Public reporting of home health care agencies' performance metrics, including patient satisfaction, care processes, and health outcomes, aims to inform customer decisions and encourage agencies to improve the quality of services. However, there is limited research that examines the heterogeneous performance of home health care agencies. The aim of this study was to analyze the performance of home health care agencies by identifying distinct subgroups of agencies with similar performance profiles and describing the relationships between agency characteristics and such subgroups. We propose a two-stage analytical approach employing unsupervised machine learning methods. First, clustering analysis is applied to performance measures, allowing the partitioning of agencies into homogeneous subgroups based on similarities in performance. Then, association rule mining is used to uncover the relationships between cluster assignments and agency characteristics. The two-stage analytical approach identified four clusters with significantly different performance profiles and agency characteristics: cost-efficient agencies with high patient satisfaction (Cluster 1), high-cost agencies with high-quality care (Cluster 2), urban agencies with low patient satisfaction (Cluster 3), and small agencies with low-quality care (Cluster 4). This study contributes to understanding agency performance in the U.S. home health care industry. By identifying distinct subgroups of agencies and understanding the factors influencing their performance, we can enhance home health care services' overall quality and effectiveness. Our study uncovered diverse performance profiles and associated characteristics among home health care agencies, highlighting the need for tailored strategies and targeted interventions to improve the quality of care across clusters. Health care administrators and policymakers should consider cluster-specific recommendations.
Read full abstract