Intergenerational Associations of Parent Adverse Childhood Experiences and Child Health Outcomes

This cross-sectional study linked data from the 2012 Southeastern Pennsylvania Household Health Survey (HHS) and the Philadelphia ACE Survey. The HHS was a population-based telephone survey in English or Spanish of residents of Philadelphia and surrounding counties.²³ Used in the survey was a geographically stratified sampling design with separate sampling of landline households and cell phone users. An adult respondent aged ≥18 years was selected randomly from each household. In households with children aged <18 years, a proxy interview about a randomly selected child was conducted with the household member with the most knowledge about the child’s health. Because the proxy adult was the child’s mother or father in 92% of cases, we refer to this person as the “parent” hereafter. The 2012 HHS included 10 018 adult interviews and 2745 child proxy interviews.

The Philadelphia ACE Survey was a follow-up telephone survey to the 2012 HHS developed by the multisectoral Philadelphia ACE Task Force.²⁴ Philadelphia residents who had participated in the 2012 HHS were recontacted between November 2012 and January 2013 to complete the survey. A total of 1784 surveys were completed (response rate = 67%).⁴ 

The sample population for this study was parent-child dyads of children aged <18 years with both 2012 HHS child proxy survey and Philadelphia ACE Survey information (N = 377). We excluded 27 dyads (7%) with missing values on any parent ACE items (n = 11) or covariates (n = 22). The final analytic sample included 350 parent-child dyads.

The Philadelphia ACE Survey questionnaire was used to assess both conventional ACEs, which are childhood experiences related to abuse, neglect, and family dysfunction, as well as an additional set of expanded ACE items used to measure community-based childhood stressors. Conventional ACE questions, adapted from the original ACE study¹ and the Behavioral Risk Factor Surveillance Survey ACE module,²⁵ were used to ask adult participants whether they were exposed to the following 9 household-level adversities during the first 18 years of life: emotional abuse, physical abuse, sexual abuse, physical neglect, emotional neglect, household substance abuse, household mental illness, domestic violence, and having an incarcerated care provider. Parental divorce or separation, included in the original ACE study, was omitted on the basis of previous research showing that this measure did not capture the complexities of parental relationships in this population.²⁶ Expanded ACE items were developed through previous qualitative studies among Philadelphia adult residents^27,28 and were used to capture the following 5 community-level childhood adversities: witnessing violence, racial and/or ethnic discrimination, living in an unsafe and/or unconnected neighborhood, bullying, and living in foster care. In keeping with previous research, responses to each ACE item were dichotomized and the items were then summed to create an expanded ACE score ranging from 0 to 14. We also created a conventional ACE score ranging from 0 to 9 including only the conventional ACE items. In Table 1 we present the ACE questions as well as thresholds used to denote exposure to each adversity. In secondary analyses, we categorized the ACE scores as 0, 1 to 3, and ≥4 for conventional ACE score and 0, 1 to 5, and ≥6 for expanded ACE score on the basis of previous studies^2,26 and the empirical distribution.

We examined 10 binary proxy-reported child health outcomes, including 3 measures of health status, 4 measures of health behaviors, and 3 measures of health care access and use. Health status measures comprised the overall health status of the child, whether the child was ever diagnosed with asthma, and obesity. In keeping with previous research in children,²⁹ we dichotomized overall health status as good, fair, or poor versus excellent or very good because of low prevalence of fair or poor health (n = 19) and the distribution of other health outcomes by health status in our sample (eg, asthma prevalence was 38% among children with good, fair, or poor health versus 16% among children with excellent or very good health). Child obesity was defined as having an age-specific BMI ≥95th percentile. BMI was calculated from proxy-reported weight and height among children aged ≥6 years. Cutoffs for poor health behaviors were based on clinical recommendations and previous research: <5 servings of fruit and vegetables on a typical day,³⁰ any past-month soda consumption,³¹ <7 days per week of at least 30 minutes of physical activity over the past month,³² and past-month excessive television watching (>1 hour per day for children aged 3–4 years; >2 hours per day for older children).³³ Questions about fruit and vegetable consumption, physical activity, and television watching were only asked about children aged ≥3 years. Measures of health care access and use were whether the child had health insurance, had a usual source of health care, and had been examined or treated by a dentist in the past 12 months.

Data on adult and child demographic characteristics were obtained from the HHS. Covariates in the adjusted analyses were selected a priori on the basis of theoretical and previous empirical associations with ACEs and the outcomes, and included adult age (years), sex, and race and/or ethnicity (white, African American, Latino, other) and child age (years) and sex. Adult and child age were modeled as continuous variables; quadratic terms were included in models in which there was evidence of nonlinearity in associations with the outcome. We also ran models including the following indicators for current socioeconomic circumstances that may mediate potential effects of parent ACEs on child health outcomes: household poverty (≤150% federal poverty level) and low parent education (less than high school, high school degree, or technical and/or vocational training versus some college or college degree).

All analyses were performed by using SAS 9.3 (SAS Institute, Inc, Cary, NC). To examine associations between parent ACEs and child outcomes, we fitted separate logistic regression models for each child outcome. Associations were modeled separately with both the conventional and expanded ACE scores. To assess the dose-response pattern, we tested models both using continuous ACE score and categories. Results were consistent with each other, so we present models using continuous scores in the main text. Statistical significance was assessed at α = .05 by using 2-tailed tests.

Table 2 shows sample characteristics. Adult participants were mostly female (80%) and were racially diverse (45% non-Hispanic African American, 39% non-Hispanic white, 10% Hispanic or Latino, and 5% other race and/or ethnicity). Among children, approximately half were female and the mean age was 9 years. As reported by the parent, 22% of children had good, fair, or poor overall health, 18% were obese, and 21% had an asthma diagnosis. The prevalence of poor child health behaviors was high: 89% of children ate <5 servings of fruits and vegetables per day, 48% drank soda within the past month, 55% did 30 minutes of physical activity <7 days per week, and 32% watched excessive television in the past month. Children overwhelmingly had health insurance (97%) and a usual source of health care (98%), and most had been to a dentist within the past year (82%).

Parent ACEs were highly prevalent (Table 2). Seventy-one percent of parents had experienced at least 1 conventional ACE and 19% had experienced ≥4; when the expanded ACE items were included, 85% had experienced at least 1 ACE and 18% had experienced ≥6. The prevalence of specific conventional ACEs ranged from 5% for having experienced physical neglect to 38% for having lived with someone who abused alcohol or drugs (Table 1). The prevalence of expanded ACEs ranged from 2% for having lived in foster care to 42% for having witnessed community violence (Table 1).

In unadjusted logistic regression models (Table 3; Model 1), parent conventional ACE score was related to higher odds of poor health for all 3 measures of child health status. For each additional ACE experienced, the results were as follows: odds ratio [OR] = 1.21 (95% confidence interval [CI]: 1.07–1.37) for good, fair, or poor health status; OR = 1.28 (95% CI: 1.10–1.50) for obesity; and OR = 1.17 (95% CI: 1.03–1.33) for asthma diagnosis. Adjusted results (Table 3; Model 2) show that parent conventional ACE score was related to higher odds of poor child overall health status (adjusted odds ratio [aOR] = 1.20; 95% CI: 1.05–1.36) but not statistically significantly related to obesity or asthma (aOR = 1.17; 95% CI: 0.98–1.40 and aOR = 1.13; 95% CI: 0.99–1.30, respectively).

Compared with the conventional ACE score, results for the expanded ACE score were similar and in fact more robust to covariate adjustment. In adjusted models, for each additional parent ACE, the results were as follows: aOR = 1.19 (95% CI: 1.07–1.32) for good, fair, or poor health status; aOR = 1.14 (95% CI: 1.00–1.31) for obesity; and aOR = 1.17 (95% CI: 1.05–1.30) for asthma (Table 3). This translates into dramatically higher odds of poorer child health across the spectrum of parent ACE exposure, as demonstrated by the models using a categorical measure of parent ACEs, although CIs were wide (see Supplemental Information). Children whose parent had experienced ≥6 expanded ACEs, compared with no ACEs, had over 6 times higher odds of good, fair, or poor health status (aOR = 6.51; 95% CI: 2.02–20.97) or asthma (aOR = 6.38; 95% CI: 1.93–21.13) (Supplemental Tables 4 and 5).

For child health behaviors, only television watching was associated with parent ACEs. In adjusted models, each additional parent conventional ACE was associated with 1.20 (95% CI: 1.06–1.36) times higher odds, and each additional parent expanded ACE was associated with 1.16 (95% CI: 1.05–1.28) times higher odds of excessive television watching. Parent ACEs were not associated with child health care access or use, although this result may reflect the lack of variability in these outcomes in this sample. For all outcomes, results were nearly identical after additional adjustment for parent education and household poverty (Table 3, Model 3).

In a population-based sample of adult-child dyads living in Philadelphia, parent ACE exposure was related to worse overall health status and higher odds of asthma and excessive television watching in their children. Parent ACEs were not related to child diet, physical activity, or health care access or use in this sample.

Previous research in which authors have examined intergenerational influences of parent ACEs on child health beyond infancy is limited. Parent ACEs have been associated with low birth weight and shorter gestational age,^18,33 maladaptive socioemotional symptoms at age 6 months,¹⁸ and poorer physical and emotional health at age 18 months.¹⁹ In addition, maternal childhood abuse has been associated with offspring newborn brain anatomy differences,²⁰ offspring emotional and behavioral problems,³⁴ and worse offspring adjustment.³⁵ 

Our intergenerational findings for offspring health outcomes mirror existing research linking ACEs with poorer health in the same individuals, including obesity and asthma.^1,2,5,36,37 ACEs have also been related to poor self-rated health status among adults, although this has not been the case in 2 recent studies of youth.^1,38,39 Pediatric research shows within-individual associations between child ACE exposure and poorer physical, psychological, and developmental health.^36,39,40 

Sparse within-individual ACE research in which authors examine the health behaviors addressed in our study has related ACEs to poorer diet and physical inactivity.^1,41,42 In a post hoc analysis of our sample, parent ACEs were not related to diet, physical activity, or television and computer time in the parents themselves (data available on request). Similarly, little existing research has examined how ACEs relate to health care access and use. ACEs were more prevalent in uninsured and publicly insured adults in Wisconsin.⁴³ Authors of another study found that ACEs were not related to having health insurance or a regular doctor but were related to barriers to health care such as medical debt or foregoing health care because of cost.⁴⁴ 

Our results using the expanded ACE score were consistent with those in which we used the conventional score, providing evidence of predictive validity of this more comprehensive adversity measure for offspring child health outcomes. Fourteen percent of our sample had no conventional ACEs but at least 1 expanded ACE, suggesting that the conventional score may fail to capture the full spectrum of relevant childhood adversities experienced by diverse urban populations.²⁶ 

Parent ACEs were consistently related to worse offspring health in our sample, but given the survey methodology, we cannot address which specific mechanisms explain how parent ACEs are associated with offspring health. Child health behaviors and health care access and use, two potential mechanisms, were generally not related to parent ACEs in our study. Parent ACEs may affect other aspects of parenting behaviors or parent risk behaviors that were not measured in our study but that influence offspring health. Intergenerational continuity of ACEs, and in particular child maltreatment, has been well documented.¹⁷ More generally, parent ACEs may lead to a “chain of risk” throughout the parent’s life, resulting in stressful or unhealthy living circumstances that could affect offspring health.⁴⁵ For example, maternal childhood abuse is associated with early family life stressors, some offspring interim life stressors,³⁵ offspring family dysfunction,¹⁷ and offspring victimization before age 2 years.⁴⁶ Parent ACEs were also related to household food insecurity in a Philadelphia-based study.¹⁵ Finally, parent ACEs may have detrimental effects on fetal physiologic or epigenetic development, leaving offspring vulnerable to poorer health.^18,19,33 

We also did not measure protective family or community characteristics that may buffer potential effects of parent ACEs on child health. Most parents who were abused during their own childhoods do not later abuse their children.⁴⁷ Research is needed in which authors focus not only on intergenerational risk factors, but also on resilience factors that enable effective parenting and promote child health despite past adversities experienced by parents. Indeed, some potential resilience-promoting factors, such as connected communities or emotional support, can be thought of as the flip side of ACEs.

Strengths of our analysis included the population-based sample and our ability to control for both parent- and child-level confounders. However, our analysis was subject to several limitations. We relied on parent reports of their ACE exposure and the child outcomes, although the ACE and child health questions were asked in separate surveys months apart. As in other studies in which authors use the summary ACE score, we did not account for differences in the severity of different ACEs or possible interactions between ACEs. Although parent ACEs clearly predate child health, our cross-sectional data did not allow for the examination of specific timing of development of the child outcomes. Our relatively small sample may have limited our ability to detect associations, particularly for the health care access and use outcomes with limited sample variability. Our results also may not be generalizable to other populations. Finally, we did not have information on important potential mechanisms such as parenting behaviors or offspring ACE exposure. Particularly given our null results for the health behavior and health care outcomes, other, more proximal family or community factors that mediate or act independently of parent ACEs are likely more important predictors of child health.

Our results suggest that the full scope of health effects of ACEs is not limited to the exposed individual. The lasting effects of childhood adversity are well known, and with our study, we extend these effects intergenerationally, at least in our high-risk urban sample. Future research is needed to determine how generalizable our results are to other populations; hone in on specific mechanisms that drive these associations, as well as buffering mechanisms that modify them; and determine the salience of ACEs experienced by 1 or both parents, as well as other caretakers and household members.

Emphasized in the life course health development framework is that effective health care and health policy require an approach that is integrated across different life stages.⁴⁸ The American Academy of Pediatrics (AAP), in recognition of the importance of early-life adversity and toxic stress for lifelong health, recommends that physicians use an eco-bio-developmental framework to develop active screening for factors causing toxic stress among their patient populations, including ACEs.⁴⁹ In the 2013 AAP 85th Periodic Survey of nonretired AAP members, most endorsed the link between childhood adversity and future poor health, but few reported asking about parent ACEs.⁵⁰ This may represent a missed opportunity for physicians to improve population health by using a 2-generation approach to simultaneously address health effects of parent ACEs on both parent and child.⁵¹ Although parent ACEs cannot be undone because they already occurred, interventions may promote resilience and mitigate any impact of parents’ past experiences on their own and their children’s well-being.^52,–54 

AAP

American Academy of Pediatrics

ACE

adverse childhood experience

aOR

adjusted odds ratio

CI

confidence interval

HHS

Southeastern Pennsylvania Household Health Survey

OR

odds ratio

We thank the members of the Philadelphia ACE Task Force for allowing us to use the Philadelphia ACE Study data.