Propensity score matching is a statistical method that is often used to make inferences in observational studies. In recent years, there has been widespread use of the technique in the cardiothoracic surgery literature. However, the small sample size and the strong dependence of the treatment assignment on the baseline covariates that often characterize this setting make the evaluation using standard approaches challenging. We propose to use propensity score matching in combination with oversampling and replacement to face these issues. The idea behind the approach is to increase the initial sample size to eventually improve the statistical power that is needed to detect the effect of interest. In this study, we review the proposed approach in small sample size settings. We evaluate the method using Monte Carlo simulations, and we illustrate it using a real case study from the cardiac surgery literature. In some scenarios, we find some improvements regarding balance and bias reduction when each unit is matched with replacement with 2 or 3 units. Nevertheless, the observed benefits may still be unsatisfactory in small sample size settings.