Child disability status has been robustly associated with increased rates of maltreatment within the United States and internationally (e.g., Maclean et al., 2017; Spencer et al., 2005; Sullivan & Knutson, 2000). While ID has consistently been associated with heightened maltreatment risk (e.g., Dion, Paquette, Tremblay, Collin-Vezina, & Chabot, 2018; Horner-Johnson & Drum, 2006), evidence is equivocal regarding ASD (see Hoover & Kaufman, 2018, for review). Samples of children with ASD receiving mental health treatment have documented elevated rates of maltreatment (Brenner, Pan, Mazefsky, Smith, & Gabriels, 2017; Mandell et al., 2005). State wide data has shown that children identified with ASD via school records are more likely to be involved in child protective services (Hall-Lande et al., 2015). Similarly, research linking child services administrative data with Medicaid claims have also found that children with ASD are at elevated risk for foster care involvement (Cidav, Xie, & Mandell, 2018). However, some population-level studies have failed to find that children with ASD are at elevated risk for maltreatment compared to population controls (Maclean et al., 2017; Spencer et al., 2005; Sullivan & Knutson, 2000).

Methodological differences may have contributed to these disparate findings (Urbano, Epstein, Cull, Vehorn, & Warren, 2017). Many of the studies documenting elevated rates of maltreatment among children with ASD have relied on parental report for children receiving mental health treatment, which may not fully represent the wider population of children with ASD (Mandell et al., 2005; Brenner et al., 2017). Most studies using population level data have relied on administrative labels to identify ASD, which may under-identify children with ASD (Sullivan & Knutson, 2000; Spencer et al., 2005). Therefore, examining the prevalence and characteristics of objectively identified maltreatment among a large, population-level sample of children identified with ASD and ID would significantly advance research in this area.

To explicate the unique risk associated with ASD and/or ID, it is essential to adjust for socio-demographic correlates of maltreatment. Poverty is a robust risk factor for child maltreatment (Drake & Jonson-Reid, 2014; Pelton, 2015), and socioeconomic factors underlie ethnic differences in maltreatment risk (Kim & Drake, 2018; Putnam-Hornstein, Needell, King, & Johnson-Motoyama, 2013). Further, low parental education has consistently been associated with heightened maltreatment risk (Sidebotham, Heron, et al., 2006). Given concern that females with ASD may be uniquely vulnerable to abuse and emotional difficulties (Bargiela, Steward, & Mandy, 2016), it is also important to consider child gender in relation to maltreatment risk.

To comprehensively identify appropriate supports for children with ASD and/or ID who have experienced maltreatment, it is important to understand multiple characteristics of maltreatment. Maltreatment experiences differ across several dimensions, including whether or not they are substantiated (i.e., whether sufficient evidence was found to conclude that the child was abused or neglected). Substantiated and unsubstantiated maltreatment cases do not differ in terms of existing risk factors, recidivism risk, or associated child developmental or behavioral outcomes (e.g., Kohl, Jonson-Reid, & Drake, 2009). Although substantiation is often required for service provision, allegations are the better estimate for maltreatment risk (Kohl et al., 2009). Thus, considering both reports (all investigated cases of maltreatment) and substantiations is important for understanding maltreatment risk and the likelihood of service provision.

Maltreatment experiences also vary by subtype and perpetrator relationship to the victim. Physical neglect is differentiated from abuse, which is typically classified as sexual, physical, or emotional (Barnett et al., 1993). Perpetrators may vary in closeness to the victim, including whether they are immediate or extended family versus nonfamilial individuals (Sullivan & Knutson, 2000). Research identifying whether children with ASD and/or ID are at particular risk for certain subtypes and perpetrators would enable more targeted prevention and intervention efforts, as most prior research has focused on the presence or absence of maltreatment.

Little is known regarding behavioral correlates of maltreatment exposure in ASD, which is essential for developing assessment and treatment approaches for traumatic stress in ASD (Hoover, 2015; Brenner et al., 2017; Kerns, Newschaffer, & Berkowitz, 2015). Children with ASD and abuse may show heightened behavioral difficulties, such as aggression, self-injury, tantrums, and fears, compared to children with ASD without abuse (Howlin & Clements, 1995). Given that these studies are cross-sectional, it is also possible that these behavioral difficulties increase child risk for maltreatment, rather than being the consequence of abuse or neglect.

The present study aimed to evaluate the odds of experiencing maltreatment among children with ASD and/or ID in comparison to population controls. Differences in broader maltreatment characteristics were examined among all maltreated children with and without ASD and/or ID. In addition, the association between maltreatment and behavioral difficulties was evaluated for children with ASD.

To assess lifetime maltreatment experiences within the study cohort, ADDM and birth certificate records for the 4,988 individuals were linked to South Carolina Department of Social Services (DSS) records from 2000–2016 via the SC Office of Revenue and Fiscal Affairs (RFA) Health and Demographics Division. Ethical approval for data linkage and retention was obtained via the local institutional review board and agreement to use DSS data stored by SC RFA was approved by SC DSS. Participants ranged in age from two to eight years at the beginning of the study and all participants had turned 18 by the end of the study period. Extracted variables obtained from DSS included maltreatment type, presence of a screened-in alleged report and whether the report was substantiated, and perpetrator relationship.

Maltreatment experiences were classified into established operationalized subtypes according to the Maltreatment Classification System (MCS; Barnett et al., 1993): (1) sexual abuse, (2) physical abuse, (3) emotional abuse, and (4) physical neglect. Primary analyses were conducted on whether an individual had ever experienced a report (0 = none, 1 = at least one report) or substantiation (0 = none; 1 = at least one substantiation) overall, and then by subtype.

For analyses regarding maltreatment characteristics, the number of total cases involving reports and substantiations of each maltreatment category were summed to ascertain the frequency of distinct maltreatment experiences. Counts were arrayed by case number to account for multiple reports from the same incident, such that distinct subtypes from the same case number were only counted once. For example, a score of 0 for physical abuse indicated that zero cases involved a report of physical abuse, whereas a score of 3 indicated that three separate cases involved a report of physical abuse. To assess subtype comorbidity, we evaluated the number of lifetime subtypes experienced from 1 (e.g., physical abuse only) to 4 (physical, sexual, and emotional abuse, and physical neglect).

Perpetrators were categorized based on closeness to the victim as (1) immediate family members in a parental role, (2) extended family members (e.g., aunt, cousin, grandparent), (3) extrafamilial individuals (e.g., babysitter, child care provider, public employee, residential staff), or (4) siblings/other children (e.g., biological sibling, foster child, etc.), in accordance with prior research (Barnett et al., 1993; Sullivan & Knutson, 2000). The total number of cases involving a report or substantiation by each type of perpetrator were summed, such that distinct perpetrators from the same case number were only counted once for each case. For example, a score of 0 for the extrafamilial category indicated that no cases had an extrafamilial individual named as a perpetrator, whereas a score of 3 would indicate that three separate cases had an extrafamilial individual named as a perpetrator. Lastly, the number of distinct perpetrators across all reports was summed; a score of 1 indicated one perpetrator only (e.g., all charges came from mother) whereas higher scores indicated multiple perpetrators (e.g., score of 3 might indicate that charges came from mother, father, and aunt).

Multinomial logistic regression was utilized to quantify the relationship between group status (ASD-only, ASD+ID, ID-only) and the probability of maltreatment overall and by subtype, relative to the PC group. Gender, ethnicity, maternal education, and income were entered as covariates. To evaluate whether the ASD-only, ASD+ID, and ID-only groups had different maltreatment experiences relative to other maltreated youth in the PC group, a sub-group analysis among children with at least one report or substantiation was conducted. Group differences in subtype frequency and perpetrator characteristics, while controlling for gender, ethnicity, maternal education, and income, were evaluated using ANCOVA with Bonferroni-corrected posthoc comparisons if overall F-tests were significant at the p < .001 level, given the high number of statistical tests. Posthoc comparisons were also evaluated amongst the ASD-only, ASD+ID, and ID-only groups to evaluate whether risk is greater for certain disability groups. A second sub-analysis among the ASD (ASD-only, ASD+ID) groups using multinomial logistic regressions was used to evaluate whether maltreatment related to behavioral indicators, controlling for child gender.

Information regarding missing data is provided in Table 1. There was a significant portion (10%) of missing data for maternal education, which was estimated using multiple imputation for substantive analyses. Sensitivity analyses were conducted and confirmed that the overall interpretation of findings was not affected by the inclusion or exclusion of maternal education. Thus, reported analyses are adjusted for maternal education.

The adjusted odds ratios of experiencing overall reported and substantiated maltreatment by group are presented in Table 2. The ASD-only, ASD+ID, and ID-only groups were more likely to have reported and substantiated maltreatment relative to the PC group. The adjusted odds ratios of experiencing maltreatment subtypes are presented in Table 3. For sexual abuse, the ASD+ID and ID-only groups were more likely to have reports compared to controls, whereas only the ID-only group was more likely to have substantiations. For physical abuse, all groups (ASD-only, ASD+ID, ID-only) were more likely to have reports than controls, whereas only the ID-only group was more likely to have substantiations. For emotional abuse, the ASD-only and ID-only groups were more likely to have reports and the ID-only group was more likely to have substantiations relative to controls. For physical neglect, all groups (ASD-only, ASD+ID, ID-only) were more likely to have reports and substantiations than controls.

Evaluation of group differences in maltreatment characteristics among maltreated youth is presented in Table 4. Regarding subtypes, the ASD-only (mean diff = .90, p=.002) and ID-only groups (mean diff = .46, p=.001) had significantly more cases with alleged physical abuse than controls. The ASD-only (mean diff = .92, p=.044) and ID-only (mean diff = 1.05, p=.000) groups also had significantly more cases involving alleged and substantiated physical neglect than controls (mean diff for ASD-only = 1.89, p=.002; mean diff for ID-only = 1.21, p=.000). Regarding perpetrator variables (see Table 4), the ASD+ID (mean diff = .63, p=.028) and ID-only (mean diff = .82, p=.000) had more distinct perpetrators reported across cases. The ID-only group (mean diff = .72, p=.000) had more reported cases with immediate family members named as perpetrator relative to controls. For substantiations, the ASD-only (mean diff = 1.26, p=.014) and ID-only groups (mean diff = 1.06, p=.000) experienced more substantiated perpetrators across cases than controls. Further, the ASD-only (mean diff = .98, p=.036) and ID-only groups (mean diff = .87, p=.000) had more cases with immediate family caregivers substantiated as perpetrators than controls. No pairwise comparisons between the ASD-only, ASD+ID, and ID-only groups were significant.

The associations between maltreatment report status and behavioral features among children with ASD is presented in Table 5. Adjusting for gender, maltreatment status was associated with higher likelihood of aggression, hyperactivity, and tantrums.

Even compared to other maltreated youth, children with ASD and/or ID are at risk for more frequent and complex maltreatment. Children with ASD-only and ID-only experienced more cases involving reports of physical abuse and neglect compared to maltreated children without developmental disabilities. Further, the ASD-only and ID-only groups had more cases with an immediate family caregiver named as perpetrator than maltreated children in the PC group. This is consistent with extant research documenting that the majority of maltreatment experiences are perpetrated by a caregiver (US DHHS, 2017), and highlights the power of parent-child prevention and intervention approaches for children with ASD and/or ID as well as for typically developing children. The ASD-only and ID-only groups also had a higher number of different perpetrators who were substantiated across all of their cases, suggesting that children with ASD and ID may be at risk for complicated maltreatment perpetrated by multiple individuals.

For children with ASD, maltreatment was associated with a higher likelihood of hyperactivity, aggression, and temper tantrums. This is consistent with prior research documenting that children with ASD who have experienced abuse may have heightened disruptive behaviors (Howlin & Clements, 1995) and difficulties similar to those of typically developing maltreated children (Brenner et al., 2017). However, these behavioral measures relied on descriptors within extracted records rather than validated symptom measures, and it was not possible to identify the direction of effects (e.g., whether behaviors placed children at risk or were a consequence of maltreatment). Much work remains to be done in the identification and measurement of trauma-related symptoms in ASD (Kerns et al., 2015).

Although this study showed significantly higher rates of maltreatment among children with ASD and/or ID, the study design could not identify bidirectional relationships between ASD and maltreatment. Moreover, the control group was not reevaluated for late ASD diagnosis after age 8, which may have under-identified children with subtle ASD presentations and potentially led to misclassification. Longitudinal, prospective research is critical in this area. There are several additional limitations to the current investigation. Maltreatment data from state social services only reflects cases that were formally reported, which may underestimate the true prevalence of maltreatment and non-caregiver perpetrated maltreatment. In addition, this study did not incorporate other stressful (e.g., bullying) or adverse childhood experiences (e.g., community violence), nor involvement with adult protective services. Lastly, limited information was available regarding family socio-economic status, and future research should examine how other poverty indicators relate to maltreatment for children with disabilities (e.g., Putnam-Hornstein, & Needell, 2011).

In summary, the current investigation represents the first empirical evidence from population-level data that the prevalence of maltreatment is elevated in ASD relative to the general population. Establishing that children with ASD, who are already a population vulnerable to elevated family stress and emotional and behavioral difficulties, are also at heightened risk for maltreatment emphasizes the urgent need to consider developmental disability status in child maltreatment research and refine clinical supports for those doubly vulnerable families affected by both maltreatment and ASD and/or ID (Hoover & Kaufman, 2018).