Propensity Score Methods and Target Trial Emulation in Orthopaedic Surgery: A Practical Guide to Causal Inference from Observational Data.

» The propensity score is a patient's modeled probability of receiving a treatment given measured baseline characteristics (covariates). Matching patients with similar scores balances those covariates between groups, approximating the comparability achieved by randomization. » Target trial emulation specifies the protocol of the ideal randomized trial and then reproduces it in observational data, enforcing an explicit eligibility window, time zero (the single point at which eligibility, treatment assignment, and the start of follow-up coincide), and analysis plan. » Regression adjusts for confounders within the outcome model, whereas propensity score matching balances those confounders before outcomes are examined. Propensity matching is preferred for rare outcomes or many confounders, where regression tends to overfit. » Neither propensity matching nor target trial emulation addresses unmeasured confounding, and both lose validity in small single-center cohorts with limited overlap (few treated and untreated patients who resemble one another closely enough to be fairly compared); a randomized trial remains the reference standard when feasible. » When appraising studies using propensity matching or target trial emulation, look for 3 key features: (1) Postmatching covariate balance: Are the treatment groups well balanced after matching? (standardized mean difference <0.10). (2) Positivity: Is there enough overlap in patient characteristics that treated and untreated patients can be fairly compared? (3) Unambiguous "time zero": Did follow-up begin at the same well-defined starting point for all patients?