Impact of exposure measurement error in multi-pollutant models: a simulation study in the time-series framework

Abstract

BACKGROUND AND AIM Exposure measurement error in air pollutants can lead to biased health effect estimates and effect transfer from more poorly to better measured exposures. While some studies have estimated the impact of error in single-exposure models we aimed to quantify the effect of measurement error in multi-pollutant models, specifically in time-series analysis of PM₂.₅, NO₂, and mortality using simulations, under various plausible scenarios for exposure errors. METHODS Measurement error was defined as the difference between ambient concentrations and personal exposure from outdoor sources. Error-free exposures with their consequent health outcome and error-prone exposures of various error types (classical/Berkson/mixture) were generated using information from the literature. Bias was quantified as the relative difference in effect estimates of the error-free and error-prone exposures. RESULTS NO₂ was assumed to be more prone to error than PM₂.₂, based on separate work, and, thus, its mortality effect estimates were more affected by measurement error bias. The observed bias was up to 37% towards the null, when low ratios of the true exposure variance over the error variance were assumed. Higher ratios resulted in smaller, but still substantial health effect underestimation, up to 19% for both pollutants. Significant effect transfer was observed from NO₂ to PM₂.₅ as the general underestimation due to measurement error was worsened for the former and lessened for the latter compared to single-pollutant model estimates. CONCLUSIONS Our results indicated an underestimation of true independent health effects of multiple exposures due to measurement error. Effect transfer showed that less precise measurements for one pollutant yield more bias, while the co-pollutant associations were found closer to the true. Using error parameter information in future epidemiological studies should provide more accurate concentration-response functions for health impact assessment. KEYWORDS Air pollution, Measurement error, Mixture error, Effect transfer, Simulations