Methodologic issues for pooling dietary data.

This paper presents the methodologic issues to consider when conducting a pooled analysis of data from nutritional epidemiologic studies. Because a wide variety of dietary assessment methods have been used in nutritional epidemiology, pooled analyses often combine data collected under very different circumstances. Whether these different methods influence the results obtained in a pooled analysis has not yet been examined. One of the main concerns in a pooled analysis is whether the data to be pooled are comparable. Questions to consider regarding data comparability and methods to address these questions in a pooled analysis are discussed.