Article Text
PDF
Abstract
Background Existing evidence on stress and asthma prevalence has disproportionately focused on pregnancy and postpregnancy early life stressors, largely ignoring the role of childhood adversity as a risk factor. Childhood adversity (neglect, stressful living conditions and maltreatment) may influence asthma prevalence through mechanisms on the hypothalamic-pituitary axis.
Methods Data from the Center for Disease Control's (CDC's) Behavioral Risk Factor Surveillance System (BRFSS) surveys were used to examine cross-sectional associations of adverse childhood experiences (ACE) with lifetime and current asthma prevalence. Information on childhood adversity was available from 84 786 adult respondents in 10 US states. Poisson regression models (with robust SE) were used to estimate prevalence ratios (PRs) relating overall ACE score and dimensions of exposure ACE to asthma prevalence, adjusting for socioeconomic status.
Results Greater ACE was associated with a higher prevalence of asthma (adjusted PRcat 4=1.78 (95% CI 1.69 to 1.87), adjusted PRcat 1=1.21 (95% CI 1.16 to 1.27)). Reported experiences of sexual abuse (adjusted PR=1.48* (1.42 to 1.55)) and physical abuse (adjusted PR=1.38* (1.33 to 1.43)) were associated with a higher asthma prevalence. No clear socioeconomic gradient was noted, but those reporting lowest education and income levels reported high rates of asthma and adversity. Sensitivity analyses indicated that ACE exposures were interrelated.
Conclusions Report of childhood adversity predicts asthma prevalence among US adults. Frameworks for asthma prevention need to recognise and integrate aspects related to childhood adversity. Further investigation into specific time periods of exposure would provide meaningful inferences for interventions.
- Asthma
- Asthma Epidemiology
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 3.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/3.0/
Statistics from Altmetric.com
Strengths and limitations of this study
-
Greater adversity in childhood associated with increased risk of current and lifetime asthma prevalence.
-
Dimensions of adversity (parental discord, stressful living conditions, verbal, physical and sexual abuse) were, singularly and in combinations, associated with increased asthma risks.
-
Lowest SES groups at higher risks of adversity, but unclear socioeconomic gradients.
Introduction
The majority of research linking stress with asthma has focused on pregnancy and postpregnancy early life stressors, with little investigation into childhood adversity and its long-term impacts. Three sets of epidemiological studies have dominated the discussion. First, studies examining in utero effects of stress during pregnancy and immediate postpregnancy risk factors have assessed evidence for programming of respiratory health.1–4 Second, studies have drawn associations between reported or measured parental stress as predictors of wheezing among children.5–7 Third, associations between mental distress and asthma have been examined.5 ,8–11
Assessing the role of childhood adversity investigates and relates to the research around physiological mechanisms for stress and asthma research. Evidence from stress biology finds links to asthma (along with bronchial hypersensitivity, eczema and wheezing) through immune and inflammatory responses, acting through the hypothalamic-pituitary axis.1 ,12–14 For instance, extreme stress and emotional states such as trauma have been associated with asthma through immune dysregulation as well as nervous system hyper-responsiveness. The role of such mechanisms may be more pronounced in childhood compared to pregnancy and postpregnancy. Further, hormones and inflammation (eg, interleukin (IL)-4, IL-5 and IL-13) related to stress can lead to contractions of smooth muscle and excess of mucus production, heightening risks for asthma morbidity.14 Hormonal mechanisms influencing weight gain and obesity have also been associated with stress and asthma.15 ,16
Research on childhood adversity may also address some limitations in the current epidemiological research on stress and asthma. Childhood adversity, often reported, examines the qualitative nature of experiencing stress in social interactions instead of focusing on stress biomarkers. In previous epidemiological studies, household or parental measures of stress were equated to the individual's own stress. This ignores the role of factors such as personality, resilience and the interactions between individual-social environments.
Childhood adversity has been operationalised by adverse childhood experiences (ACE), a multi-domain marker measuring diverse aspects of adversity including neglect, stressful living conditions, maltreatment and abuse. A number of studies have examined associations between ACE and health behaviours (including smoking, alcohol and drug initiation), suicide ideation and mental health outcomes.17–27 A handful of studies relating separate aspects of adversity to asthma outcomes have highlighted the potential for this relationship. For instance, a prospective follow-up of African-American women showed higher incidence rate ratios (IRRs) for physician-diagnosed asthma associated with any childhood abuse (IRR=1.24 (95% CI 1.06 to 1.45) and any adolescent abuse (IRR =1.10 (95% CI 0.88 to 1.36).28 Respondent's child protection agency history (a proxy for a stressful home environment or parental discord) was associated with higher elevated odds of lifetime asthma (adjusted OR=2.26, 95% CI 1.33 to 3.83); the association did not hold, however, for self-reported measures for the same exposure.29 Hyland et al30 showed that reported insults and physical punishment during childhood was associated with a higher likelihood of asthma (RR: 1.6).
In the context of these gaps, this study examined the relationships between overall ACE as well as independent ACE dimensions with asthma prevalence, using data from 10 US states between 2009 and 2011. We hypothesised positive relationships between adversity (measured through ACE) and asthma prevalence, even as relationships with markers of socioeconomic status (SES) may be variable for adversity and asthma.
Materials and methods
Data and sample
Behavioral Risk Factor Surveillance System (BRFSS) is a nationally representative annual telephone health survey designed and administered by the Center for Disease Control (CDC) that collects data on health risk behaviours, preventive health practices and access to healthcare across US states.31 As a repeated cross-sectional survey, it was designed to help in tracking preventive health goals and for highlighting emerging issues.
Information on self-reported physician-diagnosed asthma prevalence was available for 1 384 116 respondents across the USA between 2009 and 2011; however, information on childhood adversity measures was available for 84 796 respondents (2009: 12 088, 2010: 24 334 and 2011: 48 364). Hence, the total sample ‘n’ for this study was 84 796 (men: 40.1%, women: 59.9%) from 10 US states, representing all geographic regions: Arkansas, District of Columbia (DC), Hawaii, Louisiana, Minnesota, Montana, Nevada, Vermont, Washington and Wisconsin.
Variables
The main outcomes analysed in the study were self-reported physician-diagnosed asthma prevalence among adults (measured by lifetime and current asthma). Respondents in the survey were asked: “(Ever told) by a doctor, nurse or other health professional that you had asthma” followed by “Do you still have asthma?”. These outcomes have not been systematically tested for outcome reliability and validity. A review on self-reported asthma has estimated a mean sensitivity of 68% (range 48–100%) and a mean specificity of 94% (range 78–100%), with an increase if physician-diagnosis was reported.32
The main exposure of interest was childhood adversity operationalised as an ACE score. The analysis used both overall ACE score (categorised by adversity type) and five separate dimensions of ACE. Eleven questions from a submodule administered in 10 states gathered data from respondents regarding ACE (details in online supplementary table S2). Five separate dimensions were considered and represented inter-related aspects of experience of stress before age 18 years (details discussed in online supplementary table S2). These included: (1) parental marital discord, (2) stressful living conditions, (3) experience of verbal abuse, (4) experience of physical abuse and (5) experience of sexual abuse. ACE questions used in the BRFSS have been tested previously for reliability.21 Weighted-κ for individual questions and accumulated dimensions ranged from 0.52 to 0.86 with the reliability estimate for overall ACE being 0.64 (0.36 to 0.6).21 The inter-relatedness of ACE dimensions has also been tested previously, with 81–98% of respondents reporting more than one dimension of ACE.33 The type of adversity measured by ACE was categorised into an exploratory gradient. The categories considered included: ‘none’ (category 0), ‘parental marital discord or verbal abuse’ (category 1), ‘discord, verbal abuse and stressful living conditions’ (category 2), ‘discord, verbal abuse, stressful living conditions and physical abuse’ (category 3) and ‘discord, verbal abuse, stressful living conditions, physical and sexual abuse, and sexual abuse only’ (category 4; see online supplementary table S1).
Other covariates in the analysis included respondent household income and education, age, sex, race, survey year and state of residence. Respondents did not provide data on parents, and hence their self-reported education and income were used as controls and proxies for lifetime SES. Household income was categorised into six categories (less than US$15 000, US$15 000–35 000, US$35 000–50 000, US$50 000–75 000, >US$75 000 and not reported or refused) and education was categorised as ‘Less than high school’, ‘Finished high school’, ‘Some college’ and ‘College or more’. Respondent self-reports of race/ethnicity were categorised as Non-Hispanic White (NHW), Non-Hispanic Black (NHB), Hispanic and Other (including multiracial).
Analysis
The study estimated prevalence rates for lifetime and current asthma, along with dimensions of childhood adversity in the sample. Bivariate associations (along with χ2 tests) were estimated for asthma prevalence and ACE (overall and dimensions) with socioeconomic covariates of interest. Multivariable Poisson regression models with robust SE were used to estimate prevalence ratios (PRs) for asthma by ACE, adjusted for SES. PRs were preferred as risk ratio estimates over ORs due to the high prevalence of asthma in several categories.34 ,35 Regression models included unadjusted models (M1); models adjusted for age, sex, race and survey year (M2); models adjusted for age, sex, race, survey year and SES (M3) and models adjusted for age, sex, race, survey year, SES and state (M4). All analyses were conducted using STATA V.12.
Results
Prevalence rates for ACE and asthma
Table 1 shows the differences in asthma prevalence (lifetime and current) by socioeconomic and demographic covariates, including age, sex, race/ethnicity, income and education levels. Higher asthma prevalence was seen among women, those in the lower age groups, among NHW populations and those reporting lower income and education levels. Stressful living conditions was the most commonly reported dimension of ACE (34.46%) followed by any verbal abuse (32.62%) and parental marital discord (20.67%; table 2). Nearly 12.29% reported any experience of sexual abuse.
Sample characteristics and associations between asthma prevalence and socioeconomic and demographic variables
Dimensions of childhood adversity and prevalence of asthma
Exposure to experiences of adversity was associated with higher asthma prevalence. Experiencing parental marital discord and living in stressful conditions were respectively associated with higher lifetime asthma prevalence (parental marital discord: 16.77% vs 12.2%; p<0.0001, and stressful living conditions: 17.14% vs 11.12%; p<0.0001). Asthma prevalence was higher among those reporting an experience of verbal abuse (16.77% vs 11.46%: p<0.0001), any physical abuse (17.7% vs 11.87%: p<0.0001) or any sexual abuse (20.34% vs 12.16%: p<0.0001), compared to those with no experience of the specific adversity.
ACE and asthma prevalence
High adversity in childhood was associated with greater asthma prevalence, with evidence for a stepwise gradient (table 3). Each ACE category was seen to be positively related to asthma current and lifetime prevalence. Small differences in the asthma PRs were noted for this exploratory gradient between respondents reporting categories 2 and 3 adversity. Highest risks for asthma prevalence were seen among respondents reporting highest ACE compared to those reporting no adversity (PR=2.03 (95% CI 1.94 to 2.14)). Adjusting for race/ethnicity, SES, demographic covariates and state of residence led to proportionally greater attenuations in asthma prevalence (∼30%) among those who reported all dimensions of ACE (adjusted PR (APR)=1.78 (95% CI 1.69 to 1.87) compared to lower categories of ACE. Similar patterns were noted for current asthma with unadjusted PR=2.23 (95% CI 2.09 to 2.36) reducing on covariate adjustment to APR=1.88 (1.77 to 2.006).
Prevalence ratios from Poisson regression models for lifetime asthma by overall ACE score
Analyses for separate ACE dimensions showed positive relationships between childhood adversity and asthma prevalence, adjusted for covariates (table 4). Respondents who reported parental marital discord showed a 37% higher prevalence (PR=1.37, 95% CI 1.32 to 1.43). Adjusted for covariates, PRs reduced by 35%. For current asthma, parental discord was associated with a 40% high prevalence (APR=1.40 (95% CI 1.34 to 1.47). Respondents reporting stressful living conditions in childhood showed a 54% higher asthma prevalence (PR=1.54, 95% CI 1.49 to 1.59), which attenuated by 18% on covariate adjustment. Stressful living conditions were associated with a 47% higher prevalence of current asthma (APR=1.47 (95% CI 1.41 to 1.54)). Respondents reporting verbal or physical abuse showed 38% higher adjusted risks for asthma prevalence (APR=1.38, 95% CI 1.33 to 1.43). Respondents reporting sexual abuse had a 67% higher asthma prevalence (PR=1.67, 95% CI 1.59 to 1.74); covariate adjustment moderately attenuated the estimated risk (APR=1.48, (95% CI 1.42 to 1.55)). Respondents reporting any experience of sexual abuse were noted to have a 53% higher prevalence of current asthma (APR=1.53 (95% CI 1.45 to 1.62)).
Results from Poisson regression models (PRs and 95% CI) for lifetime asthma prevalence by independent ACE dimensions
Additionally, combined experiences of adversity provided greater risks of asthma lifetime and current prevalence, particularly for respondents reporting physical and sexual abuse (APR=2.15 (95% CI 2.009 to 2.31)), verbal and sexual abuse (APR=1.68 (1.48 to 1.90)) and verbal and physical abuse (APR=1.65 (95% CI 1.54 to 1.76); table 5). Among the singular dimensions of asthma, adjusted for covariates and other dimensions of adversity, stressful living conditions and experience of sexual abuse in childhood predicted the highest risks of asthma prevalence (Stressful Living Conditions: APR=1.35 (95% CI 1.24 to 1.46); Sexual Abuse: APR=1.37 (95% CI 1.18 to 1.59)). The risks for asthma among those reporting parental marital discord only (APR=1.24 (95% CI 1.12 to 1.38)) and parental marital discord with verbal abuse (APR=1.26 (95% CI 1.04 to 1.53)) were similar.
Results from Poisson regression models on combinations of adversity experiences
Social determinants of adversity
Higher rates of parental marital discord (separation or divorce) were reported by racial/ethnic minorities (NHB: 36.88%, Hispanic: 28.48% vs NHW: 18.93%; table 6). Hispanics reported higher rates of adversity seen through rates of living in stressful conditions (38.5%) compared to NHW (34.6%) and NHB (32.8%) populations. Higher rates of verbal and physical abuse were also reported by Hispanics (verbal abuse: 36.3%, physical abuse: 34.24%) compared to NHW (verbal abuse: 32.7%, physical abuse: 21.4%) and NHB (verbal abuse: 26.4%, physical abuse: 26.08%). Hispanics reported higher rates of sexual abuse (14.26%) compared to NHB (12.1%) and NHW (12.05%) populations.
Distribution of ACE dimensions by race/ethnicity, income and education in the BRFSS
The lowest income and educational categories reported higher rates of adversity but a clear SES gradient could not be established (see online supplementary table S3). Respondents with less than high school education reported greater parental marital discord (29.28%) compared to college educated respondents (15.82%); similar rates were seen for any physical abuse (<high school education: 30.48% vs college educated: 18.97%). Respondents with lowest incomes (income<US$15 000) reported higher adversity, with small differences between the remaining groups. Lowest income respondents (<US$15 000) reported higher parental marital discord (29.27%) and stressful living conditions (42.6%) compared to respondents with income between US$15 000 and US$35 000 (parental marital discord: 22.78%, stressful living conditions: 35.52%). Those in the lowest income group also reported higher rates of sexual abuse (19.02%) compared to 10.77% in the highest income group.
Discussion
Our study reports three major findings. First, we found a positive relationship between childhood adversity and asthma prevalence that persisted after adjusting for covariates. Second, we observed increasing asthma by categories of adversity, which has not been investigated in previous research. Studies on childhood adversity need to investigate the scope and meaning of this observed gradient, including whether these relationships hold for other health outcomes. Third, separate dimensions of ACE were associated with asthma prevalence, with the strength of the relationship varying by exposure.
The linkages between childhood adversity and asthma may be explained by the physiological evidence on the role of sociobiological and neurobiological mechanisms from stress and adversity. These mechanisms go on to impact immune and inflammatory responses, muscular and nervous system mechanisms, and hormonal and obesity-related mechanisms, all of which relate to asthma.12–15 Studies on asthma prevalence have also attributed some of the disease burden to environmental risk factors, including poor control of pollutants (toxins, dust and mites) in the household.36–42 There is also some evidence on clustering or co-occurrence of risks of disadvantages (socioeconomic, adversity and environmental risks) within households.14 ,36 ,43 ,44 Studies have, in the past, examined the relationships between stress during pregnancy and postpregnancy and asthma among children.9 The scope for connecting pregnancy and early life exposures to childhood adversity was not available in the BRFSS surveys, and needs to be investigated in future analyses.
An aspect needing more examination in studies of adversity pertains to the interrelatedness of dimensions.33 Previous assessments of adversity have shown high degrees of interconnectedness, with the presence of one dimension of adversity increasing the probability of additional ACE exposures (adjusted odds between 2 and 17.7 times, median 2.8).33 In our study, 12% of the respondents reported having experienced all dimensions of adversity and 23% reported the lowest category of adversity—experiencing parental discord and verbal abuse (see online supplementary table S1). This co-occurrence of multiple ACE dimensions also varied by gender. In our study, the reported experience of the overall ACE score was higher among women compared to men (15.08% vs 6.5%), but when sexual abuse was excluded, men (16.2%) reported higher ACE compared to women (12.44%). Lower levels of adversity did not show gender differences.
In addition to understanding the level of co-occurrence of experiences of adversity, examining the role of combinations of adverse experiences may potentially be another area for research in the future. Research needs to examine the health effects of combinations of adversity, and whether certain experiences singularly or in consonance with others have greater impacts than others. Our additional analyses (table 5) showed that experiencing multiple types of abuse (verbal, physical and sexual) can magnify risks for asthma, beyond the experience of a singular adversity dimension.
We adjusted for race/ethnicity and SES in the relationships between childhood adversity and asthma prevalence. Several studies have shown the racial/ethnic and SES patterning of asthma prevalence rates in the USA.44–48 A review on maltreatment highlighted the potential role of parental poverty and low educational achievement.49 Since we did not have data on parental SES or SES of the respondents during childhood, we adjusted for own SES as a proxy for the same. Those in the lowest income and education groups reported higher rates on all dimensions of ACE, but since this information does not necessarily correspond to childhood SES, the patterning of the adversity–asthma relationship by SES remains unclear. Associations of smoking with adversity and asthma may also be another aspect of relevance,27 ,50 with smoking either being potentially driven by experiences of adversity, or as an independent predictor of asthma. Adjusting for smoking in the analyses (see online supplementary table S4) would only be relevant if the attempt was to disentangle the behavioural determinants from neuroendocrinal mechanisms.
Limitations
Findings from this study need interpretation in light of certain limitations. While data from the BRFSS surveys provide tremendous value in surveying major health conditions and behaviours in the USA, these surveys, based on self-reported data and recall, may be prone to several limitations. First, adversity measured through ACE in the BRFSS surveys was self-reported and based on recall. Information was unavailable in the surveys on the exact timing or duration of the exposure. Social desirability bias, limited duration of questioning and the sensitivity of ACE questions may lead to underestimation of the true burden of ACE in the population.51 A study on self-reporting of abuse in surveys has shown that self-reported accounts can be underestimates or overestimates.52 Second, our analysis is based on childhood adversity data available from a subsample of the BRFSS from 10 states over 3 years. No information was available on why these states were chosen. Third, the data were cross-sectional, and hence this analysis only examined associations and does not present a causal argument. Information on age at asthma diagnosis was available based on recall for a subset, but this could not be related to the timing of the experience of adversity. Fourth, BRFSS asthma questions have not been systematically examined for reliability and validity, and diagnostic assessments are needed to compare the strengths and limitations of self-reported physician-diagnosed asthma prevalence in population-level studies. Even as the questions in the surveys on asthma have asked respondents about physician diagnoses, careful assessment is needed on the validity of these questions. Finally, BRFSS did not provide data on parental SES. A key element for future research in the area linking adversity to asthma will pertain to the SES linkages, which may also be relevant from an intervention point of view.
Conclusion
We found positive relationships between retrospectively reported childhood adversity and current and lifetime asthma prevalence. The framework for asthma prevention needs to recognise the importance of childhood adversity as an important risk factor. Greater investigation is needed to understand the specific time periods in which this experience may have an enhanced effect on the risks for asthma, along with linking it to early life exposures.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Footnotes
-
Contributors NB, MMG, IK and SVS jointly conceptualised the study. NB led the data analysis, interpretation and writing of the manuscript. MMG, IK and SVS contributed to the interpretation and writing of the manuscript and provided overall supervision.
-
Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
-
Competing interests At the time of research, NB was a doctoral student at the Harvard School of Public Health and received the Presidential Scholarship at Harvard University (2009–2013). SVS is supported by the Robert Wood Johnson Investigator Award in Health Policy Research; he was also partially supported by the National Institutes of Health Career Development Award (NHLBI K25 HL081275) when the study was conducted.
-
Provenance and peer review Not commissioned; externally peer reviewed.