Spousal Assaulters in Outpatient Mental Health Care: The Relevance of Structured Risk Assessment

Abstract

This study examined whether a typology of perpetrators of intimate partner violence (IPV) could be replicated in a Dutch sample (N = 154) of self-referred IPV perpetrators using a structured risk assessment tool for relational violence (Brief Spousal Assault Form for the Evaluation of Risk [B-SAFER]). Our findings support the previous IPV perpetrator subtypes: low-level antisocial (LLA), family only (FO), psychopathology (PP), and generally violent/antisocial (GVA). The subtypes differed on the descriptive dimensions general criminality, substance use, and mental health problems. The prevalence rates for each subtype were roughly comparable with those in previous studies. Contrary to expectation, the prevalence of the GVA subtype was relatively high in our self-referred sample compared with court-referred samples. Our findings suggest that structured risk assessment should be an integral part of the intake procedure for IPV perpetrators entering treatment, to assess their level of risk and to arrive at a tailored risk management strategy, regardless of setting or referral source.

Keywords

domestic violence IPV batterer typologies B-SAFER risk assessment

Annually, 200,000 people in the Netherlands become victim of domestic violence (Van der Veen & Bogaerts, 2010a). Domestic violence is defined as “a physical, mental or sexual violation of the personal integrity of the victim by a person from the victim’s family circle. This includes (ex)-partners, family members and family friends of the victim” (Van Dijk, Flight, Oppenhuis, & Duesmann, 1997, p. 8). Van Dijk and colleagues (1997) reported that 45% of a Dutch community sample had been a victim of domestic violence at some time in their life. Furthermore, 27% had ever been a victim of domestic violence, which occurred weekly or daily. A more recent study reported similar results; approximately, 40% of a Dutch community sample had ever been a victim of domestic violence and/or incident (Van der Veen & Bogaerts, 2010a). Two other national prevalence studies in the Netherlands (Lünnemann & Bruinsma, 2005; Van Dijk, Veen, & Cox, 2010) reported that a little less than half (40% and 46%) of all domestic violence offenses were committed by the ex-partner and more than one third (37% and 34%) by the current partner.

Since the mid-1990s, several typologies (of perpetrators) of intimate partner violence (IPV) have been developed. There are typologies that focus more on the dominant type of violence used by a perpetrator (Emery, 2011; M. Johnson, 2008; Stark, 2007) and typologies that focus on empirically derived perpetrator types (Dixon & Browne, 2003; Holtzworth-Munroe & Stuart, 1994; R. Johnson et al., 2006; Van der Veen & Bogaerts, 2010b). As an example of a typology that focuses more on type of violence used, Stark (2007) detailed different coercive strategies that men use to control women. Likewise, M. Johnson (2008) proposed a distinction based on coercive control in intimate relationships, in which “intimate terrorism,” including violence deployed in the service of general control over one’s partner, is distinguished from “situational couple violence,” which does not include the systematic, controlling abuse associated with battering. Emery (2011) proposed a more elaborate typology based on sociological building blocks: order, power, and the relationship between the violent act and relationship norms. A different, empirically derived, IPV perpetrator typology for which there exists consistent empirical support is the typology developed by Holtzworth-Munroe and Stuart (1994). They examined previous typologies of male batterers to determine the subtypes that consistently appeared across the IPV literature. In addition, they identified three underlying descriptive dimensions that distinguished IPV perpetrators on a consistent basis. These dimensions are as follows: severity and frequency of spousal physical violence, generality of violence, and perpetrators’ psychopathology (PP) and personality disorder. Three major subtypes were uncovered: family only, dysphoric/borderline (DB), and generally violent/antisocial (GVA; see Table 1). Family only (FO) perpetrators engage in the least severe IPV. Their violence is generally restricted to family members; these perpetrators do not engage in more general violence outside the home. There is little to no PP or personality disorder found in these cases (Holtzworth-Munroe & Stuart, 1994; White & Gondolf, 2000). DB perpetrators engage in moderate to severe IPV, including psychological and sexual abuse. This subtype is characterized by psychological distress, borderline, and/or schizoid personality traits. Borderline personality characteristics are defined by emotional instability, intense and unstable relationships, and fear of rejection and jealousy. In addition, DB perpetrators may experience problems with substance use (Holtzworth-Munroe & Meehan, 2004; Holtzworth-Munroe & Stuart, 1994; White & Gondolf, 2000). The third subtype consists of GVA cases: They engage in moderate to severe IPV, including psychological and sexual abuse. Different from the other subtypes, these assaulters also engage in extra-familial aggression and are often characterized by an extensive history of criminal behavior and/or involvement with the criminal justice system. They are most likely to be diagnosed with antisocial personality disorder or psychopathy. Moreover, individuals of this type may experience problems with substance use (Holtzworth-Munroe & Stuart, 1994; White & Gondolf, 2000). In a subsequent study by Holtzworth-Munroe, Meehan, Herron, Rehman, and Stuart (2000), a fourth subtype emerged: the low-level antisocial subtype (LLA). Moderate scores on measures of seriousness of IPV, general violence, and antisociality characterize this subtype (see Table 1). On many assessment measures (e.g., drug use, violence, PP, justice involvement), the LLA type’s scores fell intermediate between the FO and GVA subtype.

Table 1.

IPV Perpetrator Typology.

Descriptive Dimension	Family Only (FO)	Dysphoric/Borderline (DB)	Generally Violent/Antisocial (GVA)	Low-Level Antisocial (LLA)
Severity of IPV	Low	Moderate-high	Moderate-high	Moderate
Psychological and sexual abuse	Low	Moderate-high	Moderate-high	Moderate
Generality of violence
Extrafamilial violence	Low	Low-moderate	High	Moderate
Criminal behavior	Low	Low-moderate	High	Moderate
Psychopathology	Low-moderate	High	Low-moderate	Moderate
Personality disorder	None or passive/dependent	Borderline or schizoid	Antisocial/psychopathy	Moderate antisociality
Alcohol/drug abuse	Low-moderate	Moderate	High	Moderate

Source. Holtzworth-Munroe and Stuart (1994); Holtzworth-Munroe, Meehan, Herron, Rehman, and Stuart (2000).

Note. IPV = intimate partner violence.

The typology proposed by Holtzworth-Munroe and Stuart (1994) has repeatedly received empirical support in court-referred samples (Cunha & Gonçalves, 2013; Dixon & Browne, 2003; Huss & Ralston, 2008; R. Johnson et al., 2006; Thijssen & de Ruiter, 2011) as well as in community samples (Holtzworth-Munroe & Meehan, 2004; Holtzworth-Munroe et al., 2000; Langhinrichsen-Rohling, Huss, & Ramsey, 2000). Furthermore, a follow-up study demonstrated reasonable long-term (i.e., 1.5 to 3 years) stability of the typology (Holtzworth-Munroe, Meehan, Herron, Rehman, & Stuart, 2003). The FO subtype, in particular, proved to be the most stable subtype in contrast to the less stable DB subtype (Holtzworth-Munroe et al., 2003). However, the LLA subtype, as proposed by Holtzworth-Munroe and colleagues (2000), was not always found (e.g., the LLA subtype was not found in the community sample of Waltz, Babcock, Jacobson, & Gottman, 2000).

In a recent study, Thijssen and de Ruiter (2011) identified the four subtypes of IPV perpetrators in a court-referred sample from the Dutch probation service. They used the three descriptive dimensions found by Holtzworth-Munroe and Stuart (1994): severity of IPV, generality of violence, and PP/personality disorder. These dimensions were scored using four risk factors of the Brief Spousal Assault Form for the Evaluation of Risk (B-SAFER; Kropp, Hart, & Belfrage, 2005): violent acts, general criminality, substance use problems, and mental health problems. Findings of Thijssen and de Ruiter (2011) showed that the majority (37%) of their sample awaiting sentencing belonged to the FO subtype. However, a significant proportion (24%) of the sample belonged to the LLA subtype and another 18% belonged to the GVA subtype. R. Johnson and colleagues (2006) showed that approximately half (47%) of their court-referred sample consisted of the antisocial group (similar to the GVA subtype). In accordance, Dixon and Browne (2003) found significantly fewer FO batterers and more antisocial men in a court-referred sample, compared with a self-referred sample.

In the current study, we will examine whether the IPV perpetrator typology, as found by Holtzworth-Munroe and Stuart (1994) and Thijssen and de Ruiter (2011), is supported in a sample of IPV perpetrators self-referred to a regional mental health care center in the southwest of the Netherlands. Furthermore, we will compare the prevalence rates of the different subtypes in our sample with those in the Dutch court-ordered sample of Thijssen and de Ruiter (2011). Research has shown that the different subtypes have different risk profiles (Cavanaugh & Gelles, 2005; De Ruiter, 2011). Knowing which subtypes occur in different samples can facilitate risk management. Several authors suggested that interventions targeted to address the individual criminogenic needs of specific subtypes may increase the effectiveness of risk management strategies (Andrews & Bonta, 2010; Cavanaugh & Gelles, 2005; Murphy & Eckhardt, 2005). In this study, the B-SAFER will be used to identify the subtypes, using the methodology of Thijssen and de Ruiter (2011). Based on the previously discussed research (Dixon & Browne, 2003; Holtzworth-Munroe & Stuart, 1994; R. Johnson et al., 2006; Thijssen & de Ruiter, 2011), we expect to replicate the four subtypes. Because the FO subtype is often found to be more prevalent in self-referred samples, we expect that the FO subgroup will be larger in our self-referred outpatient sample than in the court-ordered sample of Thijssen and de Ruiter (2011). Reversely, it is expected that the GVA subgroup will be smaller in our sample, in comparison with the court-ordered sample.

Method

Sample

The sample consisted of 163 IPV perpetrators referred to GGZ West North Brabant (WNB), a mental health care center in the southwest of the Netherlands. The majority of men were self-referred and enrolled voluntarily; however, a small group (n = 9) was referred by the Dutch probation service or other legal authority. These 9 cases were excluded for the present study, resulting in a sample of 154 self-referred IPV cases. At GGZ WNB, all men were admitted to an outpatient 12-week IPV and anger management group program, called “Niet meer door het lint” (“Not losing it anymore”; van Dam, Van Tilburg, Steenkist, & Buisman, 2009). This program is a manualized cognitive-behavioral group therapy. Perpetrators are confronted with the consequences of their aggressive behaviors and taught alternative responses and behaviors. Components such as (social) skills training and anger management are combined. One treatment round, covering a period of 3 months, consists of 12 weekly group sessions. The first three sessions focus on psycho-education about aggression, motivation for treatment, and setting treatment goals. The next sessions focus on aggression scenarios, triggers, and arousal awareness of the participants. Subsequent sessions teach techniques to prevent and cope with aggression. These sessions include relaxation training, coping strategies, assertiveness and communication training, and emotion regulation practice (van Dam et al., 2009).

Files were retrieved for all individuals who had entered the program from November 2007 until January 2012. Inclusion criteria were male, able to read and write Dutch, and having committed any type of IPV. The age of the sample ranged from 17 to 76 years, with a mean age of 37.10 years (SD = 10.06). The majority of the sample (92.20%) was Dutch, 3.90% was Turkish, 1.90% Surinamese, and 1.90% had another ethnic background. The majority (83.30%) had a relationship with an intimate partner at the time of intake; the remainder was single. Almost half (48.70%) of the sample had been in contact with police or the criminal justice system at some point in their life.

Instrument

B-SAFER (Kropp et al., 2005) is a structured professional judgment instrument for assessing the risk of spousal assault. In the present study, the Dutch version of the B-SAFER (De Ruiter, 2009) was used to identify subtypes. The B-SAFER was scored by two of the authors based on the patient’s records. The B-SAFER contains 10 risk factors that are divided into two sections. The first section includes 5 risk factors related to the perpetrator’s history of IPV, such as violent acts, threats, and court orders. The second section includes 5 risk factors related to the perpetrator’s history of psychosocial functioning, such as general criminality, intimate relationship problems, and mental health problems. The presence of these risk factors is coded using a 3-point scale, reflecting the degree to which the risk factor is present (“Y” = definitely present = 2, “?” = possibly or partially present = 1, “N” = absent = 0). When there is not enough information available on a certain risk factor or when the available information is considered invalid, the risk factor is left un-coded (i.e., omitted; Kropp et al., 2005). After scoring the individual items, a final risk judgment is made about the prospective risk in the short term (within 2 months) and in the long term (2 months and beyond). This final risk rating is coded as low (L = 0), moderate (M = 1), or high (H = 2; Kropp et al., 2005).

There is evidence for good construct and concurrent validity for the B-SAFER (Au et al., 2008). Predictive validity was found to be moderate (Au et al., 2008). However, in a more recent study, Belfrage and Strand (2012) found a poor predictive power of police risk assessments: The higher the police-assessed risk, the lower the recidivism rate. Nevertheless, this could be explained by the fact that the police only interfered when the risk was assessed as high.

The interrater reliability of the B-SAFER proved to be good (De Ruiter, de Jong, Reus, & Thijssen, 2008). In the present study, the interrater reliability of the B-SAFER was examined using the intraclass correlation coefficient (ICC). A two-way random effects model in combination with the absolute agreement type was used (McGraw & Wong, 1996; Shrout & Fleiss, 1979). For single measure ICCs, the following critical values were selected: ICC ≥ .75 = excellent, .60 ≤ ICC < .75 = good, .40 ≤ ICC < .60 = moderate, ICC < .40 = poor (Fleiss, 1986). The ICCs were calculated for 37 cases independently scored by the two raters. The single measure ICC ranged from .24 to .86 with a mean of .63, for all B-SAFER items. The B-SAFER items used in the present study (violent acts, general criminality, substance use problems, and mental health problems) demonstrated good to excellent interrater reliability (all ICCs ≥ .72; see Table 2).

Table 2.

ICCs, Single Measure (n = 37).

B-SAFER Item	ICC
Violent acts	.79
General criminality	.72
Substance use problems	.75
Mental health problems	.75

Note. ICC = intraclass correlation coefficient; B-SAFER = Brief Spousal Assault Form for the Evaluation of Risk.

Procedure

The present study is a retrospective file study. Demographic data were obtained from intake reports. Demographic data consisted of age, gender, education level, work, relationship status, and a self-reported history of criminal justice contact. All risk factors of the B-SAFER were retrospectively coded for each individual patient record. Files contained at least an intake report and treatment progress reports for every session attended. Sometimes there was an official criminal record and/or previous mental health reports. Two raters, who had received a 1-day training in coding the B-SAFER, independently coded 10 practice cases and discussed them in a consensus meeting. The two raters independently coded 37 files, for which interrater reliability was calculated. All risk factors were coded separately for the year prior to intake and for the past before this year.

In this study, the B-SAFER was used as a classification tool. Holtzworth-Munroe and Stuart (1994) identified three underlying descriptive dimensions to divide spousal assaulters into the previously discussed field-driven typology. The B-SAFER can be used to measure these descriptive dimensions (Thijssen & de Ruiter, 2011). The descriptive dimension “severity of IPV” was measured by the first risk factor, “violent acts.” Violent acts consist of actual as well as attempted physical harm. They also include actual or attempted sexual violence and actual or attempted use of weapons (Kropp et al., 2005). The second descriptive dimension “generality of violence” was measured by Item 6 “general criminality.” General criminality demonstrates itself through the engagement in persistent, frequent, or diverse antisocial behavior, of which general violence may be a part (Kropp et al., 2005). Item 10 of the B-SAFER “mental health problems” can be used to measure the third descriptive dimension “PP/personality disorders.” Mental health problems include disturbances of thought and perceptions (i.e., delusions and hallucinations), intellectual or cognitive deficits, emotional problems (i.e., depression, mania, and extreme anger or anxiety), and disorganized behaviors (i.e., impulsivity, suicidality; Kropp et al., 2005). In addition, Item 9 “substance use problems” was used to assess the PP dimension. Substance use problems consist of illegal drug use as well as the abuse of legal drugs, such as alcohol and/or prescribed medications (Kropp et al., 2005).

For all analyses, items coded as omitted (because of missing information) were replaced by a 0 (absent). By replacing the omitted items by 0, we obtain a conservative risk estimate. In total, 16 omitted items were replaced by 0, which is approximately 1% of all coded items.

Analyses

K-means cluster analysis was used to examine whether the four subtypes as proposed by Holtzworth-Munroe and colleagues (2000) could be detected. This analysis produces a designated number of clusters. We followed similar procedures to Holtzworth-Munroe et al. (2000) and Thijssen and de Ruiter (2011). We used cluster centers produced by the initial cluster in a k-means cluster analysis to confirm the original cluster membership, setting it for two, three, and four clusters, respectively. Variables selected for the cluster analysis were the four risk factors from the B-SAFER, as previously mentioned. The selection of variables is very important in cluster analysis, as they define the establishment of the resulting clusters.

Subsequent to the k-means cluster analysis, a series of one-way ANOVAs was carried out to examine differences between the clusters. In addition, post hoc comparisons were carried out for the four relevant B-SAFER items. Prevalence rates for the subtypes were determined for our self-referred sample. These prevalence rates were compared with the prevalence rates of the sample of Thijssen and de Ruiter (2011) by means of a chi-square test.

Results

The four risk factors from the B-SAFER selected for the k-means cluster analysis were effective in distinguishing the four subtypes by Thijssen and de Ruiter (2011). Several k-means cluster analyses were performed, setting k at 2, 3, or 4 clusters. Consistent with our expectation, a four-cluster solution was found to best fit the present data. The four clusters we found were labeled as follows: LLA (n = 41, 26%), FO (n = 43, 28%), PP (n = 35, 23%), and GVA (n = 35, 23%).

The four subtypes were compared by means of a series of one-way ANOVAs, carried out for the four B-SAFER items. The subtypes did not differ on the item “violent acts” (for the preceding year and past), F(3, 150) = 0.83, ns, and F(3, 150) = 1.62, ns, respectively. A significant difference across the clusters was found for “general criminality” for the preceding year, F(3, 150) = 14.22, p < .001, and for the past, F(3, 150) = 40.52, p < .001. Furthermore, a significant difference was found for the item “substance use problems” for the preceding year, F(3, 150) = 102.74, p < .001, as well as for the past, F(3, 150) = 121.3, p < .001. For the item “mental health problems,” a significant difference was found between the clusters for the preceding year, F(3, 150) = 26.37, p < .001, and for the past, F(3, 150) = 67.13, p < .001 (see Table 3).

Table 3.

Mean Scores, Standard Deviations, and Significant Differences Across the Four Clusters per B-SAFER Item.

	Low-Level Antisocial		Family Only		Psychopathology		Generally Violent/Antisocial
	(n = 41)		(n = 43)		(n = 35)		(n = 35)
B-SAFER Item	M	SD	M	SD	M	SD	M	SD	F	p
1. Violent acts
Preceding year	1.66	0.58	1.53	0.55	1.57	0.50	1.71	0.55	0.84	.48
Past	1.56	0.59	1.28	0.66	1.43	0.61	1.54	0.74	1.62	.19
6. General criminality
Preceding year	0.20^a	0.56	0.30^a	0.67	0.20^a	0.53	1.09^b	0.92	14.22	<.001
Past	0.51^a	0.84	0.49^a	0.80	0.34^a	0.73	1.94^b	0.24	40.52	<.001
9. Substance use problems
Preceding year	1.85^a	0.36	0.19^b	0.45	0.14^b	0.36	1.31^c	0.83	102.74	<.001
Past	1.95^a	0.22	0.42^b	0.76	0.11^b	0.32	1.80^a	0.58	121.30	<.001
10. Mental health problems
Preceding year	1.12^a	0.87	0.77^a	0.68	1.77^b	0.43	1,86^b	0.36	26.37	<.001
Past	1.00^a	0.92	0.07^b	0.26	1.63^c	0.49	1.77^c	0.49	67.13	<.001

Note. Superscripted letters (a, b, c) indicate the clusters that significantly differed (p < .05). B-SAFER = Brief Spousal Assault Form for the Evaluation of Risk.

Cluster 1: LLA (n = 41, 26%)

Cluster 1 (n = 41) was labeled the low-level antisocial subtype, because these individuals had moderate to high scores on “violent acts.” Individuals in this cluster had low scores on “general criminality” and differed significantly from the GVA subtype (p < .001). However, they had a significantly higher score on “substance use problems” than Clusters 2 (FO) and 3 (PP), for both the preceding year, as well as for the past (p < .001). LLA perpetrators did not differ from Cluster 4 (GVA) on “substance use problems” in the past, yet they scored significantly higher (p < .05) in the preceding year. On the item “mental health problems,” this subtype fell between Cluster 2 (FO) and Cluster 4 (GVA), for both the preceding year and the past. This cluster resembles most closely the LLA subtype found by Thijssen and de Ruiter (2011).

Cluster 2: Family Only (FO; n = 43, 28%)

Cluster 2 (n= 43) was labeled the family only subtype. Individuals in this cluster scored the lowest on most items. On the item “violent acts,” this cluster scored the lowest although the difference was not significant. Together with Clusters 1 and 3, this cluster scored significantly lower than Cluster 4 (GVA; p < .001) on “general criminality” (in the preceding year as well as in the past). On “substance use problems,” this cluster scored significantly lower than Clusters 1 (LLA; p < .001) and 4 (GVA; p < .001), but did not differ from Cluster 3 (PP; p = .072), for both the preceding year and the past. Furthermore, together with Cluster 1 (LLA), this cluster scored lowest on “mental health problems” for the preceding year (p < .001) and scored significantly lower than all three clusters for the past (p < .001). This cluster resembles the FO of Thijssen and de Ruiter (2011).

Cluster 3: PP (n = 35, 23%)

Cluster 3 (n = 35) was labeled the psychopathology subtype. Similar to the PP subtype found by Thijssen and de Ruiter (2011), this cluster could not be labeled dysphoric/borderline, because the item “mental health problems” covers all types of mental health issues. This cluster received low to moderate scores on most items but a high score on “mental health problems.” “Mental health problems” were significantly higher than Clusters 1 (LLA) and 2 (FO), for both the preceding year and the past (p < .001). Individuals in this cluster scored moderate on “violent acts.” On “general criminality,” this cluster scored low and significantly lower than Cluster 4 (GVA; p < .001), for both the preceding year and the past. Furthermore, this cluster scored lowest on “substance use problems,” significantly lower than clusters LLA and GVA, for the preceding year as well as the past (p < .001) but did not differ from cluster FO. This PP subtype resembles the DB subtype of Holtzworth-Munroe and Stuart’s (1994) typology and the PP cluster as found by Thijssen and de Ruiter (2011).

Cluster 4: GVA (n = 35, 23%)

Cluster 4 (n = 35) was labeled the generally violent/antisocial subtype, and this cluster scored high on almost all four B-SAFER items. Although not significantly different from the other clusters, they had the highest score on “violent acts.” On “general criminality,” this cluster scored significantly higher than the other clusters, for the preceding year as well as for the past (p < .001). Furthermore, this cluster scored significantly higher than clusters FO and PP on “substance use problems” for the preceding year and the past (p < .001). In addition, this cluster scored significantly higher on “mental health problems” than clusters LLA and FO, for both the preceding year as well as the past (p < .001), but did not differ from the PP subtype. This subtype resembles most closely the GVA of Thijssen and de Ruiter (2011).

Prevalence Rates

The prevalence rates of the subtypes in our self-referred outpatient sample were compared with those in the court-ordered sample of Thijssen and de Ruiter (2011; see Table 4). The LLA and PP subtype prevalence rates were roughly comparable with the sample in Thijssen and de Ruiter (2011). In contrast to our hypothesis, the FO subgroup was larger in the court-ordered sample (n = 54, 37%) than in ours (n = 43, 28%); in both studies, this subtype was the largest subgroup. In addition, the GVA seemed somewhat more prevalent in our self-referred sample (n = 35, 23%) than in the court-ordered sample (Thijssen & de Ruiter, 2011; n = 26, 18%). A chi-square test was conducted to examine the differences in the two samples. Results showed that the differences in prevalence rates were not statistically significant, χ²(3, n = 100) = 3.94, p = .27.

Table 4.

Prevalence Rates of Subtypes Compared With Thijssen and de Ruiter (2011).

Subtypes	Current Self-Referred Sample	Thijssen and de Ruiter’s (2011) Court-Ordered Sample
LLA	26% (n = 41)	24% (n = 35)
PP	23% (n = 35)	21% (n = 31)
FO	28% (n = 43)	37% (n = 54)
GVA	23% (n = 35)	18% (n = 26)

Note. LLA = low-level antisocial; PP = psychopathology; FO = family only; GVA = generally violent/antisocial.

Discussion

The present study replicated Thijssen and de Ruiter’s (2011) findings in a sample of self-referred IPV perpetrators. The four subtypes were as follows: LLA, FO, PP, and GVA, in accordance with previous research. These subtypes differed on the descriptive dimensions general criminality, substance use problems, and mental health problems. These results are in line with previous research, indicating that IPV perpetrators can be meaningfully divided into different subtypes (Cunha & Gonçalves, 2013; Dixon & Browne, 2003; Holtzworth-Munroe & Stuart, 1994; Holtzworth-Munroe et al., 2000; Holtzworth-Munroe & Meehan, 2004; Huss & Ralston, 2008; R. Johnson et al., 2006; Langhinrichsen-Rohling et al., 2000; Thijssen & de Ruiter, 2011).

Individuals in the LLA subtype engage in moderate to serious violent acts and little to no general criminality. However, our LLA subtype differed from the LLA subtype of Thijssen and de Ruiter (2011), in that our LLA subtype had significantly more substance use problems, compared with the other subtypes. This can perhaps be explained by the particular setting from which our subjects were recruited: The mental health center where treatment is offered receives referrals from the local addiction treatment clinic. Furthermore, in the geographic region of West-Brabant, 301 to 400 per 100,000 residents sought help for addiction problems in 2012, which is much higher than the national mean annual prevalence. For instance, addiction to gamma-hydroxybutyric acid (GHB) rates are as high as >25 per 100,000 residents, compared with a national mean of 5 per 100,000 (Wisselink, Kuijpers, & Mol, 2013).

The FO subtype shows moderately severe IPV and is not likely to engage in general antisocial behavior. In addition, they are the least likely to have substance use and mental health problems. The FO cluster is the largest group in our self-referred outpatient sample, which is in line with our hypothesis and previous research (Dixon & Browne, 2003; R. Johnson et al., 2006), and concurs with the prevalence rates of Thijssen and de Ruiter (2011), who also found the FO cluster to be the largest group in their Dutch court-referred sample. An explanation for this finding might be that the FO subtype is the most common subtype in IPV research, regardless of the type of referral (Cavanaugh & Gelles, 2005).

The PP subtype resembles Holtzworth-Munroe and Stuart’s (1994) DB subtype but could not be labeled as such, as the item “mental health problems” we used covers all types of mental health problems and not only borderline symptomatology. Individuals in this subtype are more likely than the LLA and FO subtypes to have mental health problems. They engage in low-to-moderately violent acts and general criminality, and are not likely to have substance use problems.

Finally, the GVA subtype demonstrates serious IPV, and they are most likely to engage in general criminal behavior, compared with the other subtypes. In addition, this subtype has serious substance use and mental health problems. The GVA subtype (23%) is the least prevalent (together with the PP subtype) in our self-referred sample. Contrary to our expectation, the GVA subtype was somewhat more prevalent than the GVA subtype (18%) in the court-referred sample of Thijssen and de Ruiter (2011). However, statistical analyses revealed that this difference was not significant. This finding could perhaps be explained by the fact that some clients reported to us they participated voluntarily in treatment to obtain a more favorable position in possible future legal proceedings. GVA perpetrators, who are charged for their IPV, may try to evade legal prosecution by engaging in voluntary treatment.

Our findings indicate it is possible to distinguish IPV perpetrators across subtypes. The B-SAFER has again shown itself to be a useful tool for distinguishing subtypes. Results showed no significant differences between the prevalence rates of the subtypes in our self-referred outpatient sample and the court-ordered sample of Thijssen and de Ruiter (2011). Unexpectedly, the prevalence of the GVA subtype was relatively high in our sample, compared with the previous court-ordered sample.

All subtypes in the present sample were indiscriminately admitted to the treatment program on the basis of a general intake procedure that did not include a structured violence risk assessment. Most current IPV treatment programs in the Netherlands use little to no mechanisms through which they match the treatment to specific individual criminogenic needs. In addition, little consideration is given to the level of risk for reoffending (Cavanaugh & Gelles, 2005; Tierolf, Lünnemann, & Steketee, 2014). Tierolf and colleagues (2014) recommended the use of structured risk assessment in all cases of IPV to be able to provide an adequate safety/risk management plan, which is not practiced at present.

As previously stated, the GVA batterers are seriously violent and are most likely to engage in general criminal behavior. In addition, this subtype has serious substance use and mental health problems. Moreover, they are 4 times more likely to drop out and have a higher risk for future IPV (Huss & Ralston, 2008; Langhinrichsen-Rohling et al., 2000; Thijssen & de Ruiter, 2011). Although we expected only a small percentage of this subtype in our self-referred sample, our findings contradict this assumption. On the basis of our study’s findings and in accordance with Tierolf et al. (2014), we recommend the implementation of structured risk assessment of domestic violence risk to assist risk management and treatment planning as a standard practice in the intake procedure for IPV interventions, regardless of the setting.

Structured risk assessment can be used to inform risk management planning, for example, in the form of matching of risk, need, and responsivity (RNR; Andrews & Bonta, 2010; Cavanaugh & Gelles, 2005; Holtzworth-Munroe et al., 2000; Huss, Covell, & Langhinrichsen-Rohling, 2006; Murphy & Eckhardt, 2005). Treatment programs that follow the principles of RNR (Andrews & Bonta, 2010) have the largest effect sizes in reducing violent recidivism. The risk principle states that the intensity of treatment should be matched to the risk level of the offender (i.e., high-risk offenders should receive high-intensity services, while low-risk offenders should receive low-intensity services). The need principle indicates that effective treatment programs should target the offender’s criminogenic needs (e.g., PP, substance abuse, procriminal attitudes). The responsivity principle states that effective treatment is tailored to the cognitive capabilities and learning style of the participant (Andrews & Bonta, 2010).

A number of scholars in the field of IPV have concluded that there exists limited scientific support for the effectiveness of treatment programs for this target group (e.g., Babcock, Green, & Robie, 2004; Saunders, 2008; Stover, Meadows, & Kaufman, 2009). A study of treatment effectiveness of IPV treatment programs in the four largest cities in the Netherlands concluded that 1½ year after intake, half of the women who experienced IPV and sought help were still a victim of serious physical violence (Tierolf et al., 2014). In addition, high attrition is a serious problem in IPV treatment; attrition rates range from 22% to 90%, with higher rates in self-referred samples than in court-ordered ones (Brown, O’Leary, & Feldbau, 1997; Jewell & Wormith, 2010; Rooney & Hanson, 2001). However, batterer subtypes differ greatly in terms of attrition rates. The characteristics of treatment dropouts (e.g., young, low income, from abusive families, prior criminal history, substance abuse) are found to be similar to the characteristics of assaulters with a high risk of recidivism, such as the GVA subtype (Bowen & Gilchrist, 2006; Buttel & Carney, 2008; Huss & Ralston, 2008; Jewell & Wormith, 2010; Rooney & Hanson, 2001). Along similar lines, research indicates that IPV perpetrators who resemble the GVA subtype were the least likely to complete treatment, in comparison with the FO and DB batterers (Langhinrichsen-Rohling et al., 2000).

Although we were able to replicate the findings of Thijssen and de Ruiter (2011) in a self-referred sample, this study has some important limitations. First, the coding of the B-SAFER was sometimes difficult as not all patient files were complete or sufficiently detailed. Still, interrater agreement for the two raters was good to excellent, indicating the quality of the files was adequate to achieve consensus. Second, the four clusters did not differ significantly on the item “violent acts.” In comparison with Thijssen and de Ruiter (2011), the average scores on this item were higher in our study. Especially, the scores of the LLA subtype and the past score of the FO subtype were higher in our sample. Possibly due to these higher scores, the differences found between the subtypes by Thijssen and de Ruiter (2011) were not replicated in our study. We can only speculate about the reasons for this divergence in severity of the violent IPV acts. Perhaps the court-referred sample tends toward minimization of the actual violent behavior toward the partner, because they have just been arrested by the police and are still awaiting the court’s decision. The self-referred sample is perhaps a bit more open about their violent behaviors, because they do not have to fear sentencing. Finally, our sample is not a random sample drawn from the community. This means we should be careful in generalizing our findings to the general population of spousal assaulters.

In summary, our findings replicate the IPV typology developed by Holtzworth-Munroe, distinguishing four subtypes: LLA, FO, PP, and GVA. Furthermore, our findings underscore the relevance of structured risk assessment for IPV, which can be used to assist in referral to tailored treatment and effective risk management.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

Author Biographies

Colinda M. B. Serie, MSc, obtained her master’s degree (cum laude) in forensic psychology from Maastricht University in 2014. She completed her master’s thesis “Comparing Subtypes of Spousal Assaulters in a Self-Referred Sample: Differences in Attrition and Treatment Outcome?” at GGZ Westelijk Noord-Brabant, the Netherlands. Her objective is to bridge the gap between science and practice, by working toward a position as a skilled scientist-practitioner in the field of forensic psychology.

Carola A. van Tilburg, MSc, is a licensed mental health psychologist in the Netherlands, who specializes in the assessment and treatment of aggression regulation problems. She is a supervisor for the Dutch Association for Cognitive and Behavioral Therapy and co-author of the manualized treatment program “Not Losing It Anymore.” She is currently conducting research toward a PhD degree.

Arno van Dam, PhD, is a licensed mental health psychologist in the Netherlands. He obtained his PhD in 2014 from Radboud University, Nijmegen. He specializes in the treatment of aggression regulation problems and is the main author of the manualized treatment “Not Losing It Anymore,” which is based on cognitive behavioral principles. He regularly supervises mental health psychologists in training. He is also an adjunct clinical professor at Tilburg University.

Corine de Ruiter, PhD, is professor of forensic psychology at Maastricht University. She is a licensed clinical psychologist in the Netherlands and offers supervision and consultation in psychological assessment. Her research of the past 15 years has focused on the relationship between mental health problems and violence, and on methods for violence risk assessment in specific. She served as president (2011-2013) of the International Association for Forensic Mental Health Services and as associate editor for the International Journal of Forensic Mental Health (2009-2014).

References

Andrews

D. A.

Bonta

(2010). The psychology of criminal conduct (5th ed.). New Providence, NJ: LexisNexis Matthew Bender.

Cheung

Kropp

Yuk-chung

Lam

G. L. T.

Sung

(2008). A preliminary validation of the Brief Spousal Assault Form for the Evaluation of Risk (B-SAFER) in Hong Kong. Journal of Family Violence, 23, 727-735. doi:10.1007/s10896-008-9198-z

Babcock

J. C.

Green

C. E.

Robie

(2004). Does batterers’ treatment work? A meta-analytic review of domestic violence treatment. Clinical Psychology Review, 23, 1023-1053. doi:10.1016/j.cpr.2002.07.001

Belfrage

Strand

(2012). Measuring the outcome of structured spousal violence risk assessments using the B-SAFER: Risk in relation to recidivism and intervention. Behavioral Sciences & the Law, 30, 420-430. doi:10.1002/bsl.2019

Bowen

Gilchrist

(2006). Predicting dropout of court-mandated treatment in a British sample of domestic violence offenders. Psychology, Crime & Law, 12, 573-587. doi:10.1080/10683160500337659

Brown

P. D.

O’Leary

K. D.

Feldbau

S. R.

(1997). Dropout in a treatment program for self-referring wife abusing men. Journal of Family Violence, 12, 365-386. doi:10.1023/A:1021906825073

Buttel

F. P.

Carney

M. M.

(2008). A large sample investigation of batterer intervention program attrition: Evaluating the impact of state program standards. Research on Social Work Practice, 18, 177-188. doi:10.1177/1049731508314277

Cavanaugh

M. M.

Gelles

R. J.

(2005). The utility of male domestic violence offender typologies: New directions for research, policy and practice. Journal of Interpersonal Violence, 20, 155-166. doi:10.1177/0886260504268763

Cunha

Gonçalves

R. A.

(2013). Intimate partner violence offenders: Generating a data-based typology of batterers and implications for treatment. The European Journal of Psychology Applied to Legal Context, 5, 131-139. doi:10.5093/ejpalc2013a2

10.

De Ruiter

de Jong

E. M.

Reus

Thijssen

(2008). Risicotaxatie bij daders van relationeel geweld: De RISc vergeleken met de B-SAFER [Risk assessment in perpetrators of intimate partner violence: Comparing the RISc and the B-SAFER]. Utrecht, The Netherlands: Trimbos-instituut.

11.

De Ruiter

(2009). B-SAFER: Gestructureerde beoordeling van het risico van relationeel geweld [B-SAFER: Structured risk assessment of intimate partner violence]. Maastricht, The Netherlands: Maastricht University.

12.

De Ruiter

(2011). Risicotaxatie bij relationeel geweldplegers in de praktijk: De B-SAFER [Risk assessment in intimate partner violence in practice: The B-SAFER]. GZ-Psychologie, 3, 24-33.

13.

Dixon

Browne

K. D.

(2003). Heterogeneity of spouse abuse: A review. Aggression and Violent Behavior, 8, 107-130. doi:10.1016/S1359-1789(02)00104-0

14.

Emery

(2011). Disorder or deviant order? Re-theorizing domestic violence in terms of order, power, and legitimacy. A typology. Aggression and Violent Behavior, 16, 525-540. doi:10.1016/j.avb.2011.07.001

15.

Fleiss

J. L.

(1986). The design and analysis of clinical experiments. New York, NY: Wiley.

16.

Holtzworth-Munroe

Meehan

J. C.

(2004). Typologies of men who are maritally violent: Scientific and clinical implications. Journal of Interpersonal Violence, 19, 1369-1389. doi:10.1177/0886260504269693

17.

Holtzworth-Munroe

Meehan

J. C.

Herron

Rehman

Stuart

G. L.

(2000). Testing the Holtzworth-Munroe and Stuart (1994) batterer typology. Journal of Consulting and Clinical Psychology, 68, 1000-1019. doi:10.1037/0022-006X.68.6.1000

18.

Holtzworth-Munroe

Meehan

J. C.

Herron

Rehman

Stuart

G. L.

(2003). Do subtypes of maritally violent men continue to differ over time? Journal of Consulting and Clinical Psychology, 71, 728-740. doi:10.1037/0022-006X.71.4.728

19.

Holtzworth-Munroe

Stuart

G. L.

(1994). Typologies of male batterers: Three subtypes and the differences among them. Psychological Bulletin, 116, 476-497. doi:10.1007/s10896-008-9151-1

20.

Huss

M. T.

Covell

C. N.

Langhinrichsen-Rohling

(2006). Clinical implications for the assessment and treatment of antisocial and psychopathic domestic violence perpetrators. Journal of Aggression, Maltreatment, & Trauma, 13, 59-85. doi:10.1300/J146v13n01_04

21.

Huss

M. T.

Ralston

(2008). Do batterer subtypes actually matter? Treatment completion, treatment response, and recidivism across a batterer typology. Criminal Justice and Behavior, 35, 710-724. doi:10.1177/0093854808316218

22.

Jewell

L. M.

Wormith

J. S.

(2010). Variables associated with attrition from domestic violence treatment programs targeting male batterers: A meta-analysis. Criminal Justice and Behavior, 37, 1087-1113. doi:10.1177/0093854810376815

23.

Johnson

(2008). A typology of domestic violence: Intimate terrorism, violence resistance, and situational couple violence. Lebanon, NH: Northeastern University Press.

24.

Johnson

Gilchrist

Beech

A. R.

Weston

Takriti

Freeman

(2006). A psychometric typology of UK domestic violence offenders. Journal of Interpersonal Violence, 21, 1270-1285. doi:10.1177/0886260506291655

25.

Kropp

P. R.

Hart

S. D.

Belfrage

(2005). The Brief Spousal Assault Form for the Evaluation of Risk: User manual. Vancouver, British Columbia, Canada: ProActive ReSolutions.

26.

Langhinrichsen-Rohling

Huss

M. T.

Ramsey

(2000). The clinical utility of batterer typologies. Journal of Family Violence, 15, 37-53. doi:10.1023/A:1007597319826

27.

Lünnemann

K. D.

Bruinsma

M. Y.

(2005). Geweld binnen en buiten: Aard, omvang en daders van huiselijk en publiek geweld in Nederland [Violence inside and outside: Nature, extent and perpetrators of domestic and public violence in the Netherlands]. Den Haag, The Netherlands: Ministry of Justice.

28.

McGraw

K. O.

Wong

S. P.

(1996). Forming inferences about some intraclass correlation coefficients. Psychological Methods, 1, 30-46. doi:10.1037/1082-989X.1.1.30

29.

Murphy

C. M.

Eckhardt

C. I.

(2005). Treating the abusive partner: An individualized cognitive-behavioural approach. New York, NY: Guilford Press.

30.

Rooney

Hanson

R. K.

(2001). Predicting attrition from treatment programs for abusive men. Journal of Family Violence, 16, 131-149. doi:10.1023/A:1011106902465

31.

Saunders

D. G.

(2008). Personality characteristics of men who batter. In Renzetti

Edleson

(Eds.), Encyclopedia of interpersonal violence (pp. 212-215). Thousand Oaks, CA: Sage.

32.

Shrout

P. E.

Fleiss

J. L.

(1979). Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin, 86, 420-428. doi:10.1037/0033-2909.86.2.420

33.

Stark

(2007). Coercive control: How men entrap women in personal life. New York, NY: Oxford University Press.

34.

Stover

C. S.

Meadows

A. L.

Kaufman

(2009). Interventions for intimate partner violence: Review and implications for evidence-based practice. Professional Psychology: Research and Practice, 40, 223-233. doi:10.1037/a0012718

35.

Thijssen

de Ruiter

(2011). Identifying subtypes of spousal assaulters using the B-SAFER. Journal of Interpersonal Violence, 26, 1307-1321. doi:10.1177/0886260510369129

36.

Tierolf

Lünnemann

Steketee

(2014). Doorbreken geweldspatroon vraagt gespecialiseerde hulp. Onderzoek naar effectiviteit van de aanpak huiselijk geweld in de G4 [Breaking through a pattern of violence needs specialized help. Research into the effectiveness of targeting domestic violence in the G4]. Utrecht, The Netherlands: Verwey-Jonker Instituut.

37.

Van Dam

Van Tilburg

Steenkist

Buisman

(2009). Niet meer door het lint: Handleiding [Not losing it anymore: Manual]. Houten, The Netherlands: Bohn Stafleu van Loghum.

38.

Van der Veen

H. C. J.

Bogaerts

(2010a). Huiselijk geweld in Nederland: Overkoepelend syntheserapport van het vangst-hervangst-, slachtoffer- en daderonderzoek 2007-2010 [Domestic violence in the Netherlands: Umbrella report of the catch-recatch, victim- and perpetrator research 2007-2010]. Den Haag, The Netherlands: Ministry of Justice.

39.

Van der Veen

H. C. J.

Bogaerts

(2010b). Het Landelijk onderzoek huiselijk geweld2010: De methode en de belangrijkste resultaten [National study on domestic violence 2010: Method and most relevant findings]. Justitiële Verkenningen, 36, 33-45.

40.

Van Dijk

Flight

Oppenhuis

Duesmann

(1997). Huiselijk geweld: Aard, omvang en hulpverlening [Domestic violence: Nature, prevalence and services]. Den Haag, The Netherlands: Ministry of Justice.

41.

Van Dijk

Veen

Cox

(2010). Slachtofferschap van huiselijk geweld: Aard, omvang, omstandigheden en hulpzoekgedrag [Victims of domestic violence: Nature, prevalence, circumstances and helpseeking behavior]. Den Haag, The Netherlands: Ministry of Justice.

42.

Waltz

Babcock

J. C.

Jacobson

N. S.

Gottman

J. M.

(2000). Testing a typology of batterers. Journal of Consulting and Clinical Psychology, 68, 658-669. doi:10.1037/0022-006X.68.4.658

43.

White

R. J.

Gondolf

E. W.

(2000). Implications of personality profiles for batterer treatment. Journal of Interpersonal Violence, 15, 467-488. doi:10.1177/088626000015005002

44.

Wisselink

D. J.

Kuijpers

W. G. T.

Mol

(2013). Kerncijfers verslavingszorg 2012 [Core numbers addiction care 2012]. Houten, The Netherlands: Stichting informatieve voorziening zorg.