It’s Not Just a Smoking-related Disease: Outdoor Pollution May Increase Risk of Chronic Obstructive Pulmonary Disease

It is well accepted that air pollution exacerbates existing respiratory diseases in adults, including asthma and chronic obstructive pulmonary disease (COPD). It is less clear if cumulative exposure to air pollution increases the incidence of new obstructive lung diseases in adults. In this issue of the Journal, Shin and colleagues (pp. 1138–1148) present a population-based study of all 5 million adults living in the Canadian province of Ontario who were free of asthma or COPD diagnosis in 2001 (1). During the 15-year study follow-up period, over 340,000 incident cases of COPD and over 218,000 incident cases of asthma were identified. Using a 3-year average of air pollution exposure with a 1-year lag, the authors found that multiple outdoor pollutants, including fine particulate matter (PM_2.5), nitrogen dioxide (NO₂), and ozone (O₃), were associated with greater COPD incidence, which is defined as a COPD hospitalization or at least three physician claims for COPD within 2 years. In contrast, there were no associations between exposure to these pollutants and new adult-onset asthma.

Ontario is a large contiguous geographic area that includes a large metropolitan area and more rural areas with sufficient variability in the long-term exposure to these pollutants to examine associations with asthma and COPD incidence, even at the relatively low pollution concentrations in Canada. A major strength of this population-based cohort study is that it included all Ontarians with minimal concern for selection bias because everyone is eligible for the publicly funded health plan. However, the use of administrative claims data has the important limitation that the authors could not directly adjust for individual-level habits such as tobacco use, which is the number one cause of COPD. The authors attempted to address this limitation by adjusting for individual-level comorbidities, fine-scale neighborhood characteristics, and indirectly adjusted for smoking and obesity using an auxiliary dataset. The findings for COPD were robust to adjustment for these potential confounding variables. Interestingly, healthier adults (those without comorbidities and those who were younger at baseline) had a larger relative risk of COPD in association with pollution than those who had more comorbidities or were older. As in other research on the health effects of pollution, residents in neighborhoods of lower socioeconomic position had a heightened risk of COPD incidence in association with PM_2.5 compared with those in more privileged neighborhoods.

The authors examined the shape of exposure–response relationships and found “supralinear functions” for PM_2.5, O₃, and the redox-weighted oxidative potential of O₂ and NO₂, meaning that the exposure–response relationship for COPD incidence was steeper at lower exposure concentrations (<10 μg/m³ PM_2.5 and <40 ppb O₃) and less steep at higher exposure concentrations. This pattern of greater health effects per increment of pollution at the lower range of exposure has been observed for other health effects of PM_2.5 and O₃ exposure, including mortality (2). For NO₂ exposure, there appeared to be a threshold effect with increasing risk for COPD only above 25 ppb. The authors importantly highlight that these adverse effects were observed for concentrations well below the current U.S. national ambient air quality standard for PM_2.5, O₃, and NO₂. Several recent reports have concluded that despite the strong plausibility that outdoor pollution causes COPD (based in part on the compelling evidence for tobacco and indoor biomass smoke), the evidence thus far has been deemed insufficient to determine causality (3–5). This study provides compelling evidence that long-term exposure to higher concentrations of ambient pollution increases the incidence of COPD.

In contrast, results for adult-onset asthma were null. There are several possible reasons for these null results: 1) there is truly no effect of air pollution during adulthood on adult-onset asthma at the pollution concentrations observed; 2) sampling error, which randomly resulted in variation from the true population effect; or 3) bias was present in the study that only (or more severely) impacted the asthma analysis and not the COPD analysis. We discuss each of these below.

Air pollution exposure may not increase the risk of asthma in adults at the concentrations observed in this study. A recent American Thoracic Society workshop concluded that traffic-related pollution causes asthma in children (3), but evidence that outdoor pollution causes adult-onset asthma is sparse and equivocal. A nationwide study of women from the United States found that each 3.6 μg/m³ of long-term PM_2.5 exposure was associated with an odds ratio of 1.20 (95% confidence interval, 0.99–1.46) for adult-onset asthma, with similar findings for NO₂ (6). A large multicohort study in Europe (ESCAPE) reported numerous positive but insignificant associations between different traffic-related exposure metrics and adult asthma incidence (7). Perhaps the window of vulnerability for asthma onset due to air pollution has already passed by age 35. Unfortunately, the authors were limited in their ability to investigate exposures earlier than this 3-year window with a 1-year lag.

Sampling error is unlikely given that very large sample size, but bias is possible. The use of the health administrative databases precluded direct adjustment for potential individual-level confounders such as smoking, obesity, or family history of asthma. Unmeasured or residual confounding could bias results. Any negative confounding variables, however, would have to only (or more strongly) affect asthma over COPD to explain the null result for asthma only. Although outcome misclassification is possible, the asthma ascertainment method used here by Shin and colleagues is similar to methods commonly applied in epidemiologic studies and not likely a major source of bias. Finally, we would expect any severe exposure misclassification to bias both the COPD and asthma analyses, thus it is not likely a major factor to explain the null asthma findings.

Air pollution effect estimates are often relatively small (hazard ratio, 1.03–1.06 in this study), highlighting the need for large, well-conducted studies, such as the one by Shin and colleagues. Air pollution exposure is ubiquitous, and even a small effect estimate can have a large impact on the population’s health, especially when it contributes to the incidence of a chronic disease that impairs quality of life and increases health care use. From a policy standpoint, an increase in the incidence of an irreversible chronic disease is even more costly to society than acute events, such as asthma or COPD hospitalizations. There is a great need for studies on long-term air pollution exposure from childhood into adulthood and the incidence of chronic respiratory disease in adults. Research such as this work that involves areas where pollution concentrations are mostly below current regulatory standards is especially informative for future policy decision-making to protect public health.