Gender-specific determinants of asthma among U.S. adults

Using 2007-2012 BRFSS data from 1,003,894 U.S. adult respondents, we quantified the association between asthma and socioeconomic and demographic factors. The strongest non-modifiable factor associated with asthma was gender, with women having an OR of 1.80 (95% CI 1.74-1.86) relative to men. We next performed gender-stratified BRFSS adjusted analyses to identify independent risk factors that varied in men and women. Similar analyses with NHANES adult data from 2007-2012 were performed to confirm BRFSS findings. While NHANES was a much smaller study (n?=?13,364), some of its data is considered more reliable than that of BRFSS, as interviews were performed in-person and measures such as height and weight were obtained during physical exams, rather than via phone interviews.

The strongest association observed for asthma in both BRFSS and NHANES was with obesity, and this risk factor was particularly important among women. Our findings are consistent with the growing body of literature on the relationship between obesity and asthma, and more generally, metabolic syndrome and lung disease [33–36]. Prospective studies have found a dose-response relationship between the odds of asthma incidents and being overweight or obese [37, 38], and data from animal and human studies have shown that increases of normal adipose tissue lead to a systemic proinflammatory state [39]. Although the precise mechanism by which obesity contributes to asthma is not known, based on our findings and those of others, decreasing obesity should be a primary goal of interventions intended to reduce asthma. Our finding that obesity is a strong risk factor for asthma particularly in women is also consistent with a recent report based on U.S. National Health and Nutrition Examination Survey data, although in contrast with our results, that report found that obesity was not a significant risk factor in men [40]. The reason why obese women are at greater risk for asthma than obese men is not known.

Current smoking was another modifiable factor that was associated with asthma and had a greater effect in women than in men. The association between smoking and increased risk of asthma is well-documented, but there is little functional evidence that directly links smoking to asthma and it is unclear why the association is more pronounced in women than men [41, 42]. The gender-specific disparity observed between women and men was larger in NHANES than in BRFSS: women who were current smokers in NHANES had an OR of 1.78 for having asthma vs. never smokers, while those in BRFSS had an OR of 1.41. We are unable to determine whether this difference is due to unequal response reliability between the two studies or a reflection of cohort size differences.

While previous studies suggest that individuals differ in their genetic predisposition to asthma owing to differences in genetic ancestry that are reflected in racial/ethnic categories [8–10, 43], our BRFSS results suggest that other factors, such as income, more prominently lead to disparities in asthma prevalence by race/ethnicity. Although black and American Indians/Alaskan Native respondents had increased odds of asthma relative to white respondents in unadjusted analyses, after adjusting for other factors and stratifying by gender, the association with black race/ethnicity was not significant in males and led to decreased odds of asthma in females, and the association with American Indians/Alaskan Natives was considerably decreased in magnitude in both men and women. In NHANES adjusted analyses, statistically significant differences in asthma prevalence were observed for Hispanic men and women, who had the lowest prevalence of asthma among racial/ethnic groups.

The categorization of Hispanic respondents is limited, especially for BRFSS, in that we do not know respondent country of origin and large differences in asthma rates are known to occur among such groups: Mexican Americans have the lowest rate of asthma in the U.S. while Puerto Rican Americans have the highest rate [7]. Because BRFSS attempts to represent the U.S. population, the Hispanic results from this study most likely reflect responses from Mexican Americans, who according to the 2010 US census represent 63% of Hispanics in the U.S. Puerto Ricans, who are the next largest group, represent 9.2% of U.S. Hispanics [44]. NHANES included a Mexican American racial/ethnic category, and respondents who were classified in this group did have lowest asthma rates consistent with previous findings [7]: NHANES Mexican American males and females had adjusted ORs for asthma of 0.28 (95% CI: 0.17-0.45) and 0.35 (95% CI: 0.24-0.52), respectively, while NHANES Other Hispanic males and females had adjusted ORs for asthma of 0.59 (95% CI: 0.36-0.95) and 0.97 (95% CI: 0.71-1.32), respectively. Studies that gather more detailed information regarding Hispanic country of origin are likely to be more informative regarding the influence of demographic and socio-economic factors and asthma among U.S. Hispanics.

In BRFSS, low yearly household income ($25,000) remained a significant risk factor for asthma in both genders after adjusting for other socioeconomic and demographic factors. Because a greater proportion of black and American Indian/Alaskan Native relative to white BRFSS respondents had low income, unadjusted analyses found stronger race/ethnicity associations within these groups. Our NHANES results also found that individuals of both genders living below the poverty line (PIR ?1) were at increased risk of having asthma. The association between poverty and asthma has been observed and explored for many years [43, 45–47], and our results support the notion that individuals living in poverty experience a higher disease burden of asthma in the U.S. Continued studies that seek to understand the environmental and healthcare access factors, including gene-environment interactions [48], exposures (e.g., mold, cockroaches, and house dust mites) [49, 50] and health literacy [12, 51], that are associated with low income status and disproportionately affect some minority groups will decrease in asthma prevalence disparities by race/ethnicity.

Our findings are consistent with previous studies that used smaller portions of BRFSS data to examine demographic factors associated with asthma, although none performed gender-specific analyses [23, 24]. A study that used 2000 BRFSS data also found that obesity, female gender, current/past smoking, and low socioeconomic status were associated with asthma [24]. Another study that used 2009-2010 BRFSS data and performed race/ethnicity stratified analyses found that asthma risk factors differed by race/ethnicity [23]. For example, female gender and low income status were associated with asthma in all groups other than Asian/Pacific Islanders, and obesity was associated with asthma in all groups other than American Indians/Alaskan Natives [23].

By using NHANES and BRFSS data while accounting for their survey design, we obtained nationally representative measures from two independent studies with a large number of diverse participants. Although results from both studies were highly consistent, both datasets are subject to limitations. We were unable to make inferences about causality of observed associations due to the cross-sectional nature of BRFSS and NHANES. While the reliability and validity of BRFSS is high [17, 18], a study comparing it to other national surveys found that BRFSS asthma prevalence estimates were higher than those of the National Health Interview Survey (NHIS) [52]. BRFSS relies solely on self-reported data, and is thus, subject to error. Further, while we used a weighted survey design, residual bias from differences between survey respondents and the U.S. adult population may still affect our results. For example, BRFSS respondents were 60% female and 40% male, a distribution that is very skewed compared to the U.S. population. Comparison of other characteristics of our BRFSS respondents to those of the general U.S. population reveal other differences: (1) our study subjects consisted of 79% non-Hispanic White, 9% non-Hispanic Black, 8% Hispanic, 3% Asian/Pacific Islander and 1% American Indian/Alaskan Native respondents, while the 2010 U.S. Census reported a population of 64% non-Hispanic White, 13% Black, 16% Hispanic, 5% Asian/Pacific Islander and 1% American Indian/Alaskan Native [53], (2) 98% of our study subjects (all at least 22 years old) completed high school or more and 68% completed some college or more, while according to the American Community Survey, in 2015, 88% of the U.S. population at least 25 years old completed high school or higher and 59% had completed some college or more [54]. Household income of our study subjects was consistent with the U.S. mean of $55,775 in 2015 according to the American Community Survey [55], as 25% of our study subjects had a household income below $25,000, 42% had a household income between $25,000 and $75,000, and 32% had a household income above $75,000. The consistency we observed for BMI association results between BRFSS and NHANES suggests that self-reported measures of BMI from BRFSS were as reliable as those for NHANES, or that owing to its larger sample size, bias errors from BRFSS did not affect overall results for at least this variable. We used complete cases for our study, which may have introduced bias related to questions not answered by some groups although our sample size was large and included many subjects in most categories.