Abstract
To the Editor:
I read with interest the recent article by Qu et al. on the association between air pollutants and adult atopic dermatitis (AD) outpatient visits in Beijing. 1 The authors present a substantial dataset spanning nearly 6 years, making a useful contribution to our understanding of environmental triggers in adult AD populations. However, several methodological aspects warrant further discussion regarding the interpretation and clinical applicability of these findings.
The study assigned air pollution exposure using concentrations from the nearest monitoring station to each patient’s residential address. While this is a practical approach for time-series work, it raises questions about exposure misclassification. Individual exposure varies considerably based on time-activity patterns, indoor pollution sources, and specific microenvironments, factors not captured here. Previous work has shown that outdoor monitoring data can misrepresent personal exposure by 30–50%, especially for people who spend most of their time indoors. 2
This becomes particularly relevant when we consider the reported age and gender differences. If elderly or female patients have systematically different activity patterns (for example, spending more time indoors), what appears to be differential biological susceptibility could partly reflect differential measurement error. The authors might consider sensitivity analyses using more spatially refined exposure models, as land-use regression or satellite-derived PM2.5 estimates have shown better performance in urban settings. 3 It would also help to know the typical distance between monitoring stations and patient residences, which would give readers a sense of the spatial resolution being assumed.
The authors examined lag effects from day 0 to day 7 and found the strongest associations at lag 6–7 for several pollutants. Biologically, particulate matter triggers oxidative stress and immune activation within hours to 2 days. 4 The progression from these early subclinical responses to symptomatic exacerbation severe enough to prompt an outpatient visit, however, may reasonably span several more days. A peak effect at lag 6–7 could therefore reflect a genuine causal pathway, though it also leaves open the possibility that unmeasured time-varying confounders such as delayed care-seeking, appointment scheduling, or other factors that happen to track pollution levels from days earlier contribute to the pattern.
The analysis also doesn’t account for cumulative effects or potential non-linearities in the exposure-response relationship. Distributed lag non-linear models, as described by Gasparrini et al., offer a more flexible framework that can capture both delayed and cumulative exposure patterns while reducing bias from model specification choices. 5 Applying this approach over, say, a 0–14 day window might clarify the temporal dynamics. Season-stratified analyses could also be informative. Do these lag patterns hold across heating and non-heating seasons? This could help separate direct pollution effects from other seasonal factors like indoor heating or humidity changes.
These methodological points notwithstanding, the authors have tackled an important public health question and assembled an impressive dataset. Refining the exposure assessment and reconsidering the lag structure would strengthen the findings and make them more directly applicable to both clinical guidance and air quality policy. The work highlights a real concern for adult AD patients in polluted urban environments and provides a foundation for more detailed investigations.