Heterogeneity's ruses: the neglected role of between-individual variability in longitudinal studies of COPD exacerbations.

Abstract

Studying the causal and temporal association between past and future exacerbations in COPD is an active area of research. Standard survival analysis techniques often used in such studies typically produce results that pertain to the overall population, whereas the greatest interest is in the study of associations within individuals. A factor that can lead to profound discrepancies between population-level and individual-level survival patterns is the between-individual heterogeneity in the rate of exacerbations. We briefly review two studies that, while reporting valid results for the overall population, drew conclusions at the individual level that could not be supported by the observations. We caution on the distinction between population and individual-level associations in survival analysis, and recommend accounting for heterogeneity in future studies.