Segmented generalized mixed effect models to evaluate health outcomes

Excerpt

Randomized placebo-controlled trials (RCTs) are considered the gold standard for assessing the effect of exposures (e.g., treatments) or interventions (e.g., policies) on a variety of outcomes. By design, randomization “controls” for confounders to yield internally valid inference. However due to high costs, feasibility issues and/or ethical considerations, the RCT study design may be unable to answer pertinent public health-related research questions (West et al. 2008). Such questions include real-world effectiveness of newly marketed medications or the evaluation of health policies. Observational studies can bridge knowledge gaps left by RCTs. The following article will explain how to extend a pre–post study design using a segmented generalized mixed model to evaluate the impact of acute individual-level exposures on health outcomes. We describe the advantages of using repeated measures over traditional pre–post designs, what exposures are appropriate to analyze, and how different impact models can be parameterized. Like all methods, this approach comes with strengths, assumptions and limitations, which we discuss. …