Introduction
Asthma is a chronic respiratory disease characterized by recurring episodes of airway inflammation, shortness of breath, wheeze, cough, and chest tightness.1 Asthma is one of the most common chronic illnesses in Canada affecting approximately 10.8% (3.8 million) of the total population, with an average of 443 new cases diagnosed every day.2 Asthma poses significant burden to the health care system and substantial adverse effects on health-related quality of life. In 2010–2011, approximately 0.8 million medical visits in Canada were attributed to asthma, constituting a major cause of productivity loss and school absenteeism.2
The etiology and pathogenesis of asthma is complex and not completely understood.1 Current scientific evidence implicates a complex interplay of prenatal and early-life exposures, genetic factors, and environmental hazards in asthma causation pathways.1,Reference Subbarao, Mandhane and Sears3 A growing body of research has identified harmful effects of early-life environmental exposures on asthma including maternal atopy and smoking, preterm birth and bronchopulmonary dysplasia, infections, and malnutrition.Reference Bobolea, Arismendi, Valero and Agustí4 Among these early-life exposures, adverse childhood experiences (ACEs) such as physical or sexual abuse, emotional maltreatment, and witnessing chronic relational violence have been linked to an increased susceptibility to developing asthma in adulthood.Reference Abajobir, Kisely and Williams5–Reference Hughes, Bellis and Hardcastle13 Childhood trauma is thought to leave a lifelong negative impact on respiratory health that is mediated through stress-induced inflammatory pathways and immune dysregulation.Reference Abajobir, Kisely and Williams5–Reference Danese and McEwen14
Evidence indicates that ACEs often occur in clusters.Reference McDonald, Kingston, Bayrampour and Tough12,Reference Hughes, Bellis and Hardcastle13 Consequently, recent studies have analyzed the cumulative effect of exposure to multiple ACEs on asthma onset.Reference Remigio-Baker, Hayes and Reyes-Salvail9–Reference McDonald, Kingston, Bayrampour and Tough12,Reference Wing, Gjelsvik, Nocera and McQuaid15 Exposure to multiple ACEs has consistently been reported to increase the risk of adult asthma in a dose–response manner.Reference Remigio-Baker, Hayes and Reyes-Salvail9,Reference McDonald, Kingston, Bayrampour and Tough12,Reference Hughes, Bellis and Hardcastle13 These findings suggest that ACEs are multidomain constructs encompassing different and overlapping adverse experiences and that ACE combinations are likely to have a cumulative effect on asthma onset.Reference Bhan, Glymour and Kawachi10,Reference Hughes, Bellis and Hardcastle13,Reference Moylan, Herrenkohl and Sousa16
As the number of ACE exposures increases, elucidating their conjoined effects on health outcomes poses some statistical challenges. For example, large sample sizes would be required to estimate the effects of large numbers of potential ACE combinations (e.g., a total of 26 different combinations theoretically would exist when 5 ACEs are to be evaluated, whereas 247 combinations could potentially exist if 8 ACEs are studied), making statistical modeling intractable. An alternative approach is to identify latent factors/constructs underlying diverse ACEs. Using this approach, a large number of ACEs can be parsimoniously represented by a smaller number of underlying latent factors.Reference Fabrigar and Wegener17 Latent factors are unobserved variables that are inferred through mathematical models from other variables and that substantially reduce data dimensionality,Reference Fabrigar and Wegener17 while providing significant insights into the relationship between ACEs and future health.Reference Lew and Xian18 To our knowledge, the use of latent factors of ACEs has not been explored as an alternative to evaluate the relationship between ACEs and adult-onset asthma. Understanding how ACE latent factors influence asthma development in adult life may provide valuable insights for the design of integrative preventive strategies targeting children from families with high vulnerability to ACEs and help to reduce the impact of ACEs on asthma burden.
Using data from the 2013 Alberta Adverse Childhood Experiences survey (ACE survey),Reference McDonald, Kingston, Bayrampour and Tough12,19 the present study evaluated relations among eight types of ACE to identify latent factors of ACEs and assess their association with the development of asthma in adulthood. Our study objectives were (1) to identify latent factors characterizing the correlation structure among ACEs and (2) to evaluate the effect of ACE latent factors on adult asthma development. We hypothesized that ACEs are highly correlated; they can be represented by latent factors and that ACE latent factors associate with asthma in adulthood.
Methods
Study design and data sources
We conducted a cross-sectional secondary data analysis of the 2013 Alberta ACE Survey conducted by the Alberta Centre for Child, Family and Community Research. Reference McDonald, Kingston, Bayrampour and Tough12,19 Briefly, the ACE survey was a computer-assisted telephone survey using a random sample of 1207 adults living in Alberta (response rate 20.9%) that gathered sociodemographic data, information about adversity during childhood (defined as <18 years of age), and poor health outcomes in adulthood, including asthma and other respiratory problems. The survey had eight questions about ACEs that assessed individual exposures to abuse, neglect, and household dysfunction during childhood. Detailed methods for the original survey study are described elsewhere. Reference McDonald, Kingston, Bayrampour and Tough12 Ethics approval for this secondary analysis was obtained from the University of Alberta’s Health Research Ethics Board (Study ID: Pro00083997). A data access agreement was signed with the data custodian (PolicyWise for Children and Families) to protect confidentiality and privacy of the original study participants.
Exposure variables
We obtained data on the following eight individual ACEs from the Alberta ACE survey: verbal insults (frequently experienced verbal insults or threats from an adult or parent in the household), physical abuse (experienced physical injury or bruise caused by a parent or adult in household), sexual abuse (experienced inappropriate sexual advances or contact by an adult or someone older to them by 5 years or more), witnessing violence against mother, alcohol/drug use at home, living in a household with someone with depression/mental illness, living in a household with someone with chronic illness/physical disability, and parental separation/divorce. Data on ACEs occurring before the age of 18 years were self-reported and measured on a binary scale (absent/present). No information about duration and severity of ACE exposure was collected in the survey. 19
Study outcome and covariates
Participants in the Alberta ACE survey were asked questions about adult health diagnostic indicators (i.e., “Have you been diagnosed with any of the following health problems as an adult?”) and if they received treatment or not (i.e., “If Yes, did you receive treatment?”). The primary outcome of interest was a self-reported diagnosis of asthma as an adult (>18 years of age). We obtained data on characteristics of the study population including sex, age, area of residence in Alberta (Edmonton/Calgary/other), and country of birth (Canada/elsewhere).
Statistical analysis
We described the characteristics of the study population using frequencies and percentages for categorical variables and mean with standard deviations for continuous variables. We performed an exploratory factor analysis to evaluate the underlying correlation structure of the eight ACEs. We used the bivariate polychoric correlation matrix as input and applied the maximum likelihood method for extracting the factor solution. Correlation values were interpreted as very high (0.90–1.00), high (0.70–0.89), moderate (0.50–0.69), low (0.30–0.49), and negligible (0.00–0.29). Reference Hinkle, Wiersma and Jurs20 The adequacy of the sample for factor analysis was estimated using the Kaiser–Meyer–Olkin test. Reference Meyers, Gamst and Guarino21
We first used a scree plot to generate a plausible range of factors and then Bayesian information criterion (BIC) Reference Preacher, Zhang, Kim and Mels22 and Akaike’s information criterion (AIC) Reference Preacher, Zhang, Kim and Mels22 to determine the number of ACE latent factors to be retained for subsequent analysis. Reference Preacher, Zhang, Kim and Mels22 We applied a varimax Kaiser normalization rotation to improve interpretability of results and to pursue orthogonality of factors. Reference Meyers, Gamst and Guarino21 Factor loading coefficients were interpreted as the correlation between individual ACEs and the extracted latent factors. We used a path flow diagram to visually represent the relationship among the eight types of ACE, the latent factors derived from the factor analysis, and how the underlying latent factors were associated with a diagnosis of asthma in adulthood. Predictive factor scores (a linear combination of factor loadings and original variables) were to be used in logistic regression models.
We built a logistic regression model using predictive factor scores for the ACE latent factors retained and calculated odds ratios (ORs) with 95% confidence intervals (CIs) having self-reported adult asthma (yes/no; the nonasthma group was chosen as the reference group in the calculation of the ORs) as the outcome variable and resulting latent factors as explanatory variables while adjusting for sex (1 = male/0 = female) and being born in Canada (1 = yes/0 = no). Model diagnostics were run to evaluate the goodness of fit of the regression model. We estimated the minimally detectable OR for the survey population (n = 1207) using the lowest prevalence of ACE exposure (physical abuse: 11%) and the prevalence of asthma in the absence of ACE (∼7%) for a power of 80% according to formulas of power and sample size calculations for logistic models. Reference Demidenko23 Our study sample had a statistical power of 80% to detect an OR of 2.17.
We followed the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for reporting study design and results. Reference Vandenbroucke, von Elm and Altman24 All statistical analyses were conducted using STATA Data Analysis and Statistical Software (Version SE 14.2; StataCorp LLC, College Station, Texas, USA).
Results
We analyzed ACE survey results from the 1207 completed interviews. The mean age of survey respondents was 51 years (range: 17–94 years). Female-to-male ratio was 1.03. The response rates for ACE questions were very high, ranging from 98.9% to 99.7%. About 80% of respondents were born in Canada, 67% resided in the two main metropolitan areas of Edmonton and Calgary, whereas 33% had residence elsewhere in Alberta. About 9% (n = 106) of the survey population reported being diagnosed with asthma in adulthood. Alcohol/drug abuse at home (21%) was the most commonly experienced ACE, and the least frequently reported ACE was physical abuse (11%). Table 1 describes the key characteristics of the survey population related to ACEs and the prevalence of individual ACEs. For detailed descriptions of the study population, please refer to the original study by McDonalds et al. Reference McDonald, Kingston, Bayrampour and Tough12
Table 1. Sociodemographic characteristics and prevalence of adverse childhood experiences reported in the 2013 Alberta Adverse Childhood Experiences survey (n respondents = 1207)
Results from bivariate polychoric correlation analysis revealed that all eight ACEs were positively correlated, although with different degrees of correlation.Reference Hinkle, Wiersma and Jurs20 For example, a high correlation was observed for physical abuse and verbal insults (0.87), moderate correlations for witnessing violence against mothers and physical abuse (0.69), sexual abuse, and verbal insults (0.52); and negligible correlations for chronic illness/physical disability at home and parental separation/divorce (0.11). Detailed polychoric correlation coefficients including the absolute number of cases per dual exposures for the eight ACE types are presented in Table 2.
Table 2. Bivariate polychoric correlation matrix showing correlation coefficients and total number of cases (n) for the eight ACEs* (n = 1207)
* This matrix was the input for factor analysis.
Factor analysis based on a bivariate polychoric correlation matrix yielded four ACE latent factors. We retained the four ACE latent factors after interpreting the scree plot criteria (Fig. 1), communalities, and the BIC and AIC values calculated for models that sequentially included the four ACE latent factors. The corresponding BIC/AIC values for the models including one to four latent factors were 924.6/883.8, 334.2/258, 178.3/71.6, and 187.5/55.5, respectively. Thus, BIC favored a three-latent-factor model, whereas AIC favored a four-latent factors model. In the three-latent-factor model, the communality of sexual abuse was 0.32 (very low), whereas the communality in the four-latent-factor model increased to 0.78. The variable-related communality is the variance in the measured variable that is explained by all the factors in the model. Thus, the low communality of sexual abuse in the three-latent-factor solution suggests that this variable is not part of the three domains suggested by the factor model. The inclusion of the four-factor substantially increased the communality for sexual abuse, suggesting that the four-factor-latent model is better compared to the three-latent-factor solution.
Fig. 1. Scree plot of latent ACE factors extracted from factor analysis.
Fig. 2. Path diagram of the relationship among the eight types of ACE, latent ACE factors, and adult-onset asthma.
As a result of combining all criteria, we retained the four-latent-factor model for further analysis. Table 3 shows the factor loadings, communality, and uniqueness of individual ACEs after rotation of the retained factors. Factor 1 – termed relational violence – had the highest positive loadings (i.e., a high correlation between the variable and the latent factor after controlling for other factors in the model) on physical abuse (0.84), verbal insults (0.85), and witnessing violence against mothers (0.52). Factor 2 was closely associated with alcohol/drug use at home (0.71), witnessing violence against mothers (0.67), and parental separation/divorce (0.60) and was termed as negative home environment factor. Factor 3 was closely associated with living in a household with someone with depression/mental illness (0.54) and chronic illness/physical disability (0.54) and was termed as illness-at-home factor. Factor 4 was mainly associated with sexual abuse (0.79) and was termed sexual abuse. Witnessing violence inflicted to the mother was associated with both factors 1 and 2. Chronic illness/physical disability and parental separation/divorce showed communalities below 0.5 (0.30 and 0.43, respectively), indicating that they were not well explained/represented by the latent factors they were related to. Fig. 2 shows a path diagram of the relationship among the eight types of ACE, latent ACE factors, and adult-onset asthma.
Table 3. Matrix of the rotated* factor loadings for the four-latent-factor solution from the eight ACEs
* We used the Kaiser orthogonal varimax method of rotation. After rotation, the proportion of variance was 36.4% for factor 1, 32.2%; for factor 2, 15.9% for factor 3, and 15.5% for factor 4. The overall Kaiser–Meyer–Olkin (KMO) value was 0.835. The minimum KMO value was 0.75 for Chronic illness/physical disability at home (a value of 0.70 or above in considered adequate for factor analysisReference Hinkle, Wiersma and Jurs20).
a Factor loadings represent the correlation between the variable and the latent factor after controlling for other factors in the model. Values range between −1 and 1 (although in oblique rotations, these coefficients can be greater than 1 or less than −1).
b Communality represents the variance in the measured variable that is explained by all the factors in the model.
c Uniqueness represents the unexplained variance in each individual item that is not explained by the factors in the model.
Results from the logistic regression analysis showed that three out of the four ACE latent factors were associated with a self-reported diagnosis of asthma in adulthood. Table 4 shows the logistic regression model for the prediction of an asthma diagnosis in adulthood. Adjusted ORs by sex (male/female), area of residence, and country of birth (Canada/elsewhere) indicated that factor 4/sexual abuse (OR: 3.23; 95% CI: 1.89, 5.53), factor 1/relational violence (OR: 1.99; 95% CI: 1.17, 3.38), and factor 2/negative home environment (OR: 1.86; 95% CI: 1.03, 3.35) were significant predictors of a diagnosis of asthma in adulthood, whereas factor 3/illness at home (OR: 1.38; 95% CI: 0.75, 2.56) did not show such association. For comparative purposes, we estimated the OR for the logistic model including the eight ACEs as explanatory variables adjusted by sex and born (yes/no) in Canada, and corresponding variance inflation factors. The results are presented in Supplementary Material (Supplementary Table S1).
Table 4. Logistic regression model for the prediction of adult-onset asthma using ACE latent factors 1, 2, and 3 derived from the exploratory factor analysis
Variance inflation factors (VIFs): factor 1 = 1.08, factor 2 = 1.07, factor 3 = 1.02, factor 4 = 1.02. These VIFs can be compared with those including the eight ACEs as explanatory variables in Supplementary Table S1. VIF values indicate how much the variance of a coefficient is “inflated” because of linear dependence with other predictors.
Factor scores are estimated from a linear combination of observed variables and factor loadings as a by-product of the factor analysis. Conceptually, factor scores represent the degree to which each individual scores high on the group of items that load high on a factor (See Supplementary Table S2).
The ORs can be interpreted as the odds of asthma increase as the factor loadings increase. An OR above 1 and excluding the 95% CI indicate that the factor loadings were higher in the asthma group compared to the reference group (see Supplementary Table S2).
** Adjusted for sex (male/female) and country of birth (Canada/elsewhere); goodness of fit (Hosmer–Lemeshow test) P-value = 0.133.
Reference category: no asthma.
a Model statement for unadjusted OR: Outcome: asthma (0 = no/1 = yes): log(odds) = Factor 1 + Factor 2 + Factor 3 + Factor 4. Goodness of fit (Hosmer–Lemeshow test) P-value = 0.15
b Model for adjusted OR: log(odds) = Factor 1 + Factor 2 + Factor 3 + Factor 4 + sex (0 = female/1 = male) + born in Canada (0 = no/1 = yes). Factors 1 to 4 are continuous variables represented by factor scores; goodness of fit (Hosmer–Lemeshow test) P-value = 0.133.
Discussion
We conducted a secondary data analysis of the 2013 Alberta ACE survey to evaluate the relationship between latent factors underlying childhood adversities and an increased risk of asthma diagnosis in adulthood. Our study confirms existing evidence that different types of ACEs frequently exist in clusters. Reference Iniguez and Stankowski11–Reference Hughes, Bellis and Hardcastle13,Reference Dong, Anda and Felitti25 We found that the eight ACEs evaluated in our study had various degrees of correlation among them. It has been previously reported that exposure to multiple types of ACEs compounds the risk of adverse health outcomes. Reference Remigio-Baker, Hayes and Reyes-Salvail9,Reference Iniguez and Stankowski11–Reference Hughes, Bellis and Hardcastle13 ACE management programs need to take into consideration the high likelihood of coexistence of various types of adversities during childhood and design multimodal interventions to prevent the negative impact of ACEs on adult health.
Sexual abuse was identified to be a different exposure compared to the other ACEs. The relationship between childhood sexual abuse and poor health in adult life has been well established Reference Hughes, Bellis and Hardcastle13 ; however, most studies have grouped sexual and physical abuse altogether. In our study, experiences of sexual abuse during childhood were identified as a strong predictor of asthma in adulthood. Childhood sexual abuse is an early-life traumatic event with important consequences to the hypothalamic–pituitary–adrenal axis. The trauma generated by these experiences, particularly if they are chronic and pervasive, can trigger specific immunological and neuroendocrine imbalances and lead to different impacts on inflammatory markers associated with asthma. Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli26
Results from the exploratory factor analysis suggest that childhood experiences of relational violence – either directly inflicted to the child or indirect experiences of witnessing violence – are highly correlated and together are significant predictors of adult asthma. These findings are consistent with previous studies suggesting that exposure to violent behaviors during childhood (e.g., domestic violence or parental interpartner violence) is linked to increased susceptibility for asthma development in adulthood. Reference Abajobir, Kisely and Williams5–Reference Goodwin, Wamboldt and Pine8 The biological mechanisms through which childhood experiences of violence continue to impact health later in life are not well understood. It has been hypothesized that ACEs impact respiratory health through stress-induced inflammatory pathways and immune dysregulation Reference Exley, Norman and Hyland6,Reference McDonald, Kingston, Bayrampour and Tough12 that trigger a myriad of behavioral and biological responses. Reference Nurius, Green, Logan-Greene and Borja27–Reference Segerstrom and Miller29 Studies suggest that exposures to early-life stress can have a deleterious impact on biological stress responses, Reference Segerstrom and Miller29 immune system, and airway development which, in turn, may lead to symptoms of airway hyper-responsiveness in adulthood. Reference Nurius, Green, Logan-Greene and Borja27,Reference Coogan, Wise and O’Connor28 A recent literature review presented a biological embedding model that posits that exposure to stress in childhood creates chronic pro-inflammatory responses and endocrine dysregulations that manifest as risky health behaviors and morbidities in adulthood. Reference Segerstrom and Miller29 The combination of experiencing both first- and second-hand violence in childhood may reflect a milieu of physical and psychological stress that exerts lingering damage and results in poor health outcomes later in life. Future work in this area should further explore biological markers associated with ACE-mediated psychophysiological stress and their role in the ACE-adult asthma causal trajectory.
A negative home environment was also found to be strongly associated with a diagnosis of asthma in adulthood. The fact that diverse ACE types (i.e., witnessing violence against mothers, parental separation/divorce and alcohol/drug use at home) were highly to moderately correlated with each other and that they had a unidimensional effect altogether suggests that the correlation structure underlying several ACEs is complex and that estimation of individual and combined effects of ACEs can be challenging due to collinearity issues. Therefore, classification of ACEs as unique individual exposures or independent groups of experiences may be unrealistic. Our analysis suggests that conceptualization of ACEs based on latent factors provides an alternative and comprehensive approach to understanding the relationships between ACEs and their long-term impact on health outcomes. Future ACE studies can use a similar approach to explore other negative health outcomes commonly associated with ACEs such as heart disease, hypertension, and mental distress. Reference Iniguez and Stankowski11,Reference Nurius, Green, Logan-Greene and Borja27 Another area that merits further attention is exploring the impact of other related asthma triggers such as maternal smoking, and socioeconomic status Reference Subbarao, Mandhane and Sears3 on the relationship between a negative home environment and a diagnosis of asthma in adulthood.
Other studies have also used factor analysis and similar techniques to create ACE clusters with similar factorial structure. Reference Ford, Merrick and Parks30,Reference Scott, Burke and Weems31 Only a few studies have evaluated the association between ACE clusters and specific long-term outcomes in adulthood (e.g., cancer Reference Brown, Thacker and Cohen32 and mental health disorders Reference Green, McLaughlin and Berglund33 ). As different factor solutions have been suggested to fit the data in these studies, it is expected that they influence how ACE clusters associate with long-term health outcomes. Our study adds novel information to this area of research as we identified a four-factor solution for ACEs that account for significant associations with asthma in adulthood. This finding highlights the need to identify common risk factors among specific adverse experiences that can convey useful information for targeted prevention and mitigation interventions.
The potential use of latent factors of ACEs as predictors of adult-onset asthma has implications for clinical practice. To date, clinical prediction models for asthma have mainly focused on genetic and environmental risk factors such as family history of asthma, parental atopy, recurrent wheezing, and maternal smoking. Reference Miller, Chen and Parker34–Reference Castro-Rodríguez, Holberg, Wright and Martinez36 Researchers developing asthma prediction tools should consider the inclusion of psychosocial stressors such as ACEs as potential asthma predictors for a comprehensive assessment of future asthma risk. It would be interesting to compare the accuracy of traditional asthma prediction models with a clinical tool incorporating evaluation of ACEs and other psychosocial stressors. Comprehensive asthma prediction tools may assist clinicians to devise timely preventive interventions and reduce asthma incidence in adult life. Reference Balemans, van der Ent and Schilder37
Our study had several strengths that included the use of population-based survey data that captured ACEs in a sample from the general population in Alberta. Our logistic regression analysis had a statistical power of 80% to detect an OR of at least 2.17. Although some of the effect sizes were smaller than 2.17 (i.e. OR: 1.86 for negative home environment), it is still clinically important to identify effects smaller than this statistical threshold. Moreover, the large sample size and the correlation structure among ACEs were adequate for investigating latent factors (as indicated by the Kaiser–Meyer–Olkin test). Using factor analysis methods, eight ACE types were reduced to four latent factors which allowed the use of factor scores into a prediction model of adult asthma while avoiding multicollinearity problems and the inclusion of too many parameters related to double and triple interactions. Reference Preacher, Zhang, Kim and Mels22
Our study had some limitations that should be considered when interpreting the results. Due to the cross-sectional nature of the study, caution should be observed while making inferences based on the study results. The use of retrospectively collected survey data has inherent potential for misclassification and recall bias. Reference McDonald, Kingston, Bayrampour and Tough12 However, our findings are supported by existing literature Reference Abajobir, Kisely and Williams5–Reference Bhan, Glymour and Kawachi10 and have implications for future ACE research, ACE management policies, and clinical practice. The 2013 Alberta ACE survey12 had a low response rate that is common among telephone surveys conducted in Alberta. Reference Kingston, McDonald and Tough38 It has been argued that a low participation rate does not necessarily imply a high risk of nonresponse bias, Reference Galea and Tracy39 but that the difference between participants and nonparticipants would determine the magnitude of bias. Authors of the 2013 Alberta ACE survey reported that the characteristics of the study sample reflected the sociodemographic profile of the province, strengthening the generalizability of the findings.
The ACE survey did not collect information about the temporality and duration of ACEs, and therefore, we were not able to evaluate whether exposures to ACEs at early ages were more likely among individuals reporting asthma in adulthood.
There is no consensus on the best approach to assess ACEs. Reference McKinney, Harris and Caetano40 The 2013 Alberta ACE survey used self-reported information to assess the prevalence of eight ACE types. Self-reported measures of childhood abuse are likely to be underreported due to hesitance to disclose sensitive information. Reference McKinney, Harris and Caetano40 This characteristic is common in social/psychological research. For example, childhood sexual abuse is likely to be underreported, especially if the child’s parents are the main perpetrators. Reference Cohen, Canino, Bird and Celedón7 It is possible that underreporting of sexual abuse in our sample accounts for the modest relationship with relational violence (factor 1). Other factors that may influence the accuracy of ACE reporting include time passed since the ACEs occurred, the frequency and intensity of adversity experience, and the type of abuse suffered. Reference McKinney, Harris and Caetano40 It is unknown whether these characteristics may have affected the ACE survey results. The 2013 Alberta ACE study was a telephone survey, administered by trained interviewers, and confidential which is likely to have facilitated the reporting of ACEs. Reference McDonald, Kingston, Bayrampour and Tough12 Further studies are needed to find optimal approaches to assess childhood adversities in research and clinical context.
Asthma was defined in the 2013 Alberta ACE survey by self-reported diagnosis as an adult. Defining asthma in epidemiological studies is a well-known challenge as it is prone to misclassification. Reference Pekkanen and Pearce41 However, self-reported diagnosis of asthma has been frequently used in population-based studies as a reliable and cost-effective alternative to lung function measurements which may not be readily available in every setting. Reference Remigio-Baker, Hayes and Reyes-Salvail9,Reference Bhan, Glymour and Kawachi10,Reference Luo, Nkoy, Stone, Schmick and Johnson35,Reference Pekkanen and Pearce41,Reference Toren, Palmqvist and Lowhagen42 Therefore, the ACE survey definition of asthma was appropriate to explore the relationship between ACEs and adult asthma.
The 2013 Alberta ACE survey did not collect important information about the characteristics of the home environment when the ACEs occurred. Therefore, we were not able to explore the role of important mediators of the relationship between ACE and asthma in adulthood (e.g., socioeconomic status, nutrition, home crowding, parental smoking, or nutrition status in childhood). Data on sociodemographic characteristics (e.g., socioeconomic status, education, and employment) were collected at the time of survey and may not reflect childhood exposures. Hence, we were not able to account for these variables in our analysis.
Using data from the 2013 Alberta ACE survey, the present study evaluated the relationship between latent factors of ACEs and asthma in adulthood. We found that childhood experiences of sexual abuse, relational violence, and a negative home environment are strongly predictive of asthma in adulthood. These findings provide valuable insights into the developmental origins of adult asthma and may help design effective ACE management strategies targeting children at risk of experiencing poor respiratory health in adulthood. Conceptualization of ACEs based on latent factors provides a valid alternative approach to understand ACE effects and merits attention in future studies exploring ACEs and their negative effects on long-term health.
Supplementary Material
To view supplementary material for this article, please visit https://doi.org/10.1017/S2040174419000886
Acknowledgments
This study used deidentified data from the Alberta Adverse Childhood Experiences study, accessed through the Secondary Analysis to Generate Evidence program, an initiative of the PolicyWise for Children and Families to ensure that existing datasets can be fully utilized to generate new research findings. The opinions, results, and conclusions reported are those of the authors. No endorsement by PolicyWise for Children & Families or any of its funders or partners is intended or should be inferred. We thank the Population Research Laboratory, University of Alberta, who is the data producer of the ACE survey 2013 on behalf of PolicyWise for Children & Families.
Financial Support
This work was supported by PolicyWise for Children & Families through a Secondary Data Use Grant (RES0041222) and the Lois Hole Hospital for Women through the Women and Children’s Health Research Institute. The funding agencies did not take part in study design; analysis and interpretation of data; writing of the paper; and decision to submit it for publication.
Conflicts of Interest
None.
