Propensity score to reduce bias in observational studies : the case of enteral sedation in ICU patients

To reduce structural biases of observational studies, the use of propensity score (PS), a subject's probability of receiving a specific treatment conditional on the observed covariates, has been proposed the use of the probability that a subject would have been treated allows to adjoust the estimation of the treatment effect, creating a "quasi randomized" trial. The three techniques in using PS (matching, stratification and regression adjustment) were here applied to compare intravenous and enteral approach for sedative therapy in ICU.