Informed Clustering of Donation Service Areas for Improved Benchmarking

Abstract

The donation community continuously strives to collaborate and share effective practices to further the mission of saving and healing lives. Donation service areas in which the Organ Procurement Organizations (OPOs) work are multifaceted in their demographics, inciting the Organ Procurement and Transplantation Network to consider a more holistic and objective measure of similarity rather than the size of population alone or locational proximity alone. This would allow OPOs, as a part of their quality improvement efforts, to learn from and mentor other organizations that are dealing with similar challenges. By incorporating multiple informative characteristics together, we can distinguish those likenesses only revealed by taking into account multiple factors simultaneously. We used statistical approaches that take many characteristics of interest describing a donation service area and purposely excluded performance measures that an OPO may be able to influence by their own practices. Unsupervised learning methods combined the original characteristics into a smaller number of new variables, eliminating correlation and overlap in information from the original characteristics, and clustered donation service areas based on the general characteristics and population of the area. This analysis is a first step in providing a different perspective for OPOs to learn from other organizations that may face similar challenges, as well as to share best practices and open new lines of communication.