Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Mar 29.
Published before final editing as: Thromb Haemost. 2026 Feb 16:10.1055/a-2806-3554. doi: 10.1055/a-2806-3554

Effects of Cocoa Extract and Multivitamin Supplementation on Venous Thromboembolism in the COSMOS Trial

Sarah Jaehwa Park 1,2, JoAnn E Manson 1,3, Eunjung Kim 1, Robert J Glynn 1,4, Pamela M Rist 1,3, Matthew A Allison 5, Howard D Sesso 1,2,3
PMCID: PMC13032829  NIHMSID: NIHMS2157890  PMID: 41651011

Abstract

Background

Epidemiological and experimental studies suggest cocoa flavanols and multivitamin-multimineral (MVM) supplements may confer arterial vascular benefits. However, their effects on clinical venous thromboembolic events have been infrequently examined.

Objective

To evaluate whether cocoa extract (CE) or MVM supplementation reduces the risk of venous thromboembolism (VTE) among older adults.

Methods

We conducted an ancillary study analysis of the COcoa Supplement and Multivitamin Outcomes Study (COSMOS), a completed randomized, double-blind, placebo-controlled, 2-by-2 factorial trial of CE and MVM supplementation for the prevention of cardiovascular disease and cancer among 21,442 older US adults. Our primary outcome was self-reported incident VTE, defined as the first reported deep vein thrombosis (DVT) or pulmonary embolism (PE) event after randomization; secondary outcomes were the individual components.

Results

Over a median follow-up of 3.5 years, 379 participants reported an incident VTE event (including 277 DVT and 165 PE). In intention-to-treat analyses, neither CE (HR: 0.88; 95% CI: 0.72, 1.08) nor MVM (HR: 0.89; 95% CI: 0.73, 1.09) significantly reduced VTE risk, with similar findings for DVT and PE. Exploratory latency and per-protocol analyses suggested potential patterns of benefit that merit further evaluation.

Conclusion

In this large trial of older adults, neither CE nor MVM supplementation significantly reduced the risk of VTE or its component parts in intention-to-treat analyses. Additional research may help clarify whether these supplements influence VTE risk in other contexts or populations.

Keywords: cocoa extract, multivitamin, venous thromboembolism, deep vein thrombosis, pulmonary embolism

Graphical Abstract

graphic file with name nihms-2157890-f0001.jpg

Introduction

Venous thromboembolism (VTE) is a common, multifactorial disease characterized by the formation of blood clots in the deep veins (deep vein thrombosis, DVT) and/or their migration to the lungs (pulmonary embolism, PE). It is the third leading cause of vascular mortality after myocardial infarction (MI) and stroke, and is responsible for an estimated 300,000 US deaths annually.1 Established risk factors for VTE include genetic predisposition, surgery, trauma, malignancy, immobilization, estrogen therapy, and obesity.2 Notably, a growing body of evidence suggests an overlap in several risk factors between venous and arterial thrombosis, with some important differences.3–8 To that end, research has sought to understand whether nutritional interventions that reduce the risk of atherosclerotic cardiovascular disease (ASCVD) could concomitantly protect against VTE.

Flavonoids are a group of plant compounds suggested to possess a wide range of potential health benefits, including a reduced risk of CVD and thrombotic complications.9–13 Cocoa, one of the richest sources of flavanols (a subclass of flavonoids), is thought to possess anti-thrombotic and anticoagulant effects.14 Small short-term mechanistic trials of cocoa flavanol intake have demonstrated significant improvements in several arterial risk factors including endothelial function,15,16 platelet reactivity,17,18 insulin sensitivity,19 and blood pressure.19,20 Indeed, in the COcoa Supplement and Multivitamin Outcomes Study (COSMOS),21 intention-to-treat analyses showed that daily flavanol-rich cocoa extract (CE) supplementation significantly reduced CVD death by 27%, but not total CVD events (including MI, stroke, coronary revascularization, CVD death, carotid artery disease, peripheral artery surgery, and unstable angina). Moreover, per-protocol analyses supported a significant 13% reduction in total CVD events among those compliant with the intervention in per-protocol analyses. However, although these findings suggest potential antithrombotic effects, no randomized clinical trials, to our knowledge, have evaluated the effects of CE supplementation on clinical venous thromboembolic events.

Multivitamin-multimineral (MVM) formulations are the most common dietary supplement taken in the US.22 Several observational studies have reported inverse associations between MVM use and incident MI,23,24 CVD death,25 and coronary heart disease,26,27 and randomized clinical trials testing individual or limited combinations of vitamins and minerals included in typical MVM formulations have reported reductions in idiopathic and unprovoked VTE.28,29 However, evidence from large-scale trials of MVM supplementation remains limited to COSMOS30 and the Physicians’ Health Study (PHS) II.31 In both studies, MVM supplementation did not significantly reduce clinical cardiovascular events, and neither evaluated its effects on VTE.

To our knowledge, neither a flavanol-rich CE supplement nor a broad-based MVM supplement have been tested for reducing clinical venous thromboembolic events. We therefore conducted an ancillary analysis of the COSMOS trial to evaluate whether long-term CE or MVM supplementation reduces self-reported incident VTE among older US adults.

Methods

Study Population: COcoa Supplement and Multivitamin Outcomes Study (COSMOS)

The COSMOS trial21,30 is a completed randomized, double-blind, placebo-controlled, 2-by-2 factorial trial that tested a CE supplement (2 capsules/d containing 500 mg cocoa flavanols/d including 80 mg (–)-epicatechin; supplied by Mars Edge) and an MVM supplement (Centrum Silver; supplied by Pfizer Consumer Healthcare) for the prevention of CVD and cancer. The COSMOS trial design has been previously described in greater detail else-where.32 Briefly, recruitment included mailings to active Women’s Health Initiative (WHI) Extension Study participants33; women and men contacted for, but not randomized into, the VITamin D and OmegA-3 TriaL (VITAL)34; and mass mailings and media outreach across the US. A total of 21,442 US adults were randomized into COSMOS between June 2015 and March 2018, including 12,666 women aged ≥65 years and 8,776 men aged ≥60 years initially free of MI, stroke, and recently diagnosed cancer (except for nonmelanoma skin cancer) within the past 2 years. The participant flowchart is summarized in Supplementary Fig. S1 (available in the online version only). All participants provided written informed consent, and the Institutional Review Board at Brigham and Women’s Hospital approved the research protocol.

Participants received follow-up questionnaires, with calendar packs containing study pills, at 6 and 12 months following randomization and semiannually thereafter to assess compliance with randomized treatments (missing ≤8 days/month of trial pills), use of non-trial CE and/or MVM supplements, potential side effects of the interventions, updated medical history, and other relevant lifestyle, clinical, and dietary risk factors. Randomized treatments continued until the scheduled end date on December 31, 2020 for a median treatment and follow-up time of 3.6 years in COSMOS.21 For the present ancillary study analysis of VTE outcomes, the median follow-up was 3.5 years; Graphical Abstract summarizes the key questions, methods, and findings of the present study.

Primary and Secondary Outcomes

The primary outcome for this ancillary study was a composite of self-reported incident VTE, defined as any new post-randomization DVT or PE event, whichever occurred first. Secondary outcomes were self-reported incident DVT and PE endpoints individually. Questionnaires asking about DVT and PE status were sent every 6 months throughout interventional follow-up. Self-reports were based on an affirmative response to either of the following questions: “In the past year, have you been newly diagnosed with deep vein thrombosis (blood clot in legs)? (no, yes)” or “In the past year, have you been newly diagnosed with pulmonary embolism (blood clot in lungs)? (no, yes).” Participants responding “yes” to either question were asked to provide the date (month/year) of the new diagnosis. When the date of diagnosis was not reported, we used the midpoint date between the date of the questionnaire indicating a new diagnosis and the date of the last completed questionnaire. Altough participants both with andwithout a history of VTE prior to randomization were included for analysis due to the limited number of events, we examined differences in the risk of developing a first and recurrent event in subgroup analyses.

Baseline Covariates

On the COSMOS baseline questionnaire, participants provided self-reported information on sociodemographic, lifestyle, and clinical risk factors including age (in years and categorized as (60–69, 70–79, 80 years), biological sex (male, female), race (white, non-white), education level (categorized as high school diploma/general educational development or less, attended or graduated college, post-college), weight and height (converted to body mass index and categorized as <25, 25–29.9, and ≥30 kg/m2), smoking status (categorized as never, past, current), menopausal hormone use (among women, categorized as never, past, current), aspirin use (no, yes), statin use (no, yes), anticoagulant use (no, yes), CE use before run-in (no, yes), MVM use before run-in (no, yes), and history of DVT (no, yes), PE (no, yes), hypertension (no, yes), CVD (no, yes), fracture within the past year (no, yes), or cancer (except for nonmelanoma skin cancer) within the past 2 years (no, yes). Participants also completed the Harvard Semiquantitative Food Frequency Questionnaire,35,36 which provided dietary data including chocolate consumption (monthly or less, weekly or daily) and Alternative Healthy Eating Index (AHEI) score37 (categorized into tertiles). In addition, of the 21,442 randomized participants, 6,867 (32.0%) participants provided baseline pre-randomization biospecimens with measurements of urinary 5-(3′,4′-dihydroxyphenyl)-[gamma]-valerolactone metabolites (gVLM, in μM) as a biomarker of baseline flavanol intake.38

Statistical Analyses

Our primary analyses were completed separately for CE and MVM based on the intention-to-treat principle for time to first event data.39 Descriptive statistics were used to compare the distribution of baseline characteristics across the intervention groups, with mean (standard deviation) for continuous variables and number and proportion for categorical variables. Cox proportional hazards models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) using an indicator variable for either CE or MVM and stratifying the baseline hazard functions by age, sex, and the other intervention assignment.

Models were constructed for each clinical outcome, where person-time for each outcome was counted as time from randomization to the first self-reported diagnosis of the designated outcome. Follow-up was censored at date of last contact, death, or end of the trial on December 31, 2020, whichever came first. Per-protocol analyses censored follow-up when a participant reported discontinuing trial pills, beginning outside non-study use of a CE or MVM supplement, and/or taking <75% of trial pills (missing >8 days/month of trial pills). We additionally censored follow-up when a participant did not return their questionnaires. HRs and 95% CIs were estimated using Cox regression models, weighted by the inverse probability weighting of dependent censoring for non-compliance.40,41

Kaplan-Meier cumulative incidence curves, interactions between randomization groups with trial time, and post hoc analyses that excluded the first 1 and 2 years of follow-up assessed whether treatment effects varied over time and the joint effects of both active treatments, paralleling previous analyses in the COSMOS trial.21,30 Formal tests of the proportional hazards assumption for CE and MVM were based on interaction terms between these treatments and study time added to models that included age, sex, and the other intervention assignment. Subgroup analyses for CE and MVM examined effect modification by baseline characteristics including age, sex, race, body mass index, smoking status, menopausal hormone use, statin use, concurrent COSMOS intervention assignment, MVM use before run-in (for MVM analyses), chocolate consumption (for CE analyses), AHEI score, gVLM concentration (for CE analyses), history of VTE (or DVT or PE for their respective analyses). Effect modification was tested using interaction terms between either CE or MVM and each variable.

Statistical significance (P ≤ 0.05) was assessed with two-sided P values. We did not adjust P values or CIs for multiple testing. Consequently, all results except for those from the primary analysis should be interpreted cautiously and considered hypothesis generating. Data on age, sex, race, and history of DVT, PE, cancer, and fracture were complete. Other baseline characteristics had missing data for ≤2% of COSMOS participants, except for chocolate consumption and AHEI score, which were missing for 8.0 and 11.5% of participants, respectively. All statistical analyses were performed with SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

Baseline characteristics for the 21,442 randomized COSMOS participants are summarized in Table 1. Overall, self-reported sociodemographic, clinical, and lifestyle factors were equally distributed between the active and placebo groups for the CE and MVM interventions at baseline. The mean age (±standard deviation; SD) was 72.1 years (±6.6). Participants were predominantly female (59.1%), non-Hispanic white (88.1%), and had post-college degrees (48.3%). The prevalence of CE and MVM use before run-in were 0.4 and 41.2%, respectively. In terms of clinical and lifestyle risk factors of VTE, 3.6% of participants reported a history of VTE prior to randomization (including 2.9 and 1.5% of DVT and PE, respectively); 27.2% had obesity; 4.0% were current smokers; and 8.3% of women currently used menopausal hormone therapy. While there was a high prevalence of aspirin use (48.9%) and statin use (42.1%), only 4.3% reported current anticoagulant use, likely reflecting the low proportion of participants with pre-randomization VTE or CVD. The median (interquartile range; IQR) baseline gVLM concentration was 3.29 (0.7–11.0).

Table 1.

Baseline characteristics of 21,442 COSMOS participants by randomized cocoa extract and multivitamin assignmentsa

Cocoa extract Multivitamin
Total
(N = 21,442)		 Active
(n = 10,719)		 Placebo
(n = 10,723)		 P 2 Active
(n = 10,720)		 Placebo
(n = 10,722)		 P 2
n (%) n (%) n (%) n (%) n (%)
Age (y), mean ± SD 72.1 ± 6.6 72.1 ± 6.6 72.1 ± 6.6 0.90 72.1 ± 6.6 72.1 ± 6.6 0.86
Age category (y) 0.99 0.99
 60–69 9,224 (43.0) 4,610 (43.0) 4,614 (43.0) 4,611 (43.0) 4,613 (43.0)
 70–79 9,525 (44.4) 4,762 (44.4) 4,763 (44.4) 4,761 (44.4) 4,764 (44.4)
 ≥80 2,693 (12.6) 1,347 (12.6) 1,346 (12.6) 1,348 (12.6) 1,345 (12.5)
Sex 0.89 0.88
 Male 8,774 (40.9) 4,382 (40.9) 4,394 (41.0) 4,382 (40.9) 4,394 (41.0)
 Female 12,666 (59.1) 6,337 (59.1) 6,329 (59.0) 6,338 (59.1) 6,328 (59.0)
Race/ethnicity 0.59 0.24
 Non-Hispanic white 18,887 (88.1) 9,429 (88.0) 9,458 (88.2) 9,415 (87.8) 9,472 (88.3)
 Other 2,555 (11.9) 1,290 (12.0) 1,265 (11.8) 1,305 (12.2) 1,250 (11.7)
Education level 0.80 0.33
 High school diploma/GED or less 2,296 (10.8) 1,141 (10.8) 1,155 (10.9) 1,180 (11.1) 1,116 (10.5)
 Attended or graduated college 8,685 (40.9) 4,328 (40.8) 4,357 (41.1) 4,315 (40.7) 4,370 (41.1)
 Post-college 10,241 (48.3) 5,147 (48.5) 5,094 (48.0) 5,104 (48.2) 5,137 (48.4)
Body mass index (kg/m2), mean ± SD 27.7 ± 5.4 27.6 ± 5.4 27.8 ± 5.5 0.08 27.6 ± 5.4 27.7 ± 5.5 0.19
Body mass index category (kg/m2) 0.26 0.16
 <25 7,070 (33.6) 3,584 (34.1) 3,486 (33.2) 3,533 (33.7) 3,537 (33.6)
 25–29.9 8,230 (39.2) 4,062 (38.7) 4,168 (39.7) 4,166 (39.7) 4,064 (38.6)
 ≥30 5,718 (27.2) 2,859 (27.2) 2,859 (27.2) 2,800 (26.7) 2,918 (27.7)
Smoking status 0.31 0.81
 Never 11,565 (54.7) 5,766 (54.6) 5,799 (54.9) 5,808 (55.0) 5,757 (54.5)
 Past 8,731 (41.3) 4,396 (41.6) 4,335 (41.0) 4,345 (41.1) 4,386 (41.5)
 Current 835 (4.0) 398 (3.8) 437 (4.1) 417 (4.0) 418 (4.0)
Menopausal hormone use 0.63 0.60
 Never 5,199 (42.0) 2,588 (41.7) 2,611 (42.3) 2,631 (42.4) 2,568 (41.6)
 Past 6,160 (49.8) 3,110 (50.2) 3,050 (49.4) 3,069 (49.5) 3,091 (50.0)
 Current 1,021 (8.3) 502 (8.1) 519 (8.4) 503 (8.1) 518 (8.4)
Aspirin use 10,379 (48.9) 5,211 (49.1) 5,168 (48.7) 0.58 5,168 (48.7) 5,211 (49.1) 0.57
Statin use 8,911 (42.1) 4,480 (42.3) 4,431 (41.9) 0.56 4,464 (42.2) 4,447 (41.9) 0.66
Anticoagulant use 910 (4.3) 473 (4.5) 437 (4.1) 0.24 451 (4.3) 459 (4.3) 0.80
Cocoa extract use before run-in 91 (0.4) 45 (0.4) 46 (0.4) 0.92 42 (0.4) 49 (0.5) 0.46
Multivitamin use before run-in 8,795 (41.2) 4,438 (41.6) 4,357 (40.8) 0.26 4,413 (41.3) 4,382 (41.0) 0.65
Chocolate consumption 0.29 0.58
 Monthly or less 6,275 (31.8) 3,095 (31.5) 3,180 (32.2) 3,167 (32.0) 3,108 (31.6)
 Weekly or daily 13,446 (68.2) 6,740 (68.5) 6,706 (67.8) 6,729 (68.0) 6,717 (68.4)
AHEI score, mean ± SD 42.4 ± 11.0 42.5 ± 10.9 42.3 ± 11.1 0.18 42.5 ± 11.0 42.3 ± 10.9 0.25
AHEI category 0.44 0.20
 Low tertile 6,144 (32.4) 3,031 (32.0) 3,113 (32.8) 3,060 (32.2) 3,084 (32.6)
 Mid tertile 6,414 (33.8) 3,237 (34.2) 3,177 (33.4) 3,176 (33.4) 3,238 (34.2)
 High tertile 6,421 (33.8) 3,209 (33.9) 3,212 (33.8) 3,275 (34.4) 3,146 (33.2)
History of venous thromboembolism 773 (3.6) 371 (3.5) 402 (3.7) 0.26 371 (3.5) 402 (3.7) 0.26
Deep vein thrombosis 618 (2.9) 294 (2.7) 324 (3.0) 0.22 298 (2.8) 320 (3.0) 0.37
Pulmonary embolism 312 (1.5) 147 (1.4) 165 (1.5) 0.31 147 (1.4) 165 (1.5) 0.31
History of hypertension 12,423 (58.1) 6,190 (57.9) 6,233 (58.3) 0.57 6,153 (57.6) 6,270 (58.6) 0.11
History of cardiovascular diseasec 1,269 (5.9) 626 (5.8) 643 (6.0) 0.63 643 (6.0) 626 (5.8) 0.62
History of fracture within past year 4,246 (19.8) 2,153 (20.1) 2,093 (19.5) 0.30 2,071 (19.3) 2,175 (20.3) 0.08
History of cancerd within past 2 years 3,550 (16.6) 1,775 (16.6) 1,775 (16.6) 0.99 1,813 (16.9) 1,737 (16.2) 0.16
gVLMe (μM), median [IQR] 3.29 [0.72, 11.00] 3.20 [0.72, 10.94] 3.40 [0.73–11.08] 0.48 3.37 [0.77–11.42] 3.21 [0.68–10.69] 0.27
Open in a new tab

Abbreviations: AHEI, Alternative Healthy Eating Index, GED, general educational development; gVLM, 5-(3′,4′-dihydroxyphenyl)-[gamma]-valerolactone metabolites; IQR, interquartile range; SD, standard deviation.

a

Percentages may not add to 100 because of rounding. Data on age, sex, race, history of deep vein thrombosis, pulmonary embolism, fracture, and cancer had no missing data. Data on other characteristics, except for chocolate consumption and AHEI score, were missing for ≤2.0% of participants. Chocolate consumption was missing for 8.0% of participants (n = 1,721); AHEI score was missing for 11.5% of participants (n = 2,463). Multivitamin refers to multivitamin-multimineral which was tested in COSMOS.

b

P values were calculated using global ANOVA for continuous variables and chi-square tests for categorical variables.

c

Defined as history at baseline of coronary artery bypass graft/percutaneous coronary intervention, unstable angina, carotid artery surgery/stenting, or peripheral artery surgery/stenting.

d

Excluding non-melanoma skin cancer.

e

The 5-(3′,4′-dihydroxyphenyl)-[gamma]-valerolactone (gVLM), including the 3′/4′-sulphated and 3′-O-glucuronidated metabolites, is a biomarker of flavanol intake. gVLM is measured in the longitudinal biospecimen subcohort. There were 6,508 participants (3,252 in the cocoa extract group versus 3,256 in the placebo group; 3,201 in the multivitamin group versus 3,307 in the placebo group) with baseline gVLM measurements.

Cocoa Extract Supplementation and Venous Thromboembolism

During a median treatment and follow-up period of 3.5 years, there were 379 self-reported incident VTE events (4.95 events per 1,000 person-years). The total count included 277 DVT and 165 PE events, wherein only the first event counted toward VTE, although some participants experienced both DVT and PE. The VTE event rates were 0.83 and 0.94% in the CE and placebo groups, respectively. Participants assigned CE had no significant benefit on the primary outcome of self-reported incident VTE events (HR: 0.88; 95% CI: 0.72, 1.08; Fig. 1). A similar lack of effect was observed for the secondary outcomes of DVT and PE.

Fig. 1.

Fig. 1

Open in a new tab

Hazard ratios (HR)1 and 95% confidence intervals (CI) of venous thromboembolism and its component parts by randomized cocoa extract and multivitamin assignments, in intention-to-treat analyses.1 Summary statistics were from Cox regression models that stratified baseline hazard functions by age, sex, and other intervention assignment. All P values were >0.05. Analyses were not adjusted for multiple comparisons. Multivitamin refers to multivitamin-multimineral which was tested in COSMOS.

Next, we examined the effect of CE supplementation on VTE based on compliance to the intervention (taking ≥75% of study pills) and completion of follow-up questionnaires, which were 78.1 and 79.3% at 36 months, respectively.21 Per-protocol analyses censoring at the first report of non-compliance (Fig. 2) did not greatly alter HRs for VTE (279 events; HR: 0.90; 95% CI: 0.71, 1.14) compared with ITT analyses (HR: 0.88; 95% CI: 0.72, 1.08), although VTE event rates were lower in both active and placebo CE arms among these compliant subjects (0.61 and 0.69% in active and placebo arms, respectively). A similar lack of effect was observed for DVT and PE.

Fig. 2.

Fig. 2

Open in a new tab

Hazard ratios (HR)1 and 95% confidence intervals (CI) of venous thromboembolism and its component parts by randomized cocoa extract and multivitamin assignments, in per-protocol analyses.1 Summary statistics were from Cox regression models that stratified baseline hazard functions by age, sex, and other intervention assignment. All P values were >0.05. Analyses were not adjusted for multiple comparisons. Multivitamin refers to multivitamin-multimineral which was tested in COSMOS.2 A participant’s follow-up was censored at the first time the participant reported taking <75% of trial pills in a month, took non-study cocoa extract or multivitamin pills, or did not respond to a semi-annual questionnaire.

The cumulative incidence curves for the effects of CE versus placebo (Fig. 3A and Supplementary Fig. S2, available in the online version only) did not differ between treatment groups on VTE (log-rank p = 0.24), DVT (log-rank p = 0.51), and PE (log-rank p = 0.51), but suggested a possible violation of proportional hazards. Specifically, hazards in the CE group appeared to start higher before crossing below those of the placebo group after year 2 of follow-up. Although cumulative incidence curves for DVT appeared similar to VTE and PE, the distinction between treatment groups diminished after year 4, which could be due to a lack of statistical power with the decrease in sample size. Consistent with these observations, the proportional hazards assumption was satisfied for CE analyses of DVT (P > 0.05), but not for VTE (p = 0.01) or PE (p = 0.03).

Fig. 3.

Fig. 3

Open in a new tab

Cumulative incidence rates1 of venous thromboembolism (VTE) comparing (A) cocoa extract versus placebo groups and (B) multivitamin versus placebo groups, according to year of follow-up.1 Summary statistics were from Cox regression models that stratified baseline hazard functions by age, sex, and other intervention assignment (intention-to-treat analyses). P value was for the effect of the intervention assignment, based on a stratified score (log-rank) test. Multivitamin refers to multivitamin-multimineral which was tested in COSMOS. CI, confidence interval; HR, hazard ratio.

To explore the implications of the proportional hazards violations, we conducted exploratory latency analyses for CE that excluded early follow-up periods (Fig. 4). For DVT, we observed no evidence of latency effects after excluding the first year (HR: 0.78; 95% CI: 0.59, 1.03) or the first 2 years of follow-up (HR: 0.79; 95% CI: 0.56, 1.10). For VTE, excluding the first year yielded a significant reduction (HR: 0.72; 95% CI: 0.57, 0.92), with a similar estimate after excluding the first 2 years (HR: 0.72; 95% CI: 0.53, 0.97). For PE, a significant reduction was observed after year 1 (HR: 0.67; 95% CI: 0.47, 0.96), whereas the estimate after year 2 was similar in magnitude but no longer statistically significant (HR: 0.64; 95% CI: 0.40, 1.01), likely reflecting reduced statistical power due to the smaller number of events.

Fig. 4.

Fig. 4

Open in a new tab

Hazard ratios (HRs)1 and 95% confidence intervals (CI) of venous thromboembolism and its component parts by randomized cocoa extract and multivitamin assignments, in latency analyses.1 Summary statistics were from Cox regression models that stratified baseline hazard functions by age, sex, and other intervention assignment. All P values were >0.05, except for those denoted with an asterisk (*). Analyses were not adjusted for multiple comparisons. Multivitamin refers to multivitamin-multimineral which was tested in COSMOS.

Among 12 sets of subgroups considered, the effect of CE on VTE was apparently modified by baseline self-reported VTE status (P-interaction = 0.005, Supplementary Fig. S3, available in the online version only). Participants without history of VTE prior to randomization taking CE versus placebo had a significant reduction in first VTE (HR: 0.77; 95% CI: 0.61, 0.97) whereas among the 773 randomized participants with a history of VTE at baseline, those assigned CE had a significantly increased risk of recurrent VTE (HR: 1.66; 95% CI: 1.07, 2.58). Similar patterns were observed for DVT (Supplementary Fig. S4, available in the online version only) and PE (Supplementary Fig. S5, available in the online version only) but the interaction was significant only for the latter endpoint (P-interaction = 0.005).

We also found possible modification of the effect of CE on VTE in analyses of the 6,508 randomized subjects with available gVLM concentrations (P-interaction = 0.03). Although stratum-specific HRs for gVLM concentration above versus below its median level did not reach nominal statistical significance, they suggested that participants assigned CE with baseline gVLM levels below the median (indicative of lower flavanol intake) had a potential reduction in VTE (HR: 0.67; 95% CI: 0.41, 1.10), while those with levels above the median may have had an increase in risk (HR: 1.56; 95% CI: 0.89, 2.73). We found no other evidence of effect modification by other covariates for CE on primary or secondary endpoints (all P-interaction >0.05). Further, exploratory analyses of the 2 × 2 factorial design showed no significant interaction between CE and MVM (Supplementary Table S1, available in the online version only).

Multivitamin-multimineral Supplementation and Venous Thromboembolism

The VTE event rates were 0.83 and 0.93% in the MVM and placebo groups, respectively (Fig. 2). MVM assignment did not significantly affect the incidence of VTE (HR: 0.89; 95% CI: 0.73, 1.09) or its component parts. Based on rates of compliance with study pills and follow-up questionnaires of 71.0 and 68.6% at 36 months among those taking MVM and placebo, respectively,21 per-protocol analyses (Fig. 3) indicated borderline nonsignificant reductions in VTE (HR: 0.78; 95% CI: 0.61, 1.00) and DVT (HR: 0.75; 95% CI: 0.56, 1.01) among participants compliant with the intervention.

Log-rank tests comparing cumulative incidence curves for MVM versus placebo (Fig. 4B and Supplementary Fig. S6, available in the online version only) supported null overall findings for VTE (log-rank p = 0.27), which extended to both DVT (log-rank p = 0.50) and PE (log-rank p = 0.81). Differing from CE analyses, latency analyses excluding the first 1 and 2 years of follow-up did not support any significant reductions for either primary or secondary endpoints (Fig. 4). The proportional hazards assumption was satisfied for all MVM analyses (all P > 0.05).

Among 11 sets of subgroup analyses for VTE (Supplementary Fig. S7, available in the online version only), we found evidence of effect modification by smoking status (P-interaction = 0.03). Specifically, there was significant reduction in VTE among never smokers (HR: 0.70; 95% CI: 0.53, 0.94), compared with a nonsignificant increase in former smokers (HR: 1.20; 95% CI: 0.89, 1.62) and nonsignificant reduction in current smokers (HR: 0.62; 95% CI: 0.22, 1.75). In subgroup analyses for secondary endpoints (Supplementary Figs. S8 and S9, available in the online version only), the effect of MVM on DVT was possibly modified by baseline DVT status (P-interaction = 0.04), with nominally significant benefits among participants with DVT prior to randomization (HR: 0.59; 95% CI: 0.35, 0.99), but not among those without (HR: 1.06; 95% CI: 0.81, 1.39). We found no other evidence of effect modification for MVM on primary or secondary endpoints by other factors considered (all P-interaction > 0.05).

Discussion

In this ancillary study analysis of the COSMOS trial among older US adults, the incidence of self-reported VTE was 4.95 events per 1,000 person-years. Our primary analyses showed that neither CE nor MVM supplementation significantly reduced self-reported incident VTE or its component manifestations over a median follow-up of 3.5 years. However, in exploratory analyses, when we excluded the first year of follow-up, CE supplementation significantly reduced the risk of VTE by 28% and of PE by 33%, suggesting potential delayed benefits. For MVM, per-protocol analyses suggested a potential 21% reduction in VTE risk among participants compliant to the intervention. Caution is needed in interpretation of analyses restricted to compliers, especially in light of the low event counts among compliant subjects in both active and placebo groups. We also found several potentially informative effects by baseline VTE status and gVLM concentration for CE supplementation, and baseline VTE and smoking status for MVM supplementation.

Prospective studies examining cocoa products based on chocolate intake,42–46 along with broader-based analyses of habitual dietary flavonoid intake,47–51 which are limited to CVD endpoints, have been inconsistent. However, some meta-analyses suggest an inverse association with CVD risk.46,51 Potential benefits ofcocoa on vascular risk factors are also supported by comparatively smaller and shorter dietary intervention trials with up to 1 year of treatment.15–20 To our knowledge, the present ancillary analysis of COSMOS is the first study to examine the long-term effects of a flavanol-rich CE supplement on clinical venous thromboembolic events.21 Although CE supplementation did not significantly reduce VTE events after a median 3.5 years of follow-up in intention-to-treat analyses, it is biologically plausible that any benefits may requirelonger follow-up to manifest into reductions in clinical event outcomes.51 In this analysis, latency analyses and cumulative incidence curves suggested a possible delayed benefit beginning after 1 and 2 years of follow-up, respectively. These findings paralleled those of the main COSMOS trial,21 which suggested cumulative benefits on total CVD and CVD death beginning after year 1. Thus, extended morbidity and mortality follow-up in COSMOS is essential to fully evaluate the long-term effects of CE supplementation on both thromboembolic and cardiovascular events.

We also found that the effect of CE on VTE may be modified by baseline VTE status. Specifically, CE supplementation significantly reduced incident VTE by 23% among participants with no self-reported history of VTE prior to randomization, indicating that CE may prevent an initial VTE event. On the other hand, CE was also associated with a 66% increase in the risk of a recurrent VTE event, which may have been related to the 3-fold increase in PE risk. Although this finding may be due to chance, as we did not adjust for multiple testing, it is important to improve our mechanistic understanding of CE and flavanol intake in the context of VTE. To our knowledge, no other studies have reported a similar effect modification in VTE risk from flavanol intake.

To our knowledge, this is also the first study to specifically examine a broad-based MVM supplement for the prevention of VTE. The lack of effect for MVM supplementation on VTE in our intent-to-treat analyses is consistent with predominantly null findings in previous cohort studies and randomized trials on clinical cardiovascular events,21,27,31,52–54including COSMOS and PHS II.21,27,31,52–54 Our finding of a potential reduction in VTE from MVM supplementation in per-protocol analyses highlights the need to evaluate the MVM intervention on VTE in PHS II.

We are otherwise limited to observational data and randomized clinical trials testing individual (or a limited combination of) vitamins and minerals contained in the MVM tested in COSMOS. Hyperhomocysteinemia55,56 and lower serum vitamin D status57,58 have been associated with an increased risk of VTE; however, randomized trials testing vitamin B6, B12, and folate supplementation have not supported reductions in VTE59,60. Conversely, daily supplementation of vitamin D (400 IU/day) plus calcium (1,000 mg/day) was found to significantly reduce unprovoked VTE but not overall VTE among postmenopausal women over 7 years in the Women’s Health Initiative.29 Our analyses of MVM supplementation (including 1,000 IU/day of vitamin D and 220 mg/day of calcium) were consistent with the lack of effect on VTE; however, we did not ask participants to differentiate between provoked and unprovoked VTE for these self-reported endpoints in COSMOS.

In the Women’s Health Study,28 randomized vitamin E supplementation (600 IU/day every other day) was significantly associated with a 21 to 27% reduction in overall VTE and unprovoked VTE events over 10 years, respectively. In subgroup analyses, investigators also found a 40 to 49% reduction in VTE among women with a high risk of VTE as defined by a prior history of VTE or the presence of prothrombotic mutations, suggesting a potential mechanism of vitamin K antagonism.28 In the present analyses, MVM supplementation (including 50 IU/day of vitamin E) among participants with a baseline history significantly reduced the risk of recurrent DVT by 41% and was associated with a nonsignificant 34% reduction in recurrent VTE; however, COSMOS did not evaluate the presence of prothrombotic conditions. Further research is needed to determine whether vitamin E in the MVM formulation may contribute to reduction in DVT or VTE, and what the optimal dose would be to drive these potential reductions.

Several potential limitations warrant consideration. First, COSMOS tested a CE supplement (containing all naturally occurring bioactive components of the cocoa bean, including 500 mg/d cocoa flavanols, 80 mg/d (–)-epicatechin, and theobromine) and a broad-based MVM supplement against a true placebo; thus, we cannot disentangle whether specific components of each intervention explained our observed effects.20 Second, generalizability may be limited, with modest racial and ethnic diversity and volunteer bias for those willing and eligible to enroll in a clinical trial.

Third, ancillary outcomes in COSMOS were based upon self-reports without medical record adjudication; thus, some misclassification of VTE events is possible due to potential reporting errors, including age-related recall limitations. Although the incidence of VTE in COSMOS was higher than estimates previously reported in the general population-based studies (approximately 1 to 2 per 1000 person-years),61,62 this was likely because COSMOS focused on an older adult population, and our definition of incident VTE included participants both with and without a history of VTE prior to randomization. When we excluded participants with pre-randomization VTE, our incidence rate decreased to 4.0 events per 1,000 person-years. Nevertheless, any such misclassification would be expected to be nondifferential with respect to treatment assignment, biasing effect estimates toward the null.

Fourth, because VTE was self-reported without verification through medical records or additional follow-up questions, we were unable to use anticoagulant therapy as a proxy to further assess VTE event validity, as its use was only self-reported as a binary variable without information on specific therapies, indication, timing, or duration. However, self-reported anticoagulant use was infrequent and evenly distributed across treatment groups and thus unlikely to have meaningfully influenced treatment comparisons. This also precluded us from the ability to further subtype events between provoked versus unprovoked VTE or to account for underlying thrombophilia as a contributing risk factor. Although COSMOS collected limited information on surgeries and traumas, these data were not systematically linked to the timing or clinical relevance of VTE reports and therefore could not be used to classify provoked events. Collecting medical records and more detailed event information would be optimal in future analyses.

Fifth, surveillance and selection bias may have affected our risk estimates due to loss to follow-up and declining compliance over time, particularly for per-protocol and latency analyses in which censoring may be informative. Finally, all results except for those of the primary analysis warrant cautious interpretation and should be considered hypothesis-generating, as we did not account for multiple testing. At the nominal 0.05 level, we would expect 1 secondary outcome and <1 interaction to be significant by chance alone.

In conclusion, randomized CE and MVM supplementation did not significantly reduce self-reported incident VTE or its component parts in older adults after a median 3.5 years of treatment and follow-up. However, there were potential reductions in VTE from CE and MVM supported in latency analyses and per-protocol analyses, respectively. Future studies with longer-term follow-up that identify clinically relevant mechanisms of effect are warranted to elucidate the effect of CE and MVM supplementation on clinical venous thromboembolic events.

Supplementary Material

Supplementary Material

Supplementary Material is available at https://doi.org/10.1055/a-2806-3554.

What is known about this topic?

  • There is evidence that cocoa flavanols and multivitamin-multimineral supplements may improve vascular risk factors.

  • In a large-scale randomized, double-blind, placebo-controlled trial, cocoa extract supplementation significantly reduced cardiovascular death.

What does this paper add?

  • Neither cocoa extract supplementation nor multivitamin-multimineral supplementation significantly reduced self-reported venous thromboembolism in this trial cohort of older US adults in intention-to-treat analyses.

  • Cocoa extract supplementation may have had delayed benefits on self-reported venous thromboembolism.

Acknowledgment

We are deeply indebted to the 21,442 COSMOS participants for their steadfast and conscientious collaboration and to the COSMOS Research Group for their commitment and perseverance to the trial. We specifically acknowledge the COSMOS Research Group for their scientific (Brigham and Women’s Hospital [BWH], Fred Hutchinson Cancer Research Center [FHCRC], Women’s Health Initiative [WHI], Data Safety and Monitoring Board [DSMB], Mars Edge) and logistical (BWH, FHCRC, DSMB, Mars Edge, Contract Pharmacal Corp, Pfizer Consumer Healthcare [now Haleon]) contributions. For a list of all the investigators who have contributed to WHI science, please visit: https://s3-us-west-2.amazonaws.com/www-whi-org/wp-content/uploads/WHI-Investigator-Long-List.pdf

Funding Information

This publication was supported in part by the National Center for Complementary and Integrative Health of the National Institutes of Health (NIH), Bethesda, MD with training grant 5T32AT000051 (SJP). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The COcoa Supplement and Multivitamin Outcomes Study (COSMOS) is supported by an investigator-initiated grant from Mars Edge (JEM, HDS), a segment of Mars dedicated to nutrition research and products, which included infrastructure support and the donation of study pills and packaging. Pfizer Consumer Healthcare (now Haleon) provided support through the partial provision of study pills and packaging (JEM, HDS). COSMOS is also supported in part by NIH grants AG050657, AG071611, EY025623, and HL157665. The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, US Department of Health and Human Services through 75N92021D00001, 75N92021D00002, 75N92021D00003, 75N92021D00004, 75N92021D00005.

Conflict of Interest

H.D.S. and J.E.M. reported receiving investigator-initiated grants from Mars Edge and Pfizer Consumer Healthcare (now Haleon) during the conduct of the COSMOS trial. The other authors report no conflicts of interest.

Data Availability Statement

The dataset(s) will be de-identified prior to release for sharing. We will make the data and associated documentation available to users only under a data-sharing agreement. Details on the availability of the study data to other investigators will be on our study Web site at https://cosmostrial.org/.

References

  • 1.Raskob GE, Angchaisuksiri P, Blanco AN, et al. ; ISTH Steering Committee for World Thrombosis Day. Thrombosis: a major contributor to global disease burden. Semin Thromb Hemost 2014;40(07):724–735 [DOI] [PubMed] [Google Scholar]
  • 2.Anderson FA Jr, Spencer FA. Risk factors for venous thromboembolism. Circulation 2003;107(23, Suppl 1):I9–I16 [DOI] [PubMed] [Google Scholar]
  • 3.Gregson J, Kaptoge S, Bolton T, et al. ; Emerging Risk Factors Collaboration. Cardiovascular risk factors associated with venous thromboembolism. JAMA Cardiol 2019;4(02):163–173 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Mahmoodi BK, Cushman M, Anne Næss I, et al. Association of traditional cardiovascular risk factors with venous thromboembolism. Circulation 2017; 135(01):7–16 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Glynn RJ, Rosner B. Comparison of risk factors for the competing risks of coronary heart disease, stroke, and venous thromboembolism. Am J Epidemiol 2005;162 (10):975–982 [DOI] [PubMed] [Google Scholar]
  • 6.Prandoni P, Bilora F, Marchiori A, et al. An association between atherosclerosis and venous thrombosis. N Engl J Med 2003;348(15):1435–1441 [DOI] [PubMed] [Google Scholar]
  • 7.Tsai AW, Cushman M, Rosamond WD, Heckbert SR, Polak JF, Folsom AR. Cardiovascular risk factors and venous thromboembolism incidence: the longitudinal investigation of thromboembolism etiology. Arch Intern Med 2002;162 (10):1182–1189 [DOI] [PubMed] [Google Scholar]
  • 8.Everett BM, Glynn RJ, Buring JE, Ridker PM. Lipid biomarkers, hormone therapy and the risk of venous thromboembolism in women. J Thromb Haemost 2009;7 (04):588–596 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Erdman JW Jr, Carson L, Kwik-Uribe C, Evans EM, Allen RR. Effects of cocoa flavanols on risk factors for cardiovascular disease. Asia Pac J Clin Nutr 2008;17(1, Suppl 1):284–287 [PubMed] [Google Scholar]
  • 10.Manach C, Mazur A, Scalbert A. Polyphenols and prevention of cardiovascular diseases. Curr Opin Lipidol 2005;16(01):77–84 [DOI] [PubMed] [Google Scholar]
  • 11.Kim Y, Je Y. Flavonoid intake and mortality from cardiovascular disease and all causes: a meta-analysis of prospective cohort studies. Clin Nutr ESPEN 2017;20:68–77 [DOI] [PubMed] [Google Scholar]
  • 12.Quintal Martínez JP, Segura Campos MR. Flavonoids as a therapeutical option for the treatment of thrombotic complications associated with COVID-19. Phytother Res 2023;37(03):1092–1114 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Vazhappilly CG, Ansari SA, Al-Jaleeli R, et al. Role of flavonoids in thrombotic, cardiovascular, and inflammatory diseases. Inflammopharmacology 2019;27 (05):863–869 [DOI] [PubMed] [Google Scholar]
  • 14.Schroeter H, Heiss C, Balzer J, et al. (−)-Epicatechin mediates beneficial effects of flavanol-rich cocoa on vascular function in humans. Proc Natl Acad Sci U S A 2006;103(04):1024–1029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Sansone R, Rodriguez-Mateos A, Heuel J, et al. ; Flaviola Consortium, European Union 7th Framework Program. Cocoa flavanol intake improves endothelial function and Framingham Risk Score in healthy men and women: a randomised, controlled, double-masked trial: the Flaviola Health Study. Br J Nutr 2015;114 (08):1246–1255 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Sun Y, Zimmermann D, De Castro CA, Actis-Goretta L. Dose-response relationship between cocoa flavanols and human endothelial function: a systematic review and meta-analysis of randomized trials. Food Funct 2019;10(10): 6322–6330 [DOI] [PubMed] [Google Scholar]
  • 17.Rein D, Paglieroni TG, Wun T, et al. Cocoa inhibits platelet activation and function. Am J Clin Nutr 2000;72(01):30–35 [DOI] [PubMed] [Google Scholar]
  • 18.Rein D, Paglieroni TG, Pearson DA, et al. Cocoa and wine polyphenols modulate platelet activation and function. J Nutr 2000;130(8S Suppl):2120S–2126S [DOI] [PubMed] [Google Scholar]
  • 19.Grassi D, Necozione S, Lippi C, et al. Cocoa reduces blood pressure and insulin resistance and improves endothelium-dependent vasodilation in hypertensives. Hypertension 2005;46(02):398–405 [DOI] [PubMed] [Google Scholar]
  • 20.Ried K, Fakler P, Stocks NP. Effect of cocoa on blood pressure. Cochrane Database Syst Rev 2017;4(04):CD008893. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Sesso HD, Manson JE, Aragaki AK, et al. ; COSMOS Research Group. Effect of cocoa flavanol supplementation for the prevention of cardiovascular disease events: the COcoa Supplement and Multivitamin Outcomes Study (COSMOS) randomized clinical trial. Am J Clin Nutr 2022;115(06):1490–1500 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Bailey RL, Gahche JJ, Lentino CV, et al. Dietary supplement use in the United States, 2003–2006. J Nutr 2011;141(02):261–266 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Rautiainen S, Åkesson A, Levitan EB, Morgenstern R, Mittleman MA, Wolk A. Multivitamin use and the risk of myocardial infarction: a population-based cohort of Swedish women. Am J Clin Nutr 2010;92(05):1251–1256 [DOI] [PubMed] [Google Scholar]
  • 24.Holmquist C, Larsson S, Wolk A, de Faire U. Multivitamin supplements are inversely associated with risk of myocardial infarction in men and women—Stockholm Heart Epidemiology Program (SHEEP). J Nutr 2003;133(08): 2650–2654 [DOI] [PubMed] [Google Scholar]
  • 25.Bailey RL, Fakhouri TH, Park Y, et al. Multivitamin-mineral use is associated with reduced risk of cardiovascular disease mortality among women in the United States. J Nutr 2015;145(03):572–578 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Rimm EB, Willett WC, Hu FB, et al. Folate and vitamin B6 from diet and supplements in relation to risk of coronary heart disease among women. JAMA 1998;279(05):359–364 [DOI] [PubMed] [Google Scholar]
  • 27.Rautiainen S, Rist PM, Glynn RJ, Buring JE, Gaziano JM, Sesso HD. Multivitamin use and the risk of cardiovascular disease in men. J Nutr 2016;146(06):1235–1240 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Glynn RJ, Ridker PM, Goldhaber SZ, Zee RYL, Buring JE. Effects of random allocation to vitamin E supplementation on the occurrence of venous thromboembolism: report from the Women’s Health Study. Circulation 2007;116 (13):1497–1503 [DOI] [PubMed] [Google Scholar]
  • 29.Blondon M, Rodabough RJ, Budrys N, et al. ; From the Women’s Health Initiative Randomized Controlled Trial. The effect of calcium plus vitamin D supplementation on the risk of venous thromboembolism. Thromb Haemost 2015;113 (05):999–1009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Sesso HD, Rist PM, Aragaki AK, et al. ; COSMOS Research Group. Multivitamins in the prevention of cancer and cardiovascular disease: the COcoa Supplement and Multivitamin Outcomes Study (COSMOS) randomized clinical trial. Am J Clin Nutr 2022;115(06):1501–1510 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Sesso HD, Christen WG, Bubes V, et al. Multivitamins in the prevention of cardiovascular disease in men: the Physicians’ Health Study II randomized controlled trial. JAMA 2012;308(17):1751–1760 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Rist PM, Sesso HD, Johnson LG, et al. ; COSMOS Research Group. Design and baseline characteristics of participants in the COcoa Supplement and Multivitamin Outcomes Study (COSMOS). Contemp Clin Trials 2022;116:106728. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Cauley JA, Crandall C. The Women’s Health Initiative: a landmark resource for skeletal research since 1992. J Bone Miner Res 2020;35(05):845–860 [DOI] [PubMed] [Google Scholar]
  • 34.Bassuk SS, Manson JE, Lee IM, et al. Baseline characteristics of participants in the VITamin D and OmegA-3 TriaL (VITAL). Contemp Clin Trials 2016;47:235–243 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 1985;122 (01):51–65 [DOI] [PubMed] [Google Scholar]
  • 36.Feskanich D, Rimm EB, Giovannucci EL, et al. Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. J Am Diet Assoc 1993;93(07):790–796 [DOI] [PubMed] [Google Scholar]
  • 37.Chiuve SE, Fung TT, Rimm EB, et al. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr 2012;142(06):1009–1018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Ottaviani JI, Borges G, Momma TY, et al. The metabolome of [2–14C](−)-epicatechin in humans: implications for the assessment of efficacy, safety, and mechanisms of action of polyphenolic bioactives. Sci Rep 2016:6:29034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Cox DR. Regression models and life-tables. J R Stat Soc B 1972;34(02):187–202 [Google Scholar]
  • 40.Robins JM, Finkelstein DM. Correcting for noncompliance and dependent censoring in an AIDS Clinical Trial with inverse probability of censoring weighted (IPCW) log-rank tests. Biometrics 2000;56(03):779–788 [DOI] [PubMed] [Google Scholar]
  • 41.Willems S, Schat A, van Noorden MS, Fiocco M. Correcting for dependent censoring in routine outcome monitoring data by applying the inverse probability censoring weighted estimator. Stat Methods Med Res 2018;27 (02):323–335 [DOI] [PubMed] [Google Scholar]
  • 42.Zhong GC, Hu TY, Yang PF, et al. Chocolate consumption and all-cause and causespecific mortality in a US population: a post hoc analysis of the PLCO cancer screening trial. Aging (Albany NY) 2021;13(14):18564–18585 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Kwok CS, Boekholdt SM, Lentjes MAH, et al. Habitual chocolate consumption and risk of cardiovascular disease among healthy men and women. Heart 2015;101 (16):1279–1287 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Greenberg JA, Manson JE, Neuhouser ML, et al. Chocolate intake and heart disease and stroke in the Women’s Health Initiative: a prospective analysis. Am J Clin Nutr 2018;108(01):41–48 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Buijsse B, Weikert C, Drogan D, Bergmann M, Boeing H. Chocolate consumption in relation to blood pressure and risk of cardiovascular disease in German adults. Eur Heart J 2010;31(13):1616–1623 [DOI] [PubMed] [Google Scholar]
  • 46.Morze J, Schwedhelm C, Bencic A, et al. Chocolate and risk of chronic disease: a systematic review and dose-response meta-analysis. Eur J Nutr 2020;59 (01):389–397 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.McCullough ML, Peterson JJ, Patel R, Jacques PF, Shah R, Dwyer JT. Flavonoid intake and cardiovascular disease mortality in a prospective cohort of US adults. Am J Clin Nutr 2012;95(02):454–464 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Mursu J, Voutilainen S, Nurmi T, Tuomainen TP, Kurl S, Salonen JT. Flavonoid intake and the risk of ischaemic stroke and CVD mortality in middle-aged Finnish men: the Kuopio Ischaemic Heart Disease Risk Factor Study. Br J Nutr 2008;100 (04):890–895 [DOI] [PubMed] [Google Scholar]
  • 49.Mursu J, Nurmi T, Tuomainen TP, Ruusunen A, Salonen JT, Voutilainen S. The intake of flavonoids and carotid atherosclerosis: the Kuopio Ischaemic Heart Disease Risk Factor Study. Br J Nutr 2007;98(04):814–818 [DOI] [PubMed] [Google Scholar]
  • 50.Cassidy A, O’Reilly EJ, Kay C, et al. Habitual intake of flavonoid subclasses and incident hypertension in adults. Am J Clin Nutr 2011;93(02):338–347 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.van Dam RM, Naidoo N, Landberg R. Dietary flavonoids and the development of type 2 diabetes and cardiovascular diseases: review of recent findings. Curr Opin Lipidol 2013;24(01):25–33 [DOI] [PubMed] [Google Scholar]
  • 52.Xu X, Shi Z, Liu G, et al. Article the joint effects of diet and dietary supplements in relation to obesity and cardiovascular disease over a 10-year follow-up: a longitudinal study of 69,990 participants in australia. Nutrients 2021;13(03):944. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Rautiainen S, Lee IM, Rist PM, et al. Multivitamin use and cardiovascular disease in a prospective study of women. Am J Clin Nutr 2015;101(01): 144–152 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Neuhouser ML, Wassertheil-Smoller S, Thomson C, et al. Multivitamin use and risk of cancer and cardiovascular disease in the Women’s Health Initiative cohorts. Arch Intern Med 2009;169(03):294–304 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Zhou K, Zhao R, Geng Z, et al. Association between B-group vitamins and venous thrombosis: systematic review and meta-analysis of epidemiological studies. J Thromb Thrombolysis 2012;34(04):459–467 [DOI] [PubMed] [Google Scholar]
  • 56.Ray JG. Meta-analysis of hyperhomocysteinemia as a risk factor for venous thromboembolic disease. Arch Intern Med 1998;158(19):2101–2106 [DOI] [PubMed] [Google Scholar]
  • 57.Hung KC, Yang SH, Chang CY, et al. Is circulating vitamin D status associated with the risk of venous thromboembolism? A meta-analysis of observational studies. Nutrients 2023;15(05):1113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Wan J, Yuan J, Li X, et al. Association between serum vitamin D levels and venous thromboembolism (VTE): a systematic review and meta-analysis of observational studies. Complement Ther Med 2020;54:102579. [DOI] [PubMed] [Google Scholar]
  • 59.Ray JG, Kearon C, Yi Q, Sheridan P, Lonn EHeart Outcomes Prevention Evaluation 2 (HOPE-2) Investigators. Homocysteine-lowering therapy and risk for venous thromboembolism: a randomized trial. Ann Intern Med 2007;146(11):761–767 [DOI] [PubMed] [Google Scholar]
  • 60.den Heijer M, Willems HPJ, Blom HJ, et al. Homocysteine lowering by B vitamins and the secondary prevention of deep vein thrombosis and pulmonary embolism: a randomized, placebo-controlled, double-blind trial. Blood 2007;109(01):139–144 [DOI] [PubMed] [Google Scholar]
  • 61.Wendelboe AM, Raskob GE. Global burden of thrombosis: epidemiologic aspects. Circ Res 2016;118(09):1340–1347 [DOI] [PubMed] [Google Scholar]
  • 62.Cushman M, Tsai AW, White RH, et al. Deep vein thrombosis and pulmonary embolism in two cohorts: the longitudinal investigation of thromboembolism etiology. Am J Med 2004;117(01):19–25 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
NIHMS2157890-supplement-Supplementary_Material.docx (2.4MB, docx)

Data Availability Statement

The dataset(s) will be de-identified prior to release for sharing. We will make the data and associated documentation available to users only under a data-sharing agreement. Details on the availability of the study data to other investigators will be on our study Web site at https://cosmostrial.org/.

RESOURCES