Evaluation of Standard Use of Intravenous Infusion in 25 General Hospitals in Shanxi Province Based on Factor Analysis and Cluster Analysis

Lin Chen¹Song Wang¹Jinju Duan²Rong Yu³Donghong Yin^2*Ruigang Hou^2*

• ¹Shanxi Medical University, Taiyuan, China
• ²Second Hospital of Shanxi Medical University, Taiyuan, China
• ³Shanxi Cardiovascular Hospital, Taiyuan, China

Introduction: Intravenous (IV) infusion is overused in Chinese hospitals, and a tool to appraise its appropriateness quantitatively and qualitatively is lacking.This study aimed to develop a multidimensional evaluation framework for assessing the appropriateness of IV infusion use in general hospitals using factor and cluster analyses. Methods: We conducted a multicentre retrospective study using data from 25 hospitals and a total of 2,064 cases were analyzed. Stratified proportional random sampling was used to select medical records, and a dedicated intravenous infusion survey form was completed for each patient. Factor analysis and cluster analysis were employed to comprehensively assess the ranking and classification of standardized intravenous infusion use among these hospitals. The Kruskal-Wallis test was used to compare disparities in performance characteristics across clusters. Nine IV infusion-related indicators were selected based on national quality control guidelines. Principal component factor analysis was used to extract common factors, and K-means cluster analysis was applied to categorize hospitals into performance tiers. Statistical comparisons were performed to validate inter-group differences. Results: The number of prescription items was classified according to ATC codes, with the top three being anti-infectives (J01), nutritional preparations (V06), and vitamins (A11). Two factors explained 81.9 % of variance: “intensity” (high loadings on DOT, LOT, volume, bottles) and “penetration” (utilization and bed-day coverage). K-means clustering classified the 25 hospitals into four categories. “Excellent” (24.00%), “good” (20.00%), “Middle”(2.00%), and “Inferior” (48.00%). Inter-group comparisons showed no statistically significant difference in the proportion of intravenous infusion orders (P = 0.131), while the other eight indicators differed significantly (P < 0.01). These results objectively reflect the distinct characteristics of intravenous infusion practices among hospitals in Shanxi Province. Discussion: The factor analysis incorporating nine evaluation indicators can serve as a method to assess the standardization of intravenous infusion in hospitals. The cluster analysis reveals a discernible pattern in the combination of intravenous-infusion -related indicators across hospitals. This study presents a practical and scalable tool for evaluating IV infusion appropriateness in hospitals. The proposed framework supports targeted quality improvement interventions and aligns with national goals for rational drug use.