Grouping of Regencies/Cities in Indonesia Based on National Health Insurance (JKN) Participants with the Ensemble ROCK Approach

Abstract

Health is a fundamental human need, and the National Health Insurance (JKN) program was established in Indonesia to provide equitable access to healthcare services for all citizens. Despite its implementation, disparities remain across regencies/cities, necessitating a comprehensive mapping of JKN participant profiles. This study aims to group 34 regencies/cities in Indonesia based on the characteristics of JKN participants, utilizing numerical and categorical data clustering. The Ensemble Robust Clustering using links (ROCK) method was employed, combining hierarchical clustering for numerical data and the ROCK method for categorical data. The study analyzed data comprising eight numerical variables (age, household size, household total expend, expend healthcare, tobacco expend, ATP, WTP, and expend insurance) and six categorical variables (living area, sex, education, reasons for joining JKN, ATP, WTP). Numerical clustering through single linkage yielded four clusters, while categorical clustering with the ROCK method at a threshold value of 0.2 produced three groups. The final ROCK ensemble analysis integrated these results, forming three quality-based clusters: low, medium, and high. Key findings revealed distinct socio-economic and demographic patterns among the clusters. For instance, the low-quality group exhibited lower household expenditures and healthcare spending, while the high-quality group had higher averages across these variables. Insights from this study can guide policy-makers in prioritizing healthcare resources and addressing regional disparities in JKN implementation.