A Short Adverse Experiences Measure Among Mothers of Young Children

Data came from cross-sectional surveys of biological mothers of children aged <4 years conducted in a pediatric emergency department in Philadelphia, PA in January 2011 through March 2020 as part of the Children's HealthWatch study.²²  This analysis was limited to biological mothers to reduce variation by caregiver role and potential biological or epigenetic mechanisms.^3,23  Surveys were conducted verbally in English or Spanish while participants were waiting to be seen by a healthcare provider. Specifically for the ACE questionnaire, participants were given the choice of reading and filling out the questionnaire on paper if they preferred. During a pilot phase, participants were additionally given the choice of completing the ACE questionnaire using audio computer-assisted self-interview software (ACASI). Participants were reminded during the interview that they were free to choose at any time whether to answer each question or to continue the survey. Interviewers were trained to reduce the likelihood of distress (eg, maintaining neutrality, being cognizant of body language, reiterating the right to refuse to answer requestions and that the same questions are asked of all participants) and in appropriate response if a participant should become distressed, including seeking the aid of provider staff. At the close of the survey, participants were given the option of receiving information on local resources for a variety of specific services, including physical, mental, and dental health; social work; domestic violence; legal; public benefits assistance; assistance with basic needs such as food, diapers, and utilities; and educational services. Mothers of children with severe injury or illness were excluded, including any child who had been triaged as level 1 on the Emergency Severity Index,²⁴  and surveys were paused or halted as needed to avoid interfering with the child’s medical care. Participants were compensated with gift cards. The study received Institutional Review Board approval. Caregivers gave verbal consent to participate. Caregiver names were not collected as part of the survey, and written documentation of consent was waived to protect participant privacy. All child data were deidentified for storage and analysis. Out of 4150 biological mothers administered the ACE questionnaire, we excluded responses for 103 (2.5%) with incomplete ACE information and 48 (1.2%) with missing covariate information, resulting in an analysis sample of 3999 mothers.

The shortened ACE measure developed by Wade et al²⁰  comprises 2 of the questions from the BRFSS 11-item ACE questionnaire. The BRFSS questionnaire assesses 2 dimensions of adversity—abuse and household stressors—experienced when the participant was age <18 years.¹⁹  The 2 questions chosen for the short measure were the most frequently reported items in each of the 2 dimensions, assessing emotional abuse (abuse) and problematic household alcohol abuse (household stressor). Compared with reporting 4+ ACEs on the full 11-item scale, sensitivity of the short measure in Wade et al’s sample was 99% and specificity was 66% when high adversity was defined as experiencing at least 1 of these 2 adversities.

For this analysis, we modified the measure for the 10-item Children’s HealthWatch ACE questionnaire,²²  which phrases all questions as “yes or no,” combines some items asked separately in the BRFSS, and—like the original ACE questionnaire¹ —includes questions about emotional and physical neglect in addition to abuse and household stressors¹  (Table 1). For the short measure, we defined high adversity as an affirmative response to at least 1 of the following questions: “Did a parent or other adult in the household often or very often swear at you, insult you, put you down, or humiliate you or act in a way that made you afraid that you might be physically hurt?” (abuse) and “Did you live with anyone who was a problem drinker or alcoholic or who used street drugs?” (household stressor).

To assess convergent validity—how related a measure is to theoretically connected constructs—we examined associations of the 2-item and standard measures with binary measures of maternal, household, and child health outcomes. Maternal outcomes were maternal health status (fair or poor versus good or excellent)²⁵  and depressive symptoms, defined as affirmative response to 2+ questions on the 3-question Kemper scale.²⁶  Household food insecurity was defined using standard coding of the validated 18-item Household Food Security Survey Module.^27,28  Child outcomes were mother-reported fair or poor health status,²⁵  risk of obesity, and developmental concerns. Risk of obesity, taken from the child’s electronic health record from the visit where the survey was conducted, was defined as weight-for-height >90th percentile.^29,30  If length data were not available, weight-for-age >95% was used. Developmental concerns were defined as 1 or more parental concerns on the Parents’ Evaluation of Developmental Status.³¹ 

We assessed sensitivity, specificity, and convergent validity of the shortened ACE measure compared with the standard of 4+ ACEs on the full 10-question questionnaire.¹  Sensitivity, the degree to which mothers who reported high adversity on the standard scale were correctly identified by the short measure, was calculated as the proportion of mothers who reported 4+ ACES on the full scale who were classified by the short measure as having experienced high adversity (ie, [number of mothers reporting high adversity on the short measure]/[number of mothers reporting 4+ ACEs on the full scale]). Specificity, the degree to which mothers who did not report high adversity on the standard scale were correctly identified by the short measure, was calculated as the proportion of mothers who reported <4 ACES on the full scale who were classified by the short measure as not having experienced high adversity (ie, [number of mothers not reporting high adversity on the short measure]/[number of mothers reporting <4 ACEs on the full scale]). We also assessed sensitivity and specificity stratified by maternal combined race and ethnicity (Latina, Black non-Latina, white non-Latina). Small sample size precluded evaluation among mothers of other racial and ethnic groups. Ninety-five percent confidence intervals were calculated for measures of sensitivity and specificity using the conservative “exact” method.^32,33 

Convergent validity was assessed by separately estimating associations of the 2-item and standard measures with maternal, household, and child health outcomes using adjusted logbinomial regression. Models were adjusted for child age (months) and sex assigned at birth (male or female); and maternal self-reported race and ethnicity (Latina, Black non-Latina, white non-Latina, other [includes Asian, American Indian or Native American, “Other”] or multiple non-Latina race and ethnicity) and education (≤high school or >high school). Associations of high maternal adversity—measured using the short measure—with poor outcomes would indicate good convergent validity, particularly if these associations were similar to associations estimated using the standard measure.

The participation rate among eligible approached caregivers was 86%. Characteristics of survey participants excluded from this analysis because of missing information (n = 151) were similar to those of included participants (Supplemental Table 5). Mothers were racially and ethnically diverse (54% Latina, 35% Black non-Latina, 8% white non-Latina), 88% were United States-born, and 36% had education beyond high school (Table 2). Ninety-four percent of children were publicly insured.

Fifty-one percent of mothers reported at least 1 ACE and 15% reported 4 or more. Thirteen percent of mothers reported emotional abuse and 16% household substance use (Table 1). Twenty-three percent reported high adversity using the short measure (ie, emotional abuse and/or household substance use), including 6% who reported both of these experiences. Thirty-two percent of mothers completed the ACE questionnaire verbally, 33% using ACASI, and 13% using paper self-administration; 22% either used multiple modes or had missing information for mode of administration. The percent reporting 4+ ACES on the full questionnaire or reporting high adversity using the short measure did not vary statistically significantly by mode of administration (Supplemental Table 6).

Compared with 4+ ACEs on the full questionnaire as the standard, the 2-item measure had sensitivity 88% ([95% confidence interval] 87%–89%) and specificity 90% (87%–92%). In stratified analyses, sensitivity was highest among Latina mothers (90% [89%–92%]), followed by Black Non-Latina (87% [85%–89%]) and white non-Latina (81% [76%–86%]), whereas specificity was highest among white non-Latina mothers (93% [85%–98%]), followed by Black non-Latina (91% [87%–95%]) and Latina (87% [81%–91%]) (Table 3).

The 2-item measure demonstrated good convergent validity in this sample for the outcomes examined (Table 4). In adjusted models, high adversity was associated with all poor maternal, household, and child outcomes examined except for child risk of obesity. Compared with estimates using the standard measure, estimates using the short measure were identical in terms of statistical significance, but point estimates were slightly lower for most outcomes.

A modified version of a previously developed 2-item short ACE measure demonstrated good sensitivity, specificity, and convergent validity in a racially and ethnically diverse sample of mothers, primarily with low incomes, seeking care for their young children in an urban pediatric emergency department. Wade et al suggest that the shortened measure may be incorporated into a 2-step screening process where the remainder of the ACE questionnaire is administered to individuals who screen positive on the short measure.²⁰  Our results suggest that, depending on the context-specific goals of collecting parent ACE information, this approach may be appropriate in populations and settings similar to our sample.

Some researchers and practitioners have argued that given the strong impact of ACEs on health and well-being, screening for ACEs is an important component of effective health care.^14,34  They have argued that ACE information provides crucial context for understanding patient populations’ physical and mental health and aids in designing trauma-informed interventions and making appropriate referrals. A handful of qualitative studies support this, with caregivers reporting that they are willing to provide ACE information and that doing so may help elucidate family needs and identify appropriate care.^35,36  However, although care provision may include decisions about how to allocate limited resources, individual patients should not be denied beneficial services based on the outcome of ACE screening. In the case of assessment of parent ACEs in pediatric settings, there is the risk of perceived or actual blaming of parents for their child’s behavior or health, undermining effective treatment and perpetuating discrimination.^9,37,38  Parents may also be concerned that they may divulge information that will trigger reporting to a child welfare agency. These concerns may contribute to underreporting of ACEs, which could in turn lead to under-provision of services if receipt of the services is tied to screening results. Administering the questionnaire may also retraumatize individuals who have experienced adversities, and care must be taken to provide adequate training for providers administering the questionnaire as well as a plan for appropriate follow-up should administration cause distress.

Another challenge is the lack of evidence-based interventions specifically to address ACEs, including parent ACEs. In a recent review, Gupta et al⁸  provide a framework for addressing parent ACEs in pediatric settings that incorporates both universal and targeted interventions. Universal interventions may include parent education programs in which ACE screening is incorporated into broader conversations with parents about the impacts of ACEs, common parenting strategies for coping, proactive identification of resilience factors, and connection with supportive resources. Of note, some universal interventions may be implemented without incorporating screening for parent ACEs. Targeted strategies for families identified as having higher risk may include provision of or referral to more intensive mental health or parenting support services.

Implementing patient or caregiver ACE screening also poses challenges for provider workflow and compensation. Some studies have shown that administering the questionnaire takes little time for the majority of patients and that individuals find providing this information acceptable.¹²  However, surveys have found pediatric providers are hesitant because of discomfort with the sensitive nature of the questionnaire and the time required to administer it.¹¹  New compensation models, including California’s recently implemented ACEs Equity ACT (SB 428), which mandates insurance compensation for ACE screening, help address time-constraint challenges.³⁹  Our results suggest that when thoughtfully used, the short ACE measure may be an appropriate tool to help address these concerns.

Given these varied considerations, we believe the greatest utility for the short measure is in clinical or research settings where ACE information provides important context but where in-depth, individual-level information from all individuals is not required. In this sense, the short measure may serve as a low-barrier screener to provide population-level information to inform program or policy planning. The good convergent validity of the short measure in our sample is encouraging in this respect, as it suggests the short measure serves as an adequate proxy for high ACE exposure in associations with outcomes. This includes maternal and household outcomes representing mechanistic pathways that may serve as important intervention targets. We recommend following up with the full ACE questionnaire in situations where a greater degree of accuracy or depth of information is needed. It is also important to clearly explain the purpose of the questionnaire before administration.

Compared with the BRFSS sample in which the short measure was developed, in this study, sensitivity was somewhat lower (88% vs 99%), whereas specificity was substantially higher (90% vs 66%). These differences likely resulted from underlying ACE exposure variations²¹  as well as differences in the BRFSS and Children’s HealthWatch ACE questionnaires, such as the frequency cutoff for reported emotional abuse (“often or very often” in Children’s HealthWatch versus “more than once” in BRFSS). Although our sample was limited to a single urban site, limiting generalizability, our results support applicability of the short measure for assessing maternal ACEs among racially and ethnically diverse populations in pediatric health care settings. Of note, our sample also included both United States-born and immigrant mothers, although we did not assess differences by maternal nativity because of sample-size constraints. We relied on mothers’ retrospective self-report of ACEs that may be subject to recall bias or other reporting errors. Future analyses incorporating additional important sources of childhood adversity, such as community violence, racial discrimination, and bullying, are needed.^40,41  In our assessment of convergent validity, we were not able to incorporate information on ACEs experienced by other adults in the household or by the children themselves. We were also not able to assess how the short ACE measure relates to important mediating or moderating maternal mechanisms beyond depressive symptoms, such as anxiety, substance use, parenting behaviors, or resilience factors.⁴²  These may be important potential intervention targets.

A short screener may support parent ACE screening in pediatric settings, facilitating care that is family-centered, trauma-informed, and strengths-based. However, the decision whether and how to screen for parent ACEs is context-specific and evidence on its clinical utility is still emerging. Decisions must incorporate consideration of the specific purpose of conducting the screening, ability to follow up with appropriate services or referrals, provider compensation and workflow challenges, and the potential for subjecting parents to retraumatization or blame.^34,37 –39 

We thank Kathryn Duncan and Saishi Cui for their valuable contributions to the development of this manuscript. A list of Children’s HealthWatch funders can be found here: https://childrenshealthwatch.org/giving/supporters/.

ACE

adverse childhood experience

BRFSS

Behavioral Risk Factor Surveillance System