Risk Factors and Services to Reduce Child Sexual Abuse Recurrence

Abstract

Child sexual abuse (CSA) is associated with a range of negative consequences for victims that are compounded when it recurs. We used the National Child Abuse and Neglect Data System to study a cohort of 42,036 children in 45 U.S. states with sexual abuse reports first confirmed by child protective services (CPS) during 2010 in order to identify children with increased risk for recurrence. A small proportion (3.6%) had a second confirmed sexual abuse report through 2015. In multivariate models, female gender, family hearing and vision problems, other child maltreatment, and other family violence were associated with increased risk of recurrence, while younger children, Hispanic families, and those with substance abuse tended to have less risk. One fourth of recurrence involved the same offender, usually a parent or caretaker. One fourth of cases were referred for any CPS services, which were more likely to be provided for families with poverty, drug or alcohol problems, or other violence. Only substance abuse services significantly reduced recurrence in multivariable models. Those trying to reduce CSA recurrence should recognize that certain case characteristics are associated with greater recurrence, and most CPS services do not significantly reduce CSA recurrence.

Keywords

child sexual abuse offenders recurrence recidivism services

Children may suffer a range of negative consequences after child sexual abuse (CSA) (Pérez-Fuentes et al., 2013), and these consequences are often worse in children exposed to multiple CSA events (Hovens, Giltay, Spinhoven, van Hemert, & Penninx, 2015; Matta Oshima, Jonson-Reid, & Seay, 2014; Papalia et al., 2017). Recurrence rates for abuse or neglect during childhood have been reported to range from 9% to 67%, depending on the age of the child, length of follow-up, type of initial or subsequent abuse, services provided, and whether the study looked at reports, hospitalizations, or actual maltreatment (Fluke & Hollinshead, 2003; Levy, Markovic, Chaudhry, Ahart, & Torres, 1995; U.S. DHHS, 2016). Risk factors for CSA include female gender (Mraovich & Wilson, 1999), concurrent physical abuse (Fleming, Mullen, & Bammer, 1997), younger child age and disability (Murray, Nguyen, & Cohen, 2014), parental substance abuse (Fleming et al., 1997; Hornor & Fischer, 2016; Murray et al., 2014), financial problems and receiving family support services (Sinanan, 2011). While one of the ways to prevent further CSA involves separation from the offender, CSA recurrence can occur with or without the same offender and may involve just CSA or CSA coupled with other forms of child maltreatment (CM) such as physical abuse, psychological maltreatment, or neglect.

The extent of CSA recurrence during childhood is not well understood. One study noted rates as high as 35.4% in an National Child Abuse and Neglect Data System (NCANDS) cohort of 3,835 children with first confirmed CSA in 2002 (Sinanan, 2011), and another found that 35% of emergency department visits for CSA involved recurrence (Hu et al., 2018). This study also noted that recurrent CSA was associated with poor school performance and more psychiatric problems, attempted suicide, treatment with antipsychotic medication, and other comorbidities (Hu et al., 2018). When considering adolescent and adult revictimization, there are even higher rates of recurrence with reports ranging from 30% to 70% (Barnes, Noll, Putnam, & Trickett, 2009; Brenner & Ben-Amitay, 2015; Classen, Palesh, & Aggarwal, 2005).

It is important to identify potential risk factors for CSA recurrence, which can be effectively addressed by child protective services (CPS). For recurrent CM of all types, young, White children and those with disabilities and parental alcohol abuse were more likely to be rereported to CPS and found to be victims again (Connell, Bergeron, Katz, Saunders, & Tebes, 2007; Fluke, Shusterman, Hollinshead, & Yuan, 2008; Fuller & Wells, 2003). For initial abuse including multiple types, child disability, poverty, emotional problems with parents or child, and violence between caretakers were associated with increased risk of any recurrence (Connell et al., 2007; Dakil, Sakai, Lin, & Flores, 2011; Fluke et al., 2008; Jonson-Reid, Emery, Drake, & Stahlschmidt, 2010; Palusci, Smith, & Paneth, 2005; White, Hindley, & Jones, 2015). Regarding CSA specifically, Hispanic ethnicity or African American race, having a disability, having the first report made by medical and law enforcement personnel, and financial problems have been associated with a statistically decreased likelihood for second reports (Fluke et al., 2008), and younger children, girls, and children who were prior victims of CM had increased risk of CSA recurrence (Sinanan, 2011). Hornor and Fischer (2016) noted that revictimized children were younger and more likely to possess a developmental delay and mental health diagnosis when compared with children with only one episode of CSA. Another study noted that older age at abuse, a criminal history, and mental health problems were uniquely predictive of more problematic and persistent revictimization trajectories (Papalia et al., 2017).

Little has been reported about whether there are certain characteristics of the offender in the first report of the child’s sexual abuse that are predictive of recurrence. One small study noted that children with recurrent CSA were more likely to have a caregiver as the offender and to have this occur in a familiar place such as at home, school, or the babysitter’s home (Hu et al., 2018). If CSA is more likely to recur with the same offender identified in a first confirmed report, several steps could be taken within the child welfare and criminal justice systems to prevent recurrence. We have not identified extensive published research examining the extent to which CPS services after CSA confirmation, if provided, are associated with decreases in CSA recurrence. Studies evaluating the effects of services on CM recurrence in general have noted that service referral is often associated with no significant change or with increased rather than decreased recurrence (Barth, Gibbons, & Guo, 2006; Fluke et al., 2008; Fluke, Yuan, & Edwards, 1999; Palusci et al., 2005). Using data from 10 states in 1994–1995, others found that neglect was most likely to recur and recurrence was found to be more likely among young children; services were also associated with increased recurrence (Fluke et al., 1999). One study (Sinanan, 2011) noted that having received family preservation services was found to be associated with a statistically decreased likelihood for second CSA reports, but this study only looked at a small number of U.S. states over 3 years.

Some services may be associated with increased CPS rereporting because of surveillance bias during service provision or because services are offered preferentially to families perceived to have greater risk of recurrence (Chaffin & Bard, 2006). Prior analyses of services may also be biased because not all services are applied to all cases, and children with long histories of maltreatment have been found to be more likely to be referred for some services, especially out-of-home care (Kinard, 2002). Certain services seem more likely to be provided after specific types of confirmed maltreatment (Garland, Landsverk, Hough, & Ellis-MacLeod, 1996) such as physical abuse with the maltreating caretaker no longer in the home (Swenson, Brown, & Sheidow, 2003) and for unsubstantiated cases when the report comes from a mandated reporter (Drake, Jonson-Reid, & Sapokaite, 2006).

We hypothesized that CSA differs from other forms of maltreatment in its recurrence and that certain predictive child, family, and offender characteristics can be identified. We were unsure as to the potential effects of CPS service referrals but hypothesized that there may be commonly used services that can reduce the risk of CSA recurrence. In previous studies, we have used NCANDS data to study reporting and recurrence of other forms of CM and believe it could also be helpful to better understand CSA (Palusci, 2011; Palusci et al., 2005; Palusci & Ondersma, 2012; Palusci & Vandervort, 2014; Palusci, Vandervort, & Lewis, 2016). The objectives of the present study are to (1) determine the recurrence rate 5 years after confirmed CSA reports to CPS; (2) identify child, family, offender, and report characteristics and their association as risk factors with the recurrence rates; and (3) assess the correlation of key CPS services with CSA recurrence rates.

Method

Data Set Preparation

NCANDS collects data from U.S. states and territories since 1990 and offers large annual samples of CM reports. State CPS agencies voluntarily submit expanded case-level information about child, family, and service characteristics for what is now called the Child File (National Data Archive on Child Abuse and Neglect [NDACAN], 2017). Although precise definitions vary from state to state, CM-type categorizations in NCANDS are based on federal guidelines for evidence of one or more instances of physical abuse, sexual abuse, psychological maltreatment, neglect, or medical neglect (U.S. DHHS, 2016). When state agencies find credible evidence that abuse or neglect has occurred, the report is labeled “substantiated” or “indicated” based on state law and the child is considered a “victim” in a CPS-confirmed report. Recent years in the data set also contain ‘alternative response’ victims which are also considered confirmed reports, although the investigation process is different (NDACAN, 2017). Information is provided regarding the type of postinvestigation services provided by CPS which adds to the ability to link the use of such services with CM recurrence.

NCANDS public use data sets were obtained for this study for 2010–2015 from the National Data Archive on Child Abuse and Neglect at Cornell University in February 2017 (NDACAN, 2017). The SAS statistical software package, Version 9.1 (SAS Institute Inc., Cary, NC), was used for data preparation and analysis. Data from multiple years were merged by state and year after confirming that child identifiers were consistent across years for that state. Victim reports could have up to four confirmed CM types (sexual abuse, physical abuse, neglect, medical neglect, and psychological maltreatment). Duplicative reports occurring on the same day were deleted. Records were sorted by state, child identification number (ID), and report date in the data set and were compared with published information to assess overall data set integrity.

Several steps have been taken in the data set preparation, distribution, and use to protect the privacy of children and families. Names and other identifying information had been removed and replaced with unique child and report identifiers prior to distribution (NDACAN, 2017). State identifiers for fatalities were masked as were county identifiers for counties with fewer than 1,000 annual reports. Files were transmitted using secure servers and were stored on secure computers. Because of these protections, the New York University human subjects committee deemed this research to be exempt from review.

Study Sample

The study sample used for this study was derived from the NCANDS Child File data sets from 2010 to 2015. A cohort was constructed from all records of children in 2010 with confirmed CSA reports that were labeled as having no prior confirmed reports. Contributions from U.S. states and territories were assessed, and those with data missing for one or more years during the study period were excluded. Only children with first CSA reports in 2010 were kept in the study cohort, but records could have one or more CM types included in addition to CSA. The following records were excluded: (1) children in states with noncontinuous data submissions, (2) children with any confirmed abuse or neglect prior to 2010, (3) unconfirmed reports, and (4) first confirmed reports with no CSA. The data set was sorted by state, child ID, and report date, and second records were matched to cohort records by state and child ID. A counter variable was created to identify the first and second confirmed reports for each child ID in the study cohort. Time to second report was calculated. Recurrence of CSA was defined as a second confirmed report of CSA reported for the same child ID in the same state during the 2010–2015 period.

Study Variables

NCANDS Child File data include fields with information concerning child age in years; gender; race (mutually exclusive categories of Native American, Asian, Black, Pacific Islander, or White); Hispanic ethnicity; CPS report date, report source and disposition; services provided; and one to four types of CM. Children and offenders are labeled with individual masked identifiers in the data. NCANDS also includes response items regarding whether the child does or does not have special medical needs, developmental delays, drug or alcohol exposure, impairments, disabilities, behavior problems, and chronic medical problems. We defined a collapsed variable for disability that included mental retardation, visual or hearing impairment, learning disability, and other special medical needs. Caretaker characteristics include the presence or absence of identified drug or alcohol problems, emotional or physical problems, mental retardation, disability, medical problems, violence between caretakers, receipt of financial assistance, or problems with money or housing. Offender characteristics are also recorded in NCANDS for each of the possible four CM types listed in a report in which the child has been found to be victimized. These child, family, and case factors are determined by the CPS worker during investigation and do not necessarily reflect the independent determinations of clinicians, courts, or other child welfare professionals involved in a case (NDACAN, 2017).

Information is provided in NCANDS regarding referrals for broad categories of services. These include counseling, family support services, family preservation services, health-related and home health services, housing services, mental health treatment, substance abuse treatment, foster care, adoption, and juvenile court petition. These postinvestigation services are entered into the child’s NCANDS record if they were provided or arranged by the CPS agency, social services agency, or the child welfare agency for the child or family within the first 90 days after the disposition of the report.

Study Design

Variables were placed in an ecological model to assess CSA recurrence. Included in our study design were child characteristics such as age and gender and family characteristics such as alcohol/drug abuse, disability, behavior problems, financial problems, and living arrangement. Report factors were added to include offender data such as relationship to victim, co-occurrence of other type of CM, and service referrals.

Data Analysis

Data analysis took place in three major steps. In the first step, we sought to calculate recurrence rates overall and stratified them by child, family, offender, and report factors. The frequencies and odds ratios (ORs) by 5-year age groups for child, family, report, and services in confirmed reports were calculated using standard methods. Given that the peak incidence for CSA reporting occurs during ages 5–9 years, statistical comparisons of the proportions of these factors among children aged 5–9 years with those aged 0–4 and 10–14 years were calculated. For child age groups, survival curves were constructed using standard methods (Marshall & English, 1999). χ² tests were used to assess the statistical significance of the association for categorical variables, and Student’s t tests were used for continuous variables. Statistical power was calculated to assess the adequacy of the sample size of this cohort for individual factors. In the second major step, we identified associations of a number of child, family, offender, and report characteristics with CSA recurrence in bivariate and multivariate models, including logistic regression and Cox proportional hazard models. In the third major step, we calculated associations between service referrals (mental health; counseling; substance abuse treatment; foster care; adoption; health-based, family preservation, and family support services) and child, family, offender, and report characteristics and then entered these factors into multivariate models of recurrence with a particular service referral. This identified a set of key variables associated with each service, allowing us to later control for whether a service was used at all in later models. To account for the 90-day time needed for service provision, recurrence for our services analysis was determined to have occurred if a second confirmed CSA report was made 90 days or more after the first report.

Logistic regression models were used to assess the contribution of multiple factors predicting CSA in second confirmed reports compared to children with no second confirmed CSA reports. To be inclusive, variables with significance .10 or less in bivariate analyses were entered into the multivariable model, and variables with the highest variance were removed one at a time and the model was recalculated until all remaining variables had p ≤ .05. ORs with 95% confidence intervals (CI) were calculated for the variables remaining in adjusted models using standard methods. To assess the effect of time on revictimization, time elapsed in each record from first confirmed report (rounded to 2 weeks due to data set blinding) to second confirmed CSA report was calculated using the report dates in the data set. Key variables associated with any service referral were first entered as covariates into multivariate, right-censored Cox regression models of time to CSA recurrence, and then the dichotomous service referral variable was entered.

We excluded records from analysis when one or more of the covariates or outcomes in a model were missing. Missing data were assessed to see if there were any significant associations in those records with child and family factors or service referrals. We did not use data imputation because, even with these missing data, the number of complete cases in the cohort retained had >.80 statistical power to detect a 10% change in recurrence while controlling for covariates. Furthermore, there does not appear to be bias in the missing data in that absent data came primarily from seven states that did not submit certain service or child factor information representing less than 10% of the cohort. Others have used a similar approach to working with NCANDS data (Fluke et al., 2008).

Results

Recurrence Rates After Confirmed CSA Reports

The merged data set for years 2010–2015 contained 22,940,468 child report records. Records in which the child was not labeled a “victim” were removed, leaving 4,285,686. After duplicate reports on the same day were removed, there remained 4,232,423 victim records from 45 U.S. states with records in all 6 years. Of these, there were 42,036 confirmed cases of CSA in 2010 with 1,496 (3.6%) matched second records from 2010 to 2015 containing confirmed sexual abuse (Figure 1). California, Georgia, North Dakota, New Hampshire, Oregon, and Puerto Rico did not have records in all years during 2010–2015 which prevented complete linkage, and records from these states were not included in our analyses. Second confirmed reports occurred on average 639 days after the first confirmed report for sexual abuse (standard deviation [SD] = 537; range = 13–2,342 days). Most (36,227/42,036; 86%) first confirmed reports contained only sexual abuse.

Figure 1.

Study cohort reports with confirmed child sexual abuse, NCANDS Child Files (2010–2015). CM = child maltreatment; CSA = child sexual abuse; NCANDS = National Child Abuse and Neglect Data System.

It is important to note that there are differences by age in the amount of time to CSA recurrence. There was a significant difference between age of first abuse and recurrence rate, and 10- to 14-year-olds had the highest recurrence rates (718/15,000; 4.54%). This group also had faster recurrence during the study period (Figure 2). Older teens (15–19 years) also recurred quickly, but they could not be followed for the full study period as they were no longer considered children in the data set when they “aged out” at 18 years.

Figure 2.

Survival analysis comparing age groups to time to recurrence in days.

Child, Family, Offender, and Report Characteristics

Child factors that were associated with higher rates of CSA recurrence were being female, having Pacific Islander race, emotional problems, learning problems, medical problems, physical disability, or any disability (mental retardation, visual impairment, learning disability, physical disability, or other special medical needs; Table 1). Mean child age at first confirmed sexual abuse was 10.1 years (SD = 4.6; 95% CI = [9.9, 10.3]). Family characteristics that were significantly associated with an increased likelihood of recurrence were family hearing and vision problems, family medical problems, and any disability (mental retardation, visual impairment, learning disability, physical disability, or other special medical needs; Table 2). Asian race, Hispanic ethnicity, family drug problems, and inadequate housing were all associated with decreased risk. For report characteristics, the majority of recurrence was confirmed as sexual abuse only (1,219), but 19% included other maltreatment (277 cases; Table 3). While the number of cases was small, there was a significantly increased risk for CSA recurrence after a first confirmed report of CSA, physical abuse, neglect, and medical neglect. Of the offender characteristics, only caretaker relationship was significantly associated with recurrence in bivariate analysis.

Table 1.

Bivariate Analyses: Child Factors and CSA Recurrence.

Factor	Factor Present, # (%)	CSA Recur., #	CSA Recur., %	OR [95% CI]
All CSA	42,036	1,496	3.56	NA
Child gender
Female	33,206 (79.0)	1,291	3.89	1.67 [1.44, 1.93]***
Male	8,680 (20.6)	202	2.33	0.59 [0.51, 0.68]***
Child age
0–4 years	5,233 (12.4)	173	3.31	0.72 [0.61, 0.85]***
5–9 years	11,752 (28.0)	430	3.66	0.87 [0.81, 0.94]***
10–14 years	15,806 (37.6)	718	4.54	1.00 [reference group]***
15–18 years	9,109 (21.7)	174	1.91	0.41 [0.46, 0.60]***
Child race
White	27,184 (64.7)	997	3.67	1.09 [0.97, 1.23]
Asian	338 (0.8)	4	1.18	0.32 [0.12, 0.86]*
Native American	465 (1.1)	18	3.87	1.09 [0.68, 1.75]
Black	8,348 (19.9)	298	3.57	1.00 [0.88, 1.14]
Pacific Islander	123 (0.3)	11	8.94	2.66 [1.43, 4.96]**
Child Hispanic ethnicity	7,079 (16.8)	209	2.95	0.77 [0.67, 0.90]***
Child alcohol exposure	68 (0.2)	0	0	NA
Child drug exposure	98 (0.2)	0	0	NA
Child mental retardation	0 (0.0)	0	0	NA
Child emotional problems	714 (1.7)	35	4.9	1.44 [1.02, 2.03]*
Child hearing or vision impaired	80 (0.2)	4	5	1.46 [0.53, 3.99]
Child learning problems	259 (0.6)	15	5.79	1.71 [1.01, 2.88]*
Child physical disability	106 (0.3)	10	9.43	2.90 [1.51, 5.57]***
Child behavior problems	1,016 (2.4)	43	4.23	1.21 [0.88, 1.65]
Child medical problems	419 (1.0)	29	6.92	2.08 [1.42, 3.05]***
Any child disability	690 (1.6)	51	7.39	2.26 [1.69, 3.02]***

Note. Any child disability includes any of mental retardation, visual impairment, learning disability, physical disability, medical need, and other disability as defined in NCANDS. CSA = child sexual abuse; CI = confidence interval; NA = not available due to small numbers; OR = odds ratio; Recur. = recurrence.

*p < .05. **p < .01. ***p < .001.

Table 2.

Bivariate Analyses: Family Factors and CSA Recurrence.

Factor	Factor Present, # (% of Sample)	CSA Recur., #	CSA Recur., %	OR [95% CI]
Family alcohol problems	661 (1.6)	18	2.72	0.76 [0.48, 1.22]
Family drug problem	757 (1.8)	13	1.72	0.47 [0.27, 0.82]**
Family mental retardation	71 (0.2)	4	5.63	1.49 [0.54, 4.11]
Family emotional problems	430 (1.0)	17	3.95	1.13 [0.70, 1.85]
Family hearing or vision impaired	109 (0.3)	10	9.17	2.76 [1.44, 5.30]**
Family physical problems	157 (0.4)	3	1.91	0.54 [0.17, 1.68]
Family learning problems	134 (0.3)	7	5.22	1.52 [0.71, 3.26]
Family medical problems	315 (0.7)	20	6.35	1.86 [1.18, 2.94]**
Any family disability	633 (1.5)	37	5.85	1.57 [1.12, 2.29]**
Military family	108 (0.3)	0	0	NA
Family domestic violence	2,008 (4.8)	86	4.28	1.25 [1.00, 1.56]
Family inadequate housing	4,302 (10.2)	123	2.86	0.80 [0.66, 0.96]*
Family financial problems	5,133 (12.2)	183	3.57	1.03 [0.88, 1.22]
Receiving public assistance	3,352 (8.0)	109	3.25	0.93 [0.77, 1.14]

Note. Any family disability is defined as any of family mental retardation, visual impairment, learning disability, physical disability, medical need, and other disability as defined in NCANDS. CSA = child sexual abuse; CI = confidence interval; NA = not available; OR = odds ratio; Recur. = recurrence.

*p < .05. **p < .01. ***p < .001.

Table 3.

Bivariate Analyses: Report Factors, Offender Characteristics, and CSA Recurrence.

Factor	Factor Present, # (%)	CSA Recur., #	CSA Recur., %	OR [95% CI]
First CM type
CSA alone	36,227 (86.2)	1,239	3.42	1.00 [reference group]
CSA + any other CM	5,809 (13.8)	257	4.42	1.31 [1.14, 1.15]***
CSA + physical abuse	562 (1.3)	29	5.16	1.52 [1.05, 2.21]*
CSA + neglect	4,422 (10.5)	189	4.27	1.26 [1.08, 1.47]***
CSA + medical neglect	11 (0.0)	3	27.27	10.58 [2.81, 39.96]***
CSA + psychological maltreatment	250 (0.6)	12	4.58	1.42 [0.80, 2.55]
Medical/police report source	3,987 (9.5)	145	3.64	1.03 [0.86, 1.22]
Offender
Male	35,407 (84.2)	1,275	3.60	1.06 [0.86, 1.31]
Female	2,815 (6.7)	96	3.41	0.95 [0.77, 1.16]
Caretaker	19,845 (47.2)	608	3.06	0.70 [0.63, 0.78]***
Parent	10,937 (26.0)	404	3.69	1.05 [0.93, 1.19]

Note. Offender types are not exclusive or exhaustive. CSA = child sexual abuse; CM = child maltreatment; CI = confidence interval; OR = odds ratio; Recur. = recurrence.

*p < .05. **p < .01. ***p < .001.

In multivariate models using logistic regression of the child, family, and report factors, variables significantly associated with increased risk for recurrence were family hearing and vision problems (OR = 3.187; 95% CI [1.584, 6.411]) and other family violence (OR = 1.487; 95% CI [1.134, 1.950]). Variables associated with decreased risk for recurrence were younger child age (OR = 0.968/year; 95% CI [0.952, 0.983]), child male gender (OR = 0.519; 95% CI [0.421, 0.638]), Hispanic ethnicity (OR = 0.784; 95% CI [0.641, 0.959]), family substance abuse (OR = 0.290; 95% CI [0.167, 0.686]), and having only sexual abuse in the initial report (OR = 0.801; 95% CI [0.648, 0.990]).

Overall, 1.07% of children with first confirmed CSA in the study cohort had a second confirmed report of sexual abuse listing the same person as the offender. However, one fourth (25%) of the offenders were the same person in both the first and second confirmed sexual abuse reports based on having the same matching ID in NCANDS. Parents comprised the highest proportion (61%) of offenders in the second confirmed CSA report. Mean offender age when CSA recurred was 37.8 years (95% CI = [35.8, 40.0]) compared to 37.9 years (95% CI = [35.9, 39.9]) with no recurrence (not statistically different; Table 3). In bivariate analysis, a child with an offender who was a "caretaker" was significantly less likely to have CSA recurrence (OR = 0.7015), but the association was not significant when the offender was labeled as a “parent.”

Key Service Referrals and CSA Recurrence Rates

Fewer than one fourth (24.7%) of children or families were referred for any services after first confirmed CSA. In bivariate analysis of referral for any service, family support, family preservation, foster care, court-appointed representatives, information and referral, and mental health services were all statistically associated with increased CSA recurrence (Table 4). However, these relationships were affected by biases in who was being referred for these services. We can speculate that cases with more perceived risk were more likely to be referred for a particular service to address that risk. Our analysis found that type of service referral varied based on type of CM, child age, race, ethnicity, emotional or physical problems, family alcohol or substance abuse, other family violence, and offender factors. Families with housing or financial problems were also differentially referred for services. In models for recurrence, service referral was added to these factors significant for service referral, and the odds for recurrence were calculated. In models controlling for the factors affecting referrals, none significantly decreased CSA recurrence, but there was a trend toward risk reduction (p = .0551) after substance abuse services referral (Table 5). CPS referrals for mental health treatment, counseling, foster care, health, family preservation, and family support were not significant (p > .05) in their effect when accounting for the variables that predicted their use.

Table 4.

Bivariate Analyses: CPS Referrals and CSA Recurrence After 90 Days.

Service	Service Present, # (% of Sample)	CSA Recur., #	CSA Recur., %	OR [95% CI]
Any services offered	10,420 (24.8)	393	3.77	1.39 [1.23, 1.57]***
Family support services	2,167 (5.2)	111	5.12	1.82 [1.49, 2.22]***
Family preservation services	2,554 (6.1)	101	3.95	1.35 [1.10, 1.67]**
Foster care services	2,697 (6.4)	126	4.67	1.65 [1.37, 1.99]***
Juvenile court petition	2,308 (5.5)	74	3.21	1.51 [0.83, 1.34]
Court-appointed representative	701 (1.7)	44	6.28	2.10 [1.54, 2.86]***
Adoption services	1,419 (3.4)	33	2.33	0.76 [0.53, 1.08]
Case management services	5,128 (12.2)	211	4.11	1.44 [1.24, 1.67]***
Counseling services	2,263 (5.4)	81	3.58	1.21 [0.96, 1.52]
Day care services	264 (0.6)	9	3.41	1.14 [0.58, 2.22]
Education and training	426 (1.0)	13	3.05	1.07 [0.61, 1.87]
Employment services	58 (0.1)	1	1.72	0.56 [0.08, 4.07]
Family planning services	66 (0.2)	0	0	NA
Health/home health services	839 (2.0)	24	2.86	0.96 [0.63, 1.43]
Home-based services	530 (1.3)	14	2.64	0.87 [0.51, 1.49]
Housing services	221 (0.5)	6	2.71	0.90 [0.40, 2.02]
Independent living services	50 (0.1)	2	4.00	1.33 [0.32, 5.46]
Information and referral	1,727 (4.1)	67	3.88	1.31 [1.02, 1.68]*
Legal services	193 (0.5)	5	2.59	0.84 [0.35, 2.05]
Mental health services	1,237 (2.9)	51	4.12	1.40 [1.05, 1.86]*
Pregnancy and parenting services	228 (0.5)	5	2.19	0.72 [0.30, 1.75]
Respite services	130 (0.3)	2	3.05	0.50 [0.12, 2.01]
Special disability services	63 (0.1)	4	6.35	2.17 [0.79, 5.98]
Special services juvenile delinquency	44 (0.1)	2	4.55	1.53 [0.37, 6.34]
Substance abuse services	374 (0.9)	9	2.41	0.79 [0.41, 1.53]
Transportation services	306 (0.7)	12	3.92	1.30 [0.73, 2.33]

Note. CPS = child protective services; CSA = child sexual abuse; CI = confidence interval; NA = not available; OR = odds ratio; Recur. = recurrence.

*p < .05. **p < .01. ***p < .001.

Table 5.

Multivariate Analyses: CPS Referrals and CSA Recurrence After 90 Days.

Service	Factors Associated With Service Referral	Service Effect on Recurrence, OR [95% CI]
Any service	Child age, White, Hispanic, emotional, physical problems, behavior; family drug, hearing and vision, learning, housing, money problems, public assistance; physical abuse, neglect, medical neglect, psychological maltreatment; offender: caretaker	1.02 [0.840, 1.23]
Mental health	Child Asian, Hispanic, emotional, behavior problems; family alcohol, emotional, violence; multiple CM, offender: caretaker, age	0.87 [0.49, 1.53]
Counseling	Child Hispanic, behavior problems; family emotional, learning, violence, housing, public assistance; multiple CM, offender: gender, relationship, age	1.09 [0.80, 1.48]
Substance abuse	Child Hispanic, family alcohol, drugs, violence; CSA only, physical abuse	0.38 [0.14, 1.04], (p = .0551)
Foster care	Child sex, American Indian, Black, White, Hispanic, alcohol, behavior; family drugs, housing, public assistance; report source, CSA only, physical abuse, multiple CM, offender: gender, relationship, age	1.29 [0.97, 1.72]
Adoption	Child White race; neglect	2.19 [1.52, 3.15], (p < .001)
Health	Child American Indian, White, Hispanic, emotional, disabled; family emotional, public assistance; physical abuse, neglect, offender: caretaker, relationship	0.58 [0.24, 1.44]
Family preservation	Child Asian race, Hispanic, emotional; family learning, violence, housing; neglect, offender: caretaker, relationship	0.83 [0.65, 1.07]
Family support	Child age, Asian race, Hispanic, behavior; family drugs, emotional, violence, disability, housing, money, public assistance; report source, multiple CM, offender: caretaker, relationship	1.28 [0.95, 1.72]

Note. OR = odds ratio; CI = confidence interval; CM = child maltreatment; CSA = child sexual abuse.

p > .05 unless shown.

Discussion

CSA Recurrence Rate After Confirmed CSA Reports

This study confirms that a small but important number of children with a confirmed CSA report have CPS-confirmed CSA recurrence. Our 5-year recurrence rate of 3.56% is higher than that of physical abuse (1.9%) but is lower than rates reported in NCANDS for psychological maltreatment (9.2%) (Palusci et al., 2005; Palusci and Ondersma, 2012). It appears that CSA is fundamentally different from other CM and is often confirmed in first and second reports as the only form of CM present. Unlike psychological maltreatment, which is more likely to recur when it is the only form of CM in the initial report, CSA is more likely to recur when first associated with other forms of CM. CPS workers should be aware that CSA co-occurring with other forms of CM is concerning and that reabuse may be more likely to occur in these settings.

Child, Family, and Offender Characteristics and CSA Recurrence

We found that certain child factors such as female gender, age, Pacific Islander race, emotional problems, disability, and medical problems were associated with recurrence risk in bivariate analysis. Family characteristics that we found associated with an increased likelihood of recurrence included hearing and vision problems, medical problems, and any disability (mental retardation, visual impairment, learning disability, physical disability, or other special medical needs). These are comparable to what others have found to be risk factors for CM overall (Connell et al., 2007; Dakil et al., 2011; Fluke et al., 2008; Fuller & Wells, 2003; Jonson-Reid et al., 2010; Palusci et al., 2005; White et al., 2015).

Some of the results of our study agree with previous findings. A study by Fluke, Shusterman, Hollinshead, and Yuan (2008) using logistic regression models found child female gender, older age, and family hearing and vision problems to be associated with increased risk of recurrence. Using a multivariate model, Hornor and Fischer (2016) were in agreement with our results that child disability was associated with increased recurrence. Another study also found older age to be associated with increased risk of recurrence (Papalia et al., 2017). Sinanan (2011) found a higher recurrence rate (35%) in a smaller hospital population but also noted that child female gender was associated with increased risk of recurrence. This much higher rate is likely related to the sample studied. CPS workers should be aware that female gender is found in most studies to be associated with increased risk of CSA recurrence.

Some studies differed from some of our findings. Fluke et al. (2008) found child disability to be associated with decreased risk of recurrence and they also identified other factors associated with decreased risk of recurrence, including child African American race, first report by a medical or law professional, and family financial problems, all of which were equivocal in our study. It seems to be less clear how to address older child age and disability given the inconsistency in the literature for these results, but as Papalia et al. (2017, p. 112) noted, “individuals exposed to CSA during adolescence may be particularly vulnerable to poorer re-victimization trajectories, characterized by multiple risk indices, and thus may warrant increased service provision.” CPS should know that these victims warrant more attention due to their increased risk of reabuse.

We were surprised to find that a significant proportion (one fourth) of children with recurrent CSA have been victimized by the same offender. In general, studies have noted that offender recidivism rates are 10–15% within 5 years, but these rates are thought to underestimate actual recidivism and do not apply specifically to CSA recurrence (Hanson, Morton, & Harris, 2003). Child molesters tend to have higher recidivism rates (>50%) over their lifetime, and while we noted recidivism with specific children, these rates do not reflect the total recidivism since offenders may have offended others. Reoffenders in our data were usually parents, and others have noted that repeat sexual offenders are more likely to be in poor and minority groups but with lower recidivism against their own biological children (<5%) and less likely to recidivate than nonsexual offenders (Greenberg, Bradford, Firestone, & Curry, 2000; Hanson, Scott, & Steffy, 1995; Patrick & Marsh, 2009). While we cannot address offender age or gender with our results, it is important to note that particular services may have differing effects on prevention, and it is important for CPS to address whether a parent was the offender and whether that parent will have continuing opportunity to victimize the child again.

Effects of CPS Referrals

Perhaps the most notable finding from this study is that, controlling for bias in referral to service, most services—including the referral for any service—were associated with no significant measurable change in recurrence. Only substance abuse services had even a trend toward positive impact, which may be related to a variety of factors in the services, how they are provided, or the substance-abusing parent themselves. Adoption services were actually found in our multivariate models to be associated with increased recurrence, but it is unclear why. Perhaps it is because children referred for adoption are in some way worse off to begin with and remain at high risk even after removal. This deserves further study. It is also unclear why service referrals were not provided for most confirmed victims in our data. When used, referrals for counseling and foster care services have been made more for children with other CM, particularly for older children and the poor. While one could postulate that young children are more likely to be referred for services because they are perceived by CPS to have greatest risk, fewer infants and young children received any services at all. High proportions of Native American, Pacific Island, and military families received services, but these numbers are difficult to interpret given the small number of these children and the number and type of services available to these populations, which are affected by geography or additional federal policy.

Our study is in agreement with most studies in the literature which have shown that referrals to CPS are not associated with a decrease in CM recurrence (Russell, Kerwin, & Halverson, 2018). It may be the case that current services are succeeding in their goal of reducing recurrence, but several factors could confound this finding in our study. One reason for the lack of impact of service referral may be a surveillance effect in which families offered services are being seen more by more professionals and any recurrence is more likely to be reported. Those with more CPS reports participating in postinvestigation services are often more disadvantaged, have other risks, and are known by the child welfare system for a longer time, although the degree to which surveillance bias affects recurrence is debated (Chaffin & Bard, 2006; Drake et al., 2006; Fuller & Nieto, 2014; Hélie & Bouchard, 2010). It is also possible that, given risk assessment procedures used in CPS agencies, CPS workers preferentially provide service referrals to families at greater actuarial risk of recurrence. While we did not use formal propensity score methods to control potential biases in service selection, it is thought that a point estimate of a treatment effect from our analysis of covariance using multivariate adjustment for treatment allocation is equal to the estimate obtained from the two-step adjustment used in propensity scores when the same sample covariance matrix is used (D’Agostino, 1998).

There are several other potential reasons why our study was unable to demonstrate that services provided to families with CPS-confirmed CSA resulted in lower overall rates of recurrence. One strong possibility is that services offered through CPS do not address CSA-specific caretaker or child needs and therefore are not effective in preventing future CSA (Chaffin, Bonner, & Hill, 2001; Kinard, 2002). The evidence base for a variety of services such as parenting classes, parent education, pediatric anticipatory guidance, and family wellness/preservation programs still remains weak (Palusci & Haney, 2010; Reynolds, Mathieson, & Topitzes, 2009). CPS practices regarding assessment, confirmation, and recording variables have not been standardized and there is variation in state CSA definitions. This is further complicated by the fact that CPS acceptance and investigation of reports varies across states and has varied over time with increased use of alternative response reports, leading to several potential differential and non-differential biases in confirmation and service referral. While reported cases have been shown to have risk profiles similar to those with confirmed victims (Hussey et al., 2005), we could not include unconfirmed reports in our analyses because NCANDS contains only limited information about unsubstantiated and unfounded reports.

Limitations

This study has several important limitations. Since it is up to CPS workers to record and enter data for each case, there are missing and potentially inaccurate data recorded in NCANDS, and it is apparent from the tables that some variables may not have been reported in an important number of records. Also, we can of course only study CSA that gets reported and we know that’s not all of it. We grouped all the states' data together, but reporting and confirmation threshold varies from state to state, so this could confound our results. CPS data in NCANDS are for referral to services so it is unknown if the families actually received the services after the first confirmed report. In addition, perhaps it is the engagement in services that is more important than the type or number of services (Hindley, Ramchandani, & Jones, 2006), and specific service programs, attendance, and other factors associated with less recurrence are not available in NCANDS (DePanfilis & Zuravin, 1998). This comes from the fact that despite being large, NCANDS is not nationally representative (U.S. DHHS, 2016) and contains information which was not collected specifically for our analyses. While this has not precluded multistate analyses (Fluke et al., 2008), several states submitted incomplete data in one or more years limiting its usefulness. Some of these deficiencies could be addressed by a prospective study design in which we could confirm that all data were recorded, variables such as disability and substance abuse were confirmed clinically, and that CPS services were actually received. It may also be that there are different interventions that could be helpful (such as directed education around sex abuse) that could be evaluated prospectively (Pulido et al., 2015), and future studies are needed to strengthen our inference of causal effects.

Conclusions

It has been noted that “re-entry into CPS is a complex interaction of risk to children and system factors tied to the intervention they receive” (Fluke et al., 2008, p. 76). An important number of children were sexually abused a second time within 5 years, and one fourth of these were sexually reabused by the same person, usually a parent. While we did find several factors associated with CSA recurrence in bivariate analyses, only older child age, female gender, family hearing and vision problems, and other violence and CM were significantly associated with increased risk of recurrence in multivariate models. Most families were not referred for services at initial identification, and for most who did receive services, they appear to have been ineffective in preventing CSA recurrence. This suggests a need for more standard risk assessment tools that not only reliably capture risk and case prioritization but also identify dynamic risks and interactions that can better guide service recommendations and referrals. We should temper any optimism for preventing CSA since there remains a great deal of work to be done when examining the key features of efficacious postinvestigation services for this important subgroup of maltreated children. The big questions of how best to prevent sexual abuse, how to reduce rates over time, and, eventually, eliminate it remain unanswered.

Footnotes

Authors’ Note

The authors are solely responsible for the content and analyses; neither the participating state agencies, WRMA, Inc., the Children’s Bureau, the National Data Archive, and Cornell University, nor their agents or employees bear any responsibility for the analyses, opinions, or interpretations presented here. The data utilized in this publication were made available by the National Data Archive on Child Abuse and Neglect, Cornell University, Ithaca, NY, and have been used with permission. Data from the National Child Abuse and Neglect Data System were supplied by state child protective services agencies to the Children’s Bureau, the Administration on Children, Youth and Families, U.S. Department of Health and Human Services. Funding for NCANDS was provided by the Children’s Bureau, U.S. Department of Health and Human Services. Technical support on NCANDS is provided under contract to the Children’s Bureau by WRMA, Inc. (formerly Walter R. MacDonald & Associates, Inc.).

Acknowledgments

The authors wish to thank Elliott Smith, Michael Dineen, and Andrés Arroyo for their assistance in obtaining and using the NCANDS data sets.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Vincent J. Palusci

References

Barnes

J. E.

Noll

J. G.

Putnam

F. W.

Trickett

P. K.

(2009). Sexual and physical revictimization among victims of severe childhood sexual abuse. Child Abuse & Neglect, 33, 412–420.

Barth

R. P.

Gibbons

Guo

(2006). Substance abuse treatment and recurrence of maltreatment among caregivers with children living at home: A propensity score analysis. Journal of Substance Abuse Treatment, 30, 93–104.

Brenner

Ben-Amitay

(2015). Sexual revictimization: The impact of attachment anxiety, accumulated trauma, and response to childhood sexual abuse disclosure. Violence and Victimization, 30, 49–65.

Chaffin

Bard

(2006). Impact of intervention surveillance bias on analyses of child welfare report outcomes. Child Maltreatment, 11, 301–312.

Chaffin

Bonner

B. L.

Hill

R. F.

(2001). Family preservation and family support programs: Child maltreatment across client risk levels and program types. Child Abuse & Neglect, 25, 1269–1289.

Classen

C. C.

Palesh

O. G.

Aggarwal

(2005). Sexual revictimization: A review of the empirical literature. Trauma Violence and Abuse, 6, 103–129.

Connell

C. M.

Bergeron

Katz

K. H.

Saunders

Tebes

J. K.

(2007). Re-referral to child protective services: The influence of child, family, and case characteristics on risk status. Child Abuse & Neglect, 31, 573–588.

D’Agostino

R. B.

(1998). Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Statistics in Medicine, 17, 2265–2281.

Dakil

S. R.

Sakai

Lin

Flores

(2011). Recidivism in the child protection system: Identifying children at greatest risk of reabuse among those remaining in the home. Archives of Pediatric and Adolescent Medicine, 165, 1006–1012.

10.

DePanfilis

Zuravin

S. J.

(1998). Rates, patterns and frequency of child maltreatment recurrences among families known to CPS. Child Maltreatment, 3, 27–42.

11.

Drake

Jonson-Reid

Sapokaite

(2006). Rereporting of child maltreatment: Does participation in other public sector services moderate the likelihood of a second maltreatment report? Child Abuse & Neglect, 30, 1201–1226.

12.

Fleming

Mullen

Bammer

. (1997). A study of potential risk factors for sexual abuse in childhood. Child Abuse and Neglect, 21, 49–58.

13.

Fluke

J. D.

Hollinshead

D. M.

(2003). Child maltreatment recurrence. Duluth, GA: National Resource Center on Child Maltreatment.

14.

Fluke

J. D.

Shusterman

G. R.

Hollinshead

D. M.

Yuan

Y. Y.

(2008). Longitudinal analysis of repeated child abuse reporting and victimization: Multistate analysis of associated factors. Child Maltreatment, 13, 76–88.

15.

Fluke

J. D.

Yuan

Y. Y. T.

Edwards

(1999). Recurrence of maltreatment: An application of the national child abuse and neglect data system (NCANDS). Child Abuse & Neglect, 23, 633–650.

16.

Fuller

T. L.

Nieto

(2014). Child welfare services and risk of child maltreatment rereports: Do services ameliorate initial risk? Children and Youth Services Review, 47, 46–54.

17.

Fuller

T. L.

Wells

S. J.

(2003). Predicting maltreatment recurrence among CPS cases with alcohol and other drug involvement. Children and Youth Services Review, 25, 553–569.

18.

Garland

A. F.

Landsverk

J. L.

Hough

R. L.

Ellis-MacLeod

(1996). Type of maltreatment as a predictor of mental health service use for children in foster care. Child Abuse & Neglect, 20, 675–688.

19.

Greenberg

Bradford

Firestone

Curry

(2000). Recidivism of child molesters: A study of victim relationship with the perpetrator. Child Abuse & Neglect, 24, 1485–1494.

20.

Hanson

R. K.

Morton

K. E.

Harris

A. J. R.

(2003). Sexual offender recidivism risk: What we know and what we need to know. Annals of the New York Academy of Science, 989, 154–166.

21.

Hanson

R. K.

Scott

Steffy

R. A.

(1995). A comparison of child molesters and nonsexual criminals: Risk predictors and long-term recidivism. Journal of Research in Crime and Delinquency, 32, 325–337.

22.

Hélie

Bouchard

(2010). Recurrent reporting of child maltreatment: State of knowledge and avenues for research. Children and Youth Services Review, 32, 416–422.

23.

Hindley

Ramchandani

P. G.

Jones

D. P. H.

(2006). Risk factors for recurrence of maltreatment: A systematic review. Archives of Disease in Childhood, 91, 744–752.

24.

Hornor

Fischer

B. A.

(2016). Child sexual abuse revictimization: Child demographics, familial psychosocial factors, and sexual abuse case characteristics. Journal of Forensic Nursing, 12, 151–159.

25.

Hovens

J. G.

Giltay

E. J.

Spinhoven

van Hemert

A. M.

Penninx

B. W.

(2015). Impact of childhood life events and childhood trauma on the onset and recurrence of depressive and anxiety disorders. Journal of Clinical Psychiatry, 76, 931–938.

26.

M-H.

Huang

G-S.

Huang

J-L.

C-T.

Chao

A-S.

F-S.

H-P.

(2018). Clinical characteristics and risk factors of recurrent sexual abuse and delayed reported sexual abuse in childhood. Medicine, 97, 1–5.

27.

Hussey

J. M.

Marshall

J. M.

English

D. J.

Knight

E. D.

Lau

A. S.

Dubowitz

Kotch

J. B.

(2005). Defining maltreatment according to substantiation: Distinction without a difference? Child Abuse & Neglect, 29, 479–492.

28.

Jonson-Reid

Emery

C. R.

Drake

Stahlschmidt

M. J.

(2010). Understanding chronically reported families. Child Maltreatment, 15, 271–281.

29.

Kinard

E. M.

(2002). Services for maltreated children: Variations by maltreatment characteristics. Child Welfare, 81, 617–645.

30.

Levy

H. B.

Markovic

Chaudhry

Ahart

Torres

(1995). Reabuse rates in a sample of children for five years after discharge from a child abuse inpatient assessment program. Child Abuse & Neglect, 19, 1363–1377.

31.

Marshall

D. B.

English

D. J.

(1999). Survival analysis of risk factors for recidivism in child abuse and neglect. Child Maltreatment, 4, 287–296.

32.

Matta Oshima

K. M.

Jonson-Reid

Seay

K. D.

(2014). The influence of childhood sexual abuse on adolescent outcomes: The roles of gender, poverty, and revictimization. Journal of Child Sex Abuse, 23, 367–386.

33.

Mraovich

L. R.

Wilson

J. F.

(1999). Patterns of child abuse and neglect associated with chronological age of children living in a midwestern county. Child Abuse and Neglect, 23, 899–903.

34.

Murray

L. K.

Nguyen

Cohen

J. A.

(2014). Child sexual abuse. Child and Adolescent Psychiatric Clinics of North America, 23, 321–337.

35.

National Data Archive on Child Abuse and Neglect (NDACAN). (2017). National Child Abuse and Neglect Data System (NCANDS) Child Files, 2010-2015 [Datasets]. Retrieved February 2017 from http://www.ndacan.cornell.edu

36.

Palusci

V. J.

(2011). Risk factors and services for child maltreatment among infants and young children. Children and Youth Services Review, 33, 1374–1382.

37.

Palusci

V. J.

Haney

M. L.

(2010). Strategies to prevent child maltreatment and integration into practice. APSAC Advisor, 22, 8–17.

38.

Palusci

V. J.

Ondersma

S. J.

(2012). Services and recurrence after psychological maltreatment confirmed by child protective services. Child Maltreatment, 17, 153–163.

39.

Palusci

V. J.

Smith

E. G.

Paneth

(2005). Predicting and responding to physical abuse in young children using NCANDS. Children and Youth Services Review, 27, 667–682.

40.

Palusci

V. J.

Vandervort

F. E.

(2014). Universal reporting laws and child maltreatment report rates. Children and Youth Services Review, 38, 20–28.

41.

Palusci

V. J.

Vandervort

F. E.

Lewis

J. M.

(2016). Does changing mandated reporting laws improve child maltreatment reporting in large U.S. counties? Children and Youth Services Review, 55, 170–179.

42.

Papalia

N. L.

Luebbers

Ogloff

J. R. P.

Cutajar

Mullen

P. E.

Mann

(2017). Further victimization of child sexual abuse victims: A latent class typology of re-victimization trajectories. Child Abuse and Neglect, 66, 112–129.

43.

Patrick

Marsh

(2009). Recidivism among child sexual abusers: Initial results of a 13-year longitudinal random sample. Journal of Child Sexual Abuse, 18, 123–126.

44.

Pérez-Fuentes

Olfson

Villegas

Morcillo

Wang

Blanco

(2013). Prevalence and correlates of child sexual abuse: A national study. Comprehensive Psychiatry, 54, 16–27.

45.

Pulido

M. L.

Dauber

Tully

B. A.

Hamilton

Smith

M. J.

Freeman

(2015). Knowledge gains following a child sexual abuse prevention program among urban students: A cluster-randomized evaluation. American Journal of Public Health, 105, 1344–1350.

46.

Reynolds

A. J.

Mathieson

L. C.

Topitzes

J. W.

(2009). Do early childhood interventions prevent child maltreatment? A review of research. Child Maltreatment, 14, 182–206.

47.

Russell

J. R.

Kerwin

Halverson

J. L.

(2018). Is child protective services effective? Children and Youth Services Review, 84, 185–192.

48.

Sinanan

A. N.

(2011). The impact of child, family, and child protective services factors on reports of child sexual abuse recurrence. Journal of Child Sexual Abuse, 20, 657–676.

49.

Swenson

C. C.

Brown

E. J.

Sheidow

A. J.

(2003). Medical, legal and mental health service utilization by physically abused children and their caregivers. Child Maltreatment, 8, 138–144.

50.

U.S. Department of Health and Human Services (U.S. DHHS), Administration for Children, Youth and Families. (2016). Child maltreatment 2015. Washington, DC: U.S. Government Printing Office.

51.

White

O. G.

Hindley

Jones

D. P.

(2015). Risk factors for child maltreatment recurrence: An updated systematic review. Medicine, Science and the Law, 55, 259–277.