Comparison of air pollution mortality effect estimates using different long-term exposure assessment modelling...

Area of Science:

• Environmental epidemiology
• Air pollution assessment
• Public health research

Background:

• Epidemiological studies assessing long-term ambient air pollution exposure use diverse methods.
• The impact of different exposure models on health effect estimates remains unclear.
• This study addresses the variability in air pollution mortality estimates.

Purpose of the Study:

• To compare mortality effect estimates using various exposure assessment methods for black carbon (BC), nitrogen dioxide (NO2), ultrafine particles (UFP), and particulate matter <2.5 microm (PM2.5).
• To evaluate the influence of different modeling and monitoring strategies on health outcomes.

Main Methods:

• Applied eight annual average air pollution exposure models to a Dutch national cohort (10.7 million adults aged ≥30 years).
• Evaluated dispersion and land-use regression models using mobile and fixed-site monitoring.
• Estimated hazard ratios (HR) for natural and cause-specific mortality via Cox proportional hazards models (2013-2019).

Main Results:

• Exposure estimates across models were highly correlated.
• Mortality effect estimates showed consistent directions but substantial magnitude differences (e.g., HR for BC ranged from 1.01 to 1.09 per 1 microg/m3).
• No consistent differences emerged between deterministic and empirical, fixed-site, and mobile models.

Conclusions:

• While different exposure models consistently indicated associations between air pollution and mortality, hazard ratios varied significantly (up to a ratio of 1.27).
• Discrepancies in exposure assessment methods likely contribute to the heterogeneity observed in systematic reviews of epidemiological studies.