ABSTRACT
Evidence-based evaluation plays a central role in health outcomes, health services, and policy research. It is the conscientious use of the best available evidence in evaluating interventions, broadly defined as medical treatments, clinical practice patterns, health behaviors, educational programs, or health policies, that may ultimately lead to improvements in healthcare quality and patient outcomes. The preferred setup for inferring causal effects is through randomized controlled experiments, as it often requires weaker assumptions and is easy to interpret (Shadish et al. 2002). In the evaluation of healthcare systems, health policies or complex clinical practices, however, random allocation of patients to different intervention arms is not feasible due to practical or ethical reasons. Instead, observational data are often used, and the participants of which are often more representative of the actual patient population. In this chapter, we will review the commonly used strategies for inferring causal relationship from observational data via propensity score adjustment. We will illustrate the methods with a real-world data example of trauma care system evaluation.
