Abstract
Background
Estimates of disease burden attributable to air pollution depend on the choice of exposure–response functions (ERFs). The World Health Organization (WHO) recommends using all-cause mortality ERFs in the primary analysis, with cause-specific outcomes as a secondary analysis. Therefore, this study aims to assess how disease burden estimates for air pollution are influenced by using ERFs for all-cause mortality (ACM) versus disease-specific (DS) mortality.
Methods
Environmental burden of disease (EBoD) estimates for PM2.5 and NO2 were calculated within the framework of the EU project BEST-COST across five countries (Belgium, Estonia, France, Norway, and Portugal) using the healthiar R package. WHO-recommended ERFs for ACM and DS mortality were applied to compare attributable Years of Life Lost (YLL). Exposure data and baseline health data were obtained from European Environment Agency and Global Burden of Disease study, respectively. Statistical differences were assessed using Wilcoxon signed-rank tests, Spearman correlations, and intraclass correlation coefficients (ICC).
Results
Across all countries, the DS approach consistently yielded 2–4 times lower YLL estimates than ACM framework (ratios: 0.26–0.52; median 0.31 for NO2, 0.42 for PM2.5). Despite these absolute differences, country rankings remained highly consistent (Spearman ρ ≥ 0.90). Wilcoxon tests did not reach significance (p = 0.0625), likely due to limited sample size. ICC indicated poor agreement for NO2 (0.15) and moderate for PM2.5 (0.46).
Conclusions
Although both DS and ACM approaches yield highly consistent disease burden country-level rankings, DS ERFs result in dramatically lower YLL estimates. It should, however, be kept in mind that the ACM approach could result in an overestimation of the disease burden due to air pollution Therefore, discussing both frameworks’ pros and cons is required to better understand the consequences of ERF choice on policy implications.
Key messages
• Disease-specific ERFs result in systematically lower YLL estimates for air pollution, with 50–70% compared to all-cause mortality approaches.
• Further discussion of WHOs recommendation of all-cause mortality in environmental decision-making may be warranted.