Why to use propensity score in observational studies? Case study based on data from the Czech clinical database AHEAD 2006–09

Randomized clinical trials represent the gold standard of the evidence based medicine research; nevertheless they may not always be feasible or ethical and the researchers have to rely on observational studies or research databases. However, obtaining reliable results from these studies requires the elimination of potential influence of confounding factors. Fortunately, several statistical methods capable of identifying and reducing the impact of confounding factors exist. One of them is the propensity score which has been frequently used in recent times to estimate relevant clinical effects adjusted for given confounders. This work aims to provide a concise and practical guide to propensity scores by means of an easily understandable case study. The case study is focused on gender differences in mortality rates of patients with acute heart failure in the Czech research database AHEAD (Acute Heart Failure Database).