Data-driven estimation of the onset of heat-related mortality in five Indian cities
En palabras de los autores
Background: Extreme heat is an increasingly pressing public health threat, particularly in South Asia. Yet the temperatures at which mortality risk begins to increase in Indian cities remain poorly characterized. Operational heat wave definitions are based on meteorological criteria and may not reflect epidemiologically relevant temperature thresholds. Widely used distributed lag non-linear models are designed to flexibly estimate expected mortality as a function of temperature, but do not directly target the onset of excess risk from heat, a quantity central to early warning and planning. Methods: We conducted a multi-city time-series study of daily mortality and temperature data in Ahmedabad (2001-2010), Chennai (2012-2017), Delhi (2011-2015), Kolkata (2011-2018), and Mumbai (2011-2015). We developed a data-driven structured modeling framework that permits likelihood-based inference for city-specific temperatures at which heat-related mortality first exceeds seasonal background levels, a parameter that is challenging to estimate and perform statistical inference from short and noisy series. Using these health-based thresholds, we also estimated heat-associated excess deaths. Results: In all five cities, the onset of excess mortality occurred at temperatures substantially lower than those implied by current operational heat wave definitions. The proposed framework yielded stable and interpretable threshold estimates and confidence intervals across heterogeneous climates and time-series lengths and identified a larger burden of heat-associated mortality than would be captured using standard meteorological criteria. Conclusions: Data-driven estimation of mortality-based heat thresholds provides epidemiologi- cally relevant targets for early warning systems and heat action plans. Reliance on meteorological definitions alone may delay intervention and underestimate population risk in rapidly warming regions.
Apareció: miércoles, 11 de noviembre. medRxiv. Preprint, todavía sin revisión por pares.