ABSTRACT
This chapter explains how to conduct statistical inference on dose-response relationships, describing how a response varies across levels of a quantitative factor, based on tables of data summarized by categorizing the quantitative factor. It provides a flexible framework to conduct a two-stage dose-response meta-analysis. The versatility of regression splines to answer complex research questions is explained in detail. Algorithms to reconstruct the missing covariances of empirical contrasts are presented. Testing hypotheses about the parameters and presentation of summary dose-response relationships are illustrated. The chapter also compares a dose-response analysis of individual participant data to that of the summarized data arising from large prospective cohort studies. Challenges of a two-stage dose-response meta-analysis are discussed with an indication of recent methodological developments. The examples shown in this chapter can be reproduced with software packages developed in Stata and R.
