Propensity Score Matching: A Robust Method for Observational Outcome Analysis in Cancer Epidemiology (Motivated by the Study on Metabolic Obesity Phenotypes and Prostate Cancer Risk by Wang et al.)

Propensity score matching in cancer epidemiology has become an essential methodological tool for improving causal inference using real-world data, particularly where randomized trials are not feasible. This report explores its application in the context of prostate cancer, drawing on a recent study examining the association between metabolic obesity phenotypes and prostate cancer risk. The investigators used a structured propensity score matching approach to control for confounding between 209 men with biopsy-confirmed prostate cancer and 209 controls from routine health screenings. Matching was performed on key clinical and demographic variables including age, ethnicity, and marital status, using a nearest neighbor approach with a caliper width of 0.05. Post-matching comparisons revealed improved covariate balance, with standardized mean differences approaching accepted thresholds. Logistic regression analysis demonstrated that hyperglycemia and elevated triglycerides were independently associated with increased prostate cancer risk. The report includes diagnostic tables and visualizations—such as Love plots and propensity score distribution curves—to illustrate matching performance. Strengths, limitations, and future directions for improving reproducibility and causal inference using advanced analytics are also discussed. This case highlights the role of propensity score matching in strengthening observational research when randomization is impractical.