To estimate a summarized dose-response relation across different exposure levels from epidemiologic data, meta-analysis often needs to take into account heterogeneity across studies beyond the variation associated with fixed effects. We extended a generalized-least-squares method and a...