Ranking hospitals' burn care capacity using cluster analysis on open government data. (August 2021)

• Record Type: Journal Article
• Authors: Ho, Hui Yan; Chuang, Sheuwen; Dai, Niann-Tzyy; Cheng, Chia-Hsin; Kao, Wei-Fong
• Is Part Of: Computer methods and programs in biomedicine. Volume 207(2021)
• Publication Date: 2021-08
• Subjects: Cluster analysis -- Hierarchical clustering -- Burn mass casualty incident -- Formosa Fun Coast Dust Explosion -- Mass casualty distribution; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2021.106166
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: Use of hierarchical clustering to differentiate hospitals' burn care capacity with open government data is a novel and alternative solution to traditional accreditation processes. Data selection and streamlining process of 116 open data variables determined 42 variables for cluster analysis. The internal and external clustering validation identified four clusters via four influential variables for Taiwanese hospitals. This approach can be used in other countries or contexts as an alternative to traditional accreditation methods for ranking hospital capacities in burn care or other domains of care. Abstract: Background and objective: To deal with burn mass casualty incidents (BMCIs), various countries have established national or regional BMCI emergency response plans (ERPs). A burn care capacity ranking model for hospitals can play an integral role in ERPs by providing essential information to emergency medical services for distributing and handling mass burn patients. Ranking models vary across countries and contexts. However, Taiwan has had no such model. The study aims to develop a ranking model for classifying hospitals' burn care capacity in preparation for the development of a national BMCI ERP. Methods: Multiple methods were adopted. An expert panel provided consultations on data selection and clustering validation. Data on 116 variables from 535 hospitals were collected via open data platforms under the Ministry of Health and Welfare. Data selection and streamlining was conducted to determine 42 variables for cluster analysis. SAS 9.4 was used to analyze the data set -via a hierarchical cluster analysis using Ward's method, followed by a tree-based model analysis to identify the criteria for each cluster. Both internal and external cluster validation were performed. Results: Four clusters of burn care capacity were determined to be a suitable number of clusters. All hospitals were arranged into capacity levels accordingly. Results of the Kruskal-Wallis test showed that the difference between clusters were significant. Tree-based model analysis revealed four determining variables, among which the refined level of emergency care responsibility hospital was found to be most influential on the clustering process. Responses from the questionnaire were used as an external validation tool to corroborate with the cluster analysis results. Conclusion: The use of open government data and cluster analysis was suitable for developing a ranking model to determine hospitals' burn care capacity levels in Taiwan. The proposed ranking model can be used to develop a BMCI emergency response plan and can also serve as a reference for using cluster analysis with open government data to rank care capacity or quality in other domains.