A critical review of definitions of rural areas in Indonesia and implications for health workforce policy and research.

BACKGROUND: Choosing the appropriate definition of rural area is critical to ensuring health resources are carefully targeted to support the communities needing them most. This study aimed at reviewing various definitions and demonstrating how the application of different rural area definitions implies geographic doctor distribution to inform the development of a more fit-for-purpose rural area definition for health workforce research and policies. METHODS: We reviewed policy documents and literature to identify the rural area definitions in Indonesian health research and policies. First, we used the health policy triangle to critically summarize the contexts, contents, actors and process of developing the rural area definitions. Then, we compared each definition's strengths and weaknesses according to the norms of appropriate rural area definitions (i.e. explicit, meaningful, replicable, quantifiable and objective, derived from high-quality data and not frequently changed; had on-the-ground validity and clear boundaries). Finally, we validated the application of each definition to describe geographic distribution of doctors by estimating doctor-to-population ratios and the Theil-L decomposition indices using each definition as the unit of analysis. RESULTS: Three definitions were identified, all applied at different levels of geographic areas: "urban/rural" villages (Central Bureau of Statistics [CBS] definition), "remote/non-remote" health facilities (Ministry of Health [MoH] definition) and "less/more developed" districts (presidential/regulated definition). The CBS and presidential definitions are objective and derived from nationwide standardized calculations on high-quality data, whereas the MoH definition is more subjective, as it allows local government to self-nominate the facilities to be classified as remote. The CBS and presidential definition criteria considered key population determinants for doctor availability, such as population density and economic capacity, as well as geographic accessibility. Analysis of national doctor data showed that remote, less developed and rural areas (according to the respective definitions) had lower doctor-to-population ratios than their counterparts. In all definitions, the Theil-L-within ranged from 76 to 98%, indicating that inequality of doctor density between these districts was attributed mainly to within-group rather than between-group differences. Between 2011 and 2018, Theil-L-within decreased when calculated using the MoH and presidential definitions, but increased when the CBS definition was used. CONCLUSION: Comparing the content of off-the-shelf rural area definitions critically and how the distribution of health resource differs when analysed using different definitions is invaluable to inform the development of fit-for-purpose rural area definitions for future health policy.