|Home | About | Journals | Submit | Contact Us | Français|
The purpose of this study was to describe policy and practice with respect to the assessment of intimate partner violence in a sample of child welfare agencies located throughout the United States and to examine the relationship of contextual characteristics and assessment practices. Telephone interviews were conducted with key informants from child welfare agencies. A snowball interviewing strategy was used to identify the best informant in each agency. Almost all of the participating agencies conducted some assessment of intimate partner violence, with most reporting that the majority of screening or assessment occurred during investigation of referrals. However, only 43.1% reported that all of the families referred to the child welfare system were assessed for intimate partner violence, and 52.8% indicated they had a written policy pertaining to screening and assessment of the problem. There was little relationship between county or agency characteristics and assessment practices. Additional research is needed to determine factors that influence assessment practices and to identify strategies to support and extend efforts to identify intimate partner violence and provide appropriate services for families in the child welfare system.
Intimate partner violence affects the lives of significant numbers of women and children over the life course. A recent national survey reported that approximately one in five women have experienced physical assault and almost one in ten have experienced sexual assault by an intimate partner at some time in their lives (Tjaden & Thoennes, 2000). Children have been found to be overrepresented in homes in which intimate partner violence occurs (Fantuzzo, Boruch, Beriama, Atkins, & Marcus, 1997), with estimates indicating that between 11% and 20% are exposed to such violence during childhood (Wolak & Finkelhor, 1998). In addition to observing violence between their parents or caregivers, these children are themselves often the victims of verbal, physical, and sexual abuse (Kerker, Horwitz, Leventhal, Plichta, & Leaf, 2000; Ross, 1996; Straus & Smith, 1995; Tajima, 2000). A review of studies examining the relationship between adult intimate partner violence and child maltreatment found a median co-occurrence of 40% (Appel & Holden, 1998). Many children from families with intimate partner violence come in contact with the child welfare system due to the abuse they have directly experienced or because of their exposure to violence between their caregivers. Recent studies have reported that intimate partner violence is a common problem among families involved with this system, with findings suggesting that 30% to 40% are affected (Edleson, 1999; English, Edleson, & Herrick, 2005; Hazen, Connelly, Kelleher, Landsverk, & Barth, 2004; Jones, Gross, & Becker, 2002; Magen, Conroy, McCartt Hess, Panciera, & Simon, 2001; Shepard & Raschick, 1999).
Recognition of the substantial overlap between intimate partner violence and child maltreatment and the high rates of intimate partner violence in families involved with child welfare has prompted several national organizations to recommend that screening for intimate partner violence should be standard practice in all stages of child protection cases (Carter & Schechter, 1997; National Association of Public Child Welfare Administrators, 2001; National Council of Juvenile and Family Court Judges, 1999). Little is currently known about child welfare practice in assessing intimate partner violence, but recent research has suggested that the problem is not always identified in families who come in contact with this system. Using data from a nationally representative sample of families investigated for child maltreatment, Kohl and colleagues (Kohl, Barth, Hazen, & Landsverk, 2005) found that 31% of female caregivers reported that they experienced intimate partner violence in the preceding year but child welfare workers identified this violence in only 12% of the families. Caregiver and child welfare worker reports overlapped in 8% of the cases, and workers did not identify intimate partner violence when the caregiver reported it in 22%. Under-identification by child welfare workers was associated with the female caregiver's drug or alcohol abuse, with a family having prior contact with the child welfare system, and with the female caregiver having a childhood history of abuse or neglect. A small number of studies suggest that there are limitations in the assessment of intimate partner violence among families involved with child welfare. In a study of referrals investigated by child protective services in Washington State, 40% to 50% were not assessed for intimate partner violence (English, Marshall, Brummel, & Orme, 1999). Another study found that workers conducted an assessment of intimate partner violence in only 45% of opened child welfare cases (Shepard & Raschick, 1999).
Recent studies have indicated that families with intimate partner violence are likely to have multiple referrals to the child welfare system (English et al., 1999; Hazen et al., 2004; Kohl, Edleson, English, & Barth, 2005). They also appear to have a greater likelihood of experiencing a range of risk factors, including substance use, mental health problems, and criminal involvement compared to other families in the child welfare system without intimate partner violence (Hazen et al., 2004; Kohl, Edleson, et al., 2005). These findings suggest the importance of appropriate assessment and identification of intimate partner violence for improving family safety and well-being.
The purpose of the present study was to describe policy and practice with respect to the assessment of intimate partner violence in a sample of child welfare agencies located throughout the United States and to examine the relationship of contextual characteristics and assessment practices.
Data for this study come from the Children and Domestic Violence Services (CADVS) study (Smith et al., 2005). The CADVS study was designed to collect contextual data at the state and local levels on child welfare policy and practice related to intimate partner violence and on the relationship between child welfare agencies and community domestic violence services. CADVS is a supplemental study of the National Survey of Child and Adolescent Well-Being (NSCAW), a national probability study of over 6,000 children who came in contact with the child welfare system between October 1999 and December 2000. NSCAW used a stratified two-stage sampling procedure with the first stage involving the selection of Primary Sampling Units (PSUs), which were generally defined as geographic areas that encompass the population served by a single child welfare agency (typically a single county). The second stage involving the selection of children from lists of closed investigations from the sampled agencies. Data were collected in 92 of the 100 PSUs sampled. (For additional detail on the NSCAW methods, see (NSCAW Research Group, 2002)).
The CADVS study used the same sampling frame as the NSCAW study. Child welfare services (CWS) agencies in the 92 PSUs involved in the NSCAW study were contacted to participate in CADVS. A key informant in each CWS agency had been obtained from a prior study also linked to NSCAW, the Caring for Children in Child Welfare (CCCW) study, which examined the relationship between variations in the organization and financing of mental health services and the use of mental health and other services by children involved with the child welfare system. A snowball interviewing strategy was used in which the key informant identified by the CCCW study was initially contacted and asked to identify individuals who would be the best informants for the various issues inquired about in the CADVS interview. Nominated individuals received information on the study, an interview summary, and a copy of the consent form. Research assistants then contacted these individuals by telephone to confirm receipt of the study materials, determine their willingness to participate, and ensure that they were the best available informants.
Interviews were conducted by telephone with informants from the CWS agencies. Prior to commencing the interview, a research assistant reviewed the consent form and obtained verbal consent from the respondent. No child or case specific data were obtained during the interviews. The duration of the interview was approximately one hour.
In some agencies, multiple informants were needed to complete each interview. Participants were encouraged to identify alternate informants for specific questions for which they were not the best informant. These additional informants were contacted and consented using the procedures described above. Interviewing proceeded in this manner until all interview questions had been completed or it was determined that the information was not available from any known informant. (More detailed description of the CADVS methods is provided in (Smith et al., 2005)). All study procedures were approved by the Institutional Review Board of San Diego Children's Hospital.
Several steps were undertaken to develop the CADVS interview. First, relevant literature on child welfare policy and practice on co-occurring child maltreatment and intimate partner violence was reviewed, including Effective Intervention in Domestic Violence and Child Maltreatment Cases: Guidelines for Policy and Practice (otherwise known as the Greenbook; (National Council of Juvenile and Family Court Judges, 1999) and the Model Code (National Council of Juvenile and Family Court Judges, 1993). Policy and practice recommendations relevant to the study were identified from this literature, and those recommendations that could be operationalized into interview questions measurable at the local level were retained. Nineteen recommendations were used in the development of the interview and were categorized into four domains: (1) local policies and practices for co-occurrence of child maltreatment and intimate partner violence; (2) assessment and screening for intimate partner violence; (3) training and cross-training; and (4) local coordination. In addition to assessment of formal recommendations, questions were developed on other specific issues of interest such as how agencies addressed the issue of child exposure to intimate partner violence. Interview questions corresponding to the four domains of inquiry were organized into separate interview modules. A panel of national experts reviewed the interview questions, and pilot testing was conducted in counties not participating in the NSCAW study. Modifications to the interview were made based on feedback from the expert panel and pilot interviews.
The current analyses focused on information gathered with the module on assessment and screening, which inquired about policies and practice relating to the assessment of intimate partner violence by CWS.
Information was available from the NSCAW study on the following PSU and child welfare agency characteristics: urbanicity of the PSU, poverty level of the PSU, type of agency administration, and county size. Urbanicity was based on Census definitions with counties with greater than 50% of the population living in an urban area classified as urban. The remaining counties were classified as rural. Seventy-one percent of the counties were classified as urban and 29.2% as rural. Poverty level was defined according to the percentages of families in the PSU living at the poverty level according to Census data. PSUs were considered non-poor if 5% or less of families with children were living below the 50% poverty level, and poor if more than 5% of families with children were living below the 50% poverty level. Forty-four percent were categorized as non-poor and 55.6% as poor. Type of administration referred to whether the child welfare agency was administered by the state or county. Fifty-eight percent were state administered and 41.7% were county administered. County size was defined by the number of children in the county with small counties having less than 5,000 children, medium counties, 5,000 to 24,999 children, and large counties, 25,000 children or more. Thirty-five percent were categorized as small, 34.7% as medium, and 30.6% as large.
Key informants in 75 of the 92 PSUs (81.5%) provided information on CWS policies and practices related to the assessment of intimate partner violence. The 75 PSUs that participated in the study represented 72 different child welfare agencies. (One additional agency participated but was not included in the analyses due to the large amount of missing data, and two others were not included because the key informants reported that their agencies did not assess any families for intimate partner violence and as a result, the vast majority of the interview questions were not administered.) The number of agencies and the number of PSUs represented are not equivalent due to the fact that some individual agencies in large counties accounted for more than one PSU and some PSUs were made up of multiple counties and therefore had more than one child welfare agency. Most of the informants were child welfare managers and supervisors.
As shown in Table 1, 31 (43.1%) agencies reported that all families referred to the child welfare system were assessed for intimate partner violence and 48 (66.7%) indicated that at least 75% of referred families were assessed. The vast majority (98.6%) conducted screening or assessment during investigation of referrals. At least half reported that assessment of intimate partner violence occurred during screening of referrals (50.0%), assessment (following investigation) (77.8%), case opening (56.9%), service plan review (65.3%), and case closure (58.3%), and slightly fewer than half (44.4%) indicated that such assessment occurred at placement. Only 19 (26.4%) agencies indicated that assessment or screening occurred at all stages of a case. Most respondents (75.0%) reported that the majority of assessments for intimate partner violence took place at investigation.
Respondents were asked if specific questions regarding intimate partner violence appeared on agency forms used in different stages of a case. Sixty-three (90.0%) agencies reported that such questions were present on their risk assessment tools and 37 (56.1%) indicated there were questions on investigation forms. Fewer agencies indicated that their intake (33.8%), case service plan (31.0%), and case monitoring (31.7%) forms contained questions on intimate partner violence. Twenty-three (31.9%) respondents reported that their risk assessment protocol included specific guidelines to address potential physical and mental health effects on children who witness intimate partner violence.
Thirty-eight (52.8%) of the 72 agencies indicated they had a written policy pertaining to screening and assessment of intimate partner violence. Among agencies with a written policy, 71.1% indicated that the policy included guidelines pertaining to safety considerations when conducting assessments such as protections for privacy and sharing of confidential information.
Forty (55.6%) agencies had mechanisms in place to monitor whether cases were assessed for intimate partner violence. Seventy-seven percent of these agencies indicated that they conducted some form of monitoring of assessments conducted at investigation, 61.5% did so at service plan review, and 59.0% at case closure. Only 10.5% conducted monitoring of cases at all stages inquired about (i.e., screening, investigation, assessment following investigation, case opening, placement, service plan review, and case closure), and 18.4% conducted monitoring only on assessment conducted during investigation. The types of monitoring activities described by informants included peer, supervisor and administrative case reviews and audits by state agencies. Of the 40 agencies that had monitoring mechanisms, 26 (65.0%) indicated that the information obtained was formally used for purposes such as evaluating staff performance and providing training, and for monitoring agency-level indicators.
Logistic regression analyses were used to examine the relationship between PSU (county) and child welfare agency characteristics and assessment practices. Predictors included in the models were PSU urbanicity (urban, rural), poverty level (non-poor, poor), type of agency administration (county, state), and county size (small, medium, large). The outcome variables examined in the multiple logistic regression models were: whether all families referred to child welfare were reportedly assessed for intimate partner violence (relative to less than all families assessed), whether there was a written policy for assessment of intimate partner violence (relative to no written policy), whether the risk assessment contained guidelines pertaining to children exposed to intimate partner violence (relative to report of no guidelines), and whether the agency had mechanisms to monitor the assessment of intimate partner violence (relative to report of no mechanisms for monitoring). The odds ratios and 95% confidence intervals for each model are presented in Table 2. The Hosmer and Lemeshow statistics suggested that the logistic regression models had a reasonable fit.
None of the PSU/agency characteristics were associated with reports of the existence of a written policy regarding assessment of intimate partner violence or with the existence of mechanisms for monitoring assessment practices. Urbanicity was related to reports of all families being assessed for intimate partner violence with urban counties having lower odds (OR = 0.24, p < 0.05) of assessing all families relative to rural counties. Type of administration was related to the issue of whether the child welfare risk assessment contained guidelines for addressing children exposed to intimate partner violence with county administered agencies having significantly greater odds (OR = 3.66, p < 0.05) of reporting such guidelines compared to state administered agencies.
The goal of this paper was to examine policy and practice relating to the assessment of intimate partner violence among a sample of child welfare agencies in counties throughout the United States. Guidelines issued in recent years by national policy organizations (National Association of Public Child Welfare Administrators, 2001; National Council of Juvenile and Family Court Judges, 1999) have suggested that screening or assessment for intimate partner violence should occur during all stages of a child protection case from intake to case closure. Almost all of the agencies in the current study conducted some assessment of intimate partner violence but there was variability in the extent to which it occurred. Only 43% indicated that all families referred to the child welfare system were assessed for this problem. Most of the agencies reported that staff conducted assessment during investigation of referrals, but only about one-quarter indicated that it occurred at all stages of a case. Routine, on-going assessment across the stages of a case is important as violence could begin at any point in a family's involvement with the child welfare system (Carter & Schechter, 1997) and the disclosure of intimate partner violence can be influenced by multiple factors such as stigma, embarrassment, and concerns about the repercussions of reporting violence including fear of reprisal from an abusive partner and apprehension about the response by child welfare and the criminal justice system (DeVoe & Smith, 2003; Rodriguez, Sheldon, Bauer, & Perez-Stable, 2001). An emphasis on assessment during the earlier stages of contact with the child welfare system may contribute to under recognition of the problem in families who are initially reticent to disclose a problem of this nature or who only begin to experience such violence later in their involvement with the system. It is also important that other caregivers who may become involved at different stages of a case, such as guardians, kin, foster parents, and potential adoptive parents, are assessed for intimate partner violence. Screening should include criminal and civil records checks and protocols should outline the procedures to follow if intimate partner violence is identified (National Association of Public Child Welfare Administrators, 2001; National Council of Juvenile and Family Court Judges, 1999).
Besides recommending that assessment for intimate partner violence occur throughout the stages of a case, policy groups have suggested that screening and assessment tools be used as part of standard practice in child welfare settings (National Council of Juvenile and Family Court Judges, 1999). In the current study, 90% of child welfare agencies reported that questions on intimate partner violence were present on risk assessment forms and 56% indicated there were questions on their investigation forms. It was less common to have such questions on other forms such as intake and case service plans. Information was unfortunately not obtained on the nature of the questions contained on these forms, but it is likely that standardized screening and assessment tools that include appropriate questions to gather information from family members are not being widely used (Friend, 2000). More often, the questions on these forms are prompts for the workers to record information about the violence. Experience from the health care sector suggests that intimate partner violence is often undetected without the implementation of standardized screening procedures that providers can readily apply in their contacts with patients (Augustyn & Groves, 2005; Campbell et al., 2001; Connelly, Newton, Landsverk, & Aarons, 2000; Feldhaus et al., 1997; McNutt, Waltermaurer, McCauley, Campbell, & Ford, 2005; Thompson & Krugman, 2001; Waalen, Goodwin, Spitz, Petersen, & Saltzman, 2000). There is some evidence that such procedures can be effectively implemented in child welfare agencies. In a pilot study conducted in New York City, the use of a standardized questionnaire completed by child welfare caseworkers during investigations resulted in a substantial increase in the identification of intimate partner violence (Magen et al., 2001). Comparable results have also been achieved with the use of a similar intake questionnaire in community-based child welfare preventive services agencies (Magen, Conroy, & Del Tufo, 2000).
Various mechanisms, such as written policy, staff training, and case monitoring protocols, are important for supporting the establishment and maintenance of optimal assessment practices (National Association of Public Child Welfare Administrators, 2001; National Council of Juvenile and Family Court Judges, 1999; Shepard & Raschick, 1999) yet the current findings suggest that such procedures have not been universally adopted or have limited scope. Only 52.8% of the child welfare agencies participating in this study reported that they had written policies pertaining to the assessment of intimate partner violence, and 55.6% indicated that some form of monitoring of assessments was in place. Previous findings from the CADVS study indicated that while most participating child welfare agencies covered a range of topics concerning assessment and intervention for intimate partner violence in their staff training, relatively few (less than 20%) reported mandatory training for all staff (Nuszkowski et al., in press).
This study has several limitations. First, the psychometric properties of the interview questions are unknown. The measures were developed with consultation from an expert panel and pilot tested, but formal psychometric testing was not conducted. Thus, the reliability of the information and relationship to actual practice within the agencies is not known. Second, the range of topics that were surveyed limited the depth of information that could be gathered. It would have been useful to obtain more comprehensive information on assessment practices such as the number and types of questions included on agency forms (e.g., risk assessments) and whether any standardized screening tools were being used. Information was also not obtained on issues such as cultural and other diversity considerations related to assessment practices and policy. Finally, there was little relationship between assessment practices and the county and agency variables available for examination. Future research might investigate other agency- and community-related (e.g., availability of resources) and staff-related variables (e.g., experience, values) that can potentially influence policy and practice (e.g., Wells, Lyons, Doueck, Brown, & Thomas, 2004).
In summary, this study provides valuable information on policy and practice related to the assessment of intimate partner violence in a large number of child welfare agencies. The majority of agencies reported a range of assessment-related efforts, but clearly not all families coming to the attention of child welfare were being assessed. Additional work is needed to develop and evaluate assessment methods and tools to identify affected families and to respond in ways that protect safety and well-being throughout all phases of involvement with the child welfare system. Further research on factors that influence practice is critical to the development of strategies to promote more widespread adoption and to support on-going utilization of appropriate assessment and intervention procedures for intimate partner violence.
The work described in this paper was supported, in part, by National Institute of Justice grant 2002-WG-BX-0014 (K.J.K.), National Institute of Mental Health Mentored Research Scientist Development Award K01-MH65454 (A.L.H.), and National Institute on Drug Abuse Mentored Research Scientist Development Award K01-DA15145 (C.D.C). We gratefully acknowledge the data collection efforts of staff at the Child and Adolescent Services Research Center and the assistance of Ms. Mary Mueggenborg with manuscript preparation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.