Varraso et al. Respond to "Diet and Venous Thromboembolism"

Abstract

We appreciate the thoughtful comments by Dr. Lutsey (1) about our analysis (2) of data from the Nurses’ Health Study and the Health Professional Follow-up Study. We agree with her conclusion that the association between diet and the risk of venous thromboembolism (VTE) is likely weak. In the invited commentary (1), Dr. Lutsey discussed the potential for dietary measurement error. We agree that measurement of dietary intakes is difficult, especially because they change over time. However, in addition to considering errors in the measurement of the dietary exposures, one might also examine the heterogeneity of the outcome (in this case, VTE). In most of the studies cited in Table 1 of the invited commentary (1), the authors investigated all VTEs (i.e., all cases of pulmonary embolism (PE) or deep vein thrombosis) without differentiating between idiopathic and provoked VTE. If one studies all types of VTE, he or she also studies the risk factors for VTE (e.g., cancer). Thus, if one wants to study novel risk factors for VTE, one might preferentially study cases of idiopathic VTE (i.e., cases that were unrelated to classic risk factors for malignancy, recent surgery, and trauma). Although our data did not permit us to distinguish between idiopathic and provoked VTE, we could specifically analyze idiopathic PEs. Over the follow-up period, we identified 308 women and 109 men who had an incident idiopathic PE. After adjusting for 17 potential confounders (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), we found no association among dietary patterns, foods, or nutrients and the risk of idiopathic PE in either women or men. The pooled relative risk of PE for persons in the highest quintile of prudent diet intake compared with persons in the lowest was 0.85 (95% confidence interval (CI): 0.42, 1.72; P for trend = 0.78); for Western diet intake, the corresponding relative risk was 1.14 (95% CI: 0.53, 2.45; P for trend = 0.94). These results provide further evidence that there is little (if any) association between diet and VTE. In our study (2), we investigated several foods and nutrients in relation to VTE, but we did not study coffee intake. A recent study in which it was reported that moderate coffee consumption might be associated with a reduced risk of VTE (3) led us to conduct further analyses regarding the potential role of coffee intake in the development of VTE. However, we found no association between coffee intake and the risk of VTE (for the highest quintile of coffee intake compared with the lowest, relative risk (RR) = 1.02, 95% CI: 0.87, 1.22) or idiopathic PE (RR = 0.94, 95% CI: 0.44, 1.18). As mentioned by Dr. Lutsey, diet may also influence VTE risk through its relation to obesity, which is a strong risk factor for VTE. We previously showed that increasing body mass index had a strong linear association with the risk of idiopathic PE in the Nurses’ Health Study (4). This association remained statistically significant after further adjustment for dietary patterns (for each 1-unit increase in body mass index, RR = 1.09, 95% CI: 1.07, 1.10; P < 0.001). In addition to obesity, physical activity or inactivity might also modulate the association of diet, obesity, and VTE. Recently, we reported that in the Nurses’ Health Study, the risk of idiopathic PE was more than 2-fold higher in women who spent the most time sitting than in women who spent the least time sitting, independent of obesity, diet, and other factors (5). The interrelations between nutritional factors and lifestyle factors (diet, obesity, physical activity, and physical inactivity) provide important challenges for epidemiologic researchers. Accordingly, there is a need for further research to disentangle the independent effect of each nutritional factor on VTE risk by using, for example, causal models (6). This analytic approach allows for testing of different models of causality in the epidemiology of disease arising out of environmental factors and analysis of the interplay between these factors. Such an approach might help us find a stronger scientific basis to support nutritional recommendations that lower the risk of VTE.