Abstract

The authors examined the impact of smoking status on the relation between body mass index (weight (kg)/height (m)2) and mortality across a group of 15 diverse observational studies. The studies included a heterogeneous sample of national samples, cohort studies with mortality follow-up, and clinical trials. Consideration of the data according to natural strata resulted in the formation of 42 analytic cohorts. The authors examined survival through the end of follow-up for each study, as influenced by body mass index, age, and current smoking status at baseline, using a proportional hazards model to describe the relation between body mass index and mortality with control for age and smoking status. In this paper, the authors demonstrate that the estimated body mass index of minimum mortality changes when data are analyzed while ignoring smoking status; but they also demonstrate through a simulation study that eliminating smokers from the data sets prior to analysis produces results similar to those expected from the elimination of numerically similar random proportions of the data sets prior to analysis. Based on the results of these analyses, the authors find no support for the commonly held practice of eliminating smokers from a data set prior to examining the body mass index-mortality relation. Am J Epidemiol 1999;150: 1297-308.

Author notes

*A complete list of coinvestigators is provided in the Acknowledgment.