We must rely on observational data to determine the long-term relationship between body weight and mortality. Individual analyses of the observational studies concerning the relationship between body mass index (BMI-weight(KG)/height(m)2) and mortality have produced inconsistent and confusing results. The association between BNI and mortality has been described a positive, J-shaped, inverse J-shaped, U-shaped, nonexistent, and even inverse. We are conducting a study to examine both the relationship of BMI to various health endpoints and the reasons for the inconsistencies among published reports. During our initial funding, we collected data from 19 studies containing over 350,000 observations and approximately 58,000 deaths, involving several racial/ethnic groups, both genders, and varied baseline risk. We obtained results that were much more consistent when we subjected multiple studies to uniform analytic protocols. And we demonstrated that some common methods of analysis (such as the deletion of large segments of the sample prior to analysis) produce invalid inferences. We have to date focused our efforts on examining the relationship between BMI and general mortality. Our data, however, are extensive, and we propose to expand our analyses to include the following issues: the effect of BMI on the prognoses for hypertensives, the role of BMI as a risk factor for CHD morbidity and mortality, the role of BMI in determining the risk of CHD morbidity and mortality among diabetics, and the role of age in the BMI-mortality relationships. We will also examine the relationship between BMI and the following specific causes of mortality: CHD, CVD, cerebrovascular disease, diabetes, cancer, and all other causes mortality. We will apply uniform methodologies to these divers datasets to attempt to eliminate or explain the inconsistencies that appear in the literature concerning these questions involving BMI.
