Results from the Spatial Discontinuity Models.

How do patterns of socioeconomic inequality shape the risk of child fatality in urban areas? Research has shown that both intentional and unintentional child deaths are concentrated in areas of social disadvantage. Yet in densely populated cities, child fatality risk may not follow smooth spatial patterns and instead exhibit abrupt shifts across neighborhood boundaries. This study applies a dissimilarity-based Bayesian spatial conditional autoregressive (CAR) model to detect localized discontinuities in child mortality risk associated with structural inequality. Using a continuous dissimilarity function based on differences in Economic Hardship and Inequality (EHI) scores between adjacent census tracts, the model allows spatial smoothing to attenuate across sharp transitions. The model identified 413 spatial boundaries—termed social frontiers—where sharp structural discontinuities in EHI were associated with a 22% increase in the relative risk of child mortality, even after adjusting for racial segregation, concentrated disadvantage, residential mobility, and immigrant concentration. The significance of identifying neighborhoods characterized as social frontiers, where children may benefit from additional preventive interventions, is discussed in this context.