[Win ratio in randomized clinical trials: a key tool for complex endpoints and their clinical interpretation]

Randomized controlled clinical trials (RCTs) are a fundamental tool for evaluating various intervention strategies. Their design is a critical aspect that defines the validity of their conclusions. Of the numerous factors to be considered, the endpoint is of primary importance. It represents the effect of the intervention on the outcome of the clinical entity considered. Mortality, from any cause or disease-specific, complications of a certain severity, and hospitalizations, are clinically unquestionable. When the statistical method is time to event (Cox model), only the first episode is considered. If this is hospitalization and the subsequent, more significant event is death, it will not be considered. Furthermore, the Cox model does not allow for the inclusion of repeated episodes. However, why is it necessary to include multiple components in the endpoint? The objective is for the trial to achieve sufficient power, avoiding false negatives. In the case of rare diseases, the number of patients can be increased or the observation period extended, with the resulting complications and economic implications. The win ratio (WR) method circumvents these drawbacks, as it prioritizes events in a hierarchical order within a combined outcome, also allowing for the computation of repeated episodes. With the WR, it is also possible to include some surrogate for major complications. This manuscript analyzes the clinical interpretation of the effect of a pharmacological intervention expressed by the WR in a fictitious RCT, describing the conceptual mechanics of the statistical procedure.