IntroductionThe rising global incidence of atopic dermatitis (AD) is due in part to air pollution, however evidence from arid interior northwest China is scarce.MethodsThis time-series analysis included 79,526 AD outpatient visits in Shihezi from 2015 to 2024, which were linked with six contaminants and climatic data. The DLNM and ARIMAX models evaluated short-term nonlinear lagged associations using two-pollutant adjustment, stratified analyses (gender, age, heating season), interaction testing, and sensitivity analysis.ResultsPer IQR increase, PM10 same-day visits increased by 6.0% (RR = 1.060, 95%CI 1.027–1.095), cumulative lag 6 RR = 1.110 (1.032–1.194), SO2 cumulative 7-day RR = 1.051 (1.024–1.078), and CO acute same-day RR = 1.037 (1.008–1.067). Two-pollutant models confirmed independent effects of PM10 and SO2, but O3 adjustment attenuated the associations for NO2 and CO. Significant multiplicative interactions (p < 0.05) were found for PM10 and CO with gender, age, and heating season. Stratified estimates revealed larger relationships for females, working-age persons (13–60 years), and heating season(November to April of the following year), but caution is advised due to limited power and multiple testing. ARIMAX models were consistent with the pollution effects and showed decreased recorded AD visits during COVID-19 lockdowns, which may mainly reflect changes in healthcare-seeking behaviour rather than a true decrease in AD incidence. Sensitivity analyses confirmed resilience.DiscussionIn this arid desert-industrial metropolis, short-term exposure to PM10 and SO2 is associated with an increase in the number of AD’s outpatient visits with permanent cumulative delays. Although effect modification was identified, specific sensitive subgroups could not be proven due to ecological design limitations. This research covers data gaps in northwest dry zones, promoting targeted air pollution regulation and dermatological public health initiatives.