New Open Source Analytic Resource: R Package {polymatching} Enables Triplet Matching for Injury and Trauma Research

Comparing outcomes across three or more treatment groups—such as trauma center levels—presents methodological challenges in observational studies. Traditional propensity score matching is limited to binary comparisons and does not easily extend to multi-arm settings. A newly released R package, {polymatching}, from former SAVIR President Henry Xiang and team helps address this gap.

 CRAN link: https://cran.r-project.org/web/packages/polymatching/

 The package implements a triplet-matching algorithm that forms optimal 1:1:1 matched sets across three treatment groups, enabling more rigorous and transparent causal inference. This method was developed through collaboration between Dr. Henry Xiang (Nationwide Children’s Hospital and The Ohio State University), Dr. Bo Lu (OSU College of Public Health), and former PhD student Giovanni Nattino. It was first applied in studies evaluating trauma center effectiveness, including a national analysis of emergency department mortality for severely injured patients. 

 The methodology is detailed in the team’s Journal of the American Statistical Association publication and earlier work in Epidemiology (2016;27:624–632) that introduced the foundational design and sensitivity analysis. Now available on CRAN, {polymatching} provides a user-friendly interface for applying this advanced method in public health and injury research.

 Researchers interested in using or extending the package are encouraged to connect with the development team by contacting Henry Xiang, MD, MPH, PhD, MBA at Henry.Xiang@NationwideChildrens.org.