Cynthia D. Morris, Ph.D., M.P.H.a, Susan Carson, M.P.H.a
The authors of this article are responsible for its contents, including any clinical or treatment recommendations. No statement in this article should be construed as an official position from the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services.
Address correspondence to: Cynthia D. Morris, Ph.D., M.P.H., Oregon Health & Science University Evidence-based Practice Center, Department of Medical Informatics and Clinical Epidemiology; E-mail: morrisc@ohsu.edu.
This chapter first appeared as an article in Ann Intern Med 2003;139(1):56-70.
Epidemiology
Methods
Role of the Funding Source
Results
Discussion
Acknowledgments
References
Notes
Approximately 60 million persons in the United States have some form of cardiovascular disease (CVD), and in 1996 CVD accounted for 41.4% of all deaths in the United States.1 While hypertension, diabetes, dyslipidemia, and smoking are leading risk factors, nutritional status plays a substantial role in the development of atherosclerotic CVD. Antioxidant nutrients, including vitamin C, vitamin E, and beta-carotene, are thought to play a role in atherosclerosis.2-5 Some experts believe that mild to moderate deficiencies of these vitamins, although not severe enough to cause classic deficiency diseases, may be involved in the development of CVD.3,4,6 Therefore, it is thought that antioxidant supplementation may help reduce the incidence or progression of atherosclerotic CVD. The biological basis of antioxidant use to prevent atherosclerotic heart disease is based largely on the oxidative modification hypothesis of atherosclerosis.6 According to this hypothesis, lipid peroxidation or oxidative modification of low-density lipoprotein (LDL) is the initiator of atherosclerosis. Antioxidants capable of inhibiting lipid peroxidation should support primary and secondary prevention and consequently reduce cardiovascular events, including myocardial infarction.
This evidence review was conducted for the U.S. Preventive Services Task Force (USPSTF) to serve as the foundation for its recommendations on vitamin supplementation for disease prevention. This article addresses a key question posed by the Task Force:
We searched the Cochrane Controlled Trials Registry and MEDLINE® for relevant papers published in English from 1966 to September 2001, using Medical Subject Headings and keywords for the individual nutrients (vitamin A, vitamin C, vitamin E, beta-carotene, folic acid), and for multivitamin and antioxidant supplements, combined with terms for CVD, coronary artery disease, myocardial infarction, and related risk factors (blood pressure, hypertension, hyperlipidemia, homocysteine). We examined reference lists of review articles6-15 and asked experts for additional references. Finally, we searched MEDLINE® using the acronyms or full titles of the major trials and cohort studies to identify additional publications.
The scope of this review was developed with input from the USPSTF. We included reports of randomized trials and prospective cohort studies from U.S. and European populations that assessed use of vitamin supplements and reported the incidence of or death from cardiovascular events. We included only studies that measured intake of vitamins from supplements, not from foods; most supplements provide single or limited nutrient combinations whereas dietary sources are nutritionally complex in nature and complicate data interpretation. Only cohorts that reported specifically on vitamin supplement use with risk ratios independent of dietary intake were included. Both primary and secondary prevention trials were considered, but were analyzed separately. Studies conducted in specific populations that were not widely generalizable were excluded, such as a cohort with end-stage renal disease. Only cohort studies rated as being of good to fair quality by predetermined criteria from a system developed by the current USPSTF were included.16 Studies were excluded if they contained no original data, were not relevant (eg, addressed vitamin deficiency disease), did not report data on the specified outcomes, or took place in an acute care setting. Case-control studies were excluded because of retrospective data collection.
Two reviewers read titles and abstracts of 2,758 identified articles and selected 306 as possibly relevant. Full-text articles of these citations were retrieved for further review. Of these, 38 articles, representing 10 cohort studies and 20 randomized, controlled trials (RTCs), were selected for inclusion in evidence tables. An additional 25 articles were included for background and context.
We abstracted the following descriptive information: population, setting, sample size, supplement (dose, formulation, and frequency), control group intervention, length of follow-up, follow-up rate, confounding factors, factors controlled for in analyses, method of ascertaining compliance, compliance rate, and adverse effects. We recorded data on the following outcomes: cardiovascular events, myocardial infarction, restenosis, change in angina, cardiovascular mortality, and all-cause mortality. Study quality was assessed using the standards of the current USPSTF.16 For randomized controlled trials (RCTs), we summarized study quality using the Jadad score, which rates trials on a scale of 1 to 5 on the basis of adequacy of randomization method, blinding, and other criteria.17 Data abstraction and quality assessment were conducted independently by at least 2 reviewers. Disagreements were resolved by consensus or by a third reviewer. Finally, we summarized the strength, level, and quality of the overall evidence tables for the effectiveness of each of the vitamin supplements to prevent CVD.
We included 11 reports from 10 prospective cohort studies,18-28 12 reports from 10 randomized trials of primary prevention of CVD,29-40 and 15 reports from 12 randomized trials of secondary prevention of CVD.41-55 The principal epidemiologic cohort studies (Table 1) included the Nurses' Health Study (87,245 female nurses followed for more than 10 years),18,19 the Health Professionals' Follow-up Study (39,910 male health professionals followed for 4 years),20 the Iowa Women's Health Study (34,486 Iowa women followed for 7 years),21 and a cohort of 83,639 male physicians invited to participate in the Physicians' Health Study.27 Other studies include the Established Populations for Epidemiologic Studies of the Elderly,23 (11,178 men and women older than 65 years), the first National Health and Nutrition Examination Survey (NHANES-I),22 which followed a similar number of participants, and a study of more than 1 million people recruited for a mortality study by the American Cancer Society (ACS).25 These studies were all rated as "good" or "fair" using the USPSTF rating scale.16 Most had follow-up rates above 90% after 4 or more years, used well-defined outcomes, and adjusted for relevant confounders.
The RCTs of primary prevention (Table 2) principally comprised large factorial trials, including an antioxidant supplement with a cointervention (angiotensin-converting enzyme inhibitor, aspirin, or lipid lowering agent) examining CVD incidence or mortality. Of these 9 trials, only 4 were undertaken with the primary objective of preventing CVD: HOPE (Heart Outcomes Prevention Evaluation),35 the Women's Health Study,30 the Primary Prevention Project,36 and the Heart Protection Study.40 The others were designed to test whether antioxidants would prevent cancer or reduce progression of age-related eye disease; cardiovascular events were analyzed as a secondary endpoint. With one exception,36 all primary prevention trials were double-blind and placebo-controlled.
Among secondary prevention trials, 10 of 12 examined the effects of vitamin supplementation in patients enrolled on the basis of pre-existing CVD (Table 3). Two studies analyzed subgroups with prior coronary disease.41-44 In addition, 2 major trials analyzed primary and secondary prevention together.35,40 Jadad scores for these studies ranged from 3 to 5 on the 5-point scale, indicating fair to good quality.
This research was funded by the Agency for Healthcare Research and Quality (AHRQ) under a contract to support the work of the USPSTF. AHRQ staff and Task Force members participated in the initial design of the study and also reviewed interim analyses and the final manuscript.
One good quality cohort study evaluated the effect of vitamin A supplementation on incident coronary death.21 Subjects in the highest quartile of vitamin A supplement use did not have a lower risk for coronary death compared with those in the lowest quartile. There are no data from clinical trials on the effect of vitamin A supplements on atherosclerotic CVD.
In 3 of 4 cohort studies, vitamin C supplementation was not associated with coronary heart disease mortality21-23 or all-cause mortality.23 In 2 studies of older samples,21,23 vitamin C use did not protect against coronary disease or all-cause mortality after adjustment for relevant confounders.23 Similarly, vitamin C had no impact on cardiovascular or coronary heart disease mortality.27 However, in a good quality follow-up study of the NHANES-I,22 regular use of a vitamin C supplement reduced the standardized mortality ratio for cardiovascular mortality by 48% and for all-cause mortality by 26%. In a cohort analysis of a secondary prevention trial of cholesterol reduction or coronary stenosis, vitamin C use (>250 mg/day) had no appreciable effect on progression of stenosis.28
No randomized clinical trial of primary prevention has evaluated the effect of supplementation with vitamin C alone on CVD outcomes. One small, poorly controlled trial of secondary prevention with vitamin C suggested a reduced rate of coronary restenosis and reintervention.54
Three good-quality cohort studies reported statistically significant associations between the use of vitamin E supplements and lower rates of CVD mortality23 and nonfatal CVD events.18,19 In the all-female Nurses' Health Study, use of a vitamin E supplement was associated with an adjusted risk reduction of 37% for major coronary heart disease (nonfatal myocardial infarction and coronary disease mortality) after 8 years of follow-up.18 Less than 2 years of use had no significant impact on cardiovascular risk and a minimum dose of 100 IU per day was necessary to observe risk reduction. In the all-male Health Professionals' Follow-up Study, a similar adjusted risk reduction of 25% was observed for incident coronary disease (coronary disease mortality, nonfatal myocardial infarction, coronary artery bypass graft surgery, or angioplasty) when supplement users were compared with nonusers after 4 years.20 Men who took a supplement containing at least 100 IU per day for at least 2 years had greater adjusted risk reduction than nonusers (relative risk [RR], 0.63; 95% confidence interval [CI], 0.47 to 0.84). This significant reduction in death from coronary heart disease and all-cause mortality was also observed in an elderly U.S. population.23 In a cohort analysis of a secondary prevention trial of aggressive cholesterol reduction, participants who used at least 100 IU per day of vitamin E had significantly less progression of stenosis than those who did not.28 However, 2 good-quality cohorts demonstrated no effect of supplementation on CVD mortality.21,27 Supplemental vitamin E was not associated with coronary heart disease mortality in the Iowa Women's Health Study21 or with cardiovascular or coronary heart disease mortality in the Physicians' Health Study.27
Clinical trials have generally not demonstrated that vitamin E supplementation is beneficial to CVD outcomes. There are 3 major clinical trials of primary prevention, 2 of good quality (the Alpha-Tocopherol Beta-Carotene Cancer Prevention Study Group [ATBC]32 and the HOPE trial35) and 1 of fair quality (the Primary Prevention Project36). In the ATBC Study, which comprised male smokers, vitamin E had no significant effect on coronary heart disease outcomes.34 The incidence of myocardial infarction, cardiovascular events, and cardiovascular mortality did not differ in participants randomly assigned to vitamin E compared with those assigned to placebo. The incidence of new-onset angina pectoris in vitamin E users was lower than in all patients assigned to beta-carotene or placebo (RR, 0.91; 95% CI, 0.83 to 0.99), but not when compared only to those assigned to placebo (RR, 0.97; 95% CI, 0.85 to 1.10).33
The HOPE trial tested the effect of 400 IU of vitamin E and an angiotensin-converting enzyme inhibitor versus placebo in subjects at high risk for cardiovascular events.35 Individuals older than 55 with a history of coronary artery disease, peripheral artery disease or diabetes, plus a cardiovascular risk factor, were included. The trial therefore encompassed both primary and secondary prevention. After 4.5 years of vitamin E supplementation, no reduction was observed in myocardial infarction, cardiovascular events, cardiovascular mortality, or all-cause mortality overall or in any subgroup, including participants who had diabetes and those who smoked.
In the Primary Prevention Project, 4,495 men and women older than 50 attending general practice or hypertension clinics were randomly assigned to receive 300 IU of synthetic vitamin E or aspirin in an open-label, non-placebo-controlled, 2x2 factorial trial.36 After a median of 4 years, vitamin E supplementation had no significant impact on myocardial infarction, cardiovascular events, cardiovascular mortality, or all-cause mortality. Last, in a small study investigating progression of carotid artery intimal thickening, no difference in myocardial infarction, cardiovascular events and mortality, or all-cause mortality was observed with 300 IU of vitamin E given for 3 years.39
Of the 7 RCTs of vitamin E supplementation for secondary prevention of cardiac events,43-49 only 148 demonstrated a strongly beneficial effect. Three small trials of relatively short duration evaluated evidence of restenosis of coronary arteries or improvement in angina.45-47 In 2 of these studies, 1 of which was of good quality45 and one of which was of fair quality,46 angina did not improve with vitamin E supplementation. In another, there was a suggestion of a reduced rate of restenosis with vitamin E supplementation compared with placebo, but the difference did not reach statistical significance.47
The Cambridge Heart Antioxidant Study (CHAOS) was the only RCT to describe a significantly reduced risk for myocardial infarction and all cardiac events after 1.5 years of vitamin E supplementation.48 However, it was encumbered by design problems, including unbalanced randomization, incomplete follow-up, and a mid-study change in the vitamin E dose (800 IU to 400 IU per day).48,56
Other studies did not demonstrate reduced risk for CVD or cardiac events. In a subgroup of patients with mild angina at baseline in the ATBC Study, vitamin E supplementation had no effect on progression to severe angina, major coronary events, or fatal coronary heart disease.43 In a separate analysis of an ATBC subgroup of patients with previous myocardial infarction, vitamin E supplementation significantly reduced nonfatal myocardial infarction, but not fatal myocardial infarction.44 Considered together, all myocardial infarctions, cardiovascular events, and cardiovascular mortality were not affected by vitamin E supplementation. The GISSI-P (Gruppo Italiano per lo Studio della Sopavivienza nell'Infarto miocarciso - Prevenzione) study is the largest of the secondary prevention trials. It included with more than 11,324 men and women in Italy who were randomly assigned in an open-label, non-placebo-controlled 2x2 factorial study of vitamin E or n-3 polyunsaturated fatty acid.49 Combined primary and cardiovascular endpoints were not significantly reduced in the 2-way or 4-way analyses comparing vitamin E supplements with no treatment. In secondary analyses, vitamin E supplementation significantly reduced cardiovascular death, including cardiac, coronary, and sudden deaths in the 4-way but not the 2-way analysis. Although this study was not blinded, the large sample size and practice setting lend credibility to the results.
In 6 publications from 4 RCTs of primary prevention, beta-carotene supplementation did not reduce risk for cardiovascular disease events or death.29-34 The Women's Health Study's primary purpose was preventing incident cancer and CVD.30 The beta-carotene arm of this trial was discontinued after a median of 2.1 years and patient follow-up continued for 2 additional years. Beta-carotene supplementation (50 mg on alternate days) had no significant effect on incident myocardial infarction, cardiovascular events, or all-cause and cardiovascular mortality. A subsequent nested case-control analysis of 130 women from the Women's Health Study showed no effect of treatment on outcome according to plasma beta-carotene level at baseline.57 In the all-male Physician's Health Study, there was no evidence of a direct effect of beta-carotene (50 mg on alternate days) on incident myocardial infarction, cardiovascular events, or all-cause or cardiovascular mortality.29 Three separate analyses of the ATBC study32-34 have indicated that 5 to 8 years of beta-carotene supplementation (20 mg/day) had no effect on incident non-fatal myocardial infarction, incident angina, all major coronary events, or cardiovascular mortality. However, all-cause mortality was significantly increased by 8% with beta-carotene, principally because of increased rates of ischemic heart disease and lung cancer. Last, in the Skin Cancer Prevention Study,31 beta-carotene (50 mg/day) had no significant impact on all-cause or cardiovascular mortality after a median of 8.2 years.
In analyses of secondary prevention in the ATBC study, beta-carotene supplementation had no significant effect on the development of severe angina, major coronary events, or fatal coronary heart disease in a subgroup with angina at baseline.43 In a subgroup of patients with prior myocardial infarction, the incidence of fatal coronary heart disease was significantly increased with beta-carotene.44 Although the overall risk for myocardial infarction was not affected, incidence of fatal myocardial infarction also increased significantly with beta-carotene. Finally, an analysis of the Physician's Health Study indicated a significant reduction in major coronary events in a sample of patients with angina pectoris at baseline41; with longer follow-up, beta-carotene had no demonstrated effect on cardiovascular events or mortality.42 No observational study has analyzed the use of beta-carotene supplements for the prevention of cardiovascular events.
Three good-quality23,24,26 observational studies and 1 observational study of fair quality25 have evaluated the effect of an antioxidant combination on cardiovascular events without attempting to separate the individual components. In a study in the Netherlands of persons older than 55 years of age, use of an antioxidant supplement was associated with a significantly reduced 4-year risk for myocardial infarction compared with nonuse.24 Similarly, in a U.S. cohort study in elderly persons, use of vitamins C and E reduced coronary death and all-cause mortality.23 A good-quality cohort study of Finnish residents showed no significant effect of an antioxidant supplement on coronary mortality;26 however, only 3% of the study sample used an antioxidant supplement. One fair-quality study of more than 1 million men and women in the United States demonstrated modest reductions in CVD mortality among women using antioxidant supplementation who had no history of CVD. Similar reductions were not demonstrated among men.25 All-cause mortality was significantly reduced among women but was not reduced among men.
Three large, good quality primary prevention trials of antioxidant combination therapy have reported no effect on major cardiovascular end points. In the Carotene and Retinol Efficacy Trial, the rate of cardiovascular death after 5.5 years of follow-up was not significantly higher with the supplement combination of beta-carotene and retinol than with placebo.37 All-cause mortality was significantly increased, however, in people taking the supplement combination. This trial was terminated before completion because of an increase in lung cancer incidence and death in the group receiving the supplement.
The Age Related Eye Disease Study Group reported results of a trial of vitamin C, vitamin E, and beta-carotene to prevent progression of age-related cataracts and macular degeneration.38 Morbidity and mortality outcomes were also collected. Vitamin C, vitamin E, and beta-carotene had no effect on all-cause mortality; however, participants reported chest pain much less frequently with antioxidant supplementation than without it.
The Heart Protection Study40 included more than 20,000 subjects in a double-blind, placebo-controlled, 2x2 factorial trial of a combination supplement of vitamin C, vitamin E, and beta-carotene. Data on cardiovascular events, vascular mortality, and all-cause mortality were collected. The cointervention was simvastatin. The antioxidant supplement group and the placebo group did not differ in all-cause mortality, coronary mortality, and all-vascular mortality, or in coronary events, coronary revascularization, or all major vascular events. There was no difference in the actuarial rate of events in early versus late follow-up between groups. When the trial was separated into primary and secondary prevention cohorts, the lack of any significant effect on major cardiovascular events remained.
Three good quality studies of antioxidant vitamin supplementation for secondary prevention of CVD have been reported.50-52 In the Multivitamins and Probucol (MVP) Study,50,51 investigators tested an antioxidant medication (probucol), an antioxidant vitamin supplement (vitamin C, vitamin E, and beta-carotene), and placebo. This study was terminated at the interim analysis because those in the probucol group had achieved the critical benefit threshold for the primary endpoint, restenosis rate. The rate of restenosis in the antioxidant-only group did not differ from that in the placebo group. In a second study,52 patients with coronary heart disease, low levels of high-density lipoprotein cholesterol, and normal levels of low-density lipoprotein cholesterol were assigned to a 2x2 factorial trial of simvastatin, antioxidants (vitamin C, vitamin E, natural beta-carotene, and selenium), or placebo. Antioxidant therapy did not significantly affect the rate of restenosis or cardiovascular events 3 years later.
The Women's Angiographic Vitamin and Estrogen (WAVE) trial studied the progression of minimal luminal diameter in coronary arteries with 15% to 75% stenosis at baseline.55 A supplement of vitamins C and E was compared with placebo in a 2x2 factorial trial with a cointervention of estrogen. Progression of stenosis did not differ between patients randomly assigned to the vitamin supplement and those assigned to no supplement, nor did myocardial infarction, cardiovascular events, or cardiovascular mortality. However, all-cause mortality was significantly increased with the antioxidant supplement. In summary, cohort studies show an association between antioxidant supplementation and reduced coronary heart disease morbidity and mortality, but no reduced risk is demonstrated in clinical trials of primary and secondary prevention.
Four cohort studies analyzed the relationship between the use of multivitamins and CVD. One good quality study reported a significant reduction in coronary events with multivitamin use,19 2 good quality studies reported no significant effect on mortality,23,27 and a fair quality trial reported an increase in all-cause mortality among men.25 Discrepancies in these results may be from unreported differences in the multivitamin combinations used. In an early analysis of the Nurses' Health Study with follow-up to 8 years, multivitamin use had no significant effect on coronary events.18 However, a subsequent analysis of this same cohort after 14 years showed an association between multivitamin use and reduced risk of coronary events.19 Women who reported using a multivitamin supplement on most days for at least 5 years had the lowest risk. In the Physicians' Health Study, there was no impact on cardiovascular or coronary heart disease mortality after 4 years.27 Similarly, a fair quality report from a cohort of more than 1 million men and women demonstrated no benefit on cardiovascular mortality.25 All-cause mortality in this study was increased by multivitamin supplement use in men only. However, when a multivitamin supplement plus antioxidant was considered, cardiovascular mortality was reduced in both men and women.
A secondary prevention trial of coronary heart disease studied a multivitamin combination (folate, vitamin B12, and pyridoxine) and placebo on restenosis at 6 months after angioplasty.53 The rate of restenosis was significantly reduced, as was the rate of cardiovascular events; neither incident myocardial infarction nor cardiovascular mortality was not affected significantly.
Adverse effects of vitamin supplementation are best measured in clinical trials. In most studies of vitamin supplementation, adverse effects were not reported, as might be expected in a pharmacologic trial. The Heart Protection Study reported no difference in cognitive impairment, respiratory disease, and fracture when comparing antioxidant therapy with placebo.40 However, an increase in levels of plasma triglycerides, low density lipoprotein cholesterol, and plasma total cholesterol was observed.40 Only a non-significant increase in triglyceride levels was noted in CARET.58 In the ATBC32 study and CARET,59 a significant increase in lung cancer incidence and lung cancer mortality was observed in smokers and was ascribed primarily to beta-carotene supplementation. A non-significant increase in non-hemorrhagic stroke with vitamin E supplementation was observed in the ATBC study, although this was not reported in other studies.32 Heavy smokers may represent a subgroup of the population who should use antioxidants with caution.