What is propensity score matching (PSM)?

From the Guidelines

Propensity matching is a statistical technique used to reduce bias in observational studies by creating comparable treatment and control groups, as evident in a systematic review of cancer and cancer surgical studies published in the Journal of the National Cancer Institute in 2017 1.

Definition and Purpose

Propensity matching works by pairing subjects from different groups who have similar characteristics or "propensities" to receive a particular treatment. The process involves calculating a propensity score for each subject based on their baseline characteristics, then matching subjects with similar scores across groups. This method helps researchers estimate treatment effects more accurately by mimicking randomization when a randomized controlled trial isn't possible.

Application and Benefits

For example, when studying a medication's effectiveness using existing medical records, propensity matching would pair patients who received the medication with similar patients who didn't, based on factors like age, gender, comorbidities, and disease severity. This approach controls for confounding variables that might otherwise skew results, though it can only account for measured characteristics and not unknown confounders.

• Key aspects of propensity matching include:
  • Calculating propensity scores based on baseline characteristics
  • Matching subjects with similar propensity scores across treatment and control groups
  • Controlling for confounding variables to estimate treatment effects more accurately
• The technique is particularly valuable in healthcare research, economics, and social sciences where randomized experiments may be impractical or unethical, as noted in the study published in the Journal of the National Cancer Institute 1.

Quality of Reporting and Guidelines

The systematic review evaluated the use and reporting quality of propensity score analysis in oncology studies and found that while the use of propensity score analysis has increased, there is substantial room for improvement in the quality of reporting, even in top journals 1. The study proposed reporting guidelines for propensity score analyses that are broadly applicable to different areas of medical research, which would allow for better evaluation and comparison across studies applying this approach. Therefore, it is crucial to adhere to these guidelines to ensure the validity and reliability of propensity matching in observational studies.

From the Research

Definition of Propensity Matching

• Propensity score matching (PSM) is a statistical method used to remove confounding bias from observational cohorts where randomization is not possible 2, 3, 4.
• It aims to equate treatment groups with respect to measured baseline covariates to achieve a comparison with reduced selection bias 4.
• PSM estimates the probability for a subject to receive a treatment conditional on a set of baseline characteristics (confounders) 2.

Purpose of Propensity Matching

• The primary purpose of PSM is to control for confounding by indication in observational studies 2.
• It allows for a more accurate estimation of the influence of treatment on outcome by accounting for imbalances in potential confounding variables between treatment groups 5.
• PSM can improve the quality of research and broaden the range of research opportunities by mimicking randomized controlled trials (RCTs) 4.

Implementation of Propensity Matching

• The process of PSM involves several steps, including identifying its feasibility, obtaining propensity scores, matching patients, assessing the success of matching, and applying statistical methodology using the propensity-matched data 4.
• Propensity scores are commonly estimated using logistic regression, and patients are matched based on similar propensity scores to balance confounders 2, 4.
• PSM can be used in various fields, including orthopedic surgery, anesthesia, and cardio-thoracic surgery, to analyze observational nonrandomized data 6, 3, 5.

Advantages and Limitations of Propensity Matching

• PSM offers certain advantages over traditional regression methods, including the ability to model the relationship between confounders and treatment assignment 2.
• However, PSM requires the removal of data and works primarily on binary treatments, which can be a limitation 3.
• Despite its limitations, PSM is a powerful tool for analyzing observational nonrandomized data and can increase the level of evidence of a study 5, 4.