Cereal fiber and coronary heart disease: a comparison of modeling approaches for repeated dietary measurements, intermediate outcomes, and long follow-up.

The appropriate manner of incorporating incident angina, coronary artery bypass surgery (CABG), percutaneous coronary intervention (PCI), diabetes, hypertension, and hypercholesterolemia, which may be both intermediate events and time-dependent confounders, into analyses of diet and coronary heart disease (CHD) is not clear. Using data from 72,266 women in the Nurses' Health Study between 1984 and 2006, the authors examined the relation between report of intermediate events and change in cereal fiber intake and used different proportional hazards models to evaluate the association between cereal fiber and CHD. Cereal fiber intake increased significantly among participants who reported hypercholesterolemia (0.42 g/day; 95% CI: 0.34, 0.51 g/day) and diabetes (0.07 g/day with each additional 2-year increment; 95% CI: 0.01, 0.13 g/day). However, angina, CABG/PCI, and hypertension were not associated with a change in cereal fiber intake and thus were not important time-dependent confounders. Cereal fiber intake was inversely associated with risk of CHD in all proportional hazards models, but results varied modestly depending on the approach used to incorporate multiple measures of diet. Because stopping the updating of dietary variables when an intermediate event is diagnosed may lead to misclassification, future analyses should consider updating diet even after these diagnoses to best represent long-term intake. To best evaluate associations with incidence of disease, considerations should include the temporal trends in diet, changes in intake following intermediate events, and latency patterns. Sensitivity analyses can also be useful.