Geographic variation in potentially preventable hospitalisations in Indonesia: a multilevel analysis using National Health Insurance Sample Data

ObjectivesIndonesian data on potentially preventable hospitalisations (PPH), a widely used indicator for measuring primary health care (PHC) quality, are lacking. This study aimed to describe Indonesias PPH rates and to quantify variation in PPH across districts/cities and their associations with individual sociodemographic characteristics. Design and settingCross-sectional study conducted in Indonesia using the 2023 National Health Insurance Sample Data. ParticipantsSample of 2.3 million individuals, representing 266.8 million national health insurance-Jaminan Kesehatan Nasional (JKN)-enrolees. Outcome measuresPPH overall and by type: vaccine-preventable, acute, and chronic. MethodsWe quantified crude and age-standardised rates per 10,000 JKN enrolees. We applied multilevel negative binomial regression and estimated: 1) median rate ratios (MRRs) to quantify variation in PPH across districts/cities, unadjusted and adjusted for sociodemographic characteristics, and 2) incidence rate ratios (IRRs) to quantify associations between PPH and sociodemographic characteristics. ResultsThere were 4.83 million overall PPHs in 2023, an overall rate of 180.9 per 10,000 JKN enrolees (acute: 94.8, chronic: 93.6, vaccine-preventable: 2.8). Overall PPH rates were generally highest in districts/cities in Java-Bali, Sulawesi, Sumatera, Kalimantan, lower in Nusa Tenggara, Maluku, and lowest in Papua. Variation in PPH across districts/cities was similar for all PPH types (MRRs overall: 1.62, vaccine-preventable: 1.83, acute: 1.72, chronic: 1.53) and persisted after adjusting for sociodemographic characteristics (MRRs of 1.52, 1.82, 1.61, 1.41, respectively). Higher PPH was strongly associated with older age (55-64 and 65-74 years, particularly chronic) and young age (0-14 years, vaccine-preventable, acute), modestly associated with being married or divorced (acute, chronic, overall), and being non-subsidised (all types), and weakly associated with sex (all types). ConclusionsPPHs varied across districts/cities as well as by sociodemographic characteristics. While PPH has potential for measuring PHC quality in Indonesia, context-specific interpretation is required as lower PPH may reflect limited accessibility to hospital-care rather than better PHC quality. Strengths and limitations of this study{Rightarrow} This is the first study to describe potentially preventable hospitalisations (PPH) in Indonesia, one of the most populous countries in the Asia-Pacific region and globally, using a large-scale sample from routine national health insurance (Jaminan Kesehatan Nasional or JKN, one of the worlds largest health insurance schemes) data. {Rightarrow}We applied two-level multilevel analysis to account for the hierarchical structure of the data (individuals nested within districts/cities). {Rightarrow}Our study did not capture all PPH in Indonesia as around 28% of hospitalisations are not through JKN scheme. {Rightarrow}Indonesia has not yet developed its own list of conditions for defining PPH. We modified lists of conditions from other countries, which may not fully reflect the Indonesian context.