Considerable evidence supports the hypothesis

Transcription

1 Relation of Consumption of Vitamin E, Vitamin C, and Carotenoids to Risk for Stroke among Men in the United States Alberto Ascherio, MD, DrPH; Eric B. Rimm, ScD; Miguel A. Hernán, MD, MPH; Edward Giovannucci, MD, ScD; Ichiro Kawachi, MD, DrPH; Meir J. Stampfer, MD, DrPH; and Walter C. Willett, MD, DrPH Background: Antioxidants increase the resistance of lowdensity lipoprotein to oxidation and may thereby reduce risk for atherosclerosis. Objective: To determine whether intake of vitamin E, vitamin C, or carotenoids predict risk for total or ischemic stroke. Design: Prospective observational study. Setting: The Health Professionals Follow-up Study. Participants: men 40 to 75 years of age who did not have cardiovascular disease or diabetes. Measurements: Repeated and validated dietary assessments were done by using a self-administered 131-item food-frequency questionnaire, which included questions on dose and duration of vitamin supplement use. The follow-up period was 8 years. Results: A total of 328 strokes occurred: 210 ischemic, 70 hemorrhagic, and 48 unclassified. After adjustment for age, smoking, hypertension, hypercholesterolemia, body mass index, physical activity, parental history of myocardial infarction, alcohol consumption, and total energy intake, the relative risk for ischemic stroke in the top quintile of vitamin E intake (median, 411 IU/d) compared with the bottom quintile (5.4 IU/d) was 1.18 (95% CI, 0.77 to 1.82). The relative risk for ischemic stroke in the top quintile of vitamin C intake (1167 mg/d) compared with the bottom quintile (95 mg/d) was 1.03 (CI, 0.66 to 1.59). Results for total stroke were similar. Associations of vitamin intake with hemorrhagic stroke were also nonsignificant, but the CIs were wide. Neither dose nor duration of vitamin E or vitamin C supplement use was related to risk for total or ischemic stroke. The relative risk for ischemic stroke was 1.16 (CI, 0.81 to 1.67) in men using 250 IU or more of vitamin E supplementation per day compared with men who used no vitamin E supplements and was 0.93 (CI, 0.60 to 1.45) in men using 700 mg or more of vitamin C supplementation per day compared with men who used no vitamin C supplements. A significant inverse relation between lutein intake and risk for ischemic stroke was seen but was not independent of other dietary factors. Conclusions: Vitamin E and vitamin C supplements and specific carotenoids did not seem to substantially reduce risk for stroke in this cohort. Modest effects, however, cannot be excluded. Ann Intern Med. 1999;130: From Harvard School of Public Health, Harvard Medical School, and Brigham and Women s Hospital, Boston, Massachusetts. For current author addresses, see end of text. Considerable evidence supports the hypothesis that vitamin E reduces the risk for atherosclerosis (1). Vitamin E inhibits the proliferation of smooth-muscle cells in vitro (2) and increases the resistance of low-density lipoprotein to oxidation when it is added to plasma or administered to humans (3, 4). Furthermore, platelets from persons taking vitamin E supplements had markedly reduced adhesiveness to collagen (5). Consistent with these findings, vitamin E intake has been inversely associated with risk for coronary heart disease (6 8) and for peripheral arterial disease (9) and with carotid artery wall thickness (10, 11). Risk for ischemic stroke also increases with atherosclerosis and might therefore be reduced by vitamin E. However, few studies have addressed the relation between vitamin E and risk for ischemic stroke, and those that have been done were small, had little variation in vitamin E intake, or did not distinguish between ischemic and hemorrhagic stroke (12 15). Risk for hemorrhagic stroke is unlikely to be reduced by high doses of vitamin E and could even be increased by the reduced platelet adhesion that these doses cause (5). The most informative previous studies a large randomized trial in men (16, 17) and a prospective investigation in women (6) supported a modest beneficial effect of vitamin E with respect to ischemic stroke, but the results were not statistically significant. Moreover, in the trial in men (16), this apparent benefit was offset by an increased risk for hemorrhagic stroke. Like vitamin E, vitamin C and carotenoids have antioxidant properties (18, 19) and have been hypothesized to reduce risk for stroke, but epidemiologic evidence to support these hypotheses is limited (14 17, 20 22). Of the carotenoids, we considered -carotene, -carotene, lutein, and lycopene because they are present in relatively large amounts in the diet and have distinct antioxidant properties (23). Our primary hypothesis was that high intake of these antioxidants would reduce risk for ischemic stroke. We report here the associations between intakes of vitamin E, vitamin C, and specific carotenoids and risks for ischemic and hemorrhagic stroke in a large cohort of men in the United States American College of Physicians American Society of Internal Medicine 963

2 Study Sample Methods The Health Professionals Follow-up Study began in 1986, when health professionals 40 to 75 years of age (24) completed a 131-item food-frequency questionnaire and supplied information about medical history and lifestyle. Follow-up questionnaires were sent every 2 years so that we could update information on potential risk factors, including use of vitamin supplements, and identify newly diagnosed cases of stroke and other diseases. We excluded from analysis 1595 men who did not have 1) a daily caloric intake between 800 and 4200 kcal and 2) fewer than 70 blanks (among 131 listed food items) on the food-frequency questionnaire. We also excluded men with previous myocardial infarction, angina, coronary artery surgery, stroke, transient ischemic attack, peripheral arterial disease, or diabetes. We followed the eligible men for incidence of stroke for 8 years. The average response rate for the 2-year follow-up cycles was more than 94%. Nonresponding participants who were not matched to the National Death Index were assumed to be alive. This study was approved by the institutional review board of the Harvard School of Public Health. Assessment of Diet and Other Exposure Variables On the 1986 questionnaire, we asked about 1) the average frequency of intake over the previous year of specified portions of 131 foods and 2) the use of vitamin and mineral supplements. We specifically asked about dose and duration of use of multivitamin, vitamin E, and vitamin C supplements. Nutrient calculations took into account the specific brands of breakfast cereal and multivitamin supplements reported by each participant. We assessed the questionnaire s validity in a random sample of 127 men who completed two 1-week diet records (25); 121 of these men also provided blood samples (26, 27). The correlations between the food-frequency assessments of intake and plasma concentrations were 0.51 for -tocopherol, 0.47 for -carotene and lycopene, 0.35 for -carotene, and 0.40 for lutein. These correlations were similar to those seen between plasma levels of these nutrients and intake of these nutrients as estimated by the two 1-week diet records (25), and they support the validity of the food-frequency questionnaire. Even if dietary measurements were perfect, vitamin intake and plasma levels would not be highly correlated because multiple factors contribute to between-person variations in plasma vitamin levels. The correlation between the food-frequency and diet-record assessments was 0.92 for both vitamin C and vitamin E (25). In 1986, in addition to reporting on diet, participants were asked to report their usual systolic and diastolic blood pressures and whether they had physiciandiagnosed hypertension. The validity of these variables has been documented (28). Other potential risk factors for stroke that were assessed at baseline and in the follow-up questionnaire included smoking history, body mass index, history of hypercholesterolemia, physical activity, parental history of myocardial infarction, and alcohol consumption. Events End points were fatal or nonfatal stroke occurring between the return of the baseline questionnaire and 31 January If a participant reported an incident stroke on a follow-up questionnaire, we asked for permission to review that participant s medical records. Strokes were confirmed if they were characterized by a typical neurologic defect of sudden or rapid onset that lasted at least 24 hours and was attributable to a cerebrovascular event. Strokes caused by infection or neoplasia were excluded. Reviews were conducted by physicians who had no knowledge of participants risk factor status. Strokes were subclassified, according to the criteria of the National Survey of Stroke, as due to ischemia (embolism or thrombosis), subarachnoid hemorrhage, intracerebral hemorrhage, or an unknown cause (29). If no records could be obtained, a stroke was considered probable if it necessitated hospitalization and was corroborated by additional information provided in a letter or interview. Deaths were initially reported by next-of-kin, coworkers, postal authorities, or the National Death Index. Fatal strokes were confirmed by medical records (84%) or autopsy reports (2%) or were considered probable if medical records or autopsy reports could not be obtained and stroke was listed as the underlying cause of death on the death certificate (14%). Statistical Analysis Participants contributed follow-up time from the return of the 1986 questionnaire to the occurrence of a confirmed stroke or death or 31 January Men who had nonfatal myocardial infarction or coronary surgery during the follow-up period were retained in the analysis. Intake of vitamin E and other nutrients was energy-adjusted to 2000 kcal/d (30). Relative risks were calculated by dividing the incidence of stroke among men in each quintile of energy-adjusted vitamin E intake at baseline by the incidence of stroke among men in the lowest quintile of intake. Similar calculations were done for the other nutrients. Quintiles were used to avoid assumptions about the shape of the dose response relation and to provide sufficient power to compare men in the extreme categories of intake. For vita June 1999 Annals of Internal Medicine Volume 130 Number 12

3 Table 1. Age-Standardized Baseline Characteristics According to Quintile of Total Intake of Vitamin E, Vitamin C, and -Carotene* Variable Vitamin E Intake Vitamin C Intake -Carotene Intake Quintile 1 Quintile 5 Quintile 1 Quintile 5 Quintile 1 Quintile 5 Risk factors Current smoker, % Body mass index, kg/m Hypertension, % High cholesterol level, % Aspirin use, % Exercise, MET/wk Mean daily intake Alcohol, g Calories from fat, % Dietary fiber, g Cholesterol, mg Folate, g Vitamin E, IU Vitamin C, mg Carotene, IU Carotene, g Lutein, g Lycopene, g Use of supplements, % Multivitamin Vitamin E Vitamin C * MET metabolic equivalent. Quintile 1 is the quintile with lowest intake; quintile 5 is the quintile with highest intake. min E, vitamin C, and -carotene, additional analyses separately considered supplemental and dietary intake. We adjusted relative risks for age (in 5-year categories) (31) and used the Mantel extension test (32) to test for linear trends. To adjust for other risk factors, we used pooled logistic regression with 2-year intervals. Where the probability of an event within an interval is small, this method is equivalent to a Cox proportional hazards analysis (33). The multivariate models included the following as covariates: calendar time (in 2-year intervals), total energy intake (continuous variable), smoking (current; past; or 1 to 14, 15 to 24, and 25 cigarettes/ d), alcohol consumption ( 5,5to9,10to14,15to 29, or 30 g/d), history of hypertension, history of hypercholesterolemia, parental history of myocardial infarction before 65 years of age, profession, and quintiles of body mass index and physical activity. In these models, we evaluated monotonic trends by using the median value of each category and modeling it as a continuous variable. Analyses were also conducted to update dietary intake and other covariates during follow-up. In these analyses, the incidence of stroke in was related to vitamin and carotenoid intake reported in 1986, whereas the incidence of stroke in was related to the average intake reported in 1986 and Because dietary changes made after the development of intermediate end points, such as coronary heart disease, peripheral arterial disease, transient ischemic attacks, diabetes, and hypercholesterolemia, may confound the associations between diet and disease (34), we repeated the analyses without updating information on diet for men who reached an intermediate end point before completing the 1990 dietary questionnaire. All P values are two-sided. Role of the Funding Source The National Institutes of Health had no role in the collection, analysis, or interpretation of the data or in the decision to submit the paper for publication. Results During person-years of follow-up, we documented 328 strokes: 210 ischemic, 70 hemorrhagic, and 48 unclassified. Fifty of the 328 strokes were fatal. The distribution of risk factors for stroke differed somewhat by vitamin E, vitamin C, and -carotene intake (Table 1). Men in the top quintiles of vitamin E or vitamin C intake most of whom used multivitamin or specific vitamin supplements were less likely to smoke, were more likely to use aspirin, and were more physically active than men in the bottom quintiles. High intake of -carotene (Table 1) and other carotenoids (data not shown) was associated with a lower prevalence of smoking and a higher level of physical activity but not with use of aspirin. To account for these differences, we controlled for these variables in all multivariate analyses. In addi- 15 June 1999 Annals of Internal Medicine Volume 130 Number

4 Table 2. Relative Risk for Stroke According to Quintile of Intake of Vitamin E, Vitamin C, and Carotenoids Nutrient Median Intake Relative Risk (95% CI) Total Stroke (n 328) Ischemic Stroke (n 210) Hemorrhagic Stroke (n 70) Age-Adjusted* Multivariate Age-Adjusted* Multivariate Age-Adjusted* Multivariate Vitamin E Quintile IU/d Reference Reference Reference Reference Reference Reference Quintile IU/d ( ) ( ) ( ) Quintile IU/d ( ) ( ) ( ) Quintile IU/d ( ) ( ) ( ) Quintile IU/d ( ) ( ) ( ) P value for trend Vitamin C Quintile 1 95 mg/d Reference Reference Reference Reference Reference Reference Quintile mg/d ( ) ( ) ( ) Quintile mg/d ( ) ( ) ( ) Quintile mg/d ( ) ( ) ( ) Quintile mg/d ( ) ( ) ( ) P value for trend Carotene Quintile IU/d Reference Reference Reference Reference Reference Reference Quintile IU/d ( ) ( ) ( ) Quintile IU/d ( ) ( ) ( ) Quintile IU/d ( ) ( ) ( ) Quintile IU/d ( ) ( ) ( ) P value for trend Carotene Quintile g/d Reference Reference Reference Reference Reference Reference Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) P value for trend Lutein Quintile g/d Reference Reference Reference Reference Reference Reference Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) P value for trend Lycopene Quintile g/d Reference Reference Reference Reference Reference Reference Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) Quintile g/d ( ) ( ) ( ) P value for trend * Five-year categories. Covariates were calendar time (2-year intervals), total energy intake (continuous variable), smoking (current; past; and 1 to 14, 15 to 24, or 25 cigarettes/d), alcohol consumption ( 5, 5 to 9, 10 to 14, 15 to 29, or 30 g/d), history of hypertension, parental history of myocardial infarction before 65 years of age, profession, and quintiles of body mass index and physical activity, in addition to age. tion, intake of -carotene and intake of other carotenoids were directly associated with consumption of dietary fiber and other nutrients that are found in high concentrations in fruit and vegetables. Average daily servings of fruit and vegetables were 1.6 and 3.3, respectively, in the bottom quintile of -carotene intake and 1.9 and 5.8, respectively, in the top quintile. The age-adjusted and multivariate relative risks for stroke according to quintiles of intake of vitamin E, vitamin C, and carotenoids are shown in Table 2. Despite the large contrasts in intake (from 5.4 IU/d in the bottom quintile for vitamin E intake to 411 IU/d in the top quintile and from 95 mg/d in the bottom quintile for vitamin C intake to more than 1000 mg/d in the top quintile), neither vitamin was significantly associated with risk for total stroke, ischemic stroke, or hemorrhagic stroke. Men in the top quintile of vitamin E intake had an age-adjusted relative risk for ischemic stroke of 1.03 (after multivariate adjustment, it was 1.18 [95% CI, 0.77 to 1.82]; P 0.2 for trend). The corresponding ageadjusted and multivariate relative risks for vitamin C intake were 0.93 and 1.03 (CI, 0.66 to 1.59; P 0.2 for trend). With further adjustment for fiber intake, these relative risks increased slightly but remained nonsignificant. Risk for hemorrhagic stroke was slightly but not significantly higher in the top quintile of vitamin E intake than in the bottom quintile (multivariate relative risk, 1.31 [CI, 0.62 to 2.76]) and changed little after adjustment for fiber intake. In age-adjusted analyses, -carotene intake and lutein intake were inversely related to risk for total and ischemic stroke (Table 2), but these associations were attenuated in multivariate analyses. Only the association between lutein intake and risk for June 1999 Annals of Internal Medicine Volume 130 Number 12

5 ischemic stroke remained marginally significant; the relative risk for the top quintile of lutein intake compared with the bottom quintile was 0.63 (CI, 0.40 to 0.99; P 0.10 for trend). No significant associations were seen for the other carotenoids. Because of the inverse associations between intake of potassium, magnesium, and cereal fiber and risk for stroke previously reported in this cohort (35), we further adjusted the carotenoid analyses for these nutrients. As expected, intake of carotenoids was positively correlated with intake of potassium, and adjustment for potassium attenuated the carotenoid stroke associations. The strongest inverse association in these analyses was between lutein intake and risk for ischemic stroke (relative risk for the top quintile of lutein intake compared with the bottom quintile, 0.72 [CI, 0.45 to 1.16]). Because changes in diet over the follow-up period may have diluted the possible associations between nutrient intake and risk for stroke, we repeated the analyses described above after updating intake of vitamin E, vitamin C, carotenoids, and other covariates after the first 4 years of follow-up. The results were similar to those shown in Table 2 (data not shown). Analyses that did not update the diets of men who developed intermediate outcomes also yielded almost the same results. In addition, to explore the possibility that subgroups of the study sample might be differentially affected by higher intake of antioxidant vitamins or carotenoids, we conducted analyses stratified by the major risk factors for stroke: age, smoking, alcohol consumption, history of hypertension, body mass index, and use of aspirin. All differences in relative risks estimated in these analyses were compatible with chance variation but because the number of events in some strata was small, we cannot exclude the possibility of interactions. For example, the relative risk for the top quintile of -carotene intake compared with the bottom quintile was 0.29 (CI, 0.06 to 1.33) among current smokers and 1.20 (CI, 0.63 to 2.67) among never-smokers, but neither group showed a significant trend. We further examined the associations of supplemental vitamin E, vitamin C, and -carotene intake with risk for stroke. We considered the amounts of vitamins ingested in multivitamin or specific vitamin supplements, but not those in fortified cereals, to be supplemental. These analyses take advantage of the high proportion of men in our cohort who used supplements (many had done so for years before the inception of the study). No significant associations were found between dose of supplements (updated after the first 4 years of follow-up) (Table 3) or duration of use of supplements (Table 4) and stroke of any type. The multivariate relative risk for total stroke in men who reported, at baseline, 10 years or more of vitamin E supplementation compared with men who had never used vitamin E supplements was 0.96 (CI, 0.63 to 1.46). The multivariate relative risk for total stroke in men who reported, at baseline, 10 years or more of vitamin C supplementation compared with men who had never used vitamin C supplements was 0.97 (CI, 0.67 to 1.39). The combined use of vitamin E ( 100 IU/d) and vitamin C ( 400 mg/d) was not associated with risk for total stroke; the relative risk for men using both types of supplement compared with men who did not use either supplement was 1.01 (CI, 0.73 to 1.41). Table 3. Multivariate Relative Risks for Stroke According to Use of Vitamin Supplements Supplement Use Person-Years Multivariate Relative Risk (95% CI)* Total Stroke (n 328) Ischemic Stroke (n 210) Hemorrhagic Stroke (n 70) Vitamin E 0 IU/d Reference Reference Reference 25 IU/d ( ) 0.88 ( ) 0.74 ( ) IU/d ( ) 0.89 ( ) 0.93 ( ) IU/d ( ) 0.63 ( ) 2.00 ( ) 250 IU/d ( ) 1.16 ( ) 1.03 ( ) P value for trend Vitamin C 0 mg/d Reference Reference Reference 100 mg/d ( ) 1.11 ( ) 0.65 ( ) mg/d ( ) 0.84 ( ) 1.67 ( ) mg/d ( ) 1.11 ( ) 0.64 ( ) 700 mg/d ( ) 0.93 ( ) 0.67 ( ) P value for trend Carotene Nonusers Reference Reference Reference Users ( ) 0.97 ( ) 1.30 ( ) * Covariates were calendar time (2-year intervals), total energy intake (continuous variable), smoking (current; past; and 1 to 14, 15 to 24, or 25 cigarettes/d), alcohol consumption ( 5, 5 to 9, 10 to 14, 15 to 29, or 30 g/d), history of hypertension, parental history of myocardial infarction before 65 years of age, profession, and quintiles of body mass index and physical activity, in addition to age. Updated after 4 years of follow-up. 15 June 1999 Annals of Internal Medicine Volume 130 Number

6 Table 4. Multivariate Relative Risk for Stroke According to Duration of Supplement Use Supplement Use Person-Years Multivariate Relative Risk (95% CI)* Total Stroke (n 303) Ischemic Stroke (n 196) Hemorrhagic Stroke (n 66) Vitamin E None Reference Reference Reference 0 1 year ( ) 1.88 ( ) 1.75 ( ) 2 4 years ( ) 0.84 ( ) 1.52 ( ) 5 9 years ( ) 1.09 ( ) 1.51 ( ) 10 years ( ) 1.06 ( ) 0.61 ( ) P value for trend Vitamin C None Reference Reference Reference 0 1 year ( ) 0.85 ( ) 1.21 ( ) 2 4 years ( ) 1.17 ( ) 1.00 ( ) 5 9 years ( ) 1.05 ( ) 1.70 ( ) 10 years ( ) 1.07 ( ) 0.55 ( ) P value for trend Multivitamins None Reference Reference Reference 0 1 year ( ) 1.91 ( ) 0.00 ( ) 2 4 years ( ) 1.89 ( ) 0.22 ( ) 5 9 years ( ) 0.75 ( ) 0.79 ( ) 10 years ( ) 1.28 ( ) 1.00 ( ) P value for trend * Covariates were calendar time (2-year intervals), total energy intake (continuous variable), smoking (current; past; and 1 to 14, 15 to 24, or 25 cigarettes/d), alcohol consumption ( 5, 5 to 9, 10 to 14, 15 to 29, or 30 g/d), history of hypertension, parental history of myocardial infarction before 65 years of age, profession, and quintiles of body mass index and physical activity, in addition to age. To address the hypothesis that use of vitamin supplements may benefit men with poor dietary intake, we restricted analyses to men who reported daily intake from food of less than 9.2 IU of vitamin E or less than 126 mg of vitamin C (that is, men in the lowest tertiles for vitamin E and vitamin C intake). We still found no associations between use of vitamin E or vitamin C supplements and risk for ischemic stroke. Discussion In this large cohort of U.S. men, we found no associations between vitamin E and vitamin C consumption and risk for total or ischemic stroke. Of the carotenoids, only lutein was significantly associated with risk for stroke, but this inverse association might be due to other dietary factors. Previous evidence about the associations between vitamin E or vitamin C intake and risk for stroke is inconsistent. In the Alpha-Tocopherol-Beta-Carotene (ATBC) trial in Finland (16), men taking tocopherol, 50 IU/d, had an increased risk for death from hemorrhagic stroke (relative risk, 1.5) and a nonsignificantly decreased risk for death from ischemic stroke (relative risk, 0.84) compared with men taking placebo; the combined relative risk for fatal stroke was In the Nurses Health Study (6), a moderate but nonsignificant inverse association was seen between use of vitamin E supplements and risk for ischemic stroke (relative risk, 0.71 [CI, 0.39 to 1.31]). In a cohort in Norway, intake of vitamin C was inversely related to risk for hemorrhagic but not ischemic stroke (36). Other studies did not distinguish between ischemic and hemorrhagic events but are nevertheless important because they address the public health question of whether increased intake of antioxidants may reduce overall risk for stroke or death from stroke. In addition, most of these studies were conducted in populations characterized by little use of vitamin supplements and higher prevalences of vitamin deficiencies; thus, they address the association between antioxidants and stroke at very low levels of antioxidant intake. In the Linxian Trial (37), -tocopherol, 60 IU/d, administered with selenium (50 mg) and -carotene (15 mg) did not reduce risk for death from stroke, and neither did the combination of vitamin C, 180 mg/d, and molybdenum, 30 g. No significant associations were found between vitamin E or vitamin C intake and death from stroke in prospective studies done in Shanghai (15) and the Netherlands (14) or between vitamin C intake and incidence of total stroke in the Western Electric Study (20). In a study among elderly people in the United Kingdom (22), intake of vitamin C was inversely related to death from stroke (relative for the top quintile of vitamin C intake compared with the bottom quintile, 0.5 [CI, 0.3 to 0.8]), but whether this association was due to vitamin C itself or to other components of fruit and vegetables remains uncertain. A similar limitation applies to the increased risk for death from stroke seen in men with low plasma levels of vitamin C and -carotene in the Basel Prospective Study (21) June 1999 Annals of Internal Medicine Volume 130 Number 12

7 The results of the ATBC trial (16) raised concerns that vitamin E may increase risk for hemorrhagic stroke. No increased risk for hemorrhagic stroke was found in our cohort among men in the highest categories of vitamin E intake or duration of vitamin E supplementation, but the CIs did not exclude a modest potential adverse effect. Further follow-up is needed. In the meantime, it should be noted that rates of hemorrhagic stroke in our cohort of mostly nonsmoking men were low the rate of fatal hemorrhagic stroke was 0.84 for person-years, approximately eight times lower than the rate seen in the ATBC trial (16). Thus, even a modestly increased relative risk for vitamin E use would mean a small absolute risk. Large supplementation trials have shown no benefit of -carotene with respect to risk for stroke. In the ATBC trial (16, 17), men receiving 20 mg of -carotene per day had a higher risk for stroke than men receiving placebo, whereas no effect was found with the use of 50 mg of -carotene on alternate days in the Physicians Health Study (17). Thus, the weak and nonsignificant inverse association between intake of -carotene and risk for stroke in our study is probably due to confounding by the beneficial effects of other nutrients whose intake is correlated with intake of -carotene. Lutein found mostly in dark-green leafy vegetables was the only carotenoid that had a significant inverse association with risk for stroke. However, intake of lutein is correlated with intake of potassium, folate, and several other nutrients that may have beneficial effects. Because of errors in the measurement of long-term diet, even in a large study like this, it is difficult to assess the independent effects of these nutrients (38). This difficulty could be addressed, in part, by pooling the data of several prospective investigations to achieve more power and a wider range of intake. Although the lack of association between antioxidant intake and risk for stroke in our study could be due partly to errors in dietary measurement, the results of our validation study (26) and the strong inverse association that we found in our cohort between use of vitamin E supplements and risk for coronary heart disease (7) support the validity of our estimates of intakes. Because men with high intake of vitamins C and E have healthier risk profiles than men with low intakes of these vitamins (Table 1), it is unlikely that the lack of association is due to confounding by other aspects of lifestyle. A possible protective effect of antioxidant vitamins could also have been diluted if some men started taking vitamin supplements because of preclinical symptoms of ill health that may have been predictive of future stroke. However, the lack of benefit in men who took vitamin supplements for more than 5 years before the inception of the study and in men who reported using supplements both in 1986 and 1990 argues against this explanation. The participants in our study are a self-selected group of health professionals, and their lifestyles and diets are healthier than those of average men. Thus, we cannot exclude the possibility that the use of vitamin E or C supplements reduces risk for stroke in populations with diets deficient in these vitamins. In addition, men with a history of cardiovascular disease were excluded from our study, and we therefore did not address the possible benefits of antioxidants in secondary prevention (39). Although a benefit of vitamin E or vitamin C with respect to risk for ischemic stroke is plausible because of the importance of low-density lipoprotein oxidation in atherosclerosis, a substantial proportion of ischemic strokes are not directly related to atherosclerosis and may be unaffected by antioxidant intake. Other risk factors for atherosclerosis, such as high blood cholesterol levels, are more predictive of coronary heart disease than of ischemic stroke. Thus, the different pathologies of coronary heart disease and ischemic stroke may partly explain the discordant results seen in our cohort. Given the uncertainty about the beneficial effects of antioxidants with respect to stroke, it is more appropriate from a public health point of view to emphasize the preventive benefits of diets rich in fruit and vegetables and to avoid claims for specific nutrients. Higher intakes of fruit and vegetables have been found to be inversely associated with risk for stroke in an ecological analysis (40), in the Framingham study (41), in a prospective study in vegetarians in the United Kingdom (42), and in our cohort (35). In conclusion, our results suggest that use of vitamin E or vitamin C supplements is unlikely to reduce risk for stroke in adult men with no history of cardiovascular disease or diabetes. Although some carotenoids may reduce risk for stroke, our data do not provide adequate support for recommending an increased intake of specific compounds. Rather, dietary prevention of stroke should be achieved by increasing intake of fruit and vegetables, as previous investigations in our cohort and other cohorts have suggested. Acknowledgments: The authors thank the participants of the Health Professionals Follow-up Study. They also thank Al Wing, Mira Kaufman, Karen Corsano, and Steve Stuart for computer assistance; Jill Arnold, Betsy Frost-Hawes, Kerry Demers, and Mitzi Wolff for assistance with compilation of data and manuscript preparation; and Laura Sampson and Helaine Rockett for maintaining food composition tables. Grant Support: In part by research grants HL and CA from the National Institutes of Health. Dr. Kawachi is supported by a Career Development Award from the National Heart, Lung, and Blood Institute. 15 June 1999 Annals of Internal Medicine Volume 130 Number

8 Requests for Reprints: Alberto Ascherio, MD, Department of Nutrition, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115; , Current Author Addresses: Drs. Ascherio, Hernán, Rimm, and Willett: Department of Nutrition, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA Drs. Giovannucci and Stampfer: Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women s Hospital, 181 Longwood Avenue, Boston, MA Dr. Kawachi: Department of Health and Social Behavior, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA References 1. Stampfer MJ, Rimm EB. Epidemiologic evidence for vitamin E in prevention of cardiovascular disease. Am J Clin Nutr. 1995;62(Suppl 6):1365S-9S. 2. Boscoboinik D, Szewczyk A, Hensey C, Azzi A. Inhibition of cell proliferation by -tocopherol. Role of protein kinase C. J Biol Chem. 1991;266: Princen HM, van Poppel G, Vogelezang C, Buytenhek R, Kok FJ. Supplementation with vitamin E but not beta-carotene in vivo protects low density lipoprotein from lipid peroxidation in vitro. Effect of cigarette smoking. Arterioscler Thromb. 1992;12: Esterbauer H, Dieber-Rotheneder M, Striegl G, Waeg G. Role of vitamin E in preventing the oxidation of low-density lipoprotein. Am J Clin Nutr. 1991;53(Suppl 1):314S-21S. 5. Steiner M. Influence of vitamin E on platelet function in humans. J Am Coll Nutr. 1991;10: Stampfer MJ, Hennekens CH, Manson JE, Colditz GA, Rosner B, Willett WC. Vitamin E consumption and the risk of coronary disease in women. N Engl J Med. 1993;328: Rimm EB, Stampfer MJ, Ascherio A, Giovannucci E, Colditz GA, Willett WC. Vitamin E consumption and the risk of coronary heart disease in men. N Engl J Med. 1993;328: Kushi LH, Fee RM, Sellers TA, Zheng W, Folsom AR. Intake of vitamin A, C, and E and postmenopausal breast cancer. Am J Epidemiol. 1996;144: Donnan PT, Thomson M, Fowkes FG, Prescott RJ, Housely E. Diet as a risk factor for peripheral arterial disease in the general population: the Edinburgh Artery Study. Am J Clin Nutr. 1993;57: Kritchevsky SB, Shimakawa T, Tell GS, Dennis B, Carpenter M, Eckfeldt JH, et al. Dietary antioxidants and carotid artery wall thickness. The ARIC Study. Atherosclerosis Risk in Communities Study. Circulation. 1995;92: Azen SP, Qian D, Mack WJ, Sevanian A, Selzer RH, Liu CR, et al. Effect of supplementary antioxidant vitamin intake on carotid arterial wall intimamedia thickness in a controlled clinical trial of cholesterol lowering. Circulation. 1996;94: Kok FJ, de Bruijn AM, Vermeeren R, Hofman A, van Laar A, de Bruin M, et al. Serum selenium, vitamin antioxidants, and cardiovascular mortality: a 9-year follow-up study in the Netherlands. Am J Clin Nutr. 1987;45: Gey KF, Moser UK, Jordan P, Stahelin HB, Eichholzer M, Ludin E. Increased risk of cardiovascular disease at suboptimal plasma concentrations of essential antioxidants: an epidemiological update with special attention to carotene and vitamin C. Am J Clin Nutr. 1993;57(Suppl 5):787S-97S. 14. Keli SO, Hertog MG, Feskens EJ, Kromhout D. Dietary flavonoids, antioxidant vitamins, and incidence of stroke: the Zutphen study. Arch Intern Med. 1996;156: Ross RK, Yuan JM, Henderson BE, Park J, Gao YT, Yu MC. Prospective evaluation of dietary and other predictors of fatal stroke in Shanghai, China. Circulation. 1997;96: The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. N Engl J Med. 1994;330: Hennekens CH, Buring JE, Manson JE, Stampfer M, Rosner B, Cook NR, et al. Lack of effect of long-term supplementation with beta carotene on the incidence of malignant neoplasms and cardiovascular disease. N Engl J Med. 1996;334: Di Mascio P, Kaiser S, Sies H. Lycopene as the most efficient biological carotenoid singlet oxygen quencher. Arch Biochem Biophys. 1989;274: Frei B, England L, Ames BN. Ascorbate is an outstanding antioxidant in human blood plasma. Proc Natl Acad Sci U S A. 1989;86: Daviglus ML, Orencia AJ, Dyer AR, Liu K, Morris DK, Persky V, et al. Dietary vitamin C, beta-carotene and 30-year risk of stroke: results from the Western Electric Study. Neuroepidemiology. 1997;16: Gey KF, Stähelin HB, Eichholzer M. Poor plasma status of carotene and vitamin C is associated with higher mortality from ischemic heart disease and stroke: Basel Prospective Study. Clin Invest. 1993;71: Gale CR, Martyn CN, Winter PD, Cooper C. Vitamin C and risk of death from stroke and coronary heart disease in cohort of elderly people. BMJ. 1995;310: Bendich A. Symposium conclusions: biological actions of carotenoids. J Nutr. 1989;119: Rimm EB, Giovannucci EL, Willett WC, Colditz GA, Ascherio A, Rosner B, et al. A prospective study of alcohol consumption and risk of coronary disease in men. Lancet. 1991;338: Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett WC. Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol. 1992;135: Ascherio A, Stampfer MJ, Colditz GA, Rimm EB, Litin L, Willett WC. Correlations of vitamin A and E intakes with the plasma concentrations of carotenoids and tocopherols among American men and women. J Nutr. 1992;122: Michaud DS, Giovannucci EL, Ascherio A, Rimm EB, Forman MR, Sampson L, et al. Associations of plasma carotenoid concentrations and dietary intake of specific carotenoids in samples of two prospective cohort studies using a new carotenoid database. Cancer Epidemiol Biomarkers Prev. 1998;7: Ascherio A, Rimm EB, Giovannucci EL, Colditz GA, Rosner B, Willett WC, et al. A prospective study of nutritional factors and hypertension among US men. Circulation. 1992;86: Walker AE, Robins M, Weinfeld FD. The National Survey of Stroke. Clinical findings. Stroke. 1981;12(Suppl 1):I Willett WC, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol. 1986;124: Rothman KJ, Greenland S. Modern Epidemiology. 2d ed. Philadelphia: Lippincott Raven; Mantel N. Chi-square tests with one degree of freedom: extensions of the Mantel-Haenszel procedure. Journal of the American Statistical Association. 1963;58: D Agostino RB, Lee ML, Belanger AJ, Cupples LA, Anderson K, Kannel WB. Relation of pooled logistic regression to time dependent Cox regression analysis: the Framingham Heart Study. Stat Med. 1990;9: Shekelle RB, Stamler J, Paul O, Shryock AM, Liu S, Lepper M. Dietary lipids and serum cholesterol level: change in diet confounds the cross-sectional association. Am J Epidemiol. 1982;115: Ascherio A, Rimm EB, Hernán MA, Giovannucci EL, Kawachi I, Stampfer MJ, et al. Intake of potassium, magnesium, calcium, and fiber and risk of stroke among US men. Circulation. 1998;98: Vollset SE, Bjelke E. Does consumption of fruit and vegetables protect against stroke? [Letter] Lancet. 1983;2: Mark SD, Wang W, Fraumeni JF Jr, Li JY, Taylor PR, Wang GQ, et al. Do nutritional supplements lower the risk of stroke or hypertension? Epidemiology. 1998;9: Rosner B, Spiegelman D, Willett WC. Correction of logistic regression relative risk estimates and confidence intervals for measurement error: the case of multiple covariates measured with error. Am J Epidemiol. 1990;132: Steiner M, Glantz M, Lekos A. Vitamin E plus aspirin compared with aspirin alone in patients with transient ischemic attacks. Am J Clin Nutr. 1995; 62(Suppl 6):1381S-4S. 40. Acheson RM, Williams DR. Does consumption of fruit and vegetables protect against stroke? Lancet. 1983;1: Gillman MW, Cupples LA, Gagnon D, Posner BM, Ellison RC, Castelli WP, et al. Protective effect of fruits and vegetables on development of stroke in men. JAMA. 1995;273: Key TJ, Thorogood M, Appleby PN, Burr ML. Dietary habits and mortality in 11,000 vegetarians and health conscious people: results of a 17 year follow up. BMJ. 1996;313: June 1999 Annals of Internal Medicine Volume 130 Number 12