Intimate Partner Violence Risk Assessment Tools: A Review

Appendix A

Examples of Intimate Partner Violence Risk Assessment Tools

Structured Clinical Judgment Tools

The Spousal Assault Risk Assessment Guide (SARA)

The Spousal Assault Risk Assessment Guide (SARA) was developed in Canada at the British Columbia Institute on Family Violence and is used in 15 countries around the world (Braff and Sneddon 2007; Helmus and Bourgon 2011). The SARA is designed to predict both intimate partner violence and lethality (Guo and Harstall 2008). The SARA consists of 20 items which focus on criminal history, psychosocial adjustment, spousal assault history and information on the alleged offence. Information for this tool is collected from a number of sources, including from the accused, the victim, standardized measures of psychological and emotional abuse, and other records such as police reports (Kropp 2008). The evaluator considers the items and determines whether the accused is at low, medium, or high risk of causing imminent harm to their intimate partner or to another individual (Helmus and Bourgon 2011). The SARA is also used for other purposes, including for determining treatment plans for offenders, intervention strategies and levels of supervision (Braff and Sneddon 2007). Because the SARA includes clinical measures, it is recommended that mental health professionals conduct the assessments; however, other professionals can use the tools if they have access to the accused’s psychological or psychiatric reports (Kropp and Hart 1997).

The SARA has a number of strengths. First, it has been shown to have good validity and inter-rater reliability (Helmus and Bourgon 2011; Kropp 2008). Also, the validity of the SARA has been tested more than any other tool (Helmus and Bourgon 2011). These validation studies have been conducted by individuals that are independent from the developers and with different types of adult male offenders (e.g., offenders in prison and on community supervision orders). Furthermore, its ease of use renders it accessible by a number of different kinds of professionals and because the SARA was developed based on diverse empirical research, it can be used with different kinds of offenders. Finally, it “includes a variety of static and dynamic factors related to both general violence and intimate partner violence, which makes it flexible, sensitive to offender change, and useful for treatment planning” (Helmus and Bourgon 2011, 73).

There are, however, limitations to the SARA. First, it has been argued that some of the research on which the scale is based may be dated, and as such, some of the items may need revision (Helmus and Bourgon 2011). It has also been noted that it is not clear which of the items are meant to be used to predict recidivism and which are to be used for treatment planning. Also, because the SARA manual recommends that information be gathered from a number of sources, it may not be practical for use in the field, as there is often limited time and resources. Furthermore, there are no restrictions on what kinds of offenders the scale can be used with despite the fact validation studies have been limited in terms of the age and gender of the offenders on which the SARA is tested (i.e., the majority of the studies have been conducted with Caucasian male offenders, aged 30 to 40 years old). Also, it has been argued that the coding criteria does not allow for differentiation between offenders based on their levels of risk. Finally, although the SARA is meant to be used as a prediction tool for re-assault and lethality, only the tool’s capacity to predict re-assault has been validated (Guo and Harstall 2008).

The SARA was recently condensed into a shorter risk assessment tool, the Brief Spousal Assault Form for the Assessment of Risk (B-SAFER). The B-SAFER was developed in Canada and is used in some jurisdictions within Canada (Millar 2009). The B-SAFER is a 10-item risk management tool that includes items that focus on the accused’s history of intimate partner violence, as well as difficulties with psychological and social adjustment. It was developed for use with professionals within the criminal justice system, including law enforcement (Kropp and Hart 2005). The tool’s developers indicate that users of the tool should have expertise in conducting assessments and in the area of intimate partner violence toward women. The B-SAFER includes fewer items than the SARA, does not require as many resources and uses less technical jargon.

The Domestic Violence Screening Inventory (DVSI)

The Domestic Violence Screening Inventory (DVSI) was developed in the United States by the Colorado Judicial Department and is used in the United States (Williams and Grant 2006; Wong and Hisashima 2008). The DVSI is a 12-item risk assessment tool that is designed to assess the likelihood of the occurrence of intimate partner violence (Guo and Harstall 2008; Hilton and Harris 2005; Kropp 2008; Wong and Hisashima 2008). It includes both static and dynamic factors, such as items pertaining to domestic violence, criminal history, current employment, relationship status, treatment history, and information on the current offence. The DVSI provides the evaluator with a risk score that is used to determine the offender’s risk level relative to other offenders.  It is used in a number of capacities in the criminal justice system, including at intake, at probation, and for case supervision. Its concurrent and predictive validity are satisfactory, but it has not been validated independently (Kropp 2008).

The Danger Assessment (DA)

The Danger Assessment (DA) was developed by Jacquelyn Campbell in the United States and is used throughout the United States and Canada (Guo and Harstall 2008; Millar 2009). The DA is a structured clinical assessment tool that was originally designed for use by emergency room nurses to assess the likelihood of intimate partner homicide (Guo and Harstall 2008; Hanson et al. 2007). It is now also used to predict domestic violence recidivism, but not in low-risk or medium violence cases (Guo and Harstall 2008). It is used in a number of settings, including for the purposes of victim education and awareness, safety planning and determining the conditions of services. The DA is comprised of two parts. The first is a calendar on which the victim indicates the severity and frequency of instances of domestic violence that she experienced within the last 12 months. The second part is a 20-item checklist of risk factors that are related to intimate partner homicide (Millar 2009). Both sections are completed in collaboration with the victim (Hanson et al. 2007). Guo and Harstall (2008) state that the most appropriate users of the DA are victim advocates, social workers or clinicians in various settings, such as women’s shelters and hospitals.

The strengths of the DA are that it has strong test-re-test reliability, good inter-rater reliability and construct validity, and correlates strongly with other measures of domestic violence recidivism (Kropp 2008). In addition, it is a good tool to use with victims as it allows victims to better understand the risk that the relationship may pose to them and what risk management options are available (Guo and Harstall 2008). It may also serve as a useful instrument when information is difficult to obtain or when the offender cannot be interviewed. The accuracy of the DA, however, is not as strong as other tools and it does not provide the evaluator with a means of assessing the risk level posed by the accused (Hilton and Harris 2005).

Actuarial Tools

The Ontario Domestic Assault Risk Assessment (ODARA)

One of the most common actuarial tools used in cases of intimate partner violence is the Ontario Domestic Assault Risk Assessment (ODARA). The ODARA was developed by the Mental Health Centre in Penetanguishene, Ontario and the Ontario Provincial Police (Millar 2009). It is used in a number of provinces across Canada and is used in at least one state in the United States (Maine) (Kercher et al. 2010; Millar 2009). The ODARA is a 13-item actuarial tool that includes the domains of criminal history, number of children, substance abuse and the barriers that victims face in terms of accessing support (Guo and Harstall 2008; Hanson et al. 2007). The ODARA is used to predict future violence against a spouse, as well as the frequency and severity of the violence (Millar 2009). This tool is used by police officers, victim services, domestic violence case workers, and probation and correctional services in many provinces across Canada. Guo and Harstall (2008) state that the ODARA may be most appropriately used by criminal justice personnel as they are able to access the needed information and it is relatively shorter than other tools.

The ODARA has been shown to predict domestic violence recidivism at a comparable level of accuracy as the VRAG (discussed below), but in a more efficient and timely manner, and correlates with the Danger Assessment and the SARA (Hilton and Harris 2005; Kropp 2008). It is, however only used with male offenders and because there were no cases of homicide in the sample used in the tool’s development, the tool may not be appropriate for use in predicting lethality (Kropp 2008; Hart 2010). Others have also argued that the content and scoring mechanisms are questionable and that it is overly rigid (Hart 2010). Furthermore, many of the items require gathering information from criminal justice databases, which may be more easily accessible for some professionals (e.g., law enforcement), but not for others (e.g., victim advocates) (Guo and Harstall 2008). Finally, because the ODARA was designed to predict recidivism and not for the purposes of risk management, those using this tool will likely need to seek assistance when making decisions involving overall case management. 

The Domestic Violence Risk Appraisal Guide (DVRAG)

The Domestic Violence Risk Appraisal Guide (DVRAG) was developed in Canada by Hilton et al. (2008) (Guo and Harstall 2008). It is a 14-item risk assessment tool designed to assess the risk of intimate partner violence recidivism among male offenders with a criminal record for intimate partner violence (Rice et al. 2010). It is based on the same items as the ODARA, with the addition of the offender’s score on the Psychopathy Checklist – Revised (PCL-R). The offender’s score is converted into a percentile score, which is then compared against similar offenders. It is recommended that the DVRAG only be used when the assessor has access to detailed clinical or correctional data of the offender. It is also recommended that the assessor have access to the psychosocial history of the offender, has been trained to use the PCL-R or have access to the offender’s PCL-R score, and that the assessor can conduct the DVRAG with reasonable inter-rater reliability. One limitation of the DVRAG is that its predictive accuracy has not yet been evaluated independently.

Risk Assessment Tools for General and Violent Offending

The Violence Risk Appraisal Guide (VRAG)

The Violence Risk Appraisal Guide (VRAG) was developed in Canada by Quinsey et al. (2006) (Guo and Harastall 2008). It is 12-item risk assessment tool that was designed for predicting violent recidivism and is used in a number of capacities including with patients in forensic and non-forensic settings, sex offenders, and offenders in prison (Hilton and Harris 2005). It is used by clinicians, courts, and parole officials (Guo and Harstall 2008). It includes items relating to demographics and childhood history, and includes a psychiatric assessment. It has been shown to accurately predict intimate partner violence recidivism (Hanson et al. 2007). One limitation of the VRAG is that it requires a great deal of time, access to offender history and ability to conduct clinical assessments (Hilton and Harris 2005).

Level of Service Inventory – Revised (LSI-R)

The Level of Service Inventory – Revised (LSI-R) was developed in Canada by Andrews and Bonta (1995) and is used in various jurisdictions in Canada and elsewhere (Guo and Harstall 2008; Millar 2009). The LSI-R is a 54-item actuarial risk assessment tool that was designed for use with offenders over the age of 16 (Andrews and Bonta 2004). The LSI-R is used to predict parole outcome, as well as recidivism. It is used by probation and parole officers, as well as correctional officers working in jails and detention facilities. The LSI-R includes scales related to criminal history, alcohol/drug problems, education/employment and financial issues, among several other scales. It has been shown to accurately predict general and violent recidivism among male offenders of intimate partner violence (e.g., Hanson and Wallace-Capretta 2000). Research evaluating the ability of the LSI-R to predict intimate partner violence recidivism specifically has shown mixed results; some studies have found small predictive effects for intimate partner violence by the LSI-R (e.g., Hendricks et al. 2006), while other studies have found no association between LSI-R scores and intimate partner violence recidivism (e.g., Hilton et al. 2010). One limitation of the LSI-R is that the tool may contain more items than is necessary to determine risk of recidivism and that the use of the LSI-R is only necessary if the offender is the focus of service planning (Andrews et al. 2010).