Data-driven estimation of the onset of heat-related mortality in five Indian cities
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
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 epidemiologically 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.
Key messages
-
We estimated the temperature at which mortality risk begins to increase above seasonal background levels in warm-climate settings and compared these estimates with current operational heat wave definitions in five Indian cities.
-
Using a structured modeling approach motivated by observed patterns in the data, we found that mortality risk began to increase at temperatures below current operational heat wave thresholds, with higher-intensity heat events associated with a larger share of excess mortality.
-
These findings suggest that epidemiologically derived thresholds may complement existing meteorological definitions in characterizing heat-related risk and setting triggers for action plans and warning systems.