Second-Generation Research on Adverse Childhood Experiences: Defining and Measuring Childhood Adversity

Background: A large body of empirical work has shown that adverse childhood experiences (ACEs) are prevalent and that they are associated with poor health and well-being over the life course. Despite the proliferation of ACE research, significant gaps in the literature remain. First, economically disadvantaged populations are understudied, even though they are likely to be at a high risk of ACEs. Second, questions remain about how to define and measure ACEs. For example, 10 conventional indicators of child maltreatment and household dysfunction are widely accepted as ACEs, but there may be other childhood adversities such as extreme poverty and community violence that are equally salient. Third, advances in measurement have been hindered by inconsistent item construction and analytic approaches. Fourth, there are also lingering doubts as to whether retrospective accounts of ACEs are reliable.

Methods: This paper addresses the four scientific gaps listed above using data collected from a diverse, low-income sample of women receiving home visiting services (N = 1,241). We assessed the self-reported prevalence of ten conventional ACEs and seven other potential ACEs: chronic family financial problems, food insecurity, homelessness, parental absence, parent/sibling death, bullying, and violent crime. Next, we examined bivariate correlations between ACEs. We then ran bivariate and multivariate regression models that tested associations between ACEs and two health-related outcomes, perceived stress and lifetime smoking, controlling for race and maternal age. Third, we performed two exploratory factor analysis (EFA), one for the 10 conventional ACEs and a second with an expanded set of ACEs. Finally, using data from a subsample of 83 participants, we assessed the test-retest reliability of individual items and total scores for ACE indices.

 Results: As expected, prevalence rates for all ACEs were higher in this sample, compared to studies of more representative samples. With the exception of parent/sibling death, all ACEs were intercorrelated and associated at the bivariate level with two health-related outcomes: perceived stress and smoking (p<.01). In two multivariate models that included 10 items and 17 items respectively, the magnitude of association between ACEs and both outcomes was greatly reduced. EFA results confirmed that conventional ACEs loaded on two factors, child maltreatment and household dysfunction, though a more complex 4-factor solution emerged once other potential ACEs were introduced. All 17 ACEs demonstrated acceptable test-retest reliability (K= .41-.77 for items; 10-item ICC = .82; 17-item ICC= .84), thereby increasing our confidence in the study results.

Conclusions and Implications:  Our results showed that ACEs were especially prevalent in a sample of low-income women. In addition, the observed associations between ACEs and health-related outcomes suggest that the influence of childhood adversity may manifest above and beyond the net effects of poverty. Expanding the ACE framework to include other indicators such as extreme poverty, parental absence, bullying, and violent crime victimization may increase its predictive validity as well as its ecological and cross-cultural validity. Additional second-generation ACE research is needed to build classification schemata and develop standard screening and assessment protocols that can be applied across diverse populations.