Browse the corpus

Introduction Body-mass index (BMI) is a reasonably good measure of general adiposity,1 and raised BMI is an established risk factor for several causes of death, including ischaemic heart disease,2 stroke,3 and cancers of the large intestine, kidney, endometrium, and postmenopausal breast.4,5 In many populations, the average BMI has been rising by a few percent per decade,6 fuelling concern about the effects of increased adiposity on health. Some uncertainties persist, however, about the relation between BMI and mortality, including whether some of the reported positive or inverse associations have been distorted by weight loss because of pre-existing disease (reverse causality) or by inadequate control for the effects of smoking; whether the shape and strength of associations with specific diseases differ between smokers and non-smokers; how the relative and absolute risks for BMI compare with, and also combine with, those for smoking; whether the relative risks differ much by sex or age (and whether any substantial association continues into old age7); how the absolute excess risks for vascular disease compare with those for neoplastic or respiratory disease; and the extent to which some less common causes of death, such as kidney8 or liver9 disease, are associated with BMI.

differ much by sex or age (and whether any substantial association continues into old age7); how the absolute excess risks for vascular disease compare with those for neoplastic or respiratory disease; and the extent to which some less common causes of death, such as kidney8 or liver9 disease, are associated with BMI. Some of these uncertainties can best be addressed by large prospective observational studies—or by large collaborative analyses of individual data from such studies, as in this report—that follow up generally healthy adults for many years and identify large numbers of deaths from specific causes. In the Prospective Studies Collaboration (PSC), the investigators of 61 prospective studies have shared individual data for a million adults.10,11 The collaboration was established chiefly to assess the relevance of blood pressure10 and blood cholesterol11 to cause-specific mortality, but for 57 of the studies information was also available for BMI (although generally not for waist circumference). This PSC report examines the relevance of BMI to cause-specific mortality 5 or more years after recruitment into these studies.

ance of blood pressure10 and blood cholesterol11 to cause-specific mortality, but for 57 of the studies information was also available for BMI (although generally not for waist circumference). This PSC report examines the relevance of BMI to cause-specific mortality 5 or more years after recruitment into these studies. Methods Data collection Previous PSC reports10,11 have described the methods of study selection, data collection, and statistical analysis, and similar methods were used in this report. BMI was calculated as weight in kg divided by the square of height in m. In three studies of US health professionals, height and weight were self-reported by participants. Following WHO convention, BMI of 30 kg/m2 or more is termed obese. Individuals with missing data for age, sex, or BMI were excluded, as were those with BMI less than 15 kg/m2 (188 excluded) or 50 kg/m2 or more (297), those with a baseline history of heart disease (54 347) or stroke (4349), and those with no follow-up in the age range 35–89 years (25 949), leaving 894 576 participants. Information about blood pressure (882 032) and total cholesterol (814 109) was available for most participants, as was information about tobacco smoking (849 723) and diabetes mellitus (698 255). Information was, however, available for relatively few of the participants about alcohol drinking (297 584), HDL cholesterol (114 939), and LDL cholesterol (42 937). Of current drinkers, 78% (155 900) had information about grams of alcohol consumed per day, and of current cigarette smokers, 57% (166 724) had information about daily number of cigarettes smoked.

latively few of the participants about alcohol drinking (297 584), HDL cholesterol (114 939), and LDL cholesterol (42 937). Of current drinkers, 78% (155 900) had information about grams of alcohol consumed per day, and of current cigarette smokers, 57% (166 724) had information about daily number of cigarettes smoked. Generally, the underlying cause of death was obtained from the death certificate (information about contributory causes was not available), but in many studies confirmation was then sought from other sources, such as medical records and autopsy findings. The cause of death was coded to 3 digits using any of International Classifications of Diseases (ICD) 6–10.

as obtained from the death certificate (information about contributory causes was not available), but in many studies confirmation was then sought from other sources, such as medical records and autopsy findings. The cause of death was coded to 3 digits using any of International Classifications of Diseases (ICD) 6–10. Statistical analysis Cross-sectional associations between BMI and risk factors were estimated by multiple linear regression or logistic regression, with adjustment for study, baseline age (in 10-year groups), and baseline smoking (three groups: current cigarette smoker [32%]; never smoked any type of tobacco regularly [35%]; and other smoker, ex-smoker of any type of tobacco, or unknown [8%, 19%, and 5%, respectively]). Any individuals with missing values of BMI or the particular risk factor were excluded. Associations between baseline BMI and mortality were estimated by Cox regression, with stratification for study, sex, age at risk (in 5-year groups), and baseline smoking (as above), but not for blood pressure, blood lipids, or diabetes (since these are mechanisms by which BMI affects vascular mortality). The resulting relative risks were not corrected for the regression dilution bias,12 since one BMI measurement is highly correlated with the long-term usual BMI (self-correlation 0·90 between baseline BMI and a re-measurement of BMI some 6 years later; webappendix p 11). In categorical analyses, the boundaries of BMI categories were always multiples of 2·5 kg/m2, and the boundaries used in particular analyses are indicated by tick marks in the figures. Values exactly on a boundary went above it. In continuous analyses, log risk was regressed on BMI as a continuous variable within the range 15–25 kg/m2 (termed the lower range), 25–50 kg/m2 (upper range), or 15–50 kg/m2 (full range), yielding in each range the hazard ratio per 5 kg/m2 higher BMI (HR). To limit effects of pre-existing disease on baseline BMI, the main analyses exclude all person-years and deaths in the first 5 years of follow-up.

the range 15–25 kg/m2 (termed the lower range), 25–50 kg/m2 (upper range), or 15–50 kg/m2 (full range), yielding in each range the hazard ratio per 5 kg/m2 higher BMI (HR). To limit effects of pre-existing disease on baseline BMI, the main analyses exclude all person-years and deaths in the first 5 years of follow-up. Relative risks for different BMI categories are presented as floating absolute risks by multiplying all of them by a common constant to make their inverse-variance-weighted average match the uniformly age-standardised death rate per 1000 person-years at ages 35–79 years (ie, the simple mean of the nine age-specific rates at ages 35–39 years to 75–79 years) either in the PSC population, or in the European Union (EU) population in 2000 (ie, the combined population of 15 western European countries).13 Multiplication of all the relative risks by this common constant means that the floating absolute risks (and the SEs of the log risks) do not depend at all on the choice of an arbitrary reference group. Hence, an appropriate SE and CI can be assigned to the log of the floating absolute risk in each BMI category.14 This SE—described more simply as the SE of the log rate (Julian Peto, London School of Hygiene and Tropical Medicine, London, UK; personal communication)—does not depend on the common constant that was chosen, and is roughly equivalent to the coefficient of variation of the risk in that one group.

each BMI category.14 This SE—described more simply as the SE of the log rate (Julian Peto, London School of Hygiene and Tropical Medicine, London, UK; personal communication)—does not depend on the common constant that was chosen, and is roughly equivalent to the coefficient of variation of the risk in that one group. The webappendix provides further information, including details of the collaborating studies, endpoint definitions, BMI re-measurement results, and findings for specific causes of death. Role of the funding sources The funders of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. GW, PS, and SL had full access to the data in the study, and the writing committee had final responsibility for the decision to submit for publication. Results In the 57 studies with information for BMI, individual records with information about BMI were available for 894 576 adults. 92% of these participants were in Europe, Israel, the USA, or Australia; the remaining 8% (accounting for just 3% of the deaths) were in Japan. 85% (763 274) of participants were recruited during the 1970s or 1980s. Across all studies, median year of recruitment was 1979 (IQR 1975–85), mean recruitment age was 46 (SD 11) years, and 61% (n=541 452) were male. Mean BMI was 24·8 (SD 3·8) kg/m2, but it was lower in the European and Israeli studies (24·7 [3·6] kg/m2) than in the US and Australian studies (25·6 [4·3] kg/m2), and lower still in the Japanese studies (22·8 [2·9] kg/m2).

as 1979 (IQR 1975–85), mean recruitment age was 46 (SD 11) years, and 61% (n=541 452) were male. Mean BMI was 24·8 (SD 3·8) kg/m2, but it was lower in the European and Israeli studies (24·7 [3·6] kg/m2) than in the US and Australian studies (25·6 [4·3] kg/m2), and lower still in the Japanese studies (22·8 [2·9] kg/m2). For both sexes, mean BMI at baseline was greatest between 50 years and 69 years of age (webappendix p 11). For people with a re-measurement of BMI more than 5 years after baseline, the rate of change between the two measurements showed an increase in BMI in early adult life and middle age (particularly in women), a levelling off in late middle age, and a slight decrease in old age (webappendix p 11). The greatest rate of increase (about 1·5 kg/m2 per decade) was in men younger than 40 years and women younger than 50 years. Apart from these age-related trends, the changes on re-measurement were slight, and were consistent with regression to the mean having had only a minor effect that negated any even more minor tendency for usual BMI values to disperse (webappendix p 12).

was in men younger than 40 years and women younger than 50 years. Apart from these age-related trends, the changes on re-measurement were slight, and were consistent with regression to the mean having had only a minor effect that negated any even more minor tendency for usual BMI values to disperse (webappendix p 12). At baseline, several vascular risk factors were strongly related to BMI (figure 1). Throughout the full range (15–50 kg/m2), BMI was associated positively and nearly linearly with systolic and diastolic blood pressure (SBP, DBP; figure 1A). On average across all ages (15–89 years), every 5 kg/m2 higher BMI was associated with at least 5 mm Hg higher SBP (male 5·8 mm Hg, female 5·2 mm Hg) and about 4 mm Hg higher DBP (male 4·9 mm Hg, female 3·3 mm Hg). In the range up to 30 kg/m2, BMI was associated inversely with HDL cholesterol (male 0·16 mmol/L, female 0·14 mmol/L lower per 5 kg/m2), positively with non-HDL cholesterol (male 0·50 mmol/L, female 0·39 mmol/L higher per 5 kg/m2), and therefore strongly positively with the ratio of non-HDL to HDL cholesterol (male 0·85, female 0·54 higher ratio per 5 kg/m2, from analyses of individual ratios: figure 1). These relations with blood pressure and with the non-HDL/HDL cholesterol ratio were about a third weaker in the 2% of participants aged 70–89 years (data not shown). Above 30 kg/m2, BMI was only weakly associated with either cholesterol fraction (figure 1B). Obesity was strongly associated with diabetes (figures 1C and 1D), with the sex-specific prevalences rising more than five-fold over the full BMI range.

third weaker in the 2% of participants aged 70–89 years (data not shown). Above 30 kg/m2, BMI was only weakly associated with either cholesterol fraction (figure 1B). Obesity was strongly associated with diabetes (figures 1C and 1D), with the sex-specific prevalences rising more than five-fold over the full BMI range. Smoking and drinking may affect BMI, and, after adjustment for age and study, the mean BMI was slightly lower in current smokers than in never-smokers (male 0·3 kg/m2, female 0·9 kg/m2 lower), and in regular alcohol users than in others (male 0·1 kg/m2, female 1·2 kg/m2 lower). Hence, in both sexes the prevalences of smoking and (especially in females) drinking tended to be high in those with low BMI (figures 1C and 1D). In regular smokers or drinkers with relevant data (webappendix p 13), daily cigarette or alcohol consumption was not strongly dependent on BMI.

g/m2, female 1·2 kg/m2 lower). Hence, in both sexes the prevalences of smoking and (especially in females) drinking tended to be high in those with low BMI (figures 1C and 1D). In regular smokers or drinkers with relevant data (webappendix p 13), daily cigarette or alcohol consumption was not strongly dependent on BMI. Of the 894 576 participants with baseline measurements of BMI, 15 996 died in the first 5 years of follow-up, and 852 824 were still alive and under observation at the start of year 5. During 6·5 million person-years of subsequent follow-up (mean 8 [SD 6] years per person), 72 749 deaths were identified. Most (90%) of this additional follow-up and more than half (58%) of these deaths were at ages 35–69 years; 9% of the follow-up and 29% of the deaths were at ages 70–79 years; and 2% and 12%, respectively, were at ages 80–89 years (table 1). Among the 72 749 who died, 54 703 (75%) were males, median year of birth was 1918 (IQR 1910–25), and, of the 62 055 deaths for which the exact year of death is available, 60 153 (97%) occurred between 1970 and 1999 (median 1986). 6197 (9%) of the deaths were from an unknown cause (6% at ages 35–69 years, 9% at 70–79 years, and 16% at 80–89 years). For the remainder, mean age at death was 67 (SD 10) years.

910–25), and, of the 62 055 deaths for which the exact year of death is available, 60 153 (97%) occurred between 1970 and 1999 (median 1986). 6197 (9%) of the deaths were from an unknown cause (6% at ages 35–69 years, 9% at 70–79 years, and 16% at 80–89 years). For the remainder, mean age at death was 67 (SD 10) years. In both sexes (and at all ages: webappendix p 1), all-cause mortality was lowest at about 22·5–25 kg/m2 (figure 2). Above this minimum, mortality was on average about 30% higher for every 5 kg/m2 higher BMI. Although the proportional increase was somewhat greater at younger ages (35–59 years), each 5 kg/m2 higher BMI was still associated with almost 30% higher mortality at 70–79 years of age (table 1). In the lower BMI range (15–25 kg/m2), the inverse association of mortality with BMI (HR per 5 kg/m2 higher BMI 0·79 [95% CI 0·77–0·82]) became less extreme either when the analysis was restricted to lifelong non-smokers (0·87 [0·81–0·94]) or when a further 10 years of follow-up were excluded (0·85 [0·81–0·91]; webappendix p 14).

er BMI range (15–25 kg/m2), the inverse association of mortality with BMI (HR per 5 kg/m2 higher BMI 0·79 [95% CI 0·77–0·82]) became less extreme either when the analysis was restricted to lifelong non-smokers (0·87 [0·81–0·94]) or when a further 10 years of follow-up were excluded (0·85 [0·81–0·91]; webappendix p 14). Ischaemic heart disease accounted for more than a quarter of all deaths of known cause. BMI and mortality from ischaemic heart disease were associated strongly, positively, and roughly log-linearly throughout the BMI range from 20 to 40 kg/m2, and perhaps at even greater BMI (figure 3). In the upper BMI range (25–50 kg/m2), each 5 kg/m2 higher BMI was associated with about 40% higher ischaemic heart disease mortality (figure 4, table 2). The association was a little stronger in middle than in old age, but was still definite even at ages 80–89 years (HR 1·30 [1·17–1·45]; figure 4A). In the European and Israeli studies the association was about as strong as it was in the US and Australian studies (HRs 1·38 [1·33–1·44] and 1·40 [1·32–1·50], respectively, for deaths at ages 35–89 years; only 44 deaths from this disease occurred in the Japanese studies), and there was no significant heterogeneity (p=0·13) across the HRs for the 16 larger studies (with >200 deaths from ischaemic heart disease) and the aggregated smaller studies (webappendix p 5).

1·32–1·50], respectively, for deaths at ages 35–89 years; only 44 deaths from this disease occurred in the Japanese studies), and there was no significant heterogeneity (p=0·13) across the HRs for the 16 larger studies (with >200 deaths from ischaemic heart disease) and the aggregated smaller studies (webappendix p 5). Stroke accounted for a third as many deaths as ischaemic heart disease did. In the upper BMI range (25–50 kg/m2) each 5 kg/m2 higher BMI was, as for ischaemic heart disease, associated with about 40% higher mortality (table 2), largely irrespective of follow-up period (after the first 5 years), smoking status, or stroke subtype (figure 4B). As with ischaemic heart disease, there was no significant heterogeneity (p=0·64) across the HRs from the eight larger studies (with >100 stroke deaths) and the aggregated smaller studies (webappendix p 5). By contrast with ischaemic heart disease, however, the association of BMI with stroke was much stronger in middle than in old age (figure 4; for both diseases, allowance for the older age at death of women eliminates the apparent relevance of sex). Furthermore, in the lower BMI range (15–25 kg/m2) there was no evidence of a positive association between BMI and stroke (figure 3, table 2). Nor was there any evidence of a positive association in this lower range after participants who had ever smoked were excluded (HR 0·98 [0·78–1·23]), or after the analysis was restricted to haemorrhagic (0·76 [0·58–1·00]) or to ischaemic (0·87 [0·68–1·10]) stroke; the number with confirmation of subtype by imaging is, however, unknown.

y evidence of a positive association in this lower range after participants who had ever smoked were excluded (HR 0·98 [0·78–1·23]), or after the analysis was restricted to haemorrhagic (0·76 [0·58–1·00]) or to ischaemic (0·87 [0·68–1·10]) stroke; the number with confirmation of subtype by imaging is, however, unknown. For the aggregate of all other vascular causes of death (table 2), the association with BMI was similar to that for stroke. In the lower BMI range (15–25 kg/m2) there was, if anything, a slightly inverse association, but in the upper range each 5 kg/m2 higher BMI was again associated with about 40% higher mortality. Among particular other vascular causes, the associations in the upper BMI range were particularly strong for mortality attributed to heart failure (HR 1·86 [1·55–2·23]) and to hypertensive disease (2·03 [1·75–2·36]). In the upper range (25–50 kg/m2), BMI was associated strongly and positively with mortality attributed to diabetes, to non-neoplastic kidney disease, and to non-neoplastic liver disease (table 2), which was mainly cirrhosis (HR 1·79 [1·54–2·08]).

For the aggregate of all other vascular causes of death (table 2), the association with BMI was similar to that for stroke. In the lower BMI range (15–25 kg/m2) there was, if anything, a slightly inverse association, but in the upper range each 5 kg/m2 higher BMI was again associated with about 40% higher mortality. Among particular other vascular causes, the associations in the upper BMI range were particularly strong for mortality attributed to heart failure (HR 1·86 [1·55–2·23]) and to hypertensive disease (2·03 [1·75–2·36]). In the upper range (25–50 kg/m2), BMI was associated strongly and positively with mortality attributed to diabetes, to non-neoplastic kidney disease, and to non-neoplastic liver disease (table 2), which was mainly cirrhosis (HR 1·79 [1·54–2·08]). In the range 25–50 kg/m2, neoplastic disease accounted for nearly two-thirds as many deaths as did vascular disease, but the association with BMI was much weaker: only 10% higher neoplastic mortality, compared with 40% higher vascular mortality, for each 5 kg/m2 higher BMI (neoplastic HR 1·10 [1·06–1·15]). Even across the full BMI range (15–50 kg/m2) the 95% CIs for site-specific cancers were wide (webappendix p 15), but nonetheless there were positive associations for several sites, including the liver (HR 1·47 [1·26–1.71]), kidney (1·23 [1·06–1·43]), breast (1·15 [1·02–1·31] for deaths at ages 60–89 years and, identically, for deaths at 35–59 years), endometrium (1·38 [1·08–1·77]), prostate (1·13 [1·02–1·24]), and large intestine (1·20 [1·12–1.28]; male 1·29 [1·19–1·40], female 1·05 [0·94–1·18]).

including the liver (HR 1·47 [1·26–1.71]), kidney (1·23 [1·06–1·43]), breast (1·15 [1·02–1·31] for deaths at ages 60–89 years and, identically, for deaths at 35–59 years), endometrium (1·38 [1·08–1·77]), prostate (1·13 [1·02–1·24]), and large intestine (1·20 [1·12–1.28]; male 1·29 [1·19–1·40], female 1·05 [0·94–1·18]). In the lower range (15–25 kg/m2), BMI was associated inversely with mortality from cancer as a whole, mainly because of steep inverse associations with the cancers most strongly related to smoking (table 2; upper aerodigestive cancer includes oesophagus cancer, which had HR 0·52 [95% CI 0·38–0·72], and cancer of the mouth, pharynx, and larynx). The inverse association of BMI with lung and upper aerodigestive cancer (combined) weakened with increasing duration of follow-up but was still definite during years 10–14 (0·65 [0·53–0·79]) and years 15 and more (0·75 [0·62–0·91]). Even among lifelong non-smokers, there was a definite inverse association for upper aerodigestive cancer (0·35 [0·16–0·74]), although not for lung cancer (0·90 [0·48–1·68]).

kened with increasing duration of follow-up but was still definite during years 10–14 (0·65 [0·53–0·79]) and years 15 and more (0·75 [0·62–0·91]). Even among lifelong non-smokers, there was a definite inverse association for upper aerodigestive cancer (0·35 [0·16–0·74]), although not for lung cancer (0·90 [0·48–1·68]). Respiratory disease accounted for an eighth as many deaths as did vascular disease (table 2). In the lower range, BMI was strongly and inversely associated with mortality from each main type of respiratory disease, of which the most common was chronic obstructive pulmonary disease (COPD). After exclusion of the first 5 years of follow-up, COPD mortality in this BMI range was four times higher for 5 kg/m2 lower BMI (HR 0·26 [95% CI 0·22–0·30]). In the first 5 years (excluded from all the main analyses), the inverse association at low BMI was even greater (0·11 [0·08–0·16]), since COPD can cause weight loss (ie, there is reverse causality). Exclusion of an additional 10 years further attenuated the inverse association in the lower BMI range, but a strong inverse association remained more than 15 years after the baseline BMI measurement (COPD HR 0·31 [0·24–0·40]). Relatively few deaths were attributed to tuberculosis, which can cause chronic wasting and was (in the lower BMI range) strongly inversely associated with BMI even after exclusion of the first 10 years of follow-up (HR 0·09 [95% CI 0·04–0·19]). In the upper BMI range (25–50 kg/m2), overall respiratory mortality was about 20% higher for each 5 kg/m2 higher BMI (table 2).

s, which can cause chronic wasting and was (in the lower BMI range) strongly inversely associated with BMI even after exclusion of the first 10 years of follow-up (HR 0·09 [95% CI 0·04–0·19]). In the upper BMI range (25–50 kg/m2), overall respiratory mortality was about 20% higher for each 5 kg/m2 higher BMI (table 2). The remaining mortality was divided into three categories (other specified diseases, external causes [mainly injury], and unknown causes), each of which had a U-shaped association with baseline BMI and a minimum at about 22·5–25 kg/m2 (table 2). For each category, 5 kg/m2 higher BMI in the 25–50 kg/m2 range was associated with about 20% higher mortality. The other specified diseases were a broad range of disorders and were not dominated by any particular causes of death. External causes of death could not be usefully analysed, since little information was available about the specific circumstances of these deaths. Absolute excess mortality depends not only on relative risks but also on absolute mortality rates. To indicate the absolute excess risks at different BMI levels, figure 5 is plotted on an additive scale and applies the PSC relative risks at ages 35–79 years. The absolute difference in mortality between 35–50 kg/m2 and 22·5–25 kg/m2 was five times as great for vascular as for neoplastic disease in males, and twice as great for vascular as for neoplastic disease in females. Below 22·5–25 kg/m2, there was a pronounced excess of lung cancer, upper aerodigestive cancer, and respiratory disease, particularly in males (figure 5).

kg/m2 and 22·5–25 kg/m2 was five times as great for vascular as for neoplastic disease in males, and twice as great for vascular as for neoplastic disease in females. Below 22·5–25 kg/m2, there was a pronounced excess of lung cancer, upper aerodigestive cancer, and respiratory disease, particularly in males (figure 5). Both in current cigarette smokers and in lifelong non-smokers, overall mortality was lowest at about 22·5–25 kg/m2, but the excess mortality below this range was both relatively and absolutely much greater in smokers (figure 6, again plotted on an additive scale). In both smoking groups, the excess at 25–27·5 kg/m2 was slight, and although the excess at 30–35 kg/m2 was substantial, it was still much less than the excess attributable in this study to cigarette smoking itself (as shown in figure 6 by the vertical separation of the curves; the excess attributable to persistent cigarette smoking throughout adult life would be even greater than this). Throughout the range 25–50 kg/m2, the effects of BMI and smoking seemed to be roughly additive, rather than multiplicative, both for vascular mortality (webappendix p 8) and for all-cause mortality (figure 6).

he excess attributable to persistent cigarette smoking throughout adult life would be even greater than this). Throughout the range 25–50 kg/m2, the effects of BMI and smoking seemed to be roughly additive, rather than multiplicative, both for vascular mortality (webappendix p 8) and for all-cause mortality (figure 6). Discussion In this collaborative analysis of data from almost 900 000 adults in 57 prospective studies, overall mortality was lowest at about 22·5–25 kg/m2 in both sexes and at all ages, after exclusion of early follow-up and adjustment for smoking status. Above this range, each 5 kg/m2 higher BMI was associated with about 30% higher all-cause mortality (40% for vascular; 60–120% for diabetic, renal, and hepatic; 10% for neoplastic; and 20% for respiratory and for all other mortality) and no specific cause of death was inversely associated with BMI. Below 22·5–25 kg/m2, the overall inverse association with BMI was predominantly due to strong inverse associations for smoking-related respiratory disease (including cancer), and the only clearly positive association was for ischaemic heart disease.

ity) and no specific cause of death was inversely associated with BMI. Below 22·5–25 kg/m2, the overall inverse association with BMI was predominantly due to strong inverse associations for smoking-related respiratory disease (including cancer), and the only clearly positive association was for ischaemic heart disease. In laboratory studies, BMI is moderately strongly correlated (30–50%) with fat-free mass, but it is much more strongly correlated (60–90%) with fat mass.15 BMI is also strongly correlated (80–85%) with measured waist circumference;16,17 in the EPIC prospective study of 360 000 adults in Europe, for example, the two variables have about an 85% correlation, so each has a similar association with mortality.16 (In EPIC, waist-to-hip ratio was not quite as strongly related either to BMI or to mortality.) In such populations, either measurement can thus be used to help assess the causal relevance of obesity to mortality, and each could well add some predictive information to the other. Neither, however, directly measures visceral fat. Although BMI and waist circumference are not directly causal, both are closely correlated in such populations with aspects of adiposity that directly affect blood pressure, lipoprotein particles, and diabetes (figure 1). Effective interventions for weight loss lower blood pressure, favourably affect lipoprotein particles, and increase insulin sensitivity,18 and drugs that substantially lower blood pressure19 or LDL particle numbers20 reduce vascular disease. At least some of the major adverse effects of obesity are, therefore, reversible.

ective interventions for weight loss lower blood pressure, favourably affect lipoprotein particles, and increase insulin sensitivity,18 and drugs that substantially lower blood pressure19 or LDL particle numbers20 reduce vascular disease. At least some of the major adverse effects of obesity are, therefore, reversible. For ischaemic heart disease, the magnitude of the positive association with BMI in this study can be largely accounted for by blood pressure, lipoprotein particles, and diabetes. The associations of baseline BMI with baseline measurements of SBP and of the non-HDL/HDL cholesterol ratio (figure 1) can be taken as the associations of BMI with the usual levels of these variables over the past and next few years, so they would predict at least a doubling of mortality from ischaemic heart disease between 20 kg/m2 and 30 kg/m2 (if the combined effects of SBP and the ratio of cholesterol fractions were approximately additive10,11), which is what was observed. (Merely adjusting regression analyses for single measurements of blood pressure and total cholesterol would underestimate the mediating effects of blood pressure and, especially, of lipoprotein particles.21,22) Above 30 kg/m2, further increases in BMI have little further effect on the ratio of cholesterol fractions (figure 1B), but could be associated with other adverse changes in lipoprotein particles that cannot be inferred from cholesterol fractions (eg, an increase in the number of small dense LDL particles). Diabetes becomes particularly important at BMI greater than 30 kg/m2 (figures 1C and 1D). Other hypothesised intermediate factors (eg, fibrinogen, C-reactive protein, obstructive sleep apnoea) were not assessed.

cannot be inferred from cholesterol fractions (eg, an increase in the number of small dense LDL particles). Diabetes becomes particularly important at BMI greater than 30 kg/m2 (figures 1C and 1D). Other hypothesised intermediate factors (eg, fibrinogen, C-reactive protein, obstructive sleep apnoea) were not assessed. Confounding by diet, physical activity, or socioeconomic status could have somewhat affected the ischaemic heart disease results. The cardioprotective effects of physical activity might not be due solely to reduced adiposity,23 so variation in physical activity could have caused the independent effects of adiposity to be somewhat overestimated. Confounding by socioeconomic status could have caused the independent effects to be either overestimated or underestimated. In the three prospective studies of US health professionals, however, there would have been relatively little socioeconomic confounding, yet for all-cause mortality in the upper BMI range (there were too few deaths to subdivide by cause), the association seemed to be broadly similar across these three studies to that in the PSC as a whole (webappendix p 14). The weakening of the association between BMI and mortality from ischaemic heart disease above age 70 years is probably a result of the weaker associations at older ages of blood pressure and cholesterol with risk,10,11 and the slightly weaker associations of BMI with these intermediate variables. (At older ages, BMI might depend increasingly on muscle loss.24)

MI and mortality from ischaemic heart disease above age 70 years is probably a result of the weaker associations at older ages of blood pressure and cholesterol with risk,10,11 and the slightly weaker associations of BMI with these intermediate variables. (At older ages, BMI might depend increasingly on muscle loss.24) For stroke, the findings in the upper and lower BMI ranges were quite different from each other. In the upper range, BMI was associated positively with ischaemic, haemorrhagic, and total stroke, and each of these associations can be largely accounted for by the effects of BMI on blood pressure. In the lower BMI range, however, there was no evidence of a positive association for ischaemic, haemorrhagic, or total stroke, despite the strong positive association between BMI and blood pressure. (For a specific blood pressure, therefore, BMI in this lower range would actually be inversely related to stroke.) These findings for stroke in the lower BMI range were not materially affected by exclusion of participants who had ever smoked (by contrast with the findings reported from a large Chinese prospective study25). The evidence from previous large studies of BMI and stroke subtype is not as consistent as might be expected,1,3,26–28 but generally suggests that the association of BMI with stroke risk is strongly positive at BMI greater than 25 kg/m2 for both ischaemic and haemorrhagic stroke, and, less definitely, that at BMI less than 25 kg/m2 it is still positive for ischaemic but not for haemorrhagic stroke. In the lower BMI range, however, the PSC found no evidence of an association for ischaemic stroke (although the possibility of a weak positive association is not excluded), and found only slight evidence of an inverse association for haemorrhagic stroke. These findings for stroke in the lower BMI range are not fully explained.

ower BMI range, however, the PSC found no evidence of an association for ischaemic stroke (although the possibility of a weak positive association is not excluded), and found only slight evidence of an inverse association for haemorrhagic stroke. These findings for stroke in the lower BMI range are not fully explained. For kidney and liver disease,8,9 the positive associations with BMI could have resulted mainly from the effects of adiposity on blood pressure, diabetes, and blood lipids. Central adiposity can cause non-alcoholic fatty liver disease, which could predispose to cirrhosis or hepatocellular carcinoma (the commonest type of liver cancer).9 The positive associations of BMI with cirrhosis and liver cancer are unlikely to have been due to confounding by alcohol, since drinking was not strongly related to BMI in males and was inversely related to it in females.

h could predispose to cirrhosis or hepatocellular carcinoma (the commonest type of liver cancer).9 The positive associations of BMI with cirrhosis and liver cancer are unlikely to have been due to confounding by alcohol, since drinking was not strongly related to BMI in males and was inversely related to it in females. For cancer, the evidence of several positive associations complements that from other million-person prospective studies (eg, the Cancer Prevention Study-II4 and the Million Women Study5). Possible mechanisms by which obesity could cause cancer at particular sites have been summarised elsewhere.29 The overall inverse association with cancer mortality in the lower BMI range (15–25 kg/m2) was mainly due to inverse associations with cancers of the lung and oesophagus. Most of the oesophageal cancer deaths occurred before the 1990s, so most are likely to have been squamous cell carcinomas,30 which are reported to be associated inversely with BMI, rather than adenocarcinomas, which are reported to be associated positively:5,31 histological subtype is, however, not available in the PSC. The inverse association for lung and upper aerodigestive cancer combined was still strongly negative even after exclusion of the first 10 years of follow-up, implying that it was not chiefly a result of reverse causality.

orted to be associated positively:5,31 histological subtype is, however, not available in the PSC. The inverse association for lung and upper aerodigestive cancer combined was still strongly negative even after exclusion of the first 10 years of follow-up, implying that it was not chiefly a result of reverse causality. For COPD and other respiratory diseases, the inverse associations with BMI in the range 15–25 kg/m2 were remarkably strong. In each sex, the inverse association for respiratory mortality accounted for about 60% of the difference in all-cause mortality between 15–20 kg/m2 and 22·5–25 kg/m2 (figures 2 and 5). COPD can cause weight loss over many years, so the inverse association (even after exclusion of the first 15 years of follow-up) might have been due mainly to reverse causality (ie, to low BMI being an indicator of progressive COPD). However, some close correlate of low BMI itself could increase COPD progression and, hence, mortality.

eight loss over many years, so the inverse association (even after exclusion of the first 15 years of follow-up) might have been due mainly to reverse causality (ie, to low BMI being an indicator of progressive COPD). However, some close correlate of low BMI itself could increase COPD progression and, hence, mortality. The inverse associations with COPD, lung cancer, and upper aerodigestive cancer were much steeper in smokers than in non-smokers. Smoking is a major cause of all three diseases, and the greater steepness in smokers might have been due at least partly to uncontrolled confounding by smoking intensity. Smoking can cause weight loss,32 and if greater intensity of smoking were to cause increased weight loss, then there would be a substantially greater proportion of intensive smokers in the lower BMI categories, who would be at greater risk of these conditions (both through direct effects and, possibly, as a result of being less likely to quit). Although cigarettes smoked per day varied little with BMI in this study, other evidence (webappendix p 13) suggests that, for a specific number of cigarettes per day, leaner smokers have substantially higher blood cotinine concentrations than other smokers do (and also substantially more lung cancer, upper aerodigestive cancer, and COPD). Hence, smoking intensity might confound associations with BMI even in the absence of an association between BMI and daily cigarette consumption. Alternatively, lower BMI might somehow exacerbate the effects of smoking on respiratory cancer or other respiratory disease. The steep inverse associations for these diseases among smokers are still largely unexplained.

ciations with BMI even in the absence of an association between BMI and daily cigarette consumption. Alternatively, lower BMI might somehow exacerbate the effects of smoking on respiratory cancer or other respiratory disease. The steep inverse associations for these diseases among smokers are still largely unexplained. This study did not assess measures of central obesity, but other large epidemiological studies have done so. In the ten-country EPIC prospective study (with 12 000 deaths of known cause, of which 3000 were vascular [vs 36 000 vascular deaths in the PSC]),16 waist circumference improved the ability of BMI to predict vascular and all-cause mortality. In the 52-country INTERHEART case–control study of acute myocardial infarction (with 12 000 cases),33 a difference of 5 kg/m2 in BMI seems, for reasons that are not clear, to be of much less relevance to heart disease (odds ratio ∼1·12 [95% CI 1·08–1·16]) than it was in the PSC (HR 1·39 [1·34–1·44]) or in EPIC (HR ∼1·4), and hence to be of much less relevance than measures of central obesity are. Since case–control studies have greater potential for some types of bias, disentangling the interdependent associations of closely correlated anthropometric variables with particular diseases might need prospective studies that are even larger than this PSC study.

ch less relevance than measures of central obesity are. Since case–control studies have greater potential for some types of bias, disentangling the interdependent associations of closely correlated anthropometric variables with particular diseases might need prospective studies that are even larger than this PSC study. This report cannot quantify the effects of present levels of childhood obesity on adult mortality over the next few decades; the relevance of obesity to mortality in different ethnic groups; the substantial effects of obesity on disability, quality of life, or non-fatal disease (eg, osteoarthritis, obstructive sleep apnoea); or the positive effects of some types of adiposity on prognosis after some chronic disorders (eg, heart failure,34 respiratory disease35) have already developed. It does, however, quantify particularly reliably both the excess mortality associated with low BMI (much of which could be non-causal) and that associated with high BMI (which would be even greater if full allowance could be made for the extent to which chronic disease can cause weight loss). If the overall inverse association at low BMI is partly non-causal, then the real optimum BMI might be somewhat lower than the apparent optimum of about 23 kg/m2 or 24 kg/m2.

ciated with high BMI (which would be even greater if full allowance could be made for the extent to which chronic disease can cause weight loss). If the overall inverse association at low BMI is partly non-causal, then the real optimum BMI might be somewhat lower than the apparent optimum of about 23 kg/m2 or 24 kg/m2. The absolute excess mortality at BMI greater than 22·5–25 kg/m2 was mainly vascular, but also partly neoplastic, and was probably largely causal (ie, due to causal factors closely associated with BMI). Figure 7 shows, for different BMI levels in middle age, estimates of the lifetime probabilities of surviving from age 35 years, which are calculated by applying the relative risks that were considered likely to be causal (webappendix p 18) to disease-specific mortality rates at ages 35–79 years from the EU in 2000.13 (The year 2000 EU probability of surviving from birth to age 35 years is 98%.) For both sexes, the median survival (figure 7A) is reduced by 0–1 year for people who would, by about age 60 years, reach a BMI of 25–27·5 kg/m2, by 1–2 years for those who would reach 27·5–30 kg/m2, and by 2–4 years for those who would become obese (30–35 kg/m2). Much less information was available for BMI greater than 35 kg/m2 (hence the dashed lines in figure 7B), but the median survival seems to be reduced by about 8–10 years in those who would become morbidly obese (40–50 kg/m2, which in the PSC is mainly 40–45 kg/m2).

4 years for those who would become obese (30–35 kg/m2). Much less information was available for BMI greater than 35 kg/m2 (hence the dashed lines in figure 7B), but the median survival seems to be reduced by about 8–10 years in those who would become morbidly obese (40–50 kg/m2, which in the PSC is mainly 40–45 kg/m2). The extreme reduction in survival with morbid obesity is about as great as the 10-year reduction caused by persistent cigarette smoking in male British doctors born in 1900–30, for whom the cigarette smoker versus non-smoker mortality rate ratio was about 2·5 not only at 35–69 years but also at 70–79 years of age.13,36 In the present report, the smoker versus non-smoker mortality rate ratio is slightly less than 2·5 for men aged 35–69 years, and much less than 2·5 for women aged 70–79 years (webappendix p 7). In both cases this was partly because many who were current smokers at baseline did not smoke as many cigarettes when young as the British doctors did (or, indeed, as young smokers do nowadays), and partly because many who were current smokers at baseline in the PSC would have quit during follow-up (which is taken account of in the doctors' study,36 but not in the PSC). The difference in mortality between smokers and non-smokers in figure 6 therefore underestimates the effects of smoking throughout adult life, but it could likewise underestimate the effects of becoming obese well before middle age.

ring follow-up (which is taken account of in the doctors' study,36 but not in the PSC). The difference in mortality between smokers and non-smokers in figure 6 therefore underestimates the effects of smoking throughout adult life, but it could likewise underestimate the effects of becoming obese well before middle age. These PSC relative risks for BMI, combined with recent population BMI values,37,38 suggest that in the present decade, about 29% of vascular deaths and 8% of neoplastic deaths in late middle age in the USA (where mean BMI6 at age 50 years was 28·5–29 kg/m2 in 2000) would have been attributable to having a BMI greater than 25 kg/m2; for the UK (where mean BMI38 at that age was about 1 kg/m2 lower), the corresponding proportions would have been about 23% and 6%, respectively. In both countries, as elsewhere, these proportions will probably increase if average BMI in middle age continues to rise, even if rates of vascular and neoplastic mortality continue to fall because of decreases in smoking, improvements in treatment, or other reasons. Moreover, since BMI is an imperfect measure of visceral and other adiposity, the number of vascular and other deaths attributable to all adiposity-related factors is probably appreciably greater than these calculations suggest.

ntinue to fall because of decreases in smoking, improvements in treatment, or other reasons. Moreover, since BMI is an imperfect measure of visceral and other adiposity, the number of vascular and other deaths attributable to all adiposity-related factors is probably appreciably greater than these calculations suggest. In adult life, it may be easier to avoid substantial weight gain than to lose that weight once it has been gained. By avoiding a further increase from 28 kg/m2 to 32 kg/m2, a typical person in early middle age would gain about 2 years of life expectancy. Alternatively, by avoiding an increase from 24 kg/m2 to 32 kg/m2 (ie, to a third above the apparent optimum), a young adult would on average gain about 3 extra years of life. Web Extra Material Supplementary webappendix Acknowledgments The Prospective Studies Collaboration has been supported by the UK Medical Research Council, British Heart Foundation, Cancer Research UK, EU BIOMED programme, NIA grant P01 AG17625-01, and CTSU overheads. Merck (F Walker) helped support the 1996 meeting of collaborators. Gary Whitlock was supported by a Girdlers' Health Research Council of New Zealand Fellowship. Sarah Lewington had a British Heart Foundation Fellowship to coordinate the project. Sarah Parish supplied unpublished analyses of BMI and cotinine (webappendix).

ck (F Walker) helped support the 1996 meeting of collaborators. Gary Whitlock was supported by a Girdlers' Health Research Council of New Zealand Fellowship. Sarah Lewington had a British Heart Foundation Fellowship to coordinate the project. Sarah Parish supplied unpublished analyses of BMI and cotinine (webappendix). Contributors All members of the writing committee contributed to the collection and analysis of the data, and to the preparation of the report. All collaborators had an opportunity to contribute to the interpretation of the results and to the re-drafting of the report. The writing committee accepts full responsibility for the content of this paper. Members of the Prospective Studies Collaboration Writing Committee—Gary Whitlock, Sarah Lewington, Paul Sherliker, Robert Clarke, Jonathan Emberson, Jim Halsey, Nawab Qizilbash, Rory Collins, Richard Peto. Steering Committee—S Lewington (coordinator and statistician), S MacMahon (chair), R Peto (statistician), A Aromaa, C Baigent, J Carstensen, Z Chen, R Clarke, R Collins, S Duffy, D Kromhout, J Neaton, N Qizilbash, A Rodgers, S Tominaga, S Törnberg, H Tunstall-Pedoe, G Whitlock.

Members of the Prospective Studies Collaboration Writing Committee—Gary Whitlock, Sarah Lewington, Paul Sherliker, Robert Clarke, Jonathan Emberson, Jim Halsey, Nawab Qizilbash, Rory Collins, Richard Peto. Steering Committee—S Lewington (coordinator and statistician), S MacMahon (chair), R Peto (statistician), A Aromaa, C Baigent, J Carstensen, Z Chen, R Clarke, R Collins, S Duffy, D Kromhout, J Neaton, N Qizilbash, A Rodgers, S Tominaga, S Törnberg, H Tunstall-Pedoe, G Whitlock. PSC Collaborators—Atherosclerosis Risk in Communities (ARIC): L Chambless; Belgian Inter-university Research on Nutrition and Health (BIRNH): G De Backer, D De Bacquer, M Kornitzer; British Regional Heart Study (BRHS): P Whincup, S G Wannamethee, R Morris; British United Provident Association (BUPA): N Wald, J Morris, M Law; Busselton: M Knuiman, H Bartholomew; Caerphilly and Speedwell: G Davey Smith, P Sweetnam, P Elwood, J Yarnell; Cardiovascular Health Study (CHS): R Kronmal; CB Project: D Kromhout; Charleston: S Sutherland, J Keil; Copenhagen City Heart Study: G Jensen, P Schnohr; Evans County: C Hames (deceased), A Tyroler; Finnish Mobile Clinic Survey (FMCS): A Aromaa, P Knekt, A Reunanen; Finrisk: J Tuomilehto, P Jousilahti, E Vartiainen, P Puska; Flemish Study on Environment, Genes and Health (FLEMENGHO): T Kuznetsova, T Richart, J Staessen, L Thijs; Research Centre for Prevention and Health (Glostrup Population Studies): T Jørgensen, T Thomsen; Honolulu Heart Program: D Sharp, J D Curb; Imperial College and Oxon Epidemiology Limited: N Qizilbash; Ikawa, Noichi and Kyowa: H Iso, S Sato, A Kitamura, Y Naito; Centre d'Investigations Preventives et Cliniques (IPC), Paris: A Benetos, L Guize (deceased); Israeli Ischaemic Heart Disease Study: U Goldbourt; Japan Railways: M Tomita, Y Nishimoto, T Murayama; Lipid Research Clinics Follow-up Study (LRC): M Criqui, C Davis; Midspan Collaborative Study: C Hart, G Davey Smith, D Hole (deceased), C Gillis; Minnesota Heart Health Project (MHHP) and Minnesota Heart Survey (MHS): D Jacobs, H Blackburn, R Luepker; Multiple Risk Factor Intervention Trial (MRFIT): J Neaton, L Eberly; First National Health and Nutrition Examination Survey Epidemiologic Follow-up Study (NHEFS): C Cox; NHLBI Framingham Heart Study: D Levy, R D'Agostino, H Silbershatz; Norwegian Counties Study: A Tverdal, R Selmer; Northwick Park Heart Study (NPHS): T Meade, K Garrow, J Cooper; Nurses' Health Study: F Speizer, M Stampfer; Occupational Groups (OG), Rome: A Menotti, A Spagnolo; Ohasama: I Tsuji, Y Imai, T Ohkubo, S Hisamichi; Oslo: L Haheim, I Holme, I Hjermann, P Leren; Paris Prospective Study: P Ducimetiere, J Empana; Perth: K Jamrozik, R Broa

Heart Study (NPHS): T Meade, K Garrow, J Cooper; Nurses' Health Study: F Speizer, M Stampfer; Occupational Groups (OG), Rome: A Menotti, A Spagnolo; Ohasama: I Tsuji, Y Imai, T Ohkubo, S Hisamichi; Oslo: L Haheim, I Holme, I Hjermann, P Leren; Paris Prospective Study: P Ducimetiere, J Empana; Perth: K Jamrozik, R Broa dhurst; Prospective Cardiovascular Munster Study (PROCAM): G Assmann, H Schulte; Prospective Study of Women in Gothenburg: C Bengtsson, C Björkelund, L Lissner; Puerto Rico Health Heart Program (PRHHP): P Sorlie, M Garcia-Palmieri; Rancho Bernardo: E Barrett-Connor, M Criqui, R Langer; Renfrew and Paisley study: C Hart, G Davey Smith, D Hole (deceased); Saitama Cohort Study: K Nakachi, K Imai; Seven Cities China: X Fang, S Li; Seven Countries (SC) Croatia: R Buzina; SC Finland: A Nissinen; SC Greece (Greek Islands Study): C Aravanis, A Dontas, A Kafatos; SC Italy: A Menotti; SC Japan: H Adachi, H Toshima, T Imaizumi; SC Netherlands: D Kromhout; SC Serbia: S Nedeljkovic, M Ostojic; Shanghai: Z Chen; Scottish Heart Health Study (SHHS): H Tunstall-Pedoe; Shibata: T Nakayama, N Yoshiike, T Yokoyama, C Date, H Tanaka; Tecumseh: J Keller; Tromso: K Bonaa, E Arnesen; United Kingdom Heart Disease Prevention Project (UK HDPP): H Tunstall-Pedoe; US Health Professionals Follow-up Study: E Rimm; US Physicians' Health Study: M Gaziano, J E Buring, C Hennekens; Värmland: S Törnberg, J Carstensen; Whitehall: M Shipley, D Leon, M Marmot; Clinical Trial Service Unit (CTSU): R Clarke, R Collins, J Emberson, J Halsey, S Lewington, A Palmer (deceased), S Parish, R Peto, P Sherliker, G Whitlock.

ollow-up Study: E Rimm; US Physicians' Health Study: M Gaziano, J E Buring, C Hennekens; Värmland: S Törnberg, J Carstensen; Whitehall: M Shipley, D Leon, M Marmot; Clinical Trial Service Unit (CTSU): R Clarke, R Collins, J Emberson, J Halsey, S Lewington, A Palmer (deceased), S Parish, R Peto, P Sherliker, G Whitlock. Conflict of interest statement All the writing committee (except NQ) work in the CTSU, which has a policy of staff not accepting fees, honoraria, or consultancies. The CTSU is involved in clinical trials with funding from the UK Medical Research Council, British Heart Foundation, and/or various companies (AstraZeneca, Bayer, Merck, Schering-Plough, Solvay) as research grants to (and administered by) the University of Oxford. NQ is a former Director of Epidemiology at GlaxoSmithKline and now works in Oxon Epidemiology, which has undertaken consultancy work for several pharmaceutical companies (including GlaxoSmithKline, Pfizer, Lilly, Roche, Gilead, and Grunenthal). Figure 1 Vascular risk factors versus BMI at baseline in the range 15–50 kg/m2

Conflict of interest statement All the writing committee (except NQ) work in the CTSU, which has a policy of staff not accepting fees, honoraria, or consultancies. The CTSU is involved in clinical trials with funding from the UK Medical Research Council, British Heart Foundation, and/or various companies (AstraZeneca, Bayer, Merck, Schering-Plough, Solvay) as research grants to (and administered by) the University of Oxford. NQ is a former Director of Epidemiology at GlaxoSmithKline and now works in Oxon Epidemiology, which has undertaken consultancy work for several pharmaceutical companies (including GlaxoSmithKline, Pfizer, Lilly, Roche, Gilead, and Grunenthal). Figure 1 Vascular risk factors versus BMI at baseline in the range 15–50 kg/m2 Adjusted for baseline age, baseline smoking status (apart from the smoking findings), and study. Numerical values are shown for 20–22·5 kg/m2, for 30–32·5 kg/m2, and for the extreme BMI groups. Boundaries of BMI groups are indicated by tick marks. 95% CIs are not shown, but most are narrower than the heights of the plotted symbols. (A) Blood pressure (in 533 242 males and 348 790 females). (B) Blood cholesterol fractions (in 62 364 males and 52 575 females with total and HDL cholesterol both measured); dashed line indicates the ratio of mean non-HDL cholesterol to mean HDL cholesterol (mean of the individual ratios would be about 8–12% greater). (C) Prevalences in males for alcohol drinking (168 283), cigarette smoking (334 496), and diabetes (378 854). (D) Prevalences in females for alcohol drinking (129 301), cigarette smoking (226 307), and diabetes (319 401).

cholesterol to mean HDL cholesterol (mean of the individual ratios would be about 8–12% greater). (C) Prevalences in males for alcohol drinking (168 283), cigarette smoking (334 496), and diabetes (378 854). (D) Prevalences in females for alcohol drinking (129 301), cigarette smoking (226 307), and diabetes (319 401). Figure 2 All-cause mortality versus BMI for each sex in the range 15–50 kg/m2 (excluding the first 5 years of follow-up) Relative risks at ages 35–89 years, adjusted for age at risk, smoking, and study, were multiplied by a common factor (ie, floated) to make the weighted average match the PSC mortality rate at ages 35–79 years. Floated mortality rates shown above each square and numbers of deaths below. Area of square is inversely proportional to the variance of the log risk. Boundaries of BMI groups are indicated by tick marks. 95% CIs for floated rates reflect uncertainty in the log risk for each single rate. Dotted vertical line indicates 25 kg/m2 (boundary between upper and lower BMI ranges in this report). Figure 3 Ischaemic heart disease and stroke mortality versus BMI in the range 15–50 kg/m2 (excluding the first 5 years of follow-up)

Relative risks at ages 35–89 years, adjusted for age at risk, smoking, and study, were multiplied by a common factor (ie, floated) to make the weighted average match the PSC mortality rate at ages 35–79 years. Floated mortality rates shown above each square and numbers of deaths below. Area of square is inversely proportional to the variance of the log risk. Boundaries of BMI groups are indicated by tick marks. 95% CIs for floated rates reflect uncertainty in the log risk for each single rate. Dotted vertical line indicates 25 kg/m2 (boundary between upper and lower BMI ranges in this report). Figure 3 Ischaemic heart disease and stroke mortality versus BMI in the range 15–50 kg/m2 (excluding the first 5 years of follow-up) Relative risks at ages 35–89 years, adjusted for age at risk, sex, smoking, and study, were multiplied by a common factor (ie, floated) to make the weighted average match the PSC mortality rate at ages 35–79 years. Floated mortality rates shown above each square and numbers of deaths below. Area of square is inversely proportional to the variance of the log risk. Boundaries of BMI groups are indicated by tick marks. 95% CIs for floated rates reflect uncertainty in the log risk for each single rate. Figure 4 Ischaemic heart disease (A) and stroke mortality (B) versus BMI in the upper BMI range (25–50 kg/m2) only (excluding the first 5 years of follow-up, except as indicated)

Relative risks at ages 35–89 years, adjusted for age at risk, sex, smoking, and study, were multiplied by a common factor (ie, floated) to make the weighted average match the PSC mortality rate at ages 35–79 years. Floated mortality rates shown above each square and numbers of deaths below. Area of square is inversely proportional to the variance of the log risk. Boundaries of BMI groups are indicated by tick marks. 95% CIs for floated rates reflect uncertainty in the log risk for each single rate. Figure 4 Ischaemic heart disease (A) and stroke mortality (B) versus BMI in the upper BMI range (25–50 kg/m2) only (excluding the first 5 years of follow-up, except as indicated) Hazard ratios are per 5 kg/m2—eg, 30 kg/m2 versus 25 kg/m2—and are, when appropriate, adjusted for age at risk, sex, smoking, and study. Mean ages at death are given, but the dependence of the HR on mean age at death is not corrected for in analyses of factors other than age. The area of each square is inversely proportional to the variance of the log hazard ratio. White squares include the first 5 years of follow-up; black squares and white diamonds do not. Subarachnoid=subarachnoid haemorrhage (not included in haemorrhagic stroke). Figure 5 Mortality rates at ages 35–79 years for main disease categories versus BMI in the range 15–50 kg/m2 (excluding the first 5 years of follow-up)

Hazard ratios are per 5 kg/m2—eg, 30 kg/m2 versus 25 kg/m2—and are, when appropriate, adjusted for age at risk, sex, smoking, and study. Mean ages at death are given, but the dependence of the HR on mean age at death is not corrected for in analyses of factors other than age. The area of each square is inversely proportional to the variance of the log hazard ratio. White squares include the first 5 years of follow-up; black squares and white diamonds do not. Subarachnoid=subarachnoid haemorrhage (not included in haemorrhagic stroke). Figure 5 Mortality rates at ages 35–79 years for main disease categories versus BMI in the range 15–50 kg/m2 (excluding the first 5 years of follow-up) Relative risks at ages 35–79 years, adjusted for age at risk, smoking, and study, were multiplied by a common factor (ie, floated) to make the weighted average match the age-standardised European Union (15 countries) mortality rate at ages 35–79 years in 2000. Neoplastic mortality is split into the types most strongly associated with smoking (cancers of the lung and upper aerodigestive tract) and all other specified types. By contrast with figures 2–4, risk is indicated on an additive rather than multiplicative scale, with floated mortality rates shown above or below each symbol. The estimates for 35–50 kg/m2 are based on limited data, so lines connecting to those estimates are dashed. Boundaries of BMI groups are indicated by tick marks. 95% CIs for floated rates reflect uncertainty in the log risk for each single rate.

tive scale, with floated mortality rates shown above or below each symbol. The estimates for 35–50 kg/m2 are based on limited data, so lines connecting to those estimates are dashed. Boundaries of BMI groups are indicated by tick marks. 95% CIs for floated rates reflect uncertainty in the log risk for each single rate. Figure 6 All-cause mortality at ages 35–79 years versus BMI in the range 15–50 kg/m2, by smoking status (excluding the first 5 years of follow-up) Relative risks at ages 35–79 years, adjusted for age at risk, sex, and study, were multiplied by a common factor (ie, floated) so that the mean for all participants (including ex-smokers and anyone with missing smoking data) matches the European rate at ages 35–79 years in 2000. Results for ex-smokers and those with missing smoking data not shown (but are, taken together, only slightly above those for never smokers). Note that many smokers were at only limited risk, since they had not smoked many cigarettes during early adult life, or had stopped shortly after the baseline survey. Risk is indicated on an additive rather than multiplicative scale. The estimates for 35–50 kg/m2 are based on limited data, so lines connecting to those estimates are dashed. Floated mortality rates shown above each square and numbers of deaths below. Area of square is inversely proportional to the variance of the log risk. Boundaries of BMI groups are indicated by tick marks. 95% CIs for floated rates reflect uncertainty in the log risk for each single rate. Figure 7 BMI versus lifespan in western Europe, year 2000

Relative risks at ages 35–79 years, adjusted for age at risk, sex, and study, were multiplied by a common factor (ie, floated) so that the mean for all participants (including ex-smokers and anyone with missing smoking data) matches the European rate at ages 35–79 years in 2000. Results for ex-smokers and those with missing smoking data not shown (but are, taken together, only slightly above those for never smokers). Note that many smokers were at only limited risk, since they had not smoked many cigarettes during early adult life, or had stopped shortly after the baseline survey. Risk is indicated on an additive rather than multiplicative scale. The estimates for 35–50 kg/m2 are based on limited data, so lines connecting to those estimates are dashed. Floated mortality rates shown above each square and numbers of deaths below. Area of square is inversely proportional to the variance of the log risk. Boundaries of BMI groups are indicated by tick marks. 95% CIs for floated rates reflect uncertainty in the log risk for each single rate. Figure 7 BMI versus lifespan in western Europe, year 2000 Estimated effects of the BMI that would be reached by about 60 years of age on survival from age 35 years, identifying European Union (EU) mortality rates in 2000 with those for BMI 25–30 kg/m2 and combining the disease-specific EU mortality rates with disease-specific relative risks (for details, see webappendix pp 18–20). The absolute differences in median survival (but probably not in survival to age 70 years) should be robust to changes in mortality rates, and therefore generalisable decades hence. (A) 3 main BMI categories. (B) 3 main and 2 higher BMI categories. (The 2 higher BMI categories account for just 2% of PSC participants, and so are indicated by dashed lines.)

rvival (but probably not in survival to age 70 years) should be robust to changes in mortality rates, and therefore generalisable decades hence. (A) 3 main BMI categories. (B) 3 main and 2 higher BMI categories. (The 2 higher BMI categories account for just 2% of PSC participants, and so are indicated by dashed lines.) Table 1 All-cause mortality versus baseline BMI in the ranges 15–25 kg/m2 and 25–50 kg/m2 All participants Never smokers only 15–25 kg/m2 25–50 kg/m2 15–25 kg/m2 25–50 kg/m2 Deaths HR (95% CI) Deaths HR (95% CI) Deaths HR (95% CI) Deaths HR (95% CI) Overall 35 256 0·79 (0·77–0·82) 37 493 1·29 (1·27–1·32) 7054 0·87 (0·81–0·94) 9849 1·32 (1·28–1·36) Male 26 720 0·79 (0·76–0·82) 27 983 1·32 (1·29–1·36) 3694 0·87 (0·78–0·97) 4811 1·44 (1·36–1·53) Female 8536 0·80 (0·75–0·85) 9510 1·26 (1·23–1·30) 3360 0·87 (0·78–0·97) 5038 1·27 (1·22–1·32) Age at risk (years) 35–59 9333 0·76 (0·71–0·81) 8386 1·37 (1·31–1·42) 1665 0·88 (0·74–1·04) 1667 1·43 (1·32–1·55) 60–69 11 514 0·77 (0·73–0·82) 13 007 1·32 (1·27–1·36) 1782 0·88 (0·75–1·03) 2841 1·36 (1·28–1·45) 70–79 10 078 0·82 (0·77–0·87) 11 358 1·27 (1·23–1·32) 2116 0·93 (0·80–1·06) 3364 1·33 (1·25–1·40) 80–89 4331 0·89 (0·80–0·97) 4742 1·16 (1·10–1·23) 1491 0·86 (0·74–1·01) 1977 1·15 (1·07–1·25) Hazard ratio per 5 kg/m2 higher BMI (HR). HR less than 1 if BMI inversely associated with risk. All analyses exclude the first 5 years of follow-up and adjust for study and age at risk (in 5-year groups). The overall and age-specific analyses also adjust for sex, and the all-participant analyses also adjust for baseline smoking status.

5 kg/m2 higher BMI (HR). HR less than 1 if BMI inversely associated with risk. All analyses exclude the first 5 years of follow-up and adjust for study and age at risk (in 5-year groups). The overall and age-specific analyses also adjust for sex, and the all-participant analyses also adjust for baseline smoking status. Table 2 Cause-specific mortality versus baseline BMI in the ranges 15–25 kg/m2 and 25–50 kg/m2

5 kg/m2 higher BMI (HR). HR less than 1 if BMI inversely associated with risk. All analyses exclude the first 5 years of follow-up and adjust for study and age at risk (in 5-year groups). The overall and age-specific analyses also adjust for sex, and the all-participant analyses also adjust for baseline smoking status. Table 2 Cause-specific mortality versus baseline BMI in the ranges 15–25 kg/m2 and 25–50 kg/m2 15–25 kg/m2 25–50 kg/m2 Deaths HR (95% CI) Deaths HR (95% CI) Ischaemic heart disease 7461 1·22 (1·13–1·32) 10 783 1·39 (1·34–1·44) Stroke 2964 0·92 (0·82–1·03) 3164 1·39 (1·31–1·48) Other vascular disease 2648 0·84 (0·75–0·95) 3396 1·47 (1·39–1·56) Diabetes 171 0·96 (0·59–1·55) 393 2·16 (1·89–2·46) Kidney disease (non-neoplastic) 197 1·14 (0·74–1·77) 217 1·59 (1·27–1·99) Liver disease (non-neoplastic) 489 0·69 (0·52–0·91) 603 1·82 (1·59–2·09) Lung cancer 2959 0·71 (0·63–0·79) 2040 0·98 (0·88–1·09) Upper aerodigestive cancer 685 0·49 (0·39–0·61) 471 0·98 (0·79–1·20) Other specified cancer 6134 0·94 (0·87–1·02) 6190 1·12 (1·06–1·18) Respiratory disease* 2426 0·31 (0·28–0·35) 1344 1·20 (1·07–1·34) Other specified disease 2049 0·62 (0·54–0·71) 1823 1·20 (1·10–1·31) External cause 2112 0·82 (0·71–0·95) 1720 1·19 (1·08–1·32) Unknown cause† 4961 0·72 (0·66–0·79) 5349 1·22 (1·16–1·28) All causes 35 256 0·79 (0·77–0·82) 37 493 1·29 (1·27–1·32) Hazard ratio per 5 kg/m2 higher BMI (HR). HR less than 1 if BMI inversely associated with risk. Analyses exclude the first 5 years of follow-up, and adjust for study, sex, age at risk (in 5-year groups), and baseline smoking status. For analyses restricted to those who had never smoked, see webappendix p 17.

1·27–1·32) Hazard ratio per 5 kg/m2 higher BMI (HR). HR less than 1 if BMI inversely associated with risk. Analyses exclude the first 5 years of follow-up, and adjust for study, sex, age at risk (in 5-year groups), and baseline smoking status. For analyses restricted to those who had never smoked, see webappendix p 17. * HR 0·37 (95% CI 0·30–0·44) in the range 15–25 kg/m2 after exclusion of the first 15 years of follow-up (leaving 956 deaths). † Includes 4113 deaths from cancer of unspecified site.

on or occlusive stroke while not on aspirin would probably then have started long-term aspirin to avoid recurrence, so the mortality results from those trials can help to decide between the policies of immediate versus deferred aspirin (ie, deferral of the start of long-term aspirin until there is evidence of disease). In view of the limitations of the analyses underlying current guidelines, and the large populations affected by these guidelines, a collaborative meta-analysis of individual participant data was established involving the principal investigators of all large trials of primary prevention with aspirin. Meta-analyses of previously obtained individual participant data from 16 secondary prevention trials of aspirin were also undertaken to compare the proportional and absolute effects of aspirin in these two treatment settings.1,2

Introduction In patients who are at high risk because they already have occlusive vascular disease, long-term antiplatelet therapy (eg, with aspirin) reduces the yearly risk of serious vascular events (non-fatal myocardial infarction, non-fatal stroke, or vascular death) by about a quarter.1,2 This decrease typically corresponds to an absolute reduction of about 10–20 per 1000 in the yearly incidence of non-fatal events, and to a smaller, but still definite, reduction in vascular death. Against this benefit, the absolute increase in major gastrointestinal or other major extracranial bleeds is an order of magnitude smaller. Hence, for secondary prevention, the benefits of antiplatelet therapy substantially exceed the risks.

atal events, and to a smaller, but still definite, reduction in vascular death. Against this benefit, the absolute increase in major gastrointestinal or other major extracranial bleeds is an order of magnitude smaller. Hence, for secondary prevention, the benefits of antiplatelet therapy substantially exceed the risks. For primary prevention, however, the balance is less clear because the risks without aspirin, and hence the absolute benefits of aspirin, are generally an order of magnitude lower than in secondary prevention. Previous meta-analyses of primary prevention trials were not based on individual participant data, so they could not compare reliably the benefits and risks of aspirin in prognostically important groups (such as older people and others at increased risk of coronary heart disease), and could not quantify reliably the extent to which people at increased risk of coronary heart disease might also be at increased risk of bleeding. Current guidelines largely ignore any differences in bleeding risk, and recommend that aspirin be used widely for primary prevention in those at moderately raised risk of coronary heart disease.3–5 It has also been suggested that, since age is a major determinant of the risk of coronary heart disease, daily aspirin should be started in all people above a specific age, either alone or in combination with other drugs.6–8

widely for primary prevention in those at moderately raised risk of coronary heart disease.3–5 It has also been suggested that, since age is a major determinant of the risk of coronary heart disease, daily aspirin should be started in all people above a specific age, either alone or in combination with other drugs.6–8 The alternative to primary prevention is deferral of the start of long-term aspirin until some evidence of occlusive vascular disease is noted. The main disadvantage of deferral is that the first manifestation of disease might be a disabling or fatal event, but the main advantage is that it could avoid decades of slightly increased risk of cerebral haemorrhage or major extracranial bleeding. In the primary prevention trials, most controls who had a non-fatal myocardial infarction or occlusive stroke while not on aspirin would probably then have started long-term aspirin to avoid recurrence, so the mortality results from those trials can help to decide between the policies of immediate versus deferred aspirin (ie, deferral of the start of long-term aspirin until there is evidence of disease).

volving the principal investigators of all large trials of primary prevention with aspirin. Meta-analyses of previously obtained individual participant data from 16 secondary prevention trials of aspirin were also undertaken to compare the proportional and absolute effects of aspirin in these two treatment settings.1,2 Methods Trial eligibility Primary or secondary prevention trials were eligible only if they involved a randomised comparison of aspirin versus no aspirin (with no other antiplatelet drug in either group). Primary prevention trials excluded individuals with any history of occlusive disease at entry. (Subsequent enquiry showed that 2% did in fact have some evidence of previous vascular disease, but they remain in all analyses apart from those estimating the absolute effects of aspirin.) Primary prevention trials were sought only if they recruited at least 1000 non-diabetic participants with at least 2 years of scheduled treatment. Individual participant data were provided from all six published trials.9–14 Unpublished trials were sought through electronic searches and discussions, but none was identified.

ary prevention trials were sought only if they recruited at least 1000 non-diabetic participants with at least 2 years of scheduled treatment. Individual participant data were provided from all six published trials.9–14 Unpublished trials were sought through electronic searches and discussions, but none was identified. Secondary prevention trials were included in analyses if they involved individuals with previous myocardial infarction (six trials) or stroke or transient cerebral ischaemia (ten trials), and had contributed individual participant data to the 2002 Antithrombotic Trialists' (ATT) report (webappendix pp 11–23).1,2 Two further trials for which only tabular data were available are shown in the webappendix but do not contribute to analyses. Electronic searches established that no similar trials of aspirin had been reported since 2002.

ual participant data to the 2002 Antithrombotic Trialists' (ATT) report (webappendix pp 11–23).1,2 Two further trials for which only tabular data were available are shown in the webappendix but do not contribute to analyses. Electronic searches established that no similar trials of aspirin had been reported since 2002. Prespecified analyses The comparisons were intention-to-treat analyses of first events during the scheduled treatment period in all participants allocated aspirin versus all those allocated control (irrespective of any other treatment allocated factorially). The main outcomes were serious vascular event, defined as myocardial infarction, stroke, or death from a vascular cause (including sudden death, pulmonary embolism, haemorrhage, and, for secondary prevention trials only, death from an unknown cause); major coronary event (myocardial infarction, coronary death, or sudden death); any stroke (haemorrhagic or probably ischaemic [ie, definitely ischaemic or of unknown type]); death from any cause; and major extracranial bleed (mainly gastrointestinal and usually defined as a bleed requiring transfusion or resulting in death). In the primary prevention trials, myocardial infarctions and strokes were classified as fatal or non-fatal in accordance with each trial's definitions. In the secondary prevention trials, as previously,2 these outcomes were regarded as non-fatal only if the patient was alive at the end of the trial or died of a non-vascular cause. Five of the primary prevention trials classified stroke subtypes on the basis of either clinical examination10 or CT imaging.9,11,13,14 In the sixth trial,12 imaging information was available only for strokes that had been confirmed as cerebral bleeds. In most secondary prevention trials little information about stroke causes was available (webappendix pp 15–18).1,2

stroke subtypes on the basis of either clinical examination10 or CT imaging.9,11,13,14 In the sixth trial,12 imaging information was available only for strokes that had been confirmed as cerebral bleeds. In most secondary prevention trials little information about stroke causes was available (webappendix pp 15–18).1,2 Statistical analysis The log-rank observed minus expected (o–e) statistics, one from each trial, and their variances (v), were summed to produce, respectively, a grand total observed minus expected (G) and its variance (V). The one-step estimate of the log of the event rate ratio is G/V. The χ2 test statistic (χ2n–1) for heterogeneity between n trials is S–(G2/V), where S is the sum over all the trials of (o–e)2/v. Heterogeneity of rate ratios among multiple subgroups defined by baseline characteristics was investigated by a global heterogeneity test, which helps to avoid misinterpreting false positive results arising from multiple comparisons. (For each characteristic [eg, age] a χ2 test for trend on 1 degree of freedom was calculated. If there are 11 different χ2 tests and none of the trends is real, then their sum has expectation 11 and has variance at least as great as that of χ211.)

isinterpreting false positive results arising from multiple comparisons. (For each characteristic [eg, age] a χ2 test for trend on 1 degree of freedom was calculated. If there are 11 different χ2 tests and none of the trends is real, then their sum has expectation 11 and has variance at least as great as that of χ211.) For trials that randomised unequally (ie, 2:1), we multiplied the control group by two when displaying adjusted control totals and describing the total amount of information available, but not in other calculations. For the purposes of discussion, we calculated what the absolute effects of aspirin allocation would be on outcome at 5 years (only two trials9,14 had much longer follow-up) if the yearly event rates were constant and the proportional effects of aspirin were independent of age, sex, and other risk factors.

heart disease. *Myocardial infarction, stroke, or vascular death. Vascular death is coronary heart disease death, stroke death, or other vascular death (which includes sudden death, death from pulmonary embolism, and death from any haemorrhage, but in the primary prevention trials excludes death from an unknown cause). Figure 2 Serious vascular events in primary prevention trials—subgroup analyses Actual numbers for aspirin-allocated trial participants, and adjusted numbers for control-allocated trial participants, are presented, together with the corresponding mean yearly event rates (in parentheses). Rate ratios (RRs) for all trials are indicated by squares and their 99% CIs by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. A global test for heterogeneity (χ2 on 11 degrees of freedom) is provided. Unknown values are not plotted. SBP=systolic blood pressure. DBP=diastolic blood pressure. BMI=body-mass index. CHD=coronary heart disease. *Excluding patients with a history of vascular disease. Figure 3 Selected outcomes in primary and secondary prevention trials of aspirin, by sex

calculations. For the purposes of discussion, we calculated what the absolute effects of aspirin allocation would be on outcome at 5 years (only two trials9,14 had much longer follow-up) if the yearly event rates were constant and the proportional effects of aspirin were independent of age, sex, and other risk factors. To identify risk factors for various outcomes in people in the primary prevention trials without any known history of vascular disease, we used Poisson regression, stratified by trial, to estimate the common linear dependence of the log of the event rate on age, sex, diabetes, current cigarette smoking, total cholesterol, mean (of systolic and diastolic) blood pressure, body-mass index, and allocation to aspirin or control (webappendix p 24). Additionally, the results of this model for major coronary events in control participants only, together with the absolute event rates in the controls of each trial, were used to classify the baseline risks of all participants (including those allocated aspirin) as very low (predicted 5-year risk of coronary heart disease without aspirin <2·5%), low (2·5–5%), moderate (5–10%), or high (≥10%). Further details of the trials and of the analyses are available in the webappendix. Role of the funding sources The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The secretariat had full access to all the data in the study and the writing committee had final responsibility for the decision to submit for publication.

Further details of the trials and of the analyses are available in the webappendix. Role of the funding sources The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The secretariat had full access to all the data in the study and the writing committee had final responsibility for the decision to submit for publication. Results Six primary prevention trials were available (95 000 individuals, 3554 serious vascular events; table 1, figure 1).9–14 One12 recruited people with hypertension, and two9,13 recruited people with coronary risk factors (although without overt disease). The results were contrasted with those from the 16 secondary prevention trials (17 000 individuals, 3306 vascular events).2

dividuals, 3554 serious vascular events; table 1, figure 1).9–14 One12 recruited people with hypertension, and two9,13 recruited people with coronary risk factors (although without overt disease). The results were contrasted with those from the 16 secondary prevention trials (17 000 individuals, 3306 vascular events).2 In the primary prevention trials, 1671 serious vascular events occurred during 330 000 person-years (0·51% per year) in people allocated aspirin compared with an adjusted total of 1883 events during 330 000 person-years (0·57% per year) in those allocated control. This small absolute reduction (only 0·07% per year) represented a 12% proportional reduction (rate ratio [RR] 0·88 [95% CI 0·82–0·94], p=0·0001; figure 1), with no significant heterogeneity between the prespecified subgroups (global test for heterogeneity p=0·7; figure 2). The proportional reduction seemed similar (p=0·9) in men and women, and did not differ significantly (p trend=0·3) between those with predicted 5-year risk of coronary heart disease less than 2·5%, 2·5–5%, 5–10%, or 10% or more (figure 2). The apparently unpromising result in the 2% of all participants in the primary prevention trials who were in the highest risk group was statistically unreliable because it involved small numbers (131 [3·59% per year] vs 127 [3·44% per year] vascular events; RR 1·07 [99% CI 0·76–1·50], p=0·6; figure 2). The most important predictor of risk was age; the mean age at entry for those in the highest risk group was 69 years (SD 6) (so their mean age during the trials was over 70 years). The proportional reduction in risk of any serious vascular event did not differ significantly between primary and secondary prevention trials, but the absolute risk reduction was much smaller in primary than in secondary prevention (table 2).

was 69 years (SD 6) (so their mean age during the trials was over 70 years). The proportional reduction in risk of any serious vascular event did not differ significantly between primary and secondary prevention trials, but the absolute risk reduction was much smaller in primary than in secondary prevention (table 2). Since major coronary events and strokes accounted for a large proportion of serious vascular events, the effects of aspirin on each outcome were assessed separately. In the primary prevention trials, allocation to aspirin yielded an 18% proportional reduction in major coronary events, but only a small absolute reduction (0·28% vs 0·34% per year; RR 0·82 [95% CI 0·75–0·90], p<0·0001; figure 1). Most of this decrease derived from a 23% proportional reduction in non-fatal myocardial infarction (0·18% vs 0·23% per year; RR 0·77 [99% CI 0·67–0·89], p<0·0001; figure 1), with no clear reduction in mortality from coronary heart disease (0·11% vs 0·12% per year; RR 0·95 [99% CI 0·78–1·15], p=0·5), although the CI for the proportional reduction in mortality from coronary heart disease is wide. The proportional reduction in major coronary events seemed to be similar in primary and secondary prevention (RR 0·82 [95% CI 0·75–0·90] primary and RR 0·80 [0·73–0·88] secondary), but the absolute benefit differed by an order of magnitude (absolute benefits 0·06% per year primary and 1·00% per year secondary) (table 2).

e proportional reduction in major coronary events seemed to be similar in primary and secondary prevention (RR 0·82 [95% CI 0·75–0·90] primary and RR 0·80 [0·73–0·88] secondary), but the absolute benefit differed by an order of magnitude (absolute benefits 0·06% per year primary and 1·00% per year secondary) (table 2). For major coronary events in the primary prevention trials, the 11 separate tests for trend or heterogeneity of effect yielded only one, between men and women, that was marginally significant (p=0·03; figure 3 and webappendix p 4) if considered in isolation, but it was no longer significant if multiplied by 11 to allow for multiple comparisons (p=0·33). Furthermore, the 16 secondary prevention trials also suggested no heterogeneity of the effect on major coronary events between men and women (figure 3). Conversely, for ischaemic stroke, the six primary prevention trials suggested a greater proportional risk reduction in women than in men (p=0·08 for heterogeneity of effect between men and women if considered in isolation, but p=0·88 [not significant] if multiplied by 11 to allow for multiple comparisons); again, however, the 16 secondary prevention trials suggested no such heterogeneity of effect. For the aggregate of all serious vascular events, gender was of no apparent relevance in either type of trial to the proportional reduction produced by allocation to aspirin (figure 3).

y 11 to allow for multiple comparisons); again, however, the 16 secondary prevention trials suggested no such heterogeneity of effect. For the aggregate of all serious vascular events, gender was of no apparent relevance in either type of trial to the proportional reduction produced by allocation to aspirin (figure 3). Aspirin seemed to increase the incidence of haemorrhagic stroke both in the primary and in the secondary prevention trials (p=0·05 and p=0·07, respectively; p=0·01 when analysed together). Conversely, aspirin seemed to reduce the incidence of ischaemic stroke in both types of trial (p=0·05 and p=0·04, respectively; p=0·005 when analysed together). The proportion of strokes of known cause that were haemorrhagic was greater in the primary than in the secondary prevention trials (23% vs 15%; table 2); this was probably also true for strokes of unknown cause. For, at least 84% (893/1060: webappendix p 18) of the strokes in the secondary prevention trials were in patients with a previous history of ischaemic stroke or transient cerebral ischaemia, who would be at high risk of recurrence. In the primary prevention trials, aspirin had no net effect on strokes of known cause (haemorrhagic plus ischaemic), on strokes of unknown cause, or on the aggregate of all strokes (table 2). In the secondary prevention trials, however (in which a smaller proportion of the strokes of known cause were haemorrhagic), aspirin significantly reduced the aggregate of all strokes.

ect on strokes of known cause (haemorrhagic plus ischaemic), on strokes of unknown cause, or on the aggregate of all strokes (table 2). In the secondary prevention trials, however (in which a smaller proportion of the strokes of known cause were haemorrhagic), aspirin significantly reduced the aggregate of all strokes. As most strokes do not cause death, and haemorrhagic strokes may be more dangerous than ischaemic strokes, the proportional effects of aspirin on overall stroke mortality and on non-fatal stroke could differ. In figure 4 the results are therefore subdivided not only by cause but also by outcome (any outcome, or fatal). For strokes with any outcome (figures 4A, 4C, 4E, and 4G) the results are those from table 2. The subtotals in figure 4 combine the primary and secondary trial results. Taking both types of trial together, there was evidence of an adverse effect on haemorrhagic stroke (RR 1·39 [95% CI 1·08–1·78], p=0·01; figure 4A) but a protective effect on ischaemic stroke (RR 0·83 [0·73–0·95], p=0·005; figure 4C).

m table 2. The subtotals in figure 4 combine the primary and secondary trial results. Taking both types of trial together, there was evidence of an adverse effect on haemorrhagic stroke (RR 1·39 [95% CI 1·08–1·78], p=0·01; figure 4A) but a protective effect on ischaemic stroke (RR 0·83 [0·73–0·95], p=0·005; figure 4C). If attention is restricted to the fatal strokes just in the primary prevention trials, then fatal haemorrhagic strokes outnumber fatal ischaemic strokes (82 vs 53) and there is, if anything, an adverse effect on overall stroke mortality (119 vs 98 fatal strokes, p=0·18 [non-significant]; figure 4H). For, there was a significant excess of fatal haemorrhagic strokes in participants allocated aspirin (52 vs 30; RR 1·73 [99% CI 0·96–3·13], p=0·02; figure 4B) plus similar numbers of other fatal strokes (24 vs 29 ischaemic plus 43 vs 39 unknown cause; figures 4D and 4F).

al strokes, p=0·18 [non-significant]; figure 4H). For, there was a significant excess of fatal haemorrhagic strokes in participants allocated aspirin (52 vs 30; RR 1·73 [99% CI 0·96–3·13], p=0·02; figure 4B) plus similar numbers of other fatal strokes (24 vs 29 ischaemic plus 43 vs 39 unknown cause; figures 4D and 4F). Since allocation to aspirin had no significant effect on fatal stroke, fatal coronary heart disease or other vascular causes of death, there was no significant reduction in overall vascular mortality in the primary prevention trials (RR 0·97 [95% CI 0·87–1·09], p=0·7; figure 1). Since there was also no significant effect on non-vascular mortality (RR 0·93 [95% CI 0·85–1·02], p=0·1) or on mortality from unknown causes (RR 0·96 [99% CI 0·70–1·30], p=0·7), there was no significant effect on total mortality (RR 0·95 [95% CI 0·88–1·02], p=0·1; figure 5). By contrast, in the secondary prevention trials, aspirin seemed to reduce vascular mortality (RR 0·91 [0·82–1·00], p=0·06) and had no significant effect on other mortality (RR 0·85 [0·66–1·08], p=0·2), yielding a 10% reduction in total mortality (RR 0·90 [0·82–0·99], p=0·02).

0·95 [95% CI 0·88–1·02], p=0·1; figure 5). By contrast, in the secondary prevention trials, aspirin seemed to reduce vascular mortality (RR 0·91 [0·82–1·00], p=0·06) and had no significant effect on other mortality (RR 0·85 [0·66–1·08], p=0·2), yielding a 10% reduction in total mortality (RR 0·90 [0·82–0·99], p=0·02). The main hazard of aspirin is haemorrhage and, apart from any effect on intracerebral haemorrhage, aspirin increased major gastrointestinal and other extracranial bleeds by about half in the primary prevention trials (0·10% vs 0·07% per year; RR 1·54 [1·30–1·82], p<0·0001; table 2 and webappendix pp 9, 10). This increase was non-significantly greater in participants with high cholesterol (p=0·02 for trend if considered in isolation, but p=0·22 if multiplied by 11; webappendix p 10). The excess risk was chiefly of non-fatal bleeds; perhaps by chance, there were actually fewer fatal bleeds in participants allocated aspirin than in controls (nine vs 20; figure 5). Major bleeds were recorded in only five of the 16 secondary prevention trials, so a meta-analysis might be unreliable. There was again, however, a significant excess of major bleeds among those allocated aspirin (RR 2·69 [1·25–5·76], p=0·01; table 2 and webappendix p 20), with no significant heterogeneity between the relative risks in the six primary and these five secondary prevention trials (p=0·2; table 2).

ysis might be unreliable. There was again, however, a significant excess of major bleeds among those allocated aspirin (RR 2·69 [1·25–5·76], p=0·01; table 2 and webappendix p 20), with no significant heterogeneity between the relative risks in the six primary and these five secondary prevention trials (p=0·2; table 2). The absolute yearly incidence of vascular events and of major extracranial bleeds varied substantially among participants in the primary prevention trials. Poisson regression in 93 918 individuals without known vascular disease at entry within the primary prevention trials indicated that age (per decade), male sex, diabetes, current smoking, and mean blood pressure (per 20 mm Hg) were each associated with about a two-fold increased risk of major coronary events, whereas total cholesterol (per 1 mmol/L) and body-mass index (per 5 kg/m2) were more weakly associated with such events (table 3). Measurements of cholesterol fractions were not sought, and body-mass index acts mainly as a determinant of other cardiac risk factors15 so is of little independent relevance in these multivariate analyses. The main risk factors for coronary events in table 3 were also associated with haemorrhagic events, although for most the associations were slightly weaker for bleeding than for occlusive events.

ainly as a determinant of other cardiac risk factors15 so is of little independent relevance in these multivariate analyses. The main risk factors for coronary events in table 3 were also associated with haemorrhagic events, although for most the associations were slightly weaker for bleeding than for occlusive events. Discussion Previous meta-analyses have shown that aspirin is of substantial net benefit in secondary prevention,1,2 but the balance of beneficial effects and bleeding hazards in primary prevention was less clear. The availability of individual participant data for the present meta-analysis has allowed more reliable comparison of the benefits and hazards of aspirin in apparently healthy people. All four of the proportional reductions in major coronary events and in ischaemic stroke in the primary and in the secondary prevention trials were similar to each other (figure 3). Vascular mortality was not significantly reduced in the primary prevention trials, although a proportional reduction comparable with that in the secondary prevention trials could not be excluded. Whether or not the proportional benefits are similar, however, the absolute benefits of aspirin are an order of magnitude smaller in the primary than in the secondary prevention trials (table 2). (We have ignored the hypothesis16,17 of an eventual reduction in cancer mortality, since it would be expected to have little effect on our analyses of mortality during the scheduled treatment period.)

benefits of aspirin are an order of magnitude smaller in the primary than in the secondary prevention trials (table 2). (We have ignored the hypothesis16,17 of an eventual reduction in cancer mortality, since it would be expected to have little effect on our analyses of mortality during the scheduled treatment period.) In the primary prevention trials, the proportional reduction in serious vascular events did not depend significantly on age or sex (and the suggestion, on the basis of the primary prevention trials,18 that the proportional reductions in particular vascular outcomes might differ between men and women was not supported by the secondary prevention trials; figure 3). Nor did it depend significantly on smoking history, blood pressure, total cholesterol, body-mass index, history of diabetes, or predicted risk of coronary heart disease. In particular, there was no significant trend in the proportional effects of aspirin in people at very low, low, moderate, and high estimated risk of coronary heart disease. If the proportional risk reductions in these different subgroups really are similar, then the absolute risk reductions will depend chiefly on an individual's absolute risk without treatment.

d in the proportional effects of aspirin in people at very low, low, moderate, and high estimated risk of coronary heart disease. If the proportional risk reductions in these different subgroups really are similar, then the absolute risk reductions will depend chiefly on an individual's absolute risk without treatment. Figures 6 and 7 provide hypothetical calculations of what the absolute effects of aspirin allocation on 5-year outcome would be (in the absence of non-vascular causes of death) if the yearly event rates and the proportional effects of aspirin were as in the primary and as in the secondary prevention trials, and if these proportional effects were independent of age, sex, and other risk factors. Long-term low-dose aspirin had significant effects on both fatal and non-fatal events in people who already had occlusive vascular disease (table 2). Figure 6 suggests that, in this secondary prevention setting, aspirin would be of substantial net benefit (irrespective of age or sex), that it would reduce non-fatal vascular events by much more than it would increase major extracranial bleeds, and that—despite any adverse effects on cerebral haemorrhage—it would reduce overall vascular mortality (a result that is strongly reinforced by meta-analyses of all of the trials of any antiplatelet regimen in secondary prevention1,2). Nowadays, however, many patients with a history of occlusive stroke or myocardial infarction would have their risks of recurrence reduced substantially by statins, other modern drugs, and, when appropriate, vascular procedures. For occlusive vascular events, the relative risks produced by these other interventions and by aspirin might well be approximately multiplicative. If so, and if the other interventions approximately halve the recurrence risk, then the absolute benefit of adding aspirin to these other methods might be only about half as great as that of giving aspirin alone. Still, figure 6 suggests that, for secondary prevention, the net benefits of adding aspirin would substantially exceed the bleeding hazards, irrespective of age or sex.

Actual numbers for aspirin-allocated trial participants, and adjusted numbers for control-allocated trial participants, are presented, together with the corresponding mean yearly event rates (in parentheses). Rate ratios (RRs) for all trials are indicated by squares and their 99% CIs by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. A global test for heterogeneity (χ2 on 11 degrees of freedom) is provided. Unknown values are not plotted. SBP=systolic blood pressure. DBP=diastolic blood pressure. BMI=body-mass index. CHD=coronary heart disease. *Excluding patients with a history of vascular disease. Figure 3 Selected outcomes in primary and secondary prevention trials of aspirin, by sex Actual numbers for aspirin-allocated trial participants, and adjusted numbers for control-allocated trial participants, are presented together with the corresponding mean yearly event rate (in parentheses). Rate ratios (RRs) for all trials are indicated by squares and their 99% CIs by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. *Myocardial infarction, stroke (haemorrhagic or other), or vascular death. Figure 4 Stroke subtypes in primary and secondary prevention trials

recurrence risk, then the absolute benefit of adding aspirin to these other methods might be only about half as great as that of giving aspirin alone. Still, figure 6 suggests that, for secondary prevention, the net benefits of adding aspirin would substantially exceed the bleeding hazards, irrespective of age or sex. In the primary prevention trials, however, the absolute risk of a serious vascular event among people of a given age and sex was an order of magnitude less than in the secondary prevention trials. Figure 6 suggests that (irrespective of age or sex) the absolute reduction in occlusive events would be only about twice as large as the absolute increase in bleeding. Moreover, these trials of aspirin were mainly in people who were not taking statin therapy, which would have reduced both myocardial infarction and ischaemic stroke with little hazard.19,20 Generic statins are now widely available at low cost and, because of their efficacy and safety, primary prevention by a statin could well be preferred to primary prevention only by aspirin. If so, then one of the main questions for aspirin in primary prevention nowadays is whether it is worthwhile to add it to a statin (or to some statin-based combination of measures). If the risk of occlusive vascular disease is already approximately halved by statins or other measures, then the further absolute benefit of adding aspirin could well be only about half as large as was suggested by these primary prevention trials, but the main bleeding hazards could well remain. In that case, the benefits and hazards of adding long-term aspirin in people without pre-existing disease might be of approximately similar magnitude.

te benefit of adding aspirin could well be only about half as large as was suggested by these primary prevention trials, but the main bleeding hazards could well remain. In that case, the benefits and hazards of adding long-term aspirin in people without pre-existing disease might be of approximately similar magnitude. There is, of course, still the possibility that the primary prevention trials have, by chance, somewhat underestimated the main effects on mortality in the populations they studied (as is shown by the CIs in figure 5). There is also the possibility that some particular category of individuals will eventually be identified in which primary prevention with aspirin is of definite net benefit. One particularly important such category might be adults with diabetes but no known vascular disease, for whom aspirin is at present recommended.21 Although the evidence from the six primary prevention trials reviewed here is consistent with some net benefit in such patients (figure 2), the evidence from three other primary prevention trials in diabetes has been unpromising.22–24 Two much larger trials are, however, now recruiting only patients with diabetes.25,26 More generally only 9% of participants in the six primary prevention trials had predicted coronary heart disease incidence rates above 1% per year, so the present results among them are not, on their own, reliable (figure 2). Two major new trials in such moderately high-risk individuals are, however, now being undertaken,27,28 which will eventually yield more reliable evidence.

evention trials had predicted coronary heart disease incidence rates above 1% per year, so the present results among them are not, on their own, reliable (figure 2). Two major new trials in such moderately high-risk individuals are, however, now being undertaken,27,28 which will eventually yield more reliable evidence. To maximise the excess of benefit over hazard in primary prevention, most current guidelines3–5 recommend that aspirin be given to those with risk of coronary heart disease exceeding a particular threshold. These guidelines implicitly assume, however, either that the absolute risk of bleeding remains approximately constant irrespective of risk of coronary heart disease,4,5 or that it depends solely on age,3 whereas the present analyses showed that other risk factors for this disease are also risk factors for bleeding (table 3). As a result, even for people at moderately increased risk of coronary heart disease, the major absolute benefits and hazards of adding aspirin to a statin-based primary prevention regimen could still be approximately evenly balanced, as is suggested by the calculations in figure 7.

risk factors for bleeding (table 3). As a result, even for people at moderately increased risk of coronary heart disease, the major absolute benefits and hazards of adding aspirin to a statin-based primary prevention regimen could still be approximately evenly balanced, as is suggested by the calculations in figure 7. A non-fatal stroke or heart attack is more likely to result in long-term disability than is a non-fatal gastrointestinal (or other extracranial) bleed, but in primary prevention the net absolute reduction in disabling or fatal occlusive events is likely to be small, and at least partially offset by a small increase in serious intracranial and extracranial bleeds. Thus, although it might cost little to add aspirin to any other drugs that are being used for the primary prevention of vascular disease, the additional effectiveness against fatal or disabling outcomes has not been reliably demonstrated for men or women of any age who do not yet have any relevant disease (and, if effectiveness is uncertain then detailed estimates of cost-effectiveness29 are of limited relevance). Moreover, drug safety (like vaccine safety) is of particular importance in public health recommendations for large, apparently disease-free populations; there should be good evidence that benefits exceed risks by an appropriate margin. Hence, although the currently available trial results could well help inform personally appropriate judgments by individuals about their own use of long-term aspirin, they do not seem to justify general guidelines advocating the routine use of aspirin in all apparently healthy individuals above a moderate level of risk of coronary heart disease.3–8

available trial results could well help inform personally appropriate judgments by individuals about their own use of long-term aspirin, they do not seem to justify general guidelines advocating the routine use of aspirin in all apparently healthy individuals above a moderate level of risk of coronary heart disease.3–8 Web Extra Material Supplementary webappendix Acknowledgments We thank the trial participants and investigators. Sources of funding of each individual trial are described in its publications. The CTSU is supported by a core grant from the UK MRC, the BHF, and Cancer Research UK, and has previously received funding from the European Community Biomed Programme. C Baigent is supported by the MRC, R Collins by a BHF Personal Chair, and J Emberson by a BHF Intermediate Research Fellowship. This paper is dedicated to the memory of Richard Doll (1912–2005), in collaboration with whom the first primary prevention trial10 was undertaken. Contributors All members of the writing committee contributed to the collection or analysis of the data, or both, to the interpretation of the results, and to the preparation of the report.

Acknowledgments We thank the trial participants and investigators. Sources of funding of each individual trial are described in its publications. The CTSU is supported by a core grant from the UK MRC, the BHF, and Cancer Research UK, and has previously received funding from the European Community Biomed Programme. C Baigent is supported by the MRC, R Collins by a BHF Personal Chair, and J Emberson by a BHF Intermediate Research Fellowship. This paper is dedicated to the memory of Richard Doll (1912–2005), in collaboration with whom the first primary prevention trial10 was undertaken. Contributors All members of the writing committee contributed to the collection or analysis of the data, or both, to the interpretation of the results, and to the preparation of the report. Primary prevention working group of the ATT Collaboration British Doctors Study—Rory Collins, Richard Peto, Charles Hennekens, Richard Doll (deceased). US Physicians Study—Vadim Bubes, Julie Buring, Rimma Dushkesas, Michael Gaziano, Charles Hennekens. Thrombosis Prevention Trial—Patrick Brennan, Tom Meade, Alicja Rudnicka. Hypertension Optimal Treatment Study—Lennart Hansson (deceased), Ingrid Warnold (AstraZeneca), Alberto Zanchetti. Primary Prevention Project—Fausto Avanzini, Maria Carla Roncaglioni, Gianni Tognoni. Women's Health Study—Julie Buring, Marilyn Chown, Michael Gaziano, Charles Hennekens. Secretariat (CTSU)—Colin Baigent, Ian Barton, Alex Baxter, Neeraj Bhala, Lisa Blackwell, Jill Boreham, Louise Bowman, Georgina Buck, Rory Collins, Jonathan Emberson, Jon Godwin, Heather Halls, Lisa Holland, Patricia Kearney, Richard Peto, Christina Reith, Kate Wilson. Writing Committee—Colin Baigent, Lisa Blackwell, Rory Collins, Jonathan Emberson, Jon Godwin, Richard Peto (CTSU, Oxford University, Oxford, UK); Julie Buring (Brigham and Women's Hospital, Harvard University, Boston, MA, USA); Charles Hennekens (Charles E Schmidt College of Biomedical Science and Center of Excellence, Florida Atlantic University, Boca Raton, FL, USA); Patricia Kearney (Department of Medical Gerontology, Trinity College Dublin, Dublin, Ireland); Tom Meade (London School of Hygiene and Tropical Medicine, London University, London, UK); Carlo Patrono (Catholic University School of Medicine, Rome, Italy); Maria Carla Roncaglioni (Mario Negri Institute, Milan, Italy); and Alberto Zanchetti (Istituto Auxologico Italiano, University of Milan, Milan, Italy).

reland); Tom Meade (London School of Hygiene and Tropical Medicine, London University, London, UK); Carlo Patrono (Catholic University School of Medicine, Rome, Italy); Maria Carla Roncaglioni (Mario Negri Institute, Milan, Italy); and Alberto Zanchetti (Istituto Auxologico Italiano, University of Milan, Milan, Italy). Conflicts of interest The Clinical Trial Service Unit and Epidemiological Studies Unit (CTSU), where the ATT secretariat is located, has a policy of staff not accepting fees, honoraria, or paid consultancies. The CTSU continues, however, to be involved in clinical trials of cholesterol modification therapy and of antiplatelet therapy with funding from the Medical Research Council (MRC), British Heart Foundation (BHF), and/or various companies (Bayer, Merck, Merck Schering Plough, Solvay and, for the 1978 study of aspirin in British Doctors, the Aspirin Foundation) as research grants to (and administered by) Oxford University. J Buring received grant support in the form of pills and packaging, as well as one speaker's honorarium from Bayer. T Meade and C Patrono received consultancy or speaker fees, grant support, or both, from Bayer. C Hennekens receives investigator-initiated research grant support from Bayer and serves as an independent scientist on the Data and Safety Monitoring Board for the ARRIVE trial. Figure 1 Serious vascular events in primary prevention trials—proportional effects of aspirin allocation

Conflicts of interest The Clinical Trial Service Unit and Epidemiological Studies Unit (CTSU), where the ATT secretariat is located, has a policy of staff not accepting fees, honoraria, or paid consultancies. The CTSU continues, however, to be involved in clinical trials of cholesterol modification therapy and of antiplatelet therapy with funding from the Medical Research Council (MRC), British Heart Foundation (BHF), and/or various companies (Bayer, Merck, Merck Schering Plough, Solvay and, for the 1978 study of aspirin in British Doctors, the Aspirin Foundation) as research grants to (and administered by) Oxford University. J Buring received grant support in the form of pills and packaging, as well as one speaker's honorarium from Bayer. T Meade and C Patrono received consultancy or speaker fees, grant support, or both, from Bayer. C Hennekens receives investigator-initiated research grant support from Bayer and serves as an independent scientist on the Data and Safety Monitoring Board for the ARRIVE trial. Figure 1 Serious vascular events in primary prevention trials—proportional effects of aspirin allocation Actual numbers for aspirin-allocated trial participants, and adjusted numbers for control-allocated trial participants, are presented, together with the corresponding mean yearly event rate (in parentheses). Participants can contribute only once to the total of serious vascular events. Rate ratios (RRs) for all trials are indicated by squares and their 99% CIs by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. MI=myocardial infarction. CHD=coronary heart disease. *Myocardial infarction, stroke, or vascular death. Vascular death is coronary heart disease death, stroke death, or other vascular death (which includes sudden death, death from pulmonary embolism, and death from any haemorrhage, but in the primary prevention trials excludes death from an unknown cause).

Actual numbers for aspirin-allocated trial participants, and adjusted numbers for control-allocated trial participants, are presented together with the corresponding mean yearly event rate (in parentheses). Rate ratios (RRs) for all trials are indicated by squares and their 99% CIs by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. *Myocardial infarction, stroke (haemorrhagic or other), or vascular death. Figure 4 Stroke subtypes in primary and secondary prevention trials Actual numbers for aspirin-allocated trial participants, and adjusted numbers for control-allocated trial participants, are presented. Rate ratios (RRs) for all trials are indicated by squares and their 99% CIs by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. *Haemorrhagic, ischaemic, or unknown cause. Figure 5 Mortality by cause in primary prevention trials Actual numbers for aspirin-allocated trial participants, and adjusted numbers for control-allocated trial participants, are presented together with the corresponding mean yearly event rate (in parentheses). Rate ratios (RRs) for all trials are indicated by squares and their 99% CIs by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. CHD=coronary heart disease. GI=gastointestinal.

esponding mean yearly event rate (in parentheses). Rate ratios (RRs) for all trials are indicated by squares and their 99% CIs by horizontal lines. Subtotals and their 95% CIs are represented by diamonds. Squares or diamonds to the left of the solid line indicate benefit. CHD=coronary heart disease. GI=gastointestinal. Figure 6 Predicted 5-year absolute effects of allocation to aspirin in different categories of age and sex in the primary and secondary prevention trials (ignoring non-vascular mortality) Results are generally for otherwise untreated individuals; other risk reduction measures might approximately halve the vascular event rates in both aspirin (A) and control (C) groups. Three outcomes were analysed: non-fatal gastrointestinal (GI) (or other non-cerebral) bleeds in the primary prevention trials only; non-fatal vascular events in the primary trials and in the secondary trials; and vascular mortality (including any fatal bleeds) in the primary trials and in the secondary trials. For every outcome, the overall risk ratio (aspirin vs control in all participants, irrespective of age or sex) was combined with the absolute yearly risk among the controls in these four categories of sex and age. The risk ratios are those resulting from allocation to daily aspirin, so they underestimate the effects of actually taking aspirin for the whole 5-year period. MI=myocardial infarction.

ts, irrespective of age or sex) was combined with the absolute yearly risk among the controls in these four categories of sex and age. The risk ratios are those resulting from allocation to daily aspirin, so they underestimate the effects of actually taking aspirin for the whole 5-year period. MI=myocardial infarction. Figure 7 Predicted 5-year absolute effects of allocation to aspirin in the primary prevention trials in different categories of 5-year risk (if untreated) of coronary heart disease (CHD) (ignoring non-vascular mortality) Three outcomes were analysed in aspirin (A) and control (C) groups: non-fatal gastrointestinal (GI) (or other non-cerebral) bleeds when aspirin is given alone; non-fatal vascular events when aspirin is given alone and when aspirin is added to other drugs that halve risk; and vascular mortality (including any fatal bleeds) when aspirin is given alone and when aspirin is added to other drugs that halve risk. For every outcome, the overall risk ratio, irrespective of risk of coronary heart disease, was combined with the absolute yearly risk among the controls in three categories of predicted 5-year risk of a major coronary event (<5%, 5–10%, >10%). Absolute effects are estimated both directly from the data (middle column) and in the hypothetical situation in which risk is halved by statins and other primary prevention measures (right-hand column). Table 1 Design and eligibility criteria of primary prevention trials

Three outcomes were analysed in aspirin (A) and control (C) groups: non-fatal gastrointestinal (GI) (or other non-cerebral) bleeds when aspirin is given alone; non-fatal vascular events when aspirin is given alone and when aspirin is added to other drugs that halve risk; and vascular mortality (including any fatal bleeds) when aspirin is given alone and when aspirin is added to other drugs that halve risk. For every outcome, the overall risk ratio, irrespective of risk of coronary heart disease, was combined with the absolute yearly risk among the controls in three categories of predicted 5-year risk of a major coronary event (<5%, 5–10%, >10%). Absolute effects are estimated both directly from the data (middle column) and in the hypothetical situation in which risk is halved by statins and other primary prevention measures (right-hand column). Table 1 Design and eligibility criteria of primary prevention trials Dates of recruitment Participating countries Year of main publication Number of participants Mean duration of follow-up (years) Target population Eligible age range (years) at entry Aspirin regimen Randomised factorial comparison Placebo control British Doctors' Study10 Nov 1978–Nov 1979 UK 1988 5139 5·6 Male doctors 19–90 500 mg daily None No US Physicians' Health Study11 Aug 1981–Apr 1984 USA 1988 22071 5·0 Male doctors 45–73 325 mg alternate days β carotene vs placebo Yes Thrombosis Prevention Trial9 Feb 1989–May 1994 UK 1998 5085 6·7 Men with risk factors for CHD 45–69 75 mg daily Warfarin vs placebo Yes Hypertension Optimal Treatment Trial12 Oct 1992–May 1994 Europe, North and South America, Asia 1998 18790 3·8 Men and women with DBP 100–115 mm Hg 50–80 75 mg daily Three blood pressure regimens Yes Primary Prevention Project13 June 1993–Apr 1998 Italy 2001 4495 3·7 Men and women with one or more risk factors for CHD 45–94 100 mg daily Vitamin E vs open control No Women's Health Study14 Sep 1992–May 1995 USA 2005 39876 10·0 Female health professionals ≥45 100 mg alternate days Vitamin E vs placebo Yes CHD=coronary heart disease. DBP=diastolic blood pressure.

98 Italy 2001 4495 3·7 Men and women with one or more risk factors for CHD 45–94 100 mg daily Vitamin E vs open control No Women's Health Study14 Sep 1992–May 1995 USA 2005 39876 10·0 Female health professionals ≥45 100 mg alternate days Vitamin E vs placebo Yes CHD=coronary heart disease. DBP=diastolic blood pressure. Table 2 Comparison of proportional and absolute effects of aspirin in primary and secondary prevention trials

98 Italy 2001 4495 3·7 Men and women with one or more risk factors for CHD 45–94 100 mg daily Vitamin E vs open control No Women's Health Study14 Sep 1992–May 1995 USA 2005 39876 10·0 Female health professionals ≥45 100 mg alternate days Vitamin E vs placebo Yes CHD=coronary heart disease. DBP=diastolic blood pressure. Table 2 Comparison of proportional and absolute effects of aspirin in primary and secondary prevention trials Number of events (aspirin vs control) Rate ratio (95% CI) (aspirin vs control) Yearly absolute difference (% per year) Primary prevention (660 000 person-years) Secondary prevention (43 000 person-years) Primary prevention Secondary prevention p value for heterogeneity Primary prevention Secondary prevention Major coronary event 934 vs 1115 995 vs 1214 0·82 (0·75–0·90) 0·80 (0·73–0·88) 0·7 −0·06 −1·00* Non-fatal MI 596 vs 756 357 vs 505 0·77 (0·69–0·86) 0·69 (0·60–0·80) 0·5 −0·05 −0·66 CHD mortality 372 vs 393 614 vs 696 0·95 (0·82–1·10) 0·87 (0·78–0·98) 0·4 −0·01 −0·34 Stroke 655 vs 682 480 vs 580 0·95 (0·85–1·06) 0·81 (0·71–0·92) 0·1 −0·01 −0·46* Haemorrhagic 116 vs 89 36 vs 19 1·32 (1·00–1·75) 1·67 (0·97–2·90) 0·4 0·01 ..† Ischaemic 317 vs 367 140 vs 176 0·86 (0·74–1·00) 0·78 (0·61–0·99) 0·5 −0·02 ..† Unknown cause 222 vs 226 304 vs 385 0·97 (0·80–1·18) 0·77 (0·66–0·91) 0·1 −0·001 ..† Vascular death 619 vs 637 825 vs 896 0·97 (0·87–1·09) 0·91 (0·82–1·00) 0·4 −0·01 −0·29 Any serious vascular event 1671 vs 1883 (0·51% vs 0·57% per year) 1505 vs 1801 (6·69% vs 8·19% per year) 0·88 (0·82–0·94) 0·81 (0·75–0·87) 0·1 −0·07 −1·49* Major extracranial bleed 335 vs 219 23 vs 6 1·54 (1·30–1·82) 2·69 (1·25–5·76) 0·2 0·03 ..† MI=myocardial infarction. CHD=coronary heart disease. Non-fatal MI definitions vary; see methods.

nt 1671 vs 1883 (0·51% vs 0·57% per year) 1505 vs 1801 (6·69% vs 8·19% per year) 0·88 (0·82–0·94) 0·81 (0·75–0·87) 0·1 −0·07 −1·49* Major extracranial bleed 335 vs 219 23 vs 6 1·54 (1·30–1·82) 2·69 (1·25–5·76) 0·2 0·03 ..† MI=myocardial infarction. CHD=coronary heart disease. Non-fatal MI definitions vary; see methods. * Major coronary event rates (percent per year, aspirin vs control) 6·0 vs 7·4 in post-MI trials and 2·4 vs 3·0 in post-cerebral vascular disease trials; corresponding rates of stroke (mainly of unknown cause) 0·6 vs 0·8 in post-MI trials and 3·9 vs 4·7 in post-cerebral vascular disease trials (webappendix pp 14–18). † Stroke causes, and extracranial bleeds, very incompletely reported. Table 3 Rate ratios (95% CI) associated with risk factors for selected outcomes in people with no known vascular disease in primary prevention trials

* Major coronary event rates (percent per year, aspirin vs control) 6·0 vs 7·4 in post-MI trials and 2·4 vs 3·0 in post-cerebral vascular disease trials; corresponding rates of stroke (mainly of unknown cause) 0·6 vs 0·8 in post-MI trials and 3·9 vs 4·7 in post-cerebral vascular disease trials (webappendix pp 14–18). † Stroke causes, and extracranial bleeds, very incompletely reported. Table 3 Rate ratios (95% CI) associated with risk factors for selected outcomes in people with no known vascular disease in primary prevention trials Major coronary event Probably ischaemic stroke Haemorrhagic stroke Major extracranial bleed Age (per decade) 1·84 (1·74–1·95) 2·46 (2·27–2·65) 1·59 (1·33–1·90) 2·15 (1·93–2·39) Male sex* 2·43 (1·94–3·04) 1·44 (1·14–1·82) 1·11 (0·52–2·34) 1·99 (1·45–2·73) Diabetes mellitus 2·66 (2·28–3·12) 2·06 (1·67–2·54) 1·74 (0·95–3·17) 1·55 (1·13–2·14) Current smoker 2·05 (1·85–2·28) 2·00 (1·72–2·31) 2·18 (1·57–3·02) 1·56 (1·25–1·94) Mean blood pressure (per 20 mm Hg)† 1·73 (1·59–1·89) 2·00 (1·77–2·26) 2·18 (1·65–2·87) 1·32 (1·09–1·58) Cholesterol (per 1 mmol/L) 1·18 (1·12–1·24) 1·02 (0·95–1·09) 0·90 (0·77–1·07) 0·99 (0·90–1·08) Body-mass index (per 5 kg/m2) 1·09 (1·03–1·15) 1·06 (0·98–1·14) 0·85 (0·71–1·02) 1·24 (1·13–1·35) * Analyses are stratified by trial. The relevance of male sex can therefore be assessed only in the two trials that included both men and women, so the 95% CIs for it are wide, particularly for stroke. † Mean of systolic and diastolic blood pressure. Associations with measured values are not corrected for the effects of regression dilution.

Introduction By west European standards, Russian adults, particularly men, have a very high risk of premature death, which has fluctuated sharply in recent decades. At 2005 mortality rates, for example, only 7% of UK men but 37% of Russian men would die before the age of 55 years (appendix pp 3–5).1 Strong alcoholic drink, mainly vodka, is a major cause of the high risk of premature death in Russian adults.2–11 Since 2005, Russian consumption of spirits and male mortality before age 55 years both decreased by about a third (appendix pp 5–6), but are still substantial. The effects of alcohol on mortality need to be assessed both by large retrospective studies of people who have already died (in which information on previous consumption is sought retrospectively from informants who knew the dead person) and by large prospective studies. Both retrospective and, particularly, prospective studies are liable to under-estimate the individual risks and the alcohol-attributable population risks, but prospective studies avoid some of the potential biases of retrospective studies.

vely from informants who knew the dead person) and by large prospective studies. Both retrospective and, particularly, prospective studies are liable to under-estimate the individual risks and the alcohol-attributable population risks, but prospective studies avoid some of the potential biases of retrospective studies. A retrospective study of 50 000 deaths in three typical Russian cities (Barnaul, Byisk, and Tomsk) sought information from surviving family members about the drinking habits of the dead person.10 Such reports may well understate actual consumption, but even in those who had died of diseases that were deemed unlikely to be much related to alcohol use, 47% of the men and 11% of the women were reported to have drunk at least a bottle of vodka a week, and many were reported by their families to have drunk on average about a bottle of vodka a day. The proportions reported to have done so were, however, substantially larger than this in those who had died from external causes (accident, poisoning, suicide, or homicide) or eight particular disease groupings (cancer of the upper aerodigestive tract or liver, other liver disease, tuberculosis, pneumonia, acute pancreatitis, acute ischaemic heart disease other than myocardial infarction [ICD-10 I2412], and ill-specified conditions, appendix p 10), showing that mortality from these causes was substantially increased in drinkers.

ncer of the upper aerodigestive tract or liver, other liver disease, tuberculosis, pneumonia, acute pancreatitis, acute ischaemic heart disease other than myocardial infarction [ICD-10 I2412], and ill-specified conditions, appendix p 10), showing that mortality from these causes was substantially increased in drinkers. Both retrospective and prospective studies are limited by the unreliability of the information they obtain about alcohol use and by the variability of drinking patterns, but have complementary strengths. Retrospective studies can obtain information on large numbers of deaths relatively quickly, and approach the ideal of sampling all deaths (although informants may be unavailable, or misleading). Conversely, although heavy drinkers may well be under-represented in prospective studies, such studies have the great advantage that the information recorded at baseline about alcohol use cannot be distorted by the subsequent onset of disease. This allows unbiased comparison between the reported habits of those who do and do not die.

hough heavy drinkers may well be under-represented in prospective studies, such studies have the great advantage that the information recorded at baseline about alcohol use cannot be distorted by the subsequent onset of disease. This allows unbiased comparison between the reported habits of those who do and do not die. We report a large prospective study of self-reported alcohol (mainly vodka) consumption and mortality in the three Russian cities that we studied retrospectively. Since the number of deaths is smaller than that in the retrospective study, the analyses subdivide all-cause mortality into only two parts: those causes previously found to be alcohol-related (external causes and the eight disease groupings) and those not related to alcohol. The present analyses use the same age and vodka consumption categories as did the retrospective analyses, and hence test prospectively the retrospective findings;10 the two should be considered together.

usly found to be alcohol-related (external causes and the eight disease groupings) and those not related to alcohol. The present analyses use the same age and vodka consumption categories as did the retrospective analyses, and hence test prospectively the retrospective findings;10 the two should be considered together. Methods Study design and participants In this prospective study of mortality in 200 000 Russian adults, participants were recruited from three west Siberian industrial cities with predominantly European populations, and adult mortality rates similar to Russia as a whole:10 Barnaul (2002 population 0·7 million), Tomsk (0·5 million), and Byisk (0·2 million). Throughout 1999 (the first phase of recruitment, restricted to Barnaul and Tomsk), interviewers visited households randomly selected from electoral lists. During 2002–08 (the main phase of recruitment, including all three cities), interviewers visited households where an adult who had died several years ago (in 1990–2001) used to live, partly to ask surviving family members about the dead person for our retrospective study of alcohol use10 and partly to recruit adults still living there into the present prospective study.

ding all three cities), interviewers visited households where an adult who had died several years ago (in 1990–2001) used to live, partly to ask surviving family members about the dead person for our retrospective study of alcohol use10 and partly to recruit adults still living there into the present prospective study. Since almost all heavy drinkers were male smokers, our main analyses are of male smokers, although analyses of male non-smokers and women are also provided (appendix p 13). Results at ages 35–54 and 55–74 years are presented separately. We excluded people with no follow-up at 35–74 years, or with diseases that might alter drinking patterns (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, cirrhosis or chronic hepatitis), or who had stopped smoking or drinking due to illness (appendix pp 7–8). Ethics approval was obtained from the WHO IARC Ethical Review Committee, the Research Ethics Committee of the Moscow Russian Cancer Research Centre, and the local research ethics committees.

Since almost all heavy drinkers were male smokers, our main analyses are of male smokers, although analyses of male non-smokers and women are also provided (appendix p 13). Results at ages 35–54 and 55–74 years are presented separately. We excluded people with no follow-up at 35–74 years, or with diseases that might alter drinking patterns (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, cirrhosis or chronic hepatitis), or who had stopped smoking or drinking due to illness (appendix pp 7–8). Ethics approval was obtained from the WHO IARC Ethical Review Committee, the Research Ethics Committee of the Moscow Russian Cancer Research Centre, and the local research ethics committees. Procedures In both phases, teams of local general practitioners known to the local population (and hence generally trusted by households) and trained in objective interview methods did the interviews. All registered residents of visited households who were at least 30 years old and were present when the household was visited were invited to join the study. Those who accepted answered questions (about smoking, drinking, education, work, and their history of serious illness), and had their blood pressure, height, weight, waist and hip circumference measured. Ex-smokers and ex-drinkers were asked whether they had stopped because of illness.

d were invited to join the study. Those who accepted answered questions (about smoking, drinking, education, work, and their history of serious illness), and had their blood pressure, height, weight, waist and hip circumference measured. Ex-smokers and ex-drinkers were asked whether they had stopped because of illness. In 1999 the alcohol questions were only about current (ie, in the year before the baseline interview) consumption of vodka. In 2002–08, however, the questions were about both current and past consumption not only of vodka, but also of other strong alcoholic drinks, beer, and wine; the main source of alcohol was, however, vodka. Total consumption (frequency times amount, with beer taken as 0·125 and wine as 0·25 times vodka in strength) was described in units of half-litre bottles of vodka per week.10 The present analyses relate mortality only to current vodka consumption at baseline, and hence can include participants recruited in both phases.

otal consumption (frequency times amount, with beer taken as 0·125 and wine as 0·25 times vodka in strength) was described in units of half-litre bottles of vodka per week.10 The present analyses relate mortality only to current vodka consumption at baseline, and hence can include participants recruited in both phases. Vodka consumption (in half-litre bottles per week) was divided into low (never drinkers, ex-drinkers who had not quit because of illness, and men drinking less than 1 bottle per week or women drinking less than 0·25 of a bottle per week), middle (men drinking 1 to <3 bottles per week or women drinking 0·25 to <1 bottle per week) and high (men drinking ≥3 bottles per week or women drinking ≥1 bottle per week). The appendix includes translations of the 1999 and of the 2002–08 questionnaires (pp 14–20). We noted a strong digit preference in the self-reported number of bottles of vodka per week, so we did not analyse consumption as a continuous variable. Some 12 000 participants were interviewed twice (mainly because a participant could be interviewed both in the first and in the main phase). We used only the first interview in the prospective analyses of mortality; we used the second interview only to help assess the reproducibility of self-reported vodka consumption.

Some 12 000 participants were interviewed twice (mainly because a participant could be interviewed both in the first and in the main phase). We used only the first interview in the prospective analyses of mortality; we used the second interview only to help assess the reproducibility of self-reported vodka consumption. Long-term follow-up to Jan 1, 2010, was through local state mortality records, which include full name, address, date of birth, date of death and causes of death (underlying and proximal, in text and ICD-10 coded12) and are essentially complete. Record linkage was by a specially developed probabilistic algorithm, based on matching first name, father's name (traditionally used in Russia as an individual or personal identifier), second name, date of birth and address. It is unlikely that linkage on each of name, date of birth and address would be missed completely (and, a few missed linkages should not bias the findings appreciably). Where linkage was partial, additional checks by hand (occasionally including visits to the presumed decedent's address) were pursued until linkage was definitely confirmed or refuted. Causes of death were divided into those prespecified as alcohol-related (based on the retrospective study results) and those not; the appendix (p 10) gives the detailed ICD-10 codes.12

hand (occasionally including visits to the presumed decedent's address) were pursued until linkage was definitely confirmed or refuted. Causes of death were divided into those prespecified as alcohol-related (based on the retrospective study results) and those not; the appendix (p 10) gives the detailed ICD-10 codes.12 Statistical analysis For all-cause mortality and for the prespecified alcohol-related causes (with rates for other causes calculated as the difference), we calculated absolute death rates in the low, middle, and high alcohol groups in three stages. First, relative risks (RRs) were estimated by Poisson regression, adjusted for city, recruitment phase, attained age at risk (in 5-year bands), education (four levels), and cigarettes smoked per day (≤10, 11–19, 20, >20 cigarettes), all as categorical variables, with RR=1 for the high-alcohol group. Next, just for the high-alcohol group, we calculated in each 5-year age range the mortality rate as the number of deaths divided by person-years; we then defined the uniformly age-standardised rate in the age range 35–54 years (or, likewise, 55–74 years) as the average of the four age-specific rates within that age range. (If in a 20-year age range the uniformly age-standardised annual rate per 1000 persons is R, then exp[–20R/1000] is the conditional probability that someone who has survived to the start of that age range will survive to the end of it.) Finally, we multiplied these three RRs by the mortality rate in the high-alcohol group to obtain the mortality rates in each of the three alcohol groups, standardised to the pattern of risk factors in the high-alcohol group.

ty that someone who has survived to the start of that age range will survive to the end of it.) Finally, we multiplied these three RRs by the mortality rate in the high-alcohol group to obtain the mortality rates in each of the three alcohol groups, standardised to the pattern of risk factors in the high-alcohol group. We used Plummer's method13 to estimate the variance of the log risk in each of the three alcohol groups. From these variances we derived 95% CIs that describe for the log risk in each of the three groups, including the reference group, the effects of the play of chance in the data just for that group. If, for a group with death rate R, the variance of the log risk is v, then the 95% CI for R runs from R/k to R×k, where k=exp(1·96√v). (In the case of only three groups, Plummer's method simplifies: if the log relative risks in the low-alcohol and middle-alcohol groups have respective variances a and b and covariance c, then the log risks in the low, middle, and high groups have respective variances a–c, b–c and c.) To determine which diseases contributed most to the excess mortality among drinkers, we subcategorised underlying causes of death more finely, and for each subcategory we compared the numbers of deaths observed in the high-alcohol and middle-alcohol categories with the numbers expected at the age-specific death rates of the low-alcohol category (<1 bottle per week at baseline) in male smokers (appendix p 12). Analyses used SASv9.3; figure-plotting used Rv3.0.

To determine which diseases contributed most to the excess mortality among drinkers, we subcategorised underlying causes of death more finely, and for each subcategory we compared the numbers of deaths observed in the high-alcohol and middle-alcohol categories with the numbers expected at the age-specific death rates of the low-alcohol category (<1 bottle per week at baseline) in male smokers (appendix p 12). Analyses used SASv9.3; figure-plotting used Rv3.0. Role of funding sources The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding authors had full access to all the data in the study and had final responsibility for the decision to submit for publication. Results We obtained baseline interview data from 210 002 adults in 73 060 households. After exclusion of those too old or young to have any follow-up at ages 35–74 years, those with previous illness, and those who had stopped smoking or drinking because of illness (appendix pp 7–8), 151 811 remained for prospective follow-up of mortality.

baseline interview data from 210 002 adults in 73 060 households. After exclusion of those too old or young to have any follow-up at ages 35–74 years, those with previous illness, and those who had stopped smoking or drinking because of illness (appendix pp 7–8), 151 811 remained for prospective follow-up of mortality. Table 1 (also appendix p 9) shows the participants' characteristics, subdivided by their self-reported vodka consumption. Relatively few (compared with controls in our retrospective study)10 reported drinking one or more bottles per week of vodka: 19% (15 347 of 79 311) of the men and 2% (1209 of 72 500) of the women. Among them, intake of alcohol from other sources was only about a tenth that from vodka. Since intake of alcohol from other sources was much less than (and correlated with) intake from vodka, its consequences could not be studied. Although vodka consumption correlated with manual work and lack of education (table 1), its main correlate was smoking (figure 1); moreover, mean cigarette consumption per smoker was greater in the high-vodka than in the low-vodka group. Mean body-mass index (BMI) was slightly lower in the high- than in the low-vodka group, and in men (though not women) mean blood pressure was slightly higher (table 1). Our main mortality analyses do not adjust for these small differences. We rated cooperation with the baseline interview as having been good in about 90% of participants, largely irrespective of reported vodka consumption.

low-vodka group, and in men (though not women) mean blood pressure was slightly higher (table 1). Our main mortality analyses do not adjust for these small differences. We rated cooperation with the baseline interview as having been good in about 90% of participants, largely irrespective of reported vodka consumption. Self-reported vodka consumption can vary substantially over just a few years, increasing in some individuals and decreasing in others. Among the 6822 men who happened to be re-interviewed about 3 years after their baseline interview (table 2), 321 had reported drinking three or more bottles of vodka per week at baseline. When these heavy drinkers were re-interviewed, only 13% (41 of 321) still reported drinking three or more bottles per week, while 58% (185 of 321) reported drinking one bottle or less per week. Consequently, although their mean self-reported vodka consumption was 5·2 (SD 3·1) bottles per week at baseline, it was only 1·2 (SD 1·9) bottles per week at re-interview. Conversely, 14% (738 of 5435) of those who reported drinking less than one bottle per week at baseline reported drinking ≥1 bottle a week when re-interviewed about 3 years later. We noted similar patterns in women (appendix p 11).

bottles per week at baseline, it was only 1·2 (SD 1·9) bottles per week at re-interview. Conversely, 14% (738 of 5435) of those who reported drinking less than one bottle per week at baseline reported drinking ≥1 bottle a week when re-interviewed about 3 years later. We noted similar patterns in women (appendix p 11). Of the 151 811 participants with no previous illness who were followed up for mortality at ages 35–74 years, 5412 of 79 311 men (7%) and 2297 of 72 500 women (3%) died in this age range. Of them, 2807 men and 782 women died from external causes and the eight disease groupings that had been prespecified as alcohol-related (introduction and appendix p 10). Table 3 gives for these causes, other causes, and all causes the numbers of deaths and the age-standardised death rates, according to vodka consumption at baseline. Results, subdivided by age (35–54 or 55–74 years), are given separately for male smokers, male non-smokers, female smokers, and female non-smokers. Appendix p 13 gives similar analyses, but with male ex-smokers and never-smokers subdivided.

d the age-standardised death rates, according to vodka consumption at baseline. Results, subdivided by age (35–54 or 55–74 years), are given separately for male smokers, male non-smokers, female smokers, and female non-smokers. Appendix p 13 gives similar analyses, but with male ex-smokers and never-smokers subdivided. There were too few deaths to determine reliably whether a little vodka consumption provided a slight protective effect or a slight hazard. In each of the eight combinations of age, sex, and smoking status, however, we noted a striking excess overall mortality in those who had reported drinking one or more bottles of vodka per week, with particularly high rates in those who had reported drinking three or more bottles per week. This excess overall mortality was driven by a strong association with mortality from those causes that had been prespecified10 as alcohol-related. Since most people who reported drinking one or more bottle per week at baseline were male smokers (figure 1), it was only in the male smokers that our mortality analyses were statistically stable. Figure 2 plots mortality at ages 35–54 and 55–74 years in male smokers versus baseline vodka consumption, combining never drinkers, ex-drinkers, and those consuming less than one bottle per week. These absolute death rates were adjusted for amount smoked (and other factors: see Methods) to the values in the high-alcohol group. Hence, the absolute mortality rates in the high-alcohol group are the crude age-standardised rates actually observed in that group.

ers, and those consuming less than one bottle per week. These absolute death rates were adjusted for amount smoked (and other factors: see Methods) to the values in the high-alcohol group. Hence, the absolute mortality rates in the high-alcohol group are the crude age-standardised rates actually observed in that group. In both age ranges vodka consumption was strongly related to overall mortality, mainly because of its relation to the causes of death that had been prespecified as alcohol-related. For male smokers of age 35–54 years, the uniformly age-standardised annual death rate per 1000 men was 21·3 (95% CI 18·7–24·3) in the highest vodka consumption category, compared with 8·5 (CI 8·0–9·1) in the lowest category; for male smokers of age 55–74 years, the corresponding rates in the highest and lowest vodka consumption categories were 51·1 (CI 44·5–58·7) and 34·4 (CI 32·6–36·3). These uniformly age-standardised annual death rates translate into 20-year mortality risks for high, middle and low self-reported vodka consumption of 35% (31–39), 20% (18–22), and 16% (15–17) at ages 35–54 years (trend p<0·0001), and 64% (59–69), 54% (51–57) and 50% (48–52) at ages 55–74 years (trend p<0·0001).

36·3). These uniformly age-standardised annual death rates translate into 20-year mortality risks for high, middle and low self-reported vodka consumption of 35% (31–39), 20% (18–22), and 16% (15–17) at ages 35–54 years (trend p<0·0001), and 64% (59–69), 54% (51–57) and 50% (48–52) at ages 55–74 years (trend p<0·0001). Figure 2 relates risk over a period of several years to self-reported consumption on just one occasion (ie, at the baseline survey). Table 2 showed, however, that many who described themselves as being in the high alcohol consumption category did not continue to drink heavily, whereas some who described themselves as being in the low alcohol consumption category later drank appreciably greater amounts of vodka. If, therefore, it had been possible to compare people who continued to drink heavily with people who never did so, the high-vodka group would have had even higher risks, the low-vodka group would have had even lower risks, and the relationship with mortality would have been much steeper than in figure 2.

unts of vodka. If, therefore, it had been possible to compare people who continued to drink heavily with people who never did so, the high-vodka group would have had even higher risks, the low-vodka group would have had even lower risks, and the relationship with mortality would have been much steeper than in figure 2. The numbers of deaths from specific diseases and specific external causes were generally too small for similar analyses to yield statistically reliable absolute risks. Nevertheless, when underlying causes of death were subcategorised more finely (appendix p 12), the differences between the numbers observed among drinkers and the numbers expected at non-drinker death rates indicated that about half of the excess risk in heavy drinkers involved external causes, and that most of the remainder involved the somewhat strange ICD-10 category I24 (acute ischaemic heart disease that is not myocardial infarction). Discussion This prospective study of alcohol and mortality in Russia provides strong, unbiased confirmation of the already striking findings from smaller prospective studies,6,11 retrospective studies,8,10 autopsy studies,9 and national mortality trends2–5,7,9,10 that vodka (or other strong alcoholic drink) is a major cause of death in Russia (panel).

of alcohol and mortality in Russia provides strong, unbiased confirmation of the already striking findings from smaller prospective studies,6,11 retrospective studies,8,10 autopsy studies,9 and national mortality trends2–5,7,9,10 that vodka (or other strong alcoholic drink) is a major cause of death in Russia (panel). A retrospective study of 50 000 deaths in the same cities10 had found a marked excess of heavy vodka use in those whose death was attributed to external causes (accident, suicide, violence, and alcohol poisoning) or eight particular disease groupings (cancer of the upper aerodigestive tract, tuberculosis, pneumonia, liver cancer, other liver disease, pancreatic disease, acute ischaemic heart disease that is not myocardial infarction (I24), and ill-specified disease). Sharp fluctuations in Russian mortality rates from these causes during the 1990s were the main reason for the sudden large fluctuations in premature mortality in women and, particularly, men (figure 3; appendix pp 3–5).9,10 It is unclear what medical conditions predominate among the Russian deaths ascribed to the ICD-10 category I24 (used commonly in Russia but rarely in the UK), yet I24 dominates the high level and sharp fluctuations of Russian mortality from supposedly vascular causes.9 Many whose deaths are ascribed to it have high post-mortem alcohol concentrations, so it probably includes some deaths from acute effects of alcohol, perhaps superposed on chronic effects.9

rarely in the UK), yet I24 dominates the high level and sharp fluctuations of Russian mortality from supposedly vascular causes.9 Many whose deaths are ascribed to it have high post-mortem alcohol concentrations, so it probably includes some deaths from acute effects of alcohol, perhaps superposed on chronic effects.9 Retrospective studies of alcohol and mortality seek a family member or other informant who can describe the alcohol consumption of the dead person. They have the great advantage that they can begin with a reasonably representative sample of all deaths and can study large numbers of deaths quickly, but could be subject to two sources of bias. First, if the dead person had become socially isolated because of heavy drinking it might be difficult to find a suitable informant. Second, if the informant thought the person had died because of drink then this knowledge might bias their description of the dead person's drinking habits.

ubject to two sources of bias. First, if the dead person had become socially isolated because of heavy drinking it might be difficult to find a suitable informant. Second, if the informant thought the person had died because of drink then this knowledge might bias their description of the dead person's drinking habits. Prospective studies avoid these disadvantages, but take longer to observe large numbers of deaths (partly because those at high risk of death might not join such a study, and because, to avoid reverse causality, the analyses exclude those with prior disease). More than a decade since it began, our prospective study of 200 000 adults has accumulated only 8000 deaths at ages 35–74 years among the 151 000 participants with no previous disease at baseline and some follow-up in this age range. This is not enough to yield reliable dose-response results for each separate disease. To avoid unduly data-dependent groupings we therefore divided underlying causes into only two categories: those prespecified from the retrospective study findings10 as strongly alcohol-related, and all others.

ow-up in this age range. This is not enough to yield reliable dose-response results for each separate disease. To avoid unduly data-dependent groupings we therefore divided underlying causes into only two categories: those prespecified from the retrospective study findings10 as strongly alcohol-related, and all others. Likewise, we used the same alcohol consumption categories as the retrospective study.10 Adjustment for our measures of amount smoked, educational and social factors made little difference to the hazards associated with alcohol, so major residual confounding by such factors is unlikely, and adjustment for recruitment phase allows for any differences in questionnaire design (appendix pp 14–20) or in drinking prevalence. Alcohol intake from vodka (sold legally or illegally) predominated. Information was not sought on use of illicit drugs or on surrogate (eg, industrial or medicinal8) ethanol products that are not taxable as alcoholic drinks.

lows for any differences in questionnaire design (appendix pp 14–20) or in drinking prevalence. Alcohol intake from vodka (sold legally or illegally) predominated. Information was not sought on use of illicit drugs or on surrogate (eg, industrial or medicinal8) ethanol products that are not taxable as alcoholic drinks. One limitation of our prospective study is that most heavy drinkers were male smokers, so our main description of the effects of heavy drinking is restricted to male smokers (although such information as we have suggests that females and non-smokers would have similar absolute risks from drinking heavily if they were to do so). Another is that heavy drinkers are under-represented, so this study cannot directly estimate the alcohol-attributed fraction of overall mortality in Russia. (The proportions of men and women reportedly drinking at least one bottle of vodka a week were only 19% and 2% in the present study, as against 47% and 11% among the controls in our retrospective study in the same three cities.10)

cannot directly estimate the alcohol-attributed fraction of overall mortality in Russia. (The proportions of men and women reportedly drinking at least one bottle of vodka a week were only 19% and 2% in the present study, as against 47% and 11% among the controls in our retrospective study in the same three cities.10) The main limitation, however, is that self-described drinking habits are inaccurate, and in addition a resurvey showed that they can vary greatly over just a few years (as can Russian consumption of spirits,14,15 appendix p 6). Hence, even the extreme differences in overall mortality that we found (eg, figure 2, table 3) substantially under-estimate the real hazards of prolonged heavy consumption of vodka (or other strong alcoholic drinks) in Russia. Exact allowance for the effects of such inaccuracies is impossible, but even approximate allowance for them would make the relative risks for death from external causes or from the eight alcohol-related disease groupings substantially steeper than those in figure 2, and hence more closely consistent with the extreme relative risks noted in the retrospective study.10 Thus, our findings strongly reinforce the past and present findings from the Global Burden of Disease project16,17 about the importance of alcohol as a cause of death.

ings substantially steeper than those in figure 2, and hence more closely consistent with the extreme relative risks noted in the retrospective study.10 Thus, our findings strongly reinforce the past and present findings from the Global Burden of Disease project16,17 about the importance of alcohol as a cause of death. Nevertheless, even within the relatively low-risk circumstances of a prospective study, we found large differences in overall mortality between the top and bottom alcohol consumption categories that were driven mainly by the causes prespecified as alcohol-related. This provides strong confirmation, free from the disadvantages of retrospective mortality analyses, that alcohol, particularly vodka, is a major determinant of mortality from these causes, and hence of the high and sharply fluctuating Russian national mortality rates. This online publication has been corrected. The corrected version first appeared at thelancet.com on April 25, 2014 Supplementary Material Supplementary appendix Acknowledgments This study is funded by CTSU core support from the UK Medical Research Council, Cancer Research UK, and British Heart Foundation, by IARC collaborative research agreements (GEE/07/05 and GEP/08/07), and by EU research grants (ICA2-1999-10139 and ICA2-CT-2001-10002). We thank the fieldworkers for their help and the interviewees for their cooperation.

re support from the UK Medical Research Council, Cancer Research UK, and British Heart Foundation, by IARC collaborative research agreements (GEE/07/05 and GEP/08/07), and by EU research grants (ICA2-1999-10139 and ICA2-CT-2001-10002). We thank the fieldworkers for their help and the interviewees for their cooperation. Dedication This Article is dedicated to Gary Whitlock (1964–2013), who died after the final revision. An appropriate epidemiological memorial is his excellent and enjoyable personal website of 18 200 mortality trends in many different countries, http://www.mortality-trends.org/ (accessed Oct 18, 2013). Contributors DZ, RP, PBo, and PBr designed the study. DZ and AB coordinated its conduct. RK, AL, IK, VI, and TT coordinated the fieldwork. AB, JB, SL, GS, PS, and XK prepared the final database. SL, RP, PS, XK, and GW undertook the statistical analyses. SL, RP, DZ, and PBr drafted the paper, and all authors contributed to the final version. Conflicts of interest We declare that we have no conflicts of interest. Figure 1 Prevalence of current smoking (%) in 151 811 participants versus mean vodka use self-reported at baseline

Contributors DZ, RP, PBo, and PBr designed the study. DZ and AB coordinated its conduct. RK, AL, IK, VI, and TT coordinated the fieldwork. AB, JB, SL, GS, PS, and XK prepared the final database. SL, RP, PS, XK, and GW undertook the statistical analyses. SL, RP, DZ, and PBr drafted the paper, and all authors contributed to the final version. Conflicts of interest We declare that we have no conflicts of interest. Figure 1 Prevalence of current smoking (%) in 151 811 participants versus mean vodka use self-reported at baseline Excludes people with no follow-up at ages 35–74 years or with evidence at baseline of pre-existing disease (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, liver cirrhosis, or chronic hepatitis) or who had already quit drinking or smoking due to illness. *Current drinkers were subdivided into prespecified categories of vodka consumption (for men <1, 1 to <3, ≥3 bottles per week, for women <0·25, 0·25 to <1, ≥1 bottle per week), but results are plotted against total consumption of alcohol, vodka or other, in units of 200 mL per week, as in table 1. Figure 2 Mortality rates at ages 35–54 years and 55–74 years for causes prespecified as alcohol-related, other causes, and all causes, by vodka use self-reported at baseline in 57 361 male current smokers without previous disease

Excludes people with no follow-up at ages 35–74 years or with evidence at baseline of pre-existing disease (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, liver cirrhosis, or chronic hepatitis) or who had already quit drinking or smoking due to illness. *Current drinkers were subdivided into prespecified categories of vodka consumption (for men <1, 1 to <3, ≥3 bottles per week, for women <0·25, 0·25 to <1, ≥1 bottle per week), but results are plotted against total consumption of alcohol, vodka or other, in units of 200 mL per week, as in table 1. Figure 2 Mortality rates at ages 35–54 years and 55–74 years for causes prespecified as alcohol-related, other causes, and all causes, by vodka use self-reported at baseline in 57 361 male current smokers without previous disease The mortality rate in each 20-year age range is the mean of the four rates in the four 5-year age groups within that range, and the 20-year risk of death is conditional on reaching the start of the age range. Excludes people with no follow-up at ages 35–74 years or with evidence at baseline of pre-existing disease (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, liver cirrhosis, or chronic hepatitis) or who had already quit drinking or smoking because of illness. Causes prespecified as alcohol-related: external (includes assault, suicide, accident, alcohol poisoning, etc), liver diseases (neoplastic or not), upper aerodigestive cancer, tuberculosis, pneumonia, non-myocardial infarction acute ischaemic heart disease (ICD-10 I24), non-neoplastic pancreatic disease, and ill-defined causes. *Current drinkers were subdivided into prespecified categories of vodka consumption (for men <1, 1 to <3, ≥3 bottles per week), but results are plotted against total consumption of alcohol, vodka or other, in units of 200 mL per week, as in table 1.

CD-10 I24), non-neoplastic pancreatic disease, and ill-defined causes. *Current drinkers were subdivided into prespecified categories of vodka consumption (for men <1, 1 to <3, ≥3 bottles per week), but results are plotted against total consumption of alcohol, vodka or other, in units of 200 mL per week, as in table 1. Figure 3 All-cause mortality, males aged 15–54 years, in Russia from 1980–2012 and in the UK from 1980–2010 *Mean of male age-specific death rates in the eight component 5-year age groups (15–19 to 50–54 years). †Probability that a 15-year-old male individual would die before age 55 years, if exposed over next 40 years to male age-specific death rates of one particular calendar year. Table 1 Characteristics and overall mortality of the 151 811 participants, by sex and vodka use self-reported at baseline*

*Mean of male age-specific death rates in the eight component 5-year age groups (15–19 to 50–54 years). †Probability that a 15-year-old male individual would die before age 55 years, if exposed over next 40 years to male age-specific death rates of one particular calendar year. Table 1 Characteristics and overall mortality of the 151 811 participants, by sex and vodka use self-reported at baseline* Men Women <1 half-litre bottle of vodka per week† 1 to <3 half-litre bottles of vodka per week ≥3 half-litre bottles of vodka per week <0·25 of a half-litre bottle of vodka per week† 0·25 to <1 half-litre bottle of vodka per week ≥1 half-litre bottle of vodka per week Number of interviewees 63 964 12 505 2842 67 288 4003 1209 Mean (SD) vodka consumption, bottles per week 0·2 (0·2) 1·3 (0·4) 4·9 (2·5) 0·0 (0·1) 0·4 (0·1) 2·0 (1·8) Mean (SD) non-vodka consumption, units of 200 mL alcohol per week‡ 0·2 (0·3) 0·3 (0·5) 0·5 (0·9) 0·1 (0·2) 0·2 (0·2) 0·3 (0·5) Mean (SD) drinking frequency, days per week 0·8 (0·9) 1·8 (1·1) 4·2 (1·8) 0·3 (0·5) 1·1 (0·6) 2·5 (1·6) Mean (SD) age, years 46·6 (11·3) 46·5 (10·0) 47·3 (9·8) 48·4 (12·0) 45·7 (10·0) 46·6 (10·4) Never smoker 13 758 (21·5%) 1173 (9·4%) 198 (7·0%) 54 953 (81·7%) 2115 (52·8%) 415 (34·3%) Ex-smoker§ 6125 (9·6%) 594 (4·8%) 102 (3·6%) 2477 (3·7%) 241 (6·0%) 21 (1·7%) Current smoker 44 081 (68·9%) 10 738 (85·9%) 2542 (89·4%) 9858 (14·7%) 1647 (41·1%) 773 (63·9%) Mean (SD) number of cigarettes per day 16·4 (6·7) 19·1 (6·9) 21·3 (8·0) 9·3 (5·4) 11·2 (5·5) 13·4 (6·6) Mean (SD) BMI, kg/m2¶ 26·0 (3·4) 25·8 (3·5) 24·9 (3·5) 27·0 (4·6) 27·2 (4·6) 26·0 (5·0) Mean (SD) SBP, mmHg¶ 128 (15·4) 128 (16·0) 131 (17·4) 128 (19·2) 128 (18·4) 128 (19·4) Mean (SD) DBP, mmHg¶ 82 (8·9) 82 (9·3) 84 (10·3) 81 (10·9) 82 (10·8) 81 (11·7) No education beyond primary school 3965 (6·2%) 1030 (8·2%) 437 (15·4%) 5834 (8·7%) 258 (6·4%) 201 (16·6%) Manual worker 34 154 (53·4%) 8660 (69·3%) 1928 (67·8%) 15 723 (23·4%) 1348 (33·7%) 559 (46·2%) Good cooperation‖ 57 982 (90·6%) 10 769 (86·1%) 2638 (92·8%) 62 288 (92·6%) 3256 (81·3%) 1062 (87·8%) Mean (SD) person-years at ages 35–74 years 5·2 (3·0) 5·5 (2·8) 6·3 (3·1) 6·3 (3·3) 5·2 (2·6) 6·3 (2·8) Number of deaths at ages 35–74 years 3844 1052 516 2063 109 125 Data are n (%) unless otherwise specified. Mean consumption includes all drinks, and is expressed in units of 200 mL of pure alcohol per week (the approximate alcohol content of one bottle of vodka). BMI=body-mass index. SBP=systolic blood pressure. DBP=diastolic blood pressure.

s 35–74 years 3844 1052 516 2063 109 125 Data are n (%) unless otherwise specified. Mean consumption includes all drinks, and is expressed in units of 200 mL of pure alcohol per week (the approximate alcohol content of one bottle of vodka). BMI=body-mass index. SBP=systolic blood pressure. DBP=diastolic blood pressure. * Excludes people with no follow-up at ages 35–74 years or with evidence at baseline of pre-existing disease (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, liver cirrhosis, or chronic hepatitis) or who had already quit drinking or smoking due to illness. † Includes never-drinkers, ex-drinkers who did not quit because of illness, and low drinkers (<1 half-litre bottle of vodka per week for men or <0·25 of a half-litre bottle vodka per week for women): figure 1 separates never, ex and low· ‡ Phase 2 only (58 387 men, 41 738 women), no non-vodka consumption ascertained in phase 1. § Did not quit because of illness (see footnote *). ¶ Only 58 387 males and 41 738 females, as height, weight, and blood pressure were not measured during the first phase of recruitment. ‖ Good cooperation during interview, as assessed by interviewer. Table 2 Vodka use self-reported at baseline and at unintended re-interview (mean 3 years later), 6822 men*

§ Did not quit because of illness (see footnote *). ¶ Only 58 387 males and 41 738 females, as height, weight, and blood pressure were not measured during the first phase of recruitment. ‖ Good cooperation during interview, as assessed by interviewer. Table 2 Vodka use self-reported at baseline and at unintended re-interview (mean 3 years later), 6822 men* <1 half-litre bottle of vodka per week (n=5435) 1 to <3 half-litre bottles of vodka per week (n=1066) ≥3 half-litre bottles of vodka per week(n=321) Half-litre bottles of vodka per week reported at re-interview <1 4697 (86%) 727 (68%) 185 (58%) 1 to <3 653 (12%) 290 (27%) 95 (30%) ≥3 85 (2%) 49 (5%) 41 (13%) Mean (SD) vodka consumption reported at baseline, in units of 200 mL alcohol per week 0·1 (0·2) 1·3 (0·5) 5·2 (3·1) Mean (SD) vodka consumption reported at re-interview, in units of 200 mL alcohol per week 0·4 (0·7) 0·7 (1·1) 1·2 (1·9) Mean (SD) alcohol consumption reported at re-interview, in units of 200 mL alcohol per week 0·6 (0·9) 1·0 (1·1) 1·4 (2·1) Data are n (%). Groupings in this table are defined only by vodka consumption, but mean alcohol consumption includes all drinks, and is expressed in units of 200 mL of pure alcohol per week (the approximate alcohol content of one bottle of vodka).

rview, in units of 200 mL alcohol per week 0·6 (0·9) 1·0 (1·1) 1·4 (2·1) Data are n (%). Groupings in this table are defined only by vodka consumption, but mean alcohol consumption includes all drinks, and is expressed in units of 200 mL of pure alcohol per week (the approximate alcohol content of one bottle of vodka). * 6822 men were unintentionally interviewed twice, generally once during phase 1 and once during phase 2 of recruitment; this number excludes men with no follow-up at ages 35–74 years, or with evidence at baseline (ie, at the first interview) of pre-existing disease (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, liver cirrhosis, or chronic hepatitis), or who at baseline had already quit drinking or smoking due to illness. Table 3 Mortality from causes prespecified as alcohol-related, other causes, and all causes, by sex, smoking habit at baseline, age at risk, and vodka use self-reported at baseline among 151 811 participants*

* 6822 men were unintentionally interviewed twice, generally once during phase 1 and once during phase 2 of recruitment; this number excludes men with no follow-up at ages 35–74 years, or with evidence at baseline (ie, at the first interview) of pre-existing disease (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, liver cirrhosis, or chronic hepatitis), or who at baseline had already quit drinking or smoking due to illness. Table 3 Mortality from causes prespecified as alcohol-related, other causes, and all causes, by sex, smoking habit at baseline, age at risk, and vodka use self-reported at baseline among 151 811 participants* Prespecified as alcohol-related Other causes All causes Deaths Annual rate per 1000 (95% CI) Deaths Annual rate per 1000 (95% CI) Deaths Annual rate per 1000 (95% CI) Male smokers (n=57 361) Age at risk: 35–54 years Never-drinker 38 4·9 (3·5–6·7) 21 2·8 (1·8–4·3) 59 7·7 (5·9–10·0) Ex-drinker† 47 6·2 (4·6–8·2) 20 2·5 (1·6–3·9) 67 8·7 (6·8–11·0) <1 half-litre bottle of vodka per week 675 5·5 (5·1–6·0) 339 3·1 (2·7–3·4) 1014 8·5 (8·0–9·1) 1 to <3 half-litre bottles of vodka per week 313 8·1 (7·2–9·0) 107 3·0 (2·5–3·6) 420 11·0 (10·0–12·1) ≥3 half-litre bottles of vodka per week 197 17·3 (14·9–20·0) 46 4·1 (3·0–5·5) 243 21·3 (18·7–24·3) Age at risk: 55–74 years Never-drinker 30 11·0 (7·7–15·9) 76 25·6 (20·4–32·2) 106 36·7 (30·2–44·5) Ex-drinker† 34 14·5 (10·3–20·3) 68 26·2 (20·6–33·3) 102 40·6 (33·4–49·4) <1 half-litre bottle of vodka per week 595 14·0 (12·8–15·2) 874 19·9 (18·5–21·3) 1469 33·8 (32·1–35·7) 1 to <3 half-litre bottles of vodka per week 230 17·7 (15·5–20·1) 258 21·0 (18·6–23·7) 488 38·7 (35·4–42·3) ≥3 half-litre bottles of vodka per week 115 26·3 (21·8–31·8) 100 24·8 (20·3–30·3) 215 51·1 (44·5–58·7) Male non-smokers (15 129 never-smokers, 6821 ex-smokers†) Age at risk: 35–54 years Never-drinker 17 2·2 (1·4–3·6) 20 3·1 (2·0–4·9) 37 5·3 (3·8–7·4) Ex-drinker† 9 3·2 (1·7–6·2) 3 1·2 (0·4–3·6) 12 4·4 (2·5–7·7) <1 half-litre bottle of vodka per week 135 3·3 (2·8–3·9) 81 2·3 (1·8–2·9) 216 5·6 (4·9–6·4) 1 to <3 half-litre bottles of vodka per week 27 4·9 (3·3–7·2) 9 2·1 (1·1–4·1) 36 7·0 (5·1–9·8) ≥3 half-litre bottles of vodka per week 18 18·8 (11·7–30·3) 4 5·4 (2·0–14·4) 22 24·2 (15·8–37·1) Age at risk: 55–74 years Never-drinker 31 5·4 (3·8–7·7) 75 14·2 (11·2–17·8) 106 19·5 (16·1–23·7) Ex-drinker† 19 8·1 (5·1–12·8) 31 13·7 (9·6–19·6) 50 21·8 (16·5–28·9) <1 half-litre bottle of vodka per week 201 6·3 (5·5–7·2) 405 14·1 (12·8–15·6) 606 20·4 (18·8–22·1) 1 to <3 half-litre bottles of vodka per week 50 12·5 (9·5–16·6) 58 17·2 (13·3–22·3) 108 29·7 (24·6–36·0) ≥3 half-litre bottles of vodka per week 26 24·6 (16·6–36·3) 10 12·0 (6·4–22·4) 36 36·6 (26·3–50·8) Female smokers (n=12 278) Age at risk: 35–54 years Never-drinker 16 3·7 (2·3–6·1) 15 3·1 (1·8–5·2) 31 6·8 (4·8–

2·1) 1 to <3 half-litre bottles of vodka per week 50 12·5 (9·5–16·6) 58 17·2 (13·3–22·3) 108 29·7 (24·6–36·0) ≥3 half-litre bottles of vodka per week 26 24·6 (16·6–36·3) 10 12·0 (6·4–22·4) 36 36·6 (26·3–50·8) Female smokers (n=12 278) Age at risk: 35–54 years Never-drinker 16 3·7 (2·3–6·1) 15 3·1 (1·8–5·2) 31 6·8 (4·8– 9·8) Ex-drinker† 2 2·8 (0·7–11·1) 3 4·2 (1·3–13·1) 5 6·9 (2·9–16·8) <0·25 of a half-litre bottle of vodka per week 74 3·5 (2·7–4·4) 51 2·2 (1·6–3·0) 125 5·7 (4·7–6·9) 0·25 to <1 half-litre bottle of vodka per week 18 4·2 (2·6–6·7) 13 2·9 (1·7–5·1) 31 7·1 (5·0–10·2) ≥1 half-litre bottle of vodka per week 32 8·9 (6·1–12·8) 6 1·6 (0·7–3·6) 38 10·5 (7·5–14·6) Age at risk: 55–74 years Never-drinker 8 8·3 (4·0–17·1) 18 13·1 (8·1–21·1) 26 21·4 (14·3–31·9) Ex-drinker† 2 8·4 (2·1–34·1) 7 26·9 (12·5–57·8) 9 35·2 (18·1–68·8) <0·25 of a half-litre bottle of vodka per week 28 7·7 (5·2–11·3) 63 14·6 (11·2–18·9) 91 22·3 (17·9–27·6) 0·25 to <1 half-litre bottle of vodka per week 7 11·4 (5·3–24·5) 8 8·9 (4·4–17·9) 15 20·3 (12·1–34·0) ≥1 half-litre bottle of vodka per week 24 32·9 (21·1–51·1) 11 12·0 (6·5–22·1) 35 44·8 (31·6–63·7) Female non-smokers (57 481 never-smokers, 2739 ex-smokers†) Age at risk: 35–54 years Never-drinker 24 0·6 (0·4–0·9) 68 1·5 (1·2–1·9) 92 2·1 (1·7–2·6) Ex-drinker† 3 1·0 (0·3–3·0) 4 1·2 (0·4–3·2) 7 2·2 (1·0–4·6) <0·25 of a half-litre bottle of vodka per week 137 1·1 (0·9–1·3) 158 1·1 (0·9–1·3) 295 2·2 (1·9–2·5) 0·25 to <1 half-litre bottle of vodka per week 7 1·0 (0·5–2·2) 6 0·8 (0·3–1·7) 13 1·8 (1·0–3·1) ≥1 half-litre bottle of vodka per week 10 6·3 (3·4–11·9) 3 1·5 (0·5–4·8) 13 7·9 (4·6–13·6) Age at risk: 55–74 years Never-drinker 131 2·4 (2·0–2·9) 437 8·0 (7·2–8·8) 568 10·4 (9·5–11·3) Ex-drinker† 19 3·2 (2·0–5·1) 51 7·4 (5·6–9·8) 70 10·6 (8·3–13·4) <0·25 of a half-litre bottle of vodka per week 190 2·1 (1·8–2·5) 554 6·8 (6·3–7·4) 744 9·0 (8·3–9·6) 0·25 to <1 half-litre bottle of vodka per week 26 7·1 (4·8–10·6) 24 7·1 (4·7–10·6) 50 14·2 (10·8–18·8) ≥1 half-litre bottle of vodka per week 24 20·1 (13·4–30·2) 15 15·2 (9·1–25·2) 39 35·3 (25·8–48·5) Vodka use is given in half-litre bottles of vodka per week.

week 190 2·1 (1·8–2·5) 554 6·8 (6·3–7·4) 744 9·0 (8·3–9·6) 0·25 to <1 half-litre bottle of vodka per week 26 7·1 (4·8–10·6) 24 7·1 (4·7–10·6) 50 14·2 (10·8–18·8) ≥1 half-litre bottle of vodka per week 24 20·1 (13·4–30·2) 15 15·2 (9·1–25·2) 39 35·3 (25·8–48·5) Vodka use is given in half-litre bottles of vodka per week. * Excludes people with no follow-up at ages 35–74 years or with evidence at baseline of pre-existing disease (self-reported cancer, myocardial infarction, angina, heart failure, rheumatic heart disease, stroke, diabetes, tuberculosis, liver cirrhosis, or chronic hepatitis), or who had already quit drinking or smoking due to illness. † Did not quit due to illness. Panel Research in context Systematic review Alcohol, particularly vodka, is a major cause of premature death in Russia. A large retrospective study10 suggested its main effects are on mortality from external causes and eight particular disease groupings. We searched in PubMed using the search words (“Russia”) AND (“alcohol” OR “vodka”) AND (“death” OR “mortality”), for articles in English published before 18 October 2013, but found only retrospective studies and small prospective studies. The present large prospective study now confirms vodka use is strongly predictive of premature death, mainly from causes expected from the findings of the large retrospective study. Interpretation

Alcohol, particularly vodka, is a major cause of premature death in Russia. A large retrospective study10 suggested its main effects are on mortality from external causes and eight particular disease groupings. We searched in PubMed using the search words (“Russia”) AND (“alcohol” OR “vodka”) AND (“death” OR “mortality”), for articles in English published before 18 October 2013, but found only retrospective studies and small prospective studies. The present large prospective study now confirms vodka use is strongly predictive of premature death, mainly from causes expected from the findings of the large retrospective study. Interpretation Studies of national mortality trends, retrospective studies, and large prospective studies have complementary strengths, and with the present study now reinforce each other2–10 as evidence that the extraordinarily high risk of premature death among Russian adults, particularly men, is chiefly due to the use of vodka and other strong alcoholic drink. As almost all Russians who drink more than one bottle of vodka a week also smoke, any studies of the hazards of drinking have to be largely restricted to its effects among smokers, and any studies of the hazards of smoking will have to be largely restricted to its effects in those drinking less than one bottle of vodka a week.