Exposure range matters: considering nonlinear associations in the meta-analysis of environmental pollutant exposure using examples of per- and polyfluoroalkyl substances and birth outcomes.

This article focuses on the limitations of conventional linear-based meta-analysis in environmental epidemiology, particularly its assumption of linear exposure-outcome relationships and neglect of exposure range variability. Through simulation studies and reanalysis of two published meta-analyses on prenatal exposure to per- and polyfluoroalkyl substances (PFAS)—specifically perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS)—and their associations with birth weight and preterm birth, the authors demonstrate that ignoring nonlinear dose-response patterns can lead to misleading summary effect estimates. They propose subgroup meta-analyses stratified by exposure levels as a practical approach to detect potential nonlinearities and heterogeneity in effect sizes across exposure ranges, while acknowledging challenges in selecting appropriate cut-offs. The article emphasizes the need for improved reporting of exposure distributions in individual studies and encourages the development and application of advanced nonlinear meta-analytic methods to better inform environmental health research and policy.