Five Steps to Successfully Implement and Evaluate Propensity Score Matching in Clinical Research Studies.

Area of Science:

• Clinical Research
• Biostatistics
• Anesthesia Research

Background:

• Nonrandomized retrospective data is increasingly available for clinical research.
• Comparability of treatment groups in retrospective data is a challenge due to confounding variables.
• Randomized controlled trials (RCTs) are the gold standard for evidence but are not always feasible.

Purpose of the Study:

• To present a 5-step methodology for implementing propensity score matching (PSM) in anesthesia research.
• To demonstrate how PSM can reduce selection bias and confounding in observational studies.
• To enhance the quality and generalizability of research using retrospective data.

Main Methods:

• Propensity score matching (PSM) involves equating treatment groups based on measured baseline covariates.
• Key steps include assessing feasibility, obtaining propensity scores via logistic regression, matching patients, assessing balance, and applying statistical analysis.
• An example using the 2015 Pediatric National Surgical Quality Improvement Program database is provided.

Main Results:

• PSM can create "apples to apples" comparisons, mimicking RCTs with observational data.
• Successful PSM leads to balanced treatment groups, reducing bias from confounding variables.
• This methodology can improve the evaluation of treatment effects and increase the level of evidence.

Conclusions:

• PSM is a valuable statistical methodology for improving anesthesia research quality.
• It offers a more intuitive approach to analyzing treatment effects compared to regression models alone.
• Implementing PSM can strengthen the validity and generalizability of findings from observational studies.