LAUSR

The Sustainable Development Goals brought attention to the lack of geographically disaggregated data in low- and middle-income (LMIC) countries as national health surveys are generally not designed to provide estimates beyond the first administrative level. Advances on small area estimation methods have allowed for more granular estimates which support targeting and implementation of local health interventions. This study aims to summarize literature on the advances in spatial modelling of reproductive, maternal, newborn and child health (RMNCH). We carried out a comprehensive literature search in five databases to identify studies that: 1) performed small area estimation; 2) were carried out in LMIC settings; 2) focused on RMNCH outcomes; 3) were based on national health surveys. Studies modelling travel time were not considered. We included 70 out of 4266 studies. The state-of-art approach to small area estimation has been through a Bayesian framework with INLA for administrative level or up to 5x5 km estimates. Malaria and child mortality outcomes are predominant, closely followed by anthropometry, vaccination and contraceptives use. Spatial modelling has become a suitable approach for local estimates where data at finer resolutions is not available. However, there is not a consensus on how to report and interpret uncertainties. Local estimates are essential to identify and act on the most disadvantaged areas. Properly communicating uncertainty is a big challenge for researchers and they must accompany estimates to avoid misinterpretation.