Abstract P344: Hypothetical Interventions on Fish Consumption in Mid and Later Life and the Risk of Coronary Heart Disease in Women: An Application of the Parametric G-formula

Abstract

Background_Long-term randomized trials of interventions on fish and primary prevention of coronary heart disease (CHD) are unavailable. The parametric g-formula uses observational data to estimate the effect of changes on fish consumption that would have been observed under hypothetical interventions. Methods and Results_Among 53,799 women from the Nurses’ Health Study aged 39-67 and free of CHD at baseline in 1986, 1,891 developed incident CHD (1,145 non fatal, 830 fatal) by 2008. Women provided information on risk factors and medical conditions every 2 years and on diet every 4 years. We used the parametric g-formula to evaluate hypothetical interventions from baseline forward on fish intake. Specifically, we estimated the CHD risks if all women had changed their fish intake to zero servings per week, and if all women had changed their fish intake to at least 1, 2, 3 and 5 servings per week during the entire follow-up. We adjusted for age and family history of CHD, and for baseline and time-varying menopausal status, use of oral contraceptives, postmenopausal hormone therapy, aspirin, multivitamin and vitamin E supplements, diabetes, hypertension, hypercholesterolemia, stroke, angina or CABG, BMI, physical activity, smoking and intake of calories, trans fatty acids, cereal fiber and red meat. The estimated CHD risk was 4.13 if fish intake had been changed to zero after 1986, and 3.77, 3.53, 3.32 and 3.35, if all women had changed their fish intake to at least 1, 2, 3 and 5 servings per week, respectively. Compared with changing fish intake to zero, the risk ratio of CHD was 0.85 (95% CI 0.75, 0.91) for changing intake to at least 2 servings/week, and 0.80 (95% CI 0.71, 0.97) for changing intake to at least 3 servings. This estimated protective effect of increased fish consumption was restricted to fatal CHD. Results were not appreciably altered when we specified that red meat was replaced by fish intake. Conclusion_Our results suggest that increasing fish consumption in mid or later life may lower the 22-year risk of CHD in women by 20%. Using the parametric g-formula allowed to explicitly specify hypothetical interventions to evaluate the effectiveness of diet change over a long period. Like all observational estimates, the parametric g-formula estimates may be affected by unmeasured confounding, model misspecification, and measurement error.