Association between neighborhood socioeconomic disadvantage and chronic obstructive pulmonary disease prevalence among U.S. Veterans

Abstract: Background: The contribution of individual socioeconomic status to chronic obstructive pulmonary disease (COPD) burden is well documented, and increasing attention is being placed on the impact of neighborhood disadvantage on health outcomes. Prior research shows that Medicare beneficiaries living in in the most disadva neighborhoods, as measured by the Area Deprivation Index had higher rates of chronic pulmonary disease raged erivation Indes (ADI). However, this analysis did not look specifically at COPD and was not adjusted for confounders such as rurality, race, and tobacco me. Rurality is associated with COPD prevalence, even after adjusting for poverty. Rurality is also associated with tobacco use, the major cause of COPD. Therefore, if its adjusted association with COPD prevalence persists, the ADI could inform interventions within bigh unities across rural and densely populated urban areas. Such health systems interventions are needed to account for the socioeconomic, environmental, and structural inequities that perpetuate adverse exposures and behaviors. Our study, therefore, tests the hypothesis that neighborhood disadvantage is associated with prevalent COPD after adjusting for important neighborhood- level confounders. Methods: Our primary outcome was COPD prevalence at the census block group (CBG) level in 2018, calculated as the ratio of veterans with COPD to the total number of veterans per CBG. With approval from the VA Puget Sound institutional review board, we constructed a cohort of veterans who were diagnosed with COPD between 2010 and 2018 on the basis of two or more COPD outpatient visits and two dispensations of a long-acting bronchodilator. We determined the total number of veterans per CBG using veteran population data from the 2014-2018 U.S. Census Bureau 5-year estimates. Our primary exposure was the ADI, a composite indicator of neighborhood socioeconomic disadvantage at the CBG level. The ADI ranges from 1 (lowest disadvantage) to 100 (highest disadvantage). We used latitude and longitude of patient residence at time of cohort entry to determine their CBG. To account for neighborhood-level covariates, we controlled a priori for census-aggregated veteran population variables. We obtained CBG-averaged data on veteran age and sex, as well as census tract-level data on veteran race and ethnicity from the 2014-2018 U.S. Census Bureau estimates. We measure rurality using the Department of Agriculture's Rural and Urban Community Area Codes. We obtained the census tract-level proportion of smokers for 2018 from PolicyMap. We used descriptive statistics to estimate means and proportions for all study variables and performed Poisson regression analyses to assess the association between ADI and CBG prevalent COPD. We modeled the ADI as a continuous variable scaled to 10% and as a categorical variable that was based on quintile. We performed a sensitivity analysis to examine rurality as a potential effect modifier. Analyses were conducted using Stata, Microsoft Excel, R Studio, and ArcGIS Pro. Results We identified 725,494 veterans with COPD in our cohort. To capture prevalent COPD as of 2018, we excluded veterans who died between 2010 and 2018 (n = 246,236). We excluded veterans without Federal Information Processing System codes (n = 56,511) for ADI matching (e.g., those who lived in U.S. territories that were not included in ADI or those without a home address). The remaining 422,747 veterans with COPD were geolocated to a total of 217,667 CBGs. CBG-level veteran population characteristics and COPD cohort characteristics are displayed in Table 1. -५० 1- Prevalent COPD at the CBG level ranged from 5% (SD = 9) in the lowest ADI quintile to 12% (SD = 20) in the highest ADI quintile. The adjusted risk of prevalent COPD in the highest ADI quintile was approximately 3.1 times greater (risk ratio, 3.14; 95% confidence interval: 3.14, 3.28) than in the lowest ADI quintile (Figure 1). A 10% increase in the continuous ADI score corresponded with 15% increased adjusted risk of prevalent COPD (risk ratio, 1.15; 95% confidence interval: 1.14, 1.15). The association between ADI and prevalent COPD did not differ significantly by rural status (Pinteraction = 0.85), indicating that rurality did not modify this rom 0 two relationship. Discussion: Our results build on prior research to examine the role that neighborhood disadvantage plays in chronic lung disease outcomes. We have filled a knowledge gap by showing that neighborhood disadvantage accounts for higher COPD prevalence at the CBG level after adjusting for potential confounders. There are multiple plausible explanations for our findings, including increased pollution, occupational hazards, higher stress, and early life exposures within disadvantaged neighborhoods. Our findings have important policy implications. By focusing on COPD prevalence at the census block group level, our results reveal geographic patterns of risk and disadvantage that can inform allocation of resources to address health disparities within a health care system. The ADI has the potential to identify at-risk communities not only in rural areas where we know COPD burden is high but also in high- population areas where interventions can impact large numbers of patients.