Child welfare as a gateway to domestic violence services

Children and Youth Services Review 27 (2005) 1203 – 1221 www.elsevier.com/locate/childyouth

Child welfare as a gateway to domestic violence services Patricia L. Kohla,*, Richard P. Bartha, Andrea L. Hazenb, John A. Landsverkc a

School of Social Work, University of North Carolina, Chapel Hill, NC 27599-3550 United States b Child and Adolescent Services Research Center, United States c San Diego State University, United States Available online 24 June 2005

Abstract This paper uses data from the National Survey of Child and Adolescent Well Being to examine the identification of domestic violence (DV) by child welfare workers during investigations of maltreatment and determine how this contributes to the receipt of DV services. The study focuses on female caregivers of children remaining in the home following the investigation (n = 3165). While child welfare workers indicate that active DV is present in only 12% of families investigated for maltreatment, 31% of caregivers reported DV victimization in the past year. The sensitivity of reports of DV is low between caregivers and workers, with both reporting active or recent DV in only 8% of families. Substance abuse by the primary caregiver is a strong predictor of under identification of DV by the child welfare worker (OR = 7.6). Overall, about half of the caregivers with active DV identified by the worker received DV services over the 18 months following the investigation. Logistic regression analyses examined whether receipt of child welfare services (CWS) increases the likelihood that a referral will be made to DV services and whether caregivers will then obtain these services. Both the identification of DV by the worker and having an open CWS case are significant contributors to receipt of DV services. D 2005 Elsevier Ltd. All rights reserved. Keywords: Child welfare; Domestic violence; Child maltreatment; Substance abuse

* Corresponding author. E-mail address: [email protected] (P.L. Kohl). 0190-7409/$ - see front matter D 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.childyouth.2005.04.005

1204

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1. Introduction Previous research on the co-occurrence of domestic violence (DV) and child maltreatment strongly suggests that children who have been involved with child welfare services (CWS) are often exposed to DV. Child welfare workers concluded that DV was present in 28% of the 125 caregivers indicating a current or recent relationship in a sample drawn from families in CWS in New York City (Magen, Conroy, Hess, Panciera, & Levy Simon, 2001). In another study of 74 CWS cases, workers expressed a belief that DV had happened or had a high risk of happening in 32% of the cases (Shepard & Raschick, 1999). More than half (57%) of caregivers with a history of severe DV victimization entering the child welfare system have had previous contact with CWS (Hazen, Connelly, Kelleher, Landsverk, & Barth, 2004). On the basis of the recognition of this co-occurring risk, states have begun to change their risk assessment procedures and laws to ensure preemptive attention to children in families with DV. Yet, the estimates on which new policies and practices are based are from small samples drawn from local agency case record reviews and do not capture the majority of child welfare cases. CWS procedures regarding DV are emerging as the child welfare system grapples with growing public and professional pressure to find more balanced and effective means to respond to families with co-occurring DV and child maltreatment. Some have called for greater vigilance and higher levels of intervention in DV cases. A trend toward including exposure to DV as a form of maltreatment that warrants child welfare intervention is evident (Barnett, Miller-Perrin, & Perrin, 1997; Edleson, 1999). Calls for more vigilance and CWS involvement with DV cases are countered by a growing sense of unfairness about the penalties to mothers who are victims of DV when they are charged with maltreatment or have their children placed into foster care. Some states (e.g., New York) now require that courts must consider the presence of DV in the home when determining whether the need to place a child can be eliminated by removing the abuser from the home (New York State Adoption and Safe Families Act, 1999); however, New York’s efforts to remove children and prosecute the victim of DV for child neglect were challenged in a class action lawsuit against child protective services in New York City (Allen & Bisell, 2004; Kantor & Little, 2003). In early 2002, the United States District Court ruled, in Nicholson v. Williams, that the presence of DV is bnot sufficient grounds for taking children away from their mothersQ (Kaufman, 2003). Another concern is that referrals to child welfare services for a child’s exposure to DV will inundate an already stressed system. In addition to New York, a few other states have also unsuccessfully attempted to implement legislation mandating across the board responses to all cases entering the child welfare system with co-occurring domestic violence and child maltreatment. For instance, Florida required that all cases with DV automatically be flagged as high risk, and a child protective services case be opened and filed with the court (Weithorn, 2001); however, the state was inundated with cases and had to amend the statute. Minnesota had a similar experience with a policy that included exposure to DV as a form of maltreatment and had to repeal the statute until additional funds could be found to provide an adequate CWS response (Edleson, 2004).

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1205

The National Council of Juvenile and Family Court Judges recognized the need for a system change in the response to families with both maltreatment and DV. The results of their deliberations, Effective Intervention in Domestic Violence and Child Maltreatment Cases: Guidelines for Policy and Practice, often referred to as the bGreenbookQ, outlines a guiding framework to assist communities in responding to co-existing DV and maltreatment (Schechter & Edleson, 1999). The recommendations for child welfare services include collaboration with domestic violence organizations, the development of assessment procedures, protocols, and training to aid in the identification and response to DV. 1.1. Child welfare worker identification of domestic violence in child welfare Despite the lack of consistent policies about the response of CWS to children’s exposure to DV, it is very important that child welfare workers assess for the presence of DV during investigations of child maltreatment. The identification of DV by the worker would appear to be crucial to ensuring better outcomes for children (Magen et al., 2001). Children exposed to domestic violence are at greater risk of having multiple contacts with CWS for child maltreatment. Unresolved parent problems, including DV, appear to increase the likelihood of reentry into CWS. A significant association was found between DV and re-referral to CWS in a study of 12,329 referrals to child welfare services in Washington State (English, Marshall, Brummel, & Orme, 1999). Consistent with the concept of multiple contacts with CWS, DV may also predispose families toward having children placed into foster care, making re-entry into foster care, following reunification, more common. Violence between adults was identified as a parent problem for 56% of the children who re-entered out of home care following reunification with their families (n = 62 families) in Indiana (Hess, Folaron, & Jefferson, 1992). This cyclical nature speaks to the necessity of child welfare workers identifying and addressing the DV, as well as the child maltreatment. Little is known, however, about typical national CWS practices regarding the identification of DV. Some evidence indicates that it is often not considered by the child welfare worker during an investigation of child maltreatment. In a study of 74 cases in a progressive child protection agency, workers reported inquiring directly of the caregiver for the presence of DV in only 38% of the cases (Shepard & Raschick, 1999). English and colleagues (1999) found that DV risk was not assessed in about half of the referrals to CWS included in their study. There are many plausible explanations for this lack of identification. In a qualitative study, caseworkers revealed some reluctance to ask about DV (Magen et al., 2001). Barriers identified by health services research, which may generalize to CWS, include practitioner’s fear of offending the woman, a sense of powerlessness and lack of control over the situation, and beliefs about the woman’s lack of initiative (Fogarty, Burge, & McCord, 2002). Lack of both formal education about DV and institutional support are additional barriers to social workers assessing for the presence of DV; however, agency in-service training has been shown to increase the screening for DV (Tower, 2003) and increase the likelihood that child welfare workers believe that their role should include assessing for domestic violence (Mills & Yoshihama, 2002). However, it has also been found that

1206

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

incorporating child welfare worker training on DV is a challenging undertaking (Mills et al., 2000). Another possible explanation for this under identification in cases already reported to the child protective system may be that inconsistent policies present barriers to women disclosing their abuse. Women may be hesitant to disclose due to fear of their partner, shame, or embarrassment (Fogarty et al., 2002). Assessment by a child welfare worker may exacerbate the battered woman’s fears due to concerns that her child could be removed from the home (Carter, Weithorn, & Behrman, 1999). Given the generally nonvoluntary nature of many responses to DV (Mills, 2003) and child welfare’s reputation for coercion (Hutchison, 1987), this is a reasonable expectation. Further information about the under identification of domestic violence can be ascertained through comparisons of caregivers self-report of DV victimization and the child welfare worker’s determination of DV toward the caregiver, presumably during the risk assessment. Although previous studies trying to link self-reported problems and worker responses have not identified high levels of convergence (e.g., Gibbons & Barth, submitted for publication), some convergence should exist. Given the aforementioned disincentives of disclosing domestic violence victimization, such as fear of a punitive CWS response, divergence between the caregiver and worker responses should yield an undercount (more false negatives) by the worker. Prior research offers little guidance on this prediction, although available findings suggest little correspondence between child welfare worker assessment of DV and caregiver’s self-report of DV victimization. In a Washington State study, in addition to workers identifying a history of domestic violence on the risk assessment, caregivers were asked whether they had been bhit, slapped, beaten, pushed, or sexually assaulted by a partner as an adultQ (English & Graham, 2000, p. 918). The authors report that a significant correspondence was only found for sexual assault by a partner as an adult. 1.2. Child welfare and domestic violence services Although the goal of risk assessment is to identify the risk of future harm, an additional aim is to determine service needs (Cash & Berry, 2002). The identification of DV during child maltreatment investigations must be followed by the provision of appropriate domestic violence services. In a study of 115 families, Cash and Berry (2002) found that caregivers receiving family preservation services due to the risk of having their child placed in foster care were not provided with services related to domestic violence, despite its presence. Similarly, Shephard and Raschick (1999) report that child welfare workers in their study did not refer battered women with open CWS to DV services. In response to vignettes about families entering CWS, workers indicated that domestic violence treatment would be a primary intervention in 8.5% of the families with caregivers victimized by DV (Jones & Gross, 2000). Even though children in homes with DV and child maltreatment are at high risk of a variety of untoward outcomes, identification of domestic violence does not appear to consistently lead to a caregiver’s referral to DV services. In summary, the caregivers of children investigated for maltreatment have often been victims of DV; however little is known about their pathway through the child welfare system. The purpose of this study is to examine the identification of DV by child welfare

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1207

workers during investigations of child maltreatment and to determine how this contributes to the receipt of DV services. The specific aims are to: 1. Determine how often child welfare workers identify DV in cases entering the child welfare system due to allegations of child maltreatment and assess how the worker’s identification of DV compares to caregiver self-report of DV victimization. 2. Examine what factors are associated with the child welfare worker’s under identification of DV in cases already reported to the child protection system, in which the caregiver has reported this violence. 3. Determine the levels of domestic violence service use over the 18-month period following the investigation of maltreatment. 4. Examine what factors are associated with referral and receipt of DV services, including the opening of a CWS case.

2. Methods 2.1. Study design The National Survey of Child and Adolescent Well-Being (NSCAW) is a national, probability study of children entering CWS and draws on information from the child welfare workers and caregivers. The NSCAW sample includes 5504 children, ages birth to 15, who were investigated by CWS following an allegation of maltreatment. NSCAW has a two stage stratified design, with the first level being primary sampling units (i.e., county child welfare agencies) and the second being children investigated for maltreatment. The children were selected from 92 primary sampling units proportionate to size in 97 counties, located in 36 states (see NSCAW Research Group, 2002 for further information on the sampling design). 2.2. Sample characteristics Analyses for this study involve the permanent caregivers of children remaining in the home following allegations of maltreatment. Nearly all (91%) of this group are female; therefore, for purposes of this paper, analysis will be limited to female, permanent caregivers (N = 3135). Permanent caregivers may include relative caregivers who are the legal custodians of children (i.e., they are not in kinship foster care). Caregivers were included in the study regardless of the outcome of the child maltreatment investigation. This allows for comparisons between caregivers in families who did and did not receive CWS. Families receiving ongoing CWS had some level of follow up contact with the child welfare agency following the investigation, while those without CWS were closed to services. In this sample of female caregivers of children remaining at home, 27% received CWS and 73% did not get these services. With regard to demographic characteristics, the female caregivers reporting any victimization by recent DV were similar to those without reported DV victimization (see Table 1). The racial/ethnic mix of the sample is 25% African American, 51% White, 17%

1208

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

Table 1 Demographic characteristics of female caregivers of children remaining in-home (N = 3135) Total

DV reported by CG

No DV

a

Caregiver age 24 or younger 25–34 35–44 45 and older Caregiver race/ethnicityb Black White Hispanic Other Poverty level 50% or less 50–99% 100–149% 150–199% 200% or more Intimate partnerc Caregiver education Less than high school/GED High school/GED Vocational/associate/RN Bachelor degree or higher Other Total

20.2% 44.7 27.9 7.2

24.0% 38.7 33.9 3.3

18.5% 47.3 25.3 9.0

25.2% 50.8 16.7 7.3

24.7% 53.9 12.5 9.0

25.4% 49.4 18.6 6.6

26% 34 16 10 15 47.8%

23% 30 20 13 15 42.5%

32.4% 43.6 19.6 2.5 1.8 30.5

29.2% 44.9 20.4 3.2 2.3 69.5

24% 31 18 12 15 44% 30.2% 44.5 20.1 3.0 2.2 100.0

Unweighted sample size reported; however, proportions based on weighted data. a v 2 = 15.1, p b .01. b Unweighted sample size equals 3132 for this analysis due to 3 missing values on the race/ethnicity variable. c Indicates whether a spouse or other intimate partner was living in the home at the time of the investigation.

Hispanic, and 7% other. The sample is disproportionately undereducated with 30% having less than a high school education and nearly 45% having only a high school diploma or GED. There were no significant differences by race or ethnicity, poverty level, education, or presence of spouse or other intimate partner in the home; however, the two groups were significantly different on age (v 2 (3, N = 3135) = 15.1, p b .01). 2.3. Procedures Institutional Review Board approval was sought and obtained from several institutions, including the Research Triangle Institute, four states, San Diego Children’s Hospital, and the University of North Carolina at Chapel Hill. The indicators for DV used in this study come from two sources: child welfare worker and caregiver interviews. An NSCAW Field Representative conducted face to face interviews with the permanent caregiver of children remaining in home with CWS or without CWS at baseline and 18-months. The field representative read an informed consent to the caregiver, which summarized the study purpose, data confidentiality protections, and other appropriate details of the study (Dowd et al., 2003). The caregiver was given a $50 cash incentive for participating in each

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1209

interview. The child welfare worker also provided informed consent and participated in a face-to-face interview at baseline, 12 months, and 18 months following the investigation. 2.4. Measures 2.4.1. Child welfare worker risk assessment The child welfare worker risk assessment asks if active domestic violence toward the caregiver is present and if there is a history of DV in the home. These items were completed by the worker who had the greatest knowledge of the case, between 2 and 6 months after the beginning of the investigation. This time delay was necessitated by the need to have monthly samples drawn from the population of all investigations and to then complete interviews with workers. Therefore, the child welfare worker should have had the benefit of more time for the determination of the presence or history of DV. At the same time, because of turn over among workers, some of the respondents were working from the case record, and had not been the investigating worker. Additional information about other risks in the family was also obtained from the child welfare worker using a checklist of potential risks. The worker indicated characteristics of the caregiver, such as alcohol or drug abuse, childhood history of abuse or neglect, and emotional or mental health problems. If a second adult caregiver lived in the home, the worker also indicated risks for this adult, such as alcohol and drug abuse. Another risk factor assessed by the worker is whether the family had received CWS prior to the allegations which lead to inclusion in NSCAW. 2.4.2. Conflict Tactics Scale The Conflict Tactics Scale (CTS; Straus, 1990) is a self-report measure designed to assess the strategies family members use to respond to conflicts and is the most frequently employed measure of DV. The physical violence scale was used in this study to assess caregivers’ victimization by an intimate partner. The reliability (a = .88) of the physical violence section was established on a national probability sample (N = 2143) and the validity of the measure has been well documented (Straus, 1990). The physical violence scale is divided into two subscales: minor or less severe violence (e.g., being pushed, grabbed, shoved, or slapped) and severe violence (e.g., being choked, beaten up, and threatened with a knife or gun). The CTS was administered via audio computer-assisted self-interview to all caregivers of children remaining in the home following the maltreatment investigation. In our use of the CTS, a score for any domestic violence was generated as an aggregate of the severe and less severe items. 2.4.3. Domestic violence services Data were also collected from the caregiver and the child welfare worker on referral and service receipt of DV services. Following the questions about DV victimization on the caregiver interview, the women were asked about DV services. Because the interview may have been several months after the investigation which lead to inclusion in the NSCAW study, caregiver report of DV service related issues included responses on both the baseline and 18 month follow-up. First, she was asked if she had been breferred to domestic violence services, like a battered women’s shelter or a program to help deal with

1210

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

an abusive partner in the last 12 months.Q The next question inquired if the caregiver had stayed in a shelter or received other services during the same time period. If she responded that she had not received services, she was then asked to rate her need for DV services: (1) a lot, (2) somewhat, (3) a little, or (4) not at all. For purposes of data analysis, this variable was dichotomized into 1=need and 2=no need. Finally, the caregiver was questioned about possible reasons for not receiving DV services, such as services not available in the area, wait-listed, or transportation problems. The child welfare worker was asked about the provision of DV services for the caregiver at both 12 and 18 month follow-up. In examining worker report of referral and receipt of DV services, responses from both time points were considered. The worker indicated if the caregiver had a need for DV services in the previous 12 months and whether or not agency staff made a referral to these services. When a referral was made, a follow up question inquired as to whether the referral resulted in the receipt of services. 2.5. Data analysis approach Descriptive statistics were calculated on demographic characteristics of the overall sample and for caregivers who did and did not report DV victimization within the 12 months preceding the baseline interview. Next, analyses were conducted to identify the level of agreement between caregiver report of DV and child welfare worker report of DV. The rates for sensitivity and specificity were determined. In epidemiologic research, sensitivity refers to true positives and specificity is true negatives (Mausner & Kramer, 1985). The standard criterion for this study is caregiver report of DV victimization; therefore, a true positive is defined as instances when the worker identified DV in families where the caregiver had also reported recent DV victimization. Logistic regression analysis was then used to examine the factors influencing the under identification of DV by the worker. Cases were coded as under identified when the caregiver reported DV victimization, but the worker indicated the absence of DV on the risk assessment item. The predictor variables included in the model were caregiver age and race, CWS status (receipt of CWS or no CWS following the investigation), most serious maltreatment type, presence of an intimate partner in the home, substance abuse by the primary caregiver, substance abuse by the secondary caregiver, prior receipt of CWS, caregiver’s childhood history of abuse and neglect, and a cumulative risk score. This cumulative risk score was derived from 22 items responded to by the worker about child, caregiver, and environmental risks, including DV. This score sought to ascertain the additive nature of risk factors in families undergoing investigations for child maltreatment. The calculation took into account the presence of a secondary caregiver in the home. The number of risks identified by the child welfare worker was summed and then divided by the total number of items applicable to each family. This score was categorized into three levels: low (b20% of the total applicable risks present), moderate (20% to b 40% of the total applicable risks present) and high (z 40% of the total applicable risks present). The correlations between the cumulative risk score and the other risk assessment items used for these analyses ranged from .33 (prior experiences with CWS) to .48 (caregiver’s history of child abuse and neglect). To ensure that multicollinearity was not a problem for the logistic regression model, the variance inflation factors were obtained. All independent

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1211

variables were well below the critical value of the variance inflation factor = 10, with the highest being the cumulative risk score (VIF = 2.2). Analyses then focused on ascertaining the level of DV service receipt in the 18 months following the investigation of child maltreatment. Two separate logistic regression analyses were carried out to model referral to DV services and DV service receipt. The same predictor variables were used in these models as in the above logistic regression. In addition, child welfare worker identification of DV on the risk assessment and whether the family had trouble meeting basic needs were included in these models. The values of the variance inflation factors were 2.0 or lower for all of the independent variables in both of the models. To account for the complex sampling design of NSCAW, which uses clustering and stratification, all analyses were conducted in SUDAAN which correctly adjusts the standard errors. The use of weights allows for national generalizations to all families whose children remain in home at the time of the investigation for child maltreatment. The unweighted sample sizes are reported in the tables; however, in all instances analysis was conducted on the weighted data. Although the overall sample size is 3132, analysis often relied on smaller sample sizes due to missing data on particular items. Analyses for this study used both caregiver report of DV victimization and child welfare worker identification of DV on the risk assessment. A noted limitation in making comparisons between worker and caregiver reports is the varying time frames. The child welfare worker risk assessment asks about active and history of DV, while the CTS assesses DV victimization within the past 12 months and ever. Preliminary analysis indicated that severity of violence reported by the caregiver was not associated with referrals and service receipt. Therefore, the presence of any domestic violence, either severe or minor, was used in all analyses. Analyses focused on active or recent DV because this is a more likely predictor of current need for DV services.

3. Results 3.1. Caregiver and caseworker reports of domestic violence Overall, 43% of the caregivers report lifetime prevalence of less severe DV victimization and 29% have experienced such violence within the 12 months prior to the baseline interview. About a fifth (19%) of all caregivers in the study report victimization by severe DV within the past year and a third (33%) report lifetime prevalence of severe DV victimization. When considering the aggregate score of any DV victimization, 31% of female caregivers self-reported at least one incident of DV in the past 12 months and 45% reported lifetime DV on this measure. Analyses were undertaken to examine how often child welfare workers identify DV and how this identification compares with caregiver self-report. Caregivers reported the presence of DV at much higher rates than was identified by the workers on the risk assessment. While 31% of caregivers reported domestic violence victimization in the past year, workers only identified this violence in 12% of all cases. Overall, the sensitivity between worker and caregiver report of DV victimization is very low. Both the caregiver and the worker reported active or recent DV in only 8% of the families (see Table 2).

1212

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

Table 2 Level of agreement between caregiver and child welfare worker report of active domestic violencea

Sensitivity (true positive) (CG=Yes, CWW=Yes) Specificity (true negative) (CG=No, CWW=No) Over identified (CG=No, CWW=Yes) Under identified (CG=Yes, CWW=No) Total

Total

In-home no CWS

In-home with CWS

8.2% 65.4 4.4 22.0 100.0

8.0% 67.6 3.4 21.0 100.0

9.0% 59.3 6.8 24.9 100.0

Unweighted N’s are reported; however proportions based on weighted data. CG=caregiver, CWW=child welfare worker. av 2 = (1, N = 2819) = 11.03, p b .05.

Under identification occurred in nearly a quarter of the families—the worker did not identify DV when the caregiver had reported DV in 22% of the cases. An open CWS case only slightly increased the accurate identification of DV (9% compared to 8%); however; under identification rates were actually higher for caregivers receiving CWS (25%) than families without services (21%). 3.2. Predictors of caseworker under identification of domestic violence To better understand the level of agreement between caregiver and child welfare worker report of DV, logistic regression analysis was carried out to determine factors associated with the under identification of DV in families where the caregiver reported any DV victimization on the CTS. Table 3 shows regression coefficients, Wald statistics, odds ratios, and 95% confidence intervals for each of the predictors included in the model. Forty two percent (42%) of the variance in under identification of DV during maltreatment investigations was accounted for by the predictors in the model (Cox and Snell pseudo R 2 = .42). Identification of caregiver substance abuse problems by the child welfare worker on the risk assessment is a strong predictor of under identification of DV by the worker when controlling for other variables in the model.1 When substance abuse was indicated, the worker was over seven times more likely to under identify DV (OR = 7.6). They were also less likely to accurately identify DV when the family has had prior experience with CWS (OR = 3.5) or the caregiver has a childhood history of abuse or neglect (OR = 4.8). Some factors reduce the likelihood of under identification of DV in families already reported to the child welfare system. Substance abuse by the secondary caregiver reduces the likelihood that the child welfare worker will miss the presence of DV in the home. The worker was only 29% as likely to under identify DV when a substance abuse problem was indicated for the secondary caregiver (OR = .29). Child welfare workers were also less likely to overlook DV when the most serious maltreatment type was classified in the bOtherQ category (which included maltreatment types, such as emotional maltreatment, abandonment, educational neglect, and exploitation) as compared to physical abuse. In an effort to control for the additive nature of risk in many families in CWS, the cumulative risk score was included as a predictor variable in the model. In contrast to families with low risk, there was no probability of the child welfare worker under identifying DV when the 1

In comparable models, with different variations in the risk score (not shown), the substance abuse of the primary caregiver was not significant.

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1213

Table 3 Logistic regression analysis predicting child welfare worker under identification of domestic violence from caregiver, family, and maltreatment characteristicsa Predictor variables (Intercept) Caregiver age Caregiver race/ethnicity Black White Hispanic Other Child setting In-home no CWS In-home with CWS Most serious maltreatment type Physical abuse Sexual abuse Neglect: failure to provide Neglect: failure to supervise Other** Prior CWS Caregiver’s childhood history of CAN Intimate partner in the home Substance abuse PCG** Substance abuse SCG* Cumulative risk count Low Medium High***

b 2.3* .01

Wald F

Odds ratio

95% CI Upper

Lower

.99

.95

1.0

1.3

.44

4.0

4.8 1.3

.46 .34

50.7 4.7

1.6

.56

4.7

2.6 3.2 1.7 .09 3.5 4.8 1.5 7.6 .29

.58 .69 .56 .03 1.4 1.3 .62 1.8 .12

12.0 14.3 4.9 .33 8.8 17.4 3.8 32.7 .71

.13 .00

.03 .00

.68 .01

16.2*** .7

.3 Reference group 1.6 .2 .8 Reference group .5 5.5*** Reference group 1.0 1.2 .5 2.4*** 1.2* 6.9* 1.6* 5.7* .43 .9 2.0** 7.5** 1.2** 7.6** 21.7*** Reference group 2.0* 6.2***

Unweighted N is reported; however analysis conducted on weighted data. Reference group is absence of risk for dichotomous variables. CAN = child abuse and neglect, PCG = primary caregiver, SCG = secondary caregiver. *p b .05, **p b .01, ***p b .001. a Cox and Snell pseudo R 2 = .42, N = 507.

cumulative risk score was assessed as high—an indication that in families with the highest number of risks, workers are more likely to also identify the presence of DV. 3.3. Caregiver and caseworker reports of need, referral, and service receipt Efforts were undertaken to better understand child welfare worker and caregiver report of need, referral, and receipt of DV services over the 18-month period following the investigation of maltreatment. This analysis was conducted on smaller sample sizes due to the fact that questions on referral and receipt of DV services were not administered to workers if the case was no longer open to CWS. Sample sizes are noted in the table along with the results (Table 4). No significant differences were found between the group without CWS and the group receiving CWS on either the worker or caregiver indicators of DV service related items. Overall, when the worker identified active DV toward the caregiver, a service need was indicated for 57% of the caregivers and a referral was made

1214

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

Table 4 Caregiver and child welfare worker report of domestic violence service need, referral, and service receipt by 18month follow-upa CWW identified active DV

CWW indicated: Service need (N = 292) Referral (N = 289) Service receiptb (N = 144) Caregiver indicated: Service need (N = 785) Referral (N = 924) Service receiptc (N = 893)

Caregiver identified recent DV

Total

No CWS

W/CWS

56.8 59.6 83.2

52.9 58.6 83.4

63.1 60.9 83.0

Total

No CWS

W/CWS

47.2 38.3 19.9

43.5 37.6 20.4

55.3 39.9 18.6

Unweighted N’s are reported; however proportions based on weighted data. a Female Caregivers of children remaining in-home. b Child welfare worker reported service receipt only when a referral was indicated. c All caregivers responded to question about DV services.

for 60% of the women. The child welfare worker reported that services were received by 83% of the caregivers for whom a referral was made. The proportions of caregivers who reported DV victimizations and also indicated a service need, referral, and service receipt were lower than those reported by the worker. Only 47% of the battered women identified the need for services and 38% reported that she had been referred for DV services. Of all caregivers with DV victimization, regardless of referrals, 20% stayed in a battered women’s shelter or received some other type of DV service over the 18-months. Reasons the caregiver did not receive DV services following the identification of a need for services were determined. In the interview at the end of the study period, the women were asked about seven possible reasons they did not get DV services. Scheduling and child care problems (26%) and lack of transportation (18%) were frequently indicated as reasons that the caregiver did not participate in DV services. The unavailability of services was cited by 11% of the caregivers; 3% indicated that they were put on a waiting list when they attempted to get DV services; and 14% did not have the financial means to cover the cost of the service. The bOtherQ category was the most frequently indicated reason; however this was difficult to interpret because there was not a way to discern the specific reason. 3.4. Predictors of referral and domestic violence service receipt Additional multivariate analyses were conducted to determine factors associated with the referral to DV services and DV service receipt. Both caregiver and child welfare worker report of referral and use of services were considered in an effort to capture all referrals and services. It was counted as a referral or DV service receipt if either the worker or caregiver reported a referral or service use. Identical predictor variables were included in the separate logistic regression models on referrals to and receipt of DV services (see Table 5). The full model accounted for 24% of the variance in referrals to DV services reported by either the worker or caregiver (Cox and Snell pseudo R 2 = .24). Child welfare worker identification of active DV was a significant contributor to the report of referrals. Holding all other variables in the model constant, a referral was over nine times more

Table 5 Logistic regression analyses predicting referral to domestic violence services and domestic violence service receipt reported by the child welfare worker or caregiver from caregiver, family, and maltreatment characteristics Predictor variables

Dependent variable: referral to DVSa 95% CI

b

1.0

.97, 1.0

3.1* .03

.46

.20, 1.1

.46 .77 9.3

.21, 1.0 4.6, 18.8

1.3 .74 1.4

.7, 2.2 .39, 1.4 .68, 3.1

1.4 1.6 1.1 2.2 1.1

.58, .72, .55, .82, .45,

.00

.98

.44, 2.2

1.4

5.7* 3.9*

2.2

.94, 3.8

1.3^

1.9 .84

.94, 3.8 .30, 2.3

Wald F

Odds ratio

1.5* 0

.05 1.8

.78 Reference group .77 .26 2.2*** 40.0*** .64 Reference group .22 .30 .85 .37 .93 .93 Reference group .36 .50 .06 .79 .06 .02 .02 .78* Reference group .63 .018

3.5 3.8 2.0 5.9 2.5

Wald F

Odds ratio

95% CI

.42 .88

.98

.90, 1.1

.45

.07, 2.8

2.6 .08 7.5

.42, 16.6 2.5 1.4, 41.0

3.7

1.0, 13.7

.99

.30, 3.3

.82 .08 .36 1.4 .48

.07, .01, .06, .12, .11,

3.9

.24

.06, 1.0

3.0^ .49

.27

.06, 1.22

3.0 3.4

.28, 31.2 .29, 40.0

.79 Reference group .97 .81 2.0* 5.6* 4.1* Reference group 1.3* .78 1.1 .01 0 2.7 Reference group .20 2.5** 1.0 .34 .73 .96

Reference group .92 .98

1215

Unweighted N’s are reported; however analysis conducted on weighted data. Reference group is absence of risk for dichotomous variables. DVS=Domestic Violence Services. ^p b .10, *p b .05, **p b .01, ***p b .001. a Unweighted N = 976, Pseudo R 2 = .24. b Unweighted N = 283, Pseudo R 2 = .17.

9.1 .42 2.2 16.5 2.1

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

b (Intercept) Caregiver age Caregiver race/ethnicity Black White Hispanic Other CWW report of active DV Child setting In-home no CWS In-home with CWS Intimate partner Any prior CWS Most serious maltreatment type Physical abuse Sexual abuse Neglect: failure to provide Neglect: failure to supervise Other Substance abuse primary caregiver Substance abuse secondary caregiver Difficulty meeting basic needs Cumulative risk count Low Medium High

Dependent variable: DVS receiptb

1216

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

likely to be made for caregivers when the worker identified the presence of active DV (OR = 9.3). A referral to DV services was two times more likely for families assessed by the worker as having trouble meeting their basic needs compared to families who are better able to meet these needs. The predictors in the second model explained 17% of the variance in DV service receipt by caregivers of children investigated for child maltreatment (Cox and Snell pseudo R 2 = .17). The identification of DV by the worker is also a predictor of service receipt. When the worker specified active DV on the risk assessment, the likelihood of the caregiver receiving DV services was more than seven times that of caregivers without DV identified by the worker (OR = 7.5). Further, controlling for the other variables in the model, the provision of CWS was significantly associated with the caregiver’s receipt of DV services (OR = 3.7). A trend worth noting is that families who have difficulty paying for basic necessities were only 27% as likely to receive DV services ( p = .09).

4. Discussion This study sought to determine factors impacting identification of DV by the child welfare worker during investigations of maltreatment. It also identified how a worker indication of DV on the risk assessment and the opening of CWS contributed to referral and receipt of DV services. Important information was gleaned from this study. Child welfare workers often miss recent DV on the risk assessment during investigations of child maltreatment. Nearly a third of all female caregivers of children remaining in the home following an investigation for child maltreatment report any recent DV victimization; however workers reported this in only 12% of the families. The accurate identification of DV by the worker when the caregiver has also identified this violence happens in less than 10% of the families. Further, under identification of DV is evident in nearly a quarter of the families. Although workers accurately identify alcohol or drug dependence at higher rates when the family is receiving in-home CWS (Barth, 2003), this is not the case with DV. The provision of in-home CWS following the investigation does not improve the rate of identification; in fact, under identification is slightly higher when the family receives CWS. One would expect that the worker would gain more knowledge about families when a case was open for services; however, these findings are counterintuitive to this. One possible explanation is that families with DV often have a broader range of problems (Shipman, Rossman, & West, 1999), including substance abuse, which may overshadow the caregiver’s DV victimization. Clearly, the presence of DV is missed in many families undergoing CWS investigations. The apparent role of familial substance abuse in the under identification of DV by the worker is striking. When the child welfare worker assesses the caregiver as having a problem with substance abuse, they are much more likely to miss the DV. In contrast, substance abuse by the secondary caregiver greatly reduces the likelihood of under identification of DV. This suggests that the worker may be more likely to assess for the presence of DV when substance abuse is identified in an intimate partner. Yet, when the battered woman has been assessed as having a problem with drug or alcohol abuse, the worker may believe that this is the primary and singular problem that needs to be addressed. Therefore, once substance abuse is

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1217

identified the worker does not assess for any further risks, resulting in the under identification of DV in cases already reported to the child welfare system. Domestic violence is more likely to be under identified in another, possibly stigmatizing situation, as well—prior CWS experiences. When the family has had prior experiences with CWS, the worker is three times more likely to miss the presence of DV. Given that a higher proportion of families with active DV experience multiple referrals to CWS (Hazen et al., 2004; English et al., 1999), the consequences of not identifying the violence toward the caregiver may perpetuate an unsafe home environment. Indeed, although we could not test the possibility in these analyses, the failure to identify active DV may be one of the reasons for repeated contact with CWS. Recognition of the presence of DV in families investigated for child maltreatment must extend beyond simply identifying the violence. It should serve as a gateway to appropriate DV services. As recommended in the Greenbook (Schechter & Edleson, 1999), case planning must focus on the safety and well-being of all victims within a family. The opening of CWS following an investigation for child maltreatment implies some level of ongoing contact between the agency and the family. The resulting case planning should include interventions to address the range of family risks that contribute to the risk of harm to children. An open CWS case is associated with referrals to DV services and receipt of these services. Caregivers with DV victimization in families receiving CWS are more likely to be connected to DV services; yet, overall, domestic violence services are underutilized as a resource for child welfare workers. Even when the worker identifies DV, they often do not take the next step to ensure that the woman gets appropriate intervention to deal with an abusive partner. While workers identify a DV service need and make a referral for about 60% of the caregivers with worker indicated active DV, connection to appropriate DV services does not occur for the other 40% of the women recognized by the worker as having active DV. Although identification of DV by the worker leads to more referrals and receipt of services, efforts need to be undertaken to increase the level of referral and service use in all DV indicated cases. As with all interventions in human services, barriers to service receipt are evident. The availability of DV services did not appear to be a major obstacle to use of services. Only 11% of the caregivers identified this as the reason they did not get needed services and only 3% indicated that a waiting list for services was a barrier. Yet, many caregivers report that they could not use available services. Lack of transportation and scheduling or child care problems were frequently identified as barriers to receiving services. Although many DV services are provided at no cost, families with the most economic difficulties still had greater barriers to service use. This is evidenced by the findings that families who had difficulty paying for the basic necessities were more likely to be referred to DV services, but less likely to have received services. Of those referred to DV services, it appears that the poorest families have more difficulty accessing services.

5. Implications and future directions Overlooking domestic violence may impede efforts to protect children and provide appropriate services to families. When DV is unidentified by the child welfare system,

1218

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

service needs of children and mothers are more likely to be unmet. The efforts to develop interagency policies and procedures to address issues related to children exposed to DV depend on accurate initial determinations of the co-occurrence of DV and child maltreatment. When this assessment is faulty, the benefit of interagency protocols is reduced. The findings of this study confirm studies from many decades (e.g., Geismar & Ayers, 1958; Gordon, 1988; Lawder, Poulin, & Andrews, 1984) indicating that problems of families involved with child welfare services are multidimensional and often involve coexisting DV and substance abuse. It highlights the need for child welfare workers to assess for all possible risks. Given that prior experiences with CWS are associated with the worker overlooking DV in families, this may be especially important when there are multiple experiences with CWS due to child maltreatment. The risk assessment must also be a tool in case planning when the investigation warrants an open CWS case. Further, when DV is indicated during the investigation a referral should be made even when the family is closed to ongoing CWS. Services should include all types of appropriate intervention, such as substance abuse treatment and DV services. Every CWS agency needs to be sure that domestic violence is assessed at all stages of contact with a family. An emerging strategy that holds promise is the use of domestic violence specialists in child protection agencies (Findlater & Kelly, 1999; Schechter & Edleson, 1999). This specialist can assist families in overcoming barriers to DV service receipt, such as child care and transportation. In addition to serving as a direct resource to battered women, the DV specialist can provide support and training to child welfare workers, and work to strengthen relationships with other service providers in the community, including DV advocates. Training related to the co-occurrence of DV and child maltreatment is needed by child welfare workers, although very few training protocols have been formally tested. One noted exception is the evaluation of two training curricula in California (Mills & Yoshihama, 2002). Their findings were encouraging. After participating in the training, workers were more likely to recognize the importance of assessing for DV and they also felt more confident in their ability to respond to DV. Evidence from previous CWS training evaluations on risk assessment implementation suggest that training should be ongoing and not limited to a one time, short course (Cash, 2001). Training goals should go beyond risk assessment and should seek to: (1) increase the worker’s recognition that DV is related to the safety of the child and interventions targeting the DV and supporting the adult victim will ultimately increase the child’s safety; (2) increase identification of DV in child maltreatment; and (3) increase referrals to appropriate DV services. In general, increased amounts of agency in-service training tended to increase the screening for DV compared to CEU’s for Masters level social workers in Florida (Tower, 2003). This may suggest that the nature of the agency and/or policies within individual agencies impact the attention given to DV more so than an individual worker’s knowledge of DV. Change efforts cannot focus solely on the child welfare worker. Effectively responding to co-occurring maltreatment and DV requires a system change. Child welfare agencies must embrace the framework outlined in the Greenbook and develop strategies that work for their agency and larger community. Change must involve improving their internal capacity to address DV, as well as building relationships with DV organizations in their community. Workers will only be able to consistently and effectively intervene in these families when changes occur at the agency and policy level.

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1219

Recognition of DV among caregivers in CWS must go beyond increasing the identification of DV during the investigation of maltreatment. If the process stops there, the safety and well-being of the child is no better off. Battered caregivers must be given the opportunity to participate in interventions geared at addressing their safety and wellbeing, as well. In order for this to effectively happen, additional system wide issues must be addressed. Risk assessment technology, including reliable and valid standardized instruments based on current research, must be developed and empirically tested (Rycus & Hughes, 2003). Determining risk for case planning should involve the conjoint assessment of all potential risk factors, such as domestic violence and substance abuse. The barriers between multiple systems that address issues of domestic violence and child maltreatment must be overcome to increase the ability to provide appropriate services to caregivers of allegedly maltreated children who have experienced DV victimization. The legal mandate of the child welfare system in regards to co-existing DV and maltreatment must be clarified so that it can serve as a guide in the development of agency level policies and procedures. Placing an expectation of child welfare agencies to change their response to cases with DV, without providing clear and consistent governing policies, makes successful implementation of innovative strategies very difficult. Future research should continue to disentangle the relationship of DV and substance abuse among caregivers in CWS. One pressing issue is to understand how under identification of DV in families with substance abuse problems effects case planning and outcomes. Under identification may contribute to multiple episodes of involvement with CWS. This needs to be examined with more rigorous program evaluation following the institution of interventions to increase awareness of the co-occurrence of these two problems. Acknowledgements Support for the study comes from the National Survey of Child and Adolescent WellBeing (Barth), funded by the Administration for Children and Families of the U.S. DHHS, a Children’s Bureau pre-doctoral fellowship (Kohl), and the National Institute on Justice (Hazen and Landsverk). Conclusions do not necessarily represent those of ACF or NIJ. References Allen, M., & Bisell, M. (2004, Winter). Safety and stability for foster children: The policy context [Electronic version]. The Future of Children, 14, 49 – 73. Barnett, O., Miller-Perrin, C., & Perrin, R. (1997). Family violence across the lifespan: An introduction. Thousand Oaks, CA7 Sage. Barth, R.P. (2003, December). Substance abuse and child welfare services: Research update and needs. Paper presented at the National Center on Substance Abuse and Child Welfare Researcher’s Forum, Washington DC. Carter, L. S., Weithorn, L. A., & Behrman, R. E. (1999, Winter). Domestic violence and children: Analysis and recommendations. The Future of Children, 9, 4 – 20.

1220

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

Cash, S. J. (2001). Risk assessment in child welfare: The art and science [Electronic version]. Children and Youth Services Review, 23, 630 – 811. Cash, S. J., & Berry, M. (2002). Family characteristics and child welfare services: Does the assessment drive service provision [Electronic version]. Families in Society: The Journal of Contemporary Human Services, 83, 499 – 507. Dowd, K., Kinsey, S., Wheeless, S., Thissen, R., Richardson, J., & Suresh, R., et al. (2003). National survey of child and adolescent well-being (NSCAW): Combined waves 1–3 data file user’s manual restricted release version. Ithaca, NY7 National Data Archive on Child Abuse and Neglect. Edleson, J. L. (1999). Children’s witnessing of adult domestic violence [Electronic version]. Journal of Interpersonal Violence, 14, 839 – 870. Edleson, J. L. (2004). Should exposure to adult domestic violence be defined as child maltreatment under the law? In P. G. Jaffe, L. L. Baker, & A. Cunningham (Eds.), Protecting children from domestic violence: Strategies for community intervention. New York, NY7 Guilford Press. English, D. J., & Graham, J. C. (2000). An examination of relationships between Children’s Protective Services social worker assessment of risk and independent LONGSCSN measures of risk constructs. Children and Youth Services Review, 22, 897 – 933. English, D. J., Marshall, D., Brummel, S., & Orme, M. (1999). Characteristics of repeated referrals to Child Protective Services in Washington state. Child Maltreatment, 4, 297 – 309. Findlater, J. E., & Kelly, S. (1999). Child protective services and domestic violence. Future of Children, 9(3), 84 – 96. Fogarty, C. T., Burge, S., & McCord, E. C. (2002). Communicating with patients about intimate partner violence: Screening and interviewing approaches [Electronic version]. Family Medicine, 34, 369 – 375. Geismar, L., & Ayers, B. (1958). Families in trouble. St. Paul, MN7 Family-Centered Project. Gibbons, C., & Barth, R. P. (submitted for publication). Prevalence of substance abuse among in-home caregivers in a U.S. child welfare population: Caregiver vs. child welfare worker report. Gordon, L. (1988). Heroes of their own lives: The politics and history of family violence. New York7 Viking Press. Hazen, A., Connelly, C. D., Kelleher, K., Landsverk, J., & Barth, R. (2004). Intimate partner violence among female caregivers of children reported for child maltreatment. Child Abuse and Neglect, 28, 301 – 319. Hess, P. M., Folaron, G., & Jefferson, A. B. (1992). Effectiveness of family reunification services: An innovative evaluation model. Social Work, 37, 304 – 311. Hutchison, E. D. (1987). Use of authority in direct social-work practice with mandated clients. Social Service Review, 61(4), 581 – 598. Jones, L. P., & Gross, E. (2000). Perceptions and practice with domestic violence among child protective service workers. Children and Youth Services Review, 22, 355 – 371. Kantor, G. K., & Little, L. (2003). Defining the boundaries of child neglect: When does domestic violence equate with parental failure to protect? Journal of Interpersonal Violence, 18, 338 – 355. Kaufman, L. (2003, November 28). Abused mothers keep children in a test of rights and safety. The New York Times, pp. A1, C10. Lawder, E. A., Poulin, J. E., & Andrews, R. G. (1984). Helping the multi-problem family: A study of services to children in their own homes (SCOH). Philadelphia, PA7 Children’s Aid Society of PA. Magen, R. H., Conroy, K., Hess, P. M., Panciera, A., & Levy Simon, B. (2001). Identifying domestic violence in child abuse and neglect investigations [Electronic version]. Journal of Interpersonal Violence, 16, 580 – 601. Mausner, J. S., & Kramer, S. (1985). Epidemiology: An introductory text. Philadelphia7 W.B. Saunders Company. Mills, L. G. (2003). Insult to injury: Rethinking our responses to intimate abuse. Princeton, NJ7 Princeton University Press. Mills, L. G., Friend, C., Conroy, K., Fleck-Henderson, A., Krug, S., & Magen, R. H., et al. (2000). Child protection and domestic violence: Training, practice, and policy issues. Children and Youth Services Review, 22, 315 – 332. Mills, L. G., & Yoshihama, M. (2002). Training children’s services workers in domestic violence assessment and intervention: Research findings and implications for practice. Children and Youth Services Review, 24, 561 – 581. New York State Adoption and Safe Families Act, A.962-A Green §.2346 Saland (1999).

P.L. Kohl et al. / Children and Youth Services Review 27 (2005) 1203–1221

1221

NSCAW Research Group. (2002). Methodological lessons from the national survey of child and adolescent wellbeing: The first three years of the USA’s First National Probability Study of Children and Families Investigated for Abuse and Neglect. Children and Youth Services Review, 24, 513 – 541. Rycus, J. S. & Hughes, R. C. (2003). Issues in risk assessment in child protective services. Retrieved March 18, 2004 from http://www.ihs-trainet.com Schechter, S., & Edleson, J. L. (1999). Effective intervention in domestic violence and child maltreatment cases: Guidelines for policy and practice. Reno, NV7 National Council of Juvenile and Family Court Judges. Shepard, M., & Raschick, M. (1999). How child welfare workers assess and intervene around issues of domestic violence. Child Maltreatment, 4, 148 – 156. Shipman, K. L., Rossman, B. B. R., & West, J. C. (1999). Co-occurrence of spousal violence and child abuse: Conceptual implications. Child Maltreatment, 4, 93 – 102. Straus, M. (1990). Measuring physical and psychological maltreatment of children with the conflict tactics scales. Durham, NH7 University of New Hampshire Family Research Laboratory. Tower, L. E. (2003). Domestic violence screening: Education and institutional correlates [Electronic version]. Journal of Social Work Education, 39, 479 – 494. Weithorn, L. A. (2001, November). Protecting children from exposure to domestic violence: The use and abuse of child maltreatment statutes [Electronic version]. Hastings Law Journal, 53, 1 – 153.