Three handy tips and a practical guide to improve your propensity score models

What is it about?

Real-world data are increasingly available to investigate ‘real-world’ safety and efficacy. But in contrast to treatments in randomized clinical trials, treatments in (real world) observational studies are not randomly allocated. Therefore, confounding by indication may occur. This means that differences in patient and disease characteristics may influence the choice of treatment as well as the response to treatment. A popular statistical method to adjust for this type of bias is the use of propensity sores (PS). A PS is a score between 0 and 1 that reflects the likelihood per patient to receive one of the treatment categories of interest, conditional on a set of variables. At least in theory, in patients with similar PS, the treatment prescribed will be independent of this set of variables. But researchers using PS sometimes fail to recognize important methodological flaws which can lead to spurious conclusions. In this viewpoint we will discuss the most commonly encountered flaws and provide a stepwise description on the estimation and use of PS, such that in future publications these flaws can be avoided.

Featured Image