Geospatial patterns and a two-year comparison of outpatient hospital utilization in Indonesia: An ecological analysis of 38 provinces, 2021–2024

Abstract

Provincial disparities in hospital outpatient utilization remain a concern in Indonesia. The aim of this study was to examine the geographic distribution and provincial correlates of hospital outpatient utilization and compare utilization between 2023 and 2024. Official aggregate data were analyzed for all 38 provinces in 2024 and 34 provinces with comparable 2023–2024 data. Total utilization was defined as the sum of public- and private-hospital outpatient utilization percentages, rather than the unduplicated proportion of residents receiving care. Analyses included paired comparisons, multivariable and Bayesian regression, and global and local spatial autocorrelation statistics. In 2024, median total utilization was 19.01% (interquartile range (IQR): 15.36–23.37; range: 7.85–31.19). Among comparable provinces, utilization was higher in 24 provinces and lower in 10 provinces in 2024 than in 2023. The mean difference was 0.91 percentage points (95% confidence interval (CI): −0.22 to 2.04; paired t-test p=0.110), whereas the median difference was 1.25 percentage points (Wilcoxon p=0.026). Previous-year utilization showed the strongest bivariate correlation with 2024 utilization (r=0.872). In the primary multivariable model, prior hospital facility preference was positively associated with utilization (b=1.28; 95%CI: 0.70–1.86), while total insurance coverage was not significantly associated. Bayesian estimates supported the relative ordering of adjusted associations. Total utilization showed no significant global spatial autocorrelation (Moran's I=0.027; p=0.600), whereas public-hospital share (I=0.503; p<0.001) and poverty (I=0.629; p<0.001) were spatially clustered. Local analyses identified high public-hospital dependence and poverty clusters in the Papua region and lower public-hospital dependence in Java. Substantial provincial disparities therefore persist, highlighting the need for context-specific investigation of healthcare access, service availability and public-sector dependence. Ecological associations should not be interpreted as individual-level or causal effects.