 (24.05%) White 4592 (85.23%) 4606 (85.44%) 1485 (89.62%) 1518 (90.57%) Weight, mean (SD) 81.9 (17.1) 81.6 (16.8) 82.5 (16.6) 82.5 (16.0) History of hypertension 4183 (77.65%) 4175 (77.44%) 1278 (77.13%) 1309 (78.1%) History of hypercholesterolaemia 4122 (76.52%) 4179 (77.52%) 1258 (75.97%) 1272 (75.94%) Current smoker 939 (17.43%) 865 (16.06%) 267 (16.11%) 278 (16.59%) History of diabetes 1774 (32.93%) 1710 (31.72%) 534 (32.25%) 547 (32.64%) Multivessel coronary artery disease 3313 (61.5%) 3300 (61.21%) 877 (52.99%) 913 (54.47%) History of PCI 4584 (85.09%) 4563 (84.66%) 1295 (78.15%) 1274 (76.01%) History of second prior MI 884 (16.41%) 900 (16.69%) 284 (17.15%) 288 (17.18%) History of PAD 301 (5.59%) 317 (5.88%) 67 (4.04%) 87 (5.19%) eGRR <60 mL/min/1.73 m2 1178 (22.16%) 1239 (23.25%) 369 (22.51%) 410 (24.77%) Qualifying event Years since MI, median (IQR) 1.5 (1.2–1.9) 1.5 (1.2–1.9) 2.5 (2.3–2.8) 2.5 (2.3–2.8) Type of MI NSTEMI 2209 (41.04%) 2177 (40.43%) 633 (38.32%) 666 (39.81%) STEMI 2872 (53.35%) 2928 (54.38%) 885 (53.57%) 881 (52.66%) Unknown 302 (5.61%) 279 (5.18%) 134 (8.11%) 126 (7.53%) Medications at baseline Aspirin 5381 (99.87%) 5382 (99.83%) 1655 (99.88%) 1675 (99.94%) Statin 4999 (92.78%) 5049 (93.66%) 1496 (90.28%) 1534 (91.53%) Beta blocker 4462 (82.81%) 4518 (83.81%) 1334 (80.51%) 1360 (81.15%) ACE-I or ARB 4310 (79.99%) 4341 (80.52%) 1321 (79.72%) 1356 (80.91%) There were no statistically significant differences in baseline characteristics by treatment arm within the EU and non-EU label subgroups.

) 5049 (93.66%) 1496 (90.28%) 1534 (91.53%) Beta blocker 4462 (82.81%) 4518 (83.81%) 1334 (80.51%) 1360 (81.15%) ACE-I or ARB 4310 (79.99%) 4341 (80.52%) 1321 (79.72%) 1356 (80.91%) There were no statistically significant differences in baseline characteristics by treatment arm within the EU and non-EU label subgroups. In addition, there were 5374 patients randomized to ticagrelor 90 mg within 2 years from qualifying MI or within 1 year from prior ADP receptor inhibitor treatment (see Supplementary material online, Table S2). Efficacy In the EU label population, the composite of CV death, MI, or stroke occurred in 373 patients (KM rate 7.9%) in the ticagrelor 60 mg group and in 463 patients in the placebo group (KM rate 9.6%; Figure 1); HR 0.80 (95% CI 0.70–0.91; P = 0.001), when compared with HR 1.00 (95% CI 0.77–1.30; P = 0.98) among the non-EU label population (P-value for interaction 0.12). The absolute risk reduction over 3 years was 1.7%, leading to a NNT of 58. In the EU label population, corresponding HRs for the components of the primary composite endpoint were 0.71 (95% CI 0.56–0.90; P = 0.0041) for CV death, 0.83 (95% CI 0.70–0.99; P = 0.041) for MI, and 0.74 (95% CI 0.55–1.01; P = 0.058) for stroke (Table 2). The HR for coronary heart disease death was 0.72 (95% CI 0.53–0.97; P = 0.03) and for all-cause death 0.80 (95% CI 0.67–0.96; P = 0.018) (see Figure 2).

endpoint were 0.71 (95% CI 0.56–0.90; P = 0.0041) for CV death, 0.83 (95% CI 0.70–0.99; P = 0.041) for MI, and 0.74 (95% CI 0.55–1.01; P = 0.058) for stroke (Table 2). The HR for coronary heart disease death was 0.72 (95% CI 0.53–0.97; P = 0.03) and for all-cause death 0.80 (95% CI 0.67–0.96; P = 0.018) (see Figure 2). Figure 1 Primary endpoint for ticagrelor 60 mg vs. placebo, European label and non-European label patients. T60 EU: ticagrelor 60 mg according to European label. Placebo EU: placebo treatment according to European label. T60 N-EU: ticagrelor 60 mg to non-European label patients. Placebo N-EU: placebo treatment to non-European label patients. Figure 2 All-cause death for ticagrelor 60 mg vs. placebo, European label and non-European label patients. T60 EU: ticagrelor 60 mg according to European label. Placebo EU: placebo treatment according to European label. T60 N-EU: ticagrelor 60 mg to non-European label patients. Placebo N-EU: placebo treatment to non-European label patients. Table 2 Efficacy of ticagrelor 60 mg vs. placebo in the European label population

Figure 2 All-cause death for ticagrelor 60 mg vs. placebo, European label and non-European label patients. T60 EU: ticagrelor 60 mg according to European label. Placebo EU: placebo treatment according to European label. T60 N-EU: ticagrelor 60 mg to non-European label patients. Placebo N-EU: placebo treatment to non-European label patients. Table 2 Efficacy of ticagrelor 60 mg vs. placebo in the European label population Outcomes Ticagrelor 60 mg bid (N = 5388) Placebo (N = 5391) HR (95% CI) P-value Number of events KM rate (%) Number of events KM rate (%) Composite of CV death/MI/Stroke 373 7.85 463 9.56 0.80 (0.70–0.91) 0.0011 CV death 119 2.58 167 3.58 0.71 (0.56–0.90) 0.0041 Coronary heart disease death 75 1.59 104 2.15 0.72 (0.53–0.97) 0.0282 MI 230 4.85 274 5.59 0.83 (0.70–0.99) 0.0406 Stroke 71 1.52 95 2.04 0.74 (0.55–1.01) 0.0583 All-cause mortality 206 4.44 256 5.39 0.80 (0.67–0.96) 0.0183 The efficacy results were virtually identical when comparing patients who would qualify for the EU label but were randomized to ticagrelor 90 mg to placebo, with a HR for CV death, MI, or stroke of 0.80 (95% CI 0.70–0.92; P = 0.0015) (Supplementary material online, Table S3). We also did a further subgroup analysis examining the efficacy of ticagrelor in patients with just one or both of the EMA requirements (MI within 2 years and within 1 year from stopping previous P2Y12 receptor inhibitor). Among patients who qualified on both points, the benefit of ticagrelor was most apparent (Supplementary material online, Table S4). In patients who qualified on just one of the EMA requirements, the difference between ticagrelor and placebo did not reach statistical significance although the trend towards the benefit is still present.

ho qualified on both points, the benefit of ticagrelor was most apparent (Supplementary material online, Table S4). In patients who qualified on just one of the EMA requirements, the difference between ticagrelor and placebo did not reach statistical significance although the trend towards the benefit is still present. Safety Thrombolysis in myocardial infarction major bleeding occurred in 94 patients (KM rate 2.5%) in the ticagrelor 60 mg group and in 43 patients (KM rate 1.1%) in the placebo group; number needed to harm 76, HR 2.36 (1.65–3.39, P < 0.001; Table 3), when compared with HR 2.13 (95% CI 1.03–4.43; P = 0.04) among non-EU label patients (P-value for interaction 0.81). The corresponding HRs for fatal or intracranial bleeding were 1.17 (0.68–2.01; P = 0.58) in the EU label subgroup, when compared with HR 1.36 (95% CI 0.41–4.46; P = 0.61) among the non-EU label patients. The HR for major bleeding for patients who would qualify for the EU label but were randomized to ticagrelor 90 mg vs. placebo was 2.59 (95% CI 1.81–3.70) (Supplementary material online, Table S5). Table 3 Safety of ticagrelor 60 mg vs. placebo in the European label population

Safety Thrombolysis in myocardial infarction major bleeding occurred in 94 patients (KM rate 2.5%) in the ticagrelor 60 mg group and in 43 patients (KM rate 1.1%) in the placebo group; number needed to harm 76, HR 2.36 (1.65–3.39, P < 0.001; Table 3), when compared with HR 2.13 (95% CI 1.03–4.43; P = 0.04) among non-EU label patients (P-value for interaction 0.81). The corresponding HRs for fatal or intracranial bleeding were 1.17 (0.68–2.01; P = 0.58) in the EU label subgroup, when compared with HR 1.36 (95% CI 0.41–4.46; P = 0.61) among the non-EU label patients. The HR for major bleeding for patients who would qualify for the EU label but were randomized to ticagrelor 90 mg vs. placebo was 2.59 (95% CI 1.81–3.70) (Supplementary material online, Table S5). Table 3 Safety of ticagrelor 60 mg vs. placebo in the European label population Outcomes Ticagrelor 60 mg bid (N = 5322) Placebo (N = 5331) HR (95% CI) P-value Number of events KM rate (%) Number of events KM rate (%) TIMI major bleeding 94 2.46 43 1.14 2.36 (1.65–3.39) <0.0001 TIMI minor bleeding 49 1.39 15 0.39 3.50 (1.96–6.25) <0.0001 Fatal bleeding 9 0.29 11 0.33 0.88 (0.37–2.13) 0.7825 Intracranial haemorrhage 23 0.68 18 0.49 1.38 (0.74–2.55) 0.3085 Fatal bleeding or intracranial haemorrhage 27 0.79 25 0.67 1.17 (0.68–2.01) 0.5777 Net clinical benefit The number of events prevented and caused per 1000 patients is shown in Figure 3. Treating 1000 patients with ticagrelor 60 mg b.i.d. for 3 years would be expected to prevent 24 major adverse CV events, including 10 CV deaths, 9 MIs, and 5 strokes, while causing 10 major bleeds but no cases of intracranial or fatal bleeding. Thus, the NNT over 3 years to prevent one irreversible event (CV death, MI, stroke, intracranial haemorrhage, or fatal bleeding) was 42.

rs would be expected to prevent 24 major adverse CV events, including 10 CV deaths, 9 MIs, and 5 strokes, while causing 10 major bleeds but no cases of intracranial or fatal bleeding. Thus, the NNT over 3 years to prevent one irreversible event (CV death, MI, stroke, intracranial haemorrhage, or fatal bleeding) was 42. Figure 3 Clinical events prevented and caused per 1000 patients initiated on ticagrelor 60 mg b.i.d. and followed for 3 years.

rs would be expected to prevent 24 major adverse CV events, including 10 CV deaths, 9 MIs, and 5 strokes, while causing 10 major bleeds but no cases of intracranial or fatal bleeding. Thus, the NNT over 3 years to prevent one irreversible event (CV death, MI, stroke, intracranial haemorrhage, or fatal bleeding) was 42. Figure 3 Clinical events prevented and caused per 1000 patients initiated on ticagrelor 60 mg b.i.d. and followed for 3 years. Discussion The present analysis defines the clinical efficacy of dual antiplatelet treatment with 60 mg b.i.d. ticagrelor post-MI when initiated according to the CHMP-EMA approved EU label, i.e. that after the initial year of treatment with 90 mg ticagrelor b.i.d., the patient is shifted to 60 mg b.i.d. without, or with only a briefer interruption.7 In such a population, ticagrelor reduced the risk of the primary endpoint of CV death, MI, or stroke by 20%, coronary heart death by 28%, CV mortality by 29%, and all-cause mortality by 20%. Various mechanisms may explain the apparently enhanced benefit of ticagrelor in this population: for example, there is an increased MACE rate in the first 3 months after cessation of P2Y12 receptor inhibitor6,10 so ticagrelor-treated patients in the EU label group would have received some protection during this higher-risk period. Furthermore, there was a higher proportion of patients with multivessel coronary artery disease in the EU label group and we have shown these patients to have a greater absolute risk reduction with ticagrelor, including for coronary heart disease-related death, compared to patients with single-vessel disease.11 Our analysis aims to guide prescribing clinicians by aiding prediction of the benefit that might be achieved when patients meeting the EU label criteria are switched to ticagrelor 60 mg b.i.d. after 1 year of treatment with ticagrelor 90 mg b.i.d. The US label for ticagrelor is different from the EU label in that it suggests down-shifting from 90 to 60 mg b.i.d. after 12 months of treatment but otherwise has no suggested time limits or time-based guidance.

bel criteria are switched to ticagrelor 60 mg b.i.d. after 1 year of treatment with ticagrelor 90 mg b.i.d. The US label for ticagrelor is different from the EU label in that it suggests down-shifting from 90 to 60 mg b.i.d. after 12 months of treatment but otherwise has no suggested time limits or time-based guidance. Any consideration of prolonged antithrombotic therapy must also take safety into account. As expected, prolonged antithrombotic therapy was associated with more bleeding, but fatal or intracranial bleeding was not increased in this population in whom a high risk of life-threatening bleeding had been excluded at enrolment. Thus, in terms of net clinical benefit, for every 1000 eligible patients treated for 3 years, 24 major CV events would be avoided at the cost of only 10 major bleeds. Moreover, there was no excess of fatal or intracranial bleeds, and all-cause mortality was reduced.

-threatening bleeding had been excluded at enrolment. Thus, in terms of net clinical benefit, for every 1000 eligible patients treated for 3 years, 24 major CV events would be avoided at the cost of only 10 major bleeds. Moreover, there was no excess of fatal or intracranial bleeds, and all-cause mortality was reduced. Decisions on the safety and effectiveness of a drug in its intended use can be guided by an evaluation of the balance between the benefits and risks. The appropriate approach to such evaluation needs to depend on the severity of the disease and the intervention studied. The full complement of efficacy outcomes and safety evaluations in PEGASUS encompasses a range of event types with varying clinical significance; however, the assessment of the benefit-risk profile of ticagrelor used focuses primarily on those events with the greatest clinical importance. This approach has been supported as the most appropriate one for the assessment of benefit-risk balance since it compares endpoints of similar clinical impact, and integrates clinical judgement supported by quantitative analysis.9 While other therapeutic options may be considered such as the combination of low-dose factor Xa treatment12 and aspirin that combination has only been reported on patients further away from their index infarction. There are no directly comparative studies between long-term dual antiplatelet treatment with ticagrelor or treatment with low-dose factor Xa inhibitors on top of aspirin.

mbination of low-dose factor Xa treatment12 and aspirin that combination has only been reported on patients further away from their index infarction. There are no directly comparative studies between long-term dual antiplatelet treatment with ticagrelor or treatment with low-dose factor Xa inhibitors on top of aspirin. Limitations This is, by definition, a post hoc analysis since this subset was defined by regulators and not prospectively. The statistical analysis did not account for multiplicity of testing and this was not a prespecified analysis so therefore per se is hypothesis generating. However, the present analysis is of major clinical relevance to physicians and patients since this defines the benefits and risks to be expected in routine clinical practice. Conclusions In PEGASUS-TIMI 54, treatment with ticagrelor 60 mg b.i.d. in patients more recent to their MI or ADP receptor blocker discontinuation, as recommended in the EU label, reduced the risk of CV death, MI, or stroke by 20%, CV death by 29%, and all-cause mortality by 20%. Overall TIMI major bleeding was increased, but fatal or intracranial bleeding were not significantly different from placebo. There appears to be a favourable benefit-risk balance for long-term ticagrelor 60 mg b.i.d. in this population. Funding This study was supported by a grant from AstraZeneca.

Conclusions In PEGASUS-TIMI 54, treatment with ticagrelor 60 mg b.i.d. in patients more recent to their MI or ADP receptor blocker discontinuation, as recommended in the EU label, reduced the risk of CV death, MI, or stroke by 20%, CV death by 29%, and all-cause mortality by 20%. Overall TIMI major bleeding was increased, but fatal or intracranial bleeding were not significantly different from placebo. There appears to be a favourable benefit-risk balance for long-term ticagrelor 60 mg b.i.d. in this population. Funding This study was supported by a grant from AstraZeneca. Conflict of interest: M.D. discloses the following relationships: Advisory board: Novo Nordisk, AstraZeneca, Boehringer Ingelheim, Bayer. Speakers fee: AstraZeneca, Boehringer Ingelheim, Bayer. R.F.S. discloses the following relationships: institutional research grants from AstraZeneca and PlaqueTec; consultancy fees from Actelion, AstraZeneca, Avacta, Bayer, Bristol Myers Squibb/Pfizer, Idorsia, Novartis, PlaqueTec, and Thromboserin; and honoraria from AstraZeneca and Bayer. D.L.B. discloses the following relationships—Advisory Board: Cardax, Elsevier Practice Update Cardiology, Medscape Cardiology, Regado Biosciences; Board of Directors: Boston VA Research Institute, Society of Cardiovascular Patient Care, TobeSoft; Chair: American Heart Association Quality Oversight Committee; Data Monitoring Committees: Baim Institute for Clinical Research (formerly Harvard Clinical Research Institute, for the PORTICO trial, funded by St.

ciences; Board of Directors: Boston VA Research Institute, Society of Cardiovascular Patient Care, TobeSoft; Chair: American Heart Association Quality Oversight Committee; Data Monitoring Committees: Baim Institute for Clinical Research (formerly Harvard Clinical Research Institute, for the PORTICO trial, funded by St. Jude Medical, now Abbott), Cleveland Clinic, Duke Clinical Research Institute, Mayo Clinic, Mount Sinai School of Medicine (for the ENVISAGE trial, funded by Daiichi Sankyo), Population Health Research Institute; Honoraria: American College of Cardiology (Senior Associate Editor, Clinical Trials and News, ACC.org; Vice-Chair, ACC Accreditation Committee), Baim Institute for Clinical Research (formerly Harvard Clinical Research Institute; RE-DUAL PCI clinical trial steering committee funded by Boehringer Ingelheim), Belvoir Publications (Editor in Chief, Harvard Heart Letter), Duke Clinical Research Institute (clinical trial steering committees), HMP Global (Editor in Chief, Journal of Invasive Cardiology), Journal of the American College of Cardiology (Guest Editor; Associate Editor), Population Health Research Institute (for the COMPASS operations committee, publications committee, steering committee, and USA national co-leader, funded by Bayer), Slack Publications (Chief Medical Editor, Cardiology Today’s Intervention), Society of Cardiovascular Patient Care (Secretary/Treasurer), WebMD (CME steering committees); Other: Clinical Cardiology (Deputy Editor), NCDR-ACTION Registry Steering Committee (Chair), VA CART Research and Publications Committee (Chair); Research Funding: Abbott, Amarin, Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Chiesi, Eisai, Ethicon, Forest Laboratories, Idorsia, Ironwood, Ischemix, Lilly, Medtronic, PhaseBio, Pfizer, Regeneron, Roche, Sanofi Aventis, Synaptic, The Medicines Company; Royalties: Elsevier (Editor, Cardiovascular Intervention: A Companion to Braunwald’s Heart Disease); Site Co-Investigator: Biotronik, Boston Scientific, St. Jude Medical (now Abbott), Svelte; Trustee: American College of Cardiology; Unfunded Research: FlowCo, Merck, PLx Pharma, Takeda. C.H. discloses the following relationships: Advisory Board and speakers fee from AstraZeneca. R.G.K. discloses the following relationships: speakers bureau: Boehringer Ingelheim, Pfizer, Merck, Bayer AG. M.C. discloses the following relationships: speakers bureau and advisory boards: AstraZeneca. J.S. reports no relationships to disclose. A.H.

ing relationships: Advisory Board and speakers fee from AstraZeneca. R.G.K. discloses the following relationships: speakers bureau: Boehringer Ingelheim, Pfizer, Merck, Bayer AG. M.C. discloses the following relationships: speakers bureau and advisory boards: AstraZeneca. J.S. reports no relationships to disclose. A.H. is an employee of AstraZeneca. F.K. reports no relationships. G.K. reports no relationships. A.B. discloses the following relationships: consulting fees from: AstraZeneca, Sanofi-Aventis, Bristol Myers Squibb/Pfizer, GlaxoSmithKline, Bayer, Novartis; investigator fees from: AstraZeneca, Sanofi-Aventis, GlaxoSmithKline, Novartis, Bristol Myers Squibb/Pfizer, Eisai; honoraria for lectures from: AstraZeneca, Sanofi-Aventis, Bristol Myers Squibb/Pfizer, Novartis. P.J. is an employee of AstraZeneca. O.B. is an employee of AstraZeneca. F.V.d.W. and G.M. report no relationships. Supplementary Material pvz020_Supplementary_Data Click here for additional data file.

Introduction The adult congenital heart disease (ACHD) population is still growing and aging.1,2 Healthcare utilization is high, and drugs are more often prescribed in ACHD than in controls.3,4 Unlike other cardiovascular areas, evidence for drug therapy in ACHD is based on scarce clinical data and remains mostly empiric.5 Whether current pharmacological practice is efficient and safe in the long-term therefore remains questionable, but needs to be elucidated as drug therapy is increasingly used to address late complications. Pharmacological treatment in ACHD may start at a young age and may cumulate into chronic use of multiple medications. In elderly, it is known that the concurrent use of multiple medications, polypharmacy, is common (∼50%)6 and it is generally accepted that increased drug therapy is associated with adverse outcomes, such as adverse drug events (ADEs), hospitalizations, and death.7 However, data on polypharmacy in ACHD are lacking. Therefore, this study assessed medication use and polypharmacy in ACHD in comparison to the age- and sex-matched general population. Furthermore, we aimed to identify patterns of medication use in ACHD and to analyse the association between polypharmacy and adverse outcomes in ACHD.

on polypharmacy in ACHD are lacking. Therefore, this study assessed medication use and polypharmacy in ACHD in comparison to the age- and sex-matched general population. Furthermore, we aimed to identify patterns of medication use in ACHD and to analyse the association between polypharmacy and adverse outcomes in ACHD. Methods Study population and data collection This cohort study linked data of patients from the CONCOR registry,8 which includes adults (≥18 years) with congenital heart disease (CHD), to the national Dispensed Drug Register (DDR) of Statistics Netherlands (www.cbs.nl). For all Dutch residents, the DDR contains all dispensed outpatient drugs reimbursed by the compulsory basic Dutch health insurance. Drugs are classified following the Anatomical Therapeutic Chemical (ATC) classification (Supplementary material online, Table S1), which classifies drugs at five levels according to the organ/system on which they act (1st) and their therapeutic (2nd), pharmacological (3rd), and chemical properties (4th and 5th level).9 In the DDR, drugs are aggregated per person per year at the 3rd level of the ATC classification. Thus, specific drugs and their duration, timing, and daily doses within this 1-year window cannot be extracted. Receiving a specific drug is coded as dichotomous value for a full year, regardless of the amount of drugs dispensed. We, therefore, defined polypharmacy using the cumulative concept10 as ≥5 different drug types per calendar year, at the therapeutic (2nd) level of the ATC classification, to correct for changes in pharmacological classes.

c drug is coded as dichotomous value for a full year, regardless of the amount of drugs dispensed. We, therefore, defined polypharmacy using the cumulative concept10 as ≥5 different drug types per calendar year, at the therapeutic (2nd) level of the ATC classification, to correct for changes in pharmacological classes. Patients were matched with randomly selected age- and sex-matched reference subjects from the general population (1:10 ratio) to gain insight in the increase in medication use in ACHD compared to normal for these generally young persons (for details, see Supplementary material online, Methods and Figure S1). Subjects were followed from 2006 or CONCOR-inclusion until 2014 or death, using survival data from the national Cause of Death Register (CDR), which includes International Classification of Diseases (ICD) 10th revision coded causes of all deaths in Dutch citizens. From CONCOR, we obtained date of birth, inclusion date, sex, and main CHD, classified into mild, moderate, and severe CHD according to a much used consensus-based classification where proposed level of care and survival prospects differ per severity (Supplementary material online, Table S2).11,12

ths in Dutch citizens. From CONCOR, we obtained date of birth, inclusion date, sex, and main CHD, classified into mild, moderate, and severe CHD according to a much used consensus-based classification where proposed level of care and survival prospects differ per severity (Supplementary material online, Table S2).11,12 Additionally, data on hospitalizations for ADEs were collected via the Dutch Hospital Discharge Register (HDR) for the years 2006–12. The HDR contains person-linked discharge records of Dutch hospital admissions, including ICD-9 coded diagnoses and dates of admission. We defined hospitalizations for ADEs as admissions with ICD-9 codes 960–979 (poisoning by drugs, medicinal, and biological substances) as main diagnosis. The CDR was subsequently reviewed for ADEs as cause of death in all patients (ICD-10 codes T36–T50). CONCOR was approved by the ethics boards of all participating centres8 and complies with the declaration of Helsinki. Statistical analysis Statistical analyses were performed using RStudio V.1.0.153 (RStudio Team, Boston, MA, USA) and SPSS V.22 (IBM, Armonk, NY, USA). Data are summarized as n (%), mean ± standard deviation, and median [interquartile range (IQR)]. Two-sided P-values of <0.05 were considered statistically significant.

CONCOR was approved by the ethics boards of all participating centres8 and complies with the declaration of Helsinki. Statistical analysis Statistical analyses were performed using RStudio V.1.0.153 (RStudio Team, Boston, MA, USA) and SPSS V.22 (IBM, Armonk, NY, USA). Data are summarized as n (%), mean ± standard deviation, and median [interquartile range (IQR)]. Two-sided P-values of <0.05 were considered statistically significant. Drug use was described as percentage of years with dispensed drugs during the studied period. Generalized estimating equations with exchangeable working correlation and robust variance estimators were used to calculate odds ratios (ORs) for specific drugs and polypharmacy during the study in patients vs. matched referents, to determine whether sex, age, and CHD severity were independently associated with the presence of polypharmacy, and to plot predicted probability of polypharmacy by age in subsets per CHD severity. We performed subgroup analyses based on CHD type, sex, and age. A sensitivity analysis excluding sex hormones was performed to analyse the influence of oral contraceptives on the difference in polypharmacy between the sexes. We also performed sensitivity analysis excluding non-chronic drug types (including antibiotics, full list in Supplementary material online, Table S3) to test whether the cumulative definition of polypharmacy represented concurrent and continuous medication well.

ives on the difference in polypharmacy between the sexes. We also performed sensitivity analysis excluding non-chronic drug types (including antibiotics, full list in Supplementary material online, Table S3) to test whether the cumulative definition of polypharmacy represented concurrent and continuous medication well. To identify subgroups of patients with distinct patterns of medication relating to diseases of different organ systems, we used an unbiased machine learning approach. Of each patient, we determined whether drugs of the different anatomical classes of the ATC classification (1st level, Supplementary material online, Table S1) were used at year of inclusion. Hierarchical clustering was performed with the hclust and heatmap functions in R, using binary distance to calculate the dissimilarity matrix. The optimal number of clusters was estimated by maximizing the gap statistic using the gap method.13 Differences between clusters were compared using the χ2 and analysis of variance tests. Survival was assessed using the Kaplan–Meier analysis and compared between clusters using Cox hazard regression, adjusted for age, sex, and CHD severity.

clusters was estimated by maximizing the gap statistic using the gap method.13 Differences between clusters were compared using the χ2 and analysis of variance tests. Survival was assessed using the Kaplan–Meier analysis and compared between clusters using Cox hazard regression, adjusted for age, sex, and CHD severity. For survival analyses, we excluded patients who were included in 2014 or died in their year of inclusion, because the yearly aggregated data required follow-up starting the following year. Cumulative survival for patients with and without polypharmacy at inclusion was assessed per CHD using the Kaplan–Meier curves. Associations between polypharmacy and all-cause mortality were analysed using multivariable Cox regression adjusted for age, sex, and CHD severity, with polypharmacy as time-varying factor. Interaction terms were used to analyse differences between CHD severities, and between ACHD patients and referents. Similarly, Cox hazards regression was used to analyse whether polypharmacy was associated with hospitalizations for ADEs in ACHD patients. Results In total, 14 138 ACHD patients [age 35 (24–48) years, 49% male, 34% moderate, and 9% severe CHD] were followed for 8 (5–9) years (baseline characteristics in Supplementary material online, Table S4). Overall, 96 835 person-years of patients and 982 563 person-years of referents were analysed. Common drugs

For survival analyses, we excluded patients who were included in 2014 or died in their year of inclusion, because the yearly aggregated data required follow-up starting the following year. Cumulative survival for patients with and without polypharmacy at inclusion was assessed per CHD using the Kaplan–Meier curves. Associations between polypharmacy and all-cause mortality were analysed using multivariable Cox regression adjusted for age, sex, and CHD severity, with polypharmacy as time-varying factor. Interaction terms were used to analyse differences between CHD severities, and between ACHD patients and referents. Similarly, Cox hazards regression was used to analyse whether polypharmacy was associated with hospitalizations for ADEs in ACHD patients. Results In total, 14 138 ACHD patients [age 35 (24–48) years, 49% male, 34% moderate, and 9% severe CHD] were followed for 8 (5–9) years (baseline characteristics in Supplementary material online, Table S4). Overall, 96 835 person-years of patients and 982 563 person-years of referents were analysed. Common drugs Table 1 shows the most commonly dispensed drugs. Adult congenital heart disease patients had higher use of cardiovascular drugs than referents, with highest use of antithrombotics {27 vs. 6% in referents, OR = 5.83 [95% confidence interval (CI) 5.60–6.07]}, β-blockers [24 vs. 6%, OR = 4.43 (95% CI 4.26–4.61)], and renin–angiotensin–aldosterone system (RAAS) inhibitors [21 vs. 7%, OR = 3.32 (95% CI 3.17–3.47)] (Table 1A). Table 1 Dispensed drugs

Table 1 shows the most commonly dispensed drugs. Adult congenital heart disease patients had higher use of cardiovascular drugs than referents, with highest use of antithrombotics {27 vs. 6% in referents, OR = 5.83 [95% confidence interval (CI) 5.60–6.07]}, β-blockers [24 vs. 6%, OR = 4.43 (95% CI 4.26–4.61)], and renin–angiotensin–aldosterone system (RAAS) inhibitors [21 vs. 7%, OR = 3.32 (95% CI 3.17–3.47)] (Table 1A). Table 1 Dispensed drugs ACHD patients, n person-years = 96 835 (%) Matched referents, n person-years = 982 563 (%) OR (95% CI) A. Cardiovascular drugs Antithromboticsa (e.g. vitamin K antagonists, NOACs, platelet aggregation inhibitors) 26.5 5.4 5.83 (5.60–6.07) β-blockersa 23.7 6.3 4.43 (4.26–4.61) RAAS inhibitorsa 21.2 6.9 3.32 (3.17–3.47) Diureticsa 11.4 3.8 3.23 (3.07–3.40) Lipid modifiersa (e.g. statins) 10.3 6.7 1.48 (1.39–1.56) Calcium channel blockersa 6.1 2.6 2.17 (2.03–2.33) Antiarrhythmicsa 5.8 0.4 12.30 (11.23–13.47) Other antihypertensivesa 1.4 0.3 5.95 (5.14–6.90) Antihaemorrhagicsa (e.g. vitamin K, coagulation factors) 1.0 0.2 6.30 (5.61–7.05) Cardiac vasodilatorsa (e.g. nitrates) 0.3 0.2 1.72 (1.31–2.24) B. Non-cardiovascular drugs used in >10% of ACHD

Antithromboticsa (e.g. vitamin K antagonists, NOACs, platelet aggregation inhibitors) 26.5 5.4 5.83 (5.60–6.07) β-blockersa 23.7 6.3 4.43 (4.26–4.61) RAAS inhibitorsa 21.2 6.9 3.32 (3.17–3.47) Diureticsa 11.4 3.8 3.23 (3.07–3.40) Lipid modifiersa (e.g. statins) 10.3 6.7 1.48 (1.39–1.56) Calcium channel blockersa 6.1 2.6 2.17 (2.03–2.33) Antiarrhythmicsa 5.8 0.4 12.30 (11.23–13.47) Other antihypertensivesa 1.4 0.3 5.95 (5.14–6.90) Antihaemorrhagicsa (e.g. vitamin K, coagulation factors) 1.0 0.2 6.30 (5.61–7.05) Cardiac vasodilatorsa (e.g. nitrates) 0.3 0.2 1.72 (1.31–2.24) B. Non-cardiovascular drugs used in >10% of ACHD Systemic antibioticsa 37.8 19.7 2.45 (2.40–2.51) Anti-inflammatory and antirheumatic products (e.g. NSAIDs, excluding aspirin) 17.3 17.3 1.01 (0.98–1.03) Drugs for acid-related disordersa (e.g. PPIs and antacids) 15.1 10.3 1.60 (1.54–1.66) Dermatological corticosteroidsa 13.6 10.6 1.33 (1.29–1.37) Sex hormonesa (e.g. oral hormonal contraceptives) 11.2 8.6 1.33 (1.27–1.38) Drugs for obstructive airway diseasesa [includes inhalants (adrenergics, corticosteroids) and systemic adrenergics] 10.3 6.9 1.57 (1.50–1.65) Analgesicsa (e.g. opioids, aspirin) 10.2 6.7 1.58 (1.52–1.65) Ophtalmologicalsa (topical ocular drugs) 10.2 7.5 1.40 (1.35–1.46) Use of cardiovascular medication (A) and the most common non-cardiovascular medication (B) in ACHD patients compared with the use in matched referents from the general population. Drugs are presented according to the therapeutic classes of the Anatomical Therapeutic Chemical classification (Supplementary material online, Table S1).

vascular medication (A) and the most common non-cardiovascular medication (B) in ACHD patients compared with the use in matched referents from the general population. Drugs are presented according to the therapeutic classes of the Anatomical Therapeutic Chemical classification (Supplementary material online, Table S1). ACHD, adult congenital heart disease; NOAC, non-vitamin K antagonist oral anticoagulant; RAAS, renin–angiotensin–aldosterone system. a Significant at the P-value <0.001 level. Remarkably, most non-cardiovascular drugs were also used more frequently in ACHD, especially systemic antibiotics [38 vs. 20%, OR = 2.45 (95% CI 2.40–2.51)], drugs for acid-related disorders [15 vs. 10%, OR = 1.60 (95% CI 1.54–1.66)] and drugs for obstructive airway disease [10 vs. 7% OR = 1.57 (95% CI 1.50–1.65)] (Table 1B). Patients more commonly used drugs for thyroid disease than referents [3.8 vs. 2.0%, OR = 1.83 (95% CI 1.66–2.01)], especially patients with complete atrioventricular septal defects [OR = 15.69 (95% CI 9.53–25.83)] who often had Down syndrome [142 of 214 patients (67%)]. Antiepileptics also were more common [2.8 vs. 1.5%, OR = 1.84 (95% CI 1.68–2.02)], particularly in patients with transposition of the great arteries [OR = 4.58 (95% CI 2.87–7.33)] or a functionally univentricular heart [UVH; OR = 4.52 (95% CI 2.21–9.22)].

25.83)] who often had Down syndrome [142 of 214 patients (67%)]. Antiepileptics also were more common [2.8 vs. 1.5%, OR = 1.84 (95% CI 1.68–2.02)], particularly in patients with transposition of the great arteries [OR = 4.58 (95% CI 2.87–7.33)] or a functionally univentricular heart [UVH; OR = 4.52 (95% CI 2.21–9.22)]. Polypharmacy Adult congenital heart disease patients had a median of three different dispensed drugs at year of inclusion compared to a median of one in reference subjects (P < 0.001) (Figure 1). Twice as little patients were free of dispensed drugs at inclusion compared to referents (17 vs. 40%, P < 0.001) (most common drugs in polypharmacy: Supplementary material online, Table S5). Figure 1 Amount of different drugs types at inclusion in adult congenital heart disease patients and matched referents.

Polypharmacy Adult congenital heart disease patients had a median of three different dispensed drugs at year of inclusion compared to a median of one in reference subjects (P < 0.001) (Figure 1). Twice as little patients were free of dispensed drugs at inclusion compared to referents (17 vs. 40%, P < 0.001) (most common drugs in polypharmacy: Supplementary material online, Table S5). Figure 1 Amount of different drugs types at inclusion in adult congenital heart disease patients and matched referents. Mean prevalence of polypharmacy during the study was 30% in ACHD compared to 15% in referents [OR = 2.47 (95% CI 2.39–2.54)]. Polypharmacy was independently associated with older age, female sex, and CHD severity [mild: OR = 2.51 (95% CI 2.40–2.61), moderate: OR = 3.22 (95% CI 3.06–3.40), and severe: OR = 4.87 (95% CI 4.41–5.38)] (Figure 2). It was particularly present in patients with a UVH [44%, OR = 8.54 (6.62–11.02)], with many cardiovascular drugs indicating high-cardiac morbidity, and in patients with the Marfan syndrome [45%, OR = 4.60 (95% CI 3.98–5.31)], with notable use of cardiovascular drugs, ocular medication [18%, OR = 2.61 (95% CI 2.20–3.11)], and analgesics [16%, OR = 2.55 (95% CI 2.16–3.01)], reflecting ocular and skeletal problems (e.g. scoliosis) often seen in these syndromic patients. Figure 2 Factors independently associated with polypharmacy in the entire cohort, showing odds ratios (OR) for polypharmacy during the study period. CHD, congenital heart defect.

Mean prevalence of polypharmacy during the study was 30% in ACHD compared to 15% in referents [OR = 2.47 (95% CI 2.39–2.54)]. Polypharmacy was independently associated with older age, female sex, and CHD severity [mild: OR = 2.51 (95% CI 2.40–2.61), moderate: OR = 3.22 (95% CI 3.06–3.40), and severe: OR = 4.87 (95% CI 4.41–5.38)] (Figure 2). It was particularly present in patients with a UVH [44%, OR = 8.54 (6.62–11.02)], with many cardiovascular drugs indicating high-cardiac morbidity, and in patients with the Marfan syndrome [45%, OR = 4.60 (95% CI 3.98–5.31)], with notable use of cardiovascular drugs, ocular medication [18%, OR = 2.61 (95% CI 2.20–3.11)], and analgesics [16%, OR = 2.55 (95% CI 2.16–3.01)], reflecting ocular and skeletal problems (e.g. scoliosis) often seen in these syndromic patients. Figure 2 Factors independently associated with polypharmacy in the entire cohort, showing odds ratios (OR) for polypharmacy during the study period. CHD, congenital heart defect. Even in mild CHD, polypharmacy was already as common in 45-year-old female and 50-year-old male patients as in 65-year-old persons from the general population (Figure 3). Already 48% of patients with severe CHD had polypharmacy at the age of 45 years, a proportion only seen for persons aged ≥70 years in the general population. Figure 3 Probability of polypharmacy for women (A) and men (B) by age, stratified for congenital heart defect severity, and compared with age- and sex-matched referents.

nt rate of stroke/SE was 2.65/100 person-years for the apixaban group vs. 2.31/100 person-years for the rivaroxaban group (HR 1.00; 95% CI 0.89–1.14). The event rates of major bleeding were 1.76/100 person-years vs. 2.10/100 person-years in the apixaban- and rivaroxaban groups, respectively (HR 0.79; 95% CI 0.68–0.91). Subgroup analyses The risks of stroke or SE and major bleeding in selected subgroups are shown in Figure 4. No significant heterogeneity between subgroups was found with respect to risk of major bleeding. In the dabigatran–rivaroxaban-matched cohort, significant heterogeneity regarding risk of stroke/SE was seen in two subgroups; namely age <75 vs. >75 years, and patients with or without prior stroke/SE. Also, in the two other cohorts, heterogeneity was seen with respect to risk of stroke/SE in the subgroup of patients with or without prior stroke/SE. Figure 4 The risk of stroke or systemic embolism and major bleeding in selected subgroups. SE, systemic embolism. Patients initiating standard or reduced dose NOACs differed in baseline characteristics; the patients receiving reduced doses were more likely to be older and having more comorbidities than patients starting standard doses (Supplementary material online, Table S5). After propensity score re-matching on initial doses, both reduced- and standard-dose patients showed broadly consistent results to the main analysis (Figure 5).

Even in mild CHD, polypharmacy was already as common in 45-year-old female and 50-year-old male patients as in 65-year-old persons from the general population (Figure 3). Already 48% of patients with severe CHD had polypharmacy at the age of 45 years, a proportion only seen for persons aged ≥70 years in the general population. Figure 3 Probability of polypharmacy for women (A) and men (B) by age, stratified for congenital heart defect severity, and compared with age- and sex-matched referents. Overall, polypharmacy was more common in women than men [OR = 1.92 (95% CI 1.88–1.96)]. It was already present in 24% of female patients under 40 years (vs. 12% of female referents <40), with high use of antibiotics (41%) and sex hormones including contraceptives (31%). Even after exclusion of sex hormones, polypharmacy prevalence remained higher in women [OR = 1.88 (95% CI 1.74–1.78)]. In men, polypharmacy was less common at young age but showed a steep incline with age [OR = 2.3/10 years (95% CI 2.2–2.4), for women: OR = 1.6/10 years (95% CI 1.5–1.6); Pinteraction < 0.001]; 40% of male patients over 40 years had polypharmacy (vs. 19% of male referents >40), with high use of antithrombotics (46%) and RAAS inhibitors (23%). These sex- and age-specific differences were seen both in patients and referents. Mean prevalence of polypharmacy was still 25% in ACHD compared to 12% in matched referents [OR = 2.39 (95% CI 2.32–2.48)] when non-therapeutic and non-chronic drugs were excluded for sensitivity analysis.

Overall, polypharmacy was more common in women than men [OR = 1.92 (95% CI 1.88–1.96)]. It was already present in 24% of female patients under 40 years (vs. 12% of female referents <40), with high use of antibiotics (41%) and sex hormones including contraceptives (31%). Even after exclusion of sex hormones, polypharmacy prevalence remained higher in women [OR = 1.88 (95% CI 1.74–1.78)]. In men, polypharmacy was less common at young age but showed a steep incline with age [OR = 2.3/10 years (95% CI 2.2–2.4), for women: OR = 1.6/10 years (95% CI 1.5–1.6); Pinteraction < 0.001]; 40% of male patients over 40 years had polypharmacy (vs. 19% of male referents >40), with high use of antithrombotics (46%) and RAAS inhibitors (23%). These sex- and age-specific differences were seen both in patients and referents. Mean prevalence of polypharmacy was still 25% in ACHD compared to 12% in matched referents [OR = 2.39 (95% CI 2.32–2.48)] when non-therapeutic and non-chronic drugs were excluded for sensitivity analysis. Patterns of medication use The phenotype heat map created by hierarchical clustering of medication used in ACHD demonstrated heterogeneity among patients (Figure 4). The use of drugs acting on the cardiovascular and blood & blood forming organs (mainly antithrombotics) seemed to co-occur most.

Mean prevalence of polypharmacy was still 25% in ACHD compared to 12% in matched referents [OR = 2.39 (95% CI 2.32–2.48)] when non-therapeutic and non-chronic drugs were excluded for sensitivity analysis. Patterns of medication use The phenotype heat map created by hierarchical clustering of medication used in ACHD demonstrated heterogeneity among patients (Figure 4). The use of drugs acting on the cardiovascular and blood & blood forming organs (mainly antithrombotics) seemed to co-occur most. Figure 4 Medication phenotype heat map of adults with congenital heart disease. Columns represent individual patients and rows represent independent phenotypes of dispensed drugs aggregated at the anatomical level of the Anatomical Therapeutic Chemical classification. Red indicates increased value, yellow intermediate, and blue decreased value of a drug. White columns represent 2409 patients with zero drugs.

resent individual patients and rows represent independent phenotypes of dispensed drugs aggregated at the anatomical level of the Anatomical Therapeutic Chemical classification. Red indicates increased value, yellow intermediate, and blue decreased value of a drug. White columns represent 2409 patients with zero drugs. The analysis arrived at three clusters as the optimal number to reflect phenotypic variability (Supplementary material online, Figure S2). The clusters differed significantly (Supplementary material online, Table S6). As shown in Figure 5, Cluster 1 (n = 8317) had the highest proportion of patients with drugs acting on the cardiovascular and blood & blood forming systems. This cardiovascular cluster was the oldest and had most patients with severe CHD (10%) and left sided lesions (e.g. bicuspid aortic valve: 11%). Cluster 2 (n = 3501) mainly contained patients using anti-infectives and genito-urinary medication (sex hormones), but relative low use of other drugs, with polypharmacy in only 18% of patients. This low medication use cluster contained young, mainly female (70%) patients, mostly with mild defects (61%). In Cluster 3 (n = 2320), the comorbidity cluster, many patients used extra-cardiac medication. It had the highest proportion of patients with polypharmacy (36%) and genetic syndromes (7%). Figure 5 Clinical characteristics and medication use at inclusion stratified by phenogroup. Numbers represent the percentage of patients per subgroup with medication for the different organ systems used at year of inclusion.

iving reduced doses were more likely to be older and having more comorbidities than patients starting standard doses (Supplementary material online, Table S5). After propensity score re-matching on initial doses, both reduced- and standard-dose patients showed broadly consistent results to the main analysis (Figure 5). Figure 5 The risk of stroke or systemic embolism and major bleeding for patients using standard or reduced dose non-vitamin K antagonist oral anticoagulants. CI, confidence interval; Pys., person-years. Sensitivity analyses The results of the sensitivity analyses are shown in Supplementary material online, Table S6 and were in line with the primary analyses. Non-vitamin K antagonist oral anticoagulant–warfarin comparisons Comparing each NOAC with warfarin, we found no significant differences in the adjusted HRs of stroke/SE for any NOAC compared with warfarin, while dabigatran and apixaban were both associated with lower risk of major bleeding (Supplementary material online, Table S4).

The analysis arrived at three clusters as the optimal number to reflect phenotypic variability (Supplementary material online, Figure S2). The clusters differed significantly (Supplementary material online, Table S6). As shown in Figure 5, Cluster 1 (n = 8317) had the highest proportion of patients with drugs acting on the cardiovascular and blood & blood forming systems. This cardiovascular cluster was the oldest and had most patients with severe CHD (10%) and left sided lesions (e.g. bicuspid aortic valve: 11%). Cluster 2 (n = 3501) mainly contained patients using anti-infectives and genito-urinary medication (sex hormones), but relative low use of other drugs, with polypharmacy in only 18% of patients. This low medication use cluster contained young, mainly female (70%) patients, mostly with mild defects (61%). In Cluster 3 (n = 2320), the comorbidity cluster, many patients used extra-cardiac medication. It had the highest proportion of patients with polypharmacy (36%) and genetic syndromes (7%). Figure 5 Clinical characteristics and medication use at inclusion stratified by phenogroup. Numbers represent the percentage of patients per subgroup with medication for the different organ systems used at year of inclusion. After 8 years of follow-up, cumulative survival was 92% in the cardiovascular cluster, 98% in the low medication use cluster, and 95% in the comorbidity cluster. Corrected for age, sex, and CHD severity, survival was better for the low medication use vs. cardiovascular cluster [hazard ratio (HR) = 0.50 (95% CI 0.37–0.78), P < 0.001], but, despite the distinct medication patterns, did not differ between the comorbidity and cardiovascular cluster [HR = 0.89 (95% CI 0.71–1.11), P = 0.31].

d for age, sex, and CHD severity, survival was better for the low medication use vs. cardiovascular cluster [hazard ratio (HR) = 0.50 (95% CI 0.37–0.78), P < 0.001], but, despite the distinct medication patterns, did not differ between the comorbidity and cardiovascular cluster [HR = 0.89 (95% CI 0.71–1.11), P = 0.31]. Polypharmacy and outcome Survival analyses included 13 527 patients and 135 647 referents. During 7 (5–8) years, 595 (4%) patients and 2375 (2%) referents died (Figure 6). Eight-year mortality was higher in patients with polypharmacy at inclusion compared to those without polypharmacy (Figure 7). Corrected for age, sex, and defect severity, polypharmacy during the study was strongly associated with all-cause mortality in ACHD [HR = 3.94 (95% CI 3.22–4.81)]. The age- and sex-adjusted association was similar between the CHD severities (Pinteraction = 0.96 for moderate and Pinteraction = 0.70 for severe CHD compared to mild CHD) and was significantly stronger in ACHD patients than in referents (Pinteraction < 0.001). Figure 6 Kaplan–Meier survival curve of adult congenital heart disease (ACHD) patients and matched referents with and without polypharmacy at inclusion.

Polypharmacy and outcome Survival analyses included 13 527 patients and 135 647 referents. During 7 (5–8) years, 595 (4%) patients and 2375 (2%) referents died (Figure 6). Eight-year mortality was higher in patients with polypharmacy at inclusion compared to those without polypharmacy (Figure 7). Corrected for age, sex, and defect severity, polypharmacy during the study was strongly associated with all-cause mortality in ACHD [HR = 3.94 (95% CI 3.22–4.81)]. The age- and sex-adjusted association was similar between the CHD severities (Pinteraction = 0.96 for moderate and Pinteraction = 0.70 for severe CHD compared to mild CHD) and was significantly stronger in ACHD patients than in referents (Pinteraction < 0.001). Figure 6 Kaplan–Meier survival curve of adult congenital heart disease (ACHD) patients and matched referents with and without polypharmacy at inclusion. Figure 7 Eight-year cumulative mortality for patients with and without polypharmacy at inclusion per congenital heart defect. ASD, atrial septal defect; BAV, bicuspid aortic valve; cAVSD, complete atrioventricular septal defect; CoA, coarctation of the aorta; Ebstein, Ebstein’s anomaly; LVOTO, left ventricular outflow tract obstruction; pAVSD, partial atrioventricular septal defect; RVOTO, right ventricular outflow tract obstruction; TGA, transposition of the great arteries; TOF, tetralogy of Fallot; UVH, univentricular heart; VSD, ventricular septal defect.

n of the aorta; Ebstein, Ebstein’s anomaly; LVOTO, left ventricular outflow tract obstruction; pAVSD, partial atrioventricular septal defect; RVOTO, right ventricular outflow tract obstruction; TGA, transposition of the great arteries; TOF, tetralogy of Fallot; UVH, univentricular heart; VSD, ventricular septal defect. A total of 10 015 ACHD patients were uniquely identified in the HDR between 2005 and 2012. During a median of 5 (IQR 3–6) years, 21 ACHD patients were hospitalized for an ADE. Increasing drug amounts were associated with ADEs [HR = 1.20/dispensed drug (95% CI 1.10–1.32)]. Patients with polypharmacy were at markedly higher risk of hospitalization for an ADE compared to patients without polypharmacy [HR = 4.03 (95% CI 1.67–9.73)]. None of the patients that died during the study had ADEs as cause of death. Discussion This study shows that ACHD patients not only use more cardiovascular medication than the general population, but also use more extra-cardiac drugs, cumulating into polypharmacy in 30% of the patients compared to only 15% of referents. The study identified distinct medication patterns, which differed by age, sex, and CHD. Furthermore, patients with polypharmacy had an almost four-fold higher risk of all-cause mortality and almost five-fold higher risk of hospitalizations for ADEs.

polypharmacy in 30% of the patients compared to only 15% of referents. The study identified distinct medication patterns, which differed by age, sex, and CHD. Furthermore, patients with polypharmacy had an almost four-fold higher risk of all-cause mortality and almost five-fold higher risk of hospitalizations for ADEs. Recently, ACHD investigators have stressed the need for more evidence regarding drug therapy in this growing population.5 Trials investigating safety and efficacy of drugs in ACHD often remain small.14,15 The existing pool of evidence in this area therefore only grows slowly and remains largely empiric. Some epidemiologic studies have identified common drugs in ACHD cohorts.4,16 However, this study is the first to investigate polypharmacy and its associations with clinical characteristics and outcome in ACHD. Furthermore, this is the largest study comparing medication use in ACHD to the general population.

ly empiric. Some epidemiologic studies have identified common drugs in ACHD cohorts.4,16 However, this study is the first to investigate polypharmacy and its associations with clinical characteristics and outcome in ACHD. Furthermore, this is the largest study comparing medication use in ACHD to the general population. Previous studies focusing on other chronic conditions, such as diabetes mellitus, chronic kidney disease, and chronic heart failure, have shown comparably high odds for polypharmacy of these diseases.17,18 Compared to these populations, ACHD patients are special due to their young age and lifelong disease which may involve both cardiac and extra-cardiac comorbidities. Polypharmacy in 15% of the age-matched referents may seem high but is close to other findings using cumulative definitions of polypharmacy during a 1-year period.19 Not surprisingly, polypharmacy risk in our study increased with increasing CHD severity, which involves more cardiovascular complications requiring medical intervention.16,20

of the age-matched referents may seem high but is close to other findings using cumulative definitions of polypharmacy during a 1-year period.19 Not surprisingly, polypharmacy risk in our study increased with increasing CHD severity, which involves more cardiovascular complications requiring medical intervention.16,20 Apart from common use of cardiovascular drugs, use of many non-cardiovascular drugs was increased in ACHD. Previous research showed increased prevalence of drugs related to asthma and epilepsy in patients who underwent surgery for a CHD as children.4 Especially in patients with genetic syndromes, extra-cardiac comorbidities are common.4,21 In our cohort, we saw increased use of a large range of drugs, including drugs for acid-related disorders, dermatologicals, and sex hormones. This indicates high prevalence of extra-cardiac comorbidities in the ACHD population. Contra-indications for pregnancy are more common in women with cardiovascular disease22 and may explain a higher preventive use of oral contraceptives in ACHD.

uding drugs for acid-related disorders, dermatologicals, and sex hormones. This indicates high prevalence of extra-cardiac comorbidities in the ACHD population. Contra-indications for pregnancy are more common in women with cardiovascular disease22 and may explain a higher preventive use of oral contraceptives in ACHD. Interestingly, polypharmacy was even increased in mild CHD and at young age, reflecting decreased health even in these mildly affected patients. Alternatively, the increase in medication use may originate from intensive surveillance that facilitates early diagnosis and treatment.3 The particularly higher prevalence of polypharmacy in female compared to male ACHD patients at young age is in line with general sex differences that depend on differences including prevalence of morbidities and adverse drug effects, need for anticonceptives, and a lower likelihood to seek preventive healthcare in men.23

e particularly higher prevalence of polypharmacy in female compared to male ACHD patients at young age is in line with general sex differences that depend on differences including prevalence of morbidities and adverse drug effects, need for anticonceptives, and a lower likelihood to seek preventive healthcare in men.23 Cluster analysis revealed three distinct patterns of medication use in ACHD, described as cardiovascular, low medication use, and comorbidity patterns. Cluster analysis based on phenotypical data has been used previously to identify distinct subgroups within other heterogeneous populations.24,25 This unbiased approach makes it possible to identify patterns regardless of assumptions about clinical correlations. The identification of such distinct subgroups could be used to help target therapies and trials in heterogeneous syndromes such as ACHD. Clinical trials are prone to select patients without marked comorbidity, but concurrent use of different drugs is important to identify due to increased risk of drug–drug interactions and ADEs.26,27 This may be most crucial in the comorbidity subgroup.

arget therapies and trials in heterogeneous syndromes such as ACHD. Clinical trials are prone to select patients without marked comorbidity, but concurrent use of different drugs is important to identify due to increased risk of drug–drug interactions and ADEs.26,27 This may be most crucial in the comorbidity subgroup. This study showed, without implying causality, that patients with polypharmacy had a four-fold higher mortality risk (HR = 3.94), independent of age, sex, and defect severity. Furthermore, risk of hospitalization for adverse drug events was nearly five times higher in patients with polypharmacy (HR = 4.58). Interestingly, polypharmacy in the ACHD population was more associated with mortality than in the general population. Patients with polypharmacy may be sicker (needing therapy) than referents with polypharmacy (who e.g. often have statins as prevention). Whether an increased amount of drugs is an independent risk factor or a mere measure of poor health and multimorbidity, remains to be elucidated.6,7 Polypharmacy may enhance risk of adverse drug events, including bleeding due to antithrombotics,28 and increased amounts of drugs correlate with hospitalizations for adverse drug reactions.26,27 Notably, drugs often prescribed in ACHD, especially anticoagulants, are among the drugs most commonly causing ADE-related emergency department visits and hospitalizations.29,30 Benefits of prescribing may outweigh the risks of ADEs, but evidence of beneficial effects of many therapies in ACHD is still limited.5 In elderly, guidelines with criteria to start and stop certain drugs have been established to minimize inappropriate prescribing,31 and it has been suggested that deprescribing to reduce inappropriate polypharmacy can reduce mortality without harm.32,33

of beneficial effects of many therapies in ACHD is still limited.5 In elderly, guidelines with criteria to start and stop certain drugs have been established to minimize inappropriate prescribing,31 and it has been suggested that deprescribing to reduce inappropriate polypharmacy can reduce mortality without harm.32,33 Clinical implications The remarkably high prevalence of polypharmacy in ACHD shows that experience with managing polypharmacy is needed in the efficient management of these patients. Physicians should carefully judge drug indications in ACHD, especially as pharmacotherapy is often based on low-level evidence extrapolated from non-ACHD studies or small studies involving heterogeneous ACHD patients. Long-term use of some medication, e.g. amiodarone, may be suboptimal due to side effects.5 Occasionally, withdrawal of longstanding therapy with only weak indications might be an option. Trials that examine efficacy and safety of drug therapy in ACHD are warranted, and the effects of longstanding polypharmacy in these patients need to be studied further to enhance guidelines on the management of this complex population.

, withdrawal of longstanding therapy with only weak indications might be an option. Trials that examine efficacy and safety of drug therapy in ACHD are warranted, and the effects of longstanding polypharmacy in these patients need to be studied further to enhance guidelines on the management of this complex population. Methodological issues These data from national administrative databases enable insightful comparisons with the general population. Automated data collection limits recall bias seen in questionnaires and data on dispensed drugs provide more accurate information on actual drug consumption than medical records, as these prescriptions have been filled. However, actual drug consumption may be overestimated, as we have no data on compliance. Non-compliance is of importance because it is associated with mortality and increases with treatment intensity and duration,34,35 although compliance in the Netherlands is reported to be high (>80%).36

escriptions have been filled. However, actual drug consumption may be overestimated, as we have no data on compliance. Non-compliance is of importance because it is associated with mortality and increases with treatment intensity and duration,34,35 although compliance in the Netherlands is reported to be high (>80%).36 The lack of clinical detail inherent to administrative data introduces indication bias, as no information on drug indications, comorbidities, and functional status are available. We used the consensus-based severity classification to subdivide patients with different risks. However, mortality risk may vary within specific CHDs due to late complications, such as pulmonary hypertension in patients with septal defects. Therefore, these data do not provide information about individual patients, but give insight on a population level. Furthermore, appropriateness of polypharmacy is not assessed and associations with mortality have to be interpreted with caution, as polypharmacy may mark high-risk patients with multimorbidity.

defects. Therefore, these data do not provide information about individual patients, but give insight on a population level. Furthermore, appropriateness of polypharmacy is not assessed and associations with mortality have to be interpreted with caution, as polypharmacy may mark high-risk patients with multimorbidity. Other limitations inherent to the data set include unavailability of data on over-the-counter medication, and data on treatment duration, daily doses, and specific distributed drugs. Our cumulative measures of polypharmacy may overestimate the prevalence of simultaneous pharmacotherapy, due to inclusion of successive and non-chronic drugs in the observed time frame. We limited this by aggregating drugs by therapeutic class, correcting for switches in pharmacological class. Such cumulative definitions of polypharmacy are common and give comparable, clinically relevant, and as reliable results as other measures of polypharmacy.10,19,37

d non-chronic drugs in the observed time frame. We limited this by aggregating drugs by therapeutic class, correcting for switches in pharmacological class. Such cumulative definitions of polypharmacy are common and give comparable, clinically relevant, and as reliable results as other measures of polypharmacy.10,19,37 Conclusion In conclusion, ACHD patients used both more cardiovascular and non-cardiovascular medication compared to the general population, with polypharmacy in 30% of ACHD vs. just 15% of referents. Polypharmacy was even common in mild CHD at young ages. We identified different medication patterns that could be taken into account to help target therapies and trials in this heterogeneous population. As patients with polypharmacy had a four-fold higher risk of death and adverse drug events, daily clinical care of ACHD patients must include regular evaluation of their medication regimen, particularly in case of polypharmacy. Further clinical trials to investigate risks and benefits of pharmacotherapy remain needed to come to more evidence-based treatment in this population. Supplementary Material pvz014_Supplementary_Data Click here for additional data file.

Conclusion In conclusion, ACHD patients used both more cardiovascular and non-cardiovascular medication compared to the general population, with polypharmacy in 30% of ACHD vs. just 15% of referents. Polypharmacy was even common in mild CHD at young ages. We identified different medication patterns that could be taken into account to help target therapies and trials in this heterogeneous population. As patients with polypharmacy had a four-fold higher risk of death and adverse drug events, daily clinical care of ACHD patients must include regular evaluation of their medication regimen, particularly in case of polypharmacy. Further clinical trials to investigate risks and benefits of pharmacotherapy remain needed to come to more evidence-based treatment in this population. Supplementary Material pvz014_Supplementary_Data Click here for additional data file. Acknowledgements The authors thank all CONCOR participants, Lia Engelfriet, and Sylvia Mantels. This work was carried out in the context of the Parelsnoer Institute (part of and funded by the Dutch Federation of University Medical Centers). Results are based on calculations by the Amsterdam UMC—University of Amsterdam, using non-public microdata from Statistics Netherlands, which are accessible for statistical and scientific research under certain conditions. Funding This work was supported by the Dutch Heart Foundation [CVON 2014-18 project CONCOR-genes to O.I.W., F.J.M., and B.J.B.] and the Amsterdam University Fund [8532].

Acknowledgements The authors thank all CONCOR participants, Lia Engelfriet, and Sylvia Mantels. This work was carried out in the context of the Parelsnoer Institute (part of and funded by the Dutch Federation of University Medical Centers). Results are based on calculations by the Amsterdam UMC—University of Amsterdam, using non-public microdata from Statistics Netherlands, which are accessible for statistical and scientific research under certain conditions. Funding This work was supported by the Dutch Heart Foundation [CVON 2014-18 project CONCOR-genes to O.I.W., F.J.M., and B.J.B.] and the Amsterdam University Fund [8532]. Conflict of interest: B.J.M.M. and B.J.B. report grants from Actelion Pharmaceuticals, Bristol-Myers Squibb, Boehringer Ingelheim, Bayer, and Daiichi Sankyo outside this work.

Introduction Oral anticoagulants (OACs) are effective in preventing stroke and systemic embolism (SE) in patients with atrial fibrillation (AF) but are associated with an increased risk of bleeding.1 Guidelines recommend use of non-vitamin K antagonist oral anticoagulants (NOACs) over traditional therapy with vitamin K antagonists in most patients,2 and the number of patients being treated with NOACs has increased rapidly during the last few years.3 In the pivotal randomized controlled trials (RCTs) leading to their approval, each NOAC was compared with warfarin,4–6 however, no head-to-head comparison between the individual NOACs has been performed. In the absence of RCTs, observational studies utilizing data from clinical practice may add useful information regarding comparative effectiveness and safety of the individual NOACs. The aim of this study was to assess the association between the use of dabigatran, rivaroxaban, and apixaban and the risk of stroke or SE and bleeding in a nationwide cohort of patients with AF. Methods Data sources The Norwegian Patient Registry (NPR) is a nationwide registry that covers all hospital admissions and outpatient consultations as well as all specialist consultations in Norway. Each admission or consultation is assigned a primary (the disease or condition being treated) and secondary cause (relevant comorbidities). Diagnoses are coded according to the International Classification of Diseases, 10th revision (ICD-10)7 system and surgical procedures are coded according to the Nordic Medico-Statistical Committee (NOMESCO) coding system.8,9

primary (the disease or condition being treated) and secondary cause (relevant comorbidities). Diagnoses are coded according to the International Classification of Diseases, 10th revision (ICD-10)7 system and surgical procedures are coded according to the Nordic Medico-Statistical Committee (NOMESCO) coding system.8,9 The Norwegian Prescription Database (NorPD) holds information on all drug prescriptions dispensed from pharmacies nationwide. Drugs are coded according to the Anatomical Therapeutic Chemical (ATC) system.10 The Norwegian system of general reimbursement of medicine expenses requires the prescribing physician to state the relevant underlying disease warranting each drug’s reimbursement. The NorPD also contains information about date of dispensation, quantity, and strength of drugs dispensed.

apeutic Chemical (ATC) system.10 The Norwegian system of general reimbursement of medicine expenses requires the prescribing physician to state the relevant underlying disease warranting each drug’s reimbursement. The NorPD also contains information about date of dispensation, quantity, and strength of drugs dispensed. Cohort creation and study design The study cohort was generated by linkage of data from the NPR and the NorPD (Figure 1). The study population included all patients ≥18 years diagnosed with AF with at least one OAC dispensation (dabigatran 110 mg or 150 mg, rivaroxaban 15 mg or 20 mg, apixaban 2.5 mg or 5 mg, or warfarin 2.5 mg) in the study period (January 2013 to December 2017) but being anticoagulant naïve before start of the study. Patients initiating warfarin were included to enable comparisons between our findings and previous studies including patients treated with warfarin. Patients were excluded if they had mitral stenosis or mechanical prosthetic heart valves. Anticoagulant-naïve was defined as no dispensing of anticoagulants from pharmacies in the preceding 12 months before the index date. The index date was defined as the date of the first dispensation of an OAC in the study period. Due to limited usage in the study period, patients initiating edoxaban were excluded (n = 107). Patients with a history of venous thromboembolism during the last 180 days, or knee- or hip replacement surgery during the last 35 days before the index date were excluded. Details of the cohort creation procedure are shown in Figure 1, and ICD-10 codes used for inclusion- and exclusion criteria are listed in the Supplementary material online, Table S1.

hromboembolism during the last 180 days, or knee- or hip replacement surgery during the last 35 days before the index date were excluded. Details of the cohort creation procedure are shown in Figure 1, and ICD-10 codes used for inclusion- and exclusion criteria are listed in the Supplementary material online, Table S1. Figure 1 Cohort creation flow chart. AF*, atrial fibrillation in the absence of mitral stenosis or mechanical prosthetic heart valves; NPR, Norwegian Patient Registry; NorPD, Norwegian Prescription Database; OAC, oral anticoagulant; VTE, venous thromboembolism. Patients treated with a NOAC were matched with respect to propensity score, and three pairwise-matched cohorts were created: dabigatran vs. rivaroxaban; dabigatran vs. apixaban; and apixaban vs. rivaroxaban. Details of the propensity score matching (PSM) are found in the section on statistical analysis. Comorbidities Diagnoses for all hospital admissions, consultations, and procedures in the previous 5 years before the index date were retrieved from the NPR. A medication history of 5 years, including all relevant diagnosis-specific reimbursement codes, was completed from the NorPD. This information was used to compile a set of comorbidities and medication history for each patient, using primary as well as secondary codes related to each admission. The ICD-10 codes included for each diagnosis are shown in Supplementary material online, Table S1, and Supplementary material online, Table S2 shows in detail how CHA2DS2-VASc- and HAS-BLED scores were calculated.

d medication history for each patient, using primary as well as secondary codes related to each admission. The ICD-10 codes included for each diagnosis are shown in Supplementary material online, Table S1, and Supplementary material online, Table S2 shows in detail how CHA2DS2-VASc- and HAS-BLED scores were calculated. Oral anticoagulant supply For each OAC, the days of supply were computed using information on dates of dispensing, the pack-size dispensed, and the number of packages. As the NOACs are prescribed in fixed doses, to be taken once daily (rivaroxaban) or twice daily (dabigatran and apixaban), the number of days of supply strictly corresponds to the amount dispensed. The days of warfarin supply were estimated as previously described.11 To account for incomplete adherence, a 30-day gap period between the calculated end of OAC supply and the date of a new prescription was allowed, before patients were censored.

he number of days of supply strictly corresponds to the amount dispensed. The days of warfarin supply were estimated as previously described.11 To account for incomplete adherence, a 30-day gap period between the calculated end of OAC supply and the date of a new prescription was allowed, before patients were censored. Outcomes and follow-up Outcome measures of effectiveness were time to first stroke (haemorrhagic or ischaemic) or SE, and time to first ischaemic stroke. Outcome measures of safety were time to first major bleeding, clinically relevant non-major bleeding (CRNM bleeding), major or CRNM bleeding, gastrointestinal bleeding (GI bleeding), and intracranial haemorrhage. Major bleeding was defined as previously described as any bleeding into a critical area or organ, or any bleeding accompanied by blood transfusion ≤10 days after hospital admission date.11 CRNM bleeding was defined according to the International Society on Thrombosis and Haemostasis (ISTH) classification,12 as any bleeding necessitating intervention by a medical professional. ICD-10 and NOMESCO codes used for identification of outcomes are listed in Supplementary material online, Table S1. Patients were followed from the index date until discontinuation or switching of OACs, death, or end of study period (31 December 2017), whichever occurred first. For the identification of effectiveness- and safety outcomes, only primary (first listed) ICD-10 codes for each hospital stay were used.

al online, Table S1. Patients were followed from the index date until discontinuation or switching of OACs, death, or end of study period (31 December 2017), whichever occurred first. For the identification of effectiveness- and safety outcomes, only primary (first listed) ICD-10 codes for each hospital stay were used. Ethics Registration of data into the NPR and the NorPD is mandatory in Norway and legally exempt from obtainment of patient consent. This study was approved by the Regional Ethical Committee (Ref. No. 2017/410/REK North). Statistical analysis Categorical variables are reported by numbers and percent, continuous variables by means with standard deviations. Cox proportional hazards regression was used to select the strongest predictor variables for stroke/SE and major bleeding. The proportional hazards assumption was checked using Schoenfeld residuals, and by comparing the log–log transformation of the Kaplan–Meier survival curves for each variable.13

standard deviations. Cox proportional hazards regression was used to select the strongest predictor variables for stroke/SE and major bleeding. The proportional hazards assumption was checked using Schoenfeld residuals, and by comparing the log–log transformation of the Kaplan–Meier survival curves for each variable.13 To account for confounding by indication of therapy, PSM was performed. Using logistic regression, the probability of a patient being prescribed a specific NOAC was calculated on the basis of the following 16 covariates; age, gender, chronic kidney disease, hypertension, diabetes, ischaemic heart disease, peripheral artery disease, heart failure, history of stroke/SE, history of bleeding-related hospitalization, anaemia, active cancer (cancer diagnosis last 12 months), chronic lower respiratory tract disease, use of cholesterol lowering drugs, use of antiplatelet drugs, and use of non-steroidal anti-inflammatory drugs during the last 12 months. For each patient initiating a specific NOAC, initiators of another NOAC to be compared were matched 1:1 on the logit of the propensity score using calipers of width equal to 0.2 of the standard deviation of the logit of the propensity score.14 Three propensity score-matched sets were constructed; dabigatran matched with rivaroxaban, dabigatran matched with apixaban, and rivaroxaban matched with apixaban. The balance between treatment populations was assessed by investigating absolute standardized mean differences of all baseline covariates before and after the matching, using a threshold of 0.1 to indicate imbalance. Cox regression with robust sandwich estimates was utilized for evaluating the rates of stroke and bleeding in the propensity score-matched groups.15 As the matched sets were balanced, NOAC treatment was entered as the only independent variable.16,17 Subgroup analyses were performed investigating the risk of stroke and major bleeding in specific subgroups; age (<75 years vs. >75 years), gender, history of stroke, and history of bleeding. For the analyses stratified on the initial dose, de novo PSM within the initial dose defined subgroups were performed. Adjusted hazard ratios (HRs) along with P-values for interaction between treatment and the specific subgroup were calculated.

5 years vs. >75 years), gender, history of stroke, and history of bleeding. For the analyses stratified on the initial dose, de novo PSM within the initial dose defined subgroups were performed. Adjusted hazard ratios (HRs) along with P-values for interaction between treatment and the specific subgroup were calculated. Three sensitivity analyses were performed: (i) the analyses of the outcomes stroke/SE and major bleeding in the PSM cohorts were repeated restricting the follow-up time for all NOACs to 12 months; (ii) an ‘intention-to-treat’-like analysis: the analyses of the outcomes stroke/SE and major bleeding in the PSM cohorts were performed without censoring by treatment switch or discontinuation of NOACs. (iii) The comparative analyses of the outcomes stroke/SE and major bleeding were repeated in the full dataset using conventional adjustment instead of PSM to avoid exclusion of non-matched patients from the analyses. Finally, as a post hoc analysis, we performed NOAC–warfarin comparisons. The risk of stroke/SE and major bleeding were compared between users of dabigatran, rivaroxaban, apixaban, and users of warfarin, using a Cox proportional hazards model with conventional adjustment. Level of significance was set to 5%. We did not adjust for multiple comparisons. Statistical analyses were performed using SAS v.9.4 (SAS Institute, Inc.) and STATA v.15 (STATACorp LLC).

Finally, as a post hoc analysis, we performed NOAC–warfarin comparisons. The risk of stroke/SE and major bleeding were compared between users of dabigatran, rivaroxaban, apixaban, and users of warfarin, using a Cox proportional hazards model with conventional adjustment. Level of significance was set to 5%. We did not adjust for multiple comparisons. Statistical analyses were performed using SAS v.9.4 (SAS Institute, Inc.) and STATA v.15 (STATACorp LLC). Results A total of 65 563 new users of OACs were identified and included in the study population; 10 413 initiated dabigatran, 13 700 rivaroxaban, 28 363 apixaban, and 13 087 initiated warfarin (Figure 1). Baseline characteristics for the unmatched groups are shown in Supplementary material online, Table S3. New users of dabigatran were more likely to be younger than new users of the other drugs, and they also had less comorbidity. The mean CHA2DS2-VASc- and HAS-BLED scores were lowest in users of dabigatran. The standard dose for stroke prevention was used in 63.9% of dabigatran patients, 75.6% of rivaroxaban patients, and 74.6% of apixaban patients.

more likely to be younger than new users of the other drugs, and they also had less comorbidity. The mean CHA2DS2-VASc- and HAS-BLED scores were lowest in users of dabigatran. The standard dose for stroke prevention was used in 63.9% of dabigatran patients, 75.6% of rivaroxaban patients, and 74.6% of apixaban patients. Non-vitamin K antagonist oral anticoagulant–non-vitamin K antagonist oral anticoagulant comparisons After PSM in a 1:1 ratio, the cohorts used in the analyses of dabigatran vs. rivaroxaban included a total of 20 504 patients, dabigatran vs. apixaban included a total of 20 826 patients, and rivaroxaban vs. apixaban included a total of 27 398 patients. In each of the matched cohorts, baseline characteristics were well-balanced between the groups (Table 1). Plots of propensity scores before and after matching are shown in the Supplementary material, Figure S1. Figure 2 shows Kaplan–Meier curves for the risk of stroke/SE and major bleeding, whereas Figure 3 shows the incidence rates and HRs of the outcomes stroke/SE and major bleeding for the three PSM cohorts. The proportional hazard assumption was fulfilled for all primary analyses. Figure 2 Cumulative incidence of stroke or systemic embolism (A) and major bleeding (B) in the propensity score-matched groups. CI, confidence interval; HR, hazard ratio; OAC, oral anticoagulant; SE, systemic embolism.

Non-vitamin K antagonist oral anticoagulant–non-vitamin K antagonist oral anticoagulant comparisons After PSM in a 1:1 ratio, the cohorts used in the analyses of dabigatran vs. rivaroxaban included a total of 20 504 patients, dabigatran vs. apixaban included a total of 20 826 patients, and rivaroxaban vs. apixaban included a total of 27 398 patients. In each of the matched cohorts, baseline characteristics were well-balanced between the groups (Table 1). Plots of propensity scores before and after matching are shown in the Supplementary material, Figure S1. Figure 2 shows Kaplan–Meier curves for the risk of stroke/SE and major bleeding, whereas Figure 3 shows the incidence rates and HRs of the outcomes stroke/SE and major bleeding for the three PSM cohorts. The proportional hazard assumption was fulfilled for all primary analyses. Figure 2 Cumulative incidence of stroke or systemic embolism (A) and major bleeding (B) in the propensity score-matched groups. CI, confidence interval; HR, hazard ratio; OAC, oral anticoagulant; SE, systemic embolism. Figure 3 Number of events, incidence rates, and hazard ratios for primary and secondary outcomes in the three propensity score-matched cohorts. CI, confidence interval; CRNM, clinically relevant non-major; GI, gastrointestinal; Pys., person-years; SE, systemic embolism. Table 1 Baseline characteristics, propensity-matched groups

Figure 3 Number of events, incidence rates, and hazard ratios for primary and secondary outcomes in the three propensity score-matched cohorts. CI, confidence interval; CRNM, clinically relevant non-major; GI, gastrointestinal; Pys., person-years; SE, systemic embolism. Table 1 Baseline characteristics, propensity-matched groups Dabigatran–rivaroxaban-matched cohort (n = 20 504) Dabigatran–apixaban-matched cohort (n = 20 826) Apixaban–rivaroxaban-matched cohort (n = 27 398) Dabigatran (n = 10 252) Rivaroxaban (n = 10 252) SMD Dabigatran (n = 10 413) Apixaban (n = 10 413) SMD Apixaban (n = 13 699) Rivaroxaban (n = 13 699) SMD Age Mean (SD) 70.9 (10.95) 70.9 (11.21) 0.004 70.6 (11.18) 70.6 (11.67) <0.001 72.7 (11.66) 72.7 (11.08) <0.001 Median 71 71 71 71 73 73 <65 years 2526 (24.6) 2614 (25.5) 2687 (25.8) 2794 (26.8) 2934 (21.4) 2786 (20.3) 65–74 years 3869 (37.7) 3748 (36.6) 3869 (37.2) 3759 (36.1) 4596 (33.5) 4805 (35.1) ≥75 years 3857 (37.6) 3890 (37.9) 3857 (37.0) 3860 (37.1) 6169 (45.0) 6108 (44.6) OAC dose Standard dose 6498 (63.4) 8115 (79.2) 6652 (63.9) 8514 (81.8) 10 508 (76.7) 10 362 (75.6) Reduced dose 3754 (36.6) 2137 (20.8) 3761 (36.1) 1899 (18.2) 3191 (23.3) 3337 (24.4) Male gender 6286 (61.3) 6313 (61.6) 0.005 6433 (61.8) 6447 (61.9) 0.003 7946 (58.0) 7943 (58.0) 0.000 Hypertension 6656 (64.9) 6628 (64.7) 0.006 6693 (64.3) 6641 (63.8) 0.010 9376 (68.4) 9288 (67.8) 0.014 Ischaemic heart disease 2107 (20.6) 2089 (20.4) 0.004 2119 (20.3) 2101 (20.2) 0.004 3050 (22.3) 3061 (22.3) 0.002 Vascular disease 743 (7.2) 757 (7.4) 0.005 743 (7.1) 720 (6.9) 0.009 1265 (9.2) 1262 (9.2) 0.001 Heart failure 2103 (20.5) 2100 (20.5) 0.001 2140 (20.6) 2079 (20.0) 0.015 3029 (22.1) 3043 (22.2) 0.002 Chronic kidney disease 245 (2.4) 258 (2.5) 0.008 245 (2.4) 255 (2.4) 0.006 657 (4.8) 627 (4.6) 0.010 Diabetes mellitus 1318 (13.2) 1352 (13.2) 0.010 1324 (12.7) 1294 (12.4) 0.009 1923 (14.0) 1887 (13.8) 0.008 Chronic lower respiratory tract diseases 1137 (11.1) 1122 (10.9) 0.005 1141 (11.0) 1128 (10.8) 0.004 1654 (12.1) 1632 (11.9) 0.005 Active cancer (diagnosis last 12 months) 769 (7.5) 770 (7.5) 0.000 770 (7.4) 773 (7.4) 0.001 1276 (9.3) 1263 (9.2) 0.003 History of stroke/SE 1341 (13.1) 1330 (13.0) 0.003 1356 (13.0) 1322 (12.7) 0.010 1860 (13.6) 1792 (13.1) 0.015 History of anaemia 456 (4.4) 447 (4.4) 0.004 458 (4.4) 432 (4.1) 0.012 801 (5.8) 757 (5.5) 0.014 History of bleeding 1142 (11.1) 1142 (11.1) 0.000 1144 (11.0) 1097 (10.5) 0.015 1723 (12.6) 1715 (12.5) 0.002 Use of antiplatelet drugs last

3.1) 1330 (13.0) 0.003 1356 (13.0) 1322 (12.7) 0.010 1860 (13.6) 1792 (13.1) 0.015 History of anaemia 456 (4.4) 447 (4.4) 0.004 458 (4.4) 432 (4.1) 0.012 801 (5.8) 757 (5.5) 0.014 History of bleeding 1142 (11.1) 1142 (11.1) 0.000 1144 (11.0) 1097 (10.5) 0.015 1723 (12.6) 1715 (12.5) 0.002 Use of antiplatelet drugs last 12 months 5109 (49.8) 5079 (49.5) 0.006 5125 (49.2) 5016 (48.2) 0.021 7312 (53.4) 7207 (52.6) 0.015 Use of NSAIDs last 12 months 2485 (24.2) 2467 (24.1) 0.004 2512 (24.1) 2492 (23.9) 0.004 3047 (22.2) 3148 (23.0) 0.018 Use of cholesterol lowering drugs 4603 (44.9) 4598 (44.8) 0.001 4629 (44.5) 4516 (43.4) 0.022 6356 (46.4) 6315 (46.1) 0.006 Mean CHA2DS2-VASc score (SD) 2.99 (1.73) 2.98 (1.71) 0.006 2.96 (1.74) 2.93 (1.72) 0.017 3.23 (1.74) 3.22 (1.71) 0.006 Mean HAS-BLED score (SD) 2.30 (1.14) 2.29 (1.12) 0.009 2.25 (1.15) 2.25 (1.16) 0.000 2.43 (1.15) 2.43 (1.12) 0.000 Values are expressed as numbers (percent), unless otherwise stated. NSAIDs, non-steroidal anti-inflammatory drugs; SD, standard deviation; SE, systemic embolism; SMD, absolute standardized mean difference. Dabigatran–rivaroxaban-matched cohort The median follow-up time was 18.6 months for dabigatran and 18.2 months for rivaroxaban. In the dabigatran group, stroke/SE occurred with an event rate of 1.84/100 person-years compared with 2.21/100 person-years in the rivaroxaban group [HR 0.88; 95% confidence interval (CI) 0.76–1.02]. A major bleeding event occurred at a rate of 1.40/100 person-years in the dabigatran group, and 1.93 in the rivaroxaban group (HR 0.75; 95% CI 0.64–0.88).

antagonist oral anticoagulant–warfarin comparisons Comparing each NOAC with warfarin, we found no significant differences in the adjusted HRs of stroke/SE for any NOAC compared with warfarin, while dabigatran and apixaban were both associated with lower risk of major bleeding (Supplementary material online, Table S4). Discussion In this study, we compared the risks of stroke or SE and major bleeding associated with use of dabigatran, rivaroxaban, and apixaban in a large nationwide cohort of anticoagulant-naïve patients with AF. In propensity score-matched analyses, we found no statistically significant differences in the risk of stroke or SE between NOACs, but dabigatran and apixaban were associated with significantly lower risk of major bleeding compared with rivaroxaban. The reduction of bleeding risk associated with dabigatran and apixaban was consistent for CRNM bleeding, major or CRNM bleeding, and intracranial bleeding. Dabigatran and rivaroxaban were associated with a significantly higher risk of GI bleeding compared with apixaban. Clinical trials and recent meta-analyses have shown that the NOACs are at least as effective as warfarin in stroke prevention and are associated with a similar or reduced risk of bleeding.4–6,18,19 In registry-based observational studies comparing the NOACs with warfarin, very similar results have been found.20–23 As the proportion of patients with AF being started on a NOAC instead of warfarin is increasing,24 knowledge of the comparative effectiveness and safety profiles of the different NOACs in clinical practice is needed.

sed observational studies comparing the NOACs with warfarin, very similar results have been found.20–23 As the proportion of patients with AF being started on a NOAC instead of warfarin is increasing,24 knowledge of the comparative effectiveness and safety profiles of the different NOACs in clinical practice is needed. Our study is one of very few studies designed to directly compare the effectiveness and safety of three individual NOACs in clinical practice. As treatment with NOACs is the standard of care in AF today,2 such a comparison seems more relevant for the practicing clinician. The NOACs were examined pairwise in PSM analyses. A strength of our study is the inclusion of all anticoagulant-naïve new users of a NOAC from a nationwide cohort; this should eliminate selection and participation bias often present in observational cohort studies. Furthermore, the follow-up times were longer and the number of patients included in the matched cohorts larger in our study compared with most previous studies.23,25

-naïve new users of a NOAC from a nationwide cohort; this should eliminate selection and participation bias often present in observational cohort studies. Furthermore, the follow-up times were longer and the number of patients included in the matched cohorts larger in our study compared with most previous studies.23,25 Our current findings are in line with similar studies.23,25–27 In a recent Danish study by Staerk et al.,27 including 31 522 patients with AF, multivariate Cox regression was chosen over PSM. In line with our findings, dabigatran and apixaban were associated with lower bleeding risk compared with rivaroxaban, but no significant differences were seen between the NOACs in terms of effectiveness. In another Danish study by Andersson et al.,25 including 12 638 new users of NOACs, PSM was performed, and no significant differences in associated risk of stroke/SE or major bleeding were found between NOACs. However, due to the low number of patients in each matched cohort, this study might have been underpowered. Similarly, Noseworthy et al.26 found no significant differences in effectiveness between the NOACs in their PSM cohorts, and both dabigatran and apixaban were associated with significantly lower bleeding risk compared with rivaroxaban. In the largest observational study to date, Lip et al.23 studied 285 292 patients pooled from the US Centers for Medicare and Medicaid Services Medicare data and four commercial claims databases in the USA (the ARISTOPHANES study). After PSM of patients with AF treated with a NOAC, apixaban was associated with significantly lower risk of both stroke or SE and major bleeding compared with dabigatran and rivaroxaban. Dabigatran compared with rivaroxaban was associated with a similar risk of stroke/SE but significantly lower risk of bleeding. A major limitation of the ARISTOPHANES study was the very short median follow-up time in all cohorts of just over 4 months. Another limitation involves the use of healthcare claims databases, necessitating Medicare or Medicaid eligibility for patient inclusion and relying on billing codes to define all baseline characteristics and outcomes. This increases risk of selection bias and loss to follow-up bias.

all cohorts of just over 4 months. Another limitation involves the use of healthcare claims databases, necessitating Medicare or Medicaid eligibility for patient inclusion and relying on billing codes to define all baseline characteristics and outcomes. This increases risk of selection bias and loss to follow-up bias. Our post hoc analysis comparing NOACs with warfarin were also generally in line with the results from similar real-world studies,20–23 showing non-significant differences in the risk of stroke/SE associated with NOACs, and significantly lower risks of major bleeding for both dabigatran and apixaban. Comparing our results with the RCTs,4–6 we did not find the reductions in stroke risk with dabigatran 150 mg and apixaban compared with warfarin that was shown in the RE-LY and ARISTOTLE trials.5,6 This has, however, been the case in several previous real-world studies.21–23 Minor discrepancies from the RCTs are to be expected, since these are not randomized comparisons. Despite adjustments, remaining unmeasured confounders will always exist. In the subgroup analyses performed in our study, significant interactions were seen between groups using NOACs as primary or secondary stroke prophylaxis. These findings are difficult to explain. Since they represent interactions based on subgroup analyses of non-randomized comparisons, they are most likely due to chance. The risks of stroke/SE and major bleeding in the cohorts rematched on standard and reduced doses were broadly consistent with the main findings.

laxis. These findings are difficult to explain. Since they represent interactions based on subgroup analyses of non-randomized comparisons, they are most likely due to chance. The risks of stroke/SE and major bleeding in the cohorts rematched on standard and reduced doses were broadly consistent with the main findings. Strengths and limitations There are fundamental differences between observational studies and RCTs, where the higher event rates often seen in registry studies reflect some of these differences.4–6,28,29 Inclusion of data into the nationwide registries is mandatory in Norway; this eliminates selection, participation, and recall bias. It also ensures a study population large enough for robust calculations. These advantages of nationwide registries are summarized in a recent position document from the European Heart Rhythm Association.30 The Norwegian system of general reimbursement of medical expenses for the treatment of serious and prolonged chronic illnesses ensures that all patients included in the study are in fact using OACs for AF, and not venous thromboembolism or any other condition; a challenge for similar studies based on registries where information on indication for treatment is unavailable.31,32

penses for the treatment of serious and prolonged chronic illnesses ensures that all patients included in the study are in fact using OACs for AF, and not venous thromboembolism or any other condition; a challenge for similar studies based on registries where information on indication for treatment is unavailable.31,32 A well-known limitation is that conventional multivariate regression, as well as PSM cannot control for unknown or unmeasurable confounders.33 In the total study population, before PSM was performed, patients starting rivaroxaban and apixaban were generally older and sicker than patients starting dabigatran (Supplementary material online, Table S3). It seems likely that the patients starting rivaroxaban and apixaban could also have other comorbidities or underlying factors that we have not taken into account, as well as a higher degree of frailty; an element which is difficult to measure in this type of study based on nationwide administrative registries, but which in this case likely is driving the estimates in favour of dabigatran.

ld also have other comorbidities or underlying factors that we have not taken into account, as well as a higher degree of frailty; an element which is difficult to measure in this type of study based on nationwide administrative registries, but which in this case likely is driving the estimates in favour of dabigatran. The events recorded were not adjudicated. There was also very likely a certain degree of miscoding and under-reporting of comorbidities and events. Despite nationwide inclusion of patients, because of demographics the study participants were still largely White northern Europeans. This may limit the generalizability of the results. Another limitation is that the registries do not supply information on relevant laboratory analyses such as estimated glomerular filtration rate, cardiac troponins, erythrocyte count, thrombocyte count, or liver enzymes; or other important patient characteristics such as body weight, lifestyle, or smoking habits.

ults. Another limitation is that the registries do not supply information on relevant laboratory analyses such as estimated glomerular filtration rate, cardiac troponins, erythrocyte count, thrombocyte count, or liver enzymes; or other important patient characteristics such as body weight, lifestyle, or smoking habits. Dabigatran was the first, rivaroxaban the second, and apixaban the third NOAC available in Norway, and all drugs were available in the whole study period. The proportion of patients starting on apixaban increased steadily throughout the study period (Supplementary material online, Table S3). Temporal changes in prescription patterns for NOACs might influence the number of events in each group. However, we found no significant differences between the NOACs regarding associated risk of stroke/SE; and dabigatran (the first NOACs on the market) and apixaban (the last NOAC on the market) were both associated with significantly lower risks of major bleeding compared with rivaroxaban (the second NOAC on the market). In addition, we created well-balanced cohorts in terms of risk factors; thus, it seems unlikely that temporal changes have played any important role for our results. To account for the approximately 6 months average shorter follow-up time for apixaban compared with dabigatran and rivaroxaban we performed a separate sensitivity analysis restricting the follow-up time to 12 months with results in line with the main analyses.

ral changes have played any important role for our results. To account for the approximately 6 months average shorter follow-up time for apixaban compared with dabigatran and rivaroxaban we performed a separate sensitivity analysis restricting the follow-up time to 12 months with results in line with the main analyses. Evaluation of the appropriateness of the dose prescribed (standard or reduced dose of NOAC) requires knowledge not only of patient age but also of serum creatinine and body weight. The variables serum creatinine and body weight are unfortunately not available from the nationwide registries in Norway, like in many other registries.23,25–27 Although we were unable to identify users of NOACs per label regarding dose, we have attempted to compensate for this by performing de novo propensity score estimation and matching within dosage groups. Furthermore, based on a recent study from UK, there are reasons to believe that the majority of AF patients are prescribed appropriate doses of NOACs; the UK study found between 75% and 85% of patients to be appropriately dosed.34 We studied drug exposure at the level of pharmacy dispensation and have no information on patient’s real intake of OAC. However, it is unlikely to expect any differences between groups in this respect. Due to the limitations of our study, the results should be interpreted with caution and need to be confirmed by findings from NOAC vs. NOAC RCTs. This is especially the case for the subgroup analyses, where we after careful consideration did not adjust for multiple comparisons (e.g. Bonferroni correction).

We studied drug exposure at the level of pharmacy dispensation and have no information on patient’s real intake of OAC. However, it is unlikely to expect any differences between groups in this respect. Due to the limitations of our study, the results should be interpreted with caution and need to be confirmed by findings from NOAC vs. NOAC RCTs. This is especially the case for the subgroup analyses, where we after careful consideration did not adjust for multiple comparisons (e.g. Bonferroni correction). Conclusion In this large registry-based study including 65 563 anticoagulant-naïve patients with AF initiating OAC therapy, we found no statistically significant differences in risk of stroke or SE between dabigatran, rivaroxaban, and apixaban, while both dabigatran and apixaban were associated with significantly lower risks of major bleeding compared with rivaroxaban. Funding This work was supported by the South-Eastern Norway Regional Health Authority. The study received additional support through the Bristol-Myers Squibb/Pfizer-sponsored European Investigator Initiated research Program (ERISTA) [Grant number 2016-ELI-0407].

Conclusion In this large registry-based study including 65 563 anticoagulant-naïve patients with AF initiating OAC therapy, we found no statistically significant differences in risk of stroke or SE between dabigatran, rivaroxaban, and apixaban, while both dabigatran and apixaban were associated with significantly lower risks of major bleeding compared with rivaroxaban. Funding This work was supported by the South-Eastern Norway Regional Health Authority. The study received additional support through the Bristol-Myers Squibb/Pfizer-sponsored European Investigator Initiated research Program (ERISTA) [Grant number 2016-ELI-0407]. Conflict of interest: O.-C.R. reports personal fees from Merck, Bayer, Boehringer Ingelheim, Novartis, and Novo Nordisk, outside the submitted work. C.J. reports personal fees from BMS/Pfizer and Bayer, outside the submitted work. W.G. reports grants and personal fees from Bayer, MSD, Novartis, and Amgen, outside the submitted work. S.H. reports personal fees from Bristol-Myers Squibb, Bayer, Boehringer Ingelheim, Pfizer, Merck, and Daiichi-Sankyo, outside the submitted work. Supplementary Material pvz086_Supplementary_Data Click here for additional data file.

Introduction Heart failure is a common and serious complication of acute myocardial infarction (MI), and frequently requires treatment with a diuretic to relieve symptoms and fluid retention. Several studies have demonstrated an interaction between diuretic-induced hypokalaemia and complex cardiac arrhythmias,1 and patients may present with both ventricular and supraventricular arrhythmias as well as cardiac arrest.2 Displacement of serum potassium influences resting membrane potentials in cardiomyocytes, which in turn cause cellular hyperpolarity as well as increased excitability and automaticity. Hypo- and hyperkalaemia following MI are associated with increased mortality,3,4 but it is important to examine whether specific potassium intervals within the normal range also set patients at risk or appear particularly safe. At present, few if any studies provide sufficient evidence regarding the specific cut-off values of serum potassium levels to identify patients with an increased mortality risk.5 Using Danish health registries, we aim to examine the possible relationships between serum potassium levels and short-term mortality in patients requiring loop diuretic treatment after acute MI.

Hypo- and hyperkalaemia following MI are associated with increased mortality,3,4 but it is important to examine whether specific potassium intervals within the normal range also set patients at risk or appear particularly safe. At present, few if any studies provide sufficient evidence regarding the specific cut-off values of serum potassium levels to identify patients with an increased mortality risk.5 Using Danish health registries, we aim to examine the possible relationships between serum potassium levels and short-term mortality in patients requiring loop diuretic treatment after acute MI. Methods Databases We used four health databases in this study; one containing hospitalizations, one containing medication dispensations, one containing birthday as well as date and cause of death, and one with blood test results. In Denmark, all residents have a unique and permanent civil registration number that enables linkage on an individual level among nationwide administrative registries. The Danish National Patient Register includes all hospitalizations in Denmark since 1978. At discharge, each hospitalization is coded with one primary and, if appropriate, one or more secondary diagnoses, according to the International Classification of Diseases. Until 1994, the 8th revision (ICD-8) was in use, and from 1994 onwards the 10th revision (ICD-10). The National Register for Medicinal Statistics includes all dispensed prescriptions from all Danish pharmacies since 1995 and is based on the Anatomical Therapeutic Chemical (ATC) System, and the accuracy has been validated.6 Blood test results were obtained from electronic registries of laboratory data, and we had access to data covering ∼1.5 million people. Date of death, date of birth, and vital status were obtained from the Danish Register of Causes of Death and the Central Personal Registry.

ystem, and the accuracy has been validated.6 Blood test results were obtained from electronic registries of laboratory data, and we had access to data covering ∼1.5 million people. Date of death, date of birth, and vital status were obtained from the Danish Register of Causes of Death and the Central Personal Registry. Study population The study population was selected according to following criteria: (i) a first MI episode, (ii) apparent heart failure as indicated by diuretic treatment after MI, and (iii) a measurement of serum potassium level within 90 days after the MI. The date of the first potassium measurement represented time 0 of our study. The patients were censored on 31 December 2012 or after 90 days of follow-up, whichever came first. The primary outcome was all-cause mortality.

treatment after MI, and (iii) a measurement of serum potassium level within 90 days after the MI. The date of the first potassium measurement represented time 0 of our study. The patients were censored on 31 December 2012 or after 90 days of follow-up, whichever came first. The primary outcome was all-cause mortality. First, patients with a first MI were selected from The Danish National Patient Register between 2004 and 2012. Myocardial infarction (ICD-10: I21) has been validated to be accurate.7,8 Secondly, we identified patients who received a prescription for loop diuretics (ATC: C03C) within 90 days using the National Register for Medicinal Statistics (prescription database). Administration of loop diuretics as a proxy for heart failure has already been associated with increased mortality without being correlated with estimated glomerular filtration rates among heart failure patients. This implies that use of loop diuretics is most likely due to cardiac causes rather than renal pathology.9–12 Based on several studies which have shown that heart failure following MI frequently develops in the first few months, patients were allowed a window period of 90 days (from MI) to redeem the loop diuretic prescription.13–16 Finally, we obtained one serum potassium level for each patient within 90 days following heart failure defined as a prescription of loop diuretics following an MI. Potassium measurements from the first day following MI were excluded.

ed a window period of 90 days (from MI) to redeem the loop diuretic prescription.13–16 Finally, we obtained one serum potassium level for each patient within 90 days following heart failure defined as a prescription of loop diuretics following an MI. Potassium measurements from the first day following MI were excluded. Potassium intervals and comorbidities Baseline demographics and clinical characteristics were compared among groups of patients stratified according to the following potassium levels: <3.5, 3.5–3.8, 3.9–4.2, 4.3–4.5, 4.6–5.0, 5.1–5.5, and >5.5 mmol/L. The serum potassium interval of 3.9–4.2 mmol/L was used as a reference for statistical analysis. Hypokalaemia was defined as potassium <3.5 mmol/L and hyperkalaemia as >5.0 mmol/L. Based on the van Deursen et al. study on comorbid illnesses in heart failure, we selected 11 covariates prior to the analysis including age, gender, diabetes, chronic obstructive pulmonary disease (J44), stroke (ICD-10: I61, I62, I63, I64), atrial fibrillation (ICD-10: I48), hypertension, drugs with effect on renin–angiotensin system, potassium-sparing diuretics, β-blockers, and potassium supplements.17 It is important to state that we did not analyse our population's drug administration after measured serum potassium.

stroke (ICD-10: I61, I62, I63, I64), atrial fibrillation (ICD-10: I48), hypertension, drugs with effect on renin–angiotensin system, potassium-sparing diuretics, β-blockers, and potassium supplements.17 It is important to state that we did not analyse our population's drug administration after measured serum potassium. Patients with a loop diuretic prescription or diagnosed with heart failure or chronic kidney disease before their first acute MI episode were excluded. Renal insufficiency and anaemia were identified within 1 week from serum potassium measurement. Renal insufficiency was defined as a serum creatinine level of >105 and >90 µmol/L for men and women, respectively (age between 18 and 70 years). Regarding patients over 70 years, renal insufficiency was defined as serum creatinine level >125 and >105 µmol/L for men and women, respectively.18–20 Anaemia was defined by a serum haemoglobin level of <8.1 mmol/L for men and <7.5 mmol/L for women.17 Diabetes was defined as administration of glucose-lowering drugs, and hypertension as a prescription of minimum two concomitant classes of antihypertensive drugs.21 All comorbidities were defined prior to the MI episode. Drug dispensations included in the analysis had to be prescribed in the period from the MI episode and serum potassium measurement.

Patients with a loop diuretic prescription or diagnosed with heart failure or chronic kidney disease before their first acute MI episode were excluded. Renal insufficiency and anaemia were identified within 1 week from serum potassium measurement. Renal insufficiency was defined as a serum creatinine level of >105 and >90 µmol/L for men and women, respectively (age between 18 and 70 years). Regarding patients over 70 years, renal insufficiency was defined as serum creatinine level >125 and >105 µmol/L for men and women, respectively.18–20 Anaemia was defined by a serum haemoglobin level of <8.1 mmol/L for men and <7.5 mmol/L for women.17 Diabetes was defined as administration of glucose-lowering drugs, and hypertension as a prescription of minimum two concomitant classes of antihypertensive drugs.21 All comorbidities were defined prior to the MI episode. Drug dispensations included in the analysis had to be prescribed in the period from the MI episode and serum potassium measurement. Statistical analysis Kaplan–Meier cumulative mortality curves were plotted for the seven preselected potassium intervals to illustrate trends in mortality over time. Cox proportional hazard regression models were used to determine the risk of death in heart failure patients with different potassium intervals, and adjusted for the defined covariates. To validate this statistical model, we tested the three Cox proportional hazard model assumptions: proportionality, linearity, and interactions.

tional hazard regression models were used to determine the risk of death in heart failure patients with different potassium intervals, and adjusted for the defined covariates. To validate this statistical model, we tested the three Cox proportional hazard model assumptions: proportionality, linearity, and interactions. The association of potassium with mortality was also assessed using restricted cubic splines with knots at the 10th, 25th, 50th, 75th, and 90th percentiles of potassium. Relative risks are presented as hazard ratios (HRs) with 95% confidence intervals (95% CIs). P-values of <0.05 were considered significant. Analyses were performed using SAS (version 9.4, SAS Institute, Cary, NC, USA) and R statistics (version 3.0.1, R Development Core Team). Results Baseline characteristics From Danish national registries we identified 2596 patients who were discharged after an MI with a prescription of loop diuretics, where a serum potassium measurement was available within 90 days following discharge. The baseline characteristics of the population reported according to each potassium level are presented in Table 1. This study population was characterized by advanced age, there were more men than women, and ∼40% of the patients had an increased creatinine level post-MI indicating renal insufficiency. Table 1 Baseline characteristics of patients (N = 2596)

ulation reported according to each potassium level are presented in Table 1. This study population was characterized by advanced age, there were more men than women, and ∼40% of the patients had an increased creatinine level post-MI indicating renal insufficiency. Table 1 Baseline characteristics of patients (N = 2596) K <3.5 mmol/L K: 3.5–3.8 mmol/L K: 3.9–4.2 mmol/L K: 4.3–4.5 mmol/L K: 4.6–5.0 mmol/L K: 5.1–5.5 mmol/L K >5.5 mmol/L Total Sex Female 77 (60.63%) 156 (51.83%) 341 (44.87%) 210 (39.4%) 247 (38.53%) 93 (53.45%) 35 (58.33%) 1159 (44.65%) Male 50 (39.37%) 145 (48.17%) 419 (55.13%) 323 (60.6%) 394 (61.47%) 81 (46.55%) 25 (41.67%) 1437 (55.35%) Age Mean ± SD 77.40 ± 10.77 75.85 ± 11.48 75.91 ± 11.29 74.42 ± 11.82 75.57 ± 11.36 78.90 ± 9.66 79.96 ± 9.17 75.88 ± 11.33 Renal insufficiency 41 (32.28%) 108 (35.88%) 264 (34.74%) 198 (37.15%) 325 (50.7%) 117 (67.24%) 52 (86.67%) 1105 (42.56%) Diabetes 19 (14.96%) 46 (15.28%) 123 (16.18%) 89 (16.7%) 159 (24.8%) 39 (22.41%) 21 (35%) 496 (19.11%) Anaemia 68 (53.54%) 155 (51.49%) 309 (40.66%) 248 (46.53%) 296 (46.18%) 91 (52.3%) 31 (51.67%) 1198 (46.15%) Atrial fibrillation 18 (14.17%) 37 (12.29%) 87 (11.45%) 50 (9.4%) 69 (10.8%) 22 (12.64%) 7 (11.67%) 290 (11.17%) COPD 15 (11.81%) 48 (15.95%) 95 (12.5%) 51 (9.57%) 68 (10.61%) 19 (10.92%) 8 (13.33%) 304 (11.71%) Stroke 26 (20.47%) 52 (17.28%) 122 (16.05%) 83 (15.6%) 109 (17%) 39 (22.41%) 14 (23.33%) 445 (17.14%) Hypertension 70 (55.12%) 165 (54.82%) 376 (49.47%) 231 (43.34%) 308 (48.05%) 94 (54.02%) 30 (50%) 1274 (49.1%) Potassium supplements 79 (62.2%) 229 (76.08%) 595 (78.29%) 418 (78.42%) 497 (77.53%) 115 (66.1%) 50 (83.33%) 1983 (76.39%) ACEIs/ARBs 24 (18.9%) 67 (22.26%) 232 (30.53%) 201 (37.71%) 229 (35.71%) 46 (26.44%) 14 (23.33%) 813 (31.32%) Beta-blockers 44 (34.65%) 111 (36.88%) 294 (38.68%) 269 (50.47%) 286 (44.62%) 77 (44.25%) 21 (35%) 1102 (42.45%) Potassium-sparing diuretics 8 (6.3%) 17 (5.65%) 55 (7.24%) 67 (12.57%) 102 (15.91%) 32 (18.4%) 14 (23.33%) 295 (11.36%) Loop diuretics strength (mg) Mean ± SD 39.69 ± 6.03 39.90 ± 6.14 40.80 ± 19.35 40.04 ± 5.85 40.16 ± 6.21 39.43 ± 6.24 41.00 ± 6.56 40.24 ± 11.66 Loop diuretics no. of packages redeemed Mean ± SD 1.01 ± 0.09 1.02 ± 0.17 1.03 ± 0.20 1.02 ± 0.17 1.01 ± 0.13 1.01 ± 0.11 1.02 ± 0.13 1.02 ± 0.16 Loop diuretics no.

5 (11.36%) Loop diuretics strength (mg) Mean ± SD 39.69 ± 6.03 39.90 ± 6.14 40.80 ± 19.35 40.04 ± 5.85 40.16 ± 6.21 39.43 ± 6.24 41.00 ± 6.56 40.24 ± 11.66 Loop diuretics no. of packages redeemed Mean ± SD 1.01 ± 0.09 1.02 ± 0.17 1.03 ± 0.20 1.02 ± 0.17 1.01 ± 0.13 1.01 ± 0.11 1.02 ± 0.13 1.02 ± 0.16 Loop diuretics no. of drugs in a package Mean ± SD 113.97 ± 63.27 116.96 ± 62.37 113.65 ± 65.90 108.26 ± 55.67 113.98 ± 60.59 112.57 ± 62.55 117.47 ± 63.67 113.04 ± 61.77 Day at loop diuretic reclaim Mean ± SD 24.94 ± 18.71 21.45 ± 17.69 19.86 ± 16.51 19.05 ± 16.09 21.13 ± 16.86 19.33 ± 14.49 20.05 ± 14.15 20.41 ± 16.63 Day at potassium measurement Mean ± SD 41.21 ± 21.91 43.17 ± 22.43 41.69 ± 22.07 42.96 ± 22.98 40.83 ± 22.49 38.54 ± 20.14 39.70 ± 21.38 41.63 ± 22.27 Death—90 days Alive 107 (84.25%) 260 (86.38%) 704 (92.63%) 490 (91.93%) 573 (89.39%) 147 (84.5%) 37 (61.67%) 2318 (89.3%) Deceased 20 (15.75%) 41 (13.62%) 56 (7.37%) 43 (8.07%) 68 (10.61%) 27 (15.5%) 23 (38.33%) 278 (10.7%) Data are presented as mean ± SD (age) or number of patients and percentage (all others). COPD, chronic obstructive pulmonary disease; ACE-I/ARBs, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker.

of drugs in a package Mean ± SD 113.97 ± 63.27 116.96 ± 62.37 113.65 ± 65.90 108.26 ± 55.67 113.98 ± 60.59 112.57 ± 62.55 117.47 ± 63.67 113.04 ± 61.77 Day at loop diuretic reclaim Mean ± SD 24.94 ± 18.71 21.45 ± 17.69 19.86 ± 16.51 19.05 ± 16.09 21.13 ± 16.86 19.33 ± 14.49 20.05 ± 14.15 20.41 ± 16.63 Day at potassium measurement Mean ± SD 41.21 ± 21.91 43.17 ± 22.43 41.69 ± 22.07 42.96 ± 22.98 40.83 ± 22.49 38.54 ± 20.14 39.70 ± 21.38 41.63 ± 22.27 Death—90 days Alive 107 (84.25%) 260 (86.38%) 704 (92.63%) 490 (91.93%) 573 (89.39%) 147 (84.5%) 37 (61.67%) 2318 (89.3%) Deceased 20 (15.75%) 41 (13.62%) 56 (7.37%) 43 (8.07%) 68 (10.61%) 27 (15.5%) 23 (38.33%) 278 (10.7%) Data are presented as mean ± SD (age) or number of patients and percentage (all others). COPD, chronic obstructive pulmonary disease; ACE-I/ARBs, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker. Loop diuretics were reclaimed, in average, 20th day after acute MI. Further on, it can be observed that a standard loop diuretic dosage, in the period post-MI until potassium measurement, was of 40.24 mg (±11.66) in a package containing 113.04 (±61.77) pills. A total of 76.4% of the patients received potassium supplement in the period following MI and before serum potassium measurement. In addition, 19.1% of the participants had a history of diabetes, and 46.1% a history of anaemia. Approximately 30% received drugs with effect on the renin–angiotensin system, and 40% received β-blockers. After 90 days, the mortality rates in the seven strata were 15.7, 13.6, 7.3, 8.1, 10.6, 15.5, and 38.3% respectively.

addition, 19.1% of the participants had a history of diabetes, and 46.1% a history of anaemia. Approximately 30% received drugs with effect on the renin–angiotensin system, and 40% received β-blockers. After 90 days, the mortality rates in the seven strata were 15.7, 13.6, 7.3, 8.1, 10.6, 15.5, and 38.3% respectively. Univariate analysis of survival A total of 278 (10.7%) patients died within the follow-up time of 90 days. Survival curves are illustrated in Figure 1. Univariate HRs according to potassium levels are shown in Figure 2. There was a significantly increased risk of death in hypo- and hyperkalaemic patients. Furthermore, patients with low normal potassium were associated with increased mortality (HR: 1.91, 95% CI: 1.28–2.86, P < 0.01]. Figure 1 Kaplan–Meier analysis of the survival probability among the different potassium intervals (N = 2596). Figure 2 All-cause mortality in heart failure patients following myocardial infarction stratified by potassium intervals. N = 2596 (90-day follow-up). Reference interval represented by the interval K: 3.9–4.2 mmol/L.

Univariate analysis of survival A total of 278 (10.7%) patients died within the follow-up time of 90 days. Survival curves are illustrated in Figure 1. Univariate HRs according to potassium levels are shown in Figure 2. There was a significantly increased risk of death in hypo- and hyperkalaemic patients. Furthermore, patients with low normal potassium were associated with increased mortality (HR: 1.91, 95% CI: 1.28–2.86, P < 0.01]. Figure 1 Kaplan–Meier analysis of the survival probability among the different potassium intervals (N = 2596). Figure 2 All-cause mortality in heart failure patients following myocardial infarction stratified by potassium intervals. N = 2596 (90-day follow-up). Reference interval represented by the interval K: 3.9–4.2 mmol/L. Multivariate analysis of survival The results of the multivariate analysis with potassium 3.9–4.2 mmol/L used as a reference are shown in Figure 3. After adjusting the model for age, sex, biologically relevant comorbidities and medication, the overall mortality remained significantly increased for patients with hypo- and hyperkalaemia. Furthermore, mortality also remained significantly increased for patients with potassium 3.5–3.8 and 4.6–5.0 mmol/L (HR: 1.84, 95% CI: 1.23–2.76, P < 0.01 and HR: 1.55, 95% CI: 1.09–2.22, P = 0.01, respectively). Covariates with significant impact on mortality are age, stroke, and drugs with effect on renin–angiotensin system (angiotensin-converting enzyme inhibitors and angiotensin receptor blocker). The results of the analysis of interaction between the predefined potassium intervals and creatinine are shown in Supplementary material online, Figure S1. Figure 3 All-cause mortality in heart failure patients following myocardial infarction stratified by potassium intervals. N = 2596 (90-day follow-up). Model adjusted for covariates. Reference interval represented by the interval K: 3.9–4.2 mmol/L. COPD, chronic obstructive pulmonary disease; ACE-I/ARBs, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker.

following myocardial infarction stratified by potassium intervals. N = 2596 (90-day follow-up). Model adjusted for covariates. Reference interval represented by the interval K: 3.9–4.2 mmol/L. COPD, chronic obstructive pulmonary disease; ACE-I/ARBs, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker. The U-shaped restricted cubic spline curve is shown in Figure 4, indicating that the higher the serum potassium level, the greater the mortality risk. However, the spline curve showed that the HR reached an almost constant increased value when serum potassium <2.7 mmol/L. Additionally, the spline curve indicates a difference in risk within the normal potassium ranges, where a potassium interval 3.9–4.5 mmol/L is associated with the lowest risk of death. Figure 4 Restricted cubic splines showing the adjusted hazard ratios for all-cause mortality as a function of potassium concentration. Knots at the 10th, 25th, 50th, 75th and 90th percentiles of potassium. Model adjusted for age, sex, COPD, stroke, AF, DM, hypertension, potassium supplement, ACEIs/ARBs, beta-blockers, and potassium-sparing diuretics (N = 2596). COPD, chronic obstructive pulmonary disease; AF, atrial fibrillation; DM, diabetes mellitus; ACE-I/ARBs, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker.

ed for age, sex, COPD, stroke, AF, DM, hypertension, potassium supplement, ACEIs/ARBs, beta-blockers, and potassium-sparing diuretics (N = 2596). COPD, chronic obstructive pulmonary disease; AF, atrial fibrillation; DM, diabetes mellitus; ACE-I/ARBs, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker. Discussion We examined the risk of death in patients receiving diuretics after MI depending on serum potassium levels. The main result of this study is that even mild deviation in serum potassium, is associated with increased mortality in patients with heart failure following an MI. It was not surprising that potassium levels outside the normal range (K: <3.5 and >5 mmol/L) were associated with an increased mortality risk. However, the novelty of this study was the association of low and high normal potassium (K: 3.5–3.8 and 4.6–5.0 mmol/L, respectively) with an increased mortality risk in heart failure patients following MI. Comparison with other studies MacDonald et al. examined the optimal potassium levels in cardiovascular patients through a meta-analysis of four studies. They found that it is desirable to avoid hypokalaemia, and that the recommended serum potassium level in heart failure and MI patients is 4.5–5.5 mmol/L. All of the studies in this meta-analysis included patients with chronic heart failure, and the majority assessed the impact of different treatment regimens on mortality with no direct focus on potassium homeostasis and outcome.22–28

e recommended serum potassium level in heart failure and MI patients is 4.5–5.5 mmol/L. All of the studies in this meta-analysis included patients with chronic heart failure, and the majority assessed the impact of different treatment regimens on mortality with no direct focus on potassium homeostasis and outcome.22–28 Cooper et al.25 demonstrated that patients administrated non-potassium-sparing diuretics suffered from an increased risk for arrhythmic death. In the studies of aldosterone antagonists in chronic heart failure, attempts were made to elucidate on the impact of treatment according to pre-treatment potassium levels.26,27 Eplerenone was superior in patients with low potassium, using a cut-off value of 4 mmol/L.26 This is foreseeable since patient with initial higher potassium is more likely to develop hyperkalaemia. However, the same relation could not be confirmed convincingly in a study of spironolactone.27 Nolan et al.28 provided retrospective data to suggest that a serum potassium level of <4.4 mmol/L was associated with an increased risk of sudden cardiac death. Overall, none of these four studies provide any direct evidence to select optimum levels of serum potassium in disease or health. Nevertheless, the review recommended a serum potassium level of 4.5–5.5 mmol/L as optimum for patients with heart failure and MI. In contrast to MacDonald's optimal serum potassium level of 4.5–5.5 mmol/L,5 our study showed a significantly increased risk of death in patients with serum potassium between 5.1 and 5.5 mmol/L (HR: 2.0, 95% CI: 1.25–3.18, P < 0.01).

m potassium level of 4.5–5.5 mmol/L as optimum for patients with heart failure and MI. In contrast to MacDonald's optimal serum potassium level of 4.5–5.5 mmol/L,5 our study showed a significantly increased risk of death in patients with serum potassium between 5.1 and 5.5 mmol/L (HR: 2.0, 95% CI: 1.25–3.18, P < 0.01). A substudy from the Digitalis Investigation Group (DIG) trial confirmed that serum potassium levels <4 mmol/L were associated with increased mortality in heart failure patients.29 This is in agreement with our findings of an association between mortality and K <3.9 mmol/L as are findings from another study where serum potassium levels >5.0 mmol/L predicted short-term mortality (12 weeks).30 Several studies have confirmed a link between low potassium and both ventricular arrhythmias and atrial fibrillation and survival.28,31–34 In a study of atrial fibrillation risk assessment in relation to potassium, there were 11.6% of patients with atrial fibrillation and potassium was measured only at baseline.32 This is similar to our study where 11.2% of the patients were diagnosed with atrial fibrillation.

and atrial fibrillation and survival.28,31–34 In a study of atrial fibrillation risk assessment in relation to potassium, there were 11.6% of patients with atrial fibrillation and potassium was measured only at baseline.32 This is similar to our study where 11.2% of the patients were diagnosed with atrial fibrillation. Treatment of acute heart failure As mentioned in the Methods section, the population is selected between year 2004 and 2012. Throughout this period, the pharmacological therapy could be marked by stepwise changes in the international guidelines for heart failure management. Therefore, the baseline characteristics may highlight lower numbers of β-blockers and angiotensin-converting enzyme inhibitors than probably expected and stated in the current guidelines. Though it is essential to have in mind that this study does not observe the pharmaceutical adjustments after the potassium measurement. Thus, some patients may have had only transient heart failure with no further need for chronic heart failure medication. Study limitations This study is not a randomized controlled trial. However, with help of databases, we were able to extract information on comorbid illnesses and concomitant medication use. All factors that were considered possible confounders were included in the Cox multivariable analysis.

Treatment of acute heart failure As mentioned in the Methods section, the population is selected between year 2004 and 2012. Throughout this period, the pharmacological therapy could be marked by stepwise changes in the international guidelines for heart failure management. Therefore, the baseline characteristics may highlight lower numbers of β-blockers and angiotensin-converting enzyme inhibitors than probably expected and stated in the current guidelines. Though it is essential to have in mind that this study does not observe the pharmaceutical adjustments after the potassium measurement. Thus, some patients may have had only transient heart failure with no further need for chronic heart failure medication. Study limitations This study is not a randomized controlled trial. However, with help of databases, we were able to extract information on comorbid illnesses and concomitant medication use. All factors that were considered possible confounders were included in the Cox multivariable analysis. Limitations of this study are represented by the lack of information regarding the cause of death. This was not possible due to uncertainty regarding cause of death registry. Based on previous diagnosis, one patient can be attributed one or more causes of death in situations where autopsy is missing.

Study limitations This study is not a randomized controlled trial. However, with help of databases, we were able to extract information on comorbid illnesses and concomitant medication use. All factors that were considered possible confounders were included in the Cox multivariable analysis. Limitations of this study are represented by the lack of information regarding the cause of death. This was not possible due to uncertainty regarding cause of death registry. Based on previous diagnosis, one patient can be attributed one or more causes of death in situations where autopsy is missing. Last but not least, we did not differentiate between the various types of MI, which means that our population can also encompass patients with type two myocardial infarction. The authors intended to identify serum creatine kinase-MB and troponin-T in order to acknowledge the severity of the MI. We were unable to provide this information due to an increased number of missing values. Conclusion Potassium levels outside the interval 3.9–4.5 mmol/L were associated with a substantial risk of death in patients requiring diuretic treatment after an MI.

Last but not least, we did not differentiate between the various types of MI, which means that our population can also encompass patients with type two myocardial infarction. The authors intended to identify serum creatine kinase-MB and troponin-T in order to acknowledge the severity of the MI. We were unable to provide this information due to an increased number of missing values. Conclusion Potassium levels outside the interval 3.9–4.5 mmol/L were associated with a substantial risk of death in patients requiring diuretic treatment after an MI. Clinical implications This study indicates that low normal potassium levels and mild hyperkalaemia may also be associated with increased mortality, which suggests that a narrower normal interval could possibly improve outcome in heart failure patients following MI. Myocardial changes in heart failure following MI influence potassium homeostasis, which makes standard potassium levels difficult to use in assessment of heart failure patients, as a pathological myocardium does not function as in normal subjects. Potassium imbalance can quickly occur in heart failure patients either due to medication or disease pathophysiology or both. For this reason, closer monitoring of serum potassium in acute heart failure patients would probably be relevant to improve survival. Supplementary material Supplementary material is available at European Heart Journal online.

Clinical implications This study indicates that low normal potassium levels and mild hyperkalaemia may also be associated with increased mortality, which suggests that a narrower normal interval could possibly improve outcome in heart failure patients following MI. Myocardial changes in heart failure following MI influence potassium homeostasis, which makes standard potassium levels difficult to use in assessment of heart failure patients, as a pathological myocardium does not function as in normal subjects. Potassium imbalance can quickly occur in heart failure patients either due to medication or disease pathophysiology or both. For this reason, closer monitoring of serum potassium in acute heart failure patients would probably be relevant to improve survival. Supplementary material Supplementary material is available at European Heart Journal online. Funding Research grant from Aalborg University Hospital, Aalborg, Denmark. Funding to pay the Open Access publication charges for this article was provided by Department of Cardiology, Aalborg University Hospital and Department of Epidemiology, Aalborg University Hospital. Conflict of interest: none declared. Supplementary Material Supplementary material online, Figure S1

Introduction Percutaneous catheter-based interventions became a critically important part of treatment in modern cardiology, improving quality of life as well as saving many lifes. Due to the introduction of foreign materials to the circulation (either temporarily or permanently) and due to a certain damage to the endothelium or endocardium, the risk of thrombotic complications is substantial and thus some degree of antithrombotic therapy is needed during all these procedures. The intensity (dosage, combination, and duration) of periprocedureal antithrombotic treatment largely varies based on the type of procedure, clinical setting, and comorbidities. The aim of this manuscript is to review the current therapeutic approach (including guidelines whenever available) and to discuss the existing gaps of evidence and unresolved questions.

ation) of periprocedureal antithrombotic treatment largely varies based on the type of procedure, clinical setting, and comorbidities. The aim of this manuscript is to review the current therapeutic approach (including guidelines whenever available) and to discuss the existing gaps of evidence and unresolved questions. Percutaneous coronary interventions for acute myocardial infarction The antithrombotic treatment in patients with ST-elevation myocardial infarction (STEMI) as well as in those with ongoing myocardial ischaemia1 in the absence of STE should include three classes of drugs: (i) acetylsalicylic acid (ASA), (ii) intravenous anticoagulant, and (iii) P2Y12 inhibitor. These agents should be given as soon as the diagnosis is certain, frequently in the pre-hospital phase when the clinical presentation and electrocardiogram are typical and diagnostic, i.e. before coronary angiography.2,3 Controversy exists only in the optimal timing of P2Y12 inhibitors—the evidence for their upfront (pre-hospital) use is still lacking for the primary PCI strategy. It is even more controversial for the thrombolytic strategy.

sentation and electrocardiogram are typical and diagnostic, i.e. before coronary angiography.2,3 Controversy exists only in the optimal timing of P2Y12 inhibitors—the evidence for their upfront (pre-hospital) use is still lacking for the primary PCI strategy. It is even more controversial for the thrombolytic strategy. Pre-/periprocedural oral antiplatelet therapy An oral loading dose of ASA 150–300 mg (or i.v. 80–150 mg) should be given to all patients. The preferred P2Y12 inhibitors are prasugrel (60 mg p.o. loading dose) or ticagrelor (180 mg p.o. loading dose).4,5 In the STEMI subgroup of the TRITON–TIMI 38 trial, prasugrel was superior to clopidogrel (primary endpoint prasugrel 10.0% vs. clopidogrel 12.4%), without a significant increase in non-CABG-related bleeding (2.4% vs. 2.1%). There was a lower risk of stent thrombosis (1.6% vs. 2.8%), as well as of cardiovascular mortality (1.4% vs. 2.4%)6 in favour of prasugrel. Prasugrel is contraindicated in patients with prior stroke or TIA and not recommended for patients aged 75 years or more. In patients with body weight <60 kg, a maintenance dose of 5 mg is recommended. In the STEMI subgroup of the PLATO trial, ticagrelor was superior to clopidogrel (primary endpoint 9.4% vs. 10.8%)7 without higher risk of bleeding (TIMI non-CABG major 2.5% vs. 2.2%) but with a trend towards a lower risk of cardiovascular mortality at 1 year (4.7% vs. 5.4%).

Pre-/periprocedural oral antiplatelet therapy An oral loading dose of ASA 150–300 mg (or i.v. 80–150 mg) should be given to all patients. The preferred P2Y12 inhibitors are prasugrel (60 mg p.o. loading dose) or ticagrelor (180 mg p.o. loading dose).4,5 In the STEMI subgroup of the TRITON–TIMI 38 trial, prasugrel was superior to clopidogrel (primary endpoint prasugrel 10.0% vs. clopidogrel 12.4%), without a significant increase in non-CABG-related bleeding (2.4% vs. 2.1%). There was a lower risk of stent thrombosis (1.6% vs. 2.8%), as well as of cardiovascular mortality (1.4% vs. 2.4%)6 in favour of prasugrel. Prasugrel is contraindicated in patients with prior stroke or TIA and not recommended for patients aged 75 years or more. In patients with body weight <60 kg, a maintenance dose of 5 mg is recommended. In the STEMI subgroup of the PLATO trial, ticagrelor was superior to clopidogrel (primary endpoint 9.4% vs. 10.8%)7 without higher risk of bleeding (TIMI non-CABG major 2.5% vs. 2.2%) but with a trend towards a lower risk of cardiovascular mortality at 1 year (4.7% vs. 5.4%). In a pooled analysis of 48 599 patients, novel P2Y12 inhibitors prasugrel or ticagrelor have been associated with a mortality benefit and no significant excess of major bleeding among STEMI patients.8 Importantly, the more potent agents (prasugrel and ticagrelor) should not be used in patients with prior haemorrhagic stroke or with moderate-to-severe liver disease.

P2Y12 inhibitors prasugrel or ticagrelor have been associated with a mortality benefit and no significant excess of major bleeding among STEMI patients.8 Importantly, the more potent agents (prasugrel and ticagrelor) should not be used in patients with prior haemorrhagic stroke or with moderate-to-severe liver disease. When neither of these agents is available (or if they are contraindicated), clopidogrel 600 mg p.o. should be given instead9 to fulfil the requirement for dual antiplatelet therapy (DAPT). Intravenous antiplatelet therapy Trials (mostly using abciximab) performed before the era of thienopyridines pre-loading documented clinical benefits from GP IIb/IIIa inhibitors as adjunct to primary PCI performed with UFH including a significant 1-year survival benefit that was revealed in a meta-analysis of GP IIb/IIIa inhibitors with abciximab.10 However, the benefits of GPIIb/IIIa inhibitors in the era of potent novel P2Y12 inhibitors are questionable and there is a high likelihood of bleeding complications when four antithrombotic agents (aspirin, P2Y12 inhibitor, injectable anticoagulant, and GPIIb/IIIa inhibitor) are be combined simultaneously. In the event of angiographic evidence of large thrombus, slow- or no-reflow, and other thrombotic complications, use of GP IIb/IIIa inhibitors as bail-out therapy appears reasonable, although this has not been tested in a randomized trial.

Intravenous antiplatelet therapy Trials (mostly using abciximab) performed before the era of thienopyridines pre-loading documented clinical benefits from GP IIb/IIIa inhibitors as adjunct to primary PCI performed with UFH including a significant 1-year survival benefit that was revealed in a meta-analysis of GP IIb/IIIa inhibitors with abciximab.10 However, the benefits of GPIIb/IIIa inhibitors in the era of potent novel P2Y12 inhibitors are questionable and there is a high likelihood of bleeding complications when four antithrombotic agents (aspirin, P2Y12 inhibitor, injectable anticoagulant, and GPIIb/IIIa inhibitor) are be combined simultaneously. In the event of angiographic evidence of large thrombus, slow- or no-reflow, and other thrombotic complications, use of GP IIb/IIIa inhibitors as bail-out therapy appears reasonable, although this has not been tested in a randomized trial. The FINESSE study11 randomized STEMI patients to upstream abciximab at the first medical contact vs. in-cath-lab abciximab and found no significant effect on the primary endpoint (death, recurrent myocardial infarction, and heart failure), but significantly increased bleeding risk after upstream abciximab.

In the event of angiographic evidence of large thrombus, slow- or no-reflow, and other thrombotic complications, use of GP IIb/IIIa inhibitors as bail-out therapy appears reasonable, although this has not been tested in a randomized trial. The FINESSE study11 randomized STEMI patients to upstream abciximab at the first medical contact vs. in-cath-lab abciximab and found no significant effect on the primary endpoint (death, recurrent myocardial infarction, and heart failure), but significantly increased bleeding risk after upstream abciximab. Cangrelor, an intravenous rapidly acting P2Y12 inhibitor (dose 30 μg/kg bolus followed by infusion of 4 μg/kg/min), was compared with a 600-mg loading dose of clopidogrel either before or early after PCI in patients with ACS undergoing PCI in The CHAMPION PLATFORM and PCI studies. A minor benefit from cangrelor was observed, however, it is not known whether this benefit would apply also if cangrelor would be compared with prasugrel or ticagrelor.

ith a 600-mg loading dose of clopidogrel either before or early after PCI in patients with ACS undergoing PCI in The CHAMPION PLATFORM and PCI studies. A minor benefit from cangrelor was observed, however, it is not known whether this benefit would apply also if cangrelor would be compared with prasugrel or ticagrelor. Injectable anticoagulants Heparin Despite the lack of large randomized trials, unfractionated heparin (UFH) remains the cornerstone of anticoagulation treatment in STEMI patients planned to undergo primary PCI. Several other drugs have been compared with heparin, but none of them was proven to be superior. There is however one critically important issue in heparin treatment: the dose MUST be adapted to the patient body weight: if no GPIIb/IIIa inhibitors are planned (what is the current routine practice in most centres), the dose of UFH should be 70–100 units kg−1 (lower dose preferred in elderly patients especially fragile females with low body weight). A frequent mistake in the real life practice is the use of an arbitrary UFH dose (e.g. 5000 or 10 000 units) for all patients, irrespective of body weight. An arbitrary 5000 unit dose will certainly be ineffective for a 95-kg middle-aged smoker and on the other hand arbitrary 10 000 units may be an extremely dangerous overdose (risk of intracranial bleeding) for an elderly 55 kg female.

ry UFH dose (e.g. 5000 or 10 000 units) for all patients, irrespective of body weight. An arbitrary 5000 unit dose will certainly be ineffective for a 95-kg middle-aged smoker and on the other hand arbitrary 10 000 units may be an extremely dangerous overdose (risk of intracranial bleeding) for an elderly 55 kg female. Bivalirubin Bivalirubin was assessed in several randomized trials. In the HORIZONS-AMI trial,12 bivalirudin alone (with bail-out GP IIb/IIIa inhibitors in 7.2% of patients) was superior to combined therapy with UFH plus systematic GP IIb/IIIa inhibitor (mostly abciximab). However, the net adverse clinical endpoint (9.2% vs. 12.1%) included major bleeding (4.9% vs. 8.3%). Thus, it is not surprising that combination of two potent antithrombotic drugs was more harmfull than a single drug. There was no significant difference in the ischaemic endpoints, even there was a higher incidence of stent thrombosis in the bivalirudin group (1.3% vs. 0.3%). The same problem was in the design of the EUROMAX trial,13 comparing a single drug strategy (pre-hospital bivalirudin) vs. a combination strategy (heparin with optional—69% patients—GPIIb/IIIa inhibitors). The primary endpoint (death or non-CABG major bleeding at 30 days) again included bleeding (2.6% vs. 6.0%) and was significantly lower with pre-hospital administration of bivalirudin (5.1% vs. 8.5%). Again, similarly to the HORIZONS-AMI trial, there were no differences in ischaemic endpoints: death (2.9% vs. 3.1%), stent thrombosis (1.6% vs. 0.5%), and re-infarction (1.7% vs. 0.9%).

) again included bleeding (2.6% vs. 6.0%) and was significantly lower with pre-hospital administration of bivalirudin (5.1% vs. 8.5%). Again, similarly to the HORIZONS-AMI trial, there were no differences in ischaemic endpoints: death (2.9% vs. 3.1%), stent thrombosis (1.6% vs. 0.5%), and re-infarction (1.7% vs. 0.9%). The HEAT-PCI study14 compared bivalirudin vs. UHF with similar rates (15%) of GPIIb/IIIa inhibitors in both arms. The study better represents contemporary practice (restriction of GP IIb/IIIa inhibitors to bail-out situations, the use of novel P2Y12 inhibitors, radial approach and drug-eluting stent, DES, implantation). The primary efficacy endpoint (all-cause mortality, stroke, recurrent infarction, and unplanned target lesion revascularization) was higher in the bivalirudin than in the UFH group (8.7% vs. 5.7%) including an increase in stent thrombosis (3.4% vs. 0.9%), but no significant difference in mortality (5.1% vs. 4.3%). The primary safety outcome (major BARC 3–5 bleeding) was 3.5% in the bivalirudin group vs. 3.1% in the UFH group.

revascularization) was higher in the bivalirudin than in the UFH group (8.7% vs. 5.7%) including an increase in stent thrombosis (3.4% vs. 0.9%), but no significant difference in mortality (5.1% vs. 4.3%). The primary safety outcome (major BARC 3–5 bleeding) was 3.5% in the bivalirudin group vs. 3.1% in the UFH group. Thus, the entire treatment benefit of bivalirudin demonstrated in HORIZONS-AMI and EUROMAX trials was caused by the study design (low use of GPIIb/IIIa inhibitors in the bivalirudin arm) and cannot answer the question whether bivalirudin is superior to heparin or vice versa. The results of HEAT-PCI study suggest, that heparin may be even superior to bivalirudin when the same rate of GPIIb/IIIa inhibition is used. A problem with bivalirudin is also that results were better in subject at low risk (no troponin) but not in those with high troponin (high risk); therefore, bivalirudin seems to provide no advantages high-risk patients.

hat heparin may be even superior to bivalirudin when the same rate of GPIIb/IIIa inhibition is used. A problem with bivalirudin is also that results were better in subject at low risk (no troponin) but not in those with high troponin (high risk); therefore, bivalirudin seems to provide no advantages high-risk patients. Enoxaparin Enoxaparin (0.5 mg/kg i.v. bolus followed by subcutaneous treatment, dose adjustment to impaired renal function is essential) was compared with UFH in the ATOLL trial. The primary composite endpoint (30-day death, complication of myocardial infarction, procedural failure, and major bleeding) was not significantly different, but secondary endpoints suggested possible benefit from enoxaparin. In the per-protocol analysis, enoxaparin was superior to UFH in reducing mortality (RR 0.36) and major bleedings (RR 0.46) in patients undergoing primary PCI. Based on these considerations, enoxaparin may be considered as an alternative to UFH in primary PCI.15 Fondaparinux Fondaparinux in the context of primary PCI is potentially harmfull (risk of catheter thrombosis) and is therefore not recommended.16

Enoxaparin Enoxaparin (0.5 mg/kg i.v. bolus followed by subcutaneous treatment, dose adjustment to impaired renal function is essential) was compared with UFH in the ATOLL trial. The primary composite endpoint (30-day death, complication of myocardial infarction, procedural failure, and major bleeding) was not significantly different, but secondary endpoints suggested possible benefit from enoxaparin. In the per-protocol analysis, enoxaparin was superior to UFH in reducing mortality (RR 0.36) and major bleedings (RR 0.46) in patients undergoing primary PCI. Based on these considerations, enoxaparin may be considered as an alternative to UFH in primary PCI.15 Fondaparinux Fondaparinux in the context of primary PCI is potentially harmfull (risk of catheter thrombosis) and is therefore not recommended.16 Elective percutaneous coronary interventions for stable coronary artery disease Ad hoc percutaneous coronary intervention Most patients with stable coronary artery disease nowadays undergo elective coronary angiography immediately followed by ad hoc PCI. In such situation, pretreatment with aspirin is widely used (usually due to a known diagnosis of coronary artery disease and not specifically due to the diagnostic angiography) and possibly is appropriate (albeit was never tested in a randomized trial). Anticoagulation with UFH (i.v. bolus of 70–100 U/kg) remains the standard anticoagulant treatment for elective PCI.3 Heparin is usually given in the cath-lab in two separate doses: initial small dose at the beginning of diagnostic angiography and second dose after the decision for ad hoc PCI is taken. The total UFH dose should be ALLWAYS calculated per the patient body weight: 70–100 units kg−1 (see also the previous chapter). The second antiplatelet drug (P2Y12 inhibitor) is usually added in the cath-lab just prior to PCI, i.e. between angiography and PCI.

and second dose after the decision for ad hoc PCI is taken. The total UFH dose should be ALLWAYS calculated per the patient body weight: 70–100 units kg−1 (see also the previous chapter). The second antiplatelet drug (P2Y12 inhibitor) is usually added in the cath-lab just prior to PCI, i.e. between angiography and PCI. Planned elective percutaneous coronary intervention Patients with known coronary angiography scheduled for elective PCI should be pretreated with DAPT at least few hours before the procedure and UFH should be used in the way described above as well. In patients not using any chronic antiplatelet therapy, the oral loading dose of ASA should be 150–300 mg (or 80–150 mg i.v.) and clopidogrel loading dose 300–600 mg.17–19 In patients on chronic aspirin and/or clopidogrel therapy, the loading dose before an elective procedure is not needed. There is no evidence of benefit for systematic clopidogrel pre-loading before diagnostic coronary angiography in SCAD.20 Recent trials did not demonstrate additional benefit from GP IIb/IIIa inhibitors after a clopidogrel loading dose of 600 mg.21–23 Anecdotal experience, however, suggests that GP IIb/IIIa inhibitors may be beneficial in ‘bail-out’ situations (intraprocedure thrombus formation, slow flow, and threatened vessel closure).24

Recent trials did not demonstrate additional benefit from GP IIb/IIIa inhibitors after a clopidogrel loading dose of 600 mg.21–23 Anecdotal experience, however, suggests that GP IIb/IIIa inhibitors may be beneficial in ‘bail-out’ situations (intraprocedure thrombus formation, slow flow, and threatened vessel closure).24 Percutaneous coronary interventions in patients with atrial fibrillation Approximately 10% of patients undergoing PCI have another indication for long-term oral anticoagulation (OAC)—most frequently concomitant atrial fibrillation. There is an ongoing debate about the optimal antithrombotic medication in these patients theoretically requiring triple therapy: OAC permanently and DAPT for 1 year. In practice, the best approach is individual decision based on the concrete bleeding risk vs. stent thrombosis risk. Patients with increased bleeding risk should receive triple therapy during the first month after stent implantation followed by long-term dual therapy (OAC + clopidogrel or OAC + aspirin). Patients at low-bleeding risk may receive triple therapy up to 6 months, followed by long-term OAC + aspirin.

tent thrombosis risk. Patients with increased bleeding risk should receive triple therapy during the first month after stent implantation followed by long-term dual therapy (OAC + clopidogrel or OAC + aspirin). Patients at low-bleeding risk may receive triple therapy up to 6 months, followed by long-term OAC + aspirin. Interventions for structural heart disease Structural heart interventions are a heterogeneous mixture of usually elective procedures ranging from the technically simple and short patent foramen ovale closure to long and complex interventions on mitral valve. Most of structural heart interventions involve rather large devices. These devices are typically metallic (stainless steel and nitinol are the most common); Dacron type polyester fabric to promote tissue growth or pericardial tissue made valve prosthesis are often present.

ong and complex interventions on mitral valve. Most of structural heart interventions involve rather large devices. These devices are typically metallic (stainless steel and nitinol are the most common); Dacron type polyester fabric to promote tissue growth or pericardial tissue made valve prosthesis are often present. Intravenous heparin is the dominant periprocedural anticoagulant because of familiarity to all operators, availability of antidote and low cost. Level of anticoagulation can be adjusted according to activated clotting time (ACT). However, the optimal target ACT is mostly not clear. Intriguingly, one single centre study elegantly demonstrated abnormal baseline ACT values prior to transcatheter aortic valve implantation (TAVI) in typical elderly and frail population and heparin dosing adjustment lead to less bleeding.25 Access site bleeding is obviously more common after arterial puncture than venous one; it is clear from TAVI data that arterial bleeding complications lead to a significant increase in early mortality. Many structural interventions involve catheter manipulation of right and left atria (i.e. thin-walled structures) with 1–2% risk of perforation and resulting cardiac tamponade. On the other hand, the longer procedure duration and the slower blood circulation around catheters both increase the risk of thrombus formation with possible embolization leading to disabling stroke or other organ embolization. Reversal of heparin activity with protamine is generally not recommended but can be very useful in case of bleeding. Bivalirudin has been compared with heparin in a randomized BRAVO 3 study of TAVI. There was no reduction in bleeding and heparin remains the standard of periprocedural care.26

r other organ embolization. Reversal of heparin activity with protamine is generally not recommended but can be very useful in case of bleeding. Bivalirudin has been compared with heparin in a randomized BRAVO 3 study of TAVI. There was no reduction in bleeding and heparin remains the standard of periprocedural care.26 Patients on OAC have this therapy interrupted for the procedure to minimize bleeding complications. Bridging with unfractionated or low-molecular heparin should be individualized based on every patient risk of bleeding and thrombosis.27 Suitable timing of OAC restart after the procedure is not well defined and is probably best left at discretion of attending physician. The role of new oral anticoagulants (NOACs) for structural heart interventions is not yet defined; dabigatran caused harm in patients with mechanical heart valves28 and ongoing Atlantis study currently evaluates apixaban after TAVI. Antiplatelet therapy is commonly prescribed just before and continued after structural heart interventions with the aim to prevent thrombotic complications until endothelization of implanted device is completed. Only aspirin and clopidogrel have been studied in this setting. The timing is empirical and any recommendations are based on expert consensus. Even a systematic review of antiplatelet and anticoagulation medication after TAVI did not provide any clear conclusions except the need for larger studies.29 Atrial fibrillation can occur in the post-operative period and could be one of the leading causes of stroke during the first 30 days after procedure.

consensus. Even a systematic review of antiplatelet and anticoagulation medication after TAVI did not provide any clear conclusions except the need for larger studies.29 Atrial fibrillation can occur in the post-operative period and could be one of the leading causes of stroke during the first 30 days after procedure. Table 1 provides summary of the most common structural heart interventions and data extracted from the major studies. For less common interventions, the scientific evidence is even more difficult to obtain due to small numbers. Table 1 Comparison of the most common structural heart interventions from antithrombotic therapy perspective

summary of the most common structural heart interventions and data extracted from the major studies. For less common interventions, the scientific evidence is even more difficult to obtain due to small numbers. Table 1 Comparison of the most common structural heart interventions from antithrombotic therapy perspective Procedure Typical access Main catheter size (French) Procedure time (min) Left atrium manipulation Atrial fibrillation (%) Heparin Guideline or IFU target ACT (s) Authors target ACT (s) Periprocedural stroke or embolism (%) Aspirin (months) Clopidogrel (months) Warfarin new indication (months) NOAC (months) References PTMV Venous 9–12 40–50 Yes Frequent Yes None 300 0.5–5 – – – To be tested 6–8 PFO closure Venous 7–9 10–50 Yes Excluded in PC and Respect trials Yes 200 Not measured Very low 6 to 24 1–6 – – 9,10 ASD closure Venous 7–12 65 Yes 3–5 Yes 200 Not measured or 250 Very low 6 – – – 11,12 LAA occlusion Venous 12 30–60 Yes Always Yes 250 300 2–3 Lifelong 0–6 0–1.5 – 13,14 TAVI 2× arterial and 1× venous 18 60–133 No 33–47 Yes 250 250 4.6–6.7 Lifelong 3–6 – Currently tested 15–19 MitraClip Venous 24 100 Yes 34–68 Yes 250 300 0.7–2.1 6 1 – To be tested 20–22 PTMV, percutaneous transvenous mitral valvuloplasty; PFO, patent foramen ovale; ASD, atrial septal defect; LAA, left atrial appendage; TAVI, transcatheter aortic valve implantation; IFU, instructions for use; ACT, activated clotting time; NOAC, new oral anticoagulant.

34–68 Yes 250 300 0.7–2.1 6 1 – To be tested 20–22 PTMV, percutaneous transvenous mitral valvuloplasty; PFO, patent foramen ovale; ASD, atrial septal defect; LAA, left atrial appendage; TAVI, transcatheter aortic valve implantation; IFU, instructions for use; ACT, activated clotting time; NOAC, new oral anticoagulant. Electronic device implantation As with any small surgery, also implantation of cardiac implantable electronic devices (CIED) is also associated with a risk of bleeding. Hematomas following CIED implantation are quite frequent (2.9–9.5% of the cases). Although bleeding following CIED implantation is typically small and only very rarely life threatening, it prolongs hospitalization, increases the costs and surgical haematoma evacuation is associated with 15 times higher risk of infection.

matomas following CIED implantation are quite frequent (2.9–9.5% of the cases). Although bleeding following CIED implantation is typically small and only very rarely life threatening, it prolongs hospitalization, increases the costs and surgical haematoma evacuation is associated with 15 times higher risk of infection. Unfortunately, many patients indicated for CIED implantation are also indicated for antithrombotic or anticoagulant treatment. Data on contemporary populations from clinical studies and surveys indicate a rate of use of anticoagulant therapy ranging from 15% in patients with pacemakers, to 35% in patients with ICDs, reaching almost 50% in patients with cardiac resynchronization therapy. Moreover, ∼50% of these patients have an indication for single or dual antiplatelet treatment.30 Typical examples are patients with a history of atrial fibrillation (either with slow ventricular response or as a part of sick sinus syndrome) or patients after valve surgery. Moreover, antithrombotic and anticoagulant treatment has changed within last 10 years with new and more potent drugs present on the market, such as prasugrel, ticagrelor or NOAC, and some patients requires a combination of anticoagulant and antiplatelet treatment, which makes the situation even more complicated.

rgery. Moreover, antithrombotic and anticoagulant treatment has changed within last 10 years with new and more potent drugs present on the market, such as prasugrel, ticagrelor or NOAC, and some patients requires a combination of anticoagulant and antiplatelet treatment, which makes the situation even more complicated. Cardiac implantable electronic devices implantation and anticoagulation The management of anticoagulation and antiplatelet treatment during CIED implantation has changed substantially within last ten years. According to the ESC guidelines for cardiac pacing from 2007, anticoagulation treatment should have been interrupted 3–8 days pre-operatively and replaced with heparin. This is in complete contradiction with the ESC guidelines from 2013. According to these recent guidelines, the use of heparin bridging to OAC has been shown to increase the risk of bleeding and continuation of warfarin is recommended instead.30–32

ave been interrupted 3–8 days pre-operatively and replaced with heparin. This is in complete contradiction with the ESC guidelines from 2013. According to these recent guidelines, the use of heparin bridging to OAC has been shown to increase the risk of bleeding and continuation of warfarin is recommended instead.30–32 The first study reporting the feasibility of implantation with ongoing warfarin was a study by Goldstein et al.31 However, the majority of the 37 patients in this study underwent generator replacement and not leads and pacemaker de novo implantation. Since then a number of observational studies and recent randomized studies have confirmed the superiority of warfarin continuation to bridging to heparin. In the BRUISE CONTROL study,32 i.e. the largest randomized study comparing ongoing warfarin to bridging to heparin strategy, warfarin continuation was associated with significantly lower risk of bleeding (RR 0.16, 95% CI 0.08–0.32) and no difference in the risk of thromboembolic events. Recently, observational and small randomized trials have shown similar incidences of bleeding complications during CIED implantation with uninterrupted novel anticoagulants (NOACs) or warfarin.33 In patients with moderate or even high risk (such as in patients with artificial valves, recent pulmonary embolisms or in patients with history of AF and higher CHADS2VASc scores), the implantation of CIED should be done while on p.o. anticoagulation, with careful haemostasis during surgery.

The first study reporting the feasibility of implantation with ongoing warfarin was a study by Goldstein et al.31 However, the majority of the 37 patients in this study underwent generator replacement and not leads and pacemaker de novo implantation. Since then a number of observational studies and recent randomized studies have confirmed the superiority of warfarin continuation to bridging to heparin. In the BRUISE CONTROL study,32 i.e. the largest randomized study comparing ongoing warfarin to bridging to heparin strategy, warfarin continuation was associated with significantly lower risk of bleeding (RR 0.16, 95% CI 0.08–0.32) and no difference in the risk of thromboembolic events. Recently, observational and small randomized trials have shown similar incidences of bleeding complications during CIED implantation with uninterrupted novel anticoagulants (NOACs) or warfarin.33 In patients with moderate or even high risk (such as in patients with artificial valves, recent pulmonary embolisms or in patients with history of AF and higher CHADS2VASc scores), the implantation of CIED should be done while on p.o. anticoagulation, with careful haemostasis during surgery. Cardiac implantable electronic devices and antiplatelet treatment Compared with untreated patients, aspirin carries a two-fold risk of bleeding and DAPT (aspirin plus thienopyridine) carries a four-fold or according some authors even a six-fold risk of bleeding during the peri-operative period. This risk was reduced by withholding clopidogrel 4–7 days before implantation. In most cases, dual antiplatelet medications can safely be discontinued, for a period of 5–7 days.

(aspirin plus thienopyridine) carries a four-fold or according some authors even a six-fold risk of bleeding during the peri-operative period. This risk was reduced by withholding clopidogrel 4–7 days before implantation. In most cases, dual antiplatelet medications can safely be discontinued, for a period of 5–7 days. Ablation of cardiac arrhythmias Ablation and anticoagulant treatment Left-sided ablations present a high risk of periprocedural thromboembolic events due to (i) the disease itself (typically atrial fibrillation), (ii) the differences in the clot formation in the right- and left-sided atria and ventricles, and (iii) the difference in the clinical manifestation in case of the embolization of right- and left-sided cardiac chambers. While thrombi from the right-sided cardiac chambers remain mostly asymptomatic, even small thrombus from the left atrium can lead to stroke with severe neurologic disability.

ria and ventricles, and (iii) the difference in the clinical manifestation in case of the embolization of right- and left-sided cardiac chambers. While thrombi from the right-sided cardiac chambers remain mostly asymptomatic, even small thrombus from the left atrium can lead to stroke with severe neurologic disability. All left-sided ablations were previously performed on heparin and the same was true of CIED implantation, patients were bridged to heparin from p.o. warfarin. The development of ongoing warfarin during ablation has followed a similar path to its use with CIED implantation Recently, according to non-randomized observational studies and randomized trials, ongoing uninterrupted warfarin has been shown to be safe and associated with lower rates of bleeding events, and lower rate of thromboembolic events compared with bridging to heparin.34 In these trials, target international normalized ratio (INR) before and during ablation was 2.0–3.0 and was checked one day before the procedure. In patients with uninterrupted warfarin, periprocedurally heparin was given to all of them with similar target ACT values. Not surprisingly, patients on uninterrupted warfarin had lower stroke events compared with bridging strategy (OR 0.17, 95% CI 0.08–0.35) according to the meta-analysis of 12 observational and randomized trials comparing this two strategies.35 Surprisingly, the rates of major (OR 0.72, 95% CI 0.54–0.95) and minor bleedings (OR 0.33, 95% CI 0.21–0.52) were also reduced in the uninterrupted warfarin strategy. It has become clear that holding warfarin and bridging with heparin/low-molecular-weight heparin creates a gap in which thrombotic complications are increased.

singly, the rates of major (OR 0.72, 95% CI 0.54–0.95) and minor bleedings (OR 0.33, 95% CI 0.21–0.52) were also reduced in the uninterrupted warfarin strategy. It has become clear that holding warfarin and bridging with heparin/low-molecular-weight heparin creates a gap in which thrombotic complications are increased. Recently, similar and quite robust data have been published comparing uninterrupted NOAC to warfarin. The ablation on uninterrupted NOAC was associated with similar rate of stroke and bleeding as the ablation with uninterrupted warfarin.36,37 With respect to total bleeding, no significant difference was observed between dabigatran, rivaroxaban and apixaban and warfarin according to the meta-analysis of the randomized trials.37 Ablation and antiplatelet treatment There is only a few reports regarding the risk and complications of catheter ablation if performed with concomitant dual antiplatelet and anticoagulant treatment. However, available reports indicate higher incidence of bleeding and vascular complications in ablation performed with clopidogrel.38 Catheter ablation present mostly elective procedure, and so discontinuation of clopidogrel or other thienopyridines is recommended.

omitant dual antiplatelet and anticoagulant treatment. However, available reports indicate higher incidence of bleeding and vascular complications in ablation performed with clopidogrel.38 Catheter ablation present mostly elective procedure, and so discontinuation of clopidogrel or other thienopyridines is recommended. Catheter-based interventions for acute ischaemic stroke There is lack of scientific evidence and complete absence of official guidelines recommending any specific protocol for periprocedural antithrombotic treatment during acute stroke interventions. Possibly, the most comprehensive document on this subject—the American guidelines for the management of acute stroke39,40—describe the reperfusion strategies and the use (or rather no use) of anticoagulant and antiplatelet agents as the potential primary therapy for stroke (when no reperfusion strategies are used), but not as periprocedural therapy during catheter-based thrombectomy (CBT).

uidelines for the management of acute stroke39,40—describe the reperfusion strategies and the use (or rather no use) of anticoagulant and antiplatelet agents as the potential primary therapy for stroke (when no reperfusion strategies are used), but not as periprocedural therapy during catheter-based thrombectomy (CBT). Thrombolysis These 2013 guidelines only describe thrombolysis use in acute ischaemic stroke. Intravenous thrombolysis is indicated for all eligible (per guidelines) stroke patients irrespective whether subsequent endovascular intervention is planned. The 2015 update40 and the Canadian guidelines41 further specify that endovascular intervention is indicated for all eligible acute ischaemic stroke patients including patients with contraindications to thrombolysis. When i.v. thrombolysis is used, the endovascular intervention should commence immediately, without waiting for the effect of thrombolysis. Nowadays, when stent-retrievers are much faster and much more effective, i.a. use of rtPA is reserved only for patients with more distal occlusions, not accessible with stent-retrievers and as a primary therapy is abandoned. Table 2 presents the current indications for acute stroke interventions according to the use of bridging thrombolysis. Table 2 Indications for acute stroke interventions with and without bridging thrombolysis

tients with more distal occlusions, not accessible with stent-retrievers and as a primary therapy is abandoned. Table 2 presents the current indications for acute stroke interventions according to the use of bridging thrombolysis. Table 2 Indications for acute stroke interventions with and without bridging thrombolysis Facilitated intervention (bridging thrombolysis)a Direct intervention (thrombolysis not used) Moderate or severe stroke NIHSS ≥6 NIHSS ≥6 Stroke onset—treatment delayb 0–4.5 h 0–6 h (6–12 h in selected patients with significant penumbra) Contraindications for the use of thrombolytics Bridging thrombolysis not possible Remains the only option for reperfusion Native CT (ASPECTS score) ≥6 ≥6 Angiographic finding (CT-A, MR-A, or invasive angiography)c ICA, MCA-M1, BA, or VA occlusion ICA, MCA-M1, BA, or VA occlusion NIHSS, National Institutes of Health Stroke Score; CT-A, computed tomography angiogram; MR-A, magnetic resonance angiogram; ICA = internal carotid artery; MCA-M1, M1 segment of the middle cerebral artery; BA, basilar artery; VA, vertebral artery. aWhen i.v. t-PA is used, patient should proceed immediately to interventional lab (waiting for the effect of thrombolysis is not anymore acceptable in 2015!). bStart of CT scan—groin puncture time (including e.v. thrombolysis) should be <60 min in 90% of patients! cWhen native CT scan shows the hyperdense MCA sign, no angiography is necessary, patient should proceed directly to the interventional lab.

aWhen i.v. t-PA is used, patient should proceed immediately to interventional lab (waiting for the effect of thrombolysis is not anymore acceptable in 2015!). bStart of CT scan—groin puncture time (including e.v. thrombolysis) should be <60 min in 90% of patients! cWhen native CT scan shows the hyperdense MCA sign, no angiography is necessary, patient should proceed directly to the interventional lab. Anticoagulants As mentioned above, no information is given in these three guideline documents about the use of anticoagulants during CBT in patients with contraindications for thrombolysis. In general, urgent anticoagulation with the goal of preventing early recurrent stroke or improving stroke outcomes or for the management of non-cerebrovascular conditions is not recommended due to the risk of serious intracranial haemorrhage (IIIA recommendation). Anticoagulant therapy within 24 h after rtPA is not recommended (IIIB). Antiplatelet agents Similarly, no recommendation is given for periprocedural use of antiplatelet agents. Acetylsalicylic acid is not recommended as a substitute for other acute interventions (IIIB), ASA is not recommended as adjunctive therapy within 24 h of i.v. rtPA (IIIC). Oral ASA should be initiated within 24–48 h after stroke onset (IA recommendation). The usefulness of clopidogrel in acute ischaemic stroke is not well established and further research is required (IIbC). Intravenous GPIIb/IIIa inhibitors are not recommended (IIIB).

s adjunctive therapy within 24 h of i.v. rtPA (IIIC). Oral ASA should be initiated within 24–48 h after stroke onset (IA recommendation). The usefulness of clopidogrel in acute ischaemic stroke is not well established and further research is required (IIbC). Intravenous GPIIb/IIIa inhibitors are not recommended (IIIB). Similar to AHA/ASA guidelines, no information about periprocedural anticoagulation or antiplatelet treatment is provided in the important multisociety consensus paper on catheter-based interventions in acute stroke.42

s adjunctive therapy within 24 h of i.v. rtPA (IIIC). Oral ASA should be initiated within 24–48 h after stroke onset (IA recommendation). The usefulness of clopidogrel in acute ischaemic stroke is not well established and further research is required (IIbC). Intravenous GPIIb/IIIa inhibitors are not recommended (IIIB). Similar to AHA/ASA guidelines, no information about periprocedural anticoagulation or antiplatelet treatment is provided in the important multisociety consensus paper on catheter-based interventions in acute stroke.42 Periprocedural therapy in published trials and registries Most published trials or registries do not mention periprocedural antithrombotic therapy at all.43–45 A small single centre registry of 23 consecutive cases of emergency carotid stenting followed by mechanical thrombectomy found successful carotid stenting in all cases, and establishment of TICI flow 2a/2b/3 in 91%. Symptomatic intracranial haemorrhage occurred in 5/23 patients (22%). Of 13 patients receiving an intravenous loading dose of abciximab during the procedure, 4/13 had SICH (31%) compared with 1/10 (10%) of those who did not. Of seven patients who received intravenous tissue plasminogen activator prior to the procedure, none had SICH. 90-day mortality was 9/23 (39%). All patients who had SICH were above the median age.46 In the TREVO study, it was recommended that administration of anticoagulants and antiplatelets be suspended for 24 h post-thrombectomy in patients who were not in direct need of these agents. Exclusion criteria: Heparin use within previous 48 h with aPTT >2 times normal was even an exclusion criterion. Intravenous thrombolysis was used in 60% of patients prior to the endovascular procedure. Periprocedural antithrombotic treatment included i.a. rtPA in 10% and GP IIb/IIIa inhibitors in 5%.47 One study48 used neither i.v. heparin nor intra-arterial fibrinolytics at any time during the mechanical thrombectomy procedure, even if the recanalization attempt was unsuccessful. When stent placement was needed, antiplatelet management consisted of 500 mg of aspirin i.v. during the procedure, and double antiplatelet was discussed after the 24-h CT control in view of any serious haemorrhagic complications. Patients treated by direct CBT in a Turkish study49 received 100 mg ASA before CBT in the emergency department. During interventional stroke procedure, 2000 units of bolus heparin were given routinely. No further antiplatelet or heparin was administered within 24 h of procedure. A CT or MRI was performed 24 h after the procedure. If no haemorrhage was present, aspirin 300 mg/day was given.

before CBT in the emergency department. During interventional stroke procedure, 2000 units of bolus heparin were given routinely. No further antiplatelet or heparin was administered within 24 h of procedure. A CT or MRI was performed 24 h after the procedure. If no haemorrhage was present, aspirin 300 mg/day was given. Elective carotid stenting The optimal anticoagulation regimen for carotid artery stenting (CAS) remains unknown.50 Periprocedural UHF is commonly used. Dual antiplatelet therapy with aspirin and clopidogrel is recommended. Two small, randomized trials comparing aspirin alone with double antiplatelet therapy for CAS were terminated prematurely due to high rates of stent thrombosis and neurological events in the aspirin-alone group.51,52 In patients with proven intolerance to DAPT, CEA should be preferred to CAS. Newer antiplatelet agents such as prasugrel or ticagrelor have not yet been adequately tested in CAS. A recent survey among the Dutch interventional radiologists showed that almost all continue acetyl salicylic acid till the time of percutaneous interventions. Clopidogrel is stopped in 40% peripheral interventions, but not before CAS. A flushing solution on the sideport of the sheath was used routinely by 50% of radiologists during CAS, but only a minority of them (28%) used a heparinized flushing solution. Unfractionated heparin was used by almost all radiologists as a bolus (5000 IU was the most used dosage, additional smaller bolus usually repeated after 1 h in longer procedures).53

f the sheath was used routinely by 50% of radiologists during CAS, but only a minority of them (28%) used a heparinized flushing solution. Unfractionated heparin was used by almost all radiologists as a bolus (5000 IU was the most used dosage, additional smaller bolus usually repeated after 1 h in longer procedures).53 Interventions for peripheral arterial disease Antiplatelet therapy with aspirin (or clopidogrel) is recommended to reduce overall cardiovascular risk in chronic symptomatic lower extremity artery disease (LEAD) patients.50,54 Data on periprocedural antithrombotic therapy in LEAD interventions are spare. Most published data deal with pharmacological treatment following revascularization procedures, but not during such procedures. Recommendations are based on expert consensus only. Acetylsalicylic acid together with heparin (UFH) is commonly used during peripheral artery interventions. Heparin is used at the doses of 50–70 IU/kg (up to 100 IU/kg). Continuous anticoagulation therapy (UFH, low-molecular-weight heparin, LMWH) following intervention (24–48hrs) is sometimes recommended after procedures with suboptimal results, complex lesions, and infrapopliteal arteries. Acetylsalicylic acid monotherapy after balloon angioplasty (without stenting) was used in the femoropopliteal region in the Basil trial.55 This trial demonstrated non-inferiority of the interventional treatment of severe lower limb ischaemia to surgical treatment. After intervention with bare metal stent (BMS) in infrainguinal region DAPT (i.e. combination of ASA and thienopyridine) is recommended for 1 month.50 The DES, that proved superiority over BMS in femoropopliteal region, was in Zilver PTX (paclitaxel) trial using DAPT for 2 months.56 In Sirocco trial with implant sirolimus eluting stents in femoropopliteal regions DAPT was used for 1 month and no benefit was seen comparing BMSs.57 It has been proven that anticoagulation therapy with Warfarin after infrainguinal balloon PTA is not superior over ASA, but has higher bleeding risk.50 Due to the lacking data regarding antithrombotic treatment in LEAD interventions, some centres adjust their protocols adapting data from coronary interventions (PCI) and prolong DAPT therapy. Individualized therapy taking into account the diffuseness of the disease, the quality of the inflow and the outflow, the presence of critical limb ischaemia, the extent of stenting, the use of covered stents, and the stent fracture risk is reasonable.58

data from coronary interventions (PCI) and prolong DAPT therapy. Individualized therapy taking into account the diffuseness of the disease, the quality of the inflow and the outflow, the presence of critical limb ischaemia, the extent of stenting, the use of covered stents, and the stent fracture risk is reasonable.58 The survery of the British Society of Interventional Radiology was summarized with the following recommendation: heparinized saline should be used at a recognized standard concentration of 1000 IU/l as a flushing concentration in all arterial vascular interventions and that 3000 IU bolus is considered the standard dose for straightforward therapeutic procedures and 5000 IU for complex, crural, and endovascular aneurysm repair work. The bolus should be given after arterial access is obtained to allow time for optimal anticoagulation to be achieved by the time of active intervention and stenting. Further research into clotting abnormalities following such interventional procedures would be an interesting quantifiable follow-up to this initial survey of opinions and practice.59

er arterial access is obtained to allow time for optimal anticoagulation to be achieved by the time of active intervention and stenting. Further research into clotting abnormalities following such interventional procedures would be an interesting quantifiable follow-up to this initial survey of opinions and practice.59 Antithrombotic strategies in cardiac patients undergoing non-cardiac surgery Aspirin and P2Y12 inhibitors The timing of non-cardiac surgery should always be weighted individually based on the nature of surgical illness vs. the cardiac illness. Early (4 weeks) after stent implantation DAPT should be continued in all patients unless the risk of life-threatening surgical bleeding is unacceptably high. Continuation of aspirin may be considered in the peri-operative period. Stopping aspirin therapy should be considered if haemostasis may be difficult to control during surgery. In patients treated with P2Y12 inhibitors, who need to undergo surgery, postponing surgery for 5–7 days after P2Y12 inhibitor cessation should be considered unless the patient is at high risk of an ischaemic event.60

topping aspirin therapy should be considered if haemostasis may be difficult to control during surgery. In patients treated with P2Y12 inhibitors, who need to undergo surgery, postponing surgery for 5–7 days after P2Y12 inhibitor cessation should be considered unless the patient is at high risk of an ischaemic event.60 Anticoagulants The bleeding risk should be individually weighed against the benefit of anticoagulants. Patients treated with vitamin K antagonists (VKAs) should have the INR <1.5 to undergo surgery safely. In patients with a high risk of thrombo-embolism (atrial fibrillation with a CHA2DS2-VASc score ≥4 or mechanical prosthetic valves or recent venous thrombo-embolism) discontinuation of VKAs is hazardous and these patients will need bridging therapy with a therapeutic-dose of LMWH. Vitamin K antagonist treatment should be stopped 3–5 days before surgery, with daily INR measurements, until ≤1.5 is reached, and LMWH should be started 1 day after discontinuation of VKA. Low-molecular-weight heparin is resumed at the pre-procedural dose 1–2 days after surgery, depending on the patient's haemostatic status, but at least 12 h after the procedure. Vitamin K antagonists should be resumed on Day 1 or 2 after surgery—depending on adequate haemostasis—with the pre-operative maintenance dose plus a boosting dose of 50% for two consecutive days. Low-molecular-weight heparin should be continued until the INR returns to therapeutic levels.

east 12 h after the procedure. Vitamin K antagonists should be resumed on Day 1 or 2 after surgery—depending on adequate haemostasis—with the pre-operative maintenance dose plus a boosting dose of 50% for two consecutive days. Low-molecular-weight heparin should be continued until the INR returns to therapeutic levels. Direct oral anticoagulants Direct oral anticoagulants (dabigatran, rivaroxaban, apixaban, or edoxaban) have a well-defined ‘on’ and ‘off’ action, ‘bridging’ to surgery is in most cases unnecessary, due to their short biological half-lives. Summary The periprocedural antithrombotic strategies vary between different types of percutaneous cardiovascular interventions. Heparin remains the key drug for most of these interventions. Oral antiplatelet drugs are essential when stents are implanted. Oral anticoagulants used chronically are not interrupted during most interventions in electrophysiology. Thrombolysis remains important part of acute stroke treatment. Overview on the use of antithrombotic drugs in different settings is summarized in Table 3. Table 3 Summary on the periprocedural use of antithrombotic drugs

anticoagulants used chronically are not interrupted during most interventions in electrophysiology. Thrombolysis remains important part of acute stroke treatment. Overview on the use of antithrombotic drugs in different settings is summarized in Table 3. Table 3 Summary on the periprocedural use of antithrombotic drugs PCI for AMI Elective PCI Structural interventions Electronic device implantation Arrhythmias ablation Acute stroke (thrombectomy) Elective carotid stenting Peripheral arterial interventions Thrombolytics NO (exception: pre-hospital thrombolysis in patients with very long delays to PCI) NO NO NO NO YES as bridging therapy in eligible patients NO NO for most cases. YES (local thrombolysis) in selected cases Injectable anticoagulants YES (heparin or enoxaparin or bivalirudin, dosage of anticoagulant must be adopted to body weight!) YES (heparin or enoxaparin, dosage of anticoagulant must be adopted to body weight!) YES (heparin, dosage adopted to body weight and to ACT!) NO YES for patients who are not on chronic OAC. NO for temporary replacement of OAC NO for patients who received bridging rtPA. YES (low dose heparin) for patients treated with direct thrombectomy without rtPA YES YES Oral anticoagulants NO If patient with AMI is on chronic OAC, no or lower dose heparin should be used. NO If chronic use, OAC should be interrupted before PCI NO If chronic use, OAC should be interrupted before PCI NO If patient is on chronic OAC, therapy should continue (careful timing of implantation with respect to OAC dosage) YES for patients who are on chronic OAC—they should not interrupt treatment NO If patient with acute stroke is on chronic OAC, no anticoagulants should be added during mechanical intervention NO NO Acetylsalicylic acid YES YES YES NO (If chronic use should be discontinued 5–7 days before implantation) NO (If chronic use should be discontinued 5–7 days before implantation) NO (Exception: YES just prior to carotid stenting in the acute phase of stroke) YES YES P2Y12 inhibitors YES YES YES NO (If chronic use, should be discontinued 5–7 days before implantation) NO (If chronic use, should be discontinued 5–7 days before implantation) NO (If carotid stenting was part of the acute procedure, P2Y12 inhibitors should be initiated after control CT scan post-thrombectomy procedure) YES YES if stent implantation GPIIb/IIIa inhibitors Routine upfront use not indicated.

e implantation) NO (If chronic use, should be discontinued 5–7 days before implantation) NO (If carotid stenting was part of the acute procedure, P2Y12 inhibitors should be initiated after control CT scan post-thrombectomy procedure) YES YES if stent implantation GPIIb/IIIa inhibitors Routine upfront use not indicated. Selective (bail-out) use in cath-lab only NO NO NO NO NO NO NO PCI, percutaneous coronary intervention; AMI, acute myocardial infarction; GP, glycoprotein; OAC, oral anticoagulants. Funding The authors receive support from the Charles University Prague, Third Faculty of Medicine, research project P35. Funding to pay the Open Access publication charges for this article was provided by the Charles University Prague, project P35. Conflict of interest: none declared.

Introduction Heart failure (HF) is a chronic condition, punctuated by acute episodes, which affects as many as one in five people aged 70–80 years.1,2 In acute heart failure (AHF), rapid worsening of the signs and symptoms of HF results in the requirement for urgent therapy and, frequently, hospitalization.3 The frequency of AHF episodes increases with disease progression, resulting in high rates of hospitalization and an increased risk of mortality.3 As such, AHF places a significant burden on both patients and healthcare systems.4 Pathophysiologically, it is known that AHF involves both haemodynamic abnormalities and end-organ damage (Figure 1).5–12 Haemodynamic abnormalities result in early clinical features of congestion,2,13–15 whereas end-organ damage may contribute to long-term morbidity and mortality.16 Figure 1 The ‘continuum’ of pathophysiological changes associated with acute heart failure that may lead to both short- and long-term effects on the heart and other end organs.5–12 Current therapies for AHF include loop diuretics and vasodilators, agents which stimulate vasodilation and diuresis to relieve haemodynamic abnormalities.4,10,17–19 However, none of these agents have been shown to prevent end-organ damage, and their use may be associated with detrimental effects on numerous organs, thereby contributing to long-term morbidity and mortality.20–22 As a result, new therapies for the treatment of AHF should relieve congestion to improve short-term clinical consequences and provide organ protection to positively impact the long-term clinical consequences of AHF.

detrimental effects on numerous organs, thereby contributing to long-term morbidity and mortality.20–22 As a result, new therapies for the treatment of AHF should relieve congestion to improve short-term clinical consequences and provide organ protection to positively impact the long-term clinical consequences of AHF. Human relaxin-2 is the major form of the hormone relaxin, which has vital roles during pregnancy.23,24 Relaxin-2 binds primarily to relaxin family peptide receptor 1 (RXFP1), located in the heart, kidneys, and vasculature, to activate numerous cellular pathways.16,25–27 Serelaxin has been manufactured as the recombinant form of human relaxin-2 and is currently under investigation for the treatment of AHF.27,28 In this review, we briefly describe the unique mechanisms underlying the ability of serelaxin to relieve congestion and, therefore, mediate short-term beneficial effects in patients with AHF. We also examine, in detail, the novel mechanisms by which serelaxin, unlike current treatments, may limit end-organ damage and thus, provide long-term treatment benefit in patients with AHF.

rlying the ability of serelaxin to relieve congestion and, therefore, mediate short-term beneficial effects in patients with AHF. We also examine, in detail, the novel mechanisms by which serelaxin, unlike current treatments, may limit end-organ damage and thus, provide long-term treatment benefit in patients with AHF. Serelaxin for the treatment of acute heart failure: key clinical data The safety and efficacy of serelaxin for the treatment of patients with AHF has been determined in the preliminary RELAXin in Acute Heart Failure (pre-RELAX-AHF) and RELAXin in Acute Heart Failure (RELAX-AHF) clinical trials. In the phase IIb pre-RELAX-AHF trial, serelaxin (30 µg/kg/day 48-h infusion) resulted in a positive effect on dyspnoea compared with placebo.29 In the phase III RELAX-AHF trial, serelaxin (30 µg/kg/day 48-h infusion), when compared with placebo, significantly improved the primary efficacy endpoint of dyspnoea relief by the visual analogue scale area under the curve to Day 5, with numerical improvement observed in the primary endpoint of dyspnoea as assessed by the Likert scale at 6, 12, and 24 h.30 Serelaxin treatment improved signs and symptoms of congestion and length of hospital stay compared with placebo in the RELAX-AHF study, although, no significant improvement in the two secondary endpoints of days alive and out of hospital, and cardiovascular (CV) death or rehospitalization for HF or renal failure through Day 60 was observed.30 In both studies, serelaxin demonstrated favourable effects on longer-term clinical outcomes, such as CV and all-cause mortality through Day 180 compared with placebo (Figure 2).29–31 In the RELAX-AHF study, elevated levels of troponin T, cystatin C, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and N-terminal pro-B-type natriuretic peptide (NT-proBNP) were associated with an increased risk of all-cause mortality through Day 180 (Figure 3).31 Serelaxin treatment, when compared with placebo, was associated with lower levels of these biomarkers, indicating that serelaxin may protect organs from further damage following AHF hospitalization.31 Overall, serelaxin had a favourable safety and tolerability profile compared with placebo.29,30 Figure 2 Risk for all-cause mortality through Day 180 in Pre-RELAX-AHF and RELAX-AHF.31 AHF, acute heart failure; RELAX-AHF, RELAXin in Acute Heart Failure; Pre-RELAX-AHF, preliminary RELAXin in Acute Heart Failure.

ation.31 Overall, serelaxin had a favourable safety and tolerability profile compared with placebo.29,30 Figure 2 Risk for all-cause mortality through Day 180 in Pre-RELAX-AHF and RELAX-AHF.31 AHF, acute heart failure; RELAX-AHF, RELAXin in Acute Heart Failure; Pre-RELAX-AHF, preliminary RELAXin in Acute Heart Failure. Reproduced under the terms of the Elsevier user license (http://www.elsevier.com/about/open-access/open-access-policies/oa-license-policy/elsevier-user-license) for Metra et al.31 Figure 3 All-cause mortality through Day 180 in RELAX-AHF by markers of organ damage/dysfunction: troponin T (A); cystatin C (B); AST (C); ALT (D), and NT-proBNP (E).31 ALT, alanine aminotransferase; AST, aspartate aminotransferase; CI, confidence interval; HR, hazard ratio; NT-proBNP, N-terminal pro-B-type natriuretic peptide. Reproduced under the terms of the Elsevier user license (http://www.elsevier.com/about/open-access/open-access-policies/oa-license-policy/elsevier-user-license) for Metra et al.31 Although promising, pre-RELAX-AHF and RELAX-AHF studies were not powered to detect changes in mortality, thus adequately designed follow-up studies are needed. A second phase III trial, RELAX-AHF-2, is ongoing and will further investigate the safety and efficacy of serelaxin for the treatment of patients with AHF, including the mortality benefit observed in previous clinical trials.32

wered to detect changes in mortality, thus adequately designed follow-up studies are needed. A second phase III trial, RELAX-AHF-2, is ongoing and will further investigate the safety and efficacy of serelaxin for the treatment of patients with AHF, including the mortality benefit observed in previous clinical trials.32 Serelaxin and correction of haemodynamic imbalance Observations from preclinical and clinical studies indicate that serelaxin acts via multiple mechanisms to correct haemodynamic imbalance and relieve congestion, as described in Table 1.30,33–52 For instance, serelaxin is thought to stimulate vasorelaxatory systems and counteract vasoconstrictor systems, to mediate both rapid and sustained vasorelaxation53 (Figure 4),28,43,54 and thus, improve haemodynamics and alleviate congestion. Evidence suggests that serelaxin also increases arterial compliance40,42 and decreases systemic vascular resistance,35,36,44–46 which could increase capacitance to prevent fluid redistribution to the lungs and improve haemodynamic abnormalities, aiding decongestion in AHF.8 Interestingly, in contrast to vasodilators such as nitroglycerin, which primarily act via direct venodilation,55 the vasorelaxatory action of serelaxin is thought to predominantly affect arteries.45 Table 1 Effects mediated by serelaxin that may alleviate haemodynamic imbalance and relieve congestion in patients with AHF30,33–52

gly, in contrast to vasodilators such as nitroglycerin, which primarily act via direct venodilation,55 the vasorelaxatory action of serelaxin is thought to predominantly affect arteries.45 Table 1 Effects mediated by serelaxin that may alleviate haemodynamic imbalance and relieve congestion in patients with AHF30,33–52 Effect mediated by serelaxin Evidence from preclinical and clinical studies following administration of serelaxina In vitro Mice Rats Healthy subjects Patients with CHF Patients with AHF Possible clinical consequences Reduction of cardiac pressures ↓SBPb33,34 (including porcine relaxin) ↓DBPc35 ↓PCWPc35 ↓SBPc35 ↓PAPc35 ↓DBPd36 ↓PCWPd36 ↓SBPd36 ↓PAPd36 ↓JVPd30 Improved haemodynamics Relief of congestion Prevention of further stimulation of neurohumoral systems Stimulation of vasorelaxation Blunted responses of rat mesenteric arteries to vasoconstriction induced by AVP and NE37 (rat relaxin) Vasorelaxation of small human resistance arteries38 ↑Coronary flow/↑NO generation in isolated guinea pig hearts subject to IR injury39 (porcine relaxin) ↑Arterial compliance40 Blunted response to vasoconstriction and ↑BP induced by Ang II33,41 (including porcine relaxin) ↓Wall stiffness42 ↑Arterial compliance42 ↑Rapid and sustained BK-mediated vasorelaxation of mesenteric arteries43 Improved haemodynamics Relief of congestion Possible prevention of fluid redistribution Reduction of SVR ↓SVRe44–46 ↓SVRf35 ↓SVRd36 Vasorelaxation Improved haemodynamics Relief of congestion Possible prevention of fluid redistribution Preservation of diuresis and natriuresis ↑Urinary excretion of sodium47 ↓Salt sensitivityb34 (porcine relaxin) ↑Urinary flow rate47 ↑Renal clearance, fractional excretion and urinary excretion of sodiumg48 No effect on urinary flow rateg48 No effect on urinary excretion of sodium or urinary flow rateh49 Neutral effect on diuretic responsei50 Preservation of renal function Improved haemodynamics Possible prevention of fluid redistribution Increased RBF and preservation of GFR ↑GFR41,51,52 ↑RBF41,47,51,52 ↑RBFg48 No effect on GFRg48 ↑RBFh,j49 No effect on GFRh49 Preservation of renal function Possible long-term renal protection Increased cardiac output ↑COe44–46 ↑COf35 No impact on CId36 Improved haemodynamics Relief of congestion Prevention of further stimulation of neurohumoral systems AHF, acute heart failure; Ang II, angiotensin II; AVP, arginine vasopressin; BK, bradykinin; BP, blood pressure; CHF, chronic heart failure, CI, cardiac index; CO, cardiac output; DBP,

e44–46 ↑COf35 No impact on CId36 Improved haemodynamics Relief of congestion Prevention of further stimulation of neurohumoral systems AHF, acute heart failure; Ang II, angiotensin II; AVP, arginine vasopressin; BK, bradykinin; BP, blood pressure; CHF, chronic heart failure, CI, cardiac index; CO, cardiac output; DBP, diastolic blood pressure; GFR, glomerular filtration rate; IR, ischaemia reperfusion; JVP, jugular venous pressure; NE, norepinephrine; NO, nitric oxide; PAP, pulmonary artery pressure; PCWP, pulmonary capillary wedge pressure; RBF, renal blood flow; SBP, systolic blood pressure; SVR, systemic vascular resistance. aSerelaxin unless otherwise stated. bIn rat models of Ang II-induced hypertension and/or salt-sensitive hypertension, reductions in SBP were mediated by the NOS system. cSerelaxin administered at doses of 10–100 µg/kg/day and/or 960 µg/kg/day for 24 h. dSerelaxin administered at a dose of 30 µg/kg/day for 20 h. eIn hypertensive and non-hypertensive rats. fSerelaxin administered at a dose of 960 µg/kg/day for 24 h. gAn intravenous bolus of serelaxin (0.2 μg/kg) was administered over 5 min, followed by an infusion of 0.5 μg/kg per hour for 4 h. hSerelaxin administered at a dose of 30 µg/kg/day for 24 h. iSerelaxin administered at a dose of 30 µg/kg/day for 48 h. jImproved RBF observed up to 28 h post-serelaxin dose compared with placebo.

fSerelaxin administered at a dose of 960 µg/kg/day for 24 h. gAn intravenous bolus of serelaxin (0.2 μg/kg) was administered over 5 min, followed by an infusion of 0.5 μg/kg per hour for 4 h. hSerelaxin administered at a dose of 30 µg/kg/day for 24 h. iSerelaxin administered at a dose of 30 µg/kg/day for 48 h. jImproved RBF observed up to 28 h post-serelaxin dose compared with placebo. Figure 4 Time-dependent effects of intravenously administered serelaxin on vasoactive systems that result in vasorelaxation.28,43,54 A, time after serelaxin administration, when the hormone is detectable in the blood ranges from minutes to hours; B, time after serelaxin administration, when the hormone is not detected in the blood ranges from 1 to several days; Ang II, angiotensin II; AVP, arginine vasopressin; BK, bradykinin; COX2, cyclo-oxygenase 2; EDHF, endothelium-derived hyperpolarizing factor; eNOS, endothelial nitric oxide synthase; ET, endothelin; ET-BR, endothelin receptor type B; MMP, metalloproteinase; NE, norepinephrine; nNOS, neuronal nitric oxide synthase; RXFP1, relaxin/insulin-like family peptide receptor 1; TGF-β, transforming growth factor β; VEGF, vascular endothelial growth factor.

zing factor; eNOS, endothelial nitric oxide synthase; ET, endothelin; ET-BR, endothelin receptor type B; MMP, metalloproteinase; NE, norepinephrine; nNOS, neuronal nitric oxide synthase; RXFP1, relaxin/insulin-like family peptide receptor 1; TGF-β, transforming growth factor β; VEGF, vascular endothelial growth factor. In addition to inducing vasorelaxation, serelaxin treatment has been shown to reduce cardiac pressures and to preserve or improve cardiac and renal function,30,33–36,41,44–52 which is likely to help restore haemodynamics, relieve congestion (via mechanisms which may include the prevention of fluid redistribution), and prevent further stimulation of neurohumoral systems in AHF.2,56 In addition, the renal effects of serelaxin may be associated with long-term renal protection, which warrants further investigation. Serelaxin treatment and the limitation of end-organ damage Serelaxin interferes with the mechanisms underlying the development of end-organ damage In patients with AHF, haemodynamic alterations stimulate a number of systemic mechanisms, including the adrenergic system, vasoactive hormones, inflammation, and oxidative stress which, in turn, alter the local mechanisms controlling cell death, tissue repair, and vessel function, contributing to the development of cardiac, renal, hepatic, vascular, and other organ damage.2,6,57–64 The available evidence suggests that serelaxin may interfere with these systemic and local mechanisms to limit end-organ damage.

, alter the local mechanisms controlling cell death, tissue repair, and vessel function, contributing to the development of cardiac, renal, hepatic, vascular, and other organ damage.2,6,57–64 The available evidence suggests that serelaxin may interfere with these systemic and local mechanisms to limit end-organ damage. Serelaxin and inhibition of inflammation Damage to organs including the heart, kidneys, and liver occurs early in AHF and has long-term consequences.16,65 Inflammatory activation can contribute to organ injury, in addition to vascular dysfunction and fluid overload.8,16 For instance, in patients with newly diagnosed HF, levels of tumour necrosis factor alpha (TNF-α), interleukin (IL)-6, and CD14 were elevated on the third day of initial hospitalization and associated with impaired function of the left atrium and more advanced left ventricular (LV) systolic and diastolic dysfunction.66 Changes in inflammatory pathways have been determined in a number of studies following the administration of serelaxin. In human umbilical vein endothelial cells incubated with serelaxin, TNF-α-induced upregulation of vascular cell adhesion molecule 1 (VCAM-1) and platelet endothelial cell adhesion molecule was diminished, along with C-C chemokine receptor type 2 and monocyte chemotactic protein 1 levels, and monocyte adhesion to the cells.67 In addition, serelaxin inhibited basophil function, via nitric oxide synthase activation, to reduce histamine release and prevent the rise in intracellular calcium that stimulates granule release.68,69

ng with C-C chemokine receptor type 2 and monocyte chemotactic protein 1 levels, and monocyte adhesion to the cells.67 In addition, serelaxin inhibited basophil function, via nitric oxide synthase activation, to reduce histamine release and prevent the rise in intracellular calcium that stimulates granule release.68,69 In rats subjected to cardiac, renal, hepatic, or splanchnic ischaemia–reperfusion (IR) injury, treatment with serelaxin or porcine relaxin diminished myeloperoxidase activity, a marker of inflammatory leukocyte infiltration.70–74 Serelaxin treatment decreased expression of inflammatory mediators and adhesion molecules including intercellular adhesion molecule-1 (ICAM-1), IL-1β, IL-18, and TNF-α in rats subjected to renal IR injury,71 while porcine relaxin downregulated expression of adhesion molecules P-selectin, E-selectin, VCAM, and ICAM-1 in a rat model of splanchnic IR injury,70 as well as TNF-α expression in a rat model of renal IR injury.75 In addition, porcine relaxin treatment was associated with a reduction in the number of neutrophils and inhibition of mast cell granule release in a rat model of cardiac IR injury.74 Similarly, reductions in myeloperoxidase levels and cardiac mast cell degranulation were evident following the administration of serelaxin in a pig model of cardiac IR injury.76,77

associated with a reduction in the number of neutrophils and inhibition of mast cell granule release in a rat model of cardiac IR injury.74 Similarly, reductions in myeloperoxidase levels and cardiac mast cell degranulation were evident following the administration of serelaxin in a pig model of cardiac IR injury.76,77 Inhibiting the inflammatory response in patients with AHF may decrease fluid overload to relieve congestion and positively impact vascular, myocardial, renal, and hepatic injury and dysfunction8,71,73,75,77 and consequently, improve long-term outcomes. The anti-inflammatory actions of serelaxin distinguish this agent from current AHF therapies, such as nitrates, that have not been shown to improve long-term outcomes in patients with AHF21 and are therefore unlikely to inhibit inflammation. Serelaxin and reduction of oxidative stress Increased oxidative stress results from the dominance of reactive oxygen species (ROS) such as superoxide over endogenous antioxidant defence mechanisms.78 In patients with AHF, oxidative stress can result in myocardial, renal, and hepatic injury and remodelling.16 Neurohormones contribute to the activation of ROS in AHF, while mitochondrial calcium overload and dysfunction (via leaky type 2 ryanodine receptors) may lead to increased release of ROS in HF78,79 and reperfusion-induced inflammation may contribute to oxidative cardiac tissue injury.24

epatic injury and remodelling.16 Neurohormones contribute to the activation of ROS in AHF, while mitochondrial calcium overload and dysfunction (via leaky type 2 ryanodine receptors) may lead to increased release of ROS in HF78,79 and reperfusion-induced inflammation may contribute to oxidative cardiac tissue injury.24 In vitro studies, animal models and clinical studies have investigated the effects of animal relaxin and serelaxin on oxidative stress. In vitro, porcine relaxin was found to reduce the production of superoxide anions from human neutrophils.68 In rats with renal or splanchnic IR injury, serelaxin treatment was associated with increased levels of the antioxidant enzymes manganese and copper–zinc superoxide dismutase71 and diminished consumption of superoxide dismutase, lipid peroxidation, and markers of deoxyribonucleic acid (DNA) damage including 8-hydroxy-2′-deoxyguanosine and poly-ADP-ribosylated DNA.70 In addition, serelaxin decreased hydrogen peroxide and thiobarbituric acid-reactive substance (TBARs) excretion and consequently, oxidative stress, in rats with angiotensin II-induced hypertension.33 In the same experimental model, serelaxin treatment reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase activity (i.e. superoxide anion generation) and excretion of TBARs and 8-isoprostane (markers of oxidative stress), and restored nitric oxide (NO) oxidation product excretion.80 Finally, serelaxin was found to decrease levels of malondialdehyde (MDA), a marker of oxygen-free radical-mediated cell damage, in a porcine model of cardiac IR injury.76

ion generation) and excretion of TBARs and 8-isoprostane (markers of oxidative stress), and restored nitric oxide (NO) oxidation product excretion.80 Finally, serelaxin was found to decrease levels of malondialdehyde (MDA), a marker of oxygen-free radical-mediated cell damage, in a porcine model of cardiac IR injury.76 In patients with AHF, serelaxin treatment (30 µg/kg/day, 48-h infusion) significantly reduced levels of uric acid, a marker of oxidative stress, compared with placebo.31 This finding reinforces the novel mechanism of action of serelaxin and suggests that this agent may possess antioxidant properties, to prevent excess formation of superoxide, which reacts with NO to form the powerful oxidant peroxynitrite.81 Protecting against oxidative stress could prevent apoptosis/necrosis and, consequently, protect the endothelium and limit the end-organ damage associated with AHF.31,33,70,71 In contrast to serelaxin, current AHF therapies, such as nitrates, do not possess antioxidant properties and may contribute to the development of endothelial dysfunction, via NO-mediated increases in superoxide and thus, peroxynitrite.82–84

the endothelium and limit the end-organ damage associated with AHF.31,33,70,71 In contrast to serelaxin, current AHF therapies, such as nitrates, do not possess antioxidant properties and may contribute to the development of endothelial dysfunction, via NO-mediated increases in superoxide and thus, peroxynitrite.82–84 Serelaxin and inhibition of cell death Cardiac wall stress as well as the stimulation of neurohormones, oxidative stress, and release of inflammatory mediators result in cell death via apoptosis and necrosis, and ultimately, organ damage in patients with AHF.2,6,58–63 Previous studies have shown that anti-apoptotic and anti-necrotic effects are associated with end-organ preservation.75,85 Preventing organ damage by protecting cells from apoptosis and/or necrosis is therefore likely to improve long-term outcomes in patients with AHF;31,86 however, evidence suggests that current standard of treatment does not provide such benefit.3,21,22 In vitro, serelaxin has been shown to antagonize apoptosis in neonatal rat cardiomyocytes exposed to hydrogen peroxide87 and high levels of glucose.88 Serelaxin also significantly increased cell viability and diminished apoptosis and nitroxidative damage in both H9c2 rat cardiomyoblasts and primary mouse cardiomyocytes subjected to hypoxia and reoxygenation; these effects were partly due to the upregulation of Notch-1 signalling.89

ide87 and high levels of glucose.88 Serelaxin also significantly increased cell viability and diminished apoptosis and nitroxidative damage in both H9c2 rat cardiomyoblasts and primary mouse cardiomyocytes subjected to hypoxia and reoxygenation; these effects were partly due to the upregulation of Notch-1 signalling.89 In vivo studies have demonstrated beneficial effects of serelaxin and animal relaxin on apoptosis and necrosis. In rat models with renal injury, serelaxin treatment has been associated with reduced DNA damage and lipid peroxidation.71 In addition, serelaxin has been shown to protect against IR injury in the rat liver, as demonstrated by lower MDA levels in a model of isolated reperfused rat liver.72,73 Administration of porcine relaxin has resulted in diminished calcium overload and MDA levels74 and lower apoptotic cell counts, as assessed by caspase-3 expression and/or terminal deoxynucleotidyl transferase dUTP nick-end labelling (TUNEL) in rat models of cardiac IR injury, splanchnic IR injury, and renal IR injury, respectively.70,75 Decreased peroxidation products, nitration products, and markers of DNA damage were also reported following porcine relaxin treatment in a rat model of splanchnic IR injury,70 while rat relaxin-3 reduced MDA levels following myocardial injury in rats.90 Similarly, in a mouse model of cardiac IR injury, treatment with serelaxin antagonized apoptosis, as assessed by TUNEL staining.85 Finally, in pig models of cardiac IR injury, tissue calcium overload, tissue caspase-3 activity, TUNEL-positive cardiomyocytes, and mitochondrial swelling in cardiomyocytes were diminished76 and oxidative cardiac tissue injury was inhibited, as demonstrated by decreased MDA levels.77

s, as assessed by TUNEL staining.85 Finally, in pig models of cardiac IR injury, tissue calcium overload, tissue caspase-3 activity, TUNEL-positive cardiomyocytes, and mitochondrial swelling in cardiomyocytes were diminished76 and oxidative cardiac tissue injury was inhibited, as demonstrated by decreased MDA levels.77 Serelaxin and inhibition of tissue fibrosis Induction of fibrosis and remodelling of organs, including the heart, kidneys, and liver, can result from neurohumoral activation, inflammation, and oxidative stress in AHF.16 Increased levels of markers of extracellular matrix turnover, including matrix metalloproteinase (MMP)-2, tissue inhibitor of MMP (TIMP)-1, and procollagen type III N-terminal peptides, have been observed during the first 24 h of hospital admission for HF decompensation.6 In addition, failing hearts, when compared with non-failing hearts, have demonstrated dysregulation of microRNA expression, which is thought to contribute to myocardial remodelling in HF.91

d procollagen type III N-terminal peptides, have been observed during the first 24 h of hospital admission for HF decompensation.6 In addition, failing hearts, when compared with non-failing hearts, have demonstrated dysregulation of microRNA expression, which is thought to contribute to myocardial remodelling in HF.91 In vitro, serelaxin inhibited transforming growth factor beta (TGF-β) and/or TIMPs in human hepatic stellate cells and human dermal fibroblasts,92,93 and increased expression of MMPs, including MMP-1, -2, -9, and -13, via mechanisms including the NO pathway, in human dermal fibroblasts92,94 and rat renal myofibroblasts.94 Production of collagen was found to decrease in rat atrial and ventricular fibroblasts95,96 and human scleroderma fibroblasts97 following administration of serelaxin. In addition, serelaxin treatment downregulated activation of human renal fibroblasts,98 rat renal fibroblast function,99 and differentiation of rat renal fibroblasts to myofibroblasts,100 to inhibit renal fibrogenesis.

ntricular fibroblasts95,96 and human scleroderma fibroblasts97 following administration of serelaxin. In addition, serelaxin treatment downregulated activation of human renal fibroblasts,98 rat renal fibroblast function,99 and differentiation of rat renal fibroblasts to myofibroblasts,100 to inhibit renal fibrogenesis. The potential anti-fibrotic and anti-hypertrophic actions of serelaxin have also been assessed in vivo. Serelaxin treatment reduced ventricular collagen accumulation in mice,95 cardiac fibrosis in mouse models of myocardial infarction-induced IR injury,85 and isoprenaline-induced cardiac injury when compared with the angiotensin-converting enzyme inhibitor enalapril.101 In the latter study, combined administration of enalapril and serelaxin diminished cardiac fibrosis two-fold compared with enalapril alone, and the inhibitory effects of serelaxin were mediated by TGF-β downregulation.101 In ageing rats and in rat models of hypertension and diabetic cardiomyopathy, administration of serelaxin decreased LV and kidney collagen content,52,102,103 fibroblast differentiation in the left ventricle,103 and atrial remodelling,104 as well as cardiac hypertrophy via inhibition of extracellular signal-regulated kinase.105 In addition, porcine relaxin diminished renal fibrosis in a rat model of salt-sensitive hypertension34 and rat relaxin-3 ameliorated cardiac fibrosis in rats with isoproterenol-induced myocardial injury.90

atrial remodelling,104 as well as cardiac hypertrophy via inhibition of extracellular signal-regulated kinase.105 In addition, porcine relaxin diminished renal fibrosis in a rat model of salt-sensitive hypertension34 and rat relaxin-3 ameliorated cardiac fibrosis in rats with isoproterenol-induced myocardial injury.90 Inhibiting fibrosis and hypertrophy is likely to be beneficial in patients with AHF, and may be associated with reduced fibrosis in organs, including the heart, vessels, kidneys, and liver, as well as the limitation of organ damage and improvement of long-term prognosis.16,34,103 The anti-fibrotic effects of serelaxin may differentiate this agent from current treatments for AHF, such as nitrates, that do not protect end organs from further damage,2,21,22 and are therefore unlikely to inhibit tissue fibrosis. Serelaxin and stimulation of angiogenesis Using imaging techniques, significant reductions in perfused small microvessels have been demonstrated in tissues from patients with AHF106 compared with control subjects.107 In addition, the peripheral tissue oxygen extraction rate (an inverse index of tissue microvascular perfusion) is increased in patients with AHF compared with those with chronic stable HF.108 Of interest, this parameter improved with AHF therapy, in parallel with the amelioration of congestion and haemodynamic parameters.108 Therefore, alterations in microcirculation may play an important role in organ damage in AHF.

perfusion) is increased in patients with AHF compared with those with chronic stable HF.108 Of interest, this parameter improved with AHF therapy, in parallel with the amelioration of congestion and haemodynamic parameters.108 Therefore, alterations in microcirculation may play an important role in organ damage in AHF. Angiogenesis can facilitate tissue repair and serelaxin may mediate pro-angiogenic effects, unlike current treatments for AHF, as assessed in vitro and in animal models. Serelaxin has been reported to stimulate NO production from, and migration of, human endothelial progenitor cells in vitro, and to increase the number of circulating human endothelial progenitor cells and stimulate vascularization in mice.109 In addition, studies with H2 relaxin and serelaxin have observed increased expression of the angiogenic cytokine vascular endothelial growth factor (VEGF) in a cyclic adenosine monophosphate (cAMP)-dependent manner,110 stimulation of angiogenesis at ischaemic cardiac sites, and induction of expression of VEGF in rodents and pigs.85,111,112 This induction of angiogenesis could minimize further organ damage and repair injury, particularly of the myocardium, in patients with AHF.16 Serelaxin and effective protection of end-organs As previously mentioned, serelaxin treatment, in contrast to current therapies, interferes with the systemic and local mechanisms underlying the development of organ damage, and thus, may protect end organs in patients with AHF.3,21,22,75,85

Angiogenesis can facilitate tissue repair and serelaxin may mediate pro-angiogenic effects, unlike current treatments for AHF, as assessed in vitro and in animal models. Serelaxin has been reported to stimulate NO production from, and migration of, human endothelial progenitor cells in vitro, and to increase the number of circulating human endothelial progenitor cells and stimulate vascularization in mice.109 In addition, studies with H2 relaxin and serelaxin have observed increased expression of the angiogenic cytokine vascular endothelial growth factor (VEGF) in a cyclic adenosine monophosphate (cAMP)-dependent manner,110 stimulation of angiogenesis at ischaemic cardiac sites, and induction of expression of VEGF in rodents and pigs.85,111,112 This induction of angiogenesis could minimize further organ damage and repair injury, particularly of the myocardium, in patients with AHF.16 Serelaxin and effective protection of end-organs As previously mentioned, serelaxin treatment, in contrast to current therapies, interferes with the systemic and local mechanisms underlying the development of organ damage, and thus, may protect end organs in patients with AHF.3,21,22,75,85 Cardiac protection Early cardiomyocyte injury and stress and LV dysfunction result from AHF.95,113,114 Cardiomyocyte injury and loss can be detected by measuring troponin T levels, which are elevated in HF,60,61 while increased levels of NT-proBNP indicate ventricular wall stress.115 In patients with AHF, increased levels of troponin T may be detected upon hospital admission and in the 6–12 h following admission.86

,114 Cardiomyocyte injury and loss can be detected by measuring troponin T levels, which are elevated in HF,60,61 while increased levels of NT-proBNP indicate ventricular wall stress.115 In patients with AHF, increased levels of troponin T may be detected upon hospital admission and in the 6–12 h following admission.86 In vitro studies, animal models and clinical studies have investigated the cardioprotective properties of serelaxin and porcine relaxin. In vitro, administration of porcine relaxin has been reported to diminish IR injury in isolated reperfused guinea pig hearts, as determined by decreased calcium overload and MDA production,39 in addition to infarct size in a rat model of IR injury.74 Serelaxin treatment also reduced markers of cardiomyocyte damage, including troponin T, creatine kinase-MB, and myoglobin, as well as cardiac injury in pig models of IR injury.76,77 In patients with AHF, serelaxin (30 µg/kg/day for 20 or 48 h) decreased levels of troponin T and NT-proBNP.31,36 Similarly, NT-proBNP levels were diminished following serelaxin treatment (10–100 and 960 µg/kg/day for 24 h) in patients with chronic heart failure (CHF).35 These data imply that the unique mechanism of action of serelaxin may be associated with the preservation of cardiac function in patients with AHF. Although further assessment of this hypothesis is needed, this finding contrasts with the effects of nitrate treatment, which is thought to contribute to cardiac injury by reducing blood pressure and organ perfusion.2,22

tion of serelaxin may be associated with the preservation of cardiac function in patients with AHF. Although further assessment of this hypothesis is needed, this finding contrasts with the effects of nitrate treatment, which is thought to contribute to cardiac injury by reducing blood pressure and organ perfusion.2,22 In addition to protecting cardiomyocytes from injury and death, serelaxin has been reported to modulate ionic currents in cardiac cells.104,116 Although the translation of these findings into the clinic requires further studies, it is interesting to note that recently, in the RELAX-AHF study, serelaxin treatment reduced mortality from other CV causes and sudden deaths, without impact on HF deaths.117 Renal protection Renal dysfunction is common in patients with AHF62 and may be exacerbated by nitrate treatment, which can cause hypotension and subsequently, renal hypoperfusion and injury.2 Renal damage and dysfunction is a major predictor of poor outcomes in AHF27 and can be detected via increased levels of serum creatinine, cystatin C, uric acid, and blood urea nitrogen (BUN), as well as reduced estimated glomerular filtration rate (GFR).4,16,31,118 Elevated levels of serum creatinine, cystatin C, uric acid, and BUN have been reported in patients with AHF in the 48 h following hospital admission.31,119

ncreased levels of serum creatinine, cystatin C, uric acid, and blood urea nitrogen (BUN), as well as reduced estimated glomerular filtration rate (GFR).4,16,31,118 Elevated levels of serum creatinine, cystatin C, uric acid, and BUN have been reported in patients with AHF in the 48 h following hospital admission.31,119 Data from preclinical and clinical studies are available concerning the impact of serelaxin treatment on kidney function and protection. For example, in rats, serelaxin treatment increased GFR and renal blood flow, and protected against renal IR injury and glomerular dysfunction,41,47,51,52,71 whereas porcine relaxin decreased levels of creatinine and BUN in rats subjected to renal IR injury.75 In healthy subjects, serelaxin increased renal blood flow, but did not impact GFR,48 an effect also observed following administration of serelaxin (30 µg/kg/day for 24 h) in patients with CHF when compared with placebo, suggesting that serelaxin treatment reduces the increase in filtration fraction to mediate beneficial renal haemodynamic effects.49 In patients with AHF, serelaxin (30 µg/kg/day for 48 h) reduced levels of cystatin C, uric acid, BUN, and serum creatinine,31 and increased creatinine clearance (30 µg/kg/day for 20 h).36 Decreased serum creatinine was also reported after infusion of serelaxin (10–100 and 960 µg/kg/day for 24 h) in patients with CHF.35 Consequently, serelaxin may prevent worsening renal function, a property which differentiates this novel agent from vasodilator treatment in AHF.

nine clearance (30 µg/kg/day for 20 h).36 Decreased serum creatinine was also reported after infusion of serelaxin (10–100 and 960 µg/kg/day for 24 h) in patients with CHF.35 Consequently, serelaxin may prevent worsening renal function, a property which differentiates this novel agent from vasodilator treatment in AHF. Hepatic protection Hepatic injury and cell death can occur during AHF,58,120 with elevated markers of hepatic dysfunction, including AST and ALT, which are also predictors of mortality and worsening HF, reported within 48 h of hospitalization for AHF.31,121 Studies have demonstrated that serelaxin may mediate hepatic protection, as observed by diminished IR injury in rat liver72,73 and decreased levels of AST and ALT in patients with AHF following serelaxin treatment (30 µg/kg/day for 48 h).31 Vascular and other organ protection Damage to the vasculature and other organs may occur in patients with AHF57,122 and nitrate therapy may increase endothelial dysfunction further in these patients via increased oxidative stress.82

Hepatic protection Hepatic injury and cell death can occur during AHF,58,120 with elevated markers of hepatic dysfunction, including AST and ALT, which are also predictors of mortality and worsening HF, reported within 48 h of hospitalization for AHF.31,121 Studies have demonstrated that serelaxin may mediate hepatic protection, as observed by diminished IR injury in rat liver72,73 and decreased levels of AST and ALT in patients with AHF following serelaxin treatment (30 µg/kg/day for 48 h).31 Vascular and other organ protection Damage to the vasculature and other organs may occur in patients with AHF57,122 and nitrate therapy may increase endothelial dysfunction further in these patients via increased oxidative stress.82 Organ preservation and vasoprotective properties may distinguish serelaxin from classical vasodilators for the treatment of AHF and improve outcomes in these patients.16,123 For instance, treatment with serelaxin has been associated with improved endothelial function in rat aortic endothelial cells57 and decreases in vessel size, wall thickening, cross-sectional area, and collagen content in spontaneously hypertensive rats,124 while porcine relaxin has provided endothelial protection in a rat model of splanchnic IR injury.70 Furthermore, studies in the rat brain have shown that serelaxin treatment reduced ischaemic cell damage in brain slices, as well as infarct size in vivo, determined 4 h following ischaemia.125–127 In addition, administration of serelaxin has resulted in diminished IR injury in rat lungs.128,129

splanchnic IR injury.70 Furthermore, studies in the rat brain have shown that serelaxin treatment reduced ischaemic cell damage in brain slices, as well as infarct size in vivo, determined 4 h following ischaemia.125–127 In addition, administration of serelaxin has resulted in diminished IR injury in rat lungs.128,129 Conclusions and perspectives AHF poses a significant burden to patients and healthcare systems. The precise mechanisms underlying this condition are poorly understood, but it is clear that a variety of pathophysiological processes are involved, which result in both haemodynamic abnormalities and end-organ damage. Current therapies available for the treatment of AHF moderately address the haemodynamic changes associated with the short-term effects of this condition, to alleviate congestion. However, no currently approved agent has demonstrated true benefit on the long-term outcomes of AHF. As such, there is an unmet medical need in AHF; a need for therapies that address both the short- and long-term effects of this condition.

sociated with the short-term effects of this condition, to alleviate congestion. However, no currently approved agent has demonstrated true benefit on the long-term outcomes of AHF. As such, there is an unmet medical need in AHF; a need for therapies that address both the short- and long-term effects of this condition. Preclinical and clinical data have highlighted serelaxin as a promising treatment of both the short- and long-term consequences of AHF. In contrast to classical vasodilators, serelaxin may act at the vascular, cardiac, and renal level to improve haemodynamics and effectively relieve congestion. Moreover, available data suggest that serelaxin may provide organ protection via inhibition of inflammation, oxidative stress, cell death, and tissue fibrosis, and induction of angiogenesis (Figure 5),24,31,33,71–74,76,77,80,90 to improve the long-term prognosis of these patients, as observed in clinical trials to date. Figure 5 Serelaxin activates multiple pathways to improve haemodynamics and may protect cells and organs via anti-apoptotic/necrotic, anti-inflammatory, anti-fibrotic, antioxidant, and pro-angiogenic effects.24,31,33,71–74,76,77,80,90 ET, endothelin; ET-BR, endothelin receptor type B; MDA, malondialdehyde; MMP, metalloproteinase; NADPH, nicotinamide adenine dinucleotide phosphate-oxidase; NO, nitric oxide; NOS, nitric oxide synthase; TBARs, thiobarbituric acid-reactive substance; TGF-β, transforming growth factor β; TNF-α, tumour necrosis factor α; VEGF, vascular endothelial growth factor. Adapted from Teichman et al.24 Reproduced under the terms of the Creative Commons Attribution Non-commercial License for open access.

NOS, nitric oxide synthase; TBARs, thiobarbituric acid-reactive substance; TGF-β, transforming growth factor β; TNF-α, tumour necrosis factor α; VEGF, vascular endothelial growth factor. Adapted from Teichman et al.24 Reproduced under the terms of the Creative Commons Attribution Non-commercial License for open access. Additional clinical data are required to confirm the potential benefits of serelaxin for the treatment of AHF. A second phase III study, RELAX-AHF 2, began in September 2013 and will further assess the effects of serelaxin on CV mortality in patients with AHF.32 Future experimental research efforts should aim to establish animal models of AHF, in which the mechanisms underlying the efficacy of serelaxin for the treatment of this condition could be studied. Meanwhile, further preclinical studies are required to investigate the pharmacokinetic and pharmacodynamic properties of serelaxin in this patient population. Authors' Contributions J.D. and L.R. designed, jointly reviewed, and revised the initial draft and subsequent versions of the manuscript, and both agreed on the final version submitted for publication. Funding The writing/editorial support was funded by Novartis Pharma AG, Basel, Switzerland. The sponsor reviewed the initial draft and subsequent versions of the manuscript for own data accuracy and for proprietary evaluation. Funding to pay the Open Access publication charges for this article was provided by Novartis Pharma AG, Basel, Switzerland.

support was funded by Novartis Pharma AG, Basel, Switzerland. The sponsor reviewed the initial draft and subsequent versions of the manuscript for own data accuracy and for proprietary evaluation. Funding to pay the Open Access publication charges for this article was provided by Novartis Pharma AG, Basel, Switzerland. Conflict of interest: J.D. has served as an advisor and as a speaker for Novartis, Merck, Sharp and Dohme, and Abbvie. L.R. has served as an advisor and as a speaker for Novartis. Acknowledgements The authors thank Hannah Birchby and Rebecca Douglas (CircleScience, an Ashfield Company, part of UDG Healthcare plc), for providing writing/editorial assistance, which was funded by Novartis Pharma AG, Basel, Switzerland.

pooling The method used to combine results from individual studies, was based on the adjusted risk estimate and its 95% confidence intervals (CIs) obtained from each study. To obtain summary measures, a random effects model according to the DerSimonian Laird method11 was used because of the heterogeneity among studies. Sources of heterogeneity, evaluation and quantification Statistical heterogeneity was assessed with the Cochran’s Q test and its magnitude evaluated by the I2 statistics (I2 values of 25%, 50%, and 75% indicate low, moderate, and high heterogeneity, respectively).12 A series of sensitivity analyses were undertaken, including subgroup analyses and meta-regression to investigate potential sources of heterogeneity in the association between β-blocker treatment and mortality. Data were stratified according to the following study level variables; prospective vs. retrospective study design, statistical methods used to control for confounding (propensity score vs. multivariable analysis), and the following patient-level variables; country (Asia vs. US/Europe), AMI type (STEMI vs. STEMI/NSTEMI or unclear), and revascularization (only PCI treated patients vs. mixed or unclear). Subgroup analyses were extended by a random-effect meta-regression analysis that allowed the effect of the continuous covariates to be investigated (such as in years; median follow-up time and mean age, and in percent; LVEF, male sex, diabetes mellitus, hypertension, and smoking) as well as the categorical covariates used in the subgroup analysis. Meta-regression was performed to explore the influence of each covariate on the effect of β-blockers. If the covariate decreased the between-study variance, the source of heterogeneity was considered important. The estimate of τ2 in the presence of a covariate in comparison to that when the covariate is omitted allowed the proportion of the heterogeneity variance explained by the covariate to be calculated.13

Introduction Patients with atrial fibrillation (AF) are at increased risk of stroke.1 Warfarin and other vitamin K antagonists are effective treatments, reducing the risk of stroke by about two-thirds. The limitations with warfarin are a narrow therapeutic range, the need for monitoring, drug and food interactions, and risk of bleeding.2 In recent years, non-vitamin K antagonist oral anticoagulants (NOACs) (apixaban, dabigatran, and rivaroxaban) have been introduced as therapeutic alternatives to warfarin.3–5 NOACs are given at fixed doses and do not require regular monitoring. Information on the characteristics of patients being treated with NOACs in routine clinical practice in the early period after introduction is of interest to clinicians, in order to secure safe use of these drugs. In particular, there is an interest to know whether the outcomes observed in randomized clinical trials, especially the rates of bleeding events, are reflected in routine clinical practice, and whether there are differences between NOACs with regard to the risk of bleeding. Using two nationwide registries, we evaluated the bleeding outcomes in patients with AF being dispensed dabigatran, rivaroxaban, or apixaban compared with patients treated with warfarin.

nts, are reflected in routine clinical practice, and whether there are differences between NOACs with regard to the risk of bleeding. Using two nationwide registries, we evaluated the bleeding outcomes in patients with AF being dispensed dabigatran, rivaroxaban, or apixaban compared with patients treated with warfarin. Methods Data sources This study was based on data from two nationwide registries; the Norwegian Patient Registry (NPR) and the Norwegian Prescription Database (NorPD).6,7 The NPR was established in 2008 and contains all hospital visits (emergency visits, hospitalizations, and outpatient consultations), length of stay, and procedures (surgical and medical) from all hospitals in Norway. Diagnoses are coded according to the International Classification of Diseases, 10th revision (ICD10). Medical and surgical procedures are coded based on the Nordic Medico-Statistical Committee (NOMESCO) coding system. Both primary and subsequent codes related to each admission were taken into account in the analyses. The NorPD is a registry covering all prescriptions dispensed at pharmacies nationwide and data are available from 1 January 2004. Each medication is coded according to the anatomical therapeutic chemical system. The NorPD also includes information about date of dispensation, quantity and strength dispensed and the time of all-cause death. Any resident in Norway has a unique personal identifier that allows datasets to be merged on an individual level. The registry holder generated the datasets and released it in a coded and de-identified form, but with a unique identifier common to the two datasets making individual merging of the datasets possible. The two registries are mandatory in Norway and legally exempted from requirement of obtaining patient consent. The study was approved by the Regional Ethics Committee of Mid-Norway (Reference number 2015/162/REK midt).

th a unique identifier common to the two datasets making individual merging of the datasets possible. The two registries are mandatory in Norway and legally exempted from requirement of obtaining patient consent. The study was approved by the Regional Ethics Committee of Mid-Norway (Reference number 2015/162/REK midt). Study population The study included all patients ≥18 years diagnosed with non-valvular AF with at least one warfarin or NOAC dispensation in the study period (1 January 2013–30 June 2015), but being anti-coagulant naïve before start of the study. Non-valvular AF was defined in accordance with the updated American Heart Association/American College of Cardiology 2014 guidelines as AF in the absence of rheumatic mitral stenosis, a mechanical or bioprosthetic heart valve, or mitral valve repair.8 OAC naïve was defined as no OAC exposure in the preceding 180 days before index date. The index date was defined as the first dispensation of an OAC (warfarin 2.5 mg, dabigatran 110 or 150 mg, rivaroxaban 15 or 20 mg, and apixaban 2.5 or 5 mg) in the study period. Patients with venous thromboembolism during the last 180 days and those who had knee or hip replacement surgery during the last 35 days before starting OAC were excluded. A cohort creation flowchart is presented in Figure 1, and the study design in Figure 2. Figure 1 Cohort creation flow-chart. NorPD, Norwegian Prescription Database; NPR, Norwegian Patient Registry; NVAF, non-valvular atrial fibrillation; OAC, oral anticoagulant; VTE, venous thromboembolism.

starting OAC were excluded. A cohort creation flowchart is presented in Figure 1, and the study design in Figure 2. Figure 1 Cohort creation flow-chart. NorPD, Norwegian Prescription Database; NPR, Norwegian Patient Registry; NVAF, non-valvular atrial fibrillation; OAC, oral anticoagulant; VTE, venous thromboembolism. Figure 2 Study design. OAC index date was the date of the first OAC dispensation (warfarin, apixaban, rivaroxaban, dabigatran) in the study period (January 2013–June 2015). Each patient was followed from the index date to the date of discontinuation or switching of OAC therapy, date of death, or end of the study period. OAC, oral anticoagulant. Co-morbidities and co-medication (listed in Table 1) were retrieved from NPR and NorPD (see Supplementary material online, Table S1 for code definitions). We calculated CHA2DS2VASc (congestive heart failure, hypertension, age ≥65, diabetes, prior stroke/TIA, vascular disease, and female sex category) score9,10 for assessing stroke risk, and a modified HAS-BLED (hypertension, abnormal renal/liver function, stroke, bleeding history/predisposition, labile international normalised ratio (INR), elderly ≥65, and drugs/alcohol abuse) score11 as a measure of bleeding risk and a co-morbidity score (see Supplementary material online, Tables S2–S4 for definitions of scores). Table 1 Baseline characteristics of the study population according to OAC treatment

edisposition, labile international normalised ratio (INR), elderly ≥65, and drugs/alcohol abuse) score11 as a measure of bleeding risk and a co-morbidity score (see Supplementary material online, Tables S2–S4 for definitions of scores). Table 1 Baseline characteristics of the study population according to OAC treatment Warfarin Dabigatran Rivaroxaban Apixaban Number of patients 11 427 7925 6817 6506 Men 6737 (59.0) 4915 (62.0) 3711 (54.4) 3579 (55.0) Age, years Mean (SE) 74.6 (11.9) 70.8 (11.3) 74.7 (10.7) 74.5 (11.1) Median (25th–75th percentile) 76 (67–84) 71 (64–79) 75 (68–83) 75 (68-83) ≥75 years 6248 (54.7) 2967 (37.4) 3524 (51.7) 3295 (50.6) Medical history Chronic kidney disease 569 (5.0) 58 (0.73) 135 (2.0) 163 (2.5) Chronic heart failure 3316 (29.0) 1250 (15.8) 1388 (20.4) 1341 (20.6) Diabetes 1674 (14.7) 822 (10.4) 794 (11.7) 797 (12.3) Stroke, TIA, and thromboembolism 1329 (11.6) 745 (9.4) 1096 (16.1) 905 (13.9) Ischaemic heart disease 4102 (35.9) 1699 (21.4) 1736 (25.5) 1795 (27.6) Previous bleeding hospitalization 1922 (16.8) 890 (11.2) 1009 (14.8) 982 (15.1) Previous OAC (>180 days prior to index) 2910 (25.5) 900 (11.4) 748 (11.0) 527 (8.1) Active cancer (last year) 1145 (10.0) 589 (7.4) 625 (9.2) 562 (8.6) COPD 1064 (9.3) 518 (6.5) 580 (8.5) 567 (8.7) Hypertension 7654 (67.0) 4677 (59.0) 4500 (66.0) 4254 (65.4) Anaemia (last year) 553 (4.8) 155 (2.0) 203 (3.0) 201 (3.1) Viral hepatitis 25 (0.22) 16 (0.20) 7 (0.10) 11 (0.17) Hospital admission last year 7734 (67.7) 4422 (55.8) 4460 (65.4) 4412 (67.8) Co-medication Low-dose aspirin (last year) 5420 (47.4) 3687 (46.5) 3621 (53.1) 3304 (50.8) NSAID (last year) 2264 (19.8) 1937 (24.4) 1583 (23.2) 1498 (23.0) Non-aspirin anti-platelet inhibitors (last year) 278 (2.4) 185 (2.3) 231 (3.4) 189 (2.9) Risk scores Modified HAS-BLED score ≥ 3 4894 (42.8) 2934 (37.0) 3206 (47.0) 3029 (46.6) CHA2DS2-VASc score Mean 3.09 2.46 2.94 2.93 ≥2 9449 (82.7) 5785 (73.0) 5709 (83.7) 5411 (83.2) Co-morbidity score ≥ 1 7527 (65.6) 3851 (48.6) 4124 (60.5) 3916 (60.2) Reduced NOAC dose at index date NA 2758 (34.8) 1824 (26.8) 1901 (29.2) Values are numbers (percentages) unless otherwise stated.

) 3206 (47.0) 3029 (46.6) CHA2DS2-VASc score Mean 3.09 2.46 2.94 2.93 ≥2 9449 (82.7) 5785 (73.0) 5709 (83.7) 5411 (83.2) Co-morbidity score ≥ 1 7527 (65.6) 3851 (48.6) 4124 (60.5) 3916 (60.2) Reduced NOAC dose at index date NA 2758 (34.8) 1824 (26.8) 1901 (29.2) Values are numbers (percentages) unless otherwise stated. COPD, chronic obstructive lung disease; NSAID, non-steroidal anti-inflammatory drug; OAC, oral anticoagulant; NOAC, non-vitamin K oral anticoagulant; SE, standard error; TIA, transitoric ischaemic attack.

) 3206 (47.0) 3029 (46.6) CHA2DS2-VASc score Mean 3.09 2.46 2.94 2.93 ≥2 9449 (82.7) 5785 (73.0) 5709 (83.7) 5411 (83.2) Co-morbidity score ≥ 1 7527 (65.6) 3851 (48.6) 4124 (60.5) 3916 (60.2) Reduced NOAC dose at index date NA 2758 (34.8) 1824 (26.8) 1901 (29.2) Values are numbers (percentages) unless otherwise stated. COPD, chronic obstructive lung disease; NSAID, non-steroidal anti-inflammatory drug; OAC, oral anticoagulant; NOAC, non-vitamin K oral anticoagulant; SE, standard error; TIA, transitoric ischaemic attack. Definition of bleeding events and endpoints of the study Bleeding was defined as all bleeding events recorded in NPR between index date and 30 days after the calculated end of OAC supply. Bleeding events were categorized as major or clinically relevant non-major (CRNM) bleeding based on available information from NPR. Major bleeding was defined as any bleeding event which occurred in a critical area or organ or any bleeding event that was accompanied by blood transfusion ≤10 days after hospital admission date. This is a slight modification of the International Society on Thrombosis and Haemostasis (ISTH) classification of major bleeding12 because no information was available in our data set on haemoglobin levels. A CRNM bleeding was defined in accordance with the ISTH classification12 as any bleeding requiring medical intervention by a health care professional, leading to hospitalization or increased level of care or prompting a face-to-face evaluation, that did not fit the criteria for major bleeding. The bleeding events were also categorized by organ system into gastrointestinal (GI), intracranial (ICH), or bleeding from other sites. Bleeding endpoints took into account all bleeds with the pre-specified ICD10 codes and were not restricted to admissions with bleeding as the primary (first) code. The primary endpoint of the study was a composite of major or CRNM bleeding. Secondary endpoints were major bleeding, CRNM bleeding, GI bleeding, ICH, and bleeds from other organ systems. See Supplementary material online, Table S5 for further details on bleeding codes.

ssions with bleeding as the primary (first) code. The primary endpoint of the study was a composite of major or CRNM bleeding. Secondary endpoints were major bleeding, CRNM bleeding, GI bleeding, ICH, and bleeds from other organ systems. See Supplementary material online, Table S5 for further details on bleeding codes. Oral anticoagulant supply For each dispensation, the OAC days of supply were computed using information on date of dispensation, the number of packages, and the pack-size dispensed. As NOACs are prescribed in a fixed dose, the number of days of supply strictly corresponds to amount dispensed. The NorPD contains information on tablet strength, pack-size and number of packages dispensed, and we assumed, according to the labelling, twice daily dosing for apixaban and dabigatran and once daily dosing for rivaroxaban, e.g. a patient supplied one package of a 60 tablet package of apixaban will have a supply lasting for 30 days whereas a rivaroxaban patient supplied one 100 tablet package will have a supply lasting 100 days. Computing the warfarin supply is not straightforward as we lack information on both dosing instructions and international normalized reference values. We therefore first calculated a median mg/day for all patients using warfarin in the study period (4.688 mg/day) and subsequently used this in the computation of warfarin supply for each dispensation, e.g. a patient dispensed one 100 tablet package of 2.5 mg strength will have a supply lasting for 53 days. We also needed to set the end of OAC supply date during the pre-index period to be able to determine whether a patient was OAC naïve or not (≥180 days without OAC supply prior to index date). We repeated the procedure for all warfarin dispensations during the pre-index period (median mg/day was estimated to 4.388 mg/day) and used this to estimate end of supply for each warfarin dispensation. To account for incomplete adherence, a gap period of 30 days within the calculated end of OAC supply was allowed. A patient continued treatment if next dispensation for the same OAC was within the 30 days after the calculated end of OAC supply. A patient switched treatment if another OAC was dispensed within 30 days after the calculated end of supply and finally the patient discontinued index OAC treatment if next OAC dispensation was more than 30 days after the calculated end of supply. Patients were censored if discontinuing or switching OAC, death, or end of follow-up, whichever occurred first.

OAC was dispensed within 30 days after the calculated end of supply and finally the patient discontinued index OAC treatment if next OAC dispensation was more than 30 days after the calculated end of supply. Patients were censored if discontinuing or switching OAC, death, or end of follow-up, whichever occurred first. Statistical analysis Cox proportional hazard regression analyses were conducted to determine the risk of bleeding for the different NOACs vs. warfarin, both unadjusted and adjusted for known patient characteristics: age, gender, previous bleeding, previous OAC use, co-morbidities, and concomitant medications at baseline. Hence, the independent exposure of interest was which OAC patients used (with warfarin as the reference drug). Age was the only continuous patient characteristic. The linear assumption was checked by considering a model for the time to bleeding as a function of age, where the function was allowed to be non-linear (using splines). A final model was selected by backwards stepwise selection, using the Akaike information criterion as a measure of model fit. Each bleeding endpoint was compared with the entire cohort and not in contrast to non-bleeders only, e.g. for the major bleeding endpoint the comparison was with all non-major bleedings. The continuous variable (age) was described by the mean, standard deviation, median, and first and third quartiles. Categorical variables were described by the number and percentage of patients in each category. Crude incidence rates (IR) were also calculated as first bleeding episode per 100 person-years. Relative risks were given as hazard ratios (HRs) with 95% confidence intervals (CIs). Post hoc subgroup analyses, for the primary endpoint of major or CRNM bleeding, were performed for elderly patients (≥75 year) as well as for OAC dose levels at index date (standard and reduced dose) in comparison with warfarin. Power calculation was based on reported annual rates of major or CRNM bleeding in the pivotal clinical trials.3–5 These calculations indicated that with a sample size of approximately 2000 apixaban patients (apixaban was chosen because it had the shortest exposure time among the NOACs at the time of planning) and with a minimum of 1 year of follow-up, there would be acceptable levels of power (70–80%) in the comparison with warfarin. All statistical tests were two-tailed and P-values  <0.05 were considered significant.

ts (apixaban was chosen because it had the shortest exposure time among the NOACs at the time of planning) and with a minimum of 1 year of follow-up, there would be acceptable levels of power (70–80%) in the comparison with warfarin. All statistical tests were two-tailed and P-values  <0.05 were considered significant. Statistical analyses were performed using R (version 3.1.1, R Development Core Team).13 Results The study population comprised a total of 32 675 patients. The mean age was 73.6 years (median 74 years) and 58% were males. Baseline characteristics in relation to the type of OAC being prescribed are presented in Table 1. Patients treated with dabigatran were younger, more likely to be men, had a lower co-morbidity load and lower baseline risk for stroke (CHA2DS2-VASc score) and bleeding (modified HAS-BLED) than patients treated with the other OACs. Warfarin patients had to a larger extent previously been exposed to OAC (>180 days prior to index) compared with the NOAC-treated patients. Apixaban- and rivaroxaban-treated patients had higher baseline risk of bleeding (modified HAS-BLED) compared with the other OACs.

S-BLED) than patients treated with the other OACs. Warfarin patients had to a larger extent previously been exposed to OAC (>180 days prior to index) compared with the NOAC-treated patients. Apixaban- and rivaroxaban-treated patients had higher baseline risk of bleeding (modified HAS-BLED) compared with the other OACs. The median follow-up time was as follows: warfarin 156 (25th, 75th percentile; 84, 309) days, dabigatran 212 (97, 413) days, rivaroxaban 209 (105, 410) days, and apixaban 143 (73, 247) days. A total of 2081 (6.37%) patients experienced a first major or CRNM bleeding episode; 419 patients (1.28%) experienced a major bleeding and 1662 patients (5.09%) a CRNM bleeding. By organ system, 594 patients (1.82%) experienced a GI bleeding, 207 patients (0.63%) an ICH, and 1280 patients (3.92%) experienced bleeding in other sites. Number and percentages of first time bleeding events for the different OACs are presented in Table 2. Table 2 Oral anticoagulant follow-up time and number (percentage) of patients experiencing a first time bleeding episode after initiating oral anticoagulant (subsequent bleeding episodes not considered) for the different bleeding endpoints

s of first time bleeding events for the different OACs are presented in Table 2. Table 2 Oral anticoagulant follow-up time and number (percentage) of patients experiencing a first time bleeding episode after initiating oral anticoagulant (subsequent bleeding episodes not considered) for the different bleeding endpoints Warfarin (n = 11 427) Dabigatran (n = 7925) Rivaroxaban (n = 6817) Apixaban (n = 6506) Total (n = 32 675) Follow-up time (days), median (25th–75th percentile) 156 (84–309) 212 (97–413) 209 (105–410) 143 (73–247) 173 (84–340) Major or CRNM bleeding: 824 (7.21) 407 (5.14) 578 (8.48) 272 (4.18) 2081 (6.37) Severity Major bleeding 181 (1.58) 80 (1.01) 109 (1.60) 49 (0.75) 419 (1.28) CRNM bleeding 643 (5.63) 327 (4.13) 469 (6.88) 223 (3.43) 1662 (5.09) Organ system GI bleeding 199 (1.74) 150 (1.89) 175 (2.57) 70 (1.08) 594 (1.82) ICH bleeding 90 (0.79) 28 (0.35) 63 (0.92) 26 (0.40) 207 (0.63) Other bleeding 535 (4.68) 229 (2.89) 340 (4.99) 176 (2.71) 1280 (3.92) CRNM, Clinically relevant non-major; GI, gastrointestinal; ICH, intracranial haemorrhage.

.43) 1662 (5.09) Organ system GI bleeding 199 (1.74) 150 (1.89) 175 (2.57) 70 (1.08) 594 (1.82) ICH bleeding 90 (0.79) 28 (0.35) 63 (0.92) 26 (0.40) 207 (0.63) Other bleeding 535 (4.68) 229 (2.89) 340 (4.99) 176 (2.71) 1280 (3.92) CRNM, Clinically relevant non-major; GI, gastrointestinal; ICH, intracranial haemorrhage. Crude IR for first bleeding events on warfarin and NOACs, and Forest plots showing the adjusted HRs for first bleeding episode for dabigatran, rivaroxaban and apixaban compared with warfarin, are presented in Figure 3. In Supplementary material online, Table S6, using the primary endpoint of major or CRNM bleeding, HRs and P-values for background variables for four different models are provided: (i) one variable at the time model; (ii) an age and OAC adjusted model; (iii) a model based on age, gender, OAC, and risk scores (CHA2DS2VASc, HAS-BLED, and co-morbidity scores); and (iv) the chosen final and optimal fitted Cox regression model including all variables (not risk scores) using the backward stepwise elimination function in R. Supplementary material online, Table S7 gives HRs for the endpoint of major or CRNM bleeding using apixaban as reference instead of warfarin. Figure 3 Forest plots showing the adjusted hazard ratios for first bleeding episode for dabigatran, rivaroxaban, and apixaban compared with warfarin. (A) Major or CRNM bleeding. (B) GI bleeding, ICH bleeding, and bleeding from other sites. Crude IR for first bleeding episode are given as events per 100 person-years. CI, confidence interval; CRNM, clinically relevant non-major bleeding; GI, gastrointestinal; HR, adjusted hazard ratio; ICH, intracranial haemorrhage; IR, incidence rate; OAC, oral anticoagulant.

, ICH bleeding, and bleeding from other sites. Crude IR for first bleeding episode are given as events per 100 person-years. CI, confidence interval; CRNM, clinically relevant non-major bleeding; GI, gastrointestinal; HR, adjusted hazard ratio; ICH, intracranial haemorrhage; IR, incidence rate; OAC, oral anticoagulant. After adjusting for differences in baseline characteristics, both dabigatran (adjusted HR 0.74, 95% CI 0.66–0.84, P < 0.001) and apixaban (adjusted HR 0.70, 95% CI 0.61–0.80, P < 0.001) were associated with a significant lower risk of major or CRNM bleeding compared with warfarin. There was no significant difference in major or CRNM bleeding risk between rivaroxaban and warfarin (adjusted HR 1.05, 95% CI 0.94–1.17, P = 0.400). A time-restricted major or CRNM bleeding analysis with a cut-off at 180 days showed that dabigatran (adjusted HR 0.79, 95% CI 0.68–0.92, P = 0.002) and apixaban (adjusted HR 0.72, 95% CI 0.62–0.85, P < 0.001) both were associated with a significant lower risk of major or CRNM bleeding compared with warfarin. There was no significant difference in major or CRNM bleeding risk between rivaroxaban and warfarin (adjusted HR 1.06, 95% CI 0.93–1.21, P = 0.400). Dabigatran (adjusted HR 1.26, 95% CI 1.01–1.57, P = 0.037) and rivaroxaban (adjusted HR 1.37, 95% CI 1.12–1.69, P = 0.003) use were both associated with a higher risk of GI bleeding compared with warfarin. There was no significant difference in the risk of GI bleeding using apixaban compared with warfarin (adjusted HR 0.77, 95% CI 0.59–1.02, P = 0.068).

I 1.01–1.57, P = 0.037) and rivaroxaban (adjusted HR 1.37, 95% CI 1.12–1.69, P = 0.003) use were both associated with a higher risk of GI bleeding compared with warfarin. There was no significant difference in the risk of GI bleeding using apixaban compared with warfarin (adjusted HR 0.77, 95% CI 0.59–1.02, P = 0.068). The risk of ICH was lower in patients using dabigatran (adjusted HR 0.46, 95% CI 0.30–0.70, P < 0.001) and apixaban (adjusted HR 0.56, 95% CI 0.36–0.86, P = 0.009) compared with warfarin, but there was no significant difference between patients using rivaroxaban vs. warfarin (adjusted HR 0.93, 95% CI 0.67–1.29, P = 0.656).

risk of ICH was lower in patients using dabigatran (adjusted HR 0.46, 95% CI 0.30–0.70, P < 0.001) and apixaban (adjusted HR 0.56, 95% CI 0.36–0.86, P = 0.009) compared with warfarin, but there was no significant difference between patients using rivaroxaban vs. warfarin (adjusted HR 0.93, 95% CI 0.67–1.29, P = 0.656). In the total population, 35% of dabigatran (n = 2758), 27% of rivaroxaban (n = 1824), and 29% of apixaban patients (n = 1901) initiated treatment on the reduced dose for stroke prevention (e.g. dabigatran 110 mg twice daily, rivaroxaban 15 mg once daily, or apixaban 2.5 mg twice daily) (Table 1). As many as 82% of patients receiving the reduced dose were ≥75 years, they were more likely to have other comorbidities as chronic kidney disease, hypertension and/or heart failure, and also more likely to have CHA2DS2-VASc score ≥2 and HAS-BLED score ≥3 (Supplementary material online, Table S8). A subgroup analysis of major or CRNM bleeding for the reduced and standard doses of each NOAC compared with warfarin is presented in Figure 4. Both the standard and reduced doses of apixaban and dabigatran were associated with a significant reduction in the primary endpoint of major or CRNM bleeding compared with warfarin. With respect to rivaroxaban, neither the reduced nor the standard dose was significantly different from warfarin. Figure 4 Risk of major or CRNM bleeding for the reduced and standard dose of dabigatran, rivaroxaban, and apixaban compared with warfarin. Crude IR for first bleeding episode are given as events per 100 person-years. CI, confidence interval; CRNM, clinically relevant non-major bleeding; HR, adjusted hazard ratio; IR, incidence rate; OAC, oral anticoagulant.

for the reduced and standard dose of dabigatran, rivaroxaban, and apixaban compared with warfarin. Crude IR for first bleeding episode are given as events per 100 person-years. CI, confidence interval; CRNM, clinically relevant non-major bleeding; HR, adjusted hazard ratio; IR, incidence rate; OAC, oral anticoagulant. A total of 16 034 patients (49%) were ≥75 years (Table 1). In the subgroup of these elderly patients, dabigatran (adjusted HR 0.84, 95% CI 0.72–0.99, P = 0.036) and apixaban (adjusted HR 0.72, 95% CI 0.61–0.86, P < 0.001) were still associated with a lower risk of major or CRNM bleeding compared with warfarin, whereas rivaroxaban remained insignificant compared with warfarin (adjusted HR 1.14, 95% CI 0.99–1.30, P = 0.067) (Figure 5). Figure 5 Risk of major or CRNM bleeding for dabigatran, rivaroxaban, and apixaban compared with warfarin in the subgroup of patients ≥75 years. Crude IR for first bleeding episode are given as events per 100 person-years. CI, confidence interval; CRNM, clinically relevant non-major bleeding; HR, adjusted hazard ratio; IR, incidence rate; OAC, oral anticoagulant.

dabigatran, rivaroxaban, and apixaban compared with warfarin in the subgroup of patients ≥75 years. Crude IR for first bleeding episode are given as events per 100 person-years. CI, confidence interval; CRNM, clinically relevant non-major bleeding; HR, adjusted hazard ratio; IR, incidence rate; OAC, oral anticoagulant. Discussion In this large nationwide cohort of 32 675 patients (median age 74 years) with AF initiating OAC, the adjusted risk of major or CRNM bleeding was lower in patients treated with dabigatran and apixaban compared with patients treated with warfarin. In patients treated with rivaroxaban, the risk of major or CRNM bleeding was not significantly different from that of warfarin. For organ system divided analyses, this study demonstrated a similar risk of GI bleeding with apixaban compared with warfarin, whereas both dabigatran and rivaroxaban were associated with a higher risk of GI bleeding. Treatment with dabigatran and apixaban were both associated with a lower risk of ICH compared with warfarin, whereas treatment with rivaroxaban was not.

nstrated a similar risk of GI bleeding with apixaban compared with warfarin, whereas both dabigatran and rivaroxaban were associated with a higher risk of GI bleeding. Treatment with dabigatran and apixaban were both associated with a lower risk of ICH compared with warfarin, whereas treatment with rivaroxaban was not. Our findings correspond with the bleeding outcomes reported in the pivotal outcome trials of the NOACs. A lower risk of bleeding with apixaban compared with warfarin was found in the ARISTOTLE trial.3,14,15 For dabigatran, our results are similar to the findings in the RE-LY trial of a higher risk of GI bleeding and a lower risk of ICH compared with warfarin, but differ with respect to the primary bleeding endpoint where we demonstrated a significantly lower risk of major or CRNM bleeding for both doses of dabigatran compared with warfarin.4 The higher risk of GI bleeding with rivaroxaban compared with warfarin was in line with the results of the ROCKET trial.5

th warfarin, but differ with respect to the primary bleeding endpoint where we demonstrated a significantly lower risk of major or CRNM bleeding for both doses of dabigatran compared with warfarin.4 The higher risk of GI bleeding with rivaroxaban compared with warfarin was in line with the results of the ROCKET trial.5 A few observational studies assessing the comparative effectiveness and safety of dabigatran, rivaroxaban, and apixaban in comparison with warfarin in routine clinical practice have recently been reported. One recent study compared dabigatran and warfarin using US Medicare data and demonstrated a reduced risk of ischaemic stroke (dabigatran HR 0.80, 95% CI 0.67–0.96) and ICH (dabigatran HR 0.34, 95% CI 0.26–0.46), and an increased risk of major GI bleeding (dabigatran HR 1.28, 95% CI 1.14–1.44), with dabigatran compared with warfarin.16 Another study using a large US insurance database demonstrated that apixaban was associated with a lower risk of stroke or systemic embolism (apixaban HR 0.67, 95% CI 0.46–0.98, P = 0.04) and that dabigatran and apixaban were associated with a lower risk of major bleeding (dabigatran HR 0.79, 95% CI 0.67–0.94, P < 0.01; apixaban HR: 0.45, 95%CI: 0.34–0.59, P < 0.001).17 Interestingly, the findings on GI bleeding were in line with our results, showing a higher risk of GI bleeding in dabigatran- and rivaroxaban-treated patients compared with warfarin, whereas for apixaban the risk of GI bleeding was similar to warfarin. The study by Larsen et al.18 using nationwide Danish registries showed no significant difference in the rate of ischaemic stroke between any of the NOACs and warfarin. Dabigatran and apixaban were both associated with a statistically significant lower risk of major bleeding compared with warfarin, also in line with our findings. However, the study population was restricted to patients with standard doses of NOAC (dabigatran 150 mg twice daily, rivaroxaban 20 mg once daily, or apixaban 5 mg twice daily) and the study population was also younger (median age 71 years) compared with ours. Comparing the findings from our study with the reported pivotal outcome trials as well as observational studies must be interpreted with caution as there are differences in study populations, bleeding definitions and health care systems as well as other factors that are difficult to account for.

rs) compared with ours. Comparing the findings from our study with the reported pivotal outcome trials as well as observational studies must be interpreted with caution as there are differences in study populations, bleeding definitions and health care systems as well as other factors that are difficult to account for. A high proportion of patients (27–35% of patients) in our study initiated NOAC therapy on the reduced dose for stroke prevention (e.g. dabigatran 110 mg twice daily, rivaroxaban 15 mg once daily, or apixaban 2.5 mg twice daily). Due to lack of information on creatinine levels, weight and bleeding diathesis, we do not know how many of these patients that fulfilled the criteria for dose reduction for stroke prevention for the different NOACs. However, 82% of these patients were ≥75 years, and also had a high baseline risk profile with respect to bleeding.

k of information on creatinine levels, weight and bleeding diathesis, we do not know how many of these patients that fulfilled the criteria for dose reduction for stroke prevention for the different NOACs. However, 82% of these patients were ≥75 years, and also had a high baseline risk profile with respect to bleeding. There was a difference in age distribution between the OACs at baseline with the proportion of patients being  ≥75 years ranging from 55% for warfarin, 52% for rivaroxaban, 51% for apixaban, and 37% for dabigatran. This distribution is similar to what was seen in the US-based study by Yao et al.17 Notably, these data suggest that a higher proportion of NOAC-treated patients in routine clinical practice are ≥75 years compared with the pivotal outcome trials; e.g. in the ARISTOTLE trial only 31% of patients were reported being ≥75 years.15 Age is an established and strong predictor of increased bleeding risk in OAC-treated patients with higher absolute risks of bleeding reported among the elderly; however, the relative bleeding risk of NOACs in comparison to warfarin in the elderly was not markedly different from the overall population (Figure 5).

ars.15 Age is an established and strong predictor of increased bleeding risk in OAC-treated patients with higher absolute risks of bleeding reported among the elderly; however, the relative bleeding risk of NOACs in comparison to warfarin in the elderly was not markedly different from the overall population (Figure 5). Strength and limitations The strength of our study is that it retrieves data from mandatory and nationwide registries in a public health care system that covers all residents. As a result, the dataset at hand gave us a complete picture of all hospitalizations and prescriptions dispensed nationwide for the entire study period. This complete coverage of data eliminates also selection bias and recall bias that is an apparent problem using other databases being based on selected hospitals, health insurance schemes, or self-reported questionnaires. An obvious limitation of this study is that we have not considered effectiveness in stroke prevention. As stroke events occur less frequently than overall bleeding complications of OAC treatment, and also considering the still early days of NOAC exposure, this study was not planned for, and deemed sufficiently powered to, evaluating NOAC effectiveness on stroke outcomes. More research is therefore needed to address the bleeding complications of NOACs in the context of stroke prevention in an unselected clinical practice setting.

ng the still early days of NOAC exposure, this study was not planned for, and deemed sufficiently powered to, evaluating NOAC effectiveness on stroke outcomes. More research is therefore needed to address the bleeding complications of NOACs in the context of stroke prevention in an unselected clinical practice setting. Although we have adjusted for baseline differences, we are unlikely to have captured the full extent and effect of different prescribing behaviour, especially in this early phase of NOAC introduction, and some unmeasured and residual confounding is undoubtedly still present. With the exception of apixaban being granted general reimbursement 6 months after rivaroxaban and dabigatran, the same conditions for OAC prescribing were valid nationwide and throughout the study period. The time-restricted 180-day bleeding risk analysis gives support to our chosen approach and that results are robust irrespective of differences in OAC follow-up time. We did not have access to information on time in therapeutic range among warfarin users; nor did we have information on laboratory tests and other characteristics such as smoking and weight. One other caveat that influences the external validity of the results is that the AF diagnosis was retrieved from the hospital level only, meaning that AF patients that were solely managed in primary care were not included in the study. Apart from co-medication, co-morbidities from primary care can be underrepresented. There is also a risk of misclassification related to coding errors of hospital admissions; however for serious conditions like bleeding this is not very likely. No formal validation studies of the AF diagnosis in NPR against health records have been conducted. We studied drug exposure at the level of pharmacy dispensation and have no information on patient’s real OAC intake.

ng errors of hospital admissions; however for serious conditions like bleeding this is not very likely. No formal validation studies of the AF diagnosis in NPR against health records have been conducted. We studied drug exposure at the level of pharmacy dispensation and have no information on patient’s real OAC intake. Conclusion In this nationwide cohort study on AF patients being prescribed OAC, use of apixaban and dabigatran were associated with a lower risk of major or CRNM bleeding compared with the use of warfarin. The risk of GI bleeding was higher among users of dabigatran and rivaroxaban compared with warfarin, whereas users of apixaban and dabigatran had a lower risk of ICH compared with users of warfarin. The risk of stroke was not addressed in this study, and hence the optimal benefit to risk balance between stroke prevention and bleeding could not be evaluated. Disclaimer Data from the Norwegian Patient Registry (NPR) have been used in this publication. The interpretation and reporting of these data are the sole responsibility of the authors, and no endorsement by the NPR is intended nor should be inferred. Supplementary material Supplementary material is available at European Heart Journal - Cardiovascular Pharmacotherapy online. Supplementary material online, Table S1 Funding This study was sponsored by Pfizer Inc., New York, NY, USA.

Disclaimer Data from the Norwegian Patient Registry (NPR) have been used in this publication. The interpretation and reporting of these data are the sole responsibility of the authors, and no endorsement by the NPR is intended nor should be inferred. Supplementary material Supplementary material is available at European Heart Journal - Cardiovascular Pharmacotherapy online. Supplementary material online, Table S1 Funding This study was sponsored by Pfizer Inc., New York, NY, USA. Conflict of interest: S.H. reports personal fees from Amgen, Astra Zeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Pfizer, Sanofi, Merck, Novartis. C.J. reports personal fees from Pfizer, during the conduct of the study; personal fees from Pfizer and Bayer, outside the submitted work. I.F.T. reports grant from Pfizer, during the conduct of the study. P.F. is an employee of Pfizer. O.S. is an employee of Pfizer. C.H. is an employee of Bristol-Myers Squibb. W.G. reports grants and personal fees from Bayer, personal fees from Pfizer, Boehringer Ingelheim and Novartis, grants from Roche, outside the submitted work.

Introduction Unlike vitamin K antagonists (VKAs) such as warfarin that reduce hepatic synthesis of various clotting factors, non-vitamin-K oral anticoagulants (NOACs, also called direct oral anticoagulants, DOACs) act on specific factors within the coagulation cascade; dabigatran (given as the prodrug dabigatran etexilate, Pradaxa) inhibits thrombin while apixaban (Eliquis), edoxaban (Lixiana), and rivaroxaban (Xarelto) inhibit factor Xa.1–4 (see Table 1) Table 1 Non-vitamin-K oral anticoagulants

rect oral anticoagulants, DOACs) act on specific factors within the coagulation cascade; dabigatran (given as the prodrug dabigatran etexilate, Pradaxa) inhibits thrombin while apixaban (Eliquis), edoxaban (Lixiana), and rivaroxaban (Xarelto) inhibit factor Xa.1–4 (see Table 1) Table 1 Non-vitamin-K oral anticoagulants Apixaban (Eliquis) Dabigatran etexilate (Pradaxa) Edoxaban (Lixiana) Rivaroxaban (Xarelto) Date authorized in the EU 18 May 2011 18 March 2008 19 June 2015 30 September 2008 Indications in adultsa Prevention of stroke and systemic embolism with non-valvular atrial fibrillation with additional risk factors; Treatment of deep-vein thrombosis and pulmonary embolism and prevention of recurrence of these conditions; Prevention of thromboembolism following total hip or total knee replacement surgery Prevention of stroke and systemic embolism with non-valvular atrial fibrillation with additional risk factors; Treatment of deep-vein thrombosis and pulmonary embolism and prevention of recurrence of these conditions; Prevention of thromboembolism following total hip or total knee replacement surgery Prevention of stroke and systemic embolism with non-valvular atrial fibrillation with additional risk factors; Treatment of deep-vein thrombosis and pulmonary embolism and prevention of recurrence of these conditions Prevention of stroke and systemic embolism with non-valvular atrial fibrillation with additional risk factors; Treatment of deep-vein thrombosis and pulmonary embolism and prevention of recurrence of these conditions; Prevention of thromboembolism following total hip or total knee replacement surgery; Adjunct for prevention of atherothrombotic events after acute coronary syndrome Target: Factor Xa Thrombin Factor Xa Factor Xa Bioavailability (%): 50% 3-7% 62% 80–100% for 2.5 and 10 mg doses; 66% for 15 and 20 mg doses Prodrug: No Yes—activated by esterase (CES1) No No Half-life (hours): 8–15 11–13 10–14 5–13 Tmax (hours): 3–4 0.5–2 1–2 2–4 Renal clearance: 25% 80% 50% 33% Substrate of: P-gp; CYP3A4/5, CYP1A2, CYP2J2 P-gp P-gp; CYP3A4/5 P-gp; CYP3A4, CYP2J2 Protein binding: 87% 35% 55% >90% Effects of food: Tmax delayed; Cmax & AUC unchanged Tmax delayed; Cmax & AUC unchanged Food increases peak exposure to a varying extent, but has minimal effect on total exposure.

ance: 25% 80% 50% 33% Substrate of: P-gp; CYP3A4/5, CYP1A2, CYP2J2 P-gp P-gp; CYP3A4/5 P-gp; CYP3A4, CYP2J2 Protein binding: 87% 35% 55% >90% Effects of food: Tmax delayed; Cmax & AUC unchanged Tmax delayed; Cmax & AUC unchanged Food increases peak exposure to a varying extent, but has minimal effect on total exposure. Tmax delayed; Cmax and AUC unchanged at the 2.5 mg and 10 mg dose, less marked decrease than fasting state at higher doses (the 15 and 20 mg doses should be given with food to increase availability) a Abbreviated wording of authorized indications; refer to the summary of product characteristics for the full indications In contrast to VKAs, no routine coagulation monitoring is required in patients taking NOACs.1–4 This relieves both patients and health services of the burden of regular blood tests. However, dosing of NOACs must take into account factors such as patient age, renal function and accompanying haemorrhagic risk.1–5 Efficacy and safety using this dosing paradigm have been shown in a series of large clinical trials which found that the risk of intracranial bleeding was lower with NOACs than with warfarin. Evidence from these studies is reflected in recommendations in major cardiovascular guidelines.6–8

panying haemorrhagic risk.1–5 Efficacy and safety using this dosing paradigm have been shown in a series of large clinical trials which found that the risk of intracranial bleeding was lower with NOACs than with warfarin. Evidence from these studies is reflected in recommendations in major cardiovascular guidelines.6–8 However, clinical experience suggests fear of bleeding complications may lead to the selection of low doses in practice, even at the risk of decreased efficacy and so a greater thrombotic risk. There has been considerable debate on whether laboratory measurement might therefore be appropriate not only in emergency situations or overdose but even in routine management. There is also a continuing need for guidance about which tests should be used. A workshop9 at the European Medicines Agency, attended by patient representatives, clinicians, academics, regulators and participants from the pharmaceutical industry, recently discussed the evidence about laboratory measurement from formal studies, clinical experience, and the multiple perspectives on NOAC treatment, and considered how our knowledge might be further enhanced.pt?>

nt representatives, clinicians, academics, regulators and participants from the pharmaceutical industry, recently discussed the evidence about laboratory measurement from formal studies, clinical experience, and the multiple perspectives on NOAC treatment, and considered how our knowledge might be further enhanced.pt?> Which tests are available? Experimental data show that drug concentrations of NOACs correlate with their activity. Although the pharmacokinetics of the licensed NOACs differ (see Table 1), unlike warfarin they have relatively short half-lives of around 5–15 h under normal circumstances, and the steep decline in plasma concentration after the peak makes measurements very sensitive to the time of sampling with respect to the last dose. This can make interpretation difficult. Mass spectrometry is the most reliable way of measuring drug concentration but is almost never used in routine practice.10 The classical coagulation tests can be misleading for determining NOAC activity.11,12 Product information for all these medicines warns against use of INR. Activated partial thromboplastin time (aPTT) or prothrombin time (PT) also cannot be used to quantify their activity precisely: changes in clotting measures are generally small and depend on the reagents used, and patients with normal values may have levels of NOAC that produce a significant anticoagulant effect.

INR. Activated partial thromboplastin time (aPTT) or prothrombin time (PT) also cannot be used to quantify their activity precisely: changes in clotting measures are generally small and depend on the reagents used, and patients with normal values may have levels of NOAC that produce a significant anticoagulant effect. In contrast, specific assays (dTT—diluted thrombin time, or ecarin-based assays such as ecarin chromogenic assay (ECA) for dabigatran, chromogenic factor Xa assays for the factor Xa inhibitors, see Table 2) correlate well with plasma concentrations11,12 although they may be less reliable at very low concentrations and require product-specific calibration. Table 2 Coagulation tests that can be used to estimate plasma concentrations of NOACs or to estimate the relative intensity of anticoagulationa,b

nhibitors, see Table 2) correlate well with plasma concentrations11,12 although they may be less reliable at very low concentrations and require product-specific calibration. Table 2 Coagulation tests that can be used to estimate plasma concentrations of NOACs or to estimate the relative intensity of anticoagulationa,b Test Molecule(s) Utility Sensitivity/ Specificity Dependence of the reagent External quality control Cut-off for a risk of bleeding (Unit(s) of expression) LC-MS/MS Dabigatran/ Rivaroxaban / Apixaban / Edoxaban Proven: LoD and LoQ around 1 and 3 ng/mL Not applicable No Yes: Accurately estimates the plasma concentrations—results expressed in ng/mL Depends on the indication (ng/mL) for dabigatran (i.e. 200 ng/m at trough in AF) Not established for direct factor Xa inhibitors APTT Dabigatran Limited: ±100 ng/mL / No Yes Yes Yes: Poorly reflect the intensity of anticoagulation Depends on the indication and the reagent (specific values are not presented since they depend on the reagent) TT Dabigatran Limited: Too sensitive (lower LoD below 0.025 ng/mL with some methodologies) / No Yes Yes Not established Only to exclude the presence of dabigatran. Useful in the peri-operative setting dTT Dabigatran Proven: ±10 ng/mL / No No Yes Yes: Accurately estimates the plasma concentrations—results expressed in ng/mL Depends on the indication (ng/mL) ECT Dabigatran Limited: ±15 ng/mL / No Probably not but an inter-lot variability has been reported No Yes: Standardization and validation required Depends on the indication (ratio: 3xULN and seconds: >103 seconds) ECA Dabigatran Proven: ±10 ng/mL / No No Yes Yes: Accurately estimates the plasma concentrations—results expressed in ng/mL Depends on the indication (ng/mL) (i.e. 200 ng/m at trough in AF) PT Rivaroxaban/ (Edoxaban) Limited: from ± 100 to > 500 ng/mL (depending on the reagent) / No Yes Yes Not established Poorly reflect the intensity of anticoagulation Chromogenic anti-Xa assays Rivaroxaban / Apixaban / Edoxaban Proven: ± 10 ng/mL / Yes–No (depend on the anti-Xa assay) No Yes Not established Accurately estimates the plasma concentrations—results expresses in ng/mL a Based on presentations and discussions during the workshop, and information summarized in7,15 of this article.

nic anti-Xa assays Rivaroxaban / Apixaban / Edoxaban Proven: ± 10 ng/mL / Yes–No (depend on the anti-Xa assay) No Yes Not established Accurately estimates the plasma concentrations—results expresses in ng/mL a Based on presentations and discussions during the workshop, and information summarized in7,15 of this article. b None of these tests are able to discriminate between therapies. Thrombin specific tests can easily identify dabigatran but other direct thrombin inhibitors such as argatroban or hirudin can influence them. For direct factor Xa inhibitors, only the Biophen® Direct Factor Xa Inhibitor can discriminate between heparins and direct FXa inhibitors but fail to differentiate between direct FXa inhibitors. LoD, limit of detection; LoQ, limit of quantification; ULN, upper limit of normal. When might measurements be useful? With increasing use of NOACs, patients in the pivotal clinical trials may not be representative of those being treated in clinical practice; real-life patients prescribed NOACs are often older, have reduced renal function and other comorbidities, and may be taking medicines that affect P-glycoprotein transporter or CYP3A4 activity—all factors that can affect plasma concentrations of NOACs and potentially increase the risk of bleeding or thrombosis.1–4

tice; real-life patients prescribed NOACs are often older, have reduced renal function and other comorbidities, and may be taking medicines that affect P-glycoprotein transporter or CYP3A4 activity—all factors that can affect plasma concentrations of NOACs and potentially increase the risk of bleeding or thrombosis.1–4 Regular clinical evaluation5 of patients treated with NOACs, for example to monitor renal function, is required. It may be important, too, in supporting adherence, particularly as patients used to monitoring for VKAs may be concerned by the lack of monitoring when switched to a NOAC.9 Existing thrombosis services and coagulation clinics may play an important role in such management. So can specific drug assays offer a significant safety improvement over the present dose adjustment by age, renal function, and haemorrhagic risk factors? Importantly there are no established, evidence-based therapeutic ranges for plasma concentrations of these medicines, although on-therapy ranges are available from some of the large studies and have been taken into account in the EU product information.1–4 However, these on-therapy drug concentrations vary with the indication for NOAC use and with patient characteristics. No single plasma concentration range provides optimal benefit-risk for all patients.9 Specific assays are used in some centres to identify outliers—patients with extremes of drug concentration—and so decide if the selected treatment is appropriate,9 but the blood concentration is only one factor in determining bleeding risk.

gle plasma concentration range provides optimal benefit-risk for all patients.9 Specific assays are used in some centres to identify outliers—patients with extremes of drug concentration—and so decide if the selected treatment is appropriate,9 but the blood concentration is only one factor in determining bleeding risk. This limits the usefulness of routine measurements of drug concentration, and on the basis of presentations to the workshop, the large-scale clinical trials that would be needed to establish a series of evidence-based therapeutic ranges for each subgroup of patients taking these medicines seem unlikely to be forthcoming. However, testing may be useful to confirm exposure in specific clinical situations such as patients who are bleeding, thought to be overdosed or who require invasive procedures, or where other factors (particularly in combination) may affect exposure (summarized in Table 3).9,13 There may also be a role for assays in monitoring the effect of the specific antidotes which have begun to become available (idarucizumab now licensed in the EU as Praxbind, andexanet alfa in development).9 The SmPC for Praxbind recommends using test results (using aPTT, dTT, or ECT) as one of the criteria to determine the need for repeat dosing,14 but an initial dose should not be delayed while awaiting the results of clotting time tests. Table 3 Situations in which coagulation testing for NOACs may be helpful

lopment).9 The SmPC for Praxbind recommends using test results (using aPTT, dTT, or ECT) as one of the criteria to determine the need for repeat dosing,14 but an initial dose should not be delayed while awaiting the results of clotting time tests. Table 3 Situations in which coagulation testing for NOACs may be helpful Situation Comment Bleeding (spontaneous or traumatic) When emergency surgery or invasive procedures are required In perioperative management For example in patients requiring elective surgery in whom the medicine may still be active Before thrombolytic treatment During bridging from one anticoagulant to another Patients suspected of being overdosed Note: there is no evidence to support titration outside the licensed doses To assess efficacy or adherence For example if new thrombosis develops during treatment with the anticoagulant Patients with deteriorating renal function Renal function is an important determinant of NOAC dosing Patients taking other medications that affect the pharmacokinetics Although NOACs are less sensitive to drug-drug interactions than warfarin, product information warns that dabigratran etexilate and edoxaban are substrates for P-glycoprotein (P-gp) and that apixaban and rivaroxaban pharmacokinetics are affected both by P-gp and cytochrome CYP3A4. Some combinations with inhibitors or inducers of these pathways are contraindicated or discouraged in the relevant SmPCs but even those inhibitors/inducers which do not normally produce clinically significant changes may be significant in combination with other factors Patients at extremes of bodyweight Edoxaban requires a dose reduction for patients with very low body weight; extremes of body weight may be significant for other NOACs in combination with other factors Patients who have received an initial dose of a specific antidote Initial administration of a specific antidote should not be delayed to await test results if it is clinically indicated, but testing may be helpful in determining the need for subsequent doses Patients with some combination of the above factors Patients not uncommonly exhibit a combination of factors (such as poor renal function, low body weight and co-administration of potentially interacting medications) that could combine to affect NOAC activity

may be helpful in determining the need for subsequent doses Patients with some combination of the above factors Patients not uncommonly exhibit a combination of factors (such as poor renal function, low body weight and co-administration of potentially interacting medications) that could combine to affect NOAC activity Workshop participants heard that, critically, there is no evidence to support doses outside the licensed dose range for a given NOAC; if a patient cannot be managed within the recommended range it may be better to consider an alternative anticoagulant.9 What are the challenges in practice? Anticoagulant therapy requires the patient, together with the doctor, to determine the acceptability of the risk of major bleeding on the one hand and of thrombotic events on the other. After decades of using VKAs, adoption of NOACs into clinical practice inevitably requires a shift in the approach to oral anticoagulation for both prescribers and patients. Critical points on the safe use of VKAs—such as INR monitoring and care over diet to avoid changes in vitamin K intake—do not apply to NOACs. Health professionals do, however, need to recognize NOACs on a patient’s medication history to avoid mishaps and assess anticoagulation appropriately.

h prescribers and patients. Critical points on the safe use of VKAs—such as INR monitoring and care over diet to avoid changes in vitamin K intake—do not apply to NOACs. Health professionals do, however, need to recognize NOACs on a patient’s medication history to avoid mishaps and assess anticoagulation appropriately. Although the evidence supports use of specific quantitative tests rather than prothrombin time tests,9,11 the specific tests are still not routinely available in many centres.9,15 Even where available, they need relevant expertise which may not be available around the clock, and so are often not requested, even in an emergency. Moreover, they require drug-specific calibration, and, in the absence of international calibration standards for the assays, there are considerable variations between laboratories.9,12 In practice, therefore, clinicians report that they often use aPTT or PT for screening, even though these tests cannot quantify activity precisely and may sometimes give normal results despite effective anticoagulation. International standardization of the specific tests is required, and rapid, point-of-care tests for use in emergency situations would be desirable. Information on appropriate testing is already included in the NOAC product information but guidance on selection and interpretation of the tests can be refined to provide the most appropriate information to health professionals. However, while regulators should ensure product information is clear, individual judgement on the circumstances for testing must rest with the clinician.

NOAC product information but guidance on selection and interpretation of the tests can be refined to provide the most appropriate information to health professionals. However, while regulators should ensure product information is clear, individual judgement on the circumstances for testing must rest with the clinician. Where do we go from here? Non-vitamin-K oral anticoagulants are effective medicines whose approval in the EU and elsewhere reflects a consistent regulatory assessment that their benefits outweigh their risks. Their product information already contains information to allow decisions on dose selection and the use of available assays. Nonetheless, further knowledge to support best use of these medicines and dose selection in particular subgroups is desirable.

tent regulatory assessment that their benefits outweigh their risks. Their product information already contains information to allow decisions on dose selection and the use of available assays. Nonetheless, further knowledge to support best use of these medicines and dose selection in particular subgroups is desirable. Large-scale randomized trials in patient subgroups are probably not feasible, but other avenues are being explored. Many patients requiring anticoagulation also have comorbidity and deteriorating renal function, so smaller studies on subgroups such as haemodialysis patients could be useful to understand the relevance of measurements in specific populations; some studies (such as NCT01896297) are under way, and others, such as a study of apixaban in atrial fibrillation patients on dialysis, are planned.9 Structured follow-up of real life patients can also help provide further evidence. Re-analysis of existing study data may be important to clarify our knowledge of relationships between drug levels and efficacy/bleeding risk and thereby strengthen recommendations. Such clarification can only further improve the safety and efficacy of NOAC use in the wider population, and might help to assuage those concerns of physicians and patients regarding bleeding risks that can lead to sub-optimal use of these medicines.

ls and efficacy/bleeding risk and thereby strengthen recommendations. Such clarification can only further improve the safety and efficacy of NOAC use in the wider population, and might help to assuage those concerns of physicians and patients regarding bleeding risks that can lead to sub-optimal use of these medicines. However, education and dissemination of the relevant information is equally important to allow prescribers, patients and others to use these medicines as safely as possible.9 Many hospitals have local protocols for dose adjustment, while for non-specialists a variety of regularly updated guidelines are available5–8 and may be helpful, but it is important that these are based on the latest evidence.11 Product information is regularly reviewed and should be updated to refine available guidance as new information emerges. Acknowledgements The authors would also like to acknowledge the particularly valuable contribution to the text of Anna Baczynska, Catherine Drai, Giampiero Mazzaglia, and Dinesh Mehta, scientific officers at the European Medicines Agency and of Jens Heisterberg, clinical pharmacologist and former CHMP member and Chief Medical Officer of the Danish Medicines Agency, who also chaired one of the sessions of the workshop. The views expressed in this article are the personal views of the authors and may not be understood or quoted as being made on behalf of or reflecting the position of the EMA or one of its committees or working parties.

edical Officer of the Danish Medicines Agency, who also chaired one of the sessions of the workshop. The views expressed in this article are the personal views of the authors and may not be understood or quoted as being made on behalf of or reflecting the position of the EMA or one of its committees or working parties. Conflict of interest: The authors have read and understood journal policy on declaration of interests and have no relevant personal interests to declare; Prof. Dogné is also a member of the Thrombosis and Hemostasis Center of the University of Namur, which has received a grant from Bayer Pharmaceuticals to support general research in thrombosis and haemostasis.

Introduction A high level of low-density lipoprotein cholesterol (LDL-C) is a well-documented risk factor for the development of CVD and the incidence of acute cardiac events. LDL-C is a modifiable factor, the lowering of which can lead to decreased risk of CVD.1 Individuals who are at risk for CVD events are recommended high-intensity LDL-C reduction treatment therapies.2,3 Despite widespread use and success of statins in the reduction of LDL-C and prevention of CVD, however, there remain unmet clinical needs in achieving LDL-C reduction goals.4,5 This includes individuals with heterozygous familial hypercholesterolaemia (HeFH), statin intolerance, and other high-risk patients who are not meeting target goals for lipid reduction with statins and lifestyle changes alone—particularly diabetics and those who have previously experienced CVD events.3,6,7 Ezetimibe has been used with some success as a next-line-of-defence cholesterol treatment, though its use and the extent it addresses unmet need is limited.8,9

ting target goals for lipid reduction with statins and lifestyle changes alone—particularly diabetics and those who have previously experienced CVD events.3,6,7 Ezetimibe has been used with some success as a next-line-of-defence cholesterol treatment, though its use and the extent it addresses unmet need is limited.8,9 Proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors evolocumab and alirocumab have recently been approved for use as lipid modifying therapies in both the US and Europe. Both drugs have been demonstrated to be safe, well tolerated, and extremely effective at lowering concentrations of LDL-C in the blood—in most cases by 55–65%.10,11 Cost-effectiveness of these drugs has not yet been well established.5 Several analyses out of the US suggest that PCSK9 inhibitors are not cost-effective, though at least one has found that they are.4,12–14 They are significantly less expensive in Europe, however, which means PCSK9 inhibitors may be cost-effective for other countries, including Norway. The UK’s National Institute for Health and Care Excellence (NICE) has recently approved it for limited use in some high-risk patient groups.15 In Norway, PCSK9 inhibitors are only reimbursed for homozygous familial hypercholesterolaemia patients, which are approximately 10 persons in the whole country. Current unmet clinical needs in LDL-C reduction reflect the candidate populations who are targets for PCSK9 inhibitors. Randomized control trials show substantial LDL-C reductions with PCSK9 inhibitors for all patients.10,11

milial hypercholesterolaemia patients, which are approximately 10 persons in the whole country. Current unmet clinical needs in LDL-C reduction reflect the candidate populations who are targets for PCSK9 inhibitors. Randomized control trials show substantial LDL-C reductions with PCSK9 inhibitors for all patients.10,11 The objective of this article is to develop a state-transition Markov model in Microsoft Excel in order to model the cost-effectiveness of PCSK9 inhibition for the prevention of cardiovascular events and CVD in Norway. The model specifically addresses PCSK9 inhibitors as compared with ezetimibe, and focuses on high-risk individuals. Methods Model structure and input A state-transition Markov model was developed to model the incidence of atherosclerotic cardiovascular disease events—specifically myocardial infarction (MI), ischaemic stroke (IS), and death. The model estimates incidence of CVD specifically within the Norwegian population for both primary and secondary prevention settings. Individuals who are ‘well’ are at risk for experiencing first-ever CVD events; those who survive these events transition into chronic post-CVD health states, where they remain at heightened risk for further events or death (Figure 1). ‘Well’ only indicates the absence of previous CVD events—these individuals may still be at a very high-risk for CVD due to various other baseline risk factors.

nts; those who survive these events transition into chronic post-CVD health states, where they remain at heightened risk for further events or death (Figure 1). ‘Well’ only indicates the absence of previous CVD events—these individuals may still be at a very high-risk for CVD due to various other baseline risk factors. Figure 1 Simplified schematic of state transition Markov model. Rectangles represent health states while ovals represent events. Patients can move in the direction of any arrow. Patients can be in only one health state per cycle. A cohort of patients can begin the model at any age from 30 years upward and runs up to age 100 or until everyone is dead. Men and women can be modelled in separate cohorts or combined. Baseline risk factors can be adjusted upwards to reflect heightened risk, while LDL-C reduction treatments reduce baseline risk according to the absolute measure of LDL-C reduction achieved. Absolute reductions in baseline LDL-C as a result of treatment were taken from meta-analyses of PCSK9 inhibitor RCTs (Table 1). Subsequent relative risk reductions are modelled according to CTT meta-analyses of the effect of LDL-C reductions on CVD risk (Table 1). Table 1 Key treatment effect parameters

A cohort of patients can begin the model at any age from 30 years upward and runs up to age 100 or until everyone is dead. Men and women can be modelled in separate cohorts or combined. Baseline risk factors can be adjusted upwards to reflect heightened risk, while LDL-C reduction treatments reduce baseline risk according to the absolute measure of LDL-C reduction achieved. Absolute reductions in baseline LDL-C as a result of treatment were taken from meta-analyses of PCSK9 inhibitor RCTs (Table 1). Subsequent relative risk reductions are modelled according to CTT meta-analyses of the effect of LDL-C reductions on CVD risk (Table 1). Table 1 Key treatment effect parameters Relative Risks (per mmol/L LDL-C reduction) RR (SE) Source Non-fatal MI 0.74 (0.03) 1,16 CHD Death 0.81 (0.01) 1,16 Non-fatal IS 0.81 (0.05) 1,16 Fatal IS 0.91 (0.10) 1,16 Other CVD Death 0.95 (0.10) 1,16 % LDL-C Reductions from Baseline % Change (SE) Source Evolocumab 0.63 (0.01) 11 Alirocumab 0.56 (0.01) 11 Ezetimibe 0.24 (0.01) 11 LDL-C reductions from PCSK9 inhibitors and effects of LDL-C reduction on CVD risk. Individuals who experience non-fatal CVD events and transition from the primary to the secondary component of the model are at increased risk for experiencing additional CVD events or death. Relative risks for those in chronic post-CVD states are taken from a variety of sources (see Supplementary material online, Appendix S1).

als who experience non-fatal CVD events and transition from the primary to the secondary component of the model are at increased risk for experiencing additional CVD events or death. Relative risks for those in chronic post-CVD states are taken from a variety of sources (see Supplementary material online, Appendix S1). The setting of this analysis is the Norwegian healthcare sector, which is publicly financed by the Norwegian National Health Insurance scheme and as such the focus is on direct medical costs.17 A lifetime horizon was chosen for this analysis. Initiation of PCSK9 inhibitors or ezetimibe is considered here to be the initiation of lifelong treatment. A discount rate of 4% is used for both future costs and utilities, as suggested by the Norwegian Ministry of Finance.18 Costs and results were converted from 2015 Norwegian Kroner (NOK) to 2015 Euros (€) for presentation in this report. A cost-effectiveness threshold of 600 000 NOK (€67 165) per additional QALY gained is typical for Norwegian economic evaluations. Though this is widely used, it is an unofficial guideline rather than a strict rule. Treatments and drugs with ICERs higher than this may also be approved for reimbursement in Norway.19

A discount rate of 4% is used for both future costs and utilities, as suggested by the Norwegian Ministry of Finance.18 Costs and results were converted from 2015 Norwegian Kroner (NOK) to 2015 Euros (€) for presentation in this report. A cost-effectiveness threshold of 600 000 NOK (€67 165) per additional QALY gained is typical for Norwegian economic evaluations. Though this is widely used, it is an unofficial guideline rather than a strict rule. Treatments and drugs with ICERs higher than this may also be approved for reimbursement in Norway.19 Quality of life The primary health outcome of this analysis is the Quality-Adjusted Life Year (QALY). We used QALY weights based on the EQ-5D HRQoL questionnaire and Time-Trade-Off (TTO) methods and valuation based on the UK tariff. QALY values are assigned to all chronic CVD states as well as healthy, non-CVD states (see Supplementary material online, Appendix S1). Use of the same tariff for all values helps to maintain consistency across QALY estimates.20 Resource use and costs Resource-use is estimated for CVD events and health outcomes; most estimates are made according to methods described in the Norwegian Cardiovascular Disease Model (NorCaD).18 NorCaD costs are well validated, and are frequently cited in Norwegian economic evaluations and health technology assessments.21–25 Costs assigned to each individual resource or cost component are taken from publicly available information.26–29

methods described in the Norwegian Cardiovascular Disease Model (NorCaD).18 NorCaD costs are well validated, and are frequently cited in Norwegian economic evaluations and health technology assessments.21–25 Costs assigned to each individual resource or cost component are taken from publicly available information.26–29 Evolocumab and alirocumab are each administered as injections every two weeks. Evolocumab costs €1813.56 for six injections; the cost for 1 year (52 weeks) is €7858.13. Alirocumab costs €1783.61 for six injections; 1 year costs €7728.38.27 Sensitivity analysis Scenario analysis was performed on price of evolocumab and alirocumab, reducing these by 50%. Probabilistic sensitivity analysis (PSA) was undertaken according to methods laid out by Briggs et al. The model was simulated 1000 times using random draws for each input parameter according to its respective distribution; this provides the probabilistic output of the model and a clearer picture of the uncertainty surrounding point estimates and mean output. Probabilistic output is recorded and analysed within the net benefits framework, and presented as Cost-Effectiveness Acceptability Curves (CEAC).17 Expected Value of Perfect Information (EVPI) analysis was performed to estimate the value of reducing uncertainty through new research or more information; EVPI estimates are based on PSA output. Analysis of the Expected Value of Perfect Information for Parameters (EVPPI) was also undertaken to determine for which specific parameters reduced uncertainty would yield the highest estimated value of new research.17

educing uncertainty through new research or more information; EVPI estimates are based on PSA output. Analysis of the Expected Value of Perfect Information for Parameters (EVPPI) was also undertaken to determine for which specific parameters reduced uncertainty would yield the highest estimated value of new research.17 Characteristics of patients We assumed very high-risk patients would be prime initial candidates for treatment. Modelling sometimes necessitates both assumptions and limitations in terms of what is modelled, as incorporating all possible variation and complexity is impossible. Our objective was to test a wide range of risk levels across age groups in both primary and secondary prevention, so we limited our focus to four base high-risk clinical profiles (Table 2). These profiles represent high-risk patients from candidate populations with unmet clinical need. Baseline characteristics come partially from characteristics of patients in RCTs, partially from diagnostic criteria, and partially from assumption. Primary prevention is defined here as prevention for those who have never suffered a previous myocardial infarction or ischaemic stroke. Secondary prevention focuses on patients who have previously suffered myocardial infarctions. Table 2  Patient profiles and baseline risk factors

teria, and partially from assumption. Primary prevention is defined here as prevention for those who have never suffered a previous myocardial infarction or ischaemic stroke. Secondary prevention focuses on patients who have previously suffered myocardial infarctions. Table 2  Patient profiles and baseline risk factors No. Description Total Cholesterol (mmol/L) LDL-C (mmol/L) Hypertension (SBP mm/Hg) Diabetes Smoker 1 Diabetic 6.2 3.9 Y (145) Y N 2 HeFH 9.2 6.2 N N N 3 Statin Intolerant 7.3 4.9 N N N 4 Misc. High Risk 6.5 4.0 Y (145) N N ‘N’ denotes ‘No’ and ‘Y’ denotes ‘Yes’ to indicate respectively the absence or presence of a risk factor. SBP denotes systolic blood pressure and is listed for hypertensive patients only. HeFH indicates heterozygous familial hypercholesterolaemia.

3 Statin Intolerant 7.3 4.9 N N N 4 Misc. High Risk 6.5 4.0 Y (145) N N ‘N’ denotes ‘No’ and ‘Y’ denotes ‘Yes’ to indicate respectively the absence or presence of a risk factor. SBP denotes systolic blood pressure and is listed for hypertensive patients only. HeFH indicates heterozygous familial hypercholesterolaemia. Results Primary prevention Ezetimibe, alirocumab, and evolocumab each lead to QALY gains across all four patient groups (Table 3). HeFH patients see the biggest gain in QALYs (discounted), from 8.74 with only statin therapy, to 9.50 when ezetimibe is added, and increasing again 10.34 with evolocumab. QALYs for diabetic patients are 7.31 with statins and increase to 7.83 with the addition of ezetimibe, and 8.57 with the addition of evolocumab. Monotherapy for statin intolerant patients sees a slightly smaller gain, increasing from 9.90 with no treatment to 10.29 QALYs with ezetimibe, and then to 10.95 with evolocumab. QALY gains for the miscellaneous high-risk group are slightly smaller. Alirocumab leads to similar but slightly smaller health gains than evolocumab for all patient groups. Table 3 Primary prevention for 65 year-olds (all numbers discounted and per person)

10.29 QALYs with ezetimibe, and then to 10.95 with evolocumab. QALY gains for the miscellaneous high-risk group are slightly smaller. Alirocumab leads to similar but slightly smaller health gains than evolocumab for all patient groups. Table 3 Primary prevention for 65 year-olds (all numbers discounted and per person) Drug cost (€) CVD cost (€) QALYs ICER (Δ€/ΔQALY) Diabetics Standard — 46 905 7.31 — Ezetimibe 5001 43 836 7.83 3716 Alirocumab 78 377 38 709 8.45 Dominated Evolocumab 80 879 37 498 8.57 93 938 HeFH Standard — 24 583 8.74 — Ezetimibe 6092 20 283 9.50 2369 Alirocumab 95 229 14 692 10.21 Dominated Evolocumab 98 007 13 609 10.34 101 351 Statin Intolerant Standard — 21 796 9.90 — Ezetimibe 6635 19 187 10.29 10 505 Alirocumab 101 796 14 328 10.86 Dominated Evolocumab 104 357 13 495 10.95 138 943 Misc. High Risk Standard — 20 897 10.33 — Ezetimibe 6902 18 238 10.67 12 170 Alirocumab 103 773 14 848 11.05 Dominated Evolocumab 106 208 14 152 11.12 212 700 Primary prevention indicates that patients have no history of myocardial infarction or ischaemic stroke. Standard treatment reflects whatever statin regimen the patients were on prior to initiation of PCSK9 or ezetimibe therapy. No drug cost was used for standard treatment; it was assumed that statin regimens would not change according to treatment and would therefore have no bearing on an incremental comparison of costs.

Standard treatment reflects whatever statin regimen the patients were on prior to initiation of PCSK9 or ezetimibe therapy. No drug cost was used for standard treatment; it was assumed that statin regimens would not change according to treatment and would therefore have no bearing on an incremental comparison of costs. The cost per patient of treating manifest CVD decreases with each incremental treatment for all patient groups. For diabetic patients costs decrease from €46 905 on standard treatment to €43 836 with ezetimibe, and then drop again to €37 498 with evolocumab. Decreases in CVD costs for the other three patient costs are comparable. Alirocumab results in slightly less CVD costs saved than evolocumab for all patient groups. Increases in lifetime drug costs per patient are quite substantial with both PCSK9 inhibitors. Ezetimibe drug costs range from €5000 to €6900, while alirocumab is €78 000 to €103 000 and evolocumab is €81 000 to €106 000. Alirocumab is €2000–€3000 less expensive than evolocumab for all patient groups at current prices. Increments in costs divided by increments in QALYs, ICERs are for PCSK9 inhibitors amongst 65 year-olds in the range of €94 000–€213 000 per QALY gained in primary prevention. For 50 year-old patient groups, ICERs were hundreds of thousands of Euros per QALY; when analysis is stratified by gender, men have consistently lower ICERs than women (see Supplementary material online, Appendix S2).

nhibitors amongst 65 year-olds in the range of €94 000–€213 000 per QALY gained in primary prevention. For 50 year-old patient groups, ICERs were hundreds of thousands of Euros per QALY; when analysis is stratified by gender, men have consistently lower ICERs than women (see Supplementary material online, Appendix S2). Secondary prevention QALYs are somewhat lower in secondary prevention due to the increased risk in these groups. QALY gains, however, are similar to those observed in primary prevention for all treatments, and in some cases slightly larger (Table 4). Treatment costs are generally higher among secondary compared with primary prevention patients, while treatment drug costs are somewhat lower. Table 4 Secondary prevention for 65 year-olds (all numbers per person) Drug cost (€) CVD cost (€) QALYs ICER (Δ€/ΔQALY) Diabetics Standard — 64 872 4.67 — Ezetimibe 3701 64 816 5.22 6544 Alirocumab 60 937 62 348 5.91 Dominated Evolocumab 63 468 61 495 6.05 68 386 HeFH Standard — 37 679 5.85 — Ezetimibe 4863 35 489 6.85 2654 Alirocumab 81 406 30 036 7.84 Dominated Evolocumab 84 646 28 695 8.01 63 174 Statin Intolerant Standard — 38 106 7.07 — Ezetimibe 5459 36 237 7.62 6588 Alirocumab 88 304 31 016 8.44 Dominated Evolocumab 91 176 29 923 8.56 84 428 Misc. High Risk Standard — 38 594 7.55 — Ezetimibe 5799 36 311 8.05 6969 Alirocumab 90 182 32 439 8.59 Dominated Evolocumab 92 841 31 525 8.69 128 191 Secondary prevention indicates all patients have a history of myocardial infarction.

62 6588 Alirocumab 88 304 31 016 8.44 Dominated Evolocumab 91 176 29 923 8.56 84 428 Misc. High Risk Standard — 38 594 7.55 — Ezetimibe 5799 36 311 8.05 6969 Alirocumab 90 182 32 439 8.59 Dominated Evolocumab 92 841 31 525 8.69 128 191 Secondary prevention indicates all patients have a history of myocardial infarction. Standard treatment reflects whatever statin regimen the patients were on prior to initiation of PCSK9 or ezetimibe therapy. No drug cost was used for standard treatment; it was assumed that statin regimens would not change according to treatment and would therefore have no bearing on an incremental comparison of costs. PCSK9 inhibitors are cost-effective for the HeFH patient group at 65 years of age with an ICER of €63 174/QALY. Use of PCSK9 inhibitors for 65 year-old diabetics is on the border of cost-effectiveness with an ICER of €68 386/QALY. The ICERs for those aged 70 and older are higher than ICERs for the 65 year-olds for all patient groups, but are still borderline cost-effective for the diabetic and HeFH patient groups. Initiating PCSK9 therapy for those younger than 65 is not cost-effective in any patient group; when results are stratified by gender, ICERs for treating men are once again lower than those for treating women (see Supplementary material online, Appendix S2).

rline cost-effective for the diabetic and HeFH patient groups. Initiating PCSK9 therapy for those younger than 65 is not cost-effective in any patient group; when results are stratified by gender, ICERs for treating men are once again lower than those for treating women (see Supplementary material online, Appendix S2). Scenario analyses With a 50% price reduction, PCSK9 inhibitors are cost-effective for all diabetic and HeFH patients, and cost-effective or borderline for older patients and secondary patients in less severe risk groups (lower section of Table 5). Table 5  Most cost-effective alternative across multiple ages and scenario analysis of price

blockers. If the covariate decreased the between-study variance, the source of heterogeneity was considered important. The estimate of τ2 in the presence of a covariate in comparison to that when the covariate is omitted allowed the proportion of the heterogeneity variance explained by the covariate to be calculated.13 Finally, a sensitivity analysis was undertaken to investigate the influence of each study by omitting each in turn from the meta-analysis and assessing the degree to which the magnitude and significance of the exposure effect changed.14 Evaluation of publication bias or small-study effect Publication bias is known to occur in meta-analyses, as studies that show a statistically significant effect of treatment are more likely to be published. Such selective publication of studies may lead to biased estimates that appear to be precise in meta-analysis based on literature search. In order to assess potential publication bias or small-study effect, we used the funnel-plot, which is a good visual evaluation of sampling bias. Funnel plot asymmetry raises the possibility of bias, and leads to a questioning of the interpretation of the overall effect when studies are combined in a meta-analysis. Sterne et al.15 have suggested that the funnel plot should be seen as a generic means of examining small study effect, which is the tendency for smaller studies in a meta-analysis to show larger treatment effects. To avoid evaluating publication bias only according to visually judgement, this was complemented by Egger’s test of asymmetry applied on the funnel plot.16

th a 50% price reduction, PCSK9 inhibitors are cost-effective for all diabetic and HeFH patients, and cost-effective or borderline for older patients and secondary patients in less severe risk groups (lower section of Table 5). Table 5  Most cost-effective alternative across multiple ages and scenario analysis of price Diabetic HeFH Statin intolerant Misc. high risk Age Primary Secondary Primary Secondary Primary Secondary Primary Secondary Current Market Price (∼€7800 per person, per year): 50 Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe 55 Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe 60 Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe Ezetimibe 65 Ezetimibe Eze/PCSK9* Ezetimibe PCSK9 Ezetimibe Ezetimibe Ezetimibe Ezetimibe 70 Ezetimibe Eze/PCSK9* Ezetimibe Eze/PCSK9* Ezetimibe Ezetimibe Ezetimibe Ezetimibe 75 Ezetimibe Eze/PCSK9* Ezetimibe Eze/PCSK9* Ezetimibe Ezetimibe Ezetimibe Ezetimibe 50% of Price (∼€3900 per person, per year): 50 PCSK9 PCSK9 PCSK9 PCSK9 Ezetimibe Ezetimibe Ezetimibe Ezetimibe 55 PCSK9 PCSK9 PCSK9 PCSK9 Ezetimibe PCSK9 Ezetimibe Ezetimibe 60 PCSK9 PCSK9 PCSK9 PCSK9 Ezetimibe PCSK9 Ezetimibe Eze/PCSK9* 65 PCSK9 PCSK9 PCSK9 PCSK9 Eze/PCSK9* PCSK9 Ezetimibe PCSK9 70 PCSK9 PCSK9 PCSK9 PCSK9 Eze/PCSK9* PCSK9 Ezetimibe PCSK9 75 PCSK9 PCSK9 PCSK9 PCSK9 PCSK9 PCSK9 Ezetimibe PCSK9 Cells indicate the most cost-effective treatment option for the respective patient/age group at a willingness-to-pay threshold of 600 000 NOK/QALY (€67 165/QALY). Eze/PCSK9* indicates borderline cost-effectiveness, meaning the ICER for PCSK9 inhibitors was between 600 000 and 700 000 NOK/QALY (€67 165–€78 360/QALY).

K9 Cells indicate the most cost-effective treatment option for the respective patient/age group at a willingness-to-pay threshold of 600 000 NOK/QALY (€67 165/QALY). Eze/PCSK9* indicates borderline cost-effectiveness, meaning the ICER for PCSK9 inhibitors was between 600 000 and 700 000 NOK/QALY (€67 165–€78 360/QALY). Probabilistic sensitivity analysis Cost-effectiveness acceptability curves (CEACs) from probabilistic sensitivity analysis indicate that there is a very low probability that PCSK9 inhibitors are cost-effective given a 600 000 NOK/QALY (€67 165/QALY) threshold (Figure 2). Assuming this WTP threshold, the probability that evolocumab is cost-effective in primary prevention is close to or at 0% for all patient groups. The diabetic patient group has the highest probability with 3.4%. In secondary prevention, the probability that evolocumab is cost-effective at this WTP threshold is 55.2% for HeFH patients, 42.0% for diabetic patients, and 13.2% for statin intolerant patients; it is still 0% for the miscellaneous high risk group. Figure 2 Cost-effectiveness acceptability curves and expected value of perfect information for 65 year-olds. Standard societal willingness-to-pay threshold is 600 000 NOK (€67 165) per QALY. Outer axis labels apply to each individual graph according to its corresponding position in the overall picture.

Probabilistic sensitivity analysis Cost-effectiveness acceptability curves (CEACs) from probabilistic sensitivity analysis indicate that there is a very low probability that PCSK9 inhibitors are cost-effective given a 600 000 NOK/QALY (€67 165/QALY) threshold (Figure 2). Assuming this WTP threshold, the probability that evolocumab is cost-effective in primary prevention is close to or at 0% for all patient groups. The diabetic patient group has the highest probability with 3.4%. In secondary prevention, the probability that evolocumab is cost-effective at this WTP threshold is 55.2% for HeFH patients, 42.0% for diabetic patients, and 13.2% for statin intolerant patients; it is still 0% for the miscellaneous high risk group. Figure 2 Cost-effectiveness acceptability curves and expected value of perfect information for 65 year-olds. Standard societal willingness-to-pay threshold is 600 000 NOK (€67 165) per QALY. Outer axis labels apply to each individual graph according to its corresponding position in the overall picture. EVPPI analysis for 65 year-old HeFH patients at a standard willingness to pay threshold of 600 000 NOK/QALY (€67 165/QALY) suggests most value would come from reducing uncertainty around baseline risk and the treatment effect variables. EVPPI results also indicate value from reducing uncertainty around QALY values, which would likely be resolved by splitting the post-MI health state into several smaller, more specific health states (see Supplementary material online, Appendix S2). For most other patient groups, EVPPI analysis at a WTP threshold of 600 000 NOK (€67 165) yields 0 for all parameters and parameter groups, indicating that there is no uncertainty needed to reduce as long as prices are at current levels.

ler, more specific health states (see Supplementary material online, Appendix S2). For most other patient groups, EVPPI analysis at a WTP threshold of 600 000 NOK (€67 165) yields 0 for all parameters and parameter groups, indicating that there is no uncertainty needed to reduce as long as prices are at current levels. Discussion Cost of treatment with PCSK9 inhibitors over many years at current market prices will be extremely expensive. In this analysis, all patient groups are at considerable risk for lifetime CVD.3 Yet despite projections of gains in QALYs, reductions in CVD events and CVD deaths, and incremental cost-savings in terms of treating manifest CVD, our model finds that the benefits of the drugs almost never offset their high long-term costs. Cost-effectiveness was only found likely in older patients with extremely high risk and a history of myocardial infarction.

uctions in CVD events and CVD deaths, and incremental cost-savings in terms of treating manifest CVD, our model finds that the benefits of the drugs almost never offset their high long-term costs. Cost-effectiveness was only found likely in older patients with extremely high risk and a history of myocardial infarction. Cost-effectiveness in our model is therefore very sensitive to long-term price, which is why older patients are found to be more cost-effective. Our modelling results are likely at odds with what many clinicians consider to be the potential clinical value of PCSK9 inhibitors. There is no doubt that preventing a heart attack for a young patient has tremendous clinical value, more so than preventing a heart attack for a much older patient. What our model suggests is that the high cost of PCSK9 inhibitors over many years, if not decades, of treatment may not be offset by the amount of CVD prevented and the clinical value gained, given the current price of the drug and standard willingness-to-pay thresholds. Discounting is also an important consideration when modelling younger patients: if prices are high and QALY gains are further into the future, this will yield a less cost-effective result. Our model is not intended to be a clinical analysis in this sense, but rather an economic one.

ndard willingness-to-pay thresholds. Discounting is also an important consideration when modelling younger patients: if prices are high and QALY gains are further into the future, this will yield a less cost-effective result. Our model is not intended to be a clinical analysis in this sense, but rather an economic one. Because they are direct comparators in next-line-of-defence pharmacological cholesterol reduction, the total number of patients currently using ezetimibe is likely the best estimation of the wider patient population who might be immediately eligible for PCSK9 inhibitors. Given the massive budget impact if all eligible users were to immediately begin therapy with PCSK9 inhibitors, the results of this modelling exercise can provide some level of insight into which patient populations would be the most cost-effective and should perhaps receive them first. Our model suggests that as the cost of PCSK9 inhibitors decreases, increasing access should focus on younger high-risk patients rather than moderate-risk older patients (Table 5). Allocation of PCSK9 inhibitors will then fit well with the age-differentiated Norwegian prioritization guidelines as discussed in NORRISK, because treating younger, high-risk patients yields higher expected benefits and is more cost-effective than treating older patients at more moderate levels of risk.30

ts (Table 5). Allocation of PCSK9 inhibitors will then fit well with the age-differentiated Norwegian prioritization guidelines as discussed in NORRISK, because treating younger, high-risk patients yields higher expected benefits and is more cost-effective than treating older patients at more moderate levels of risk.30 The fact that the Norwegian willingness-to-pay threshold tends to be flexible between €67 165 and €78 360 (600 000 and 700 000 NOK) per QALY changes the conclusion about cost-effectiveness of use of PCSK9 inhibitors in some patient groups. In Table 5, we showed that 6 groups instead of only 1 was cost-effective if the threshold is €78 360 instead of €67 165.

to-pay threshold tends to be flexible between €67 165 and €78 360 (600 000 and 700 000 NOK) per QALY changes the conclusion about cost-effectiveness of use of PCSK9 inhibitors in some patient groups. In Table 5, we showed that 6 groups instead of only 1 was cost-effective if the threshold is €78 360 instead of €67 165. Our analysis was limited to the 140 mg dose of evolocumab, as this is the only available dose in Norway. For alirocumab we included only the highest available 150 mg dose of alirocumab and did not allow for titration from lower to higher doses; annual cost of alirocumab is not affected by titration in Norway. Our model found evolocumab to be more cost-effective than alirocumab across all patient profiles and age groups. This is most likely to be a result of how the modelling was performed, as CVD prevention was modelled through LDL-C reduction, and the estimates we used suggest the dose of evolocumab considered is slightly more effective at lowering LDL-C than that of alirocumab, with no overlap in the confidence intervals. Because our model was not designed to differentiate between the two in great detail, results should be seen as an indication as to the modelled cost-effectiveness of PCSK9 inhibitors as a class of drug, rather than a recommendation of one over the other.

that of alirocumab, with no overlap in the confidence intervals. Because our model was not designed to differentiate between the two in great detail, results should be seen as an indication as to the modelled cost-effectiveness of PCSK9 inhibitors as a class of drug, rather than a recommendation of one over the other. Cost-effectiveness analyses should be specific to the setting in which treatment is being considered. The structure, organization, and funding of the health system in question should be accounted for in order for results to be meaningful and accurate. Only a limited number of cost-effectiveness analyses have been performed on PCSK9 inhibitors to date, and most were performed in a US setting. Our report offers new insight and perspective in this regard, as Norway’s health system is more similar in nature to other countries in Western and Northern Europe than is the US system. The most recent analysis from the US found the drugs not to be cost-effective for any patient group.4,14 Researchers considered HeFH patients for both primary and secondary prevention, and general high-risk patients in secondary prevention, which includes those who are statin intolerant. It was estimated that in order for PCSK9 inhibitors to be considered cost-effective for any patient groups, the annual price would need to be reduced by roughly two-thirds. Though, as stated, analyses across different health systems are not always directly comparable due to inherent differences between them, these results appear overall to be consistent with our scenario analysis on price.

t-effective for any patient groups, the annual price would need to be reduced by roughly two-thirds. Though, as stated, analyses across different health systems are not always directly comparable due to inherent differences between them, these results appear overall to be consistent with our scenario analysis on price. The UK’s National Institute for Health and Care Excellence (NICE) has issued guidance recommending use of evolocumab in certain high-risk patient groups. Specifically, it is recommended for secondary prevention of CVD for non-familial hypercholesterolaemia patients at high-risk or very high-risk. It also recommends evolocumab for familial hypercholesterolaemia patients in both primary and secondary prevention settings when LDL-C levels remain very high. The NICE report presents the same general trends as our analyses: a general inverse relationship between risk and ICERs, including lower ICERs with higher LDL-C levels and with higher age. Note that NICE results are based on undisclosed price discounts negotiated with the pharmaceutical company.15

els remain very high. The NICE report presents the same general trends as our analyses: a general inverse relationship between risk and ICERs, including lower ICERs with higher LDL-C levels and with higher age. Note that NICE results are based on undisclosed price discounts negotiated with the pharmaceutical company.15 A US-based analysis of evolocumab performed by researchers for Amgen found the drug to be cost-effective in both primary and secondary prevention for familial hypercholesterolaemia patients, and cost-effective in secondary prevention for non-familial hypercholesterolaemia patients, regardless of tolerance to statins. Like our analyses, the Amgen analysis also utilized CTT figures to estimate the effects of LDL-C reduction on the relative risk of CVD.13 The Amgen analysis did not, however, include ezetimibe as a comparator. Because cost-effectiveness modelling focuses on incremental costs and effects between treatment options, the exclusion of ezetimibe is the most likely reason for their differing results. An analysis presented at an ISPOR conference found that PCSK9 inhibitors as an adjunct to statin therapy were not cost-effective for any patient group tested, regardless of risk or history of CVD. CTT figures were used to estimate risk reductions as a result of LDL-C lowering in this US-based analysis as well.12 Our analyses are also consistent with other analyses in that it finds ezetimibe to be a cost-effective treatment option—and at least one report has found ezetimibe is cost-effective for use specifically in Norway.31,32

An analysis presented at an ISPOR conference found that PCSK9 inhibitors as an adjunct to statin therapy were not cost-effective for any patient group tested, regardless of risk or history of CVD. CTT figures were used to estimate risk reductions as a result of LDL-C lowering in this US-based analysis as well.12 Our analyses are also consistent with other analyses in that it finds ezetimibe to be a cost-effective treatment option—and at least one report has found ezetimibe is cost-effective for use specifically in Norway.31,32 Limitations The lack of data on the actual preventative effect of PCSK9 inhibitors on CVD necessitated that we make a number of assumptions. All CVD prevention was modelled strictly through reductions in LDL-C, which required extrapolation and assumptions based on CTT meta-analyses of statins. This assumes that CVD prevention is strictly a function of LDL-C, and that reductions in LDL-C as a result from PCSK9 inhibition will have the same level of effect as reduction of LDL-C through statins. All of these are major limitations of the analysis, and as such we stress that this is a modelling exercise in potential cost-effectiveness of PCSK9 treatment, rather than a true evaluation of it. Only clinical evidence of CVD prevention as a direct result of using PCSK9 inhibitors can truly address these limitations.

All of these are major limitations of the analysis, and as such we stress that this is a modelling exercise in potential cost-effectiveness of PCSK9 treatment, rather than a true evaluation of it. Only clinical evidence of CVD prevention as a direct result of using PCSK9 inhibitors can truly address these limitations. Long-term trials for PCSK9 are currently underway, with expected publication in 2017. In the meantime the method employed in this report, though limited, is thought to be one of the best available options.5 Also, later clinical studies of ezetimibe and rosuvastatin have been shown to reduce LDL-C at levels consistent with CTT estimates, which gives credibility and validity to the extrapolation of CTT beyond their original analyses.8,33 As noted above, other cost-effectiveness analyses also make similar estimations. Secondary prevention in this analysis is restricted to post-MI patients only, so results exclude stroke patients. Post-IS patients would likely need to be modelled separately, given that they are qualitatively different and have poorer prognoses than post-MI patients. We did not include the potential for a legacy effect from PCSK9 inhibitors in our analysis, whereby benefits of treatment continue after treatment period has ended. If PCSK9 inhibitors were proven to have a legacy effect, likelihood of cost-effectiveness would presumably rise. A potential legacy effect would be particularly important in younger patients, as it would significantly reduce the long-term costs of treatment.

efits of treatment continue after treatment period has ended. If PCSK9 inhibitors were proven to have a legacy effect, likelihood of cost-effectiveness would presumably rise. A potential legacy effect would be particularly important in younger patients, as it would significantly reduce the long-term costs of treatment. In addition, our analysis does not account for any ‘saw tooth’ effect on LDL-C, whereby it rises and falls in the intervals between biweekly injections. The potential importance of this factor on long-term CVD risk is currently unknown. The effects of LDL-C reduction on relative risk of CVD used in this analysis do not differ according to age. CTT does not break down data by both heterogeneity and specific type of CVD prevented; we chose to use the latter as it was more prudent to our analyses. Some of the analyses performed and results reported are strictly deterministic, when in fact the uncertainty around any analysis should be explored through PSA.34 Specifically, not all of the age groups were tested with PSA, only the 65 and 50 year olds. Results were, however similar in deterministic and probabilistic analyses. The reason it was not undertaken for all analyses is because PSA is computationally demanding and time consuming. Inclusion of more PSA would have lead to a greater exploration of the uncertainty surrounding the results and a more robust analysis.

sults were, however similar in deterministic and probabilistic analyses. The reason it was not undertaken for all analyses is because PSA is computationally demanding and time consuming. Inclusion of more PSA would have lead to a greater exploration of the uncertainty surrounding the results and a more robust analysis. Conclusion In conclusion, our model suggests PCSK9 inhibitors would not be cost-effective for primary prevention. In secondary prevention, PCSK9 inhibitors may be cost-effective for older patients at the highest levels of absolute CVD risk. Our model suggests high lifetime drug costs may not be offset by clinical value gained when treating younger patients over many years, regardless of baseline risk. A decrease in price would mean the drugs were more likely to be cost-effective for younger high-risk patients. Future research is needed to determine the actual long-term preventative effect of PCSK9 inhibitors on CVD. Supplementary material Supplementary material is available at European Heart Journal – Cardiovascular Pharmacotherapy online Conflict of interest: Torbjørn Wisløff has received funding by Amgen in a different project relating to evolocumab. Supplementary Material Supplementary Appendix 1 Click here for additional data file. Supplementary Appendix 2 Click here for additional data file.

Introduction Atrial fibrillation or flutter (AF) is the most common type of sustained cardiac arrhythmia. To restore sinus rhythm one option is to perform a cardioversion. There are two types of cardioversions: pharmacological and electrical. The procedure is associated with a risk of thromboembolic events, most commonly ischaemic stroke.1 Adequate anticoagulation can significantly reduce the incidence of thromboembolic complications. According to current AF guidelines (ESC 20162 and AHA/ACC 20143) adequate oral anticoagulation (OAC) is recommended for at least 3 weeks prior to and for 4 weeks after cardioversion in patients with AF of  ≥ 48 h or unknown duration. As an alternative to OAC, transoesophageal echocardiography can be performed before cardioversion to rule out left atrial thrombus. Vitamin K antagonists (VKA) can prevent thromboembolic episodes during cardioversion.4 A great amount of clinical experience is available on VKA,5 but there are some limitations to their usage, such as the requirement of continuous laboratory monitoring and the numerous drug and food interactions. Since the advent of non-VKA oral anticoagulants (NOACs) there are now several alternatives used for prevention of stroke in patients with AF.6–12 The NOACs comprize dabigatran, a direct thrombin inhibitor, and rivaroxaban, apixaban and edoxaban, which are direct Xa factor antagonists.

merous drug and food interactions. Since the advent of non-VKA oral anticoagulants (NOACs) there are now several alternatives used for prevention of stroke in patients with AF.6–12 The NOACs comprize dabigatran, a direct thrombin inhibitor, and rivaroxaban, apixaban and edoxaban, which are direct Xa factor antagonists. With regards to cardioversion, X-VERT was the first prospective randomized trial to show that rivaroxaban can prevent thromboembolic complications as effectively as VKA in patients undergoing cardioversion.13 Also the recently published ENSURE-AF14 trial and the posthoc analyses of the RE-LY15 and ARISTOTLE16 trials proved that edoxaban, dabigatran, and apixaban can be safely used as well. In addition, several single-centre trials and meta-analyses confirmed the low number of thromboembolic events and safe use of NOACs during cardioversion.17-20 Aims Our aim was to create a partly prospective and partly retrospective cardioversion registry, particularly focusing on OAC strategies in different European countries (Hungary, Italy, France, Spain, Lithuania), and on emerging choices over time for anticoagulants in this setting. Methods Design and patients All patients in the study centres who underwent electrical or pharmacological cardioversion of AF or flutter were included in the registry. Patient records were collected in seven cardiology departments of six different European cities in five European countries, namely (i) Budapest—Hungary, (ii) Budapest—Hungary, (iii) Pisa—Italy, (iv) Bari—Italy, (v) Madrid—Spain, (vi) Amiens—France, and (vii) Kaunas—Lithuania.

F or flutter were included in the registry. Patient records were collected in seven cardiology departments of six different European cities in five European countries, namely (i) Budapest—Hungary, (ii) Budapest—Hungary, (iii) Pisa—Italy, (iv) Bari—Italy, (v) Madrid—Spain, (vi) Amiens—France, and (vii) Kaunas—Lithuania. Data were recorded between September 2014 and October 2015, in the 7 months prior to start date of the registry, and in the 6 months following the start date of the registry. The retrospective and prospective design was chosen in order to align the timing to the 12 months following the presentation of the X-VERT trial, in order to assess whether its publication would influence NOAC uptake in the cardioversion setting. Data collection was performed in in-patient and out-patient cardiology departments. All cardioversions during electrophysiological procedures were excluded; also data from emergency departments were not collected. Cardioversions were performed according to local protocols. The study was approved by the ethics committee in participating centres/country. For all patients a case report form was completed, and subsequently databased. Cardiovascular endpoints were not assessed in this registry.

ergency departments were not collected. Cardioversions were performed according to local protocols. The study was approved by the ethics committee in participating centres/country. For all patients a case report form was completed, and subsequently databased. Cardiovascular endpoints were not assessed in this registry. Statistical methods We used the Pearson chi-squared test to compare NOAC vs. VKA in de novo prescriptions prior to scheduled cardioversion (yes/no), duration of OAC before cardioversion (more/less than 3 weeks), and duration of OAC after cardioversion (more/less than 4 weeks). The Pearson chi-squared test was also used to compare Spain and France in use of anticoagulants (yes/no). To analyse changes in OAC usage over time, we used logistic regression with OAC (yes/no) as the outcome variable and the number of months since 1 January 2014 as explanatory variable. The plots of OAC usage over time were smoothed with 1-nearest-neighbour and 2-nearest-neighbour averaging. Stata 14 (StataCorp LP) and Matlab 2014 (Mathworks, Inc) were used to perform the statistical analyses.

h OAC (yes/no) as the outcome variable and the number of months since 1 January 2014 as explanatory variable. The plots of OAC usage over time were smoothed with 1-nearest-neighbour and 2-nearest-neighbour averaging. Stata 14 (StataCorp LP) and Matlab 2014 (Mathworks, Inc) were used to perform the statistical analyses. Results A total of 1101 patients (male/female: 742/359, mean age: 67.3 years ± 11.2) were included in the registry. Six hundred and seventy-nine retrospective and 422 prospective cases were collected. Most of the cardioversions were elective (1050 cases), acute cardioversions were done only 51 times, where acute procedure was defined as a cardioversion performed on a haemodynamically unstable patient. Nearly all cardioversions were electrical (97%), only 3% were pharmacological. 405 patients were recorded in Hungary, 200 in Italy, 193 in Lithuania, 186 in Spain, and 116 in France. Oral anticoagulants were administered in 87% of the patients. The usage of NOACs vs. VKA was 31.5% vs. 68.5%, respectively. Thirteen percent of the subjects were cardioverted without oral anticoagulants, because of an AF duration of <48 h in most cases (83.3%). OAC prescription habits were also documented. No differences were found between NOACs and VKA in de novo prescriptions prior to scheduled cardioversions (NOAC: 20% vs. VKA: 19%; P = 0.68). Also we analysed previously anticoagulated and treatment-naive patients, and there were no differences in the use of NOACs and VKA between the groups (previously prescribed: 30.2% NOACs vs. new treatment: 30.3% NOACs, P = 0.99).

de novo prescriptions prior to scheduled cardioversions (NOAC: 20% vs. VKA: 19%; P = 0.68). Also we analysed previously anticoagulated and treatment-naive patients, and there were no differences in the use of NOACs and VKA between the groups (previously prescribed: 30.2% NOACs vs. new treatment: 30.3% NOACs, P = 0.99). Also there were no differences between the duration of OAC before cardioversion, i.e. most of the patients received 3 or more weeks of OAC in both the NOACs and the VKA groups (86% vs. 84%; P = 0.51). In 87% of the patients taking NOACs and in 84% of those taking VKA, the anticoagulants were given for more than 4 weeks after the procedure (P = 0.14). When using VKA, international normalized ratio (INR) at cardioversion was above 2.0 in 76% of the cases. When examining OAC strategies over time a decline in VKA usage (P = 0.033) in elective cardioversion over approximately 1 year was observed (Figure 1A). During the observation period an increase in apixaban (P < 0.001), a slight increase in rivaroxaban (P = 0.028) and no changes in dabigatran (P = 0.34) usage for elective cardioversion was recorded (Figure 1B). Figure 1 (A) Changes in OAC usage over time. (B) Changes in NOAC usage over approximately 1 year (enlarged portion of Figure 1A). There were differences in use of anticoagulants between the countries: Spain used most VKA (89%), while France used least VKA (39%, P < 0.001). There were no large differences in VKA usage between Italy, Hungary and Lithuania (Figure 2). Figure 2 VKA usage in different European countries.

Figure 1 (A) Changes in OAC usage over time. (B) Changes in NOAC usage over approximately 1 year (enlarged portion of Figure 1A). There were differences in use of anticoagulants between the countries: Spain used most VKA (89%), while France used least VKA (39%, P < 0.001). There were no large differences in VKA usage between Italy, Hungary and Lithuania (Figure 2). Figure 2 VKA usage in different European countries. Discussion According to large randomized prospective clinical studies, NOACs were shown to be at least non-inferior to VKA in the prevention of stroke in patients with AF.6–9 Further, NOACs have been shown to be safe alternatives to VKA during cardioversion.13–16

There were differences in use of anticoagulants between the countries: Spain used most VKA (89%), while France used least VKA (39%, P < 0.001). There were no large differences in VKA usage between Italy, Hungary and Lithuania (Figure 2). Figure 2 VKA usage in different European countries. Discussion According to large randomized prospective clinical studies, NOACs were shown to be at least non-inferior to VKA in the prevention of stroke in patients with AF.6–9 Further, NOACs have been shown to be safe alternatives to VKA during cardioversion.13–16 To examine anticoagulation strategies in different European countries our research group created a cardioversion registry of 1101 patients enrolled over approximately 1 year. According to our results only approximately 3/4 of the patients receiving VKA reached therapeutic INR levels during cardioversion, which is a concern given the predominantly elective and iatrogenic nature of cardioversion and the known increased risk for thromboembolism. On the other hand, in our registry most patients received anticoagulation more than 3 weeks before and more than 4 weeks after cardioversion which is similar to the results of the European Heart Rhythm Association Survey.21 No differences were observed between the NOACs and VKA groups in the duration of OAC. In the X-VERT13 trial a significantly shorter time to cardioversion was observed in the rivaroxaban group as opposed to the VKA group. According to the ENSURE-AF14 trial there was no difference in time to cardioversion between the edoxaban and warfarin group, which was explained by optimized warfarin dosing and enoxaparin bridging therapy. In our study, the exact duration from start of treatment to the cardioversion procedure was not registered. However, the relatively low number of subjects reaching therapeutic INRs may indirectly suggest that patients underwent a longer treatment period in the VKA group, and hence a reduced time to cardioversion in the NOAC group after the initial 3 weeks.

start of treatment to the cardioversion procedure was not registered. However, the relatively low number of subjects reaching therapeutic INRs may indirectly suggest that patients underwent a longer treatment period in the VKA group, and hence a reduced time to cardioversion in the NOAC group after the initial 3 weeks. One of the conclusions of our registry is that despite several practical advantages of NOACs over VKAs, clinicians appear to hesitate to embrace their usage during cardioversions. According to a Norvegian registry in total 32 675 patients with AF 65.1% was given NOACs, but in our cardioversion registry the usage of NOACs was only 31.5%.22 One of the reasons may be that compliance (persistence and adherence) is harder to monitor in NOACs than in VKAs, which is of extreme importance when performing cardioversion.

orvegian registry in total 32 675 patients with AF 65.1% was given NOACs, but in our cardioversion registry the usage of NOACs was only 31.5%.22 One of the reasons may be that compliance (persistence and adherence) is harder to monitor in NOACs than in VKAs, which is of extreme importance when performing cardioversion. The slight increase in rivaroxaban usage over time might well be a reflection of the publication of the X-VERT trial in 2014, the first prospective trial to show the safety and efficacy of a NOAC during cardioversion when compared to warfarin. Dabigatran was the first choice in most of the countries in the first 4 months of our observational period, and it remained at a stable usage ratio. Apixaban was the least prescribed NOAC during the first few months, but a substantial increase throughout the observational period of one year was recorded. In a Danish registry between 2011 and 2015, a decreasing usage of VKA in general AF management was observed, and among NOACs a rise of apixaban and a decrease of dabigatran was noticed.23 Similarly, an expanding use of NOACs was reported in the GARFIELD-AF registry, and there were differences between countries similar to our results in the cardioversion population.24 Differences of NOAC usage between countries depend predominantly on regulatory as well as economic factors.

e of dabigatran was noticed.23 Similarly, an expanding use of NOACs was reported in the GARFIELD-AF registry, and there were differences between countries similar to our results in the cardioversion population.24 Differences of NOAC usage between countries depend predominantly on regulatory as well as economic factors. The design of this registry purposely focused on the use of oral anticoagulants, however outcome data of stroke and systemic embolism and safety endpoints such as bleeding events would be of great additional value to this work. These will be subject for future research and will receive highest priority. Conclusions Taken together, our study shows that most patients did receive oral anticoagulants during cardioversion. VKAs were used more often (2/3 vs. 1/3) than NOACs in this particular setting. However, our results show a significant decrease in VKA usage over time, while NOAC usage displays a gradual increase. As NOACs are more and more widely used in AF in general, further studies in the cardioversion setting will be needed to understand their optimal role in this aspect of AF patient management. Conflict of interest: none declared. Funding Open access to this publication was supported by Bayer AG. Conflict of interest: none declared.

Introduction Improved lifestyle, together with more effective pharmacotherapy and invasive treatment, has resulted in a decline in first-time coronary artery disease (CAD) and increased survival in patients with established CAD.1,2 Despite the decline in mortality from CAD, it remains one of the leading causes of premature death on a European scale.2 Aside from an additional risk of premature death, CAD is also associated with risk of recurrent cardiovascular events, e.g. stroke and recurrent myocardial infarction (MI), with the highest risk of recurrent events during the first year after MI.3–5 This risk is targeted by a similar guideline recommended treatment duration, but evidence has shown that the risk persists beyond the first year after MI and that the risk depends on the patient’s risk profile.4–6 With the projected increase in high-risk patients that stay event-free on the first year after MI and the associated long-term health-care burden,6 it has become even more relevant to clarify the long-term risk of recurrent events in stable post-MI patients with distinct risk profiles. Thus, high-risk patients might benefit from extended tailored treatment approach. However, it is important to highlight the fact that a considerable proportion of patients with established illness still appear to receive sub-optimal cardiac care,7 secondary prevention and cardiac rehabilitation.8 Another serious challenge is that the prevalence of coexisting chronic illnesses, which in many cases share the same risk factors as CAD,9 is high and increasing,1 but even more importantly, are associated with unfavourable prognosis in patients with CAD.1,10–14 Although CAD severity is one of the strongest risk factors for long-term outcome,15–20 its importance in late-risk stratification and in relation to co-morbidity among stable post-MI patient who have stayed event-free on the first year is not entirely clear.

ated with unfavourable prognosis in patients with CAD.1,10–14 Although CAD severity is one of the strongest risk factors for long-term outcome,15–20 its importance in late-risk stratification and in relation to co-morbidity among stable post-MI patient who have stayed event-free on the first year is not entirely clear. Recognizing that CAD extent and severity can been graded differently from the more simple usage of the number of stenosed vessels15–17 to the use of more complex scoring system21 and that the coronary angiography (CAG) lack of ability to detect prognostically important physiological stenosis,22 we investigated the long-term (1 year and beyond) impact of CAD severity, using the simple estimation of CAD severity (no obstructive CAD, 1-, 2-, 3-vessel disease [VD] or left main stenosis [LMS]), in relation to co-morbidity for recurrent events in a nationwide study population. Patients conservatively treated who were not receiving CAG, and thus with no information on CAD severity, were also included, as conservatively treated patients fare worse than invasively treated patients and deserve equal attention in terms of attaining knowledge of risk and risk factors.

nationwide study population. Patients conservatively treated who were not receiving CAG, and thus with no information on CAD severity, were also included, as conservatively treated patients fare worse than invasively treated patients and deserve equal attention in terms of attaining knowledge of risk and risk factors. Methods Data sources In Denmark, each resident has a unique and permanent identification number that enables individual-level linkage among several Danish nationwide administrative registries, allowing record linkage analysis. (i) The Civil Registrations System holds information on sex, year of birth, civil status and the unique identifier on each Danish resident since 1968.23 (ii) The Danish National Patient Registry (DNPR) holds information on dates of admission and discharge, main and secondary discharge diagnoses according to the International Classification of Diseases, 10th revision (ICD-10), from 1994, and surgical procedure codes according to the NOMESCO Classification of Surgical Procedures (NCSP) from 1996. Since 2002, the DNPR has used the Diagnosis-Related Group system for hospital reimbursement.24 (iii) The Danish Register of Causes of Death keeps records on date and cause(s) of death classified according to ICD-10 since 1994.25 (iv) The Danish Heart Registry (DHR) is a clinical quality database that keeps track on invasive examinations and treatments, which include CAG, percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) since 2000. In addition, risk factors such as diabetes mellitus (DM) are recorded as well. The DHR has national coverage on CAG since 2006. National coverage is defined as more than 89% of the produces recorded in the DHR as well as in the DNPR in relation to the number of procedures registered in the DNPR. Each hospital performing CAG, PCI, or CABG is obligated to report data on performed procedures to the DHR.26 (v) The Danish National Registry of Medicinal Product Statistics holds information on date of dispensing, quantity dispensed, strength and formulation of all partially reimbursed prescription drugs dispensed from Danish pharmacies since 1995. Each drug dispensing is classified according to the International Anatomical Therapeutic Chemical (ATC) system.27 (vi) The Integrated Database for Labour Market Research holds information on taxed income gathered by government tax authorities.

y reimbursed prescription drugs dispensed from Danish pharmacies since 1995. Each drug dispensing is classified according to the International Anatomical Therapeutic Chemical (ATC) system.27 (vi) The Integrated Database for Labour Market Research holds information on taxed income gathered by government tax authorities. Study population We identified all patients ≥18 years with first recorded primary or secondary diagnosis of index MI (ICD-10: I21) from 1 January 2004 to 31 December 2010, and who had no prior MI admissions registered in the previous 12 months (Figure 1). Patients with a recent MI history were not included, because they probably had a different risk profile per se and were most likely already in dual treatment and revascularized. Figure 1 Selection of study populations.

Study population We identified all patients ≥18 years with first recorded primary or secondary diagnosis of index MI (ICD-10: I21) from 1 January 2004 to 31 December 2010, and who had no prior MI admissions registered in the previous 12 months (Figure 1). Patients with a recent MI history were not included, because they probably had a different risk profile per se and were most likely already in dual treatment and revascularized. Figure 1 Selection of study populations. Myocardial infarction population Patients who survived index MI, without stroke or recurrent MI within 7 days from discharge, were included. The quarantine period from index MI to 7 days after discharge, at which point all were alive, was chosen to allow assessment of medication at discharge. Co-morbidities, comprising prior MI, cerebrovascular disease, peripheral arterial disease, heart failure, arrhythmia, shock, pulmonary oedema, acute renal failure, chronic renal failure, DM, cancer, respiratory insufficiency, chronic obstructive pulmonary disease, anaemia, and infection, were defined as the presence of discharge diagnoses up to 10 years prior to the index MI, but prior MI was identified if any diagnosis since 1994. Identification of important co-morbidities in an unselected MI population was based on the Ontario acute MI mortality prediction rules,28 which was modified by expanding the time window for co-morbidity identification, identifying DM with as well as without complications and specifying additional co-morbidities of importance, including for type 2 MI.29,30 The incomplete capture of discharge code-based heart failure and DM was compensated by including loop diuretics and glucose-lowering drugs. Arrhythmia indicated cardiac arrest, paroxysmal tachycardia, atrial fibrillation/flutter, and other cardiac arrhythmias. Shock indicated cardiogenic, hypovolaemic, or other/unspecified shock. Infection indicated urinary tract infection and sepsis of bacterial or fungal origin. Revascularization was measured up to 7 days after hospital discharge. Concomitant medication was defined as redeemed prescriptions from 365 days before hospital admission until 7 days after discharge. Socioeconomic status was measured by means of 5-year average index income (in quintiles) and civil status (married, living with a partner or living alone). Table 1 Baseline characteristics of the index myocardial infarction population 7 days after hospital discharge and for the stable post-myocardial infarction population 366 days after discharge

sured by means of 5-year average index income (in quintiles) and civil status (married, living with a partner or living alone). Table 1 Baseline characteristics of the index myocardial infarction population 7 days after hospital discharge and for the stable post-myocardial infarction population 366 days after discharge Index MI Stable post-MI n = 55 747 n = 43 045 Age (IQR), years 70 (20) 68 (20) Age groups, years ≤49 4992 (9.0) 4033 (9.4) 50–59 8779 (15.7) 7362 (17.1) 60–69 13 277 (23.8) 11 096 (25.8) 70–79 14 153 (25.4) 10 656 (24.8) ≥80 14 546 (26.1) 9898 (23.0) Male 35 609 (63.9) 28 130 (65.4) CAD severity No significant stenosis 4145 (7.4) 4050 (9.4) 1-VD 15 122 (27.1) 14 011 (32.5) 2-VD 7777 (14.0) 6769 (15.7) 3-VD 7020 (12.6) 5566 (12.9) LMS 1607 (2.9) 1216 (2.8) Missing data on CAD severity 4223 (7.6) 3.577 (8.3) No performed CAG 15 853 (28.4) 7856 (18.3) Co-morbidity Prior MI 4413 (7.9) 3043 (7.1) Cerebrovascular disease 5526 (9.9) 3541 (8.2) Peripheral arterial disease 2239 (4.0) 1739 (4.0) Heart failurea 19 750 (35.4) 17 035 (39.6) Arrhythmiab 10 086 (18.1) 8846 (20.6) Shockc 218 (0.4) 190 (0.4) Pulmonary oedema 686 (1.2) 526 (1.2) Acute renal failure 1073 (1.9) 830 (.9) Chronic renal failure 1330 (2.4) 1040 (2.4) Diabetes mellitusd 11 290 (20.3) 9048 (21.0) Cancer 3852 (6.9) 2959 (6.9) Respiratory insufficiency 1203 (2.2) 980 (2.3) Chronic obstructive pulmonary disease 4861 (8.7) 3803 (8.8) Anaemia 3317 (6.0) 2955 (6.9) Infectione 1245 (2.2) 1087 (2.5) Revascularization PCI 25 425 (45.6) 24 050 (55.9) CABG 2331 (4.2) 3644 (8.5) No revascularization 30 322 (54.4) 16 201 (37.6) Concomitant medicationf β-Blockers 41 712 (74.8) 30 602 (71.1) Lipid-lowering treatment 41 971 (75.3) 32 639 (75.8) Aspirin 46 540 (83.5) 34 235 (79.5) Nitrate 16 030 (28.8) 7489 (17.4) P2Y12 inhibitors 34 882 (62.6) 26 144 (60.7) Clopidogrel 34 802 (62.4) 26 037 (60.5) Ticagrelor 0 4 (0.1) Prasugrel 97 (0.2) 106 (0.3) Glucose-lowering drugs 7528 (13.5) 5901 (13.7) Loop diuretics 17 373 (31.2) 14 185 (33.0) Vitamin-K antagonist/NOAC 4277 (7.7) 2759 (6.4) Spironolactone 4562 (8.2) 3248 (7.5) NSAID 16 207 (29.1) 5016 (11.7) PPI 15 344 (27.5) 9580 (22.3) Socioeconomic factors Yearly family income in quintiles 1 10 680 (19.2) 7179 (16.7) 2 10 538 (18.9) 7322 (17.0) 3 10 962 (19.7) 8317 (19.3) 4 11 626 (20.9) 9688 (22.5) 5 (highest) 11 941 (21.4) 10 539 (24.5) Civil status Married 29 553 (53.0) 23 968 (55.7) Living with a partner 3362 (6.0) 2864 (6.7) Living alone 22 832 (41.0) 16 213 (37.7) Contin

family income in quintiles 1 10 680 (19.2) 7179 (16.7) 2 10 538 (18.9) 7322 (17.0) 3 10 962 (19.7) 8317 (19.3) 4 11 626 (20.9) 9688 (22.5) 5 (highest) 11 941 (21.4) 10 539 (24.5) Civil status Married 29 553 (53.0) 23 968 (55.7) Living with a partner 3362 (6.0) 2864 (6.7) Living alone 22 832 (41.0) 16 213 (37.7) Contin uous and categorical variables were expressed as median (IQR) and frequency (%), respectively. MI, myocardial infarction; IQR, interquartile range; CAD, coronary artery disease; VD, vessel disease; LMS, left main stenosis; CAG, coronary angiography; PCI, percutaneous coronary intervention; CABG, coronary artery bypass graft; P2Y12, antiplatelet inhibitors; NOAC, new oral anticoagulants; NSAID, non-steroid anti-inflammatory drug; PPI, proton pump inhibitor. a Heart failure was defined as any discharge code indicative of heart failure or use of loop diuretics. b Arrhythmia was defined as any discharge code indicative of cardiac arrest, paroxysmal tachycardia, atrial fibrillation/flutter, and other cardiac arrhythmias. c Shock was defined as any discharge code indicative of cardiogenic, hypovolaemic, or other/unspecified shock. d Diabetes mellitus was defined as any discharge code indicative of diabetes mellitus or use of glucose-lowering drugs. e Infection was defined as any discharge code indicative of urinary tract infection and sepsis of bacterial or fungal origin.

c Shock was defined as any discharge code indicative of cardiogenic, hypovolaemic, or other/unspecified shock. d Diabetes mellitus was defined as any discharge code indicative of diabetes mellitus or use of glucose-lowering drugs. e Infection was defined as any discharge code indicative of urinary tract infection and sepsis of bacterial or fungal origin. f Redeemed prescriptions of concomitant medication: from 365 days before until 7 days after discharge for the index MI population, from day 244 to day 365 after discharge for the post-MI population, except for loop diuretics and glucose-lowering drugs, which for the stable group covered the period from 365 before until 365 days after discharge. Stable post-myocardial infarction population For ‘stable’ post-MI patients, who survived the first 365 days after the index MI, without a stroke or recurrent MI, identification of co-morbidities (discharge- as well as drug-based codes) was extended until 365 days after discharge. Revascularization was measured up to 365 days after hospital discharge. Concomitant medication (excluding loop diuretics and glucose-lowering drugs) was defined as redeemed prescriptions from day 244 to day 365 after discharge. Socioeconomic status was measured at index MI. A full list of the ICD-10, procedure (NCSP) and ATC codes used to identify co-morbidity, revascularization and medication is provided in the Supplementary material online, Table S1.

ering drugs) was defined as redeemed prescriptions from day 244 to day 365 after discharge. Socioeconomic status was measured at index MI. A full list of the ICD-10, procedure (NCSP) and ATC codes used to identify co-morbidity, revascularization and medication is provided in the Supplementary material online, Table S1. Coronary artery disease severity Identification of CAD severity was based on the findings from CAG and PCI performed up to 7 and 365 days after hospital discharge for the index MI and stable post-MI population, respectively. CAD severity was defined according to the number of obstructive coronary arteries corresponding to 50% or more narrowing and categorized into 7 groups: no significant stenosis, 1-, 2-, 3- VD, LMS, missing angiographic data or if no CAG was performed. LMS with or without additional diseased vessels were categorized as LMS only. For patients with >1 angiography record, the record with the most severe disease was retained for the analysis. Multi-vessel CAD (MVD) was defined as 2- or 3-VD or LMS.

1-, 2-, 3- VD, LMS, missing angiographic data or if no CAG was performed. LMS with or without additional diseased vessels were categorized as LMS only. For patients with >1 angiography record, the record with the most severe disease was retained for the analysis. Multi-vessel CAD (MVD) was defined as 2- or 3-VD or LMS. Endpoint and follow-up The primary composite endpoint was defined as the first recorded primary or supplementary diagnosis of MI (ICD10: I21), ischemic stroke (ICD10: I63, I64) or fatal cardiovascular disease (CVD) (ICD10: I00-I99). MI or stroke was considered to be non-fatal regardless of subsequent death at a later point in time (i.e. ≥1 day after non-fatal MI/stroke) and non-fatal MI or stroke appearing at the same time was classified as non-fatal MI. For the index MI patients, the length of follow-up was defined as the time elapsed from day 7 after hospital discharge until the primary composite endpoint, death, emigration, or end of 1-year follow-up (day 365 after discharge). For the stable post-MI patients, the length of follow-up was defined as the time elapsed from day 366 after hospital discharge until the primary composite endpoint, death, emigration, or end of study follow-up (31 December 2012).

mposite endpoint, death, emigration, or end of 1-year follow-up (day 365 after discharge). For the stable post-MI patients, the length of follow-up was defined as the time elapsed from day 366 after hospital discharge until the primary composite endpoint, death, emigration, or end of study follow-up (31 December 2012). Statistical methods Continuous and categorical variables were presented as median (interquartile range) and as frequencies (%), respectively. Kruskal–Wallis test was used for comparison of continuous variables, whereas chi-square test was used for comparison of categorical variables. Crude incidence rates of the composite endpoint and its components were calculated per 100 person-years (PY). The overall crude incidence rate of the composite endpoint was also stratified according to CAD severity, which was a constructed ordinal variable: no significant stenosis, 1-, 2-, 3-VD, LMS, missing data on CAD severity or if no CAG. The 1-year (from day 7 until day 365 after discharge) and beyond (from day 366 until end of study) cumulative incidence curves for the composite endpoint according to CAD severity were estimated using the Nelson–Aalen estimator that account for the competing event of death from non-cardiovascular causes. Logistic regression and Cox proportional-hazards models were used to estimate the 1-year and beyond impact of CAD severity and co-morbidity on the composite endpoint, respectively. The models were adjusted for potential confounders, which included age, age groups, sex, calendar year, revascularization status, medication and socioeconomic status (income and civil status). No significant stenosis, the age group of 50-59 years, highest yearly income and married as civil status served as reference groups for the analyses. Patients with missing data on CAD severity (around 8%) were excluded in the logistic regression and Cox proportional-hazard models (complete case analyses). Three sensitivity analyses of the 1-year and beyond impact of CAD severity and co-morbidity on composite endpoint were conducted with stepwise exclusion of: (i) those with missing information on CAD; (ii) those with index MI in year 2010 because of disproportionately high occurrence of missing information on CAD severity; and (iii) those who did not receive CAG. The proportional hazard assumption, linearity of continuous variable, and lack of interaction were found to be valid unless otherwise indicated.

on on CAD; (ii) those with index MI in year 2010 because of disproportionately high occurrence of missing information on CAD severity; and (iii) those who did not receive CAG. The proportional hazard assumption, linearity of continuous variable, and lack of interaction were found to be valid unless otherwise indicated. All statistical analyses and data management were carried out using SAS, version 9.4 (SAS Institute, Cary, NC, USA) and R statistic software (version 3.1.1). P-values less than 0.05 were considered statistically significant. Ethics Register-based studies do not require ethical approval according to Danish legislation. Approval was granted by the Danish Data Protection Agency (Ref.no. 2007-58-0015/local ref. GEH-2014-014 I-Suite no: 02732). Results Study populations During the study period, January 2004 to December 2010, 68 096 MI patients aged 18 years or older were hospitalized, of whom 55 747 (81.9%) patients survived and did not experience a recurrent MI or stroke 7 days after hospital discharge and were included in the study. Of the MI population, 43 045 patients (77.2%) survived 365 days without any subsequent MI or stroke and were classified as the stable post-MI population with a mean duration of follow-up time of 3.6 years and maximum follow-up time of 9 years (Figure 1 and Table 1).

er hospital discharge and were included in the study. Of the MI population, 43 045 patients (77.2%) survived 365 days without any subsequent MI or stroke and were classified as the stable post-MI population with a mean duration of follow-up time of 3.6 years and maximum follow-up time of 9 years (Figure 1 and Table 1). Irrespective of study population, when stratified according to CAD severity, age, and co-morbidity burden increased with increasing CAD severity, but at the same time, the proportion of revascularized and treatment with P2Y12 inhibitors declined as opposed to nitrates. Nevertheless, patients who did not receive CAG were older, had a greater co-morbidity burden, and received less optimal medication (Tables 2 and 3, Supplementary material online, Tables S2 and S3). Table 2 Index myocardial infarction patients’ treatment regimen by coronary artery disease severity

s opposed to nitrates. Nevertheless, patients who did not receive CAG were older, had a greater co-morbidity burden, and received less optimal medication (Tables 2 and 3, Supplementary material online, Tables S2 and S3). Table 2 Index myocardial infarction patients’ treatment regimen by coronary artery disease severity No significant stenosis n = 4145 1-VD 2-VD 3-VD LMS Missing data on CAD severity n = 4223 No CAG n = 15 853 P-value n = 15 122 n = 7777 n = 7020 n = 1607 Revascularization PCI 131 (3.2) 13 094 (86.6) 6110 (78.6) 3407 (48.5) 649 (40.4) 2034 (48.2) 0 (0) <0.001 CABG 16 (0.4) 107 (0.7) 302 (3.9) 1189 (16.9) 393 (24.5) 324 (7.7) 0 (0) <0.001 No revascularization 3998 (96.5) 1969 (13.0) 1434 (18.4) 2574 (36.7) 623 (38.8) 1884 (44.6) 15 853 (100) <0.001 Concomitant medicationa Nitrate 873 (21.1) 3075 (20.3) 2128 (27.4) 2676 (38.1) 648 (40.3) 1101 (26.1) 5529 (34.9) <0.001 β-Blockers 2717 (65.5) 12 926 (85.5) 6423 (82.6) 5281 (75.2) 1162 (72.3) 3271 (77.5) 9932 (62.7) <0.001 Lipid-lowering treatment 2963 (71.5) 13 761 (91.0) 6910 (88.9) 5743 (81.8) 1254 (78.0) 3552 (84.1) 7788 (49.1) <0.001 Aspirin 3026 (73.0) 13 604 (90.0) 6818 (87.7) 5724 (81.5) 1290 (80.3) 3646 (86.3) 12,32 (78.4) <0.001 P2Y12 inhibitors 1918 (46.3) 12 821 (84.8) 6161 (79.2) 4364 (62.2) 930 (57.9) 2961 (70.1) 5727 (36.1) <0.001 Categorical variables expressed as frequency (%). VD, vessel disease; LMS, left main stenosis; CAD, coronary artery disease; CAG, coronary angiography; PCI, percutaneous coronary intervention; CABG, coronary artery bypass graft; P2Y12, antiplatelet inhibitors. aRedeemed prescriptions of concomitant medication: from 365 days before until 7 days after discharge. Differences between the groups were found using chi-square test.

y artery disease; CAG, coronary angiography; PCI, percutaneous coronary intervention; CABG, coronary artery bypass graft; P2Y12, antiplatelet inhibitors. aRedeemed prescriptions of concomitant medication: from 365 days before until 7 days after discharge. Differences between the groups were found using chi-square test. Table 3 Stable post-myocardial infarction patients’ treatment regimen by coronary artery disease severity No significant stenosis n = 4050 1-VD 2-VD 3-VD LMS Missing data on CAD severity n = 3577 No CAG n = 7856 P-value n = 14 011 n = 6769 n = 5566 n = 1216 Revascularization PCI 234 (5.8) 12 532 (89.4) 5715 (84.4) 3102 (55.7) 571 (47.0) 1896 (53.0) 0 (0) <0.001 CABG 27 (0.7) 189 (1.3) 559 (8.3) 1850 (33.2) 534 (43.9) 485 (13.6) 0 (0) <0.001 No revascularization 3790 (93.6) 1387 (9.9) 687 (10.1) 1007 (18.1) 244 (20.1) 1230 (34.4) 7856 (100) <0.001 Concomitant medicationa Nitrate 560 (13.8) 1786 (12.7) 1118 (16.5) 1242 (22.3) 267 (22.0) 533 (14.9) 1983 (25.2) <0.001 β-Blockers 2288 (56.5) 10 900 (77.8) 5230 (77.3) 4362 (78.4) 923 (75.9) 2641 (73.8) 4258 (54.2) <0.001 Lipid-lowering treatment 2597 (64.1) 11 944 (85.2) 5777 (85.3) 4831 (86.8) 1056 (86.8) 2941 (82.2) 3493 (44.5) <0.001 Aspirin 2702 (66.7) 11 931 (85.2) 5713 (84.4) 4581 (82.3) 1012 (83.2) 2939 (82.2) 5357 (68.2) <0.001 P2Y12 inhibitors 1356 (33.5) 11 057 (78.9) 5202 (76.9) 3540 (63.6) 756 (62.2) 2259 (63.2) 1974 (25.1) <0.001 Categorical variables expressed as frequency (%).

85.3) 4831 (86.8) 1056 (86.8) 2941 (82.2) 3493 (44.5) <0.001 Aspirin 2702 (66.7) 11 931 (85.2) 5713 (84.4) 4581 (82.3) 1012 (83.2) 2939 (82.2) 5357 (68.2) <0.001 P2Y12 inhibitors 1356 (33.5) 11 057 (78.9) 5202 (76.9) 3540 (63.6) 756 (62.2) 2259 (63.2) 1974 (25.1) <0.001 Categorical variables expressed as frequency (%). VD, vessel disease; LMS, left main stenosis; CAD, coronary artery disease; CAG, coronary angiography; PCI, percutaneous coronary intervention; CABG, coronary artery bypass graft; P2Y12, antiplatelet inhibitors. a Redeemed prescriptions of concomitant medication from 244 to 365 days after discharge; B, from day 244 to day 365 after discharge. Differences between the groups were found using chi-square test. Furthermore, utilization of CAG, revascularization, and P2Y12 inhibitors increased as opposed to nitrates over the course of time. Patients enrolled in the early period were older and had more prevalent cardiac co-morbidities (MI and heart failure) (see Supplementary material online, Tables S4 and S5).

a Redeemed prescriptions of concomitant medication from 244 to 365 days after discharge; B, from day 244 to day 365 after discharge. Differences between the groups were found using chi-square test. Furthermore, utilization of CAG, revascularization, and P2Y12 inhibitors increased as opposed to nitrates over the course of time. Patients enrolled in the early period were older and had more prevalent cardiac co-morbidities (MI and heart failure) (see Supplementary material online, Tables S4 and S5). Myocardial infarction population Around 72% of the MI patients underwent CAG of which 41.1% had MVD. A total of 20 467 primary composite endpoint events (non-fatal MI, 46.0%; non-fatal stroke, 13.6%; cardiovascular death, 40.5%) were observed [10.2/100 PYs; 95% confidence interval (CI), 10.1-10.3], of which 11 129 events (non-fatal MI, 51.9%; non-fatal stroke, 10.0%; cardiovascular death, 38.2%) occurred within the first 365 days (23.8/100 PYs; 95% CI, 23.4–24.3), which corresponded to 20.0% (19.6–20.3), when accounting for the competing risk of non-cardiovascular death (Table 4 and Supplementary material online, Table S6). The incidence rate of the primary composite endpoint increased with increasing CAD severity but was highest in patients who did not receive CAG. Similarly, the cumulative risk of primary composite endpoint the first 365 days post-index MI rose from 8.4% (7.6–9.2) in those with no significant stenosis to 26.4% (24.3–28.6) in those with LMS (Figure 2A and Table 4). A higher cumulative risk was noted in patients with unknown CAD severity due to lacking invasive examination (35.5%, 34.8–36.3). After controlling for confounders, CAD severity was the most important risk factor in relation to co-morbidity (Figure 3A). Other important risk factors were increasing age, male gender, cerebrovascular disease, peripheral artery disease, shock, not receiving revascularization, not receiving secondary preventive medication (β-blockers, statin, aspirin, and P2Y12 inhibitors) and use of nitrates (Supplementary material online, Figure S1). Table 4 Incidence rate (per 100 person-years) and cumulative incidence with 95% confidence interval of the composite endpoint according to time since index myocardial infarction and after becoming stable post-myocardial infarction

in, and P2Y12 inhibitors) and use of nitrates (Supplementary material online, Figure S1). Table 4 Incidence rate (per 100 person-years) and cumulative incidence with 95% confidence interval of the composite endpoint according to time since index myocardial infarction and after becoming stable post-myocardial infarction Index MI population Stable post-MI population n = 55 747 n = 43 045 Incidence rates Cumulative incidence Incidence rates Cumulative incidenceb Composite endpointa 23.8 (23.4–24.3) 20.0 (19.6–20.3) 6.1 (5.9–6.2) 21.0 (20.6–21.4) No significant stenosis 9.0 (8.1–10.0) 8.4 (7.6–9.2) 3.8 (3.5–4.1) 13.6 (12.5–14.8) 1-VD 8.7 (8.2–9.1) 8.2 (7.8–8.7) 3.4 (3.2–3.5) 12.6 (12.0–13.2) 2-VD 16.4 (15.5–17.4) 14.5 (13.8–15.3) 4.8 (4.6–5.1) 17.3 (16.3–18.3) 3-VD 29.5 (28.1–30.9) 24.0 (23.0–25.0) 6.8 (6.5–7.2) 24.3 (23.1–25.5) LMS 33.2 (30.2–36.5) 26.4 (24.3–28.6) 7.5 (6.7–8.4) 25.2 (22.5–27.8) Missing data on CAD severity 29.4 (26.8–32.3) 15.9 (14.7–17.0) 7.6 (6.8–8.4) 18.5 (16.5–20.5) No CAG 50.4 (49.1–51.8) 35.5 (34.8–36.3) 13.3 (12.9–13.8) 41.0 (39.8–42.1) MI, myocardial infarction; VD, vessel disease; LMS, left main stenosis; CAD, coronary artery disease; CAG, coronary angiography. aNon-fatal MI, non-fatal stroke or cardiovascular death. b Four years after becoming stable.

Index MI population Stable post-MI population n = 55 747 n = 43 045 Incidence rates Cumulative incidence Incidence rates Cumulative incidenceb Composite endpointa 23.8 (23.4–24.3) 20.0 (19.6–20.3) 6.1 (5.9–6.2) 21.0 (20.6–21.4) No significant stenosis 9.0 (8.1–10.0) 8.4 (7.6–9.2) 3.8 (3.5–4.1) 13.6 (12.5–14.8) 1-VD 8.7 (8.2–9.1) 8.2 (7.8–8.7) 3.4 (3.2–3.5) 12.6 (12.0–13.2) 2-VD 16.4 (15.5–17.4) 14.5 (13.8–15.3) 4.8 (4.6–5.1) 17.3 (16.3–18.3) 3-VD 29.5 (28.1–30.9) 24.0 (23.0–25.0) 6.8 (6.5–7.2) 24.3 (23.1–25.5) LMS 33.2 (30.2–36.5) 26.4 (24.3–28.6) 7.5 (6.7–8.4) 25.2 (22.5–27.8) Missing data on CAD severity 29.4 (26.8–32.3) 15.9 (14.7–17.0) 7.6 (6.8–8.4) 18.5 (16.5–20.5) No CAG 50.4 (49.1–51.8) 35.5 (34.8–36.3) 13.3 (12.9–13.8) 41.0 (39.8–42.1) MI, myocardial infarction; VD, vessel disease; LMS, left main stenosis; CAD, coronary artery disease; CAG, coronary angiography. aNon-fatal MI, non-fatal stroke or cardiovascular death. b Four years after becoming stable. Figure 2 Cumulative incidence of composite endpoint in the index myocardial infarction (A) and stable post myocardial infarction population (B) according to coronary artery disease severity. A: day 7 after discharge until 1-year follow-up and B: day 366 after discharge until end of follow-up. VD, vessel disease; CAD, coronary artery disease; CAG, coronary angiography.

t in the index myocardial infarction (A) and stable post myocardial infarction population (B) according to coronary artery disease severity. A: day 7 after discharge until 1-year follow-up and B: day 366 after discharge until end of follow-up. VD, vessel disease; CAD, coronary artery disease; CAG, coronary angiography. Figure 3 Multivariable analyses showing the impact of CAD severity and co-morbidity on composite endpoint event in the index myocardial infarction population (A) and stable post-myocardial infarction population (B). The multivariable analysis is based on a complete case approach and adjusted for age, age-groups, gender, calendar year, revascularization, pharmacotherapy, and socioeconomic status. OR, odds ratio; HR, hazard ratio; CI, confidence interval; CAD, coronary artery disease; VD, vessel disease; CAG, coronary angiography; LMS, left main stenosis; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease.

s, gender, calendar year, revascularization, pharmacotherapy, and socioeconomic status. OR, odds ratio; HR, hazard ratio; CI, confidence interval; CAD, coronary artery disease; VD, vessel disease; CAG, coronary angiography; LMS, left main stenosis; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease. Stable post-myocardial infarction population Among the almost 82% of the stable post-MI patients that received CAG, 38.5% had MVD. Until study completion 9338 (non-fatal MI, 38.9%; non-fatal stroke, 17.9%; cardiovascular death, 43.2%) experienced a composite endpoint event (6.1/100 PYs; 95% CI, 5.9–6.2) (Table 4 and Supplementary material online, Table S6). A similar trend was noted here. The incidence rate of the primary composite endpoint increased with increasing CAD severity, but was highest in patients who did not receive CAG. Similarly, the cumulative incidence of the composite endpoints at 4 years follow-up after becoming stable was overall 21.0% (20.6–21.4), but 13.6% (12.5–14.8) in patients with no significant stenosis and 25.2% (22.5–27.8) in patients with LMS (Table 4 and Figure 2B). For patients who did not receive invasive examination, the risk was higher (41.0%, 39.8–42.1). After adjusting for confounders, CAD severity remained as the most important risk factor, but its relative importance in relation to co-morbidity was less pronounced (Figure 3B). Additionally, important risk factors were increasing age, male gender, cerebrovascular disease, peripheral artery disease, heart failure, not receiving revascularization, not receiving statins and use of nitrates (Supplementary material online, Figure S2).

ce in relation to co-morbidity was less pronounced (Figure 3B). Additionally, important risk factors were increasing age, male gender, cerebrovascular disease, peripheral artery disease, heart failure, not receiving revascularization, not receiving statins and use of nitrates (Supplementary material online, Figure S2). Sensitivity analysis To test the robustness of our results, stepwise sensitivity analyses were carried out for both study populations. Excluding patients with missing information showed no changes for the risk of events. Data from the DHR in 2010 showed a higher proportion of missing data than in the preceding years (see Supplementary material online, Tables S4 and S5), and by restricting the analysis to the 2004–2009 period no appreciable changes were noted. By further restricting the analyses to patients with known CAD severity status only, the CAD severity-stratified estimates remained identical, but the overall estimate reduced from 20.0% (19.6–20.3) to 13.50% (13.1–13.9) in the index population. Correspondingly, the overall estimate in the stable post-MI population changed from 21.0% (20.6–21.4) to 16.3% (15.9–16.8) (data not shown). A similar stepwise approach was carried out for the multivariable analyses and no appreciable changes were noted for co-morbidity impact on outcome (see Supplementary material online, Tables S7 and S8), but for CAD severity, an increase in the estimates was noted but with overlapping CIs.

(15.9–16.8) (data not shown). A similar stepwise approach was carried out for the multivariable analyses and no appreciable changes were noted for co-morbidity impact on outcome (see Supplementary material online, Tables S7 and S8), but for CAD severity, an increase in the estimates was noted but with overlapping CIs. Discussion In this nationwide cohort study, encompassing almost 56 000 MI patients with a follow-up period of up to 9 years, we examined the risk of recurrent cardiovascular events (non-fatal MI, non-fatal stroke or cardiovascular death) in 2 distinct study populations. The key findings were: (i) one in five patients experienced a recurrent cardiovascular event the first 365 days, but for patients surviving the first 365 days without a recurrent cardiovascular event after MI, the risk remained equally high, with one in five patients experiencing an event later; (ii) CAD severity remained as the most critical factor for recurrent cadiovascular events both before and after the first 365 days; (iii) co-morbidity was a strong risk factor for cardiovascular events, but its relative importance was more pronounced in the long term.

h one in five patients experiencing an event later; (ii) CAD severity remained as the most critical factor for recurrent cadiovascular events both before and after the first 365 days; (iii) co-morbidity was a strong risk factor for cardiovascular events, but its relative importance was more pronounced in the long term. The findings from this present study add valuable knowledge to the limited evidence on impact of CAD severity on recurrent events in a nationwide post-MI cohort who had been stable for at least 1 year. Using CAD severity, which is a well-established risk factor for cardiovascular events to identify high-risk patients, we found a dose–response relationship, but since CAD severity had been modified by important factors such as revascularization and pharmacotherapy at baseline, which differed across CAD severity, we did not measure the true effect of CAD severity. Furthermore, patients with no significant stenosis and to a certain extent those conservatively treated probably had a different underlying cause for MI,22,31 but examining the extremes of MI patients is valuable, particularly because these patients are not rare and pose a clinical challenge.

ot should be seen as a generic means of examining small study effect, which is the tendency for smaller studies in a meta-analysis to show larger treatment effects. To avoid evaluating publication bias only according to visually judgement, this was complemented by Egger’s test of asymmetry applied on the funnel plot.16 We used the trim and fill simulation method to detect and control for bias.17 In the presence of publication bias, the trim and fill method could help reduce bias in pooled estimates. Even though the performance is not ideal, this method is a kind of sensitivity analysis to assess the potential impact of missing studies. This then allows an adjusted overall estimate with CI to be calculated. A test of the presence of bias could be derived from this method based on the estimated number of missing studies. The estimated effect of the missing studies provides an indication of whether the imputed missing studies affect the overall result of the meta-analysis. We followed the PRISMA guidelines for meta-analyses and systematic reviews of observational studies in reporting the present study.18 All statistical analyses were performed with Stata version 15 (Stata Corporation, College Station, TX, USA) or R Package-meta (Guido Schwarzer, R News 2007).

ue effect of CAD severity. Furthermore, patients with no significant stenosis and to a certain extent those conservatively treated probably had a different underlying cause for MI,22,31 but examining the extremes of MI patients is valuable, particularly because these patients are not rare and pose a clinical challenge. Prevalence of multi-vessel disease In this present nationwide study on MI with a small percentage of non-naïve patients, the prevalence of MVD (2-VD or higher) among CAG recipients was about 41%. This is consistent with a previous Danish study on first-time MI.32 Another study, with an almost similar setup to our study, reported a nearly identical prevalence of MVD (38%) in the index MI,4 and as expected, a lower prevalence of 30% was reported when only focusing on 3-VD, LMS, and prior CABG.3 The ICD-10 classification system does not account for the different subtypes of MI. Current evidence suggests that type 1 MI [traditionally corresponding to ST-segment elevation (STEMI) and non-ST-segment elevation (NSTEMI)] constitutes the largest group.33 In STEMI and NSTEMI studies, MVD counts for around 50%.34–36 Although almost 39% had MVD in the stable post-MI population in the present study, others have reported in the range of 34–58%,37–39 reflecting important differences in how their stable populations were selected. Altogether, the prevalence of MVD in MI and stable post-MI is comparable with national and international findings, but more importantly, this shows that a large proportion of MI patients surviving the first year without events had a high coronary atherosclerotic burden.

es in how their stable populations were selected. Altogether, the prevalence of MVD in MI and stable post-MI is comparable with national and international findings, but more importantly, this shows that a large proportion of MI patients surviving the first year without events had a high coronary atherosclerotic burden. Furthermore, we showed in both populations that age and co-morbidity burden increased with increasing CAD severity, which is in line with other studies.3,39–41 As expected, we found that MI patients being event-free on the first year had a more favourable risk profile (younger, lower proportion of MVD and less degree of prior MI and stroke), an observation also made by others.5 On that note, a considerable proportion of MI did not receive CAG. These patients were older and most likely had a higher occurrence of MVD, which in turn means that the true number of MVD was probably underestimated. Nevertheless, the complexity of MVD patients underscores the need for a coordinated effort involving optimal treatment of CAD along with appropriate management of significant co-morbidities. The importance of this is emphasized further by the expected increasing prevalence of CAD combined with the high and increasing burden of co-morbidity,1 leading ultimately to an increasing economic burden on the health-care system.

ptimal treatment of CAD along with appropriate management of significant co-morbidities. The importance of this is emphasized further by the expected increasing prevalence of CAD combined with the high and increasing burden of co-morbidity,1 leading ultimately to an increasing economic burden on the health-care system. Cardiovascular risk and risk factors We demonstrated a considerable residual risk for cardiovascular events during the first year and the first 4 years of follow-up after becoming stable post-MI. More precisely, one out of 5 MI survivors experienced an event the first year and the same risk was observed after 4 years in stable post-MI patients. Studies on long-term prognosis have used different approach to identify high-risk patients.4–6 Similar to the post hoc study on the PLATelet inhibition and patient Outcomes study (PLATO),3 we focused on CAD severity, which, apart from being the best marker for coronary atherosclerosis burden, indirectly reflects the co-morbidity burden. Thus, the study3 found a 1-year risk of 16.3% when having MVD (defined as 3-VD, LMS or prior CABG), and a study6 restricted to health insurance beneficiaries demonstrated an almost 21% risk after 4 years in high-risk (defined as DM, prior MI or/and chronic end stage kidney disease) stable post-MI patients. When focusing on overall estimates, a recent nationwide Swedish study5 reported, similarly to our study, an 18.3% risk on the first year but a higher risk in the stable post-MI population (20% after 3 years vs. 21% after 4 years). This latter difference might reflect the fact that we, as opposed to Jernberg et al.,5 excluded high-risk patients (non-fatal CV events) at baseline. In a Spanish study, the corresponding estimates after 1 and 4 years were 7.3% and 10.1%, respectively.4 These lower estimates might be expected in single-centre studies with higher rate of revascularization and with a closer clinical follow-up program. Nevertheless, we showed a relatively high residual risk and that the risk increased with increasing CAD severity not only during the first year (8.4–26.4%) but also in the successive years (13.6–25.2%).

expected in single-centre studies with higher rate of revascularization and with a closer clinical follow-up program. Nevertheless, we showed a relatively high residual risk and that the risk increased with increasing CAD severity not only during the first year (8.4–26.4%) but also in the successive years (13.6–25.2%). More importantly, those treated non-invasively had the highest risk (35.5% and 41.0%), as reported elsewhere.7,35,42,43 However, after adjusting for confounders including those related to type 2 MI, MVD appeared as the most important risk factor for cardiovascular events at 1-year and beyond. This is supported by 1-year- and partially by longer follow-up studies.3,4,38 Similar to others,4 we showed that the relative importance of MVD was greater in the first year than in the following years. At the same time, the importance of co-morbidity in relation to MVD was more pronounced in the stable post-MI population, which is in line with other studies.4,38 It is important to bear in mind that some co-morbidities (e.g., cerebrovascular disease and peripheral artery disease) share the same risk factor as for CAD, and in general, the extent of atherosclerosis in the coronary and non-coronary beds is a strong determinant of long-term prognosis.44 This suggests that the impact of factors other than MVD becomes more important during long-term follow-up after the CAD has stabilized.

ry disease) share the same risk factor as for CAD, and in general, the extent of atherosclerosis in the coronary and non-coronary beds is a strong determinant of long-term prognosis.44 This suggests that the impact of factors other than MVD becomes more important during long-term follow-up after the CAD has stabilized. Follow-up and secondary prevention We showed a slightly lower degree of initiation of secondary preventive drugs (β-blockers, aspirin P2Y12 inhibitors and statins) and utilization of invasive strategy in MI as compared to other national studies.7,32 The most likely reason for this difference lies in the shorter period for initiating the medication and recording the invasive procedures. Similar to other studies from abroad, the majority received evidence-based therapies for CAD at baseline4,5 and remained on it in the stable post-MI population.5 Although there have been improvements in initiating secondary preventive drugs, a large group of MI patients faced undertreatment (primarily in non-obstructive CAD and conservative strategy) as reported elsewhere45; thus, initiating medication at discharge is a critical factor for medication adherence.46 Performing CAG is another critical factor that plays a part in the initiation of secondary prevention drugs.7,47,48 A significant proportion of both study populations were assigned a conservative strategy, a proportion similar to other studies on MI and stable CAD.32,35,37,42,47,48

al factor for medication adherence.46 Performing CAG is another critical factor that plays a part in the initiation of secondary prevention drugs.7,47,48 A significant proportion of both study populations were assigned a conservative strategy, a proportion similar to other studies on MI and stable CAD.32,35,37,42,47,48 With increasing CAD severity, the rate of revascularization dropped and the prevalence of nitrates increased, and for those selected for a conservative strategy, a less aggressive treatment was given. As in previous studies,7,42,43 these patients also appeared to be older and with greatest co-morbidity burden, crucial factors that are usually considered in the risk benefit analysis of an invasive strategy.47,48 While some found conservative strategy as being clinically justifiable,47 others showed that the risk profile has been underestimated,49 indicating a risk-treatment mismatch where high-risk patients have a lower likelihood to receive CAG,40,42 even though the benefit of an invasive strategy increases with baseline risk.50

le some found conservative strategy as being clinically justifiable,47 others showed that the risk profile has been underestimated,49 indicating a risk-treatment mismatch where high-risk patients have a lower likelihood to receive CAG,40,42 even though the benefit of an invasive strategy increases with baseline risk.50 In this present study, we showed that the residual risk in stable post-MI patients with MVD was increased even though we adjusted for treatment. This indicates that selected patients might benefit from individualized treatment. That said, we also need to be better to achieve our treatment goals, as a large share of patients received suboptimal secondary prevention drugs (more pronounced in stable post-MI), invasive strategy (more pronounced in index MI) and according to the Euroaspire survey III do not enter cardiac rehabilitation programs after discharge.8

lso need to be better to achieve our treatment goals, as a large share of patients received suboptimal secondary prevention drugs (more pronounced in stable post-MI), invasive strategy (more pronounced in index MI) and according to the Euroaspire survey III do not enter cardiac rehabilitation programs after discharge.8 Special consideration of myocardial infarction subgroups Special attention should be made to patients with no significant stenosis and conservatively treated patients, as their underlying cause for MI is different from obstructive CAD.22,31 Both represent the opposite ends of the risk spectrum. As outlined above, the high risk associated with conservative strategy is understandable. Similarly, although the risk is lower in patients with no significant stenosis compared to those with more extensive CAD, it is still high and might be linked to undertreatment as well,41 which is in line with our results. More importantly, for both populations, there is no clear evidence of appropriate treatment recommendations even though these patients are not rare in clinical practice.

enosis compared to those with more extensive CAD, it is still high and might be linked to undertreatment as well,41 which is in line with our results. More importantly, for both populations, there is no clear evidence of appropriate treatment recommendations even though these patients are not rare in clinical practice. Limitations The strength of this study lies in the large nationwide register-based cohort design, which provided a national study population on MI with close to complete information on CAD severity (8% missing) among CAD recipients and long-term follow-up data. A number of limitations inherent to the nature of retrospective studies are present. First, given that the positive predictive of the diagnosis for MI has been found high,51 it was not possible to differentiate among STEMI, NSTEMI, and type 2 MI, which is a subgroup with emerging interest. Thus, the case mix of MI patients affect the generalizability of the study results, as STEMI and NSTEMI have different risk/prognosis. Second, in 2010, the DHR changed its data structure, and therefore, some variables had missing data, including for the CAD severity for 2010. We believe the information on CAD severity was missing at random. Another important point is that the DHR has national coverage on CAG only from 2006, which meant that a higher proportion of CAG without information on CAD (obtained from the DNPR) was noted in the preceding years. Sensitivity analyses showed no significant changes in the main results, other than an almost 7% point reduction in the overall cumulative incidence noted in both groups when conservatively treated patients were excluded. Third, some patients with LMS had evidence of 2- or 3-VD, giving them a more different risk profile. Furthermore, the estimation of CAD severity relied on the basis of the number of coronary vessels with a stenosis more than 50%, which is somehow less informative than such as the Syntax score, which was not registered in the DHR. Fourth, regardless of the number of CAG performed during index hospitalization, patients with different findings on the angiogram were categorized according to the most severe CAD severity. Fifth, during the study years, the implementation of high-sensitive assays most likely, to a certain extent, resulted in reclassification of unstable angina pectoris to NSTEMI, which meant that certain NSTEMI might have had a lower risk than NSTEMI in the preceding period.

zed according to the most severe CAD severity. Fifth, during the study years, the implementation of high-sensitive assays most likely, to a certain extent, resulted in reclassification of unstable angina pectoris to NSTEMI, which meant that certain NSTEMI might have had a lower risk than NSTEMI in the preceding period. Finally, there is a lack of information about important clinical parameters such as complete vs. incomplete revascularization, fibrinolytic agents, left ventricular ejection fraction, infarct size, blood pressure, body mass index, lipid levels, and cardiac rehabilitation programs. Thus, we were unable to account for achievement of treatment goals, and therefore, we cannot rule out the effect of unmeasured confounders while estimating the effect of CAD severity and co-morbidity. Conclusions This nationwide study showed that stable post-MI patients had increased risk of cardiovascular events beyond the first year and was strongly related to CAD severity. Despite medical treatment and that a large part of patients with multi-vessel disease underwent revascularization, CAD severity was the strongest risk factor for recurrent cardiovascular event in a long-term perspective. Supplementary material Supplementary material is available at European Heart Journal—Cardiovascular Pharmacotherapy online. Funding This work was supported by AstraZeneca.

Conclusions This nationwide study showed that stable post-MI patients had increased risk of cardiovascular events beyond the first year and was strongly related to CAD severity. Despite medical treatment and that a large part of patients with multi-vessel disease underwent revascularization, CAD severity was the strongest risk factor for recurrent cardiovascular event in a long-term perspective. Supplementary material Supplementary material is available at European Heart Journal—Cardiovascular Pharmacotherapy online. Funding This work was supported by AstraZeneca. Conflict of interest: A.D. and H.N.C. are employed by AstraZeneca. G.H.G. is supported by an unrestricted research scholarship from the Novo Nordisk Foundation and reports research grants from AstraZeneca, Pfizer, Bayer, Bristol-Myers Squibb, and Boehringer Ingelheim. Supplementary Material Supplementary Tables and Figure S1 Click here for additional data file.

Introduction Oral β-blockers have been a central component of secondary prevention pharmacotherapy following acute myocardial infarction (AMI) irrespective of its severity for decades. Recent international guidelines on the management of coronary disease, however, call into question the efficacy of β-blockers.1–4 The foremost reason for this is because studies of β-blockers among patients following AMI were conducted prior to the implementation of acute coronary revascularization and the use of modern secondary preventive treatments. Moreover, landmark studies which established the rationale for the routine use of long-term oral β-blockade after AMI were published in the early 1980s.5,6 The only randomized large-scale β-blocker trial conducted in patients following AMI in recent years,7 found no prognostic benefit of early intravenous metoprolol followed by 4 weeks of oral treatment compared with placebo. A meta-analysis of randomized, controlled trials did not find a mortality effect associated with β-blockers in studies from the reperfusion era, as opposed to a significant reduction in mortality for studies published in the pre-reperfusion era.8

s metoprolol followed by 4 weeks of oral treatment compared with placebo. A meta-analysis of randomized, controlled trials did not find a mortality effect associated with β-blockers in studies from the reperfusion era, as opposed to a significant reduction in mortality for studies published in the pre-reperfusion era.8 The incidence of AMI remains high and many patients with AMI who do not have reduced left ventricular systolic ejection fraction (LVEF) and/or heart failure (HF) receive oral β-blockers. Whilst β-blockers are considered relatively safe and inexpensive, they do have well-known side effects, and adherence to other (potentially more efficacious) secondary preventive medications may wane as a result of concomitant use of β-blockers.9 Given the absence of randomized controlled trials to test the efficacy of β-blockers in contemporary AMI patients without reduced left ventricular function or HF are lacking, meta-analyses of population-based studies are potentially of value for guiding β-blocker treatment in clinical practice. We hypothesized that the survival benefit of β-blockers observed in historical trials may not be present in the contemporary post-AMI population. As such, we aimed to estimate the effect of oral β-blockers on mortality in patients with both ST-elevation myocardial infarction (STEMI) and non-STEMI (NSTEMI) where the majority of patients did not have reduced LVEF and/or no clinical signs of HF. Methods The review protocol is registered at https://www.crd.york.ac.uk/PROSPERO/, ID: CRD42017079199.

We hypothesized that the survival benefit of β-blockers observed in historical trials may not be present in the contemporary post-AMI population. As such, we aimed to estimate the effect of oral β-blockers on mortality in patients with both ST-elevation myocardial infarction (STEMI) and non-STEMI (NSTEMI) where the majority of patients did not have reduced LVEF and/or no clinical signs of HF. Methods The review protocol is registered at https://www.crd.york.ac.uk/PROSPERO/, ID: CRD42017079199. Eligibility criteria All study types and sizes published after 1 January 2000 concerning patients following AMI were eligible for inclusion. Studies where none or only a minority of patients had a history of HF, were in Killip class ≥III or had LVEF <40% at baseline, were included. It was anticipated that not all studies would have complete data on these three categories reflecting HF and/or left ventricular (LV) systolic dysfunction. Studies that did not provide estimates between the β-blocker group and the no β-blocker group were excluded.

class ≥III or had LVEF <40% at baseline, were included. It was anticipated that not all studies would have complete data on these three categories reflecting HF and/or left ventricular (LV) systolic dysfunction. Studies that did not provide estimates between the β-blocker group and the no β-blocker group were excluded. Study selection and search The literature search strategy is presented in Supplementary material online, Tables S1 and S2. We searched the electronic bibliographic databases Embase and Medline(r) for studies written in English from inception until 18 July 2017, with an additional search undertaken per 30 October 2017. After removal of duplicate references, two members of the review team undertook initial screening of article titles and abstracts. Potentially, relevant articles were obtained in full-text and read independently by three review team members. Conflicts were resolved by consensus. Reference lists were scrutinized to identify articles not included in the original search. Table 1 Characteristics of the 16 cohort studies included in the meta-analysis

. Potentially, relevant articles were obtained in full-text and read independently by three review team members. Conflicts were resolved by consensus. Reference lists were scrutinized to identify articles not included in the original search. Table 1 Characteristics of the 16 cohort studies included in the meta-analysis First author (Publication year) Country Inclusion period β- Blocker Total cohort Control for confounding Timing of the study Follow-up (years), median Age (years), mean Men (%) STEMI (%) PCIa (%) LVEF (%) History of HF (%) Killip ≤2 (%) Diabetes (%) Hypert- ension (%) Smoking (%) Prior MI (%) ASA (%) Statins (%) ARB/ ACEi(%) Kernis (2004)19 USA/ Europe 1991–1999 1661 2442 PS adjusted Retrospective 0.5 60.6 73.7 100 100 48.9 2.3 98.7 16.6 44.9 66.2 13.8 — — — Yamada (2006)20 Japan 1994–2001 400 546 Multivariate Prospective 2.0 63.0 75.5 82.5b 61.1 54.0 — 87.7 37.4 41.3 64.3 0 92.1 31.9 51.6 Ozasa (2010)21 Japan 2004–2006 349 910 PS adjusted Retrospective 3.0 67.4 76.0 100 100 52.3 17.0 — 38.0 68.0 38.0 8 99.1 54.6 76.2 Bangalorec (2012)22 USA/ Europe 2003–2004 3379 6758 PS matched Retrospective 3.6 68.6 75.1 — — — 22.3 — 37.3 73.6 9.7 — 75.8 74.5 69.4 Bao (2013)23 Japan 2005–2007 1614 3692 Multivariate Retrospective 2.6 67.1 74.6 100 100 53.5 27.3 — 31.4 78.7 41.6 8.6 99.5 56.6 75.7 Nakatani (2013)24 Japan 1998–2011 2880 5628 PS adjusted Retrospective 3.9 64.7 77.3 100 100 — — 85.4d 32.8 59.5 65.9 10.9 94.6 44.3 77.1 Bangalorec (2014)25 USA/ Europe 2002–2003 981 1962 PS matched Retrospective 2.3 64.5 79.4 — — — 0 — 35.4 69.7 18.2 — 98.0 80.4 17.7 Choo (2014)26 Korea 2004–2009 2424 3019 PS adjusted Retrospective 3.0 61.3 73.2 58.1 100 60.4 — 93.8 40.6 50.0 44.0 3.3 99.7 90.4 81.8 Yang (2014)27 Korea 2005–2010 2650 3975 PS matched Retrospective 1.0 65.7 73.0 100 100 50.0 0.9 85.3 25.3 43.6 45.7 6.5 98.6 80.8 75.9 Lee (2015)28 Korea 2003–2009 598 901 Multivariate Retrospective 4.5 57.7 79.5 100 100 51.7 — 87.9 21.9 40.1 64.8 3.2 99.2 66.3 93.2 Raposeiras-Roubín (2015)29 Spain 2003–2012 555 1110 PS matched Retrospective 5.2 66.1 69.0 28.0 65.2 —e 10.8 — 25.9 56.7 27.6 9 87.2 82.8 58.9 Hioki (2016)30 Japan 2008–2010 251 444 PS adjusted Retrospective 2.9 65.7 81.8 81.8 100 56.1 — 100 24.1 22.7 65.1 — — 100 85.1 Konishi (2016)31 Japan 1997–2011 103 206 PS matched Retrospective 4.7 64.6 80.6 100 100 56.4 0 — 41.3 61.7 35.9 0 99.0 51.5 80.1 Leef (2016)32 Korea 2009–2013 3683 7261 Multivariate Retrospective 2.4 62.5 75.1 — 100 — 2.9 — 27.1 30.1 — 0 88.0 93.1 0 Puymiratg (2

65.7 81.8 81.8 100 56.1 — 100 24.1 22.7 65.1 — — 100 85.1 Konishi (2016)31 Japan 1997–2011 103 206 PS matched Retrospective 4.7 64.6 80.6 100 100 56.4 0 — 41.3 61.7 35.9 0 99.0 51.5 80.1 Leef (2016)32 Korea 2009–2013 3683 7261 Multivariate Retrospective 2.4 62.5 75.1 — 100 — 2.9 — 27.1 30.1 — 0 88.0 93.1 0 Puymiratg (2 016)33 France 2005 1783 2217 Multivariate Prospective 1.0 64.4 72.0 56.0 48.7 55.0 0 100 31.8 54.8 32.5 14.5 — 32.2 37.5 Dondo (2017)34 UK 2007–2013 141 097 148 314 PS adjusted Retrospective 1.0 63.5 71.0 53.0 45.9h — 0 — 11.4 33.6 62.3 0 96.7 96.3 88.3 a At index AMI. b Q-wave MI. c Subgroup of patients with known prior myocardial infarction. d Killip >1. e All patients had LVEF ≥ 50%. f Subgroup of patients who received β-blocker only vs. no drug. g One-year population. h In total, 68 095/148 314 (45.9%) had in-hospital coronary intervention (PCI/CABG) and 49 087/68 095 (72.1%) was treated with primary PCI for STEMI. HF, heart failure; LVEF, left ventricular ejection fraction; PCI, percutaneous coronary intervention; PS, propensity score; STEMI, ST-elevation myocardial infarction; UK, United Kingdom; USA, United States of America. Table 2 Subgroup analysis performed according to patient and study characteristics considered as potential sources of heterogeneity for outcome all-cause mortality

HF, heart failure; LVEF, left ventricular ejection fraction; PCI, percutaneous coronary intervention; PS, propensity score; STEMI, ST-elevation myocardial infarction; UK, United Kingdom; USA, United States of America. Table 2 Subgroup analysis performed according to patient and study characteristics considered as potential sources of heterogeneity for outcome all-cause mortality Subdivision N RR (95% CI) RR = 1, Z P-value Variation in RR due to heterogeneity, I2 (%) All studies 16 0.74 (0.64–0.85) 4.20 <0.001 67.4 ST-elevation myocardial infarction All patients 7 0.70 (0.52–0.93) 2.44 0.015 70.3 Mixed/unclear 9 0.78 (0.67–0.91) 3.16 0.002 61.9 PCI All patients 10 0.68 (0.54–0.86) 3.29 0.001 65.9 Mixed/unclear 6 0.83 (0.71–0.97) 2.39 0.017 57.8 Follow-up by quartiles 0.5–1.5 years 4 0.64 (0.41–1.01) 1.90 0.057 82.0 1.5–2.7 years 4 0.85 (0.70–1.03) 1.67 0.095 19.9 2.7–3.8 years 4 0.75 (0.51–1.10) 1.48 0.138 68.6 3.8–5.2 years 4 0.68 (0.51–0.91) 2.64 0.008 61.3 Timing of the study Prospective 2 0.65 (0.44–0.98) 2.08 0.038 0.0 Retrospective 14 0.75 (0.65–0.87) 3.86 <0.001 69.8 Control for confounding Propensity score analysis 11 0.74 (0.62–0.78) 3.48 0.001 70.0 Multivariate analysis 5 0.74 (0.56–0.97) 2.19 0.029 61.4 CI, confidence interval; RR, rate ratio.

91) 2.64 0.008 61.3 Timing of the study Prospective 2 0.65 (0.44–0.98) 2.08 0.038 0.0 Retrospective 14 0.75 (0.65–0.87) 3.86 <0.001 69.8 Control for confounding Propensity score analysis 11 0.74 (0.62–0.78) 3.48 0.001 70.0 Multivariate analysis 5 0.74 (0.56–0.97) 2.19 0.029 61.4 CI, confidence interval; RR, rate ratio. Quality assessment The Newcastle–Ottawa Scale (NOS) for cohort studies10 was used to assess the quality of the included studies according to (i) methods for study participant selection, (ii) appropriate control for confounding (comparability), and (iii) methods for assessing the outcome. We further assessed the timing of the study (prospective vs. retrospective), and methods used to control for confounding (propensity score analysis vs. multivariate analysis). Data abstraction The primary endpoint considered was all-cause mortality. Publication status, study design, patient-related characteristics, and results were extracted on a standardized form according to an a priori protocol. Investigators were contacted for additional data. Patient-related variables were mean age of the cohort, frequency of male sex, diabetes mellitus, hypertension, smoking, previous myocardial infarction (MI); treatment with acetylsalicylic acid (ASA), statins, angiotensin receptor blockers (ARBs)/angiotensin-converting enzyme inhibitors (ACEI), in addition to LVEF, Killip class, history of HF, STEMI/NSTEMI, and percutaneous coronary intervention (PCI).

es mellitus, hypertension, smoking, previous myocardial infarction (MI); treatment with acetylsalicylic acid (ASA), statins, angiotensin receptor blockers (ARBs)/angiotensin-converting enzyme inhibitors (ACEI), in addition to LVEF, Killip class, history of HF, STEMI/NSTEMI, and percutaneous coronary intervention (PCI). Quantitative data synthesis Statistical pooling The method used to combine results from individual studies, was based on the adjusted risk estimate and its 95% confidence intervals (CIs) obtained from each study. To obtain summary measures, a random effects model according to the DerSimonian Laird method11 was used because of the heterogeneity among studies.

We used the trim and fill simulation method to detect and control for bias.17 In the presence of publication bias, the trim and fill method could help reduce bias in pooled estimates. Even though the performance is not ideal, this method is a kind of sensitivity analysis to assess the potential impact of missing studies. This then allows an adjusted overall estimate with CI to be calculated. A test of the presence of bias could be derived from this method based on the estimated number of missing studies. The estimated effect of the missing studies provides an indication of whether the imputed missing studies affect the overall result of the meta-analysis. We followed the PRISMA guidelines for meta-analyses and systematic reviews of observational studies in reporting the present study.18 All statistical analyses were performed with Stata version 15 (Stata Corporation, College Station, TX, USA) or R Package-meta (Guido Schwarzer, R News 2007). Results Study selection After identifying 7529 references, 7499 were excluded due to irrelevant content and duplicate publications, leaving 30 potentially eligible studies. Fourteen of these did not fulfil the inclusion criteria (see Supplementary material online, Table S3) and 16 studies19–34 were thus included in the meta-analysis (Figure 1). Additional data were obtained for three studies.23,29,34Table 3 Meta-regression model between risk of all-cause mortality and the different patient-and study-level variables

ulfil the inclusion criteria (see Supplementary material online, Table S3) and 16 studies19–34 were thus included in the meta-analysis (Figure 1). Additional data were obtained for three studies.23,29,34Table 3 Meta-regression model between risk of all-cause mortality and the different patient-and study-level variables Covariates N Level β-Coefficient Standard error t P-value aτ2 bHeterogeneity (%) None 16 — −0.3105 0.0794 −3.91 0.001 0.05396 — ST-elevation myocardial infarction 16 1/0 −0.0649 0.1674 −0.39 0.704 0.05861 −8.63 PCI 16 1/0 −0.1325 0.1625 −0.82 0.429 0.05621 −4.16 Median follow-up time 16 Years 0.0064 0.0597 0.11 0.917 0.06240 −15.65 Mean left ventricular ejection fraction 10 Percent 0.0133 0.0379 0.35 0.734 0.08332 −18.63 Mean age of patients in the cohort 16 Years 0.0530 0.0245 2.16 0.049 0.03697 31.48 Frequency in cohort Men 16 Percent −0.0130 0.0252 −0.52 0.614 0.06012 −11.42 Diabetes mellitus 16 Percent 0.0051 0.0094 0.54 0.596 0.06396 −18.53 Hypertension 16 Percent 0.0081 0.0050 1.63 0.125 0.05842 −8.27 Smokers 15 Percent −0.0052 0.0045 −1.15 0.270 0.06757 −11.18 Previous MI 13 Percent −0.0037 0.0184 −0.20 0.843 0.06657 −13.04 Heart failure 11 Percent 0.0109 0.0080 1.35 0.209 0.04209 −14.71 ASA 13 Percent −0.0077 0.0102 −0.76 0.463 0.04577 −12.21 Statin 15 Percent −0.0011 0.0039 −0.30 0.772 0.04979 −18.09 ARB/ACEi 15 Percent −0.0014 0.0029 −0.49 0.633 0.05040 −19.55 Country (Asia vs. USA/Europe) 16 1/0 −0.0789 0.1651 −0.48 0.640 0.05978 −10.79 Prospective timing of the study 16 1/0 −0.1415 0.2901 −0.49 0.633 0.05640 −4.52 Propensity score analysis 16 1/0 −0.0160 0.1778 −0.09 0.930 0.06191 −14.73 a τ2 = between study variance.

Ei 15 Percent −0.0014 0.0029 −0.49 0.633 0.05040 −19.55 Country (Asia vs. USA/Europe) 16 1/0 −0.0789 0.1651 −0.48 0.640 0.05978 −10.79 Prospective timing of the study 16 1/0 −0.1415 0.2901 −0.49 0.633 0.05640 −4.52 Propensity score analysis 16 1/0 −0.0160 0.1778 −0.09 0.930 0.06191 −14.73 a τ2 = between study variance. b The heterogeneity accounted by the covariate included in the random effect meta-regression. Figure 1 Study selection depicted in a PRISMA flowchart. A flowchart of the different phases of the systematic review. Study characteristics Study characteristics are shown in Table 1. The pooled cohort comprised 189 385 patients with AMI. The median age was 64.6 years (range 57.7–68.6 years), 75% was men (range 69.0–81.8%), and median follow-up was 2.7 years (range 0.5–5.2 years). Of ten studies providing information, median LVEF was 53.7% (range 48.9–60.4%). Only four studies used a predefined LVEF cut-off value for inclusion, being >40% in two studies31,33 and ≥ 50% in two.26,29 Eleven studies provided information about history of HF, with a median prevalence of 1.8% (range 0–27.3%). Eight studies provided information about Killip class ≤2 with a median prevalence of 90.6% (range 85.3–100%). On average, 30% of patients had diabetes, 52% hypertension, 6% previous MI, and 45% were smokers when included. Average percentages for concomitant treatments were 94% with aspirin, 69% with statins, and 64% with ARBs/ACEIs. Ten studies were on Asian populations, and six on North American or European populations.

00%). On average, 30% of patients had diabetes, 52% hypertension, 6% previous MI, and 45% were smokers when included. Average percentages for concomitant treatments were 94% with aspirin, 69% with statins, and 64% with ARBs/ACEIs. Ten studies were on Asian populations, and six on North American or European populations. In total, 86.8% (n = 164 408) of the pooled cohort received β-blockers. Information about β-blocker type and dose was provided in two studies,23,33 five studies reported only the type prescribed at hospital discharge,24,26,28,30,32 and for the remaining nine studies no information was provided (see Supplementary material online, Table S4 for further information about study β-blocker types and doses). Follow-up information concerning dose changes, discontinuation or new β-blocker prescriptions was not available for any of the included studies. Two studies22,25 included subpopulations with prior MI eligible for inclusion in the meta-analysis. All studies were cohort by design, of which 14 were retrospective. For 11 studies, confounding was controlled for on multiple clinically relevant variables by propensity score analysis, and multivariable analysis in five studies (see Supplementary material online, Tables S5 and S6). The quality of studies according to the NOS was excellent, with seven studies achieving 9/9 and nine achieving 8/9 stars (see Supplementary material online, Table S7).

cally relevant variables by propensity score analysis, and multivariable analysis in five studies (see Supplementary material online, Tables S5 and S6). The quality of studies according to the NOS was excellent, with seven studies achieving 9/9 and nine achieving 8/9 stars (see Supplementary material online, Table S7). Quantitative data synthesis The pooled estimate from the 16 studies (Figure 2) found that oral β-blockers compared with no oral β-blockers were associated with a 26% reduction in the risk of all-cause mortality [rate ratio (RR) 0.74, 95% CI 0.64–0.85] with moderate between study heterogeneity (I2 = 67.4%). The funnel plot visually showed the possibility of bias or small-study effect (Figure 3) confirmed by the Egger’s test (P = 0.001). The trim and fill simulation method suggested seven studies as missing, and the imputed point estimate was altered (adjusted RR 0.90, 95% CI 0.77–1.04). This indicates a change in magnitude and significance of the pooled effect after correction for publication bias or small-study effect. The cumulative meta-analysis starting with the largest study showed no effect, with increasing effect as the smaller studies were accumulated (Figure 4). Figure 2 Forest plot for meta-analysis of 16 cohort studies comparing β-blocker therapy with no β-blocker in post-AMI patients on all-cause mortality during follow-up. Weights are from random effects analysis. CI, confidence interval; ES, effect size. Figure 3 Funnel plot of the 16 cohort studies included in the meta-analysis. logrr, log rate ratio; se, standard error.

Figure 2 Forest plot for meta-analysis of 16 cohort studies comparing β-blocker therapy with no β-blocker in post-AMI patients on all-cause mortality during follow-up. Weights are from random effects analysis. CI, confidence interval; ES, effect size. Figure 3 Funnel plot of the 16 cohort studies included in the meta-analysis. logrr, log rate ratio; se, standard error. Figure 4 Forest plot for cumulative random effects meta-analysis of 16 cohort studies comparing β-blocker therapy with no β-blocker in post-AMI patients on all-cause mortality during follow-up. The studies are sorted by study size, starting with the largest sized. CI, confidence interval; ES, effect size. According to the pre-specified subgroup analysis (Table 2), the stratified pooled meta-analyses demonstrated no substantial differences in effect of oral β-blockers on all-cause mortality. We extended the analyses with meta-regression, and the results are presented in Table 3. One covariate was associated with mortality risk; the patient related variable ‘mean age of the cohort’ showing decreasing effect of β-blockers on mortality with increasing age of the patients accounting for 31.5% of between study heterogeneity. Of note is that neither subtype of AMI, LVEF, history of HF, length of follow-up, concomitant medical therapy, or ethnicity of the cohort was significantly associated with mortality.

ing decreasing effect of β-blockers on mortality with increasing age of the patients accounting for 31.5% of between study heterogeneity. Of note is that neither subtype of AMI, LVEF, history of HF, length of follow-up, concomitant medical therapy, or ethnicity of the cohort was significantly associated with mortality. The robustness of the primary result obtained from the 16 studies was supported in the influential analysis. When omitting one study at a time from the meta-analysis a stable pooled estimate was shown (see Supplementary material online, Table S8). Discussion This meta-analysis of 16 cohort studies comprising 189 385 patients following AMI of whom only a minority had reduced LVEF and/or clinical HF found that the use of oral β-blockers was associated with a reduction in the risk of all-cause mortality. However, publication bias or small-study effect was found to influence the result with diluted effect seen after correction of small-study effect. Heterogeneity could be explained by the patient related variable mean age of the cohort.

oral β-blockers was associated with a reduction in the risk of all-cause mortality. However, publication bias or small-study effect was found to influence the result with diluted effect seen after correction of small-study effect. Heterogeneity could be explained by the patient related variable mean age of the cohort. β-Blockers have long since been a pharmacotherapy for the management of AMI, but currently their role in the treatment of AMI could be called into question. In the 1980’s, after a series of randomized controlled trials showed improved outcomes and reduced mortality, β-blockers were approved for the treatment of AMI.5,6,35 However, these trials preceded the reperfusion era and enrolled mainly patients with large infarcts and/or HF. Meta-analyses of recent observational studies of the impact of β-blockers after AMI on mortality suggest a beneficial effect.36,37 Misumida et al.36 included six subgroups of patients with preserved LVEF, treated with primary PCI for STEMI (n = 10 857). In this selected population, oral β-blockers compared with no oral β-blockers were associated with a 21% reduction in all-cause mortality. Huang et al.37 included ten studies of whom the majority had STEMI, and eight with early revascularization to find that β-blockers were associated with a reduced risk of all-cause mortality for all subgroups, except those with sample size ≤1000 and those with preserved LVEF. Our study contrasts from Misumida et al. and Huang et al. by including larger studies, and with both STEMI and NSTEMI patients. Therefore, as opposed to the two other studies, this meta-analysis represents a more general post-AMI population with both subtypes of MI where the majority of patients are treated with an oral β-blocker, even in the absence of HF or reduced LVEF.

l. by including larger studies, and with both STEMI and NSTEMI patients. Therefore, as opposed to the two other studies, this meta-analysis represents a more general post-AMI population with both subtypes of MI where the majority of patients are treated with an oral β-blocker, even in the absence of HF or reduced LVEF. Our results based upon the largest studies are further supported by a recent registry study of 90 869 Medicare beneficiaries aged ≥65 years who had prescriptions for ACE-inhibitors, ARBs, β-blockers, or statins and survived AMI ≥180 days.9 Only those patients who were adherent to ACE-inhibitors/ARBs and statins had similar mortality rates to those adherent to all therapies, including β-blockers—suggesting limited additional mortality benefit from β-blockers. The problem with non-adherence to a medication may have been an important confounder. If sicker patients discontinue a medication more commonly than their healthy peers, the benefits of adherence to that medication will be exaggerated. In contrast, the directionality of bias may be the opposite for β-blockers. Those with disease progression and recurrent events may be more adherent to their β-blockers, whereas younger, healthier individuals may be more susceptible to real or perceived β-blocker side effects, and thus less adherent.38

be exaggerated. In contrast, the directionality of bias may be the opposite for β-blockers. Those with disease progression and recurrent events may be more adherent to their β-blockers, whereas younger, healthier individuals may be more susceptible to real or perceived β-blocker side effects, and thus less adherent.38 The use of β-blockers following AMI is based upon historical evidence and nowadays applied to a different treatment and population landscape. Moreover, international advances in the management of AMI have resulted in a decline in deaths,39,40 and it is possible that in this context β-blockers may have lost some of their effectiveness. Limitations The lack of international consensus about the effectiveness of β-blockers following AMI among patients without HF is, in part, a reflection of the lack of contemporary randomized evidence. Consequently, inferences are left to be drawn from observational data, which have inherent bias. Beyond smaller cohort studies, which (as seen in this study) may impact upon the direction of pooled estimates, cohort studies of the effectiveness of pharmacotherapies are weakened by selection and confounding bias as well as missing data. Even though some of the studies included in our meta-analysis used propensity score methods, residual confounding may remain at play. Our investigation is further limited by publication bias, or small-study effect, which may lead to biased estimates which appear precise.13

and confounding bias as well as missing data. Even though some of the studies included in our meta-analysis used propensity score methods, residual confounding may remain at play. Our investigation is further limited by publication bias, or small-study effect, which may lead to biased estimates which appear precise.13 Compromises were made in this meta-analysis regarding the number of patients with HF in each study. A small percentage of patients had a history of HF (albeit between 20% and 30% in two studies), were in Killip class ≥3 and were assumed to have LVEF <40%. Based upon these, in part incomplete data, we have not been able to express a more clear cut-off for the definition of HF than the statement of a majority of patients being without HF and/or LV systolic dysfunction. In the meta-regression model presented in Table 3, neither a history of HF nor mean LVEF was significantly associated with mortality.

omplete data, we have not been able to express a more clear cut-off for the definition of HF than the statement of a majority of patients being without HF and/or LV systolic dysfunction. In the meta-regression model presented in Table 3, neither a history of HF nor mean LVEF was significantly associated with mortality. We did not have information from the included studies about the type, dose, persistence, and new prescription of β-blockers, which may have skewed their impact on mortality. In the study of Puymirat et al.33 neither type of β-blockers at discharge nor dose was related to mortality after adjustment for age and GRACE score. Similar findings were reported by Goldberger et al.41 who could not demonstrate increased survival in patients treated with β-blockers in doses approximating those used in prior randomized trials compared with lower doses. The authors state, however, that an important caveat for their findings is that they do not represent randomized clinical trial results. Conclusions The results from this meta-analysis of nearly 200 000 patients following AMI of whom only a minority had reduced LVEF and/or clinical signs of HF, provides evidence that the association between β-blockers and long-term survival is due to small study effect, and that there might not be a significant reduction in the risk of all-cause mortality when controlling for bias. To be conclusive as for the efficacy of β-blockers on mortality in patients without HF following AMI, randomized controlled trials are a necessary next step.

long-term survival is due to small study effect, and that there might not be a significant reduction in the risk of all-cause mortality when controlling for bias. To be conclusive as for the efficacy of β-blockers on mortality in patients without HF following AMI, randomized controlled trials are a necessary next step. Supplementary Material Supplementary File Click here for additional data file. Acknowledgements Librarian Julie Skattebu at the Hospital of Vestfold, Norway has provided valuable efforts for conducting the systematic search of studies to be included. The authors declare no interests of conflict concerning the present study. Funding This work was supported by grants from the Department for Cardiology, the hospital of Vestfold (grant number 703110, project 19440). Conflict of interest: none declared.

‘Mrs Jones, based on your risk factors for having a heart attack, I recommend that we start you on a statin’. ‘No, thank you, doctor, I’ve read too many scary things about those drugs on the internet. Plus, I worry that some in your profession make these recommendations for reasons of personal financial gain. I also found that online’. Undoubtedly, the majority of cardiologists have had conversations just like this, urging a patient to take a statin, powerful cholesterol-lowering drugs with robust mortality benefit. Part of the reason these oftentimes ‘no brainer’ recommendations are rejected derives from widely disseminated incorrect information which vastly over-states the risks of these drugs. (Of course, like anything in life, statin use is not entirely risk-free; their application should always entail a thoughtful analysis of risks vs. benefits.) Most patients do not recognize that the benefits of statin use are invisible (‘I didn’t have a heart attack or stroke this past year’.), whereas the small and typically reversible risks (e.g. muscle pain) are readily apparent. Many patients who would benefit from statin use do not take them. Cardiovascular disease is the no. 1 killer of both men and women around the world. Robust scientific advances, published in the pages of our journals, have fostered significant improvements that benefit individuals and society. Yet, cardiovascular disease continues to transform itself, emerging in new forms, such as heart failure. The struggle has shifted to new battlefields.

women around the world. Robust scientific advances, published in the pages of our journals, have fostered significant improvements that benefit individuals and society. Yet, cardiovascular disease continues to transform itself, emerging in new forms, such as heart failure. The struggle has shifted to new battlefields. These successes derive from an armamentarium of powerful tools—medicines and devices—and awareness of lifestyle-related hazards, such as high blood pressure, high cholesterol, and smoking. Sadly, however, we do not take full advantage of the tools at our disposal. One significant cause of suboptimal utilization of our prodigious tool chest is medical misinformation hyped through the internet, television, chat rooms, and social media. In many instances, celebrities, activists, and politicians convey false information; not uncommonly, authors with purely venal motives participate.

One significant cause of suboptimal utilization of our prodigious tool chest is medical misinformation hyped through the internet, television, chat rooms, and social media. In many instances, celebrities, activists, and politicians convey false information; not uncommonly, authors with purely venal motives participate. We can point to numerous other examples, including the entirely unfounded concerns regarding vaccinations. The notion MMR (measles, mumps, rubella) vaccination causes autism was based on a single, flawed study, long since refuted, and its publication retracted. Seventeen much larger and properly controlled studies have proven otherwise. Nevertheless, the internet shouts unfounded warnings. Once again, celebrities, actors, activists, and politicians with no specific knowledge or training use their fame to promote a message that causes serious harm. Individuals who are neither physicians nor scientists, but often with a specific agenda, have outsized influence over our lives. They dispute scientific evidence without ever having studied it.1 Recognizing that it is impossible to prove ‘never’, scientists appropriately couch their statements in statistical terms, which may come across to the public as equivocation. The nuanced voices of scientists often do not resonate with the public as much as the strident alarms sounded by people of fame, speaking in absolute terms.

that it is impossible to prove ‘never’, scientists appropriately couch their statements in statistical terms, which may come across to the public as equivocation. The nuanced voices of scientists often do not resonate with the public as much as the strident alarms sounded by people of fame, speaking in absolute terms. Further, scientists are appropriately sceptical, as any individual scientist or study can be wrong. Yet, science ultimately self-corrects. When a scientist gets it wrong, as happens, people sometimes vilify the entire, self-correcting scientific enterprise. We trust aeronautical science when we board an aeroplane; we trust the science buried within our cell phones; we trust mechanical engineering science when we cross a bridge; yet, many are uniquely sceptical of biological science. Sadly, we cannot exclude that some in the professions of science and medicine act based on motives driven by financial considerations; incomplete declarations of potential conflict of interest persist.2 Recent examples of dramatic price hikes for important medications have reinforced this notion. Indeed, many physicians have had conversations with patients who believe that our recommendations stem, at least in part, from the prospect of personal financial gain. We, the editors-in-chief of the major cardiovascular scientific journals around the globe, sound the alarm that human lives are at stake. Pointing to the two examples elaborated above, people who decline to use a statin when recommended by their doctor, or parents who withhold vaccines from their children, put lives in harm’s way.

ors-in-chief of the major cardiovascular scientific journals around the globe, sound the alarm that human lives are at stake. Pointing to the two examples elaborated above, people who decline to use a statin when recommended by their doctor, or parents who withhold vaccines from their children, put lives in harm’s way. The media must do a better job. It is unacceptable to posit false equivalents in these discussions, often done to foster debate and controversy. It is easy to find a rogue voice but inappropriate to suggest that voice carries the same weight as that emerging from mainstream science. (We can easily point to examples outside the medical domain, as well, such as climate change, evolution, nutraceuticals, and GMO foods where false equivalents are frequently posited.) Furthermore, recent evidence suggests that misinformation travels faster through social networks than truth.3 We must work to enhance science literacy in our world; one place to start is by doing a better job of teaching the scientific method in our schools so that the lay public is aware that science is accomplished in fits and starts, but in the end, gets it right.

ormation travels faster through social networks than truth.3 We must work to enhance science literacy in our world; one place to start is by doing a better job of teaching the scientific method in our schools so that the lay public is aware that science is accomplished in fits and starts, but in the end, gets it right. Purveyors of social media must be responsible for the content they disseminate. It is no longer acceptable to hide behind the cloak of ‘platform’. We, as editors, are charged with evaluating the validity of the science presented to us for possible publication, and we work hard to fulfil this heady responsibility. Recognizing that lives are at stake, we reach out to thought-leading experts to evaluate the veracity of each report we receive. Here, we challenge social media to do the same, to leverage the ready availability of science-conversant expertise before disseminating content that may not be reliable. Without exaggeration, significant harm, to society and individuals, derives from the wanton spread of medical misinformation. It is high time that this stop, and we lay at the feet of the purveyors of internet and social media content the responsibility to fix this. Conflict of interest: P.G. Camici is consultant for Servier. R.S. Rosenson reports research grants to his institutions from Akcea, Amgen, Astra Zeneca, Medicines Company and Regeneron. R.S. Rosenson reports speaking engagements at Amgen and Kowa, research consulting for Akcea and Regeneron, royalties from UpToDate, Inc. and stock holdings in MediMergent. All other authors have nothing to disclose.

ts research grants to his institutions from Akcea, Amgen, Astra Zeneca, Medicines Company and Regeneron. R.S. Rosenson reports speaking engagements at Amgen and Kowa, research consulting for Akcea and Regeneron, royalties from UpToDate, Inc. and stock holdings in MediMergent. All other authors have nothing to disclose. Footnotes A complete list of all journals publishing this article, along with links to the individual articles, can be found online at https://www.ahajournals.org/circ/medical-misinformation