A Modelling Framework for Estimating Underlying Lagged Effects in Aggregated Health Data

Quantifying the temporally lagged effects of environmental hazards on health risk is key in understanding the impacts of climate change on human health. We present a simple but flexible and practical statistical modelling framework to capture the temporally distributed effects on environmental factors on aggregated health count data. The framework was designed to specifically allow use temporally aggregated health data available on a coarse resolution (such as weekly), to uncover the distributed effects of hazards on a higher temporal resolution (such as daily). The aim is to enable researchers to make use of open‐access but low temporal‐resolution health data to uncover health risks that would otherwise require higher resolution but inaccessible data. We focus on the example of ambient temperature and how it impacts human mortality. We use simulation experiments and implementation to 15 open source mortality data from various cities and countries, to illustrate the functionality and limitations of the approach but to also assess the efficacy of the approach on real‐life data containing outliers and confounding factors. We illustrate how practical implementation in the R package mgcv enables very flexible model configurations.