Classification of Resilience of Turkish Health System to Extraordinary Health Crises at Provincial Level

Abstract

This study aimed to classify the provincial-level resilience of the Turkish health system using K-Means and Partitioning Around Medoids (PAM) clustering methods, utilizing data from the Ministry of Health's 2022 Health Statistics Yearbook. Prior to clustering analysis, the 15 variables used to assess health system resilience were reduced to 9 through Principal Component Analysis (PCA). Clustering analyses were subsequently performed on these remaining variables using the PAM and K-Means methods. The health systems of 81 provinces were classified into 3 distinct clusters based on their resilience. The PAM method was found to yield more optimal results compared to the K-Means method. According to the PAM method, provinces assigned to Cluster 3 demonstrated superior health system resilience compared to those in the other clusters. Based on the average values of the variables, the clusters were ranked in descending order of resilience: Cluster 3, Cluster 2, and Cluster 1. Significant disparities were observed both between and within clusters, primarily attributed to the uneven distribution of resources relative to population. Therefore, it is recommended that health system capacity be strengthened, using the highest-performing cluster as a benchmark. This approach can facilitate the construction of a more resilient and equitable provincial health system regarding service delivery supply, ultimately contributing to the establishment of a stronger national health system built upon strengthened provincial foundations.