A comparison of exposure uncertainty propagation models used in epidemiological studies

Abstract

Abstract In the field of ecological epidemiological studies, accurate estimation of the long-term health effects of air pollution is crucial. Two-step models that involve exposure assessment and health effects estimation are often used for this purpose. However, the accuracy of exposure assessment is uncertain and may not accurately reflect true exposure. Despite several proposed methods to manage this uncertainty, the impact of different approaches on air pollution inferences remains uncertain. In this study, we conduct a simulation study to compare the inferences of air pollution impact from various exposure uncertainty propagation models while investigating their health effects. The results suggest that the Without-uncertainty model and the Multi-set method are preferable to the prior method and pollution-health jointly model (without cut-off). Moreover, a case study further reinforces the evidence of a link between mortality and PM2.5 concentrations, showing that an increase of 1 μg⋅m−3 in PM2.5 concentration is likely to increase all-cause deaths in Scotland by 4.51% [95% credible interval (CI), 3.42%, 5.49%] to 7.51% (95% CI, 6.28%, 8.80%). These findings have important implications for policymakers and public health officials seeking to mitigate the harmful effects of air pollution.