Browse the corpus

er. Early AMD was defined as signs of drusen with a grid area greater than a 500-μm circle and/or pigmentary abnormalities, whereas the presence of exudative or geographic atrophy signs was defined as late AMD. If retinal images were available for both eyes, the eye with the more severe status was used in the analysis. Mortality Data Mortality data were derived from the 2015 public-access linked mortality archives. Mortality data were matched with files from the National Death Index via a probabilistic matching algorithm.33 A unique study identifier of the NHANES was used to link to the mortality data. All NHANES participants 18 years or older were available for mortality follow-up. Specific cause of death attributable to CVD was determined by codes I00 to I09, I11, I13, and I20 to I51 (diseases of heart) and I60 to I69 (cerebrovascular diseases) from the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10). Cancer mortality was defined by deaths based on ICD-10 codes C00 to C97. Those deaths not classified as CVD or cancer related were considered deaths due to other causes. Participants not matched with death certificates were considered alive. Time to death was counted from baseline to date of death or December 31, 2011, whichever came first.

Introduction Age-related macular degeneration (AMD) is the leading cause of irreversible visual impairment and blindness in the United States if subretinal neovascularization is left untreated.1 The estimated annual direct health care expenditure due to AMD in the United States is more than $4.6 billion.2 The global incidence of AMD is projected to increase, with an estimated 196 million patients with AMD by 2020.3 Despite the heavy burden of AMD, mechanisms underlying AMD remain poorly understood. Previous epidemiologic studies have identified some consistent risk factors (smoking) and systemic comorbidities, including cardiovascular disease (CVD)4,5,6,7,8,9,10,11; however, results relating to the association between AMD and survival are conflicting.12,13,14,15,16,17,18,19,20,21,22,23,24,25 Speculation has suggested that these inconsistent results arise from inadequate adjustment of important confounding factors, for example AMD-associated systemic comorbidities (CVD) that may lead to poorer survival.26 Moreover, most previous studies27 might overestimate the absolute risk of CVD mortality by failing to account for a competing risk of death. Given the current concern that injection of anti–vascular endothelial growth factor (anti-VEGF) may lead to increased risks of thromboembolic events,28,29 knowledge of the accurate influence of AMD on mortality risk, especially CVD mortality, is warranted.

mortality by failing to account for a competing risk of death. Given the current concern that injection of anti–vascular endothelial growth factor (anti-VEGF) may lead to increased risks of thromboembolic events,28,29 knowledge of the accurate influence of AMD on mortality risk, especially CVD mortality, is warranted. The National Health and Nutrition Examination Survey (NHANES) is an ongoing population-based study. This nationally representative sample of the noninstitutionalized US population provides an opportunity to investigate the association between AMD and all-cause and specific-cause mortality in the context of comprehensive demographic, health-related behaviors and comorbidities. Methods Sample and Population Led by the National Center for Health Statistics of the Centers for Disease Control and Prevention, Hyattsville, Maryland, the NHANES adopts stratified multistage sampling methods. Details of the sampling and testing methods have been described in detail elsewhere.30 Briefly, participants undergo comprehensive health-related interviews and examinations every 2 years. The NHANES purposely oversamples participants older than 60 years and Hispanic and African American individuals. In adherence to the tenets of the Declaration of Helsinki, NHANES protocols were approved by the National Center for Health Statistics research ethics review board, and participants provided written informed consent.

S purposely oversamples participants older than 60 years and Hispanic and African American individuals. In adherence to the tenets of the Declaration of Helsinki, NHANES protocols were approved by the National Center for Health Statistics research ethics review board, and participants provided written informed consent. Retinal Photography and AMD Grading During the 2005-2008 phase of NHANES, retinal images were collected among participants 40 years or older. An ophthalmic digital imaging system (CR6-45NM; Canon USA, Inc) and digital camera (EOS 10D; Canon USA, Inc) were used to capture retinal images. All fundus images were graded at the University of Wisconsin, Madison, according to the modified Wisconsin Age-Related Maculopathy Grading Classification Scheme.31,32 All images were graded by at least 2 experienced graders, with any disagreements adjudicated by a third senior grader. Early AMD was defined as signs of drusen with a grid area greater than a 500-μm circle and/or pigmentary abnormalities, whereas the presence of exudative or geographic atrophy signs was defined as late AMD. If retinal images were available for both eyes, the eye with the more severe status was used in the analysis.

efined by deaths based on ICD-10 codes C00 to C97. Those deaths not classified as CVD or cancer related were considered deaths due to other causes. Participants not matched with death certificates were considered alive. Time to death was counted from baseline to date of death or December 31, 2011, whichever came first. Covariates A broad array of information regarding demographic factors and health-related behaviors and characteristics was assessed through in-person interviews and examinations. Specifically, age was categorized by 10-year age groups as 40 to 49, 50 to 59, 60 to 69, 70 to 79, and 80 years or older. Race/ethnicity was categorized as non-Hispanic white, non-Hispanic black, Mexican American, or other. Educational attainment was dichotomized as less than a high school diploma and a high school diploma or more. Marital status (unmarried or other vs married or living with a partner) was analyzed as a 2-level categorical variable. The indicator for family income (poverty income ratio) was classified as below poverty line (<1.00) or at or above poverty line (≥1.00). Smoking status was categorized as never, former, or current. Alcohol consumption was determined from participant interviews and divided as lifetime abstainer or former drinker, current drinker consuming no more than 3 drinks per week, and current drinker consuming more than 3 drinks per week.

bove poverty line (≥1.00). Smoking status was categorized as never, former, or current. Alcohol consumption was determined from participant interviews and divided as lifetime abstainer or former drinker, current drinker consuming no more than 3 drinks per week, and current drinker consuming more than 3 drinks per week. Body mass index was calculated as weight in kilograms divided by height in meters squared and categorized as underweight (<18.5), normal to overweight (18.5-30.0), or obese (≥30.0). Diabetes was defined as self-reported physician diagnosis, use of diabetic tablets or insulin, or a glycosylated hemoglobin level of 6.5% or greater (to convert to a fraction of the total, multiply by 0.01). The presence of hypertension was characterized by self-reported history of hypertension, use of antihypertensive agents, or a systolic blood pressure of 140 mm Hg or higher and/or a diastolic blood pressure of 90 mm Hg or higher based on the mean value of 3 measurements. The presence of dyslipidemia was defined as a total cholesterol level of at least 240 mg/dL (to convert to millimoles per liter, multiply by 0.0259) or the use of a prescribed agent to lower cholesterol levels. As an indicator of vascular risk, the ratio of low-density to high-density lipoprotein cholesterol levels was calculated. A high level of C-reactive protein was defined as at least 1 mg/dL (to convert to nanomoles per liter, multiply by 9.524). A score of 10 or greater on the 9-item Patient Health Questionnaire (range, 0-27) was characterized as having depressive symptoms.34 The history of comorbid age-related ocular diseases, including cataract, glaucoma, and retinopathy, was based on the questionnaire and/or retinal images in accordance with previous studies.35,36,37 Walking disability was defined by self-report of difficult walking or need of special equipment for walking. Self-rated health status was dichotomized as poor to fair or good to excellent. Comorbid medical conditions included self-reported physician diagnosis of congestive heart failure, coronary heart disease, angina, heart attack, stroke, and cancer.

self-report of difficult walking or need of special equipment for walking. Self-rated health status was dichotomized as poor to fair or good to excellent. Comorbid medical conditions included self-reported physician diagnosis of congestive heart failure, coronary heart disease, angina, heart attack, stroke, and cancer. Statistical Analysis Data were analyzed from April 1 through 30, 2018. We combined the 2005-2006 and 2007-2008 phases of NHANES. All analysis accounted for the complex and stratified design based on NHANES analytic and reporting guidelines. Baseline characteristics of study participants, including age, sex, race/ethnicity, educational attainment, marital status, family income, smoking status, alcohol consumption, diabetes, hypertension, high cholesterol level, ratio of low-density to high-density lipoprotein cholesterol levels, body mass index, high C-reactive protein level, depressive symptoms, comorbid ocular diseases, walking disability, self-rated health, and history of congestive heart failure, coronary heart disease, angina, heart attack, stroke, or cancer were reported using means and SEs for continuous variables and numbers and weighted percentages for categorical variables.

level, depressive symptoms, comorbid ocular diseases, walking disability, self-rated health, and history of congestive heart failure, coronary heart disease, angina, heart attack, stroke, or cancer were reported using means and SEs for continuous variables and numbers and weighted percentages for categorical variables. We used the unpaired t test for the comparison of continuous variables and design-adjusted Rao-Scott Pearson χ2 test for the comparison of categorical data to compare the mortality characteristics by AMD status. Plots of survival curves of participants with early, late, and no AMD were generated using Kaplan-Meier estimates. Age- and sex-adjusted Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% CIs for mortality to determine baseline characteristics significantly associated with the end point. Covariates significantly associated with mortality and AMD status were added to final Cox proportional hazards regression models calculating HRs and population-attributable risk of AMD for mortality. Specific-cause mortality risk was estimated after multiple adjustments using the Fine and Gray competing risks regression model.38 Mortality resulting from other causes was treated as a competing risk.38 To address the nonresponse issue, we used inverse probability weighting to correct the estimates in the sensitivity analyses.39 We also conducted sensitivity analyses adjusted for age and age squared in final models to evaluate the nonlinear association of age with mortality. Interactions between covariates were tested, and no evidence of interaction was found (P > .05). The proportional hazards assumption for each variable was tested by graphically inspecting or by checking their interaction with follow-up time. No evidence suggested that any of these variables violated the assumption (P > .05). The variance inflation factors procedure was used to test collinearity for all variables, and all covariables’ variance inflation factors were less than 2.00 (mean [SE], 1.29 [0.04]). All data analysis was performed using Stata software (version 14.0; StataCorp). Two-sided P < .05 was considered significant for statistical inferences.

lation factors procedure was used to test collinearity for all variables, and all covariables’ variance inflation factors were less than 2.00 (mean [SE], 1.29 [0.04]). All data analysis was performed using Stata software (version 14.0; StataCorp). Two-sided P < .05 was considered significant for statistical inferences. Results Of the total 20 497 participants in 2005-2008 NHANES surveys, 6797 were 40 years or older. Of these, 1194 were excluded owing to missing retinal images (969 participants), ungradable images (224 participants), and missing mortality data (1 participant). The remaining 5603 participants (86.0%; 2810 [47.4%] men and 2793 [52.6%] women; 3017 [77.1%] white; mean [SE] age, 56.4 [0.4] years) were included in the final analytical sample (eFigure in the Supplement). Excluded participants were significantly older (≥80 years, 283 [18.4%] vs 464 [5.2%]; P < .001) and more likely to be black (330 [16.2%] vs 1139 [9.6%]; P < .001) when compared with study participants. Other demographic, health-related behaviors, and characteristics of excluded and included participants are shown in eTable 1 in the Supplement. Demographic characteristics, health-related behaviors, and general health comorbidities of participants overall and by AMD status are presented in Table 1. Participants with any AMD tended to be older (≥80 years, 143 [27.4%] vs 321 [3.6%]), white (314 [86.5%] vs 2703 [76.4%]), unmarried (206 [40.2%] vs 1819 [30.2%]), and former smokers (178 [42.0%] vs 1634 [30.1%]); to have hypertension (281 [59.8%] vs 2477 [42.5%]) and dyslipidemia (170 [41.3%] vs 1961 [37.6%]); to be normal or overweight (295 [68.6%] vs 3077 [61.5%]); to have walking disability (88 [18.1%] vs 507 [7.5%]); and to have comorbid CVD (eg, stroke, 56 [12.4%] vs 230 [3.5%]) and cancer (81 [16.3%] vs 617 [12.0%]). Other characteristics did not differ between the groups with and without AMD.

[41.3%] vs 1961 [37.6%]); to be normal or overweight (295 [68.6%] vs 3077 [61.5%]); to have walking disability (88 [18.1%] vs 507 [7.5%]); and to have comorbid CVD (eg, stroke, 56 [12.4%] vs 230 [3.5%]) and cancer (81 [16.3%] vs 617 [12.0%]). Other characteristics did not differ between the groups with and without AMD. Table 1. Demographics, Health Behaviors, and General Health Characteristics of Participants With and Without AMDa Characteristic Study Participants All (N = 5603) Without AMD (n = 5162) With AMD (n = 441) Age, No. (%), y 40-49 1501 (34.6) 1471 (36.2) 30 (12.0) 50-59 1327 (29.8) 1281 (30.9) 46 (15.1) 60-69 1386 (18.8) 1294 (18.7) 92 (20.5) 70-79 925 (11.6) 795 (10.6) 130 (25.0) ≥80 464 (5.2) 321 (3.6) 143 (27.4) Sex, No. (%) Male 2810 (47.4) 2581 (47.4) 229 (47.0) Female 2793 (52.6) 2581 (52.6) 212 (53.0) Race/ethnicity, No. (%) Non-Hispanic white 3017 (77.1) 2703 (76.4) 314 (86.5) Non-Hispanic black 1139 (9.6) 1103 (10.0) 36 (3.8) Mexican American 864 (5.4) 811 (5.5) 53 (4.2) Other 583 (7.9) 545 (8.1) 38 (5.5) Educational attainment, No. (%) <High school 1643 (18.0) 1511 (17.7) 132 (22.6) ≥High school 3960 (82.0) 3651 (82.3) 309 (77.4) Marital status, No. (%) Unmarried or other 2025 (30.9) 1819 (30.2) 206 (40.2) Married or living with a partner 3576 (69.1) 3341 (69.8) 235 (59.8) Poverty income ratio, No. (%) Below poverty line (<1.00) 828 (9.3) 766 (9.3) 62 (10.0) At or above poverty line (≥1.00) 4380 (90.7) 4043 (90.7) 337 (90.0) Smoking status, No. (%) Never 2648 (48.5) 2451 (48.9) 197 (43.0) Former 1812 (30.9) 1634 (30.1) 178 (42.0) Current 1141 (20.6) 1075 (21.0) 66 (15.0) Alcohol consumption, No. (%) Lifetime abstainer or former 1358 (20.6) 1233 (20.2) 125 (26.4) Current, drinks/wk ≤3 2961 (55.4) 2741 (55.6) 220 (52.4) >3 1148 (24.0) 1061 (24.2) 87 (21.2) Diabetes, No. (%) 1053 (13.6) 973 (13.4) 80 (16.4) Hypertension, No. (%) 2758 (43.6) 2477 (42.5) 281 (59.8) High total cholesterol level, No. (%) 2131 (37.9) 1961 (37.6) 170 (41.3) LDL-C:HDL-C level ratio, mean (SE) 2.30 (0.02) 2.32 (0.02) 2.11 (0.07) BMI, No. (%) <18.5 79 (1.3) 75 (1.4) 4 (0.5) 18.5-30.0 3372 (62.0) 3077 (61.5) 295 (68.6) ≥30.0 2109 (36.7) 1970 (37.1) 139 (30.9) High C-reactive protein level, No. (%) 627 (10.6) 568 (10.4) 59 (12.8) Depressive symptoms, No. (%) 463 (7.2) 433 (7.3) 30 (6.7) Comorbid ocular diseases, No. (%) 1361 (19.3) 1174 (17.8) 187 (40.0) Walking disability, No. (%) 595 (8.2) 507 (7.5) 88 (18.1) Self-rated health, No.

≥30.0 2109 (36.7) 1970 (37.1) 139 (30.9) High C-reactive protein level, No. (%) 627 (10.6) 568 (10.4) 59 (12.8) Depressive symptoms, No. (%) 463 (7.2) 433 (7.3) 30 (6.7) Comorbid ocular diseases, No. (%) 1361 (19.3) 1174 (17.8) 187 (40.0) Walking disability, No. (%) 595 (8.2) 507 (7.5) 88 (18.1) Self-rated health, No. (%) Poor to fair 1427 (18.8) 1305 (18.5) 122 (23.0) Good to excellent 4056 (81.2) 3746 (81.5) 310 (77.0) History of congestive heart failure, No. (%) 256 (3.3) 223 (3.0) 33 (7.3) History of coronary heart disease, No. (%) 319 (4.8) 275 (4.4) 44 (10.1) History of angina, No. (%) 230 (3.4) 200 (3.2) 30 (6.0) History of heart attack, No. (%) 350 (4.9) 299 (4.5) 51 (10.1) History of stroke, No. (%) 286 (4.1) 230 (3.5) 56 (12.4) History of cancer, No. (%) 698 (12.2) 617 (12.0) 81 (16.3) Abbreviations: AMD, age-related macular degeneration; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol. a All proportions, means, and SEs are weighted estimates of the US population characteristics, taking into account the complex sampling design of the National Health and Nutrition Examination Survey.

(%) Poor to fair 1427 (18.8) 1305 (18.5) 122 (23.0) Good to excellent 4056 (81.2) 3746 (81.5) 310 (77.0) History of congestive heart failure, No. (%) 256 (3.3) 223 (3.0) 33 (7.3) History of coronary heart disease, No. (%) 319 (4.8) 275 (4.4) 44 (10.1) History of angina, No. (%) 230 (3.4) 200 (3.2) 30 (6.0) History of heart attack, No. (%) 350 (4.9) 299 (4.5) 51 (10.1) History of stroke, No. (%) 286 (4.1) 230 (3.5) 56 (12.4) History of cancer, No. (%) 698 (12.2) 617 (12.0) 81 (16.3) Abbreviations: AMD, age-related macular degeneration; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol. a All proportions, means, and SEs are weighted estimates of the US population characteristics, taking into account the complex sampling design of the National Health and Nutrition Examination Survey. All-Cause Mortality Among the 5603 participants included in the current analysis, AMD was present at the baseline examination in 441 participants (6.6%), of whom 386 (5.8%) had early and 55 (0.8%) had late AMD. After a median follow-up of 4.5 years (interquartile range, 3.6-5.6 years), 433 participants (5.3%) died of all causes. Among these deceased participants, 361 (83.1%) had no AMD at baseline; 54 (11.5%), signs of early AMD; 18 (5.4%), signs of late AMD; and 72 (16.9%), any AMD. Mortality rates were higher for those participants who had early (54 [10.6%]), late (18 [35.9%]), or any (72 [13.6%]) AMD compared with no AMD (361 [4.7%]) (Table 2). The mean (SE) age at death of participants without AMD (70.9 [1.0] years) was significantly younger than that of participants with early (80.4 [1.4] years; P < .001), late (83.1 [1.7] years; P < .001), or any (81.3 [1.1] years; P < .001) AMD. The mean time to death did not differ significantly by AMD status.

(Table 2). The mean (SE) age at death of participants without AMD (70.9 [1.0] years) was significantly younger than that of participants with early (80.4 [1.4] years; P < .001), late (83.1 [1.7] years; P < .001), or any (81.3 [1.1] years; P < .001) AMD. The mean time to death did not differ significantly by AMD status. Table 2. Mortality Characteristics Overall and by AMD Statusa Characteristics All Participants (N = 5603) AMD Statusb None (n = 5162) Any (n = 441) Early (n = 386) Late (n = 55) Age at death, mean (SE), y Due to all causes 72.6 (0.9) 70.9 (1.0) 81.3 (1.1)c 80.4 (1.4)c 83.1 (1.7)c Due to CVD 73.8 (1.3) 72.8 (1.4) 79.1 (3.0) 77.9 (3.5) 83.7 (3.5)d Due to cancer 70.9 (1.1) 70.0 (1.3) 78.4 (2.9)d 77.1 (3.3) 84.8 (1.9)c Due to non-CVD and noncancer causes 72.9 (1.3) 70.3 (1.3) 82.9 (1.1)c 83.0 (1.4)c 82.8 (1.6)c Mortality rate, No. (%) Due to all causes 433 (5.3) 361 (4.7) 72 (13.6)c 54 (10.6)d 18 (35.9)d Due to CVD 117 (1.4) 102 (1.2) 15 (3.3)c 12 (3.0)d 3 (5.4)d Due to cancer 105 (1.3) 92 (1.3) 13 (2.2)c 11 (2.1)d 2 (2.9)d Due to non-CVD and noncancer causes 211 (2.6) 167 (2.2) 44 (8.2)c 31 (5.5)d 13 (27.6)d Time to death from baseline examination, mean (SE), mo Due to all causes 32.8 (1.0) 32.4 (1.2) 34.6 (2.1) 34.0 (2.5) 35.9 (5.4) Due to CVD 31.1 (1.5) 30.0 (1.8) 36.8 (4.9) 30.0 (3.5) 64.4 (4.4)c Due to cancer 35.6 (2.0) 36.0 (2.1) 32.2 (2.8) 32.2 (3.3) 32.0 (2.4) Due to non-CVD and noncancer causes 32.2 (1.3) 31.7 (1.7) 34.4 (2.8) 36.9 (3.9) 30.8 (4.8) Abbreviations: AMD, age-related macular degeneration; CVD, cardiovascular disease.

35.9 (5.4) Due to CVD 31.1 (1.5) 30.0 (1.8) 36.8 (4.9) 30.0 (3.5) 64.4 (4.4)c Due to cancer 35.6 (2.0) 36.0 (2.1) 32.2 (2.8) 32.2 (3.3) 32.0 (2.4) Due to non-CVD and noncancer causes 32.2 (1.3) 31.7 (1.7) 34.4 (2.8) 36.9 (3.9) 30.8 (4.8) Abbreviations: AMD, age-related macular degeneration; CVD, cardiovascular disease. a Mortality was assessed through December 31, 2011. All proportions, means, and SEs are weighted estimates of the US population characteristics, taking into account the complex sampling design of the National Health and Nutrition Examination Survey. b All P values were calculated using the unpaired t test for continuous variables and the design-adjusted Rao-Scott Pearson χ2 test for categorical variables. Comparisons were between each group with AMD and the group with no AMD and were unadjusted. c P < .001. d P < .05.

a Mortality was assessed through December 31, 2011. All proportions, means, and SEs are weighted estimates of the US population characteristics, taking into account the complex sampling design of the National Health and Nutrition Examination Survey. b All P values were calculated using the unpaired t test for continuous variables and the design-adjusted Rao-Scott Pearson χ2 test for categorical variables. Comparisons were between each group with AMD and the group with no AMD and were unadjusted. c P < .001. d P < .05. The association of baseline covariates with all-cause mortality is shown in Table 3. After adjustments for age and sex, the HRs increased exponentially for each decade of age. Men had an increased risk of mortality due to all causes (HR, 1.53; 95% CI, 1.20-1.96; P = .001). Age- and sex-adjusted Cox proportional hazards regression models showed covariates including race/ethnicity (HR, 1.67; 95% CI, 1.25-2.22), educational attainment (HR, 0.61; 95% CI, 0.46-0.81), marital status (HR, 0.51; 95% CI, 0.41-0.64), family income (HR, 0.43; 95% CI, 0.30-0.62), smoking status (HR for former smokers, 1.63 [95% CI, 1.12-2.36]; HR for current smoking, 3.54 [95% CI, 2.57-4.87]), alcohol consumption (HR, 0.51; 95% CI, 0.38-0.70), diabetes (HR, 2.25; 95% CI, 1.52-3.31), dyslipidemia (HR, 0.72; 95% CI, 0.58-0.91), body mass index (HR, 3.29; 95% CI, 1.61-6.75), C-reactive protein level (HR, 2.62; 95% CI, 1.64-4.20), depressive symptoms (HR, 2.16; 95% CI, 1.37-3.42), comorbid ocular diseases (HR, 2.01; 95% CI, 1.49-2.72), self-rated health status (HR, 0.36; 95% CI, 0.28-0.47), walking disability (HR, 2.99; 95% CI, 2.36-3.78), and self-reported history of CVD (eg, HR for stroke, 2.63; 95% CI, 1.98-3.50) or cancer (HR, 1.50; 95% CI, 1.13-2.01) were significantly associated with an increased risk of all-cause mortality. After controlling for variables significantly associated with mortality and AMD status, the multivariate Cox regression model (Table 4) indicated that poorer survival was associated with late AMD at baseline when compared with participants without AMD (HR, 2.01; 95% CI, 1.00-4.03; P = .049). The stratum-specific HRs increased exponentially for each decade of age, ranging from 2.59 (95% CI, 1.36-4.94) for the group aged 40 to 49 years and 19.4 (95% CI, 9.18-41.0) for the group 80 years or older. However, participants with early AMD (HR, 0.79; 95% CI, 0.57-1.11; P = .17) or any AMD (HR, 1.00; 95% CI, 0.75-1.33; P = .97) at baseline were not at greater risk of all-cause mortality compared with participants without AMD. The population-attributable risk ranged from −1.23% (95% CI, −2.56% to 0.63%) for early AMD to 0.80% (95% CI, 0.00%-2.37%) for late AMD (Table 4).

CI, 0.57-1.11; P = .17) or any AMD (HR, 1.00; 95% CI, 0.75-1.33; P = .97) at baseline were not at greater risk of all-cause mortality compared with participants without AMD. The population-attributable risk ranged from −1.23% (95% CI, −2.56% to 0.63%) for early AMD to 0.80% (95% CI, 0.00%-2.37%) for late AMD (Table 4). Multiple adjusted Kaplan-Meier curves for all-cause mortality by AMD status are shown in the Figure.

CI, 0.57-1.11; P = .17) or any AMD (HR, 1.00; 95% CI, 0.75-1.33; P = .97) at baseline were not at greater risk of all-cause mortality compared with participants without AMD. The population-attributable risk ranged from −1.23% (95% CI, −2.56% to 0.63%) for early AMD to 0.80% (95% CI, 0.00%-2.37%) for late AMD (Table 4). Multiple adjusted Kaplan-Meier curves for all-cause mortality by AMD status are shown in the Figure. Table 3. All-Cause Mortality by Demographic, Health-Related Behaviors, and General Health Characteristicsa Characteristics Participants HR (95% CI)b Survived (n = 5170) Died (n = 433) Age, No. (%), y 40-49 1473 (36.0) 28 (8.8) 1 [Reference] 50-59 1281 (30.7) 46 (15.0) 2.03 (1.13-3.64)c 60-69 1298 (18.9) 88 (16.5) 3.67 (1.89-7.12)d 70-79 803 (10.6) 122 (28.7) 10.9 (5.64-21.2)d ≥80 315 (3.8) 149 (31.0) 30.4 (16.5-55.8)d Sex, No. (%) Male 2548 (47.0) 262 (54.0) 1.53 (1.20-1.96)c Female 2622 (53.0) 171 (46.0) 1 [Reference] Race/ethnicity, No. (%) Non-Hispanic white 2742 (77.0) 275 (79.2) 1 [Reference] Non-Hispanic black 1042 (9.4) 97 (11.9) 1.67 (1.25-2.22)c Mexican American 827 (5.5) 37 (3.7) 1.00 (0.70-1.42) Other 559 (8.1) 24 (5.2) 0.97 (0.47-2.01) Educational attainment, No. (%) Less than high school 1462 (17.1) 181 (33.6) 1 [Reference] High school or more 3708 (82.9) 252 (66.4) 0.61 (0.46-0.81)c Marital status, No. (%) Unmarried or other 1800 (29.7) 225 (51.6) 1 [Reference] Married or living with a partner 3368 (70.3) 208 (48.4) 0.51 (0.41-0.64)d Poverty income ratio, No. (%) Below poverty line (<1.00) 737 (8.9) 91 (17.5) 1 [Reference] At or above poverty line (≥1.00) 4078 (91.1) 302 (82.5) 0.43 (0.30-0.62)d Smoking status, No. (%) Never 2502 (49.4) 146 (31.5) 1 [Reference] Former 1624 (30.2) 188 (43.6) 1.63 (1.12-2.36)c Current 1042 (20.4) 99 (24.9) 3.54 (2.57-4.87)d Alcohol consumption, No. (%) Lifetime abstainer or former 1222 (20.0) 136 (31.0) 1 [Reference] Current, drinks/wk ≤3 2731 (55.4) 230 (54.9) 0.87 (0.70-1.07) >3 1092 (24.6) 56 (14.1) 0.51 (0.38-0.70)d Diabetes, No. (%) No 4122 (87.2) 289 (70.5) 1 [Reference] Yes 924 (12.8) 129 (29.5) 2.25 (1.52-3.31)d Hypertension, No. (%) No 2617 (57.6) 133 (33.9) 1 [Reference] Yes 2479 (42.4) 279 (66.1) 1.32 (0.94-1.83) High cholesterol level, No. (%) No 3064 (62.1) 258 (62.5) 1 [Reference] Yes 1979 (37.9) 152 (37.5) 0.72 (0.58-0.91)c LDL-C:HDL-C level ratio, mean (SE) 2.31 (0.02) 2.19 (0.10) 1.07 (0.80-1.42) BMI, No. (%) 18.5-30.0 3081 (61.8) 291 (65.6) 1 [Reference] <18.5 66 (1.2) 13 (3.7) 3.29 (1.61-6.75)c ≥30.0 1988 (37.0) 121 (30.8) 1.03 (0.79-1.34) High C-reactive protein level, No.

5) 1 [Reference] Yes 1979 (37.9) 152 (37.5) 0.72 (0.58-0.91)c LDL-C:HDL-C level ratio, mean (SE) 2.31 (0.02) 2.19 (0.10) 1.07 (0.80-1.42) BMI, No. (%) 18.5-30.0 3081 (61.8) 291 (65.6) 1 [Reference] <18.5 66 (1.2) 13 (3.7) 3.29 (1.61-6.75)c ≥30.0 1988 (37.0) 121 (30.8) 1.03 (0.79-1.34) High C-reactive protein level, No. (%) No 4472 (90.0) 318 (78.1) 1 [Reference] Yes 539 (10.0) 88 (21.9) 2.62 (1.64-4.20)d Depressive symptoms, No. (%) No 4613 (92.9) 372 (90.6) 1 [Reference] Yes 422 (7.1) 41 (9.4) 2.16 (1.37-3.42)c Comorbid ocular diseases, No. (%) No 3895 (82.5) 190 (47.7) 1 [Reference] Yes 1139 (17.5) 222 (52.3) 2.01 (1.49-2.72)d Walking disability, No. (%) No 4703 (93.0) 305 (70.4) 1 [Reference] Yes 467 (7.0) 128 (29.6) 2.99 (2.36-3.78)d Self-rated health, No. (%) Poor to fair 1238 (17.5) 189 (41.5) 1 [Reference] Good to excellent 3822 (82.5) 234 (58.5) 0.36 (0.28-0.47)d History of congestive heart failure, No. (%) No 4993 (97.6) 354 (79.9) 1 [Reference] Yes 177 (2.4) 79 (20.1) 4.23 (3.02-5.91)d History of coronary heart disease, No. (%) No 4900 (95.5) 384 (89.1) 1 [Reference] Yes 270 (4.5) 49 (10.9) 1.08 (0.75-1.57) History of angina, No. (%) No 4975 (96.9) 398 (92.2) 1 [Reference] Yes 195 (3.1) 35 (7.8) 1.18 (0.82-1.71) History of heart attack, No. (%) No 4898 (95.8) 355 (83.1) 1 [Reference] Yes 272 (4.2) 78 (16.9) 2.02 (1.32-3.10)c History of stroke, No. (%) No 4952 (96.6) 365 (83.3) 1 [Reference] Yes 218 (3.4) 68 (16.7) 2.63 (1.98-3.50)d History of cancer, No. (%) No 4583 (88.6) 322 (72.4) 1 [Reference] Yes 587 (11.4) 111 (27.6) 1.50 (1.13-2.01)c Abbreviations: AMD, age-related macular degeneration; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HDL-C, high-density lipoprotein cholesterol; HR, hazard ratio; LDL-C, low-density lipoprotein cholesterol.

(72.4) 1 [Reference] Yes 587 (11.4) 111 (27.6) 1.50 (1.13-2.01)c Abbreviations: AMD, age-related macular degeneration; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HDL-C, high-density lipoprotein cholesterol; HR, hazard ratio; LDL-C, low-density lipoprotein cholesterol. a All-cause mortality was assessed through December 31, 2011. All proportions, means, and SEs are weighted estimates of the US population characteristics, taking into account the complex sampling design of the National Health and Nutrition Examination Survey. b Adjusted for age and sex. c P < .05. d P < .001.

(72.4) 1 [Reference] Yes 587 (11.4) 111 (27.6) 1.50 (1.13-2.01)c Abbreviations: AMD, age-related macular degeneration; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HDL-C, high-density lipoprotein cholesterol; HR, hazard ratio; LDL-C, low-density lipoprotein cholesterol. a All-cause mortality was assessed through December 31, 2011. All proportions, means, and SEs are weighted estimates of the US population characteristics, taking into account the complex sampling design of the National Health and Nutrition Examination Survey. b Adjusted for age and sex. c P < .05. d P < .001. Table 4. Cox Proportional Hazards Models for All-Cause Mortality and Fine and Gray Competing Risks Regression Models for Specific-Cause Mortality by AMD Status AMD Status Mortalitya All-Cause CVD-Specific Cancer-Specific Not Due to CVD or Cancer HR (95% CI) PAR (95% CI), % HR (95% CI) PAR (95% CI), % HR (95% CI) PAR (95% CI), % HR (95% CI) PAR (95% CI), % None 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] 1 [Reference] Any (early or late) 1.00 (0.75 to 1.33) 0.00 (−1.68 to 2.13) 0.55 (0.20 to 1.50) −3.06 (−5.57 to 3.21) 0.88 (0.43 to 1.82) −0.77 (−3.91 to 5.12) 1.33 (0.86 to 2.07) 2.14 (−0.96 to 6.60) Early 0.79 (0.57 to 1.11) −1.23 (−2.56 to 0.63) 0.47 (0.17 to 1.29) −3.20 (−5.07 to 1.65) 0.85 (0.40 to 1.78) −0.89 (−3.59 to 4.35) 0.96 (0.58 to 1.59) −0.25 (−2.52 to 3.31) Late 2.01 (1.00 to 4.03)b 0.80 (0.00 to 2.37)b 0.78 (0.14 to 4.35) −0.17 (−0.69 to 2.61) 1.27 (0.19 to 8.68) 0.21 (−0.66 to 5.79) 3.42 (1.38 to 8.49)b 1.90 (0.30 to 5.65)b Abbreviations: AMD, age-related macular degeneration; CVD, cardiovascular disease; HR, hazard ratio; PAR, population attributable risk.

.52 to 3.31) Late 2.01 (1.00 to 4.03)b 0.80 (0.00 to 2.37)b 0.78 (0.14 to 4.35) −0.17 (−0.69 to 2.61) 1.27 (0.19 to 8.68) 0.21 (−0.66 to 5.79) 3.42 (1.38 to 8.49)b 1.90 (0.30 to 5.65)b Abbreviations: AMD, age-related macular degeneration; CVD, cardiovascular disease; HR, hazard ratio; PAR, population attributable risk. a Adjusted for age, sex, race/ethnicity, educational attainment, marital status, family income, smoking status, alcohol consumption, diabetes, hypertension, high cholesterol level, body mass index, high C-reactive protein level, depressive symptoms, comorbid ocular diseases, walking disability, self-rated health, history of CVD, and cancer. b P < .05. Figure. Adjusted Kaplan-Meier Curve for All-Cause Mortality Rate Findings are stratified by age-related macular degeneration (AMD) status, using the 2005-2008 National Health and Nutrition Examination Survey data. All-cause mortality was assessed through December 31, 2011. Late AMD was associated with greater mortality rates.

Adjusted Kaplan-Meier Curve for All-Cause Mortality Rate Findings are stratified by age-related macular degeneration (AMD) status, using the 2005-2008 National Health and Nutrition Examination Survey data. All-cause mortality was assessed through December 31, 2011. Late AMD was associated with greater mortality rates. Cause-Specific Mortality Among the 433 participants who died of all causes, 117 deaths (25.6%) were CVD specific, 105 (25.2%) were cancer specific, and 211 (49.2%) were not specific to CVD or cancer. The presence of any, early, or late AMD was associated with significantly higher mortality rates for each specific cause (Table 2). Competing risk regression models for cause-specific mortality showed that late AMD was associated with a more than 3-fold higher risk of mortality not due to CVD or cancer (HR, 3.42; 95% CI, 1.38-8.49; P = .01) after multiple adjustments. However, we identified no association of any AMD or early AMD with specific-cause mortality after multivariate adjustments. The population-attributable risk results of these risk factors are listed in Table 4. Sensitivity Analyses The sensitivity analyses used inverse probability weighting to correct the estimates for nonresponse, which yielded results similar to those reported in the main analysis (eTable 2 in the Supplement). We also observed results comparable to those of the main analysis when squared age was included in the final model (eTable 3 in the Supplement).

used inverse probability weighting to correct the estimates for nonresponse, which yielded results similar to those reported in the main analysis (eTable 2 in the Supplement). We also observed results comparable to those of the main analysis when squared age was included in the final model (eTable 3 in the Supplement). Discussion In this nationally representative sample consisting of 5603 US adults 40 years or older, we report that only late AMD was associated with increased risks of all-cause mortality and mortality not specific to CVD or cancer. The presence of any AMD and early AMD were not associated with all-cause or specific-cause mortality.

this nationally representative sample consisting of 5603 US adults 40 years or older, we report that only late AMD was associated with increased risks of all-cause mortality and mortality not specific to CVD or cancer. The presence of any AMD and early AMD were not associated with all-cause or specific-cause mortality. The findings from previous population-based studies12,13,14,15,16,17,18,19,20,21,22,23,24,25,40,41 on associations between AMD and mortality are summarized in eTable 4 in the Supplement. Our finding that the presence of late AMD was associated with an increased risk of all-cause mortality is in line with the Age-Related Eye Disease Study (AREDS),14 AREDS2,15 and Beaver Dam Eye Study.16,20 Other population-based studies12,13,17,18,19 have reported significant associations between any and mild AMD and all-cause mortality in subpopulations only. In contrast, the Singapore Malay Eye study,22 Beijing Eye study,24 Rotterdam study,21 Melbourne Visual Impairment Project cohort,25 Andhra Pradesh Eye Disease study,40 a UK study,23 and the Atherosclerosis Risk in Communities Study41 did not report a significant association between AMD and survival. The differences in definitions and numbers of confounders, the assessment and grading system of AMD, the length of the follow-up period, and the number of deaths may explain varied results from these studies.

Atherosclerosis Risk in Communities Study41 did not report a significant association between AMD and survival. The differences in definitions and numbers of confounders, the assessment and grading system of AMD, the length of the follow-up period, and the number of deaths may explain varied results from these studies. In the analysis of cause-specific mortality, our results support those of a recent meta-analysis of 5 population-based studies26 that suggest that AMD is not associated with CVD mortality. However, our findings were challenged by 2 recent meta-analyses42,43 that concluded that AMD, specifically late AMD, was associated with an increased risk of CVD mortality. The inconsistent nature of these results may be attributable to a range of factors. Given that atherothrombotic events and mortality have been associated with the use of anti-VEGF agents in patients with late AMD,27,28 undocumented use of anti-VEGF therapies may overestimate poor survival resulting from CVD in patients with AMD. In addition, when estimating the specific mortality in a geriatric population with comorbidities associated with poorer survival, the use of Cox proportional hazards regression models in previous studies might lead to overestimation in the absolute risk of the specific mortality by not considering the competing risks of death.38 We used the competing risk models to deal with this methodologic issue. Last, some previous studies were subject to insufficient adjustment for important confounders, such as smoking44 and a history of CVD events,14 that could explain the significant association between AMD and CVD mortality.

ompeting risks of death.38 We used the competing risk models to deal with this methodologic issue. Last, some previous studies were subject to insufficient adjustment for important confounders, such as smoking44 and a history of CVD events,14 that could explain the significant association between AMD and CVD mortality. Our association between late AMD and mortality not specific to CVD or cancer is in agreement with several previous studies,15,45 but not others.12 The reasons underlying the association of AMD with increased risk of mortality not due to CVD or cancer are still unclear. However, growing evidence supports the association between AMD and neurodegenerative diseases (eg, Alzheimer disease),46 which increased the risk of mortality.47 The limited number of deaths due to Alzheimer disease in our analysis could not explore this hypothesis. In addition, it has been speculated that visual impairment and blindness due to AMD may lead to functional and psychological problems, such as falls,48,49 fractures,50,51 unintentional injuries,52 and a loss of independence.53,54,55 These problems may contribute to the higher risk of mortality not due to CVD or cancer when compared with unaffected individuals. However, this hypothesis has been challenged by previous analyses.14,15 We did not have sufficient unintentional injury–related mortality to examine this hypothesis. We found no statistically significant differences in depression symptoms between the groups with any AMD and no AMD. Although depression symptoms were associated with higher risk of mortality in the age- and sex-adjusted model, it did not remain significant after multiple adjustments. Further studies are needed to elucidate these associations.

significant differences in depression symptoms between the groups with any AMD and no AMD. Although depression symptoms were associated with higher risk of mortality in the age- and sex-adjusted model, it did not remain significant after multiple adjustments. Further studies are needed to elucidate these associations. Another explanation elucidating the poor survival among individuals with late AMD is that the AMD reflects systemic comorbidities associated with frailty and aging. This explanation is supported by results from the present study and previous studies.7,14,26,56 Common pathogenesis, such as chronic inflammation, atherosclerosis, oxidative stress, and lipid metabolism might be the main mechanism between AMD and systemic comorbidities.56,57,58 The implications of these findings suggest that AMD is a biomarker of frailty and aging.59,60 Alternatively, this association between AMD and mortality may be attributable to unmeasured or inadequately assessed confounding factors for AMD. Age is the most important risk factor for AMD and mortality. Inclusion of age assessed as a continuous or a categorical variable or additional inclusion of age squared in the final model did not affect the association between late AMD and mortality. Second to age, smoking is an established risk factor for AMD.4 In the models adjusted for multiple covariates, AMD remained significantly associated with mortality. In addition, interaction terms of AMD with age or smoking status were not significantly associated with mortality, suggesting no difference in the AMD-mortality association in these subgroups. Extensive evidence supported increased risks of mortality among participants with other age-related ocular diseases.61,62,63 To elaborate on the real nature of the AMD-mortality association, we also adjusted for comorbid ocular diseases in the final model, which did not affect the AMD-mortality association.

subgroups. Extensive evidence supported increased risks of mortality among participants with other age-related ocular diseases.61,62,63 To elaborate on the real nature of the AMD-mortality association, we also adjusted for comorbid ocular diseases in the final model, which did not affect the AMD-mortality association. Strengths and Limitations Strengths of these analyses include the large sample size of an elderly cohort, standardized objective methods for assessing AMD, availability of comprehensive demographic characteristics, health indicators, comorbidities, and complete death records. The study was limited by the following points. First, health behavior and comorbidities were collected at a single time point, and study participants’ behavior and comorbidity status might change during follow-up. Second, although we adjusted for a comprehensive range of confounding factors, we cannot rule out residual confounding, such as anti-VEGF therapies. Finally, participants excluded in the present analysis were older and unhealthier, which might have influenced results. Nevertheless, the inverse probability weighting model to correct the estimates for nonresponse yielded similar results, again verifying the robustness of our conclusions.

such as anti-VEGF therapies. Finally, participants excluded in the present analysis were older and unhealthier, which might have influenced results. Nevertheless, the inverse probability weighting model to correct the estimates for nonresponse yielded similar results, again verifying the robustness of our conclusions. Conclusions Our findings suggest that in a large sample of elderly participants residing in the United States, only late AMD was associated with an increased risk of all-cause morality and mortality not related to CVD or cancer. The reasons for the association of late AMD with decreased survival have yet to be confirmed; however, our results suggest that AMD may reflect some systemic pathologic comorbidities indicative of frailty and aging. Alternatively, this association may be due to unmeasured or inadequately assessed confounding factors for late AMD. Further studies are needed to confirm these findings and elucidate the possible mechanisms underlying AMD. Supplement. eFigure. Schematic Showing Inclusion Criteria for Study Participants eTable 1. Demographic, Health-Related Behaviors and General Health Characteristics of Participants Included and Excluded in the Analyses eTable 2. Cox Proportional Hazards Models for All-Cause Mortality and Fine and Gray Competing Risks Regression Models for Specific-Cause Mortality by Age-Related Macular Degeneration Status Using Inverse Probability Weighting

eTable 1. Demographic, Health-Related Behaviors and General Health Characteristics of Participants Included and Excluded in the Analyses eTable 2. Cox Proportional Hazards Models for All-Cause Mortality and Fine and Gray Competing Risks Regression Models for Specific-Cause Mortality by Age-Related Macular Degeneration Status Using Inverse Probability Weighting eTable 3. Cox Proportional Hazards Models for All-Cause Mortality and Fine and Gray Competing Risks Regression Models for Specific-Cause Mortality by Age-Related Macular Degeneration Status With Additional Adjustment for Age Squared eTable 4. Summary Description of Previous Studies on the Association Between Age-Related Macular Degeneration and Mortality Click here for additional data file.

Introduction The eye is an immunoprivileged site that can harbor viruses for years.1 Chronic infection of the positive-sense single-stranded rubella virus (RV) is a cause of the Fuchs heterochromic iridocyclitis (FHI) phenotype.2,3,4,5 Prior reports3,6,7 have noted that intraocular fluid obtained from many cases of FHI demonstrate the presence of antibodies to RV, yet reverse transcription–polymerase chain reaction (RT-PCR) has frequently failed to demonstrate the presence of RV RNA. This discrepancy has been suggested to reflect a limited period during which the virus may be detected or persist in the eye.3,6 Unbiased metagenomic deep sequencing (MDS) is a high-throughput sequencing approach that can identify all genomes present in a clinical sample. Prevous studies8,9 have demonstrated that MDS of intraocular fluid can detect fungi, parasites, and DNA and RNA viruses in patients with intraocular inflammation.8,9 We present a case series of patients with rubella-associated uveitis diagnosed with MDS and assess the utility of MDS in identifying these infections.

le. Prevous studies8,9 have demonstrated that MDS of intraocular fluid can detect fungi, parasites, and DNA and RNA viruses in patients with intraocular inflammation.8,9 We present a case series of patients with rubella-associated uveitis diagnosed with MDS and assess the utility of MDS in identifying these infections. Methods This case series included 6 patients referred to the Francis I. Proctor Foundation, University of California, San Francisco (UCSF) for evaluation of recurrent or chronic hypertensive nongranulomatous anterior uveitis or hypertensive intermediate uveitis with concern for vitreal lymphoma (eTable in the Supplement). Ethical clearance was obtained from the institutional review board at UCSF, and the study adhered to the tenets of the Declaration of Helsinki.10 Written informed consent was obtained from all patients, and all data were deidentified.

ve intermediate uveitis with concern for vitreal lymphoma (eTable in the Supplement). Ethical clearance was obtained from the institutional review board at UCSF, and the study adhered to the tenets of the Declaration of Helsinki.10 Written informed consent was obtained from all patients, and all data were deidentified. Five patients (83%) were immigrants to the United States, whereas 1 (17%) was born in the United States before the institution of routine RV vaccination. Two patients exhibited anterior uveitis, whereas the remaining 4 exhibited anterior-intermediate inflammation. Two patients had no prior topical corticosteroid exposure. Five patients had a history of ocular hypertension, with 4 patients having gonioscopy performed that revealed open angles on gonioscopy, although 1 patient featured bridging vessels. Four eyes in 4 patients had corneal sensation examined, with 3 eyes exhibiting reduced corneal sensation. All involved eyes exhibited nongranulomatous keratic precipitates, with 4 having diffuse stellate keratic precipitates. Four eyes from 3 patients exhibited both iris atrophy and iris transillumination defects, 2 patients exhibited iris heterochromia, and 1 patient had no iris defects compared with the contralateral unaffected eye. All patients maintained visual acuity, ranging from 20/20 to 20/60. Two patients had fluorescein angiography performed without vascular leakage or staining of the disc (eFigure in the Supplement).

2 patients exhibited iris heterochromia, and 1 patient had no iris defects compared with the contralateral unaffected eye. All patients maintained visual acuity, ranging from 20/20 to 20/60. Two patients had fluorescein angiography performed without vascular leakage or staining of the disc (eFigure in the Supplement). Three patients received confocal microscopic imaging in our clinic. We found that the affected eyes exhibited spotlike holes, increased intercellular spaces, and infiltration of endothelial cells (Figure 1). In addition, all affected eyes exhibited features of polymorphism and polymegathism compared with the contralateral eye (Figure 1D). Two patients also exhibited stellate keratic precipitates on confocal microscopy, and 1 patient exhibited spotlike holes and endothelial infiltration in the unaffected eye. Similar findings have previously been described for herpes simplex virus–associated endothelial involvement.11 Figure 1. Confocal Images From the Patient Cohort Confocal images of the affected right eye (A) and unaffected left eye (B) of patient 3 showing infiltration of endothelial cells with endothelial infiltration (blue arrowhead) and spotlike holes (yellow arrowhead). Confocal images of unaffected right eye (C) and affected left eye (D) of patient 5 showing polymegathism and polymorphism as well as infiltration of endothelial cells (blue arrowhead).

tient 3 showing infiltration of endothelial cells with endothelial infiltration (blue arrowhead) and spotlike holes (yellow arrowhead). Confocal images of unaffected right eye (C) and affected left eye (D) of patient 5 showing polymegathism and polymorphism as well as infiltration of endothelial cells (blue arrowhead). All 6 patients had anterior chamber paracenteses performed at the Proctor Foundation, where fluid was submitted for targeted herpes simplex virus 1 and 2, varicella-zoster virus, and cytomegalovirus PCRs, with residual fluid subjected to MDS.8 Results All samples from the 6 white male patients (age range, 36-61 years) tested negative for herpes simplex virus 1 and 2, varicella-zoster virus, and cytomegalovirus by PCR, but all tested positive for RV RNA by MDS. Figure 2 shows the heterogeneous capture of the RV genome regions among samples, reflecting the unbiased nature of the assay. The MDS coverage of the RV genome for patient 6 is described elsewhere.8 Genes in the nonstructural coding sequence were detected for all 6 samples, whereas only 3 samples tested positive for genes in the structural coding sequence. Orthogonal testing using RT-PCR for the RV E1 gene was performed on 2 samples. Only the sample that yielded the highest number of reads on MDS tested positive on RT-PCR with a low cycling threshold of 38, demonstrating the potential utility of MDS in suspected cases of FHI.8

for genes in the structural coding sequence. Orthogonal testing using RT-PCR for the RV E1 gene was performed on 2 samples. Only the sample that yielded the highest number of reads on MDS tested positive on RT-PCR with a low cycling threshold of 38, demonstrating the potential utility of MDS in suspected cases of FHI.8 Figure 2. Rubella Virus (RV) Genomes From the Patient Cohort Rubella virus sequences from 5 patients assembled against the reference RV genome. CDS indicates coding sequence; RdRp, RNA-dependent RNA polymerase.

for genes in the structural coding sequence. Orthogonal testing using RT-PCR for the RV E1 gene was performed on 2 samples. Only the sample that yielded the highest number of reads on MDS tested positive on RT-PCR with a low cycling threshold of 38, demonstrating the potential utility of MDS in suspected cases of FHI.8 Figure 2. Rubella Virus (RV) Genomes From the Patient Cohort Rubella virus sequences from 5 patients assembled against the reference RV genome. CDS indicates coding sequence; RdRp, RNA-dependent RNA polymerase. Discussion The World Health Organization declared RV elimination in the Americas in 2015 as the result of effective vaccination policies.12 Because humans are the only host for RV, the prevalence of FHI in the United States was reduced substantially after the introduction of RV vaccination.13 Rubella infection, however, remains a threat throughout other parts of the world. Five of 6 patients in this study emigrated from regions where the vaccination policies were not strictly enforced. From a clinical standpoint, asking patients who live outside the United States about their immunization status or checking RV IgG levels is of little help in the workup for FHI because all immigrants are required to have the mumps-measles-rubella vaccine before entry. Furthermore, few patients will remember if they were exposed because only 2 of the patients in our study recalled having German measles as children. Although the theoretical risk of transmission exists because the patient’s RV strain in the eye can be genetically different than that of the vaccine strain,8 in the absence of intraocular surgery or globe trauma, the functional risk is likely minimal. In the 3 patients who underwent RT-PCR testing for RV RNA in the tears, nasopharynx, and urine, all samples tested negative for viral genome, indicating that the virus was exclusively localized to the eye.

the vaccine strain,8 in the absence of intraocular surgery or globe trauma, the functional risk is likely minimal. In the 3 patients who underwent RT-PCR testing for RV RNA in the tears, nasopharynx, and urine, all samples tested negative for viral genome, indicating that the virus was exclusively localized to the eye. Molecular diagnostics for intraocular rubella infections are not routinely available in the United States. The Goldmann-Witmer coefficient assay (requiring both intraocular fluid and a serum sample) and the RT-PCR for RV, which are available in Europe, can be limited in scope.3,7,14 The RV-directed RT-PCR usually targets a small region (approximately 185-739 base pairs [bp]) of the E1 gene, which is a small fraction of the 9762 bp of the entire RV genome.8,15 However, MDS is an unbiased approach that has the potential to detect any genome region of a particular pathogen in a clinical sample (Figure 2). Furthermore, the unbiased nature of the assay allows for the detection of both common and rare pathogens in minute amounts of intraocular fluid (approximately 20-50 μL) without requiring a priori knowledge, as is the case with pathogen-directed PCR.8 In contrast to prior work that used RT-PCR,3 this study found that regardless of age, the RV genome can be detected in patients with FHI using MDS. Our findings also demonstrated that anterior chamber paracentesis for RV testing by MDS may be sufficient even if the inflammation is localized mainly in the vitreous cavity of phakic patients.

work that used RT-PCR,3 this study found that regardless of age, the RV genome can be detected in patients with FHI using MDS. Our findings also demonstrated that anterior chamber paracentesis for RV testing by MDS may be sufficient even if the inflammation is localized mainly in the vitreous cavity of phakic patients. A major limitation of this study is that the population studied is from a referred group of patients with previously undifferentiated anterior and intermediate uveitis. In such cases, other causes, including syphilis and tuberculosis, must be ruled out. All of the patients had negative treponemal antibody and interferon gamma release assay test results and normal findings on chest radiography. Conclusions In summary, these findings from MDS suggest that inflammation in patients with FHI (in the classic FHI phenotype or anterior-intermediate uveitis) is stimulated by the persistent presence of RV in the eye and that these affected eyes exhibit evidence of corneal endothelial cell damage previously not appreciated without confocal imaging. As other studies14,16 have demonstrated, patients with RV uveitis can maintain usable visual acuity. Supplement. eFigure. Posterior segment features in rubella virus associated uveitis eTable. Ocular features in patients with rubella-associated uveitis Click here for additional data file.

1485 of 1535 eligible infants (96.7%) born at GA 24 to 30 weeks from 12 US centers between 2005 and 2010 (Figure 1). The validation European group included 329 of 354 eligible infants (92.9%) born at GA 24 to 30 weeks from Freiburg, Germany, with retrospective screening data collected between 2011 and 2017 (Figure 1). Study Procedures The estimation of GA was based on fetal ultrasonographic results. The chronological (postnatal) age, postmenstrual age, and GA are defined according to the American Academy of Pediatrics’ issued policy. An SD score (SDS) of expected reference weight (birth weight SDS [BWSDS]) was calculated based on GA, sex, and birth weight for all healthy singletons born at GA at least 24 weeks between 1990 and 1999 in Sweden and registered in the Medical Birth Register (800 000 healthy infants of approximately 1 million born). Hence, BWSDS was not calculated for infants born at GA less than 24 weeks because of a lack of reference for this extremely preterm population. Infants born at GA less than 24 weeks are at high risk of severe ROP requiring treatment, partly owing to a larger proportion of avascular retinal area at birth, and prediction models are not as useful in this cohort. Therefore, a simpler prediction model was developed for this group and is presented along with the results in eAppendix 1 (which references eFigures 9-11 and eTables 7 and 8) in the Supplement. Small for GA was defined as BWSDS less than −2.

retinal area at birth, and prediction models are not as useful in this cohort. Therefore, a simpler prediction model was developed for this group and is presented along with the results in eAppendix 1 (which references eFigures 9-11 and eTables 7 and 8) in the Supplement. Small for GA was defined as BWSDS less than −2. Study Outcome The prediction model was developed to estimate risk for treatment of sight-threatening ROP. The International Classification of Retinopathy of Prematurity and Early Treatment for Retinopathy of Prematurity (ETROP) criteria for treatment were used.

Introduction Retinopathy of prematurity (ROP) is a potentially blinding disease, and screening programs for detecting sight-threatening ROP needing treatment have been established worldwide. Infants with lower gestational age (GA) have a higher risk of sight-threatening ROP; in Sweden, the recommendation is to screen infants with GA less than 31 weeks and severely ill infants if older. Data are registered in the Swedish National Registry for Retinopathy of Prematurity (SWEDROP). Between 2008 and 2015, only 5.7% of screened infants in Sweden were treated for ROP. Screening includes retinal examinations by specially trained ophthalmologists and is often stressful for the infant; without risk prediction, some infants may not be screened and treated at the appropriate time. Individualized risk estimates would allow for optimization of timing and frequency of the screening processes from the health care and economics perspectives. Improving the timing of screening visits could avoid unnecessary examinations of low-risk infants and optimize identification of those at high risk.

priate time. Individualized risk estimates would allow for optimization of timing and frequency of the screening processes from the health care and economics perspectives. Improving the timing of screening visits could avoid unnecessary examinations of low-risk infants and optimize identification of those at high risk. Risk and severity of ROP vary by prenatal and postnatal factors, including poor prenatal and postnatal weight gain. For this reason, the prediction algorithm WINROP (weight, insulinlike growth factor 1, neonatal, ROP), which is based on accumulated postnatal weight gain, has been validated and broadly used. Similar tools based on longitudinal postnatal weight gain also have been developed. The objectives of this study were to create, then to internally and externally validate, and to describe the clinical implications of a prediction model for individual momentary and cumulative risks of ROP treatment based on birth characteristics alone, including infants born at GA less than 31 weeks.

e been developed. The objectives of this study were to create, then to internally and externally validate, and to describe the clinical implications of a prediction model for individual momentary and cumulative risks of ROP treatment based on birth characteristics alone, including infants born at GA less than 31 weeks. Methods Study Population Infants born between January 1, 2007, and August 7, 2018, at GA less than 31 weeks and with completed ROP screening registered in SWEDROP were included as part of the Swedish Neonatal Quality Register, started in 2007, which has approximately 97% coverage and contains perinatal data, screening outcomes, and treatment information. All data are registered through standardized protocols, in most settings by a trained pediatric ophthalmologist who has performed the screening examination. A validation of 85 randomly selected infants screened in 2018 showed 100% correctly reported values for variables used in this study. This retrospective cohort study was approved by the ethics committee at Uppsala University, Uppsala, Sweden, who also waived written informed consent because all the data were deidentified.

alidation of 85 randomly selected infants screened in 2018 showed 100% correctly reported values for variables used in this study. This retrospective cohort study was approved by the ethics committee at Uppsala University, Uppsala, Sweden, who also waived written informed consent because all the data were deidentified. Model Development Group In total, data for 8784 infants born between January 1, 2007, and October 31, 2017, were retrieved from SWEDROP for the prediction model development. Of those, data for 1372 of 8784 infants (15.6%) were excluded for having GA at least 31 weeks at birth, and 126 of 8784 infants (1.4%) were excluded for missing data. This left 7286 of 8784 infants (82.9%) eligible for the model development group. Of those, 6947 of 7286 infants (95.3%) had GA 24 to 30 weeks (Figure 1). Figure 1. Study Flowchart GA indicates gestational age; ROP, retinopathy of prematurity; and SWEDROP, Swedish National Registry for Retinopathy of Prematurity. Validation Groups The group used for temporal validation consisted of infants born between November 1, 2017, and August 7, 2018, and registered in SWEDROP. Among infants born at GA 24 to 30 weeks, 308 of 323 (95.4%) were eligible and served as the validation temporal group (Figure 1).

Figure 1. Study Flowchart GA indicates gestational age; ROP, retinopathy of prematurity; and SWEDROP, Swedish National Registry for Retinopathy of Prematurity. Validation Groups The group used for temporal validation consisted of infants born between November 1, 2017, and August 7, 2018, and registered in SWEDROP. Among infants born at GA 24 to 30 weeks, 308 of 323 (95.4%) were eligible and served as the validation temporal group (Figure 1). The validation US group included 1485 of 1535 eligible infants (96.7%) born at GA 24 to 30 weeks from 12 US centers between 2005 and 2010 (Figure 1). The validation European group included 329 of 354 eligible infants (92.9%) born at GA 24 to 30 weeks from Freiburg, Germany, with retrospective screening data collected between 2011 and 2017 (Figure 1).

retinal area at birth, and prediction models are not as useful in this cohort. Therefore, a simpler prediction model was developed for this group and is presented along with the results in eAppendix 1 (which references eFigures 9-11 and eTables 7 and 8) in the Supplement. Small for GA was defined as BWSDS less than −2. Study Outcome The prediction model was developed to estimate risk for treatment of sight-threatening ROP. The International Classification of Retinopathy of Prematurity and Early Treatment for Retinopathy of Prematurity (ETROP) criteria for treatment were used. Statistical Analysis General Methodology Number and percentage are given for categorical variables; for continuous variables, the mean, SD, median, range, and interquartile range are provided, where applicable. For comparison between 2 groups, we used the Fisher exact test for dichotomous variables, Mantel-Haenszel χ2 trend test for ordered categorical variables, and Mann-Whitney test for continuous variables. The Jonckheere-Terpstra test was applied for identifying trends between ordered categorical and continuous variables. The crude week-specific risk of ROP treatment (number of infants with the event divided by number of infants at risk) was analyzed based on postnatal age and postmenstrual age (GA plus postnatal age) by GA at birth. The modeling process consisted of (1) prediction model development, (2) internal and external validation, and (3) clinical implication. The prediction model for ROP treatment, called DIGIROP-Birth (Digital ROP), was developed using Poisson regression for time-varying data, from which we obtained a continuous hazard function, h(t), describing momentary risk for ROP treatment. From the hazard function, the survival function and its complement, the cumulative risk function F(t) = 1 − S(t), were estimated. The 95% CI for F(t) was obtained via repeated sampling (1000 samples) of the model parameters from a multivariate normal distribution using a covariance matrix estimated by the Poisson regression models. Parameter estimates, SEs, and hazard ratios (HRs) with 95% CIs are presented. The predictive ability of the continuous cumulative risks was checked and was found to be similarly high after postnatal age 15 weeks (eFigure 1 in the Supplement). Given this information and the knowledge about the studied hazard function, the cumulative risks of ever needing ROP treatment during 20 postnatal weeks were used for interpretation.

he continuous cumulative risks was checked and was found to be similarly high after postnatal age 15 weeks (eFigure 1 in the Supplement). Given this information and the knowledge about the studied hazard function, the cumulative risks of ever needing ROP treatment during 20 postnatal weeks were used for interpretation. All tests above were 2-tailed and conducted at the .05 significance level, with no adjustments for multiple comparisons. All analyses were performed using SAS statistical software, version 9.4 (SAS Institute Inc). DIGIROP-Birth Prediction Model for GA 24 to 30 Weeks Development and Validation Based on the crude risks for ROP treatment over time stratified by GA, we found that postnatal age was the most appropriate time axis. The final model for GA 24 to 30 weeks included the following: piecewise linear current postnatal age (break points, 8 and 12 weeks), piecewise linear continuous GA given in weeks and days (break point, 27 weeks), sex, piecewise linear BWSDS (break point, −1 SDS), postnatal age × piecewise linear GA interaction, sex × GA interaction, and postnatal age × piecewise linear BWSDS interaction. The break points for the variables were selected based on graphical review of univariable hazard functions. The final model was built by gradually expanding the models, starting only with postnatal age and further keeping interactions with P < .10.

GA interaction, and postnatal age × piecewise linear BWSDS interaction. The break points for the variables were selected based on graphical review of univariable hazard functions. The final model was built by gradually expanding the models, starting only with postnatal age and further keeping interactions with P < .10. Internal, temporal, and geographical external validations were performed. The model fit and adaptation were described by the area under the receiver operating characteristic curve (hereinafter referred to as AUC) overall, by calendar periods, and by race/ethnicity. We performed cross-validation and evaluated calibration plots; calculated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV); and compared DIGIROP-Birth with 4 other published prediction models (CHOP-ROP [Children’s Hospital of Philadelphia–ROP], OMA-ROP [Omaha-ROP], WINROP [weight, insulinlike growth factor 1, neonatal, ROP], and CO-ROP [Colorado-ROP]) using GA, birth weight, and different weight gain variables in the algorithms, as described in more detail in eAppendix 2 in the Supplement.

prediction models (CHOP-ROP [Children’s Hospital of Philadelphia–ROP], OMA-ROP [Omaha-ROP], WINROP [weight, insulinlike growth factor 1, neonatal, ROP], and CO-ROP [Colorado-ROP]) using GA, birth weight, and different weight gain variables in the algorithms, as described in more detail in eAppendix 2 in the Supplement. Results Study Population Birth characteristics for the whole SWEDROP cohort, model development group, and validation temporal group, as well as by maximum ROP stage, are listed in eTable 1 in the Supplement. Among 7609 patients, 4155 (54.6%) were boys, the mean (SD) GA was 28.1 (2.1) weeks, and the mean (SD) birth weight was 1119 [353] g. Of those born at GA at least 24 weeks, 1510 of 7255 (20.8%) were small for GA. In total, 354 of 7609 (4.7%) were born at GA less than 24 weeks, and 2806 of 7609 (36.9%) were born at GA 24 to less than 28 weeks. Birth characteristics were numerically balanced between the model development group and the validation temporal group. Birth characteristics for the validation US group and the validation European group are listed in eTable 2 in the Supplement.

4 weeks, and 2806 of 7609 (36.9%) were born at GA 24 to less than 28 weeks. Birth characteristics were numerically balanced between the model development group and the validation temporal group. Birth characteristics for the validation US group and the validation European group are listed in eTable 2 in the Supplement. ROP Treatment Incidence in Screened Infants Altogether, 2427 of 7609 infants (31.9%) developed any ROP, which regressed spontaneously in 1985 of 7609 (26.1%) and was treated in 442 of 7609 (5.8%) (eTable 3 in the Supplement). Among infants with GA less than 24 weeks, 142 of 354 (40.1%) were treated, 287 of 2806 (10.2%) among those with GA 24 to less than 28 weeks and 13 of 4449 (0.3%) among those with GA at least 28 weeks. The incidence of ROP treatment for infants born at GA 24 to 30 weeks was 125 of 1485 (8.4%) in the validation US group and 17 of 329 (5.2%) in the validation European group. Momentary Individual Risk of ROP Treatment for GA Less Than 31 Weeks Figure 2A and B show crude week-specific risk of ROP treatment for the SWEDROP population. Table 1 lists the observed timing for ROP treatment applying postnatal age and postmenstrual age as time axes. The ROP treatment risk peaked at postnatal week 12 regardless of GA at birth, but no specific pattern by GA was seen for postmenstrual age. Figure 2. Crude Week-Specific and Momentary Individual Risk of Retinopathy of Prematurity (ROP) Treatment Shown is risk for gestational age (GA) less than 31 weeks.

Momentary Individual Risk of ROP Treatment for GA Less Than 31 Weeks Figure 2A and B show crude week-specific risk of ROP treatment for the SWEDROP population. Table 1 lists the observed timing for ROP treatment applying postnatal age and postmenstrual age as time axes. The ROP treatment risk peaked at postnatal week 12 regardless of GA at birth, but no specific pattern by GA was seen for postmenstrual age. Figure 2. Crude Week-Specific and Momentary Individual Risk of Retinopathy of Prematurity (ROP) Treatment Shown is risk for gestational age (GA) less than 31 weeks. Table 1. Comparison Between US Guidelines (Fierson et al) Regarding Timing of Initial Examination vs SWEDROP Data and DIGIROP-Birth Model, 2007-2018 GA at Birth, wk Postmenstrual Age at Initial Examination, wk Chronological (Postnatal) Age at Initial Examination, wk Fierson et al SWEDROP 2007-2018 Suggested Agea SWEDROP 2007-2018 Observed Age for ROP Treatment Maximum Age for Estimated Cumulative Risk <.001b Fierson et al SWEDROP 2007-2018 Suggested Agea SWEDROP 2007-2018 Observed Age for ROP Treatment Maximum Age for Estimated Cumulative Risk <.001b 21 NR 31 NA NA NR 10 NA NA Mean (SD) NA NA 34.6 (1.7) NA NA NA 12.8 (1.7) NA Median (range) NA NA 34.6 (33.4-35.9) NA NA NA 12.8 (11.6-14.0) NA No./total No. NA NA 2/2 NA NA NA 2/2 NA 22 31 31 NA NA 9 9 NA NA Mean (SD) NA NA 36.3 (3.2) NA NA NA 13.6 (3.2) NA Median (range) NA NA 35.1 (32.6-47.1) NA NA NA 12.4 (10.0-24.3) NA No./total No. NA NA 39/82 NA NA NA 39/82 NA 23 31 31 NA NA 8 8 NA NA Mean (SD) NA NA 36.5 (2.9) NA NA NA 13.1 (2.8) NA Median (range) NA NA 36.0 (32.9-51.4) NA NA NA 12.6 (9.4-28.3) NA No./total No. NA NA 101/270 NA NA NA 101/270 NA 24 31 31 NA 30.2 7 7 NA 6.2 Mean (SD) NA NA 37.1 (2.5) NA NA NA 12.7 (2.5) NA Median (range) NA NA 36.6 (32.4-45.7) NA NA NA 12.3 (8.3-21.4) NA No./total No. NA NA 117/436 NA NA NA 117/436 NA 25 31 31 NA 31.7 6 6 NA 6.7 Mean (SD) NA NA 38.2 (2.9) NA NA NA 12.9 (2.9) NA Median (range) NA NA 37.7 (33.1-47.0) NA NA NA 12.4 (7.4-21.9) NA No./total No. NA NA 92/620 NA NA NA 92/620 NA 26 31 32 NA 33.2 5 6 NA 7.2 Mean (SD) NA NA 39.7 (3.3) NA NA NA 13.3 (3.3) NA Median (range) NA NA 39.3 (33.1-52.1) NA NA NA 13.0 (7.0-25.6) NA No./total No. NA NA 58/801 NA NA NA 58/801 NA 27 31 33 NA 34.7 4 6 NA 7.7 Mean (SD) NA NA 40.3 (2.8) NA NA NA 12.9 (2.9) NA Median (range) NA NA 40.1 (35.7-45.3) NA NA NA 12.6 (7.9-17.7) NA No./total No. NA NA 20/949 NA NA NA 20/949 NA 28 32 34 NA 35.7 4 6 NA 7.7 Mean (SD) NA NA 40.8 (3.8) NA NA NA 12.4 (3.9) NA Median (range) NA NA 39.4 (36.1-47.7) NA NA NA 11.1 (7.6-18.9) NA No./total No.

Mean (SD) NA NA 40.3 (2.8) NA NA NA 12.9 (2.9) NA Median (range) NA NA 40.1 (35.7-45.3) NA NA NA 12.6 (7.9-17.7) NA No./total No. NA NA 20/949 NA NA NA 20/949 NA 28 32 34 NA 35.7 4 6 NA 7.7 Mean (SD) NA NA 40.8 (3.8) NA NA NA 12.4 (3.9) NA Median (range) NA NA 39.4 (36.1-47.7) NA NA NA 11.1 (7.6-18.9) NA No./total No. NA NA 10/1179 NA NA NA 10/1179 NA 29 33 36 NA 37.3 4 7 NA 8.1 Mean (SD) NA NA 39.7 (2.2) NA NA NA 10.1 (2.3) NA Median (range) NA NA 39.4 (37.6-42.0) NA NA NA 9.7 (8.0-12.6) NA No./total No. NA NA 3/1479 NA NA NA 3/1479 NA 30 34 37 NA 38.8 4 7 NA 8.7 No./total No. NA NA 0/1791 NA NA NA 0/1791 NA Abbreviations: DIGIROP, Digital ROP; GA, gestational age; NA, not applicable; NR, not reported; ROP, retinopathy of prematurity; SWEDROP, Swedish National Registry for Retinopathy of Prematurity. a Suggested age defined as integer value of the minimum time to ROP treatment subtracted by 1 week for safety reasons. b Given the SWEDROP population, DIGIROP-Birth model for GA 24 to 30 weeks, with its sex and birth weight SD score distribution. From the Poisson regression model based on the total SWEDROP population, including postnatal age and adjusting for GA, the risk for ROP treatment increased by 54% (HR, 1.54; 95% CI, 1.39-1.70) per week from postnatal weeks 8 through 12. Afterward, it decreased by 30% (HR, 0.70; 95% CI, 0.67-0.74) per week (Figure 2C and eTable 4 in the Supplement).

del based on the total SWEDROP population, including postnatal age and adjusting for GA, the risk for ROP treatment increased by 54% (HR, 1.54; 95% CI, 1.39-1.70) per week from postnatal weeks 8 through 12. Afterward, it decreased by 30% (HR, 0.70; 95% CI, 0.67-0.74) per week (Figure 2C and eTable 4 in the Supplement). Cumulative Individual Risk of ROP Treatment for GA 24 to 30 Weeks Table 2 summarizes the final DIGIROP-Birth model for ROP treatment in infants born at GA 24 to 30 weeks. The estimated cumulative risks were 60.0% and 35.1%, respectively, for a girl with BWSDS −3 and 0 born at GA 24 weeks and were 27.8% and 14.2%, respectively, if she was born at GA 25 weeks (Figure 3 and eFigure 2 and eTable 5 in the Supplement). Corresponding figures for a boy with the same background data were 57.7% and 33.4%, respectively, and 32.5% and 16.9%, respectively. Greater decreasing risk was observed for girls than for boys with increasing GA (P for interaction = .02), with HRs of 0.83 (95% CI, 0.64-1.07) at 25 weeks and 0.50 (95% CI, 0.33-0.76) at 27 weeks (crude incidences are shown in eFigure 3 in the Supplement, and predicted cumulative risks are shown in eFigure 4 in the Supplement). The cumulative risk estimates with 95% CIs are available online for public use, requiring input of GA in weeks and days, sex, and birth weight for the infant.

CI, 0.33-0.76) at 27 weeks (crude incidences are shown in eFigure 3 in the Supplement, and predicted cumulative risks are shown in eFigure 4 in the Supplement). The cumulative risk estimates with 95% CIs are available online for public use, requiring input of GA in weeks and days, sex, and birth weight for the infant. Table 2. Final Prediction Analysis Model for Retinopathy of Prematurity Treatment for Infants Born at GA of 24 to 30 Weeks Using Poisson Regression for Time-Varying Data Predictor Estimate (SE) P Value Intercept −20.1666 (4.9219) <.001 Postnatal age 0 to 8 wk, per 1-wk increase 1.7331 (0.6129) .005 Postnatal age >8 to 12 wk, per 1-wk increase 0.3618 (0.0992) <.001 Postnatal age >12 wk, per 1-wk increase −0.3788 (0.0857) <.001 GA 24-27 wk, per 1-wk increase −0.8210 (0.3353) .01 GA >27 wk, per 1-wk increase 0.7266 (0.7302) .32 Sex, 1 = boys, 2 = girls −0.9385 (0.3054) .002 BWSDS −1 SDS or less, per 1-SDS increase 0.1521 (0.2656) .57 BWSDS exceeding −1 SDS, per 1-SDS increase −1.0401 (0.4710) .03 INT: postnatal age in weeks by GA 24-27 wk, per 1-wk increase 0.0227 (0.0230) .32 INT: postnatal age in weeks by GA >27 wk, per 1-wk increase −0.1360 (0.0627) .03 INT: sex by GA, per 1-wk increase −0.2505 (0.1066) .02 INT: postnatal age in weeks by BWSDS −1 SDS or less −0.0371 (0.0199) .06 INT: postnatal age in weeks by BWSDS exceeding −1 SDS 0.0728 (0.0349) .04 Abbreviations: BWSDS, birth weight standard deviation score; GA, gestational age; INT, interaction term; SDS, SD score.

NT: sex by GA, per 1-wk increase −0.2505 (0.1066) .02 INT: postnatal age in weeks by BWSDS −1 SDS or less −0.0371 (0.0199) .06 INT: postnatal age in weeks by BWSDS exceeding −1 SDS 0.0728 (0.0349) .04 Abbreviations: BWSDS, birth weight standard deviation score; GA, gestational age; INT, interaction term; SDS, SD score. Figure 3. Cumulative Individual Risk for Retinopathy of Prematurity (ROP) Treatment Shown is cumulative risk (95% CI) by gestational age (GA) 24 to 30 weeks for boys and girls born with birth weight SD score (BWSDS) −3 and 0.

NT: sex by GA, per 1-wk increase −0.2505 (0.1066) .02 INT: postnatal age in weeks by BWSDS −1 SDS or less −0.0371 (0.0199) .06 INT: postnatal age in weeks by BWSDS exceeding −1 SDS 0.0728 (0.0349) .04 Abbreviations: BWSDS, birth weight standard deviation score; GA, gestational age; INT, interaction term; SDS, SD score. Figure 3. Cumulative Individual Risk for Retinopathy of Prematurity (ROP) Treatment Shown is cumulative risk (95% CI) by gestational age (GA) 24 to 30 weeks for boys and girls born with birth weight SD score (BWSDS) −3 and 0. Internal and External Validation of DIGIROP-Birth for GA 24 to 30 Weeks eFigure 5 in the Supplement shows AUCs from the internal and external validations, indicating whether the model discriminates well between receiving or not receiving treatment. The AUC for the model development group was 0.90 (95% CI, 0.89-0.92), and the AUC for the cross-validation model was 0.90 (95% CI, 0.89-0.91). The AUCs for different calendar periods ranged from 0.87 to 0.92. The calibration plots, examining overestimation or underestimation of risks in different regions, showed the model as being overall well adapted (eFigure 6 in the Supplement). Temporal validation of DIGIROP-Birth showed an AUC of 0.94 (95% CI, 0.90-0.98). Geographical external validation resulted in an AUC of 0.87 (95% CI, 0.84-0.89) for the validation US group and an AUC of 0.90 (95% CI, 0.85-0.95) for the validation European group. The AUCs for stratified analysis on race/ethnicity categories in the validation US group were 0.79 for Hispanic infants, 0.85 for Asian infants, 0.86 for non-Hispanic infants, 0.88 for white infants, and 0.90 for black infants.

alidation US group and an AUC of 0.90 (95% CI, 0.85-0.95) for the validation European group. The AUCs for stratified analysis on race/ethnicity categories in the validation US group were 0.79 for Hispanic infants, 0.85 for Asian infants, 0.86 for non-Hispanic infants, 0.88 for white infants, and 0.90 for black infants. DIGIROP-Birth for GA 24 to 30 Weeks vs Existing ROP Models (Requiring Postnatal Longitudinal Data) The comparisons of DIGIROP-Birth vs CHOP-ROP, OMA-ROP, WINROP, and CO-ROP were performed on the validation US group, enabling the use of longitudinal weight data. These results are summarized in eFigures 7 and 8 and eTable 6 in the Supplement. Applying the CHOP-ROP algorithm (AUC, 0.89; 95% CI, 0.87-0.92) and categorizing the probabilities based on the recommended cutoff of 0.0140, similar prediction ability was observed compared with DIGIROP-Birth (AUC, 0.88; 95% CI, 0.86-0.91), and a cutoff of 0.0083 obtained the same sensitivity (95 of 96 [99.0%] for both CHOP-ROP and DIGIROP-Birth. Specificity was 598 of 1346 (44.4%) vs 658 of 1346 (48.9%), respectively. Applying the same cutoff on the complete SWEDROP database, the model showed 97.7% (95% CI, 95.3%-99.1%) sensitivity and 59.5% (95% CI, 58.4%-60.7%) specificity. Applying a cutoff of 0.00083 for 100% (95% CI, 98.8%-100%) sensitivity in the cohort, a specificity of 19.0% (95% CI, 18.1%-20.0%) was obtained.

respectively. Applying the same cutoff on the complete SWEDROP database, the model showed 97.7% (95% CI, 95.3%-99.1%) sensitivity and 59.5% (95% CI, 58.4%-60.7%) specificity. Applying a cutoff of 0.00083 for 100% (95% CI, 98.8%-100%) sensitivity in the cohort, a specificity of 19.0% (95% CI, 18.1%-20.0%) was obtained. Compared with OMA-ROP (AUC, 0.77; 95% CI, 0.72-0.82), a cutoff of 23 g per day in weight gain, with a corresponding cutoff of 0.0200 for DIGIROP-Birth (AUC, 0.90; 95% CI, 0.87-0.92), a sensitivity of 90 of 92 (97.8%) was obtained. Specificity was 173 of 771 (22.4%) for OMA-ROP vs 448 of 771 (58.1%) for DIGIROP-Birth. Compared with WINROP (AUC, 0.81; 95% CI, 0.78-0.84), the alarm category of WINROP score 2 or 3 provided a sensitivity of 121 of 125 (96.8%), with a corresponding cutoff of 0.0089 for DIGIROP-Birth (AUC, 0.87; 95% CI, 0.84-0.89). Specificity was 487 of 1360 (35.8%) for WINROP vs 671 of 1360 (49.3%) for DIGIROP-Birth. The specificity was 141 of 1341 (10.5%) for the CO-ROP algorithm and 642 of 1341 (47.9%) for DIGIROP-Birth. Both had a sensitivity of 122 of 124 (98.4%).

Compared with WINROP (AUC, 0.81; 95% CI, 0.78-0.84), the alarm category of WINROP score 2 or 3 provided a sensitivity of 121 of 125 (96.8%), with a corresponding cutoff of 0.0089 for DIGIROP-Birth (AUC, 0.87; 95% CI, 0.84-0.89). Specificity was 487 of 1360 (35.8%) for WINROP vs 671 of 1360 (49.3%) for DIGIROP-Birth. The specificity was 141 of 1341 (10.5%) for the CO-ROP algorithm and 642 of 1341 (47.9%) for DIGIROP-Birth. Both had a sensitivity of 122 of 124 (98.4%). Clinical Practice Implications of DIGIROP-Birth for GA 24 to 30 Weeks Based on ROP treatment timing in the SWEDROP cohort (2007-2018) and the DIGIROP-Birth model, we compared the results with the current US recommendations, based on studies that are more than 20 years old, for postnatal age and postmenstrual age at initial examination (Table 1). The maximum age for estimated risk less than 0.001 corresponds well to the observed minimum age for ROP treatment, except for GA 24 weeks, for which a somewhat higher risk at a younger age was estimated. Recommending that the initial examination should start 1 week before the earliest observed ROP treatment per GA week in our cohort would potentially have avoided 14 867 of 135 061 visits (11.0%), assuming 1 visit per week. For GA of at least 27 weeks, with a ROP treatment incidence of 33 of 5398 (0.6%), the difference between the US recommendations and this study resulted in 14 066 of 93 052 examinations (15.1%) potentially being avoided.

r cohort would potentially have avoided 14 867 of 135 061 visits (11.0%), assuming 1 visit per week. For GA of at least 27 weeks, with a ROP treatment incidence of 33 of 5398 (0.6%), the difference between the US recommendations and this study resulted in 14 066 of 93 052 examinations (15.1%) potentially being avoided. Discussion We have created and validated the DIGIROP-Birth prediction model, available free of charge online based on 6947 infants born at GA 24 to 30 weeks, estimating the individual momentary and cumulative risks for ROP treatment. The model using only available data at birth but more advanced statistical methods was at least as accurate as 4 of the ROP prediction models now in use based on longitudinal weight measurements, which are not always readily available to ophthalmologists.

the individual momentary and cumulative risks for ROP treatment. The model using only available data at birth but more advanced statistical methods was at least as accurate as 4 of the ROP prediction models now in use based on longitudinal weight measurements, which are not always readily available to ophthalmologists. Surprisingly, the momentary risk of ROP treatment peaked at 12 weeks’ postnatal age regardless of GA at birth, while no specific pattern was observed for postmenstrual age. This observation is particularly interesting because the ETROP study found that the progression of prethreshold ROP was highly associated with postmenstrual age, similar to the finding in the CRYO-ROP (Cryotherapy for ROP) study 15 years earlier. However, it should be emphasized that infants included in the CRYO-ROP study were born at higher GA, and no GA-specific hazard functions were studied for ROP outcome. Other Swedish studies have reported that lower GA at birth is associated with lower GA at treatment, but the momentary risk in relation to postnatal age and postmenstrual age was not analyzed. Recently, in a large North American cohort, the timing of ROP treatment was presented only in relation to postmenstrual age and not postnatal age.

es have reported that lower GA at birth is associated with lower GA at treatment, but the momentary risk in relation to postnatal age and postmenstrual age was not analyzed. Recently, in a large North American cohort, the timing of ROP treatment was presented only in relation to postmenstrual age and not postnatal age. The identification of a peak risk at 12 postnatal weeks in infants with GA less than 31 weeks might be clinically useful because it was recently shown that inadequate screening or treatment was identified in 11 of 17 cases with blindness from ROP (64.7%). Hence, clinicians and parents could be alerted during this period to ensure that timely screening occurs to reduce the risk of blindness. National patient registries are valuable sources for estimation of treatment risks. Herein, the DIGIROP-Birth model was compared with a validation US group and a validation European group and showed high predictive ability and generalizability both for individuals with the same and with different reported race/ethnicity.

National patient registries are valuable sources for estimation of treatment risks. Herein, the DIGIROP-Birth model was compared with a validation US group and a validation European group and showed high predictive ability and generalizability both for individuals with the same and with different reported race/ethnicity. The ROP prediction models may also be used to reduce screening frequency in infants at low risk. The latest US policy statement for ROP screening was issued in 2018. The recommendations for the timing of the first examination were based on the CRYO-ROP study published in 1991 and the LIGHT-ROP (Light Reduction in ROP) study published in 1998. In those periods, fewer extremely preterm infants survived, more mature infants were treated, and treatment criteria were different from those used today. Based on the results of our study, if the initial examination was performed 1 week before the earliest observed postnatal age at ROP treatment, 14 867 of 135 061 stressful early examinations (11.0%) could be avoided (assuming 1 examination per week) compared with US recommendations. For GA of at least 27 weeks, with a ROP treatment incidence herein less than 1%, 14 066 of 93 052 examinations (15.1%) could have been avoided while capturing all cases of ROP treatment (100% sensitivity). Notably, reaching 100% sensitivity in such models of real-life, large data sets is accompanied by low specificity. Based on approximately as large a cohort as in our study, the updated CHOP-ROP model, which uses longitudinal weight data and birth data, achieved 11.2% specificity for 100% sensitivity and 36.4% specificity for 98.5% sensitivity; DIGIROP-Birth (using only readily obtained birth data) showed 19.0% specificity for 100% sensitivity and 53.8% specificity for 99.0% sensitivity.

, the updated CHOP-ROP model, which uses longitudinal weight data and birth data, achieved 11.2% specificity for 100% sensitivity and 36.4% specificity for 98.5% sensitivity; DIGIROP-Birth (using only readily obtained birth data) showed 19.0% specificity for 100% sensitivity and 53.8% specificity for 99.0% sensitivity. Strengths and Limitations The strengths of our study include the unique and complete cohort of preterm infants born in Sweden between January 2007 and August 2018. Also, our statistical model includes 3 basic measurements (GA, sex, and birth weight). The postnatal age for ROP treatment or censoring (discontinued follow-up) is included in the hazard function estimation but is not required as an input variable. Hence, the input data are simple, facilitating their general use, even though the method is more advanced, taking into account the underlying hazard function and the important interactions that contribute to adjustment of heterogeneity, which is novel in ROP research. The DIGIROP-Birth has shown strong predictive ability in internal, temporal, and geographical external validations. If found not acceptable in future validations among a population, a subgroup-specific model designed for optimal predictions in that population might be developed using our methods. Finally, DIGIROP-Birth has been shown to be equal to or better than 4 other ROP prediction models and is accessible online.

l validations. If found not acceptable in future validations among a population, a subgroup-specific model designed for optimal predictions in that population might be developed using our methods. Finally, DIGIROP-Birth has been shown to be equal to or better than 4 other ROP prediction models and is accessible online. Our study has some limitations. One limitation is the use of registry retrospective data. However, the registry showed high coverage and successful validation of data for 85 randomly selected infants screened in 2018. In addition, infants born at GA less than 24 weeks could not be included in the prediction model because of the lack of a reference algorithm for birth weight, preventing BWSDS calculations. Given the small sample size, only a simple model could be developed for these infants, resulting in low predictive ability. Close monitoring of such infants is mandatory irrespective of calculated risk, making prediction models less important for this group.

ce algorithm for birth weight, preventing BWSDS calculations. Given the small sample size, only a simple model could be developed for these infants, resulting in low predictive ability. Close monitoring of such infants is mandatory irrespective of calculated risk, making prediction models less important for this group. Conclusions We created and validated the DIGIROP-Birth model, an individualized early prediction model for infants with GA 24 to 30 weeks, which estimates momentary and cumulative risks for receiving ROP treatment based on simple birth characteristics. A surprising finding was that postnatal age was the best predictive variable for the temporal risk of ROP treatment. The DIGIROP-Birth model is an accessible online application that appears to be generalizable and to have at least as good test statistics as other models that require longitudinal neonatal data, which are not always readily available to ophthalmologists. Supplement. eAppendix 1. Prediction Model for Gestational Age <24 Weeks eAppendix 2. Internal and External Validation eFigure 1. AUC for DIGIROP-Birth Model Performed on Model Group for Cumulative Probabilities Estimated Over the Postnatal Age eFigure 2. Momentary and Cumulative Individual Risk With 95% CI Over Time for ROP Treatment for Gestational Age 24 and 25 Weeks, by Sex for Different Birth WeightSDS eFigure 3. Incidence of ROP Treatment by Gestational Age, Sex, and Birth WeightSDS (BWSDS) eFigure 4. Estimated Risk (%) for ROP Treatment by Gestational Age and Sex

eFigure 2. Momentary and Cumulative Individual Risk With 95% CI Over Time for ROP Treatment for Gestational Age 24 and 25 Weeks, by Sex for Different Birth WeightSDS eFigure 3. Incidence of ROP Treatment by Gestational Age, Sex, and Birth WeightSDS (BWSDS) eFigure 4. Estimated Risk (%) for ROP Treatment by Gestational Age and Sex eFigure 5. ROC Curves Obtained Based on Cumulative Individual Risk for ROP Treatment for Gestational Age at Birth ≥24 Weeks From Model Group, Validation Temporal Group, Validation US Group and Validation European Group (A), by Calendar Periods in the Model Group (B), and by Race/Ethnicity in the Validation US Group (C) eFigure 6. Calibration Plot for Observed Proportion of ROP Treatment vs Estimated Probability Obtained From the Final Prediction Model for Gestational Age at Birth ≥24 Weeks (A), and From Cross-Validation Model for Gestational Age ≥24 Weeks (B) eFigure 7. ROC Curves Obtained Based on Cumulative Individual Risk for ROP Treatment for Gestational Age at Birth ≥24 Weeks From Validation US Group. DIGIROP-Birth vs CHOP-ROP (A), DIGIROP-Birth vs OMA-ROP (B), DIGIROP-Birth vs WINROP (C), and DIGIROP-Birth vs CO-ROP (D) eFigure 8. Scatterplots for Individual Risk Predictions for ROP Treatment Performed on Validation US Group, Obtained From DIGIROP-Birth and CHOP-ROP Models (A), DIGIROP-Birth and CHOP-ROP Models Zoom-in Figure for Probabilities 0.0-0.1 (B), DIGIROP-Birth and OMA-ROP Models (C), and DIGIROP-Birth and WINROP Models (D). eFigure 9. Cumulative Individual Risk With 95% CI for ROP Treatment for Gestational Age at Birth <24 Weeks

eFigure 8. Scatterplots for Individual Risk Predictions for ROP Treatment Performed on Validation US Group, Obtained From DIGIROP-Birth and CHOP-ROP Models (A), DIGIROP-Birth and CHOP-ROP Models Zoom-in Figure for Probabilities 0.0-0.1 (B), DIGIROP-Birth and OMA-ROP Models (C), and DIGIROP-Birth and WINROP Models (D). eFigure 9. Cumulative Individual Risk With 95% CI for ROP Treatment for Gestational Age at Birth <24 Weeks eFigure 10. ROC Curves Obtained Based on Cumulative Individual Risk for ROP Treatment for Gestational Age at Birth <24 Weeks From the Main Study, Cross-Validation, and External Population Database (A) and by Calendar Periods for the Main Study Database (B) eFigure 11. Calibration Plot for Observed Proportion of ROP Treatment vs Estimated Probability Obtained From the Final Prediction Model for Gestational Age at Birth <24 Weeks (A), and From the Cross-Validation Model for Gestational Age <24 Weeks (B) eTable 1. Birth Characteristics for Total SWEDROP Cohort by Study Population and by Maximum ROP Stage eTable 2. Birth Characteristics for Validation US Group and Validation European Group eTable 3. Number, Percentage, and Follow-up Weeks for ROP Treatment, by Sex, Gestational Age at Birth, and Birth WeightSDS eTable 4. Prediction Models for ROP Treatment for Total SWEDROP Cohort Using Only Postnatal Age and Gestational Age – Poisson Regression for Time-Varying Data

eTable 2. Birth Characteristics for Validation US Group and Validation European Group eTable 3. Number, Percentage, and Follow-up Weeks for ROP Treatment, by Sex, Gestational Age at Birth, and Birth WeightSDS eTable 4. Prediction Models for ROP Treatment for Total SWEDROP Cohort Using Only Postnatal Age and Gestational Age – Poisson Regression for Time-Varying Data eTable 5. Estimated Probability for ROP Treatment With 95% CI for Selected Values of Birth WeightSDS, Gestational Age at Birth, Sex, and Postnatal Age – Final Poisson Regression Model for Gestational Age at Birth ≥24 Weeks eTable 6. DIGIROP-Birth vs CHOP-ROP, OMA-ROP, WINROP and CO-ROP. AUC, Sensitivity, Specificity, PPV and NPV eTable 7. Final Prediction Model for ROP Treatment for Infants With Gestational Age at Birth <24 Weeks – Poisson Regression for Time-Varying Data eTable 8. Estimated Probability for ROP Treatment With 95% CI for Selected Values of Birth Weight, Gestational Age at Birth, Sex, and Postnatal Age – Final Poisson Regression Model for Gestational Age at Birth <24 Weeks eReferences Click here for additional data file.