Heterogeneity in the Urban Scaling of Premature Mortality by City Definition in the United States

With more than half of the world’s population currently living in cities, there is a need to understand how city-level factors, such as city size, can influence health in cities. However, there are numerous ways to define and delineate city boundaries. We explored whether years of life lost (YLL), a measure of premature mortality, scales with population size in the United States (US) from 2010 to 2015 and examined whether these scaling patterns vary by city definition, using eight different definitions. Across all city definitions, YLL displayed a sublinear scaling behavior: larger cities have relatively lower premature mortality. The scaling coefficients were similar across all city definitions, varying from 0.94 to 0.98. In addition, while the proportion of variance explained by population size differed, it was high in all cases, ranging from 99% (New England City & Town Areas) to 90% (Census Designated Places). By examining the residuals, we found that cities in the Southern Appalachian region of the US consistently experienced higher rates of premature mortality than expected from the model across all city definitions. This study shows that, while the way we define cities changes our estimates of how much premature mortality scales with city size, these differences are not large. This study can help guide future research examining the urban scaling of health outcomes across different city definitions.