Evidence reviews for the US Preventive Services Task Force (USPSTF) use an analytic framework to visually display the key questions that the review will address to allow the USPSTF to evaluate the effectiveness and safety of a preventive service. The questions are depicted by linkages that relate interventions and outcomes. For additional information see the USPSTF Procedure Manual.9
aReasons for exclusion: Intervention relevance: Study used an excluded intervention. Outcomes: Study did not report relevant outcomes. Design: Study did not use an included design. Setting: Study not conducted in a country relevant to US practice. Population: Study not conducted in community-dwelling adults without chronic disease and without nutritional deficiencies. Language: Not available in English. Publication type: Conference abstract. Study relevance: Study aim not relevant. Harms study design: Short-term nonserious harms reported, but lack of evidence of benefit. High dose: Supplement dose greater than the tolerable upper intake level as determined by the National Academies of Sciences, Engineering, and Medicine Food and Nutrition Board. Quality: Study did not meet criteria for fair or good quality.
bStudies could be included for more than 1 key question (KQ).
Mantel-Haenszel fixed-effects model used unless otherwise specified. Percent with an event is calculated as the weighted mean percent with an event across the studies included in the analysis, weighted by the number of participants analyzed in the relevant group for each study. Results do not correspond directly to pooled odds ratios (ORs) because weighting in meta-analysis models differs from this approach.
aRestricted maximum likelihood model with the Knapp-Hartung (REML-KH) correction used.
bEvidence shown is a hazard ratio from a single study, PHS-II. PHS-II provided the number of cardiovascular disease (CVD) events rather than the number of persons experiencing a CVD event, so the percentage of participants with an event was not calculable.
cThe Peto OR was used with REML-KH correction.
dEvidence shown is from a single study, SKICAP (Skin Cancer Prevention).
eMethods. Literature Search Strategies for Primary Literature
eTable 1. Inclusion and Exclusion Criteria
eTable 2. Design-Specific Quality Rating Criteria
eTable 3. List of All Included Studies
eTable 4. Table of Study, Intervention, and Population Characteristics for Studies of Multivitamin Use, Sorted by Study Design, Then Author (Key Questions 1 and 2)
eTable 5. Table of Study, Intervention, and Population Characteristics for Studies of Beta-Carotene Use, Sorted by Study Design, Then Author (Key Questions 3 and 4)
eTable 6. Table of Study, Intervention, and Population Characteristics for Studies of Vitamin A Use, Sorted by Study Design, Then Author (Key Questions 3 and 4)
eTable 7. Table of Study, Intervention, and Population Characteristics for Studies of Vitamin E Use, Sorted by Study Design, Then Author (Key Questions 3 and 4)
eTable 8. Table of Study, Intervention, and Population Characteristics for Studies of Vitamin D Use, With or Without Calcium, Sorted by Study Design, Then Author (Key Questions 3 and 4)
eTable 9. Results of Sensitivity and Subgroup Meta-analyses for Vitamin D
eTable 10. Table of Study, Intervention, and Population Characteristics for Studies of Calcium Use Alone, Sorted by Study Design, Then Author (Key Questions 3 and 4)
Customize your JAMA Network experience by selecting one or more topics from the list below.
O’Connor EA, Evans CV, Ivlev I, et al. Vitamin and Mineral Supplements for the Primary Prevention of Cardiovascular Disease and Cancer: Updated Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA. 2022;327(23):2334–2347. doi:10.1001/jama.2021.15650
Cardiovascular disease and cancer are the 2 leading causes of death in the US, and vitamin and mineral supplementation has been proposed to help prevent these conditions.
To review the benefits and harms of vitamin and mineral supplementation in healthy adults to prevent cardiovascular disease and cancer to inform the US Preventive Services Task Force.
MEDLINE, PubMed (publisher-supplied records only), Cochrane Library, and Embase (January 2013 to February 1, 2022); prior reviews.
English-language randomized clinical trials (RCTs) of vitamin or mineral use among adults without cardiovascular disease or cancer and with no known vitamin or mineral deficiencies; observational cohort studies examining serious harms.
Data Extraction and Synthesis
Single extraction, verified by a second reviewer. Quantitative pooling methods appropriate for rare events were used for most analyses.
Main Outcomes and Measures
Mortality, cardiovascular disease events, cancer incidence, serious harms.
Eighty-four studies (N=739 803) were included. In pooled analyses, multivitamin use was significantly associated with a lower incidence of any cancer (odds ratio [OR], 0.93 [95% CI, 0.87-0.99]; 4 RCTs [n=48 859]; absolute risk difference [ARD] range among adequately powered trials, −0.2% to −1.2%) and lung cancer (OR, 0.75 [95% CI, 0.58-0.95]; 2 RCTs [n=36 052]; ARD, 0.2%). However, the evidence for multivitamins had important limitations. Beta carotene (with or without vitamin A) was significantly associated with an increased risk of lung cancer (OR, 1.20 [95% CI, 1.01-1.42]; 4 RCTs [n=94 830]; ARD range, −0.1% to 0.6%) and cardiovascular mortality (OR, 1.10 [95% CI, 1.02-1.19]; 5 RCTs [n=94 506] ARD range, −0.8% to 0.8%). Vitamin D use was not significantly associated with all-cause mortality (OR, 0.96 [95% CI, 0.91-1.02]; 27 RCTs [n=117 082]), cardiovascular disease (eg, composite cardiovascular disease event outcome: OR, 1.00 [95% CI, 0.95-1.05]; 7 RCTs [n=74 925]), or cancer outcomes (eg, any cancer incidence: OR, 0.98 [95% CI, 0.92-1.03]; 19 RCTs [n=86 899]). Vitamin E was not significantly associated with all-cause mortality (OR, 1.02 [95% CI, 0.97-1.07]; 9 RCTs [n=107 772]), cardiovascular disease events (OR, 0.96 [95% CI, 0.90-1.04]; 4 RCTs [n=62 136]), or cancer incidence (OR, 1.02 [95% CI, 0.98-1.08]; 5 RCTs [n=76 777]). Evidence for benefit of other supplements was equivocal, minimal, or absent. Limited evidence suggested some supplements may be associated with higher risk of serious harms (hip fracture [vitamin A], hemorrhagic stroke [vitamin E], and kidney stones [vitamin C, calcium]).
Conclusions and Relevance
Vitamin and mineral supplementation was associated with little or no benefit in preventing cancer, cardiovascular disease, and death, with the exception of a small benefit for cancer incidence with multivitamin use. Beta carotene was associated with an increased risk of lung cancer and other harmful outcomes in persons at high risk of lung cancer.
Cardiovascular disease and cancer are the 2 leading causes of death in the US.1 Vitamin and mineral supplementation has been proposed as a preventive strategy for both diseases because of shared disease pathways involving oxidative stress, inflammation, and methionine metabolism.2-4 Further, observational evidence has suggested associations between higher plasma levels of various vitamins and minerals and lower rates of cardiovascular disease and cancer.5,6 Vitamin and mineral supplements are commonly used in the US, with estimates in 2011 to 2014 showing that 52% of adults reported having recently used at least 1 dietary supplement.7
In 2014, the US Preventive Services Task Force (USPSTF) recommended against the use of beta carotene or vitamin E to prevent cardiovascular disease and cancer and concluded that the evidence was insufficient to assess net benefit for multivitamins or the use of single- or paired-nutrient supplements.8 This systematic review was conducted to provide current evidence on the benefits and harms of vitamin and mineral supplementation in healthy adults without known vitamin or mineral deficiencies to inform an updated recommendation by the USPSTF.
Figure 1 displays the a priori–developed analytic framework and 4 key questions (KQs) that guided this review, which was posted on the AHRQ website on September 5, 2019.10 Methodological details and findings for the B and C vitamins, folic acid, magnesium, selenium, and zinc are available in the full evidence report.11
MEDLINE, PubMed (publisher-supplied records only), Cochrane Central Register of Controlled Trials and Database of Systematic Reviews, and Embase were searched for relevant English-language articles published after the 2014 review for the USPSTF (January 1, 2013, through February 1, 2022 [eMethods in the Supplement]).12 All studies in the prior review were also evaluated,12 as well as reference lists of relevant systematic reviews. ClinicalTrials.gov and the World Health Organization International Clinical Trials Registry Platform were searched for relevant ongoing trials.
Titles, abstracts, and full-text articles were reviewed by investigators against prespecified eligibility criteria (eTable 1 in the Supplement). Discrepancies were resolved by consensus. English-language fair- and good-quality randomized clinical trials (RCTs) were included that evaluated multivitamins/minerals (KQ1 and KQ2), and single nutrients or functionally related nutrient pairs (KQ3 and KQ4) compared with placebo or no intervention and reported cardiovascular disease, cancer, mortality, serious harms, or nonserious adverse events reported by at least 5% of the intervention group. For serious harms, comparative observational studies (cohort or case-control) or postmarket surveillance data were also eligible. A stated study aim of cardiovascular disease or cancer prevention was not required for inclusion; thus, trials of supplements designed to prevent other conditions were included if outcomes of interest were reported. A minimum of 1-year follow-up was required for all-cause mortality, and no minimum follow-up was required for all other outcomes. Studies were required to be conducted in countries classified as “very high” on the 2017 Human Development Index.13 Eligible populations included community-dwelling adults 18 years or older without chronic disease and without vitamin, mineral, or nutritional deficiencies. Studies among persons with cardiovascular disease risk factors, a history of colorectal adenoma, or previous nonmelanoma skin cancer were included.
Data were extracted from each included study into standardized evidence tables by 1 investigator. Data accuracy was confirmed by a second investigator. Study characteristics, dosing details, participant demographics, and results for mortality, cardiovascular disease, cancer, and harms were extracted. Only published data were extracted; investigators were not contacted to supply missing fields. The quality of each study was assessed by 2 reviewers who independently applied USPSTF design-specific criteria (eTable 2 in the Supplement).9 Each study was assigned a quality rating of “good,” “fair,” or “poor.” Discordant quality ratings were resolved by consensus. Studies rated as poor quality were excluded.
Summary tables for all KQs were created for each supplement. Quantitative pooling was conducted when at least 3 studies of the same supplement reported the same outcome. A single effect per study was included in each meta-analysis, preferentially selecting the time point corresponding with the end of supplement use. Data for beta carotene and vitamin A are summarized together because beta carotene is a vitamin A precursor.
Peto odds ratios (ORs) with a restricted maximum likelihood (REML) model were used when events occurred in less than 5% of the sample for most studies in the analysis. When events typically occurred in 5% to 10% of the sample, a fixed-effects Mantel-Haenszel model was used as the primary analysis. When events had a higher incidence, standard ORs using a REML model were pooled, adding the Knapp-Hartung correction for pooling a small number of studies.14,15 Because absolute event rates were highly variable for most analyses, sensitivity analyses using alternative pooling methods were conducted (see full report11). The presence of statistical heterogeneity among the studies was assessed using the I2 statistic. I2 values were not generated for fixed-effects models, so I2 from random-effects sensitivity analyses are reported, if available.
Stata version 16 (StataCorp) was used for all quantitative analyses. All significance testing was 2-sided, and results were considered statistically significant at P < .05.
A total of 84 studies (N = 739 803) (eTable 3 in the Supplement) were included, comprising 78 RCTs (n = 324 837)16-93 and 6 cohort studies (n = 390 689),71,94-99 after review of 17 459 unique citations and 379 full-text articles (Figure 2). Fifty-two of the included studies were newly identified since the last review.12 The included studies addressed multivitamins; vitamins A, B3, B6, B12, C, D, and E; beta carotene; folic acid; calcium; magnesium; selenium; and zinc. The evidence for the B and C vitamins, folic acid, magnesium, selenium, and zinc was low, insufficient, or absent for all outcomes. Results for these supplements can be found in the full evidence report.11
The mean age across all included studies was 61.0 years. An estimated 65.1% of all participants were women. Most participants were White in studies conducted in the US (with data from 22 of 36 studies). An estimated 19.6% of participants were Black, among studies reporting race and ethnicity. The vitamin D trials had, on average, greater representation of Black participants. Other racial and ethnic groups had minimal representation for all supplements.
Key Question 1. What is the efficacy of multivitamin supplementation for reducing cardiovascular disease, cancer, and mortality in the general adult population?
Nine RCTs addressed KQ1 (n = 51 945) (eTable 4 in the Supplement).16,26,31,43,54,70,72,81,93 Three large studies had primary aims of cardiovascular disease and cancer prevention, were all rated as good quality, and comprised most of the evidence for this KQ.16,26,93 These were the Supplementation en Vitamines et Minéraux AntioXydants (SU.VI.MAX) study (n = 13 017),16 which examined use of an antioxidant-focused supplement among adults aged 35 to 60 years; the Physicians’ Health Study II (PHS-II [n = 14 641]), which examined a broad-spectrum supplement among male physicians 50 years or older26; and the Cocoa Supplement and Multivitamin Outcomes Study (COSMOS [n = 21 442]), which examined a broad-spectrum supplement among adults aged 60 years (men) or 65 years or older (women).93 The other 6 RCTs were small, with a variety of other aims, and 5 of these did not report a robust ascertainment process for the all-cause mortality, cardiovascular disease, or cancer outcomes. The evidence suggested small to no benefit of multivitamin use for all-cause mortality, no benefit for cardiovascular disease, and a possible small benefit for cancer outcomes (Figure 3). In pooled analyses, the association with all-cause mortality was not statistically significant (OR, 0.94 [95% CI, 0.87-1.01]; 9 RCTs [n = 51 550]; I2 = 0%). The largest trial, COSMOS, reported that 3.4% of participants taking a multivitamin had died after a median of 3.6 years of follow-up, compared with 3.6% who were taking a placebo (hazard ratio [HR], 0.93 [95% CI, 0.81-1.08] [n = 21 442]),93 and effect sizes were very similar in the other 2 trials. The pooled effect sizes were also similar for cancer mortality (OR, 0.94 [95% CI, 0.81-1.09]; 4 RCTs [n = 37 400]; I2 = 28.9%) and cancer incidence (OR, 0.93 [95% CI, 0.87-0.99]; 4 RCTs [n = 48 859]; I2 = 0%; absolute risk difference range among adequately powered trials, −0.2% to −1.2%). For cancer incidence, which showed a statistically significant pooled effect, 4.8% of participants in COSMOS taking a multivitamin had developed invasive cancer after 3.6 years, compared with 5.0% taking placebo (HR, 0.97 [95% CI, 0.86-1.09]; [n = 21 442])100; effects were slightly larger in SU.VI.MAX (RR, 0.90 [95% CI, 0.76-1.06] after 7.5 years) and PHS-II (HR, 0.92 [95% CI, 0.86-1.00] after 11.2 years). The pooled effect was also statistically significant for lung cancer, an outcome reported only by COSMOS and PHS-II (OR, 0.75 [95% CI, 0.58-0.95]; 2 RCTs [n = 36 052]; I2 = 30%; absolute risk difference, −0.2% in both studies).
Key Question 2. What are the harms of multivitamin supplementation in the general adult population?
Harms of multivitamin use were reported in 9 RCTs (n = 51 614) (eTable 4 in the Supplement)16,26,31,48,54,70,72,81 and 3 cohort studies (n = 188 027).94,96,99 Among the 4 trials reporting any adverse effects,54,81 serious adverse effects,31 or withdrawals due to adverse effects,48 no group differences were found, although there were very few serious adverse effects or withdrawals due to adverse effects. With regard to specific adverse effects, PHS-II found an increased risk of rash (29.0% among multivitamin users, 27.3% among nonusers; OR, 1.06 [95% CI, 1.01-1.12]) and nosebleeds (21.6% among multivitamin users, 19.8% among nonusers; OR, 1.09 [95% CI, 1.02-1.16]). Small increases in cataracts94,99 and hip fractures94 reported by cohort studies were not statistically significant and were not reported by any of the trials. None of the harms were replicated in COSMOS, which also found very few group differences among many other assessed potential adverse effects.93
Key Question 3. What is the efficacy of supplementation with single nutrients or functionally related nutrient pairs for reducing cardiovascular disease, cancer, and mortality in the general adult population?
Six RCTs addressed KQ3 for beta carotene and vitamin A (eTable 5 and eTable 6 in the Supplement). These studies evaluated the use of 20 to 50 mg/d of beta carotene (n = 112 820)17-21,32; 1 trial (n = 18 314) examined the combined use of beta carotene and 25 000 IU/d of vitamin A.19 Two of these studies—the original Physicians’ Health Study (PHS-I)20 and the Women’s Health Study (WHS)18—had broad cancer and cardiovascular disease prevention aims in men20 or women.18 Both were factorial design trials that also evaluated aspirin, as well as vitamin E in the WHS. Two trials, the Alpha-Tocopherol Beta Carotene Cancer Prevention (ATBC) trial21 and the Beta-Carotene and Retinol Efficacy Trial (CARET),19 had primary aims of lung cancer prevention and evaluated beta carotene supplementation in high-risk populations such as smokers and asbestos-exposed workers. ATBC was multifactorial, with additional randomization to 50 mg/d of vitamin E. The other 2 beta carotene studies were more narrowly aimed at primary17 or secondary32 prevention of skin cancer. One additional RCT examined the effect of 25 000 IU of vitamin A among adults with moderate risk for new nonmelanoma skin cancer (n = 2297).33
Pooled estimates showed statistically significant paradoxical harm associated with beta carotene use (Figure 3). The most pronounced risk increase was for lung cancer, with the pooled estimate showing a statistically significantly increased risk over 3.7 to 12 years of follow-up (OR, 1.20 [95% CI, 1.01-1.42]; 4 RCTs [n = 94 830]; I2 = 38.8%). Absolute risk differences in individual trials ranged from −0.1% to 0.6%. These estimates included trials in general populations and those at high risk of lung cancer, and the strongest evidence was from the trials of people at high risk of lung cancer. Cardiovascular disease mortality similarly showed an increased risk (OR, 1.10 [95% CI, 1.02-1.19]; 5 RCTs [n = 94 506]; I2 = 0%). Absolute risk differences in individual trials ranged from −0.8% to 0.8%. In pooled analyses, the OR for all-cause mortality associated with beta carotene use was 1.06 (95% CI, 1.00-1.12; 6 RCTs [n = 112 820]; I2 = 6.4%). When the study of vitamin A supplementation (alone) was included in the meta-analysis,33 the all-cause mortality finding became statistically significant (OR, 1.06 [95% CI, 1.01-1.12]; 7 RCTs [n = 115 117]).
Nine RCTs addressed KQ3 for vitamin E (n = 116 468) (eTable 7 in the Supplement).18,21,22,25,26,51,74,77,79 Seven RCTs (n = 86 142) had an explicit aim to prevent cardiovascular disease18,25,26,51 or related outcomes,22,74,79 most among adults at increased risk for cardiovascular disease, due to either smoking history21,79 or other cardiovascular disease risk factors.22,51,74 Three of the trials with cardiovascular disease aims also had a cancer prevention aim.18,21,25,26 Doses ranged from 50 to 300 mg/d for 3 to 10 years, and follow-up time ranged from 3 to 24 years. One trial (n = 34 888) examined vitamin E with or without 200 µg of selenium daily,25 and another small trial (n = 520) had a similar design including 500 mg of vitamin C.22
Evidence indicated that vitamin E had no benefit for mortality, cardiovascular disease, or cancer. For example, pooled evidence demonstrated no statistically significant association between vitamin E use and all-cause mortality (OR, 1.02 [95% CI, 0.97-1.07]; 9 RCTs [n = 107 772]; I2 = 0%) or the composite outcome of any cardiovascular disease event (OR, 0.96 [95% CI, 0.90-1.04]; 4 RCTs [n = 62 136]; I2 = 0%) or incidence of any cancer (OR, 1.02 [95% CI, 0.98-1.08]; 5 RCTs [n = 76 777]; I2 = 0%) (Figure 3). Effect sizes were very similar when vitamin E was used with or without selenium.25 Additionally, 221,26 of 4 trials reporting hemorrhagic stroke or hemorrhagic stroke mortality showed statistically significant increases in these rare outcomes in groups randomized to vitamin E. In PHS-II, 0.5% among those taking vitamin E and 0.3% among those taking placebo experienced a hemorrhagic stroke (HR, 1.74 [95% CI, 1.04-2.90]). In the ATBC study of smokers, risk of hemorrhagic stroke death was similarly elevated (calculated OR, 1.50 [95% CI, 1.03-2.20]; vitamin E: 0.5%, placebo: 0.3%).21
Thirty-two RCTs addressed KQ3 for vitamin D (n = 123 140) (eTable 8 in the Supplement).24,28,34,35,38-41,47,50,58-60,62-64,66-69,72,73,75,76,78,82,83,86-90 Most of the studies had aims related to bone density, fractures, or falls and were primarily limited to adults 55 years or older. However, 5 explicitly aimed to prevent cardiovascular disease,24,35,39,41,89 and 7 had a cancer prevention aim.24,28,35,38,40,41,89 The 3 largest studies were the WHI (n = 36 282)24 which examined the effects of 400 IU vitamin D and 1000 mg calcium use daily; the Vitamin D and Omega-3 Trial (VITAL, n = 25 871),41 which tested the effects of 2000 IU/d of vitamin D, with or without an omega-3 fatty acid supplement; and D-Health, which examined the use of 2000 IU daily.90 Both WHI and VITAL had specific aims of cancer and cardiovascular disease prevention among adults 50 years or older, while mortality reduction among 60- to 84-year-old adults was the aim of D-Health. Among all trials, doses ranged from 20 to 5000 IU/d for 1 month to 7 years and follow-up time ranged from 1 month to 11.9 years. The mean age was 66 years, and an estimated 75% of participants in all trials were women.
Pooling studies of vitamin D with or without calcium cosupplementation showed no significant reduction in all-cause mortality (OR, 0.96 [95% CI, 0.91-1.02]; 27 RCTs [n = 117 082]; I2 = 0%), cardiovascular disease (eg, composite cardiovascular disease event outcome: OR, 1.00 [95% CI, 0.95-1.05]; 7 RCTs [n = 74 925]; I2 = 0%), or cancer outcomes (eg, any cancer incidence: OR, 0.98 [95% CI, 0.92-1.03]; 19 RCTs [n = 86 899]; I2 = 0%) (Figure 3). No clear effect modifiers were identified in sensitivity analyses or meta-regression (eTable 9 in the Supplement). For example, point estimates for all-cause mortality did not differ for vitamin D without calcium (OR, 0.968 [95% CI, 0.92-1.05]; 20 RCTs [n = 74 398]; I2 = 0%) and vitamin D administered with calcium (OR, 0.93 [95% CI, 0.85-1.01]; 8 RCTs [n = 45 322]; I2 = 0%). Similarly, findings were almost identical to the overall findings when limited to trials that reported robust outcome ascertainment methods (OR, 0.96 [95% CI, 0.91-1.02]; 12 RCTs [n = 103 457]; I2 = 0%), rather than trials that assessed outcomes incidentally through adverse events reporting or not reporting the source of these outcomes. In addition, there was no clear association between effect size and vitamin D dose or the use of bolus dosing (eg, 100 000 IU monthly) vs daily doses (interaction P = .12).
Seven RCTs addressed KQ3 for calcium (n = 11 884, eTable 10 in the Supplement).27,28,30,35,38,52,53 The most common doses were 1000 and 1200 mg/d, and duration of use ranged from 6 months to 5 years. Follow-up time ranged from 6 months to 12 years. The largest study was the RECORD trial (n = 5292), which examined the effects of 1000 mg/d of calcium, with or without 800 IU/d of vitamin D, on cardiovascular disease and cancer outcomes among older adults with fragility fractures.35
Most of the evidence indicated that calcium had no benefit for mortality, cardiovascular disease, or cancer. Pooled effects uniformly indicated no group differences, and very few individual study findings demonstrated an effect of calcium supplementation on cancer, cardiovascular disease, or mortality. For example, pooled estimates for all-cause mortality (OR, 1.05 [95% CI, 0.92-1.21]; 6 RCTs [n = 8394]; I2 = 0%), cardiovascular disease events (OR, 1.11 [95% CI, 0.90-1.36]; 4 RCTs [n = 4076]; I2 = 0%), and any incidence of cancer (OR, 0.94 [95% CI, 0.41-2.14]; 3 RCTs [n = 5051]; I2 = 49.2%) all showed no statistically significant association with calcium use (Figure 3).
Key Question 4. What are the harms of supplementation with single nutrients in the general adult population?
Six RCTs17-21,32 and 1 cohort study94 reported on the harms of beta carotene supplementation, with or without the use of other supplements (eTable 5 in the Supplement). The most prominent harms were the paradoxical harms of increased all-cause mortality, cardiovascular disease mortality, and lung cancer described under KQ3. Other than these outcomes, there was a consistent and statistically significant increased risk of hypercarotenodermia with beta carotene use in the 4 trials reporting this adverse event at 2 to 12 years of follow-up.18,20,21,32 The only other harm for which there was a statistically significant increased risk from beta carotene in an RCT was gastrointestinal symptoms in PHS-I.20 One cohort study limited to women found no statistically significant association between hip fractures and beta carotene (adjusted risk ratio [RR], 0.91 [95% CI, 0.57-1.44]).94 In addition to the trial of vitamin A alone reporting an increase in any adverse effects (OR, 1.77 [95% CI, 1.49-2.09] [n = 2264]),33 2 cohort studies explored harms of vitamin A (eTable 6 in the Supplement).94,95 A higher but not statistically significant risk of hip fracture with vitamin A use was suggested by both the Nurses’ Health Study (NHS-I) (adjusted RR, 1.50 [95% CI, 0.99-1.99])94 and the Iowa Women’s Health Study (RR, 1.18 [95% CI, 0.99-1.41]).95 The Iowa Women’s Health Study found no statistically significant association between vitamin A use and overall fracture risk (adjusted RR, 1.00 [95% CI, 0.95-1.05]).95 The NHS-I indicated no clear association between vitamin A use and cataract extraction.
Harms of vitamin E were reported in 7 RCTs18,21,22,25,26,51,77 (n = 115 576) and 2 cohort studies94,96 (n = 149 043) (eTable 7 in the Supplement). The 3 trials that reported the total number of adverse events,51,77 serious adverse events,77 or withdrawals due to adverse events22,77 found no group differences. Trial evidence also supported no group differences in hospitalization from pneumonia,21 gastrointestinal disease,51 several bleeding outcomes,26,51 fatigue,25 nail changes,25 halitosis,25 easy bruising,26 and noncataract ophthalmic events.77 However, some of these results were based on a very small or unknown number of events. Two21,26 of 4 trials that reported an increased risk of hemorrhagic stroke or hemorrhagic stroke mortality are discussed above in KQ3.
PHS-II found no increase in the incidence of cataracts with vitamin E use at 8 years of follow-up (HR, 0.99 [95% CI, 0.88-1.11]).26 Similarly, a large cohort study of women (NHS-I) (n = 121 700) with supplement use assessed biannually found no association with cataracts.94 However, a smaller cohort study (n = 27 343) of Swedish men found a higher incidence of cataracts among men who reported any vitamin E use compared with no use on a 1-time survey at 8.4 years of follow-up (HR, 1.57 [95% CI, 1.10-2.22]).96
KQ4 outcomes for vitamin D were reported in 31 RCTs24,28,34-36,38-42,47,50,55,58,61,63-67,71-73,75,76,82,83,86,88-90 (n = 117 100) and 3 cohort studies (n = 289 659) (eTable 8 in the Supplement).94,97,98 Among RCTs reporting the percent of participants experiencing any adverse events,36,39,61,63,65,76,87,90 any serious adverse events,28,40,50,83,86,90 or withdrawal due to adverse events,40,42,50,67,71,75,83,86,90 only 1 found an increase in withdrawals due to adverse events in a trial that administered 10 000 IU/wk of vitamin D plus 1000 mg/d of calcium to postmenopausal women.75
While most trials that reported data on kidney stones had very few events, 2 of the 3 largest trials indicated a small increased risk.28,38-41,58,73 In WHI, 2.5% of participants who were taking 400 IU of vitamin D and 1000 mg of calcium daily developed kidney stones after 7 years, compared with 2.1% in the placebo group (HR, 1.17 [95% CI, 1.02-1.34]). VITAL found a similar effect size, although it was not statistically significant (HR, 1.12 [95% CI, 0.99-1.28]). After 5.3 years, 3.7% of those who were taking 2000 IU/d of vitamin D developed kidney stones vs 3.3% of those in the placebo group. However, there was no significant association found in the D-Health study (incidence rate ratio, 1.03 [95% CI, 0.82-1.28]); 1.5% in the vitamin D group vs 1.4% in the placebo group). Two of the cohort studies97,98 found an increased risk of kidney stones with use of 1000 IU/d or more of vitamin D after 20 to 26 years, compared with no vitamin D use, but only 1 of these findings was statistically significant.97 There was no suggestion of increased risk with lower doses in either of these studies. The third cohort study, NHS-I,94 found no association between any dose of vitamin D and kidney stones. No statistically significant group differences were identified among a wide array of nonserious harms.
Harms of calcium were reported in 8 RCTs27,28,30,35,38,52,53,61 (n = 12 961), and 1 cohort study (n = 121 700) (eTable 10 in the Supplement).94 Studies that reported the occurrence of any adverse events,61 any serious adverse events,28 and withdrawals due to adverse events30,53 identified very few events and found no group differences. Constipation and gastrointestinal symptoms were generally increased with calcium use, but findings were statistically significant in only 3 studies.27,35,52 Evidence from 5 trials suggested no increased risk of fractures.27,35,38,52,53 The cohort study, NHS-I, included only women and reported an increased incidence of kidney stones for any calcium use compared with no calcium use, but no dose-response trend was identified.94 Evidence on kidney stones from the trials was inconclusive due to the small numbers of events.
This updated evidence review examined the use of vitamins and minerals for primary prevention of cardiovascular disease and cancer; the evidence is summarized in the Table. The findings from 84 RCTs and 6 cohort studies suggest that most vitamin and mineral supplements provide no clinically important protective effects for cardiovascular disease, cancer, or all-cause mortality in healthy adults without known nutritional deficiencies. One exception was a slightly lower risk of cancer incidence with multivitamin use. However, the evidence for multivitamins had important limitations, including only 3 adequately powered trials, 1 with a median of only 3.6 years of multivitamin use and another that was limited to antioxidants.
Other than the new finding related to multivitamin use and lower cancer incidence, these conclusions are generally consistent with those of the previous review for the USPSTF on this topic.12 Vitamin E had the strongest body of evidence demonstrating no benefit for outcomes relevant to this review. These updated review findings also confirm the previous review’s finding that beta carotene supplementation, especially with concomitant vitamin A use, likely increases the risk of lung cancer incidence, particularly in those at high risk for lung cancer. New evidence in this update was predominately for vitamin D supplementation. Despite the new inclusion of 32 RCTs and 2 cohort studies, pooled estimates for all-cause mortality were similar to that in the prior review with confidence intervals only slightly crossing 1 and point estimates suggesting at most a very small benefit.
This review found minimal other recent synthesized evidence on the effect of multivitamin use, but another review concluded that observational studies suggest a possible lower breast cancer recurrence among breast cancer survivors using multivitamins.101 The findings for vitamin D in the current review are generally consistent with those from other reviews, for example, pooled estimates in the range of 0.93 to 0.97 that may not be statistically significant for all-cause mortality.102-104 In general, the statistical significance of an all-cause mortality benefit in pooled analyses is unstable, being sensitive to the number of included studies and which study aims are considered. For vitamin E, another review of primary prevention in adults concluded that vitamin E may reduce the risk of cardiovascular disease mortality.105 The pooled analysis for cardiovascular disease mortality was not statistically significant in the current review, although the point estimate was in the direction of benefit (OR, 0.88 [95% CI, 0.74-1.04]). The point estimate in the other review was the same but was statistically significant (RR, 0.88 [95% CI, 0.80-0.96]).105 The other review included studies of multivitamins that contained vitamin E in addition to vitamin E alone, in contrast to the meta-analysis in the current review, which was limited to intervention groups examining vitamin E alone. While this might indicate a relatively small effect that is detectable in only very large pooled analyses, the lack of association with all-cause mortality and cardiovascular disease events and the lack of statistical significance in the current review led to the conclusion that vitamin E most likely has little to no effect on cardiovascular disease mortality, although some uncertainty remains.
A general limitation of literature included in this review is that the effects of individual micronutrients on human health are very difficult to detect in generally healthy populations with adequate nutrition. Supplement exposure is complicated by exposure to nutrients through dietary intake, and some studies reported fairly high levels of independent use of supplements among their study populations. There is variability in how individuals absorb and metabolize nutrients, and interactions among nutrients and between nutrients and myriad enzymes and hormones in the human body complicate the ability to detect their effects.
There is a lack of information about whether broad-spectrum multivitamins (rather than antioxidant-focused formulations) prevent cardiovascular disease and cancer in general populations including both men and women. Other limitations of this evidence include insufficient information on the effect of vitamins and minerals in Black and Native American populations, in whom the burden of cardiovascular disease and cancer is known to be high; limitations or uncertainty about the quality of cancer and cardiovascular disease outcome data in studies that were not designed for these outcomes; and likely insufficient follow-up in most studies, since cardiovascular disease and cancer may take a decade or more to manifest.
This review has several limitations. First, there may be other benefits of some supplements that were not covered in this review owing to its focus on cardiovascular disease and cancer prevention. For example, folic acid use in women who are pregnant or soon to be pregnant is known to be valuable for prevention of neural tube defects in their offspring.106 Second, because of the focus on studies in predominantly healthy populations without known nutritional deficiencies, this review also did not cover therapeutic use of supplements in persons with physical symptoms, medical conditions, or nutritional deficits. Third, owing to the focus on serious harms, this review of nonserious harms is not comprehensive. The risks of high doses were not generally addressed here, but are comprehensively documented in an Institute of Medicine Report on dietary reference intakes that addresses setting tolerable upper limits.107 However, studies with vitamin A and vitamin D doses above the recommended upper limit were included for consistency with the previous review. Fourth, because of the large number of analyses, there is the potential for false-positive findings due to chance. Fifth, there may be other doses, formulations, or supplement combinations that could be beneficial or less harmful for which the review did not have the data to explore.
Vitamin and mineral supplementation was associated with little or no benefit in preventing cancer, cardiovascular disease, and death, with the exception of a small benefit for cancer incidence with multivitamin use. Beta carotene was associated with an increased risk of lung cancer and other harmful outcomes in persons at high risk of lung cancer.
Corresponding Author: Elizabeth O’Connor, PhD, Kaiser Permanente Evidence-based Practice Center, The Center for Health Research, Kaiser Permanente Northwest, 3800 N Interstate Ave, Portland, OR 97227 (Elizabeth.OConnor@kpchr.org).
Accepted for Publication: August 30, 2021.
Author Contributions: Dr O’Connor had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: O’Connor, Evans, Ivlev, Rushkin, Lin.
Acquisition, analysis, or interpretation of data: O’Connor, Evans, Ivlev, Rushkin, Thomas, Martin.
Drafting of the manuscript: O’Connor, Evans, Ivlev.
Critical revision of the manuscript for important intellectual content: Ivlev, Rushkin, Thomas, Martin, Lin.
Statistical analysis: O’Connor, Ivlev.
Obtained funding: O’Connor, Lin.
Administrative, technical, or material support: Evans, Rushkin, Thomas, Martin.
Supervision: Evans, Lin.
Conflict of Interest Disclosures: None reported.
Funding/Support: This research was funded under contract HHSA-290-2015-00007-I, Task Order 6, from the Agency for Healthcare Research and Quality (AHRQ), US Department of Health and Human Services, under a contract to support the US Preventive Services Task Force (USPSTF).
Role of Funder/Sponsor: Investigators worked with USPSTF members and AHRQ staff to develop the scope, analytic framework, and key questions for this review. AHRQ had no role in study selection, quality assessment, or synthesis. AHRQ staff provided project oversight, reviewed the report to ensure that the analysis met methodological standards, and distributed the draft for peer review. Otherwise, AHRQ had no role in the conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript findings. The opinions expressed in this document are those of the authors and do not reflect the official position of AHRQ or the US Department of Health and Human Services.
Additional Contributions: We gratefully acknowledge the following individuals for their contributions to this project: Tina Fan, MD, MPH, and Howard Tracer, MD, at the Agency for Healthcare Research and Quality; current and former members of the US Preventive Services Task Force who contributed to topic deliberations; the National Cancer Institute and the National Heart, Lung, and Blood Institute for providing federal partner review of the draft report; and Nadia Redmond, MSPH, Jill Pope, BA, and Melinda Davies, MAIS, for technical and editorial assistance at the Center for Health Research. USPSTF members, peer reviewers, and those commenting on behalf of partner organizations did not receive financial compensation for their contributions.
Additional Information: A draft version of this evidence report underwent external peer review from 5 content experts (Erin LeBlanc, MD, MPH, Kaiser Permanente Center for Health Research; Edward Giovannucci, MD, ScD, Harvard University; Stephanie Weinstein, PhD, National Cancer Institute; Catharine Ross, PhD, Penn State University; JoAnn Manson, MD, DrPH, MPH, Harvard University) and 2 federal partners, the National Heart, Lung, and Blood Institute and the National Cancer Institute. Comments were presented to the USPSTF during its deliberation of the evidence and were considered in preparing the final evidence review.
Editorial Disclaimer: This evidence report is presented as a document in support of the accompanying USPSTF Recommendation Statement. It did not undergo additional peer review after submission to JAMA.