Propensity Score Matching: A Robust Method for Observational Outcome Analysis in Cardiovascular Research (Motivated by the BIA-LM Registry Study on Diabetes and Left Main PCI Outcomes by Kralisz et al.)

Propensity score matching has become an essential tool for strengthening causal inference in observational research, particularly when randomized trials are unavailable or impractical. This report examines the application of propensity score matching in the analysis of long-term outcomes following percutaneous revascularization of the unprotected left main coronary artery, using data from the BIA-LM registry. By balancing baseline characteristics between patients with and without diabetes, the analysis minimized confounding and enabled more reliable comparisons of survival outcomes. Key tables and figures, including baseline characteristic comparisons, standardized mean difference plots, propensity score distributions, and subgroup-specific forest plots, demonstrated the effectiveness of the matching process. Although initial unmatched analyses suggested higher mortality among diabetic patients, matched analyses revealed no statistically significant differences, highlighting the critical role of adjustment for baseline risk profiles. Strengths of propensity score matching include its flexibility, transparency, and practical utility in clinical research, though limitations such as reliance on measured covariates and potential sample size reductions must be carefully considered. Future directions include integrating machine learning methods and expanding validation efforts. This case study illustrates how rigorous propensity score matching can yield clinically relevant, transparent, and reproducible insights in complex cardiovascular populations.