Abstract
The authors investigated diet as a risk factor for the development of venous thromboembolism (VTE) among 129,430 US women and men in the Nurses’ Health Study and Health Professionals Follow-up Study. There were 2,892 cases of VTE from 1984 through 2008. Information on participants’ dietary intakes was collected every 2–4 years using a food frequency questionnaire. Dietary patterns (prudent vs. Western), food intakes (fruit, vegetables, fish, red and processed meats, and alcohol), and nutrient intakes (omega-3 fatty acids, trans fatty acids, total fiber, and vitamins K1, B6, B12, and E) were categorized into quintiles, and the risk of VTE was compared among quintiles with the use of Cox proportional hazard models. After adjusting the results for 17 potential confounders, the authors found that adherence to the Western dietary pattern was associated with an increased risk of VTE in men (for the highest quintile vs. the lowest, relative risk = 1.43, 95% confidence interval: 1.16, 1.78; P for trend < 0.001) but not in women (relative risk = 1.14, 95% confidence interval: 0.91, 1.42; P for trend = 0.09). Favorable associations were found in the pooled analysis for intakes of vitamins E and B6 and fiber. For intakes of red and processed meat and trans fatty acids, no association was found in women, whereas a significant positive association was found in men. These results suggest a weak association between diet and the risk of VTE.
Keywords: diet, food, pulmonary embolism, venous thrombosis
The annual incidence of venous thromboembolism (VTE) in the United States is approximately 1 per 1,000 adults. Among nonfatal VTEs, approximately 60% are cases of deep vein thrombosis (DVT) and 40% are cases of nonfatal pulmonary embolism (PE) (1). Arterial thrombosis and venous thrombosis share multiple risk factors, the strongest of which are age, obesity, cigarette smoking, and hypertension (2–4). However, little is known about the relation between diet and VTE.
A relation between diet and VTE was suggested as early as World War II (5). However, to our knowledge, there have only been 2 prospective epidemiologic studies—the Atherosclerosis Risk in Communities (ARIC) Study and the Iowa Women’s Health Study (IWHS)—in which investigators have systematically examined the role of dietary patterns in the development of VTE (6, 7), and the results of those studies conflicted. There have been 8 studies in which investigators have explored the relation between alcohol and VTE; in 4 of those studies, an inverse association was reported (7–10), whereas no relation was found in the other 4 (11–14). The relation between specific nutrients and coagulation factor levels has been evaluated more broadly (6, 7, 15–17). Vitamins B6 and B12 have been shown to reduce elevated homocysteine levels, which are associated with an increased VTE risk; omega-3 fatty acids decrease platelet aggregation and factor VII and fibrinogen levels (18); and vitamin E might inhibit the biologic activity of vitamin K and vitamin K-dependent clotting factors (19). However, little is known about the potential role of these nutrients in the development of VTE.
The aim of the present study was to better define the association between dietary intake and the risk of VTE by analyzing cohort data from approximately 130,000 US women and men.
MATERIALS AND METHODS
Overview
The Nurses’ Health Study (NHS) began in 1976, when 121,701 female nurses who were 30–55 years old and living in 11 US states (California, Connecticut, Florida, Maryland, Massachusetts, Michigan, New Jersey, New York, Ohio, Pennsylvania, and Texas) responded to a mailed health questionnaire (20). The Health Professionals Follow-up Study (HPFS) began in 1986, when 51,529 male US health professionals who were 40–75 years old responded to a detailed mailed questionnaire that included a diet survey and questions about lifestyle practices and medical history. In both cohorts, follow-up questionnaires were sent every 2 years thereafter to acquire updated information on lifestyle factors and to query participants about newly diagnosed medical conditions. Women in the NHS also completed a 126-item food frequency questionnaire (FFQ) in 1984, and men in the HPFS completed a 131-item FFQ at baseline (i.e., in 1986). Similar FFQs were sent every 2–4 years thereafter. The Partners Institutional Review Board (Partners HealthCare, Boston, Massachusetts) approved the NHS and the HPFS protocols. The studies are being conducted according to the ethical guidelines of Partners HealthCare.
We excluded participants who did not complete an FFQ at baseline from the analysis. Likewise, we excluded participants with extremely high energy intakes (>3,500 kcal/day for women and >4,200 kcal/day for men) or extremely low energy intakes (<500 kcal/day for women and <800 kcal/day for men), as well as those who left more than 70 questions blank. Subjects who reported a history of VTE at baseline also were excluded from the present analysis. The final baseline population (in 1984 for women from the NHS and in 1986 for men from the HPFS) included 80,192 women and 49,238 men. In the NHS, the active follow-up rate (the number of person-years in the cohort when participants are censored after their last questionnaire response) from 1984 to 2008 was 92.2%, and in the HPFS, the follow-up rate from 1986 to 2008 was 88.0%.
Assessment of dietary intake
Dietary intake information was collected using an FFQ designed to assess average food intake over the previous 12 months. Standard portion sizes were listed with each food. For each food item, participants indicated their average frequency of consumption over the past year in terms of the specified serving size by checking 1 of 9 frequency categories that ranged from “almost never” to “≥6 times/day.”
To prepare for factor analysis, food items were grouped into 38 predefined food groups as previously described (21). Using principal component analysis, dietary patterns were identified from FFQs administrated in 1984, 1986, 1990, 1994, 1998, and 2002 in the NHS and 1986, 1990, 1994, 1998, and 2002 in the HPFS. We rotated the factors using an orthogonal transformation to achieve structure with greater interpretability and determined the number of factors to be retained using the diagram of eigenvalues. Foods with a correlation of at least 0.30 were considered to be making a contribution to the factor.
On the basis of prior published reports and biologic plausibility, we selected some specific food groups (fruits, vegetables, fish, red and processed meats, and alcohol) and nutrients (omega-3 fatty acids, trans fatty acids, total fiber, vitamin K1, vitamin B6, vitamin B12, and vitamin E) to assess their association with the risk of VTE. We assessed both dietary and total (diet plus supplements) intakes. Nutrient intakes were computed by multiplying the frequency response by the nutrient content of the specified portion sizes. Food composition values were obtained from the Harvard University Food Composition Database (Boston, Massachusetts), which was derived from US Department of Agriculture sources and supplemented with information from manufacturers. All dietary factors were adjusted for total energy using the residual method (22).
To reduce measurement error and represent long-term dietary intake, the cumulative averages of pattern scores, foods, and nutrients were calculated and then divided into quintiles. We also examined the consistency of the association between dietary patterns and VTE by using the most recent dietary intakes as predictors of VTE risk.
Assessment of VTE
Participants who reported a physician-diagnosed PE on a biennial questionnaire but did not have a prior history of malignancy received a follow-up letter requesting medical records from the facility in which they were diagnosed with a PE. A detailed review of these records was undertaken, and cases were coded. Incident cases for which the medical record included imaging that was diagnostic of a PE were considered confirmed. Imaging was considered diagnostic if the radiologist reading the study noted a ventilation/perfusion lung scan that indicated a high probability of a PE, a filling defect on a contrast-enhanced computed tomographic scan of the pulmonary vasculature, or a filling defect on a catheter-based pulmonary angiograph. Over the follow-up period, we identified 956 women and 353 men with an incident PE. Participants also reported physician-diagnosed DVT on the biennial questionnaire. Men were considered to have incident DVT if they reported a physician-diagnosed DVT between 2 consecutive questionnaires, and women were if they reported “phlebitis and thrombophlebitis” as another major illness (code 451 according to the International Classification of Diseases, Eighth Revision). A total of 584 women and 999 men were classified as having DVT under these definitions. Overall, 1,540 women and 1,352 men reported VTE.
Assessment of other variables
When possible, data on covariates were obtained from the baseline questionnaire (1984 in NHS and 1986 in HPFS) and updated every 2 years. Variables included age, physical activity level, physical inactivity level, body mass index (BMI, measured as weight (kg)/height (m)2), total caloric intake, smoking status, pack-years of smoking, race/ethnicity, spouse’s highest educational attainment (only in the NHS), parity (only in the NHS), menopausal status (only in the NHS), nonaspirin nonsteroidal antiinflammatory drug use, warfarin use, multivitamin use, hypertension, coronary heart disease, and rheumatologic disease.
Physical activity, including a variety of activities such as walking, bicycling, swimming, or playing tennis, was measured in metabolic equivalents per week and categorized in quintiles. Physical inactivity was defined as the number of hours spent sitting at home per day and was categorized in 6 categories (from the lowest amount of time to the highest). BMI was updated biennially and categorized in 4 categories: less than 25.0, 25.0–29.9, 30.0–34.9, and 35 or higher. Total caloric intake was estimated through the FFQ, expressed in kilocalories per day, and used as a continuous variable. Participants were categorized by smoking status as never smokers, ex-smokers, or current smokers. Pack-years of smoking were calculated among ever smokers, a group that included both ex-smokers and current smokers. Participants’ races/ethnicities were categorized in 2 groups (white vs. nonwhite). Among women, spousal educational attainment was categorized into 3 classes (high school, college, or graduate school); parity was categorized as nulliparous or as having 1, 2–3, or 4 or more children; and menopausal status was categorized as premenopausal versus postmenopausal and according to whether estrogen or progesterone replacement (oral or patch) had been or was being used. Other potential confounders (medication use (nonsteroidal antiinflammatory drugs, warfarin), multivitamin use, hypertension, coronary heart disease (defined as any history of angina, coronary artery stenosis, or myocardial infarction), and rheumatologic disease (defined as any history of systemic lupus erythematosus or rheumatoid arthritis)) were included as dichotomous (yes vs. no) variables.
Statistical analysis
Our strategy for analysis was to study the association between dietary intake and the risk of VTE (DVT and PE) in both men and women. Each dietary variable was analyzed using the Cox proportional hazards model. A test for trend across the quintiles of each dietary variable was calculated by treating the categories as ordinal variables in a proportional hazards model. After calculating sex-specific relative risks, we combined the ln relative risks weighted by the inverse of their variances using a random-effects model (23). We tested for between-study heterogeneity using the Q statistic, which informs about the presence or absence of heterogeneity, and we also provided the I2 index to quantify the degree of heterogeneity between studies, expressed as a percentage of total variance (23). Two-sided 95% confidence intervals were calculated.
To assess whether age-related physiologic differences might modify the association between diet and VTE (9), we further stratified our analysis according to the median age at baseline (50 years in women and 53 years in men). We also stratified analyses according to obesity (BMI ≥30) and level of physical activity (median). In addition, we performed sensitivity analyses by excluding participants with prior comorbid conditions (cancer or cardiovascular diseases). All analyses were conducted using SAS, version 9.1 (SAS Institute, Inc., Cary, North Carolina).
RESULTS
Description of the population
Through the use of principal component analysis, we identified 2 distinct major dietary patterns among both women and men. The first pattern was defined by high intakes of fruit, vegetables, fish, poultry, whole-grain products, and low-fat dairy products. The second pattern was defined by high intakes of refined grains, cured and red meats, desserts, sweets, French fries, and high-fat dairy products. In accordance with what was done in previous studies of dietary patterns in this population, we labeled the first pattern the “prudent pattern” and the second pattern the “Western pattern.” Tables 1 and 2 show baseline characteristics of the participants, which were adjusted for age and stratified according to each dietary pattern among women and men. Compared with participants with the lowest intake of foods from the prudent diet (the lowest quintile), participants with the highest intake of foods from the prudent diet (the highest quintile) were older, more physically active, less likely to be current smokers, and more frequent users of multivitamin supplements. Compared with participants with the lowest intake of foods from the Western diet, participants with the highest intake of foods from the Western diet had a higher BMI, were less physically active, were more likely to smoke, and took fewer multivitamin supplements.
Table 1.
Age-Standardized Baseline Characteristics of Women According to Dietary Pattern (n = 80,263), Nurses’ Health Study, 1984
| Characteristic | Intake of Prudent Pattern |
Intake of Western Pattern |
||||||
| Lowest Quintile (n = 16,047) |
Highest Quintile (n = 15,984) |
Lowest Quintile (n = 16,050) |
Highest Quintile (n = 16,031) |
|||||
| % | Mean (SD) | % | Mean (SD) | % | Mean (SD) | % | Mean (SD) | |
| Age, years | 48.4 (7.0) | 51.7 (7.0) | 52.2 (6.8) | 48.4 (7.1) | ||||
| Total physical activity, METs/weeka,b | 10.1 | 19.7 | 17.6 | 11.8 | ||||
| Sedentary lifestylec | ||||||||
| Lowest amount of time spent sitting | 13.1 | 14.7 | 16.9 | 10.5 | ||||
| Highest amount of time spent sitting | 2.1 | 1.9 | 1.6 | 2.5 | ||||
| Body mass indexd | 24.8 | 25.2 | 24.6 | 25.5 | ||||
| Total caloric intake, kcal/daya | 1,447 | 2,086 | 1,228 | 2,380 | ||||
| Smoking status | ||||||||
| Nonsmokers | 40.9 | 45.7 | 42.3 | 46.4 | ||||
| Ex-smokers | 25.4 | 37.8 | 37.4 | 27.8 | ||||
| Current smokers | 33.5 | 16.1 | 20.0 | 25.6 | ||||
| Missing | 0.2 | 0.4 | 0.3 | 0.2 | ||||
| Pack-years among ever smokersa,e | 25 | 18 | 20 | 22 | ||||
| White race/ethnicity | 98 | 97 | 96 | 99 | ||||
| Spouse’s educational attainment | ||||||||
| High school | 39.7 | 30.7 | 30.6 | 39.1 | ||||
| College | 19.5 | 24.4 | 22.7 | 21.6 | ||||
| Graduate school | 14.2 | 22.2 | 21.0 | 16.1 | ||||
| Missing | 26.6 | 22.7 | 25.7 | 23.2 | ||||
| Parity | ||||||||
| Nulliparous | 5.7 | 6.0 | 7.3 | 5.1 | ||||
| 1 | 7.7 | 7.3 | 7.9 | 6.5 | ||||
| 2–3 | 56.8 | 54.7 | 55.3 | 55.8 | ||||
| ≥4 | 28.4 | 30.3 | 27.8 | 31.2 | ||||
| Missing | 1.4 | 1.7 | 1.7 | 1.4 | ||||
| Menopausal status | ||||||||
| Premenopausal | 54.5 | 37.2 | 34.0 | 55.1 | ||||
| Postmenopausal and no HRT use | 20.7 | 26.7 | 28.5 | 20.3 | ||||
| Postmenopausal and past HRT use | 8.2 | 13.1 | 14.0 | 8.3 | ||||
| Postmenopausal and estrogen replacement therapy use | 7.0 | 10.7 | 11.0 | 6.7 | ||||
| Postmenopausal and estrogen-progesterone replacement therapy use | 0.5 | 1.1 | 1.1 | 0.6 | ||||
| Missing | 9.1 | 11.2 | 11.4 | 9.0 | ||||
| Nonaspirin nonsteroidal antiinflammatory drug use | 43.7 | 42.5 | 40.1 | 46.4 | ||||
| Warfarin use | 0.8 | 1.1 | 1.1 | 0.8 | ||||
| Multivitamin supplement use | 30.3 | 43.9 | 42.8 | 32.3 | ||||
| Hypertension | 7.0 | 9.3 | 9.5 | 7.2 | ||||
| Coronary heart diseasef | 1.0 | 1.5 | 1.5 | 1.0 | ||||
| Rheumatologic diseaseg | 1.4 | 2.2 | 1.8 | 1.5 | ||||
| Dietary intakea | ||||||||
| Fruit, servings/day | 0.6 | 2.4 | 1.5 | 1.4 | ||||
| Vegetables, servings/day | 1.5 | 5.8 | 3.2 | 3.4 | ||||
| Fish, servings/day | 0.2 | 0.5 | 0.4 | 0.3 | ||||
| Red and processed meat, servings/day | 1.0 | 0.9 | 0.5 | 1.5 | ||||
| Alcohol, g/day | 7.1 | 6.7 | 7.1 | 6.1 | ||||
| Omega-3 fatty acids, g/dayh | 0.07 | 0.2 | 0.4 | 0.07 | ||||
| Trans fatty acids, g/day | 3.8 | 2.8 | 2.8 | 3.8 | ||||
| Total fiber, g/day | 12.4 | 21.0 | 18.5 | 14.9 | ||||
| Vitamin K1, μg/dayh | 22.1 | 15.3 | 14.6 | 22.7 | ||||
| Vitamin B6, mg/dayh | 7.1 | 13.1 | 15.3 | 6.1 | ||||
| Vitamin B12, μg/dayh | 9.3 | 14.0 | 14.4 | 9.5 | ||||
| Vitamin E, mg/dayh | 67 | 107 | 135 | 53 | ||||
Abbreviations: HRT, hormone replacement therapy; METs, metabolic equivalents; SD, standard deviation.
Age-adjusted means.
Sum of the average time per week spent in each activity × MET value of each activity.
Time spent sitting was assessed in 1988 and in 1990 when subjects were asked to report “the number of hours per week [they] spent sitting at home.” For each questionnaire response, we categorized and assigned point scores to time spent sitting: low (<10 hours/week) = 1 point; medium (11–40 hours/week) = 2 points; and high (>40 hours/week) = 3 points. We then added each subject’s point score for time spent sitting in 1988 and 1990 to define each subject’s overall time spent sitting. Thus, each subject was assigned to 1 of 5 categories of time spent sitting, with point values ranging from 2 (low time sitting in both 1988 and 1990) to 6 (high time sitting in both 1988 and 1990).
Weight (kg)/height (m)2.
No. of packs smoked per day × no. of years smoked among past and current smokers.
Coronary heart disease included any history of myocardial infarction, coronary artery stenosis, or angina.
Rheumatologic disease included any history of systemic lupus erythematosus or rheumatoid arthritis.
Intake from foods and supplements.
Table 2.
Age-Standardized Baseline Characteristics ofMen According to Dietary Patterns (n = 49,238), Health Professionals Follow-up Study, 1986
| Characteristic | Intake of Prudent Pattern |
Intake of Western Pattern |
||||||
| Lowest Quintile (n = 9,893) |
Highest Quintile (n = 9,827) |
Lowest Quintile (n = 9,846) |
Highest Quintile (n = 9,872) |
|||||
| % | Mean (SD) | % | Mean (SD) | % | Mean (SD) | % | Mean (SD) | |
| Age, years | 51.3 (9.0) | 55.0 (9.6) | 55.2 (9.5) | 51.8 (9.3) | ||||
| Total physical activity, METs/weeka,b | 15.8 | 28.6 | 25.5 | 18.1 | ||||
| Sedentary lifestylec | ||||||||
| Lowest amount of time spent sitting | 33.2 | 39.9 | 40.8 | 33.6 | ||||
| Highest amount of time spent sitting | 0.1 | 0.1 | 0.1 | 0.2 | ||||
| Body mass indexd | 25.1 | 24.6 | 24.4 | 25.3 | ||||
| Total caloric intake, kcal/daya | 1,667 | 2,395 | 1,480 | 2,706 | ||||
| Smoking status | ||||||||
| Nonsmokers | 38.6 | 50.5 | 46.8 | 43.5 | ||||
| Ex-smokers | 41.3 | 40.4 | 44.0 | 38.4 | ||||
| Current smokers | 16.1 | 4.7 | 4.8 | 14.0 | ||||
| Missing | 4.0 | 4.4 | 4.4 | 3.1 | ||||
| Pack-years among ever smokersa,e | 12 | 9 | 10 | 11 | ||||
| White race/ethnicity | 90 | 91 | 88 | 93 | ||||
| Nonaspirin nonsteroidal antiinflammatory drug use | 5.5 | 5.0 | 4.9 | 5.7 | ||||
| Warfarin use | 1.9 | 2.6 | 2.8 | 1.9 | ||||
| Multivitamin supplement use | 57.3 | 67.2 | 66.8 | 57.5 | ||||
| Hypertension | 18.6 | 22.5 | 24.3 | 18.2 | ||||
| Coronary heart diseasef | 10.0 | 17.1 | 19.1 | 10.8 | ||||
| Rheumatologic diseaseg | 1.9 | 2.4 | 2.0 | 2.4 | ||||
| Dietary intakea | ||||||||
| Fruit, servings/day | 0.7 | 2.8 | 2.0 | 1.4 | ||||
| Vegetables, servings/day | 1.4 | 5.2 | 3.2 | 3.0 | ||||
| Fish, servings/day | 0.2 | 0.6 | 0.5 | 0.3 | ||||
| Red and processed meat, servings/day | 1.1 | 0.8 | 0.3 | 1.8 | ||||
| Alcohol, g/day | 15.4 | 8.8 | 9.8 | 12.5 | ||||
| Omega-3 fatty acids, g/dayh | 0.2 | 0.4 | 0.4 | 0.2 | ||||
| Trans fatty acids, g/day | 3.4 | 2.1 | 2.0 | 3.4 | ||||
| Total fiber, g/day | 15.0 | 28.3 | 25.3 | 18.2 | ||||
| Vitamin K1, μg/dayh | 22.1 | 17.0 | 15.7 | 23.0 | ||||
| Vitamin B6, mg/dayh | 6.3 | 12.0 | 13.0 | 5.4 | ||||
| Vitamin B12, μg/dayh | 12.1 | 13.2 | 13.5 | 11.5 | ||||
| Vitamin E, mg/dayh | 77 | 119 | 147 | 57 | ||||
Abbreviations: METs, metabolic equivalents; SD, standard deviation.
Age-adjusted means.
Sum of the average time per week spent in each activity × MET value of each activity.
Time spent sitting was assessed in 1988 and in 1990 when subjects were asked to report “the number of hours per week [they] spent sitting at home.” For each questionnaire response, we categorized and assigned point scores to time spent sitting: low (<10 hours/week) = 1 point; medium (11–40 hours/week) = 2 points; and high (>40 hours/week) = 3 points. We then added each subject’s point score for time spent sitting in 1988 and 1990 to define each subject’s overall time spent sitting. Thus, each subject was assigned to 1 of 5 categories of time spent sitting, with point values ranging from 2 (low time sitting in both 1988 and 1990) to 6 (high time sitting in both 1988 and 1990).
Weight (kg)/height (m)2.
No. of packs smoked per day × no. of years smoked among past and current smokers.
Coronary heart disease included any history of myocardial infarction, coronary artery stenosis, or angina.
Rheumatologic disease included any history of systemic lupus erythematosus or rheumatoid arthritis.
Intake from foods and supplements.
Dietary patterns and risk of VTE
After adjustment for potential confounders, only the Western pattern was associated with an increased risk of VTE (Table 3). This risk was significant in men (P < 0.001) and borderline significant in women (P = 0.09). The pooled analysis led to a similar result (for the highest quintile of Western diet vs. the lowest, multivariable pooled relative risk (RR) = 1.22, 95% confidence interval (CI): 0.87, 1.71; P for trend < 0.001; P for between-study heterogeneity = 0.06; I2 = 73.5%). When the most recently reported dietary intakes instead of the cumulative average intake were used to determine dietary patterns, we found no association with the prudent pattern and only a weak association with the Western pattern (for the highest quintile of Western diet vs. the lowest, multivariable pooled RR = 1.18, 95% CI: 0.93, 1.84; P for trend < 0.001; P for between-study heterogeneity = 0.08). When the population was restricted to participants without prior cancer or cardiovascular disease (n = 112,658 women and men), the results did not differ materially (for the highest quintile of Western diet vs. the lowest, multivariable pooled RR = 1.24, 95% CI: 0.91, 1.80; P for trend < 0.001; P for between-study heterogeneity = 0.06).
Table 3.
Association Between Dietary Pattern and Incident Venous Thromboembolism (Pulmonary Embolism or Deep Vein Thrombosis) in the Nurses’ Health Study (n = 80,263), 1984–2008, and the Health Professionals Follow-up Study (n = 49,238), 1986–2008
| Quintile | Women |
Men |
Total |
||||||||
| No. of Persons | Person-Years | Multivariate RRa | 95% CI | No. of Persons | Person-Years | Multivariate RRa | 95% CI | No. of Persons | Multivariate RRa | 95% CI | |
| Prudent Pattern | |||||||||||
| 1 | 273 | 176,241 | 1.00 | Referent | 203 | 90,571 | 1.00 | Referent | 476 | 1.00 | Referent |
| 2 | 331 | 178,785 | 1.13 | 0.96, 1.34 | 255 | 92,763 | 1.04 | 0.87, 1.26 | 586 | 1.09 | 0.97, 1.24 |
| 3 | 331 | 179,585 | 1.08 | 0.91, 1.27 | 290 | 97,185 | 1.00 | 0.83, 1.21 | 621 | 1.04 | 0.92, 1.18 |
| 4 | 306 | 179,043 | 0.96 | 0.81, 1.15 | 256 | 100,299 | 0.80 | 0.66, 0.97 | 562 | 0.88 | 0.73, 1.06 |
| 5 | 299 | 177,310 | 0.94 | 0.77, 1.13 | 348 | 102,021 | 0.96 | 0.79, 1.17 | 647 | 0.95 | 0.83, 1.09 |
| P for trend | 0.14 | 0.13 | 0.13 | ||||||||
| P valueb | 0.84 | ||||||||||
| I2c | 0 | ||||||||||
| Western Pattern | |||||||||||
| 1 | 288 | 176,669 | 1.00 | Referent | 201 | 92,789 | 1.00 | Referent | 489 | 1.00 | Referent |
| 2 | 296 | 178,532 | 1.03 | 0.87, 1.21 | 215 | 92,384 | 1.09 | 0.90, 1.33 | 511 | 1.06 | 0.93, 1.20 |
| 3 | 308 | 179,051 | 1.07 | 0.90, 1.27 | 248 | 96,056 | 1.14 | 0.94, 1.38 | 556 | 1.10 | 0.97, 1.25 |
| 4 | 336 | 178,746 | 1.20 | 1.00, 1.45 | 323 | 99,504 | 1.36 | 1.12, 1.64 | 659 | 1.28 | 1.12, 1.46 |
| 5 | 312 | 177,966 | 1.14 | 0.91, 1.42 | 365 | 102,106 | 1.43 | 1.16, 1.78 | 677 | 1.22 | 0.87, 1.71 |
| P for trend | 0.09 | <0.001 | <0.001 | ||||||||
| P valueb | 0.06 | ||||||||||
| I2c | 73.5 | ||||||||||
Abbreviations: CI, confidence interval; RR, relative risk.
Multivariate RRs were adjusted for age, total physical activity level, physical inactivity level, body mass index (weight (kg)/height (m)2), total caloric intake, smoking, pack-years of smoking, race/ethnicity, spouse’s educational attainment (in women only), parity (in women only), menopausal status (in women only), nonaspirin nonsteroidal antiinflammatory drug use, warfarin use, multivitamin supplement use, hypertension, coronary heart disease, and rheumatologic disease.
P for between-study heterogeneity (quintile 5 vs. quintile 1).
I2, degree of heterogeneity between studies expressed as a percentage of total variance (quintile 5 vs. quintile 1).
We performed several subanalyses of the association between dietary patterns and VTE. First, we stratified our analyses according to the median age at baseline (Figures 1 and 2). Among both the youngest participants (women aged 38–49 years at baseline and men aged 40–52 years at baseline) and the oldest participants (women aged 50–63 years at baseline and men aged 53–75 years at baseline), no association was found between the prudent pattern and the risk of VTE; however, the Western pattern remained weakly associated with an increased risk of VTE among both the youngest participants (borderline significant, P for trend = 0.09) and the oldest participants (P for trend < 0.001). In other subanalyses, we stratified according to BMI and by level of physical activity and found similar results in both analyses (data not shown).
Figure 1.
Pooled multivariate relative risks (RRs) of incident venous thromboembolism according to quintile of adherence to prudent dietary pattern and age at baseline in the Nurses’ Health Study and the Health Professionals Follow-up Study (n =129,501 subjects; 2,892 cases of incident venous thromboembolism). Pooled RRs were adjusted for age, total physical activity level, physical inactivity level, body mass index (weight (kg)/height (m)2), total caloric intake, smoking, pack-years of smoking, race/ethnicity, spouse’s educational attainment (in women only), parity (in women only), menopausal status (in women only), nonaspirin nonsteroidal antiinflammatory drug use, warfarin use, multivitamin supplement use, hypertension, coronary heart disease, and rheumatologic disease. A) Youngest participants, whose age at baseline was lower than the median (i.e., 38–49 years for women and 40–52 years for men). P for trend = 0.10; P for between-study heterogeneity (quintile 5 vs. quintile 1) = 0.39; I2 (degree of heterogeneity between studies expressed as a percentage of total variance, quintile 5 vs. quintile 1) = 0%. B) Oldest participants, whose age at baseline was at the median or higher (i.e., 50–63 years for women and 53–75 years for men). P for trend = 0.14; P for between-study heterogeneity (quintile 5 vs. quintile 1) = 0.70; I2 = 0%. Bars, 95% confidence interval.
Figure 2.
Pooled multivariate relative risks (RRs) of incident venous thromboembolism according to quintile of adherence to Western dietary pattern and age at baseline in the Nurses’ Health Study and the Health Professionals Follow-up Study (n =129,501 subjects; 2,892 cases of incident venous thromboembolism). Pooled RRs were adjusted for age, total physical activity level, physical inactivity level, body mass index (weight (kg)/height (m)2), total caloric intake, smoking, pack-years of smoking, race/ethnicity, spouse’s educational attainment (in women only), parity (in women only), menopausal status (in women only), nonaspirin nonsteroidal antiinflammatory drug use, warfarin use, multivitamin supplement use, hypertension, coronary heart disease, and rheumatologic disease. A) Youngest participants, whose age at baseline was lower than the median (i.e., 38–49 years for women and 40–52 years for men). P for trend = 0.09; P for between-study heterogeneity (quintile 5 vs. quintile 1) = 0.78; I2 (degree of heterogeneity between studies expressed as a percentage of total variance, quintile 5 vs. quintile 1) = 0%. B) Oldest participants, whose age at baseline was at the median or higher (i.e., 50–63 years for women and 53–75 years for men). P for trend < 0.001; P for between-study heterogeneity (quintile 5 vs. quintile 1) = 0.31; I2 = 0%. Bars, 95% confidence interval.
Food groups and nutrients and risk of VTE
Regarding food groups (Table 4) and nutrients (Table 5), there was a consistent inverse association between vitamin E intake and the risk of VTE in women (P for trend = 0.02), in men (P for trend = 0.07), and in the pooled analysis (P for trend = 0.004). For fiber and vitamin B6 intakes, there was a significant inverse association with the risk of VTE in the pooled analysis, with no significant heterogeneity between men and women, although the associations were significant only in men (Table 5). For intakes of red and processed meat and of trans fatty acids, no association with VTE was found in women, whereas a significant positive association was found in men (the percentages of total variance between studies were 88% and 84% for red and processed meat and trans fatty acids, respectively). The association in the pooled analysis was not significant for trans fatty acid intake (for the highest quintile of trans fatty acid intake vs. the lowest, multivariable RR = 1.14, 95% CI: 0.84, 1.53; P for trend = 0.33; P for between-study heterogeneity = 0.01) and was borderline significant for red and processed meat intake (for the highest quintile of red and processed meat intake vs. the lowest, multivariable RR = 1.38, 95% CI: 0.92, 2.06; P for trend = 0.07; P for between-study heterogeneity = 0.004). Analyses in which we used the most recent dietary intakes or nutrient intakes derived from foods only (not supplements) had similar results (data not shown). No associations were found between the risk of VTE and the intakes of fruit, vegetables, fish, alcohol, omega-3 fatty acids, vitamin K1, and vitamin B12 (Tables 4 and 5).
Table 4.
Association Between Specific Food Groups and Incident Venous Thromboembolism (Pulmonary Embolism or Deep Vein Thrombosis) in the Nurses’ Health Study (n = 80,263), 1984–2008, and the Health Professionals Follow-up Study (n = 49,238), 1986–2008
| Quintile | Women |
Men |
Total |
||||||||
| No. of Persons | Person-Years | Multivariate RRa | 95% CI | No. of Persons | Person-Years | Multivariate RRa | 95% CI | No. of Persons | Multivariate RRa | 95% CI | |
| Fruits | |||||||||||
| 1 | 279 | 175,258 | 1.00 | Referent | 182 | 90,447 | 1.00 | Referent | 461 | 1.00 | Referent |
| 2 | 308 | 180,706 | 1.00 | 0.85, 1.18 | 235 | 93,722 | 1.01 | 0.83, 1.22 | 543 | 1.00 | 0.88, 1.14 |
| 3 | 331 | 179,565 | 1.01 | 0.86, 1.20 | 275 | 97,973 | 1.00 | 0.82, 1.21 | 606 | 1.01 | 0.89, 1.14 |
| 4 | 343 | 178,848 | 1.02 | 0.86, 1.20 | 318 | 99,513 | 1.03 | 0.85, 1.25 | 661 | 1.02 | 0.90, 1.16 |
| 5 | 279 | 176,587 | 0.92 | 0.68, 1.12 | 342 | 101,184 | 0.99 | 0.81, 1.21 | 621 | 0.90 | 0.74, 1.08 |
| P for trend | 0.17 | 0.99 | 0.29 | ||||||||
| P valueb | 0.16 | ||||||||||
| I2c | 49.4 | ||||||||||
| Vegetables | |||||||||||
| 1 | 313 | 176,165 | 1.00 | Referent | 148 | 85,054 | 1.00 | Referent | 461 | 1.00 | Referent |
| 2 | 279 | 178,833 | 0.85 | 0.72, 1.00 | 291 | 93,836 | 1.34 | 1.10, 1.64 | 570 | 1.06 | 0.68, 1.66 |
| 3 | 331 | 179,355 | 0.98 | 0.83, 1.15 | 269 | 99,071 | 1.06 | 0.86, 1.30 | 600 | 1.01 | 0.89, 1.14 |
| 4 | 328 | 179,067 | 0.93 | 0.79, 1.10 | 292 | 101,581 | 1.05 | 0.85, 1.29 | 620 | 0.97 | 0.86, 1.11 |
| 5 | 289 | 177,544 | 0.82 | 0.68, 0.98 | 352 | 103,297 | 1.14 | 0.92, 1.41 | 641 | 0.96 | 0.69, 1.33 |
| P for trend | 0.14 | 0.63 | 0.16 | ||||||||
| P valueb | 0.02 | ||||||||||
| I2c | 82.6 | ||||||||||
| Fish | |||||||||||
| 1 | 312 | 178,085 | 1.00 | Referent | 225 | 95,225 | 1.00 | Referent | 537 | 1.00 | Referent |
| 2 | 300 | 175,614 | 0.94 | 0.80, 1.11 | 245 | 85,640 | 1.02 | 0.85, 1.22 | 545 | 0.98 | 0.87, 1.10 |
| 3 | 287 | 183,064 | 0.92 | 0.79, 1.09 | 248 | 101,977 | 0.83 | 0.69, 0.99 | 535 | 0.88 | 0.78, 0.99 |
| 4 | 323 | 176,819 | 0.96 | 0.82, 1.13 | 302 | 98,063 | 0.96 | 0.81, 1.15 | 625 | 0.96 | 0.86, 1.08 |
| 5 | 318 | 177,382 | 0.95 | 0.80, 1.11 | 332 | 101,934 | 0.96 | 0.80, 1.14 | 650 | 0.95 | 0.85, 1.07 |
| P for trend | 0.63 | 0.61 | 0.48 | ||||||||
| P valueb | 0.90 | ||||||||||
| I2c | 0 | ||||||||||
| Red and Processed Meat | |||||||||||
| 1 | 275 | 178,280 | 1.00 | Referent | 156 | 156,807 | 1.00 | Referent | 431 | 1.00 | Referent |
| 2 | 309 | 181,999 | 1.08 | 0.92, 1.27 | 221 | 91,528 | 1.31 | 1.07, 1.61 | 530 | 1.18 | 0.97, 1.43 |
| 3 | 301 | 177,270 | 1.07 | 0.90, 1.27 | 278 | 97,167 | 1.47 | 1.20, 1.79 | 579 | 1.25 | 0.92, 1.69 |
| 4 | 334 | 179,150 | 1.18 | 1.00, 1.41 | 308 | 99,583 | 1.48 | 1.21, 1.81 | 642 | 1.32 | 1.06, 1.64 |
| 5 | 321 | 174,265 | 1.12 | 0.93, 1.35 | 389 | 102,754 | 1.70 | 1.38, 2.09 | 710 | 1.38 | 0.92, 2.06 |
| P for trend | 0.13 | <0.001 | 0.07 | ||||||||
| P valueb | 0.004 | ||||||||||
| I2c | 88.0 | ||||||||||
| Alcohol | |||||||||||
| 1 | 387 | 219,405 | 1.00 | Referent | 268 | 97,521 | 1.00 | Referent | 655 | 1.00 | Referent |
| 2 | 273 | 133,850 | 1.09 | 0.93, 1.27 | 235 | 92,261 | 0.87 | 0.73, 1.03 | 508 | 0.98 | 0.78, 1.22 |
| 3 | 348 | 178,388 | 1.19 | 1.03, 1.38 | 273 | 99,092 | 0.98 | 0.83, 1.17 | 621 | 1.09 | 0.90, 1.31 |
| 4 | 271 | 181,840 | 0.97 | 0.82, 1.14 | 285 | 97,745 | 1.00 | 0.84, 1.18 | 556 | 0.98 | 0.87, 1.10 |
| 5 | 261 | 177,481 | 0.97 | 0.82, 1.14 | 291 | 96,220 | 0.89 | 0.75, 1.06 | 552 | 0.93 | 0.83, 1.05 |
| P for trend | 0.51 | 0.64 | 0.43 | ||||||||
| P valueb | 0.52 | ||||||||||
| I2c | 0 | ||||||||||
Abbreviations: CI, confidence interval; RR, relative risk.
Multivariate RRs were adjusted for age, total physical activity level, physical inactivity level, body mass index (weight (kg)/height (m)2), total caloric intake, smoking, pack-years of smoking, race/ethnicity, spouse’s educational attainment (in women only), parity (in women only), menopausal status (in women only), nonaspirin nonsteroidal antiinflammatory drug use, warfarin use, multivitamin supplement use, hypertension, coronary heart disease, and rheumatologic disease.
P for between-study heterogeneity (quintile 5 vs. quintile 1).
Degree of heterogeneity between studies expressed as a percentage of total variance (quintile 5 vs. quintile 1).
Table 5.
Association Between Specific Nutrients (From Foods and Supplements) and Incident Venous Thromboembolism (Pulmonary Embolism or Deep Vein Thrombosis) in the Nurses’ Health Study (n = 80,263), 1984–2008, and the Health Professionals Follow-up Study (n = 49,238), 1986–2008
| Quintile | Women |
Men |
Total |
||||||||
| No. of Persons | Person-Years | Multivariate RRa | 95% CI | No. of Persons | Person-Years | Multivariate RRa | 95% CI | No. of Persons | Multivariate RRa | 95% CI | |
| Omega-3 Fatty Acids | |||||||||||
| 1 | 304 | 174,037 | 1.00 | Referent | 264 | 95,631 | 1.00 | Referent | 568 | 1.00 | Referent |
| 2 | 300 | 178,876 | 0.96 | 0.82, 1.12 | 296 | 97,078 | 1.01 | 0.86, 1.19 | 596 | 0.98 | 0.88, 1.10 |
| 3 | 300 | 181,334 | 0.92 | 0.78, 1.08 | 259 | 98,136 | 0.88 | 0.74, 1.04 | 559 | 0.90 | 0.80, 1.01 |
| 4 | 325 | 180,214 | 0.98 | 0.84, 1.15 | 259 | 95,679 | 0.91 | 0.77, 1.08 | 584 | 0.95 | 0.85, 1.07 |
| 5 | 311 | 176,503 | 0.94 | 0.80, 1.10 | 274 | 96,315 | 0.91 | 0.76, 1.08 | 585 | 0.92 | 0.82, 1.07 |
| P for trend | 0.59 | 0.13 | 0.15 | ||||||||
| P valueb | 0.79 | ||||||||||
| I2c | 0 | ||||||||||
| Trans Fatty Acids | |||||||||||
| 1 | 310 | 178,155 | 1.00 | Referent | 221 | 96,529 | 1.00 | Referent | 531 | 1.00 | Referent |
| 2 | 305 | 179,976 | 0.95 | 0.81, 1.12 | 263 | 96,845 | 1.14 | 0.95, 1.36 | 568 | 1.04 | 0.87, 1.23 |
| 3 | 309 | 178,742 | 0.96 | 0.82, 1.13 | 292 | 97,276 | 1.24 | 1.05, 1.48 | 601 | 1.09 | 0.85, 1.40 |
| 4 | 314 | 178,608 | 0.99 | 0.84, 1.16 | 279 | 96,793 | 1.19 | 0.99, 1.43 | 593 | 1.08 | 0.90, 1.30 |
| 5 | 302 | 175,483 | 0.98 | 0.83, 1.15 | 297 | 95,396 | 1.33 | 1.11, 1.59 | 599 | 1.14 | 0.84, 1.53 |
| P for trend | 0.97 | 0.003 | 0.33 | ||||||||
| P valueb | 0.01 | ||||||||||
| I2c | 83.8 | ||||||||||
| Total Fiber | |||||||||||
| 1 | 292 | 173,380 | 1.00 | Referent | 266 | 94,912 | 1.00 | Referent | 558 | 1.00 | Referent |
| 2 | 300 | 179,495 | 0.92 | 0.79, 1.09 | 273 | 96,687 | 0.92 | 0.78, 1.09 | 573 | 0.92 | 0.82, 1.04 |
| 3 | 318 | 179,669 | 0.94 | 0.80, 1.11 | 284 | 97,533 | 0.90 | 0.75, 1.06 | 602 | 0.92 | 0.82, 1.04 |
| 4 | 332 | 178,823 | 0.97 | 0.82, 1.14 | 270 | 97,298 | 0.83 | 0.69, 0.99 | 602 | 0.90 | 0.77, 1.05 |
| 5 | 298 | 179,597 | 0.86 | 0.72, 1.02 | 259 | 96,409 | 0.79 | 0.66, 0.95 | 557 | 0.83 | 0.73, 0.94 |
| P for trend | 0.19 | 0.006 | 0.01 | ||||||||
| P valueb | 0.54 | ||||||||||
| I2c | 0 | ||||||||||
| Vitamin K1 | |||||||||||
| 1 | 296 | 177,886 | 1.00 | Referent | 223 | 95,846 | 1.00 | Referent | 519 | 1.00 | Referent |
| 2 | 317 | 179,210 | 1.03 | 0.88, 1.21 | 291 | 97,489 | 1.24 | 1.04, 1.47 | 608 | 1.13 | 0.94, 1.34 |
| 3 | 314 | 179,803 | 1.02 | 0.87, 1.20 | 295 | 97,739 | 1.25 | 1.05, 1.48 | 609 | 1.12 | 0.92, 1.37 |
| 4 | 283 | 178,930 | 0.93 | 0.79, 1.09 | 276 | 96,861 | 1.19 | 0.99, 1.42 | 559 | 1.05 | 0.82, 1.33 |
| 5 | 330 | 175,135 | 1.11 | 0.95, 1.30 | 267 | 94,904 | 1.17 | 0.98, 1.40 | 597 | 1.14 | 1.01, 1.28 |
| P for trend | 0.56 | 0.22 | 0.21 | ||||||||
| P valueb | 0.67 | ||||||||||
| I2c | 0 | ||||||||||
| Vitamin B6 | |||||||||||
| 1 | 290 | 178,373 | 1.00 | Referent | 259 | 97,789 | 1.00 | Referent | 549 | 1.00 | Referent |
| 2 | 306 | 174,639 | 0.95 | 0.80, 1.11 | 274 | 93,136 | 0.92 | 0.77, 1.09 | 580 | 0.93 | 0.83, 1.05 |
| 3 | 328 | 179,342 | 0.98 | 0.83, 1.16 | 272 | 97,623 | 0.82 | 0.68, 0.98 | 600 | 0.90 | 0.75, 1.07 |
| 4 | 301 | 177,735 | 0.88 | 0.74, 1.06 | 280 | 96,189 | 0.85 | 0.71, 1.03 | 581 | 0.87 | 0.77, 0.99 |
| 5 | 315 | 180,875 | 0.94 | 0.80, 1.12 | 267 | 98,102 | 0.82 | 0.68, 0.98 | 582 | 0.88 | 0.77, 1.02 |
| P for trend | 0.42 | 0.03 | 0.04 | ||||||||
| P valueb | 0.26 | ||||||||||
| I2c | 21.7 | ||||||||||
| Vitamin B12 | |||||||||||
| 1 | 273 | 175,400 | 1.00 | Referent | 226 | 98,941 | 1.00 | Referent | 499 | 1.00 | Referent |
| 2 | 309 | 175,312 | 1.06 | 0.90, 1.25 | 267 | 96,710 | 1.06 | 0.89, 1.27 | 576 | 1.06 | 0.94, 1.20 |
| 3 | 312 | 180,367 | 1.03 | 0.87, 1.22 | 294 | 94,505 | 1.14 | 0.96, 1.37 | 606 | 1.08 | 0.96, 1.22 |
| 4 | 329 | 180,630 | 1.04 | 0.88, 1.23 | 303 | 95,474 | 1.10 | 0.92, 1.31 | 632 | 1.07 | 0.94, 1.20 |
| 5 | 317 | 179,255 | 1.01 | 0.85, 1.20 | 262 | 97,209 | 0.95 | 0.79, 1.15 | 579 | 0.98 | 0.87, 1.12 |
| P for trend | 0.95 | 0.68 | 0.74 | ||||||||
| P valueb | 0.67 | ||||||||||
| I2c | 0 | ||||||||||
| Vitamin E | |||||||||||
| 1 | 309 | 171,543 | 1.00 | Referent | 237 | 92,629 | 1.00 | Referent | 546 | 1.00 | Referent |
| 2 | 321 | 176,514 | 0.95 | 0.81, 1.12 | 248 | 94,683 | 0.84 | 0.70, 1.01 | 569 | 0.90 | 0.80, 1.02 |
| 3 | 306 | 181,240 | 0.87 | 0.74, 1.03 | 296 | 97,876 | 0.93 | 0.78, 1.11 | 602 | 0.90 | 0.79, 1.02 |
| 4 | 311 | 183,081 | 0.85 | 0.72, 1.01 | 319 | 101,722 | 0.94 | 0.79, 1.13 | 630 | 0.89 | 0.79, 1.01 |
| 5 | 293 | 178,586 | 0.84 | 0.70, 0.99 | 252 | 95,929 | 0.78 | 0.64, 0.94 | 545 | 0.81 | 0.71, 0.92 |
| P for trend | 0.02 | 0.07 | 0.004 | ||||||||
| P valueb | 0.55 | ||||||||||
| I2c | 0 | ||||||||||
Abbreviations: CI, confidence interval; RR, relative risk.
Multivariate RRs were adjusted for age, total physical activity level, physical inactivity level, body mass index (weight (kg)/height (m)2), total caloric intake, smoking, pack-years of smoking, race/ethnicity, spouse’s educational attainment (in women only), parity (in women only), menopausal status (in women only), nonaspirin nonsteroidal antiinflammatory drug use, warfarin use, multivitamin supplement use, hypertension, coronary heart disease, and rheumatologic disease.
P for between-study heterogeneity (quintile 5 vs. quintile 1).
Degree of heterogeneity between studies expressed as a percentage of total variance (quintile 5 vs. quintile 1).
DISCUSSION
In a pooled analysis of men and women, we found a weak association between diet and risk of VTE. No relation was found between the prudent diet pattern and the risk of VTE; however, we did find a weak positive association between the Western diet pattern and the risk of VTE. Stratification according to age, BMI, and physical activity level led to similar results. Regarding specific foods or nutrients, no associations were found between the risk of VTE and the intakes of fruits, vegetables, fish, alcohol, omega-3 fatty acids, vitamin K1, and vitamin B12. Vitamin E had a protective association, as did fiber and vitamin B6, particularly in men. Similarly, we found that increased intakes of red and processed meat and trans fatty acids might be associated with VTE in men. These sex differences should be considered with caution, as issues of sample size and multiple testing could have influenced the results. Our use of data from 2 large prospective cohorts of health professionals who were 38–73 years of age at baseline and followed for more than 20 years enabled us to establish a temporally correct relation between dietary intake and VTE and to control for many potential confounders in our analyses.
To our knowledge, there have been only 2 previous cohort studies in which dietary patterns and the risk of idiopathic VTE were examined (6, 7). Steffen et al. (6) studied 14,962 middle-aged (45–64 years old) men and women from the ARIC Study. Although the investigators found no association between the prudent pattern and idiopathic VTE, they found that a greater consumption of the Western pattern was associated with an increased risk of idiopathic VTE. The latter finding appeared to be driven by persons in the highest quintile, which suggests a threshold effect rather than a dose-response relation between the Western pattern and idiopathic VTE. Lutsey et al. (7) studied 37,393 white women aged 59–65 years from the IWHS and reported no association between either dietary pattern and the risk of VTE.
Data relating food groups/nutrients to VTE are scant, and in the few published epidemiologic studies of this relation, investigators reported conflicting results (6, 7, 11). Our findings regarding vitamin B6 and red and processed meat intakes are consistent with those from the ARIC Study (6); of the 6 food groups and 5 nutrients investigated, only intakes of vitamin B6 and fruits and vegetables were significantly associated with a decreased risk of VTE, and only red and processed meat intake was associated with an increased risk of VTE (6). A borderline-significant negative association was reported for folate intake (6). However, in 2 large randomized trials, supplementation with vitamins B6 and B12 and folic acid did not reduce VTE incidence (16) or recurrence (15). In the IWHS, of the 11 food groups and 6 nutrients investigated, only alcohol intake was associated with a reduced risk of VTE (7). Recently, in a case-control study of 292 Asian men and women, 6 food groups were investigated, and only a high vegetable intake was negatively associated with VTE (11). Our finding regarding vitamin E is consistent with a previous long-term trial in the Women’s Health Study, in which intake of 600 IU of natural-source vitamin E on alternate days was associated with a significant 21% reduction in the hazard of VTE (17).
There are several possible explanations for the different findings in the ARIC Study, the IWHS, and our study. To explain the contrary findings between the ARIC Study and the IWHS, Lutsey et al. (7) hypothesized that diet may play a bigger etiologic role in the development of VTE in younger people than in older people. In the present study, we found similar results in strata of younger and older participants, although this limited the power of our study. Another possible explanation is race/ethnicity; approximately 25% of participants in the ARIC Study were black, whereas most subjects in the IWHS, NHS, and HPFS were white. Cultural factors influence food-related behavior and attitudes (24), and black Americans have a higher incidence of VTE than do their white counterparts (25). Tang et al. (26) recently reported that, compared with whites, blacks experienced 3 times more frequent fatal out-of-hospital PE in the New York City area and died at an age that was 10 years younger in the case of fatal PE, but they did not carry factor V Leiden and the prothrombin G20210A mutation (27). However, the impact of these black/white differences on the diet-PE association is not clear.
Another possible explanation for the discrepant findings is that the primary outcome differed across studies. In the ARIC Study, the main outcome was non-cancer-related VTE (6), and in the IWHS, the main outcome was all VTE (both idiopathic and nonidiopathic) (7). We also investigated all VTE in the present study (PE and DVT) because we were not able to distinguish between idiopathic and nonidiopathic DVT. The discrepant findings might also be related to our definition of DVT in women, which was based on a classification from the International Classification of Diseases, Eighth Revision (code 451). This code corresponds to phlebitis and thrombophlebitis and therefore might include superficial thrombophlebitis. Thus, our outcome of VTE is a mixture of unconfirmed cases that are of both idiopathic and nonidiopathic origin. Many of the latter causes (e.g., cancer, surgery, trauma) are associated with adherence to a Western diet or nonadherence to a prudent diet. Further analyses are warranted to assess whether the association between diet and VTE persists when known VTE risk factors (conditions) are removed from the study outcome.
In conclusion, we found a weak association between diet and the risk of VTE in US women and men. Nutritional factors of possible etiologic relevance include the Western dietary pattern and intakes of vitamins E and B6, fiber, red and processed meat, and trans fatty acids. Other nutritional factors were not associated with VTE in our 2 large cohorts. The unique characteristics of our cohorts might limit generalizability of the results to individuals of nonwhite race/ethnicity or lower socioeconomic status or persons with limited access to medical care. Further studies are warranted in these vulnerable populations.
Acknowledgments
Author affiliations: INSERM, CESP Centre for Research in Epidemiology and Population Health, U1018, Respiratory and Environmental Epidemiology Team, Villejuif, France (Raphaëlle Varraso); Université Paris Sud 11, UMRS 1018, Villejuif, France (Raphaëlle Varraso); Department of Emergency Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts (Christopher Kabrhel, Carlos A. Camargo, Jr.); Department of Medicine, Cardiovascular Division, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts (Samuel Z. Goldhaber); Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts (Eric B. Rimm); Department of Nutrition, Harvard School of Public Health, Boston, Massachusetts (Eric B. Rimm); and Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts (Eric B. Rimm, Carlos A. Camargo, Jr.).
This work was supported by National Institutes of Health grants R21AG031079, P01CA87969, R01HL034594, and P01 CA055075.
The authors thank Gary Chase, Karen Corsano, and Betsy Frost-Hawes for invaluable assistance with the underlying cohort studies.
Conflict of interest: none declared.
Glossary
Abbreviations
- ARIC
Atherosclerosis Risk in Communities
- BMI
body mass index
- CI
confidence interval
- DVT
deep vein thrombosis
- FFQ
food frequency questionnaire
- HPFS
Health Professionals Follow-up Study
- IWHS
Iowa Women’s Health Study
- NHS
Nurses’ Health Study
- PE
pulmonary embolism
- RR
relative risk
- VTE
venous thromboembolism
References
- 1.Raskob GE, Silverstein R, Bratzler DW, et al. Surveillance for deep vein thrombosis and pulmonary embolism: recommendations from a national workshop. Am J Prev Med. 2010;38(suppl 4):S502–S509. doi: 10.1016/j.amepre.2010.01.010. [DOI] [PubMed] [Google Scholar]
- 2.Kabrhel C, Varraso R, Goldhaber SZ, et al. Prospective study of BMI and the risk of pulmonary embolism in women. Obesity (Silver Spring) 2009;17(11):2040–2046. doi: 10.1038/oby.2009.92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Goldhaber SZ, Grodstein F, Stampfer MJ, et al. A prospective study of risk factors for pulmonary embolism in women. JAMA. 1997;277(8):642–645. [PubMed] [Google Scholar]
- 4.Everett BM, Glynn RJ, Buring JE, et al. Lipid biomarkers, hormone therapy and the risk of venous thromboembolism in women. J Thromb Haemost. 2009;7(4):588–596. doi: 10.1111/j.1538-7836.2009.03302.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Cundiff DK, Agutter PS, Malone PC, et al. Diet as prophylaxis and treatment for venous thromboembolism? Theor Biol Med Model. 2010;7(1):31. doi: 10.1186/1742-4682-7-31. (doi:10.1186/1742-4682-7-31) [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Steffen LM, Folsom AR, Cushman M, et al. Greater fish, fruit, and vegetable intakes are related to lower incidence of venous thromboembolism: the Longitudinal Investigation of Thromboembolism Etiology. Circulation. 2007;115(2):188–195. doi: 10.1161/CIRCULATIONAHA.106.641688. [DOI] [PubMed] [Google Scholar]
- 7.Lutsey PL, Steffen LM, Virnig BA, et al. Diet and incident venous thromboembolism: the Iowa Women’s Health Study. Am Heart J. 2009;157(6):1081–1087. doi: 10.1016/j.ahj.2009.04.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lindqvist PG, Epstein E, Olsson H. The relationship between lifestyle factors and venous thromboembolism among women: a report from the MISS Study. Br J Haematol. 2009;144(2):234–240. doi: 10.1111/j.1365-2141.2008.07460.x. [DOI] [PubMed] [Google Scholar]
- 9.Pomp ER, Rosendaal FR, Doggen CJ. Alcohol consumption is associated with a decreased risk of venous thrombosis. Thromb Haemost. 2008;99(1):59–63. doi: 10.1160/TH07-07-0470. [DOI] [PubMed] [Google Scholar]
- 10.Pahor M, Guralnik JM, Havlik RJ, et al. Alcohol consumption and risk of deep venous thrombosis and pulmonary embolism in older persons. J Am Geriatr Soc. 1996;44(9):1030–1037. doi: 10.1111/j.1532-5415.1996.tb02933.x. [DOI] [PubMed] [Google Scholar]
- 11.Bhoopat L, Rojnuckarin P, Hiransuthikul N, et al. Low vegetable intake is strongly associated with venous thromboembolism in Thai population. Blood Coagul Fibrinolysis. 2010;21(8):758–763. doi: 10.1097/MBC.0b013e3283403537. [DOI] [PubMed] [Google Scholar]
- 12.Tsai AW, Cushman M, Rosamond WD, et al. Cardiovascular risk factors and venous thromboembolism incidence: the Longitudinal Investigation of Thromboembolism Etiology. Arch Intern Med. 2002;162(10):1182–1189. doi: 10.1001/archinte.162.10.1182. [DOI] [PubMed] [Google Scholar]
- 13.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: 10.1093/aje/kwi309. [DOI] [PubMed] [Google Scholar]
- 14.Samama MM. An epidemiologic study of risk factors for deep vein thrombosis in medical outpatients: The Sirius Study. Arch Intern Med. 2000;160(22):3415–3420. doi: 10.1001/archinte.160.22.3415. [DOI] [PubMed] [Google Scholar]
- 15.den Heijer M, Willems HP, 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(1):139–144. doi: 10.1182/blood-2006-04-014654. [DOI] [PubMed] [Google Scholar]
- 16.Ray JG, Kearon C, Yi Q, et al. Homocysteine-lowering therapy and risk for venous thromboembolism: a randomized trial. Ann Intern Med. 2007;146(11):761–767. doi: 10.7326/0003-4819-146-11-200706050-00157. [DOI] [PubMed] [Google Scholar]
- 17.Glynn RJ, Ridker PM, Goldhaber SZ, et al. 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: 10.1161/CIRCULATIONAHA.107.716407. [DOI] [PubMed] [Google Scholar]
- 18.Miller GJ. Dietary fatty acids and the haemostatic system. Atherosclerosis. 2005;179(2):213–227. doi: 10.1016/j.atherosclerosis.2004.10.010. [DOI] [PubMed] [Google Scholar]
- 19.Booth SL, Golly I, Sacheck JM, et al. Effect of vitamin E supplementation on vitamin K status in adults with normal coagulation status. Am J Clin Nutr. 2004;80(1):143–148. doi: 10.1093/ajcn/80.1.143. [DOI] [PubMed] [Google Scholar]
- 20.Colditz GA, Martin P, Stampfer MJ, et al. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol. 1986;123(5):894–900. doi: 10.1093/oxfordjournals.aje.a114319. [DOI] [PubMed] [Google Scholar]
- 21.Hu FB, Rimm EB, Stampfer MJ, et al. Prospective study of major dietary patterns and risk of coronary heart disease in men. Am J Clin Nutr. 2000;72(4):912–921. doi: 10.1093/ajcn/72.4.912. [DOI] [PubMed] [Google Scholar]
- 22.Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol. 1986;124(1):17–27. doi: 10.1093/oxfordjournals.aje.a114366. [DOI] [PubMed] [Google Scholar]
- 23.Huedo-Medina TB, Sánchez-Meca J, Marín-Martínez F, et al. Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychol Methods. 2006;11(2):193–206. doi: 10.1037/1082-989X.11.2.193. [DOI] [PubMed] [Google Scholar]
- 24.Shepherd R. Social determinants of food choice. Proc Nutr Soc. 1999;58(4):807–812. doi: 10.1017/s0029665199001093. [DOI] [PubMed] [Google Scholar]
- 25.Heit JA, Beckman MG, Bockenstedt PL, et al. Comparison of characteristics from white and black Americans with venous thromboembolism: a cross-sectional study. CDC Thrombosis and Hemostasis Centers Research and Prevention Network. Am J Hematol. 2010;85(7):467–471. doi: 10.1002/ajh.21735. [DOI] [PubMed] [Google Scholar]
- 26.Tang Y, Sampson B, Pack S, et al. Ethnic differences in out-of-hospital fatal pulmonary embolism. Circulation. 2011;123(20):2219–2225. doi: 10.1161/CIRCULATIONAHA.110.976134. [DOI] [PubMed] [Google Scholar]
- 27.Bounameaux H, Rosendaal FR. Venous thromboembolism: why does ethnicity matter? Circulation. 2011;123(20):2189–2191. doi: 10.1161/CIRCULATIONAHA.111.031690. [DOI] [PubMed] [Google Scholar]