Shikang Liu, Zhaozhao Hui, Wen Zhang, Luping Chen, Mingxu Wang
Contrary to the assumption that lockdowns would uniformly attenuate pollution-health effects, the health associations of combustion-source pollutants (SO2, NO2) in Xi'an were consistently stronger during strict COVID-19 control than the post-pandemic period, though formal interaction tests did not reach statistical significance. This pattern is consistent with, but does not confirm, an indoor exposure amplification mechanism in which prolonged home confinement increased cumulative exposure to residential heating emissions. These findings should be regarded as hypothesis-generating and indicate that reduced outdoor air pollution does not automatically translate into reduced health risk when population behavior changes substantially.
BACKGROUND: The COVID-19 pandemic triggered unprecedented social interventions that altered ambient air quality, population behavior, and healthcare-seeking. Whether these interventions modified the short-term association between air pollutants and respiratory health, and in which direction, remains unclear.
METHODS: We conducted a time-series study of daily respiratory outpatient visits (ICD-10: J00-J99, R04-R07, and R09) from a tertiary hospital in Xi'an, China, from January 2020 to December 2025. The study period was stratified into three COVID-19 phases: strict control (2020-2022), transition (Jan-Jun 2023), and post-pandemic (Jul 2023-2025). Generalized additive models with quasi-Poisson regression evaluated associations between PM2.5, SO2, NO2, and O3 and outpatient visits, adjusting for meteorological confounders, temporal trends, and phase. Pollutant-by-phase interaction terms tested for effect modification.
RESULTS: A total of 2,052,777 respiratory outpatient visits were recorded over 2,192 days. In full-period models, SO2 (lag4: ER = 9.11, 95% CI: 3.86-14.64%) and NO2 (lag5: ER = 2.72, 95% CI: 2.05-3.40%) showed the strongest associations. Phase-stratified analysis showed that the SO2 association was stronger during the strict control period (ER = 18.40, 95% CI: 10.29-27.11%) than the post-pandemic period (ER = 2.91, 95% CI: -4.82-11.28%), and the NO2 association followed the same pattern (3.07% vs. 1.31%). Formal interaction tests supported this gradient (SO2 × phase p = 0.13, NO2 × phase p = 0.15) but did not reach statistical significance, whereas PM2.5 and O3, which lack a dominant indoor combustion source, showed essentially no evidence of interaction (both p > 0.9). A localized Air Quality Health Index identified NO2 (55.4%) and SO2 (36.5%) as the dominant contributors to respiratory health risk.
CONCLUSION: Contrary to the assumption that lockdowns would uniformly attenuate pollution-health effects, the health associations of combustion-source pollutants (SO2, NO2) in Xi'an were consistently stronger during strict COVID-19 control than the post-pandemic period, though formal interaction tests did not reach statistical significance. This pattern is consistent with, but does not confirm, an indoor exposure amplification mechanism in which prolonged home confinement increased cumulative exposure to residential heating emissions. These findings should be regarded as hypothesis-generating and indicate that reduced outdoor air pollution does not automatically translate into reduced health risk when population behavior changes substantially.