Evaluating medical service performance of hospitals in Sichuan Province, China: exploratory factor analysis and hierarchical clustering analysis based on diagnosis-related groups

Abstract

Objective: This study aims to evaluate hospital medical service performance in Sichuan Province, China.

Methods: A total of 306 secondary and tertiary general hospitals were included in the analysis. A comprehensive evaluation model was developed using exploratory factor analysis (EFA) based on diagnosis-related groups (DGR_S) indicators to assess medical service performance. Indicators were determined within the Donabedian structure-process-outcome (SPO) framework. Hierarchical clustering analysis (HCA) was applied to categorize hospitals into performance clusters, and the Kruskal-Wallis H test was used to compare disparities in performance characteristics across clusters.

Results: The comprehensive evaluation revealed that all top 10 hospitals were tertiary general hospitals (TGHs), with 40.00% located in the Chengdu region. Conversely, the bottom 10 hospitals were exclusively secondary general hospitals (SGHs), predominantly concentrated in northeastern Sichuan. TGHs were classified into three clusters: "Excellent" (30.83%), "Middle" (57.14%), and "Inferior" (12.03%), while SGHs were categorized as "Excellent" (26.01%), "Middle" (69.94%), and "Inferior" (4.05%). For TGHs, the "Excellent" cluster displayed significantly higher performance in case-mix index (CMI), number of DRG_S (ND), total weight (TW), and time efficiency index (TEI) compared to the "Middle" and "Inferior" clusters, but performed worst in cost efficiency index (CEI) and mortality of middle and low-risk group cases (MMLRG). For SGHs, "Excellent" cluster hospitals significantly outperformed others in ND and TW, while the "Inferior" cluster performed best in CMI but alarmingly worst in MMLRG.

Conclusion: Significant regional and hierarchical disparities in medical service performance were observed across Sichuan Province, with Chengdu region demonstrating optimal performance. For TGHs, hospitals in the "Inferior" cluster are recommended to enhance their medical ability and efficiency compared to those in the "Excellent" cluster. Conversely, hospitals in the "Excellent" cluster should focus on controlling medical costs compared to those in the "Inferior" cluster. For SGHs, hospitals in the "Inferior" cluster should concentrate on improving medical security and ensuring patient safety compared to those in the "Middle" and "Excellent" clusters.

Keywords: DRGS; Donabedian; exploratory factor analysis; hierarchical clustering analysis; hospital; performance evaluation.

Figure 1

Geographic and hospital-level distributions of the 306 study hospitals.

Figure 2

Determination of the optimal number of clusters using SC. TGH, tertiary general hospital; SGH, secondary general hospital; SC, Silhouette Coefficient. Panel (a) corresponds to the analysis of tertiary general hospitals (TGH), showing how the Silhouette Coefficient (SC) changes with the number of clusters (K). Panel (b) corresponds to the analysis of secondary general hospitals (SGH), illustrating the SC-K relationship for this hospital level.