Despite custodial grandchildren’s (CG) traumatic histories and risk for psychological difficulties, knowledge is scant regarding the frequencies, types, and consequences of adverse childhood experiences (ACEs) they have encountered. We examined self-reported ACEs via online surveys with 342 CG (ages 12 to 18) who were recruited to participate in an RCT of a social intelligence training program. ACEs were assessed by 14 widely used items, and risk for internalizing (ID) and externalizing (ED) difficulties were measured using 80th percentile cut-offs on the Strengths and Difficulties Questionnaire. Classification and regression tree analyses included all 14 ACEs (along with CG gender and age) as predictors of ID and ED risk separately. Given possible comorbidity, analyses were run with and without the other risk type as a predictor. Less than 9% of CG self-reported no ACEs, 48.6% reported two to five ACEs, and 30.5% reported ≥6. Irrespective of ED risk, bullying from peers strongly predicted ID risk. ED risk was peak among CG who also had risk for ID. Without ID risk as a predictor, ED risk was highest among CG who were emotionally abused, not lived with a substance abuser, and encountered neighborhood violence. The frequency and types of ACEs observed were alarmingly higher than those among the general population, suggesting that many CG have histories of trauma and household dysfunction. That a small number of ACEs among the 14 studied here were significant predictors of ID and ED risk challenges the widespread belief of a cumulative dose ACE effect.
Opmerkingen
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Children raised by custodial grandparents without the involvement of their birth parents face greater mental and physical difficulties than children in other family structures (Ge & Adesman, 2017; Smith & Palmieri, 2007; Smith et al., 2019; Xu et al., 2022). Yet, reasons for these disparities remain unclear. Given that associations between adverse childhood experiences (ACEs) and children’s physical, mental, and developmental problems have been consistently reported (Bright et al., 2016; Bright & Thompson 2018; Flaherty et al, 2009; Kan et al, 2020; Turney, 2020), ACEs experienced by custodial grandchildren (CG) may help explain their heightened risk. Indeed, the circumstances necessitating grandparental care (e.g., parental emotional abuse/neglect, substance abuse, incarceration) are themselves ACEs (American Academy of Pediatrics, 2014). Since ACEs may increase vulnerability to negative outcomes even into late adulthood, efforts to mitigate the impact of ACEs through early detection and intervention are imperative (Assmusen et al, 2020; Bellis et al. 2019; Crouch et al., 2019; Hu 2021).
Although custodial grandmothers of the CG in the present study were separately found to self-report their own ACEs at high levels (Smith et al. 2023a), empirical research on ACEs among CG has been limited to two published studies. Although Rapoport et al. (2020) found, in a study of data provided by adult informants, that children from grandparent-headed households more frequently experienced seven individual ACEs than age peers raised by parents, potential linkages between ACEs and CG’s well-being were not examined. Song et al. (2021) similarly found that, compared to peers in other family structures, children from grandparent-headed households had elevated ACEs. They also found that CG’s ACE exposure was linked to their cardiometabolic health risk. While these studies provide important information about ACE exposure among CG, they did not examine the relationship between ACEs and CG mental health outcomes. Additionally, both studies used secondary data from the National Survey of Children’s Health (NSCH) – as such, the extent to which “grandparent-headed” households represent custodial grandfamilies, as opposed to families where grandparents and parents jointly contribute to childcare, is unclear. This lack of precision about the family context was a major shortcoming of these studies in view of evidence that indices of children’s physical, psychological and social well-being are generally poorest within custodial grandfamilies (Hayslip et al. 2017). These two studies were further limited by using only adult reports of children’s ACEs (vs. self-reported ACEs), exploring only a limited number of ACEs, overlooking linkages between ACEs and children’s mental health, and by not focusing on a specific child age group.
This study augments these prior studies by examining ACEs as self-reported by adolescent CG across a broad list of 14 ACEs. Further, the study includes only CG who were clearly living in custodial grandfamilies and uses classification and regression tree analysis (CART; Breiman et al., 1984) to examine how clusters of ACEs, CG age, and CG gender are related to CG risk for internalizing (ID) and externalizing difficulties (ED). Importantly, using children’s self-reported ACEs provides a means for assessing their personal experiences that may not be attainable with data obtained from adult informants (Riley, 2004). Our focus on adolescence, as a developmental period, is especially important since parents in a NSCH population-based study reported that adolescents experienced more ACEs of every type than younger children (Bethell et al., 2017). Yet, despite concerns about obtaining reports of ACEs from caregivers rather than children themselves (Turner et al., 2020), the use of adult informants has predominated. Obtaining CG self-reported ACEs in the present study is critical because grandparents may be unaware of ACEs that occurred before they assumed custodial care. Thus, the present study is unique because it is the first to investigate ACEs experienced by adolescent CG and does so from their perspectives.
We also investigate a wider array of ACEs, in comparison to previously published studies on ACE exposure among children in grandparent-headed households (Rapoport et al., 2020; Song et al., 2021). That is, in addition to the 11 items from the widely used Adverse Childhood Experiences Module of the Behavioral Risk Factor Surveillance System (BRFSS; Centers for Disease Control and Prevention, 2008), we also consider CG exposure to neighborhood violence, bullying from peers, and parental death. Exposure to neighborhood violence and bullying from peers have both been found to have multiple adverse effects on children (Asmussen et al., 2020; Hong et al., 2018; Wright et al., 2017), and we include parental death given its high likelihood of occurrence among children within foster and/or kinship care (Turney & Wildeman, 2017).
Our focus on ID and ED in relation to CG’s ACE exposure is important for several reasons. These difficulties not only comprise one of the most widely agreed upon classifications of children’s behavior disorders (Achenbach et al., 2016), but they also represent major public health concerns that place youth at risk for subsequent adversities (e.g., academic underachievement, interpersonal problems, employment difficulties, reduced social networks, incarceration, long-term substance dependence, and persistent antisocial behavior in adulthood (Brumley & Jaffee, 2016; Memmott-Elison et al., 2020). There is extensive evidence that ACEs are associated with elevated ID and ED among children and adolescents in general (Scully et al, 2020), yet there has been limited research on the mechanisms through which ACEs affect ID and ED. So far, however, evidence points to a mediating effect of family functioning (e.g. poor parenting practices, negative family emotional climate; household chaos) on the relationship between ACEs and children’s mental health problems (Lipscomb et al, 2022; Scully et al, 2020; This seems to be particularly true in relationship to specific ACEs such as parental mental health difficulties (Scully, et al, 2020). Finally, although past studies have shown ID and ED to be more prominent among CG than age peers in the general population, the extent to which they are linked to CG ACE exposure has not been examined.
To advance the understanding of ACEs among CG, as well as address existing gaps in the literature, the present study encompassed two aims. Aim 1 examined the frequency of 14 ACEs, as self-reported by a sample of 342 adolescent CG. As part of this aim, we also examined if ACE exposure varied by CG gender. The examination of gender is essential given evidence that ACEs affect boys and girls differently (Bevilacqua et al., 2021). For Aim 2, we used Classification and Regression Tree (CART) analyses to examine how specific patterns of ACE exposure are related to CG’s risk for ID and ED, while simultaneously considering CG gender and age. CART is a non-parametric statistical technique that uses a decision tree methodology to explore how sets of independent variables cluster together to characterize individuals into mutually exclusive subgroups along a given outcome (Breiman et al., 1984). CART is particularly useful for screening and assessment purposes in public health because it produces visual trees that can be easily interpreted in a non-statistical way (Lemon et al., 2003; Toschke et al., 2005). A major appeal of CART is the ease with which high-risk subpopulations can be identified, which helps practitioners to design more nuanced approaches to assessment, intervention, and prevention (Brown et al., 2019).
Our use of CART is additionally desirable to the extent that such analyses address some key methodological concerns inherent in ACE research. Although it has been widely maintained that cumulative exposure to childhood adversity yields a “dose effect” whereby the odds of negative health outcomes increase with each additional adversity (Felitti et al, 1998; Bevilacqua et al 2021), this viewpoint erroneously assumes that ACEs exert equivalent effects on the outcome of interest and the specific patterning of co-occurring ACEs is ignored (Bevilaqua et al, 2021; Negriff 2020). In response to these concerns, the use of latent class analysis (LCA) has been increasingly used to identify distinct groups or classes of individuals who have similar patterns (or profiles) of ACEs (Lacey & Minns, 2020; Lanier et al, 2018). However, unlike CART, LCA cannot identify how specific ACEs might act alone or via interaction with other ACEs to affect a given outcome. Instead, LCA only allows investigators to identify how a broader profile (or latent class) of ACE exposure where each ACE in that profile is assumed to be equally weighted in their association with a given outcome. Thus, in comparison to using either cumulative ACEs (which results in an oversimplification of co-occurrence) or LCA classes (which gives equal weight to each ACE within a given profile) as predictors of ACE related outcomes, CART uniquely permits the identification of those ACEs which are the most highly predictive.
Although the sample in the present study is from a randomized clinical trial (RCT) funded by NIH, which examined the efficacy of an online social intelligence (SI) program for custodial grandmothers and their adolescent CG (Blinded for Review), the present study uniquely looks at data pertaining to ACEs that were obtained solely from adolescent CG at baseline. Other publications arising from this sample are primarily focused on RCT outcomes a associated with the SI intervention.
Method
Sample
Participants in this study were 342 CG, of ages 11–18 years, who were in the primary care of a custodial grandmother for at least six months (as verbally reported by grandmother), in the complete absence of the CG’s birth parents. Grandmothers also had to be fluent in English and without cognitive impairment. As noted above, CG and grandmothers were recruited nationally into an RCT on the efficacy of a social intelligence intervention (Smith et al., 2023b). If a grandmother was caring for multiple grandchildren between the ages of 11 and 18, then a target grandchild was identified. Although that study included 349 grandmother-CG pairings, seven CG did not report on ACEs at baseline and thus were dropped from the present sample. Recruitment encompassed e-mails to high school counselors and principals, social service/health providers, and advocacy/support groups. In addition, written announcements/brochures were sent via targeted household mailings. The RCT was advertised as “a study to learn how online programs can give custodial grandmothers and their grandchildren information and skills to improve health and well-being”.
As Table 1 reveals, CG were predominantly female (60.2%), White (63.5%), and non-Hispanic (86.5%). On average, they were 14 years old. The most common reason for CG to be in their grandmothers’ care was due to parental substance abuse (49.7%), and over half had been in custodial care for at least 9 years. Family income, as reported by grandmothers, was diverse, though 38% of CG lived in homes with annual household incomes of less than $26 K.
Table 1
Background Characteristics for Adolescent Grandchildren
Sample by Gender
Total Sample
Characteristic
Males
n = 136
Females
n = 206
n = 342
AGC Age (years, M | SD)
14.3 | 1.6
14.2 | 1.8
14.22 | 1.7
CGM Age (years, M | SD)
60.5 | 5.3
61.9 | 6.2
61.4 | 5.9
Race (%)
American Indian
0.7%
1.9%
1.5%
Black
24.3%
22.3%
23.1%
White
64.7%
62.6%
63.5%
Other
4.4%
2.9%
3.5%
Multiple
5.8%
10.2%
8.5%
Ethnicity (%)
Hispanic
12.5%
14.1%
13.5%
Non-Hispanic
87.5%
85.9%
86.5%
Family Yearly Income (%)
Less than $15,999
23.7%
14.9%
18.4%
$16,000–$25,999
16.3%
20.8%
19.9%
$26,000–$50,999
30.4%
30.2%
29.8%
$51000–$75,999
14.1%
18.3%
16.6%
$76,000 or More
15.6%
15.8%
15.7%
Length of Care to AGC (%)
Less than 1 year
3.7%
3.9%
3.8%
1–2 years
11.8%
12.7%
12.3%
3–4 years
14.0%
10.7%
12.0%
5–6 years
12.5%
11.2%
11.7%
7–8 years
8.8%
6.3%
7.3%
9 or More years
47.1%
54.6%
51.9%
Varies
2.2%
0.50%
1.2%
Reasons for Custodial Care (%) (multiple reasons could be reported)
Child Abuse
13.2%
12.1%
12.6%
Domestic Abuse
6.6%
8.7%
7.9%
Sexual Abuse
2.2%
2.0%
2.1%
Child Neglect
22.2%
22.8%
22.6%
Abandonment
8.1%
12.0%
10.5%
Parental Substance Abuse
47.9%
51.0%
49.7%
Parent Unwilling to Provide Care
19.9%
16.5%
17.8%
Parental Incarceration
19.1%
18.4%
18.7%
Parental Mental Illness
14.8%
14.6%
14.7%
Parental Divorce
2.9%
4.9%
4.1%
Parental Death
16.9%
16.0%
16.4%
Parental Teenage Pregnancy
6.6%
3.4%
4.7%
Parental Financial Instability
8.1%
4.9%
6.2%
Number of Cared for Grandchild (M | SD)
(Min: 1; Max: 7)
1.8 | 1.0
2.0| 1.1
1.9 | 1.5
Legal Custody of AGC (%)
Yes
88.9%
89.3%
88.9
No
11.1%
10.7%
10.7
Procedures and Measures
Demographic information was obtained from grandmothers by telephone prior to beginning the RCT. All other measures were self-reported by CG at RCT baseline through an online Qualtrics survey. The first author’s University granted IRB approval (#17-301). Grandmothers’ consent and CG assent were obtained before data collection.
ACEs
Fourteen ACEs (Table 2) were measured including the 11 items from the BRFSS (Centers for Disease Control and Prevention, 2008), and three items from the WHO (2020) ACE-IQ. Inclusion of two of the WHO ACE-IQ items (i.e., neighborhood violence and peer bullying) is justified by past research showing these two constructs are important predictors of trauma symptoms, which are often associated with ACE exposure (Ports et al., 2020; Turner et al., 2020). Our inclusion of the parental death item from the ACE-IQ (Bethell et al., 2017) is justified because parental death is one of the most stressful events that can occur in the life of a child, it often increases exposure to additional life stressors (Luecken & Roubinov, 2012), and it is a common reason for receiving grandparental care. Combining ACE indices, as done in the present study, is recommended as a means for providing a more comprehensive picture of childhood adversity (Ports et al., 2020). All 14 ACE items were scored dichotomously to indicate whether they had ever been experienced by the CG (0 = no; 1 = yes).
Table 2
Endorsement frequency of 14 ACEs for entire sample and by grandchild gender (n = 342)
% Reporting “yes”
Gender
Item
Sample
Males
Females
X2 df = 1
p
Did you live with anyone who was depressed, mentally ill or suicidal?
36.0
28.7
40.8
5.21
0.02
Did you live with anyone who was a problem drinker or alcoholic?
30.7
29.4
31.6
0.18
Did you live with anyone who used illegal street drugs or abused prescription medications?
39.5
39.7
39.3
0.01
Did you live with anyone who served time or was sentenced to serve time in a prison, jail, or other correctional facility?
45.3
47.1
44.2
0.28
Was a biological parent ever lost to you through separation, divorce, or abandonment?
58.5
63.2
55.3
2.10
Was a biological parent or guardian lost to you due to death?
21.9
21.3
22.3
0.05
Did your parents or adults in your home ever slap, hit, kick, punch or beat each other up?
30.1
29.4
30.6
0.05
Did a parent or adult in your home ever hit, beat, kick or physically hurt you in any way?
26.3
27.9
25.2
0.31
Did a parent or adult in your home ever swear at you, insult you or put you down?
55.3
55.1
55.3
0.00
Did anyone at least 5 years older than you, or an adult, ever touch you sexually?
10.2
7.4
12.1
2.04
Did anyone at least 5 years older than you, or an adult, try to make you touch them sexually?
6.7
2.9
9.2
5.15
0.02
Did anyone at least 5 years older than you, or an adult, ever force you to have sex?
2.6
0.0
4.4
6.10
0.01
Did other young people do bad or unpleasant things to you, tease you in an unfriendly way, or leave you out of things on purpose?
43.9
36.8
48.5
4.62
0.02
In the neighborhood you lived in, in real life, did you ever see or hear someone being beaten up, stabbed, shot, or being threatened with a weapon?
27.2
30.9
24.8
1.55
Total Number of ACEs Reported
X2 df = 4
0
8.8
9.6
8.3
1
11.1
12.5
10.2
2–5
48.6
47.1
49.5
6–9
24.8
24.3
25.2
10–14
6.7
6.6
6.8
Internalizing and Externalizing Difficulties
The internalizing and externalizing difficulties subscales from the adolescent version of the Strengths and Difficulties Questionnaire (SDQ; Goodman, 2001) were used to measure ED and ID. The score for ED was derived by summing the SDQ Hyperactivity-Inattention and Conduct Problems scales (potential range = 0 to 20; male α = 0.76; female α = 0.80), and the ID score was derived by summing the SDQ Emotional Symptoms and Peer Problems scales (potential range = 0 to 20, male α = 0.75; female α = 0.79). Each subscale contained five items that were self-rated by CGs on a 3-point scale ranging from 0 (not true) to 2 (certainly true). Higher scores indicate greater levels of ID or ED difficulties.
Following SDQ guidelines (www.sdqinfo.com), we calculated the 80th percentile (η.80) on both scales to derive cut-off scores such that CG at above η.80 were coded as at risk of ED and ID, respectively (0 = No Risk, 1 = Risk). SDQ scores at the 90th percentile are generally considered to be indicative of “high difficulties” and η.80 indicative of “medium difficulties” (Goodman, 2001). However, Goodman (1997) advised that, until studies are conducted to precisely define cut-off scores for different samples of children, the “medium difficulties” (η.80) cut off scores can be applied to high-risk samples while the “high difficulties” cut-offs can be used for low- risk samples. The rationale for this recommendation is that the potential identification of false positives is not a major concern when studying high-risk community samples, like CG (Stone et al., 2010). The percentage of children correctly identified by the SDQ as having a disorder is high, as is the percentage of children correctly identified by the SDQ as not having a disorder (see for review, Stone et al., 2010).
Data Analyses
Analyses were conducted using SPSS v 26. For Aim 1, frequencies and percentages for all 14 ACE items were calculated, as well as cross-tabulations and corresponding chi square test statistics for percent frequency of each ACE item by CG gender. For Aim 2, CART analyses using the SPSS v28 CHAID procedure were conducted to examine predictors of CG risk for ID and ED across four different models. In two of these models, all 14 ACE items, along with CG gender and age, were examined as predictors of ID or ED risk, respectively. These same two CART models were also run with the inclusion of risk at η.80 on the other domain, as per Achenbach et al.’s (2016) recommendation that ID and ED should each be controlled for in analyses of the other domain since they are neither mutually exclusive nor independent of one another.
CART, which classifies cases into groups according to a target (or outcome) variable based on values of predictor variables, begins with binary splits of the data into “parent nodes” that are then split into “child nodes” to determine which clusters of predictors are most strongly associated with the targeted outcome based on split criteria (see, Brieman et al., 1984). The results are then displayed as classification/decision trees. The first predictor variable at the top of the tree is the most important (i.e., most influential in predicting the value of the target variable).
In the present study, those predictor variables (i.e., all 14 ACEs, ID risk, ED risk) whose categories could be statistically significantly (p < 0.05; Chi-square-based and adjusted via the Bonferroni method) were bifurcated to best separate data into subgroups related to the either ID or ED risk (depending on which was being predicted) at each partition. This process was repeated until the sample was divided into homogeneous groups for either ID or ED risk (depending which was being predicted as the outcome), resulting in a classification tree (without pruning, due to this study’s exploratory nature). Given the asymmetrical distribution of ED and ID risk due to it being indexed at η.80, along with pragmatic concerns that false negatives were more troubling than false positives, misclassification costs were specified at a 3:1 ratio (Berk, 2008). In all models tested here, a 20-fold cross-validation was specified (to assess how well the tress structure generalizes to a larger population) and minimum cases were set at 50 for the parent node and 10 for child nodes (to specify growth limits) as options within SPSS (see IBM SPSS Decision Trees Version 28 for details).
Results
Aim 1
Table 2 reveals that over half of CG reported either losing a parent through separation, abandonment, or divorce (58.5%) or being verbally/emotionally abused (55.3%). Between one-third to nearly one half of CG reported being bullied by peers (43.9%), living with someone who either had been incarcerated at one point in time (45.3%), abused substances (39.5%), or had mental health problems (36.0%). The least frequently reported ACEs were being forced into sex (2.6%), forced to touch another sexually (6.7%), or being touched sexually (10.2%). Significant gender differences emerged across four ACEs. Higher percentages of female versus male CG reported living with someone having mental health problems, being forced into sex, being touched sexually, and being forced to touch someone sexually. As for the total number of ACEs experienced, fewer than 9% of CG reported no ACE exposure, 11.1% reported one ACE, 48.6% reported between 2 and 5 ACEs, 23.8% reported 6 to 9 ACEs, and 6.7% reported 10 or more ACEs. These percentages did not differ by gender (X2 (df = 4) = 0.689).
Aim 2
Internalizing Risk
Figures 1 and 2 reveal that 24.6% of the 342 CG were at or above η.80 regarding ID (node 0). Only a handful of 14 ACEs were identified as significant predictors of ID risk. Specifically, the left side of the classification tree in Fig. 1 shows that, when risk for ED was included a predictor of ID risk (along with CG gender, age, and all 14 ACEs), those CG who had been bullied by peers (node 1; 41.6%) were significantly more likely than non-bullied CG (node 2; 11.4%) to have ID risk. Among bullied CG, those with ED risk (node 4; 58.7%) were significantly more likely than those without ED risk (node 3; 34.0%) to reach ID at η.80. A further split emerged whereby those without ED risk were at greater risk for ID if they had not lived with a person who had ever been incarcerated (node 7; 46.5%) compared to those CG who had lived with such a person (node 8; 25.0%). The tree’s right side reveals that, among non-bullied CG, females (node 5; 17.5%) were at greater risk for ID than male CG (node 6; 3.5%). Among the latter females, those with ED risk (node 10; 43.8%) were more likely to have ID risk than those without ED risk (node 9; 13.2%). This model correctly classified 82.1% of CG as being at risk for ID.
Fig. 1
Predictors of Internalizing Risk with Externalizing Risk Included
Fig. 2
Predictors of Internalizing Risk without Externalizing Risk Included
×
×
Figure 2 depicts results for the same model tested above without ED risk being included as a predictor. Past bullying by peers was a major predictor of ID risk, with bullied CG (node 1; 41.6%) being significantly more likely than non-bullied CG (node 2; 11.4%) to have ID risk. Among bullied CG, those who had never lived with a person who had been incarcerated at one point in time (node 3; 53.7%) were more likely to have ID risk than those who had lived with such a person (node 4; 31.7%). Among the former, female CG (node 7; 65.2%) were at greater risk for ID than males (node 8; 28.6%). The tree’s right side reveals that among non-bullied CG, risk for ID was higher if CG were female (node 5; 17.8%) and lived in a home where they were emotionally abused (node 9; 31.0%). This model correctly classified 89.3% (i.e.,75 out of 84) of CG as being at risk for ID.
Externalizing Risk
Figures 3 and 4 show that 23.7% % of the 342 CG were at ED risk and that only a few of the 14 ACEs were identified as significant predictors. The right side of Fig. 3 shows that ED risk was at its overall highest among those CG with ID risk (node 1; 42.9%). The left side shows that among those without ID risk (node 2; 17.4%), risk for ID was significantly greater among CG who had been emotionally abused (node 3; 23.4%) versus those who had not (node 4; 11.5%). Among the former, ED risk was greater among CG who had never lived with a substance abuser (node 5; 31.7%) than among those who had (node 6; 16.2%). Among those CG who had not been bullied by peers, and not emotionally abused, those who were aged 13 years or older (node 7; 22%) were more likely to have ED risk than younger peers (node 8; 2.8%). This model correctly predicted 67.9% of CG as being at risk for ED.
Fig. 3
Predictors of Externalizing Risk with Internalizing Risk Included
Fig. 4
Predictors of Externalizing Risk Without Internalizing Risk Included
×
×
Figure 4 shows results for the same model as above without ID risk as a predictor. The left side of the classification tree shows that ED risk was significantly greater among CG who had been verbally abused (node 1; 30%) versus those who had not (node 2; 15.6%). Among the former, ED risk was greater among those who not lived with a substance abuse (node 3; 37.5%) than among those who had (node 4; 24.0%). Among those who had lived with a substance abuser, ED risk was greater among CG who had witnessed neighborhood violence (node 8; 38.6%) than those who had not (node 7; 12.5%). The right side of the tree shows that, among CG who had not been emotionally abused, those who had been bullied by peers (node 5; 30.0%) were significantly more likely to have ED risk than non-bullied peers (node 6; 10.5%). This model correctly classified 76.5% of CG as being at risk for ED.
Discussion
Descriptive findings from Aim 1 suggest that service professionals (pediatricians, counselors, child welfare case workers, teachers, etc.) are likely to encounter CG who have experienced particular ACEs with a considerably greater frequency than other ACEs, along with minimal gender differences in ACE exposure. More specifically, the ACEs most frequently reported by CGs were losing a parent through separation, divorce or abandonment (58.5%), being verbally/emotionally abused (55.3%), living with someone who had once been incarcerated (45.3%), being bullied by peers (43.9%), living with a substance abuser (39.5%), and living with a mentally ill person (36.0%). That over two-thirds of CG reported the latter two ACEs is noteworthy given the stance that caregiver substance abuse and mental illness may contribute to an environmental context (i.e., one that is devoid of cognitive inputs and sensory, motor, linguistic, and social experiences) that puts children at risk for the occurrence of ACEs rather than merely being ACEs in and of themselves (McLaughlin, 2016).
Collectively, the level and types of ACE exposure self-reported in the present study suggests that a substantial number of CG have lived within the presence of considerable household dysfunction. The frequencies of those ACEs indicative of household dysfunction, as self-reported here by CG, are considerably higher than those reported by adults within the original Kaiser ACE study (Felitti et al, 1998). For example, only 3.4% of adults in the Kaiser study reported having lived with a once incarcerated person compared to 45.3% of CG in the present study. That the grandmothers of the CG in the present study self-reported their own ACEs at considerably higher rates than the general adult population (Smith et al., 2023a), further suggests that multiple members of custodial grandfamilies are likely to have experienced high numbers of cumulative ACEs—and emphasizes the precedence of intergenerational challenges experienced by this population. When working with family members of a CG, it should be recognized that ACEs are often transmitted across generations (Madigan et al., 2019).
The frequency of ACEs reported by the CG in our sample were also much higher across seven similarly worded ACEs, as were reported by adults for children within grandparent-headed households in the study by Rapoport et al., (2020), with the exception of losing a parent through separation, divorce, or abandonment (58.5% in the present study vs. 56.4% in Rapoport et al. 2020). The greater frequency of parallel ACEs in the present study is likely due to our sole focus on custodial grandfamilies compared to Rappaport et al.’s (2020) broader focus on grandparent-headed families, which encompasses diverse types of grandparental care arrangements. In fact, the ACE frequencies reported by CG in the present study are like those previously reported by caregivers of children between the ages 0 and 17 living in foster care (Turney & Wideman, 2017). Our findings therefore reinforce the need for screening and investigating ACE exposure among CG (O’Connor et al., 2020; Turney & Wildeman, 2017).
Our findings further suggest an alarmingly high level of cumulative (or total) ACEs self-reported by CGs relative to the US adolescent general population. For example, among a national sample of 6483 adolescents (ages 13–17), overall exposure to at least one of 12 ACEs was reported by 58.3%, among whom 59.7% reported multiple ACEs (McClaughlin et al 2012). In contrast, 91.2% of CG in the present study reported exposure to at least one of 14 ACEs, among whom 87.8% reported multiple ACEs. In a more recent study involving data from the 2016–17 National Survey of Children’s Health, 54.4% of 29,617 adolescents between the ages of 12–17 were reported by parents or guardians to have experienced at least one ACE, with 14.8% experiencing 3 or more ACEs (Bomysoad & Francis 2020). That 67.8% of CG in the present study self-reported three or more ACEs is disturbing because of national evidence suggesting that experiencing three or more ACEs places children in a category of especially high risk for future negative outcomes (Sacks & Murphey, 2018).
Risk for Internalizing and Externalizing Difficulties by Patterns of ACE Exposure
Our findings regarding Aim 2 address the ongoing concern within the broader ACE literature regarding how to best investigate linkages between ACE exposure and indicators of well-being, given that many previously used approaches have corresponding limitations (Smith & Pollak, 2021). For instance, examining linkages across individual ACEs overlooks the likelihood of ACEs co-occurring. Similarly, relying on total (or cumulative) ACE scores erroneously assumes that each ACE has identical impact on a given outcome (McClaughlin & Sheridan, 2016). Additionally, the identification of clusters of ACEs with analytic procedures like factor or latent class analysis yields groupings of ACEs based solely on their co-occurrence, without simultaneously considering if and how they might be related to each other and to risk for adverse outcomes (Brown et al., 2019). In this study, these limitations were averted by using CART, which allowed us to identify the extent to which 14 common ACEs acted independently or in combination to predict CG risk for ID and ED in a way that is readily interpretable, thus aiding clinical utility (Brown et al., 2019; Lemon et al., 2003; Toschke et al., 2005).
Risk for Internalizing Difficulties
Irrespective of whether ED risk was included in CART analyses, having been bullied by peers highly predicted CG’s risk for ID. When ED was included as a predictor, overall risk for ID was highest among CG who had been bullied by peers and were also at risk for ED. ID risk was second highest among CG who had been bullied by peers, without ED risk, and had never lived with someone who was or had been incarcerated. The third highest risk for ID was among CG who had not been bullied by peers, but were female and at risk for ED. The overall lowest ID risk was among male CG who had never been bullied.
With ED risk absent as a predictor, the utmost risk for ID was among female CG who never lived with in a household where someone had been incarcerated and had been bullied by peers. Irrespective of gender, ID risk was also high if a CG had been bullied by person but never lived in a household where a family member had been incarcerated. Even among those CG who were not bullied by peers, risk for ID was elevated among females who had been verbally abused. ID risk was lowest for male CG who were never bullied.
Being bullied by peers as a major predictor of ID risk among CG is critical in light view of longitudinal evidence that these two variables can influence each other reciprocally (Liao et al., 2022). Children and adolescents with internalizing symptoms often exhibit behaviors (e.g., fearfulness, loneliness, low-self-esteem) which signal vulnerability to bullies looking for easy targets, and frequent victimization can produce negative self-evaluations that precede ID (Christina et al., 2021). It is especially noteworthy that both bullies and victims have very limited contact with mental health services, even though these children could benefit from targeted support (Guzman-Hoist & Bowes, 2021). The consequences of bully victimization can be grave and include self-harm and suicidal ideation, with negative effects persisting across the lifespan and into old age (Hu, 2021). The possibility of bully victimization is particularly problematic among CG is suggested by a national prevalence rate of 22.7% in comparison to 43.9% observed in our sample (Lebrun-Harris et al., 2019). Awareness of bully victimization among practitioners is critical because social forces outside of the grandfamily have the potential to be as harmful as abuse and because experiencing bullying has been found to represent a 50% increase in risk for later life mental health difficulties (Asmussen et al., 2020).
Our finding that having lived in a household where someone had once been incarcerated was a significant predictor of ID risk, in addition to being bullied by peers, is noteworthy because risk was not only high when incarceration was reported (31.7%) but even higher without it (53.7%). This paradox may be explained by recent findings that children who have experienced incarceration of a family member or other meaningful person fall into distinct subgroups based upon their developmental trajectories of internalizing problems as was as in their levels of adjustment and competence (Johnson et al., 2018; Kjellstrand et al., 2020). Most likely, such heterogeneity in outcomes or trajectories is influenced by contextual factors such as parenting quality, parental mental health, and stressful life events (Kjellstrand et al., 2020). The only other ACE that emerged as a predictor of ID risk was being verbally abused. However, this was true only for female CG and if risk for ED was not included in broader the CART analysis.
Risk for Externalizing Difficulties
Contrary to what might be expected, risk for ED was peak among those CG who were also at risk for ID, suggesting that a substantial amount of comorbidity exists within this population. This finding also reinforces methodological recommendation to control for associations between ID and ED (Achenbach et al., 2016). In the current study, when CG ID risk was included as a predictor, ED risk was highest when CG had ID risk, had experienced verbal abuse, and had never lived with a substance abuser. Risk for ED was lowest among CG without ID risk, who had not been verbally abused, and were older than age 13. Less risk among older CG likely reflects that externalizing symptoms tend to decrease throughout development (Bongers et al., 2003). Without ID risk as a predictor, ED risk was highest among CG who had been verbally abused, had not lived with a substance abuser, and had encountered neighborhood violence. ED risk was almost as high, however, among CG who had been verbally abused and had never lived with a substance abuser. Even when verbal abuse was not experienced, ED risk was high if CG had been bullied by peers. The lowest risk for ED was when both verbal abuse and bullying were not experienced.
Our finding that verbal or emotional abuse was a significant predictor of ED risk, regardless of whether or not ID risk was included in the analysis, is consistent with the prior observation that emotional abuse was the most pertinent among nine ACEs, in terms of predicting ED within a sample of 30,909 justice-involved youth (Muniz et al., 2019). Fortunately, emotional abuse is a modifiable risk factor and negative effects can be adverted if it is identified early and intervened with in a timely manner.
At first glance, the fact that the impact of verbal or emotional abuse was greater among CG who had not lived with a substance abuser seems to defy common sense. However, risk might be lower in this situation because children exposed to substance abuse might be more quickly identified and targeted for services (e.g., Plans of Safe Care supported by the Child Abuse and Prevention Treatment ACT; Child Welfare Information Gateway, 2020) than children who have experienced other ACEs (e.g., bullying by peers, emotional abuse). That exposure to bullying by peers and neighborhood violence emerged as significant predictors of ED risk underscores recent recommendations to include these items on ACE screening tools (SmithBattle et al., 2022). Neighborhood violence not only poses a direct threat but may also disrupt CG’s ability to regard their surroundings as safe and may lead them to perpetuate violence themselves (Asmussen et al., 2020).
Study Limitations
There were several measurement limitations within this study that unfortunately is characteristic of ACE research in general. These include self-reported measures dependent upon retrospective recall, a non-exhaustive list of potential childhood traumas, dichotomous ACE items that fail to capture age of exposure, type of perpetrators, chronicity or severity, and insufficient attention to current and/or subsequent adversities that may be related to initial ACE exposure (Ports et al., 2020). On the other hand, the total number of ACEs examined here was greater than in most prior studies and our use of CG self-reports provides information from their unique perspective.
Measurement of our outcome variables was also limited by the absence of such objective sources as a clinical diagnosis made by multiple informants. Also, due to data limitations, we were unable to include other important outcome domains affected by ACEs such as interpersonal relationships, physical health, self-regulation, and service utilization.
Several shortcomings were also present regarding our sample which was restricted to adolescent children, only involved participants who originally volunteered for an RCT, and was limited in racial/ethnic diversity. Because age is a proxy for years of exposure, it is logical that adolescent CG have had greater opportunity to have experienced ACES than younger CG (Crouch et al., 2019). Although there is also evidence that children of different races and ethnicities do not experience ACEs to the same degree (Sacks & Murphy, 2018) we were unable to examine these issues in the present study given the limited diversity within our sample. Data from the 2016 National Survey of Children’s Health show, for example, that 51 percent of Hispanic children have experience at least one ACE, compared with 40 percent of white non-Hispanic children and only 23% of Asian non-Hispanic children (Sacks & Murphy, 2018).
Conclusions and Practical Implications
The present findings extend the limited existing understanding of ACEs among children in grandparent-headed homes (Rapoport et al. 2020; Song et al., 2021) and have important implications for practice and policy. Yet, as recently noted by Portwood, Lawler, & Roberts (2023), arriving at such translations “is necessarily complicated by the fact that efforts to assess, to mitigate, to treat, and to prevent ACEs are occurring across disciplines and settings, each of which has its own unique characteristics, perspectives, and, often, language “ (p. 7). Nevertheless, the high frequency of ACEs self-reported here by CG clearly suggest that this group is a high-risk population, which indicates their need for additional care and resources compared to children in general. There may also be subgroups of CG ACE exposure found here that do not similarly exist within the general population.
The fact that only a smaller number of ACEs among the 14 under examination here arose as significant predictors of ID and ED risk challenges the widespread belief of an ACE cumulative dose effect regardless of outcome type (Dube et al., 2003). Instead, our findings suggest that a handful of ACEs may act in unique combinations to increase the risk of any particular outcome. Practitioners across diverse settings (e.g., schools, pediatric primary care, community mental health centers, juvenile justice system) should be aware of these combinations when undertaking screening for ACEs, and they must be sure to include ACEs such as neighborhood violence and bullying from peers which are not included on some commonly use ACE measures (Asmussen et al., 2020), Moreover, our finding that different combinations of ACEs are associated with ID verses ED risk shows a need to identify through future research which unique combinations of ACEs may best predict other key psychological, physical, educational, and social outcomes for CG.
Our findings that ACEs also interacted with CG age, gender, and comorbidity to predict outcomes illustrates the need for both practitioners s and researchers to holistically consider the context under which ACE exposure leads to either the greatest or least adversity. Contextual factors for future consideration include race/ethnicity, family SES, child disabilities, neighborhood disadvantage, and key community-level factors (Wolf et al., 2018). That custodial grandmothers are also likely to have experienced their own ACEs (Smith et al., 2023a) reinforces the emerging finding that ACEs are frequently transmitted across generations (Narayan, 2023). Although not examined here, the role of protective factors in preventing or reducing ACE-related adversity among CG is also crucial to consider in future research from an intergenerational perspective (Hayslip & Smith, 2013; Narayan, 2023). Such prevention efforts might include changes to social and structural determinants of health such as providing financial security, adequate housing, childcare and early childhood education, adequately funded community programs, as well as improved juvenile justice policies and better access to social services (Portwood et al, 2023). At the family level, resilience may be fostered by helping members of custodial grandfamilies to identify positive and health promoting childhood experiences that may have encountered in addition to, or interaction with, their experiences of childhood adversity (Narayan, 2023). In turn, this is a strengths-based strategy intended for reducing risk and offering hope in the face of intergenerational ACEs.
In closing, this study underscores the high potential for adverse behavioral and emotional outcomes among CG and points to the need for effective assessment, intervention and prevention in order to attenuate the negative impact of ACEs on CG’s future development.
Funding
The research leading to these results received funding from the National Institute on Aging under Grant Agreement No R01AG054571
Compliance with Ethical Standards
Conflict of Interest
The authors declare no competing interests.
Consent to Participate
Informed consent was obtained from all individual participants included in the study.
Ethics Approval
The questionnaire and methodology for this study was approved by the Human Research Ethics committee of Kent State University of D (Ethics approval number: #17-301.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Met BSL Psychologie Totaal blijf je als professional steeds op de hoogte van de nieuwste ontwikkelingen binnen jouw vak. Met het online abonnement heb je toegang tot een groot aantal boeken, protocollen, vaktijdschriften en e-learnings op het gebied van psychologie en psychiatrie. Zo kun je op je gemak en wanneer het jou het beste uitkomt verdiepen in jouw vakgebied.
