Propensity Scores: Confounder Adjustment When Comparing Nonrandomized Groups in Orthopaedic Surgery

Dirk R. Larson, Isabella Zaniletti, David G. Lewallen, Daniel J. Berry, Hilal Maradit Kremers

Research output: Contribution to journalReview articlepeer-review


Many studies in arthroplasty research are based on nonrandomized, retrospective, registry-based cohorts. In these types of studies, patients belonging to different treatment or exposure groups often differ with respect to patient characteristics, medical histories, surgical indications, or other factors. Consequently, comparisons of nonrandomized groups are often subject to treatment selection bias and confounding. Propensity scores can be used to balance cohort characteristics, thus helping to minimize potential bias and confounding. This article explains how propensity scores are created and describes multiple ways in which they can be applied in the analysis of nonrandomized studies. Please visit the following ( for a video that explains the highlights of the paper in practical terms.

Original languageEnglish (US)
Pages (from-to)622-626
Number of pages5
JournalJournal of Arthroplasty
Issue number4
StatePublished - Apr 2023


  • bias
  • confounding
  • inverse probability of treatment
  • propensity score
  • statistics
  • total joint arthroplasty

ASJC Scopus subject areas

  • Orthopedics and Sports Medicine


Dive into the research topics of 'Propensity Scores: Confounder Adjustment When Comparing Nonrandomized Groups in Orthopaedic Surgery'. Together they form a unique fingerprint.

Cite this