Abstract
<title>Abstract</title> <p>Background Equitable access to paediatric emergency care (EC) is fundamental to reducing child mortality, yet its conventional measures rely largely on travel time, overlooking barriers in availability, affordability, accommodation and acceptability that shape care seeking. Existing composite indices also often aggregate access dimensions additively, allowing well-performing components to mask critical deficits, and rarely account for local variation in barrier importance. We address these gaps by developing a high-resolution, multidimensional composite index of access to inpatient paediatric EC in Kenya and testing its relevance for child survival and policy targeting. Methodology: Drawing on the Penchansky & Thomas framework, we quantified five access dimensions - availability, geographic accessibility, affordability, accommodation, acceptability - as continuous 1 km × 1 km surfaces using national health facility censuses, demographic surveys, and spatial datasets. Availability was modelled using an Enhanced Two-Step Floating Catchment Area method; geographic accessibility via least-cost path modelling; affordability, accommodation and acceptability via Bayesian geostatistical methods. Spatially adaptive weights were derived using geographically weighted principal component analysis, and the composite index constructed using ordered weighted averaging (OWA), a non-compensatory method penalising severe deficits. OWA was compared with arithmetic and geometric means and validated against under-five mortality using quantile regression as a plausibility check. Results Access to inpatient paediatric EC was unevenly distributed (median composite access index 0.45, IQR: 0.35,0.57). Although 70% of children resided within 1-hour of a facility offering inpatient paediatric EC services, this threshold misclassified access: 9/20 counties within 1-hour fell into the lowest composite quartiles due to deficits in other dimensions. OWA produced systematically lower scores than arithmetic/geometric means in unbalanced areas, avoiding overestimation from compensatory effects. In quantile regression, OWA-based access showed a protective association with under-five mortality, strongest at higher mortality quantiles, whereas compensatory methods produced slopes near zero or inconsistent across quantiles. Conclusions Access barriers cluster and combine differently across space, and these patterns matter for child survival. The index and underlying approach offer a practical advance beyond travel time alone, providing a transferable way to measure access and target investments in Kenya and other sub-Saharan African settings where EC resources are scarce and inequities are wide.</p>