The Burden of Rural Chronic Obstructive Pulmonary Disease: Analyses from the National Health and Nutrition Examination Survey

To the Editor:

Urban–rural disparities among individuals with chronic obstructive pulmonary disease (COPD) in the United States are receiving increasing attention, with concern that those in rural areas face unique risk factors (1). Data from the CDC and the National Health Interview Survey (NHIS) show that the prevalence of COPD in rural communities is nearly double that in urban areas (2). Our recent publication highlighted that rural residence represents an independent risk factor for COPD throughout the United States, even among never-smokers (3). A limitation of these prior studies was their reliance on self-reported data. Studies have shown that relying on self-reports can lead to underdiagnosis of COPD, particularly in disadvantaged communities, and therefore an underestimation of disparities (4). Recently, the COPD National Action Plan highlighted the importance of objectively defining the burden of rural COPD to design a coordinated approach to reduce this disparity (5). However, the impact of rural residence on both the prevalence and morbidity of COPD has not been demonstrated in a nationally representative sample that includes objective spirometry measurements. We sought to build on prior results by 1) identifying the prevalence of spirometrically defined COPD in the rural United States and 2) describing the independent impact of rural residence on COPD prevalence and morbidity.

Methods

We studied adults above the age of 40 with data available from the 2007–2012 National Health and Nutrition Examination Survey (NHANES) (6). NHANES uses a nationally representative sample, and 2007–2012 is the most recent period in which spirometry was performed. Participant data were linked to geocoded community data in the U.S. Census Bureau’s 2010 American Community Survey and the National Center for Health Statistics (NCHS) Urban–Rural Classification of Counties. The methodology used in this study replicates and extends our previously published work (3).

The primary outcome was prevalence of COPD, as defined by airflow obstruction on prebronchodilator spirometry, a >100 cigarette smoking history, and no concomitant diagnosis of asthma. This definition was based on previous epidemiologic studies of COPD within NHANES and was chosen to limit the risk of confounding by asthma (7). In primary analyses, obstruction was defined according to American Thoracic Society (ATS)/European Respiratory Society criteria (FEV₁/FVC < lower limit of normal). A sensitivity analysis used Global Initiative for Obstructive Lung Disease (GOLD) criteria (post-bronchodilator FEV₁/FVC < 0.70) for the subset of participants with available postbronchodilator spirometry (8). Additional sensitivity analyses were performed with the primary definition of COPD altered to include never-smokers and individuals with asthma. We also performed a secondary analysis to describe predictors of COPD among never-smokers. The outcomes in the analysis of COPD morbidity were airflow obstruction severity, self-reported respiratory symptoms (chronic cough, phlegm, and wheeze), and hospitalization in the prior year.

Participant characteristics collected from NHANES included age, sex, race, education, smoking duration, household secondhand smoke, high-risk occupational exposures (agriculture and mining), and region of residence. Census-level characteristics, including census-level poverty and data on heating with solid fuels (coal and wood) in each census tract, were also analyzed. Because individual poverty data were missing for more than 10% of the participants, household education and community poverty were used as surrogates. Urban–rural status was determined using the NCHS Urban–Rural Classification of Counties, which categorizes county metropolitan status on a continuum from urban to nonmetropolitan rural counties (9). The definitions of urban–rural status and community poverty were further outlined in our previous publication (3). Two multivariable logistic regression models were designed to assess different outcomes: 1) COPD prevalence (adjusted for individual and community-level factors noted above) and 2) COPD morbidity (adjusted for lung function [FEV₁] in addition to the listed covariates). All analyses used the survey weights and strata provided by NHANES. The NCHS Ethics Review Board approved the analysis of restricted data through the NCHS Research Data Center in Hyattsville, Maryland.

Results

Data from 8,500 adults with census-tract data and spirometry that met ATS quality standards were analyzed (8); 19.5% of the participants resided in rural areas, and 29.6% resided in urban areas. The estimated national prevalence of COPD was 8.9% (95% confidence interval [CI], 7.9–9.9%). Rural areas had the highest prevalence of COPD, at 12.0% (95% CI, 9.7–14.9%), which was double that in urban communities (5.9%; 95% CI, 4.7–7.3%). The prevalence of COPD was higher in rural areas than in urban areas across all age groups (Figure 1) and in all U.S. regions captured by NHANES (Figure 2).

Figure 1.

Estimated national prevalence of chronic obstructive pulmonary disease (COPD) by urban–rural status across age groups. Prevalence estimates are listed below the graph with 95% confidence intervals.

Figure 2.

Estimated national prevalence of chronic obstructive pulmonary disease (COPD) by urban–rural status across regions. Prevalence estimates are listed below the graph with 95% confidence intervals. *Spirometry data were not available for adults from rural tracts in the Northeast and as a result, prevalence data were not available.

In adjusted models, individuals living in rural areas had greater odds of having COPD than their urban counterparts (odds ratio [OR], 2.06; P = 0.005). The relationship persisted in sensitivity analyses that defined airflow obstruction using GOLD criteria (OR, 1.62; P = 0.017). In two sensitivity analyses that altered the definition of COPD to include never-smokers (OR, 1.88; P = 0.004) and individuals with asthma in addition to never-smokers (OR, 1.53; P = 0.024), rural residence was consistently associated with greater odds of COPD, but the magnitude of the association was attenuated. In contrast to previous reports, community-level poverty was not independently associated with COPD (OR, 1.16; P = 0.22) after accounting for individual factors.

Solid fuel use was more common in rural census tracts: 4.1% (3.3–7.8%) of the participants used wood as their primary heating source, as opposed to 0.6% (0.3–0.9%) in urban tracts. Use of coal for residential heating was rare in both rural and urban tracts. Neither use of coal nor use of wood for heating was associated with higher COPD prevalence among individuals with a smoking history. However, in the analysis of never-smokers, there was a significant association between wood combustion and COPD prevalence in multivariable models (OR, 1.12; P < 0.001), such that a 1% increase in the number of homes that used wood as the primary heating source was linked to a 12% higher odds of COPD.

Among individuals with COPD, those living in rural areas experienced greater morbidity than their urban counterparts. Rural residence was linked to higher odds of reporting respiratory symptoms (cough, phlegm, or wheeze [OR, 1.87; P = 0.031]), along with higher odds of both moderate (FEV₁ < 80% predicted; OR, 1.66; P < 0.001) and severe (FEV₁ < 50% predicted; OR, 3.69; P = 0.018) airflow obstruction. However, there was no difference in odds of hospitalization (OR, 1.82; P = 0.114) between urban and rural residents.

Discussion

Using a nationally representative sample with objective spirometry measurements from NHANES, we were able to demonstrate a profound urban–rural disparity among individuals with COPD, with greater prevalence and increased morbidity among those residing in rural areas. This study builds on recent work describing urban–rural differences in COPD prevalence in the NHIS, now leveraging spirometry to substantiate prior findings based on self-report. It also increases our understanding of the effects of rural residence on COPD-related respiratory morbidity, demonstrating greater disease severity and respiratory symptoms among individuals living in rural areas. The results suggest that increased neighborhood use of wood for heating, an exposure that is more prevalent in rural areas, is associated with greater COPD risk among never-smokers in the United States. These results reinforce prior findings from the NHIS that neighborhood use of solids fuels for heating is associated with higher COPD prevalence among never-smokers (3).

This study has limitations, some of which highlight the unique challenges of studying rural COPD. Although NHANES was designed to be representative of U.S. demographics, it does not sample all 50 states. Rural residents faced a greater burden of COPD in the West, Midwest, and South, but data were not available for adults living in isolated rural areas of the Northeast (Figure 2). This underscores the need to build a more robust infrastructure to investigate COPD in understudied rural areas. We also note that although GOLD guidelines recommend post-bronchodilator spirometry to confirm airflow obstruction, the primary analyses used prebronchodilator spirometry to capture the most participant data; post-bronchodilator spirometry was only available for a subset of participants. Multiple large epidemiologic studies have described a strong correlation between pre- and post-bronchodilator spirometry in predicting outcomes, and lower-limit-of-normal values, which define obstruction according to ATS criteria, are determined using prebronchodilator spirometry (10). Furthermore, a sensitivity analysis that applied the GOLD criteria of FEV₁/FVC < 0.70 for individuals with post-bronchodilator spirometry produced results consistent with those obtained in the primary analysis. Lastly, there is a need for individual-level environmental exposure assessments in future studies to better quantify the contribution of factors such as secondhand smoke and heating with solid fuels to disease development in rural areas—a limitation of the present study.

Despite these limitations, by using an approach that allowed us to uniquely link multiple nationally representative studies, we confirmed that individuals living in rural areas are at increased risk for spirometry-defined COPD and face greater respiratory morbidity. Further studies are now needed to better understand the risk factors that are unique to rural regions and enable the development of strategies to improve respiratory health and reduce disparities.