The use of single and multiple pollutant models in health impact assessment of air pollution: What difference does it make and where?

Background Ambient air pollution is the largest environmental risk factor for morbidity and premature mortality worldwide. Studies that have quantified the health effects of air pollution typically use single-pollutant models and do not account for concurrent population exposure to other air pollutants. We compare and provide estimates of single and multi-pollutant attributable mortality for 824 European cites. Methods We conducted quantitative single and multi-pollutant health impact assessment (HIA) to estimate the attributable mortality to PM 2.5 and NO 2 on adult residents (aged ≥ 20 years) in 824 European cities, in 2018. Exposure response functions from PM 2.5 and NO 2 single-pollutant models were adjusted by a coefficient difference approach in a multi-pollutant HIA. We quantified pollutant dominance and identified intra-city pollutant distribution patterns in relation to potential double counting and multi-pollutant burden. Findings Non-compliance with the WHO 2021 guidelines resulted in a total of 213 889 deaths [95% CI: 89 224 – 344 223] jointly attributable to PM 2.5 and NO 2 . This was 76 833 less deaths than the sum of single-pollutant attributable estimates (PM 2.5 : 145 747 [95% CI: 104 999 – 189 535]; NO 2 : 143 646 [95% CI: 87 398 – 198 313]). Cities with the highest multi-pollutant age-standardised attributable mortality were PM 2.5 dominant, and mostly found in Eastern Europe. The majority of cities were NO 2 dominant, however varied by intra-city pollutant profiles and magnitude of multi-pollutant burden. Interpretation Quantifying attributable burden in a multi-pollutant HIA provides interpretable burden estimates for policy-makers and minimises potential double counting. Our findings provide critical evidence for tailoring urban and transport planning policies to city-specific contexts.