The Lethality Screen

Abstract

This research is an examination of the predictive validity of the Lethality Screen, a tool used in conjunction with the Lethality Assessment Program (LAP). This intimate partner violence (IPV) risk assessment is an 11-item version of the Danger Assessment (DA) that was designed to be user-friendly for first responders and to maximize sensitivity. Participants (N = 254) were recruited into the study at the scene of police-involved IPV incidents in one Southwestern state and subsequently participated in two structured telephone interviews approximately 7 months apart. These analyses provide evidence that the Lethality Screen has considerable sensitivity (92%-93%) and a high negative predictive value (93%-96%) for near lethal and severe violence. However, specificity was low (21%). The Lethality Screen also has good agreement with the DA and IPV survivors’ perception of risk. The high sensitivity and low specificity should be considered carefully when determining whether the Lethality Screen is appropriate for particular areas of practice with IPV survivors and/or perpetrators.

Keywords

Lethality Assessment Program risk assessment domestic violence predictive validity

It is estimated that 35% of women in the United States will experience intimate partner violence (IPV) in their lifetimes, and that 25% of women will experience severe IPV in their lifetimes (Black et al., 2011). In addition to the injury suffered by women due to the violence inflicted on them by their partners, women who have experienced IPV are more likely to report physical and mental health problems (Campbell, 2002; Devries et al., 2013; Ruiz-Pérez, Plazaola-Castaño, & del Río-Lozano, 2007; Tadegge, 2008); these difficulties are exacerbated for women who experience severe IPV (Campbell et al., 2003). Physical IPV has been found to be a precursor to intimate partner femicide (the killing of women) in 65% to 80% of cases (Campbell, Glass, Sharps, Laughon, & Bloom, 2007; Campbell et al., 2003; Moracco, Runyan, & Butts, 1998; Sharps, Campbell, Campbell, Gary, & Webster, 2001). This is particularly significant as a large proportion (30%-70%) of femicide victims are killed by an intimate partner (Campbell et al., 2007; Catalano, Smith, Snyder, & Rand, 2009; Puzone, Saltzman, Kresnow, Thompson, & Mercy, 2000; Rennison & Welchans, 2000; Violence Policy Center, 2012).

IPV Risk Assessment

IPV risk assessment has been recommended for use across criminal justice and advocacy settings to determine which offenders are at risk for perpetrating future violence and homicide (Bennett, Goodman, & Dutton, 2000; Campbell, 2004; Hilton et al., 2004; Kress, Protivnak, & Sadlak, 2008; Roehl & Guertin, 2000). There are five stand-alone IPV risk assessments that have been created and tested for predictive validity in multiple research studies. These are the Danger Assessment (DA; Campbell et al., 2003), the Domestic Violence Screening Inventory (DVSI, DVSI-R; Williams & Grant, 2006; Williams & Houghton, 2004), the Kingston Screening Instrument for Domestic Violence (K-SID; Gelles & Straus, 1990, as cited in Campbell, O’Sullivan, Roehl, & Webster, 2005), the Spousal Assault Risk Assessment (SARA; Kropp & Hart, 2000), and the Ontario Domestic Assault Risk Assessment (ODARA; Hilton et al., 2004). A meta-analysis examining the average predictive validity of these five IPV risk assessments found the ODARA to have the greatest predictive validity followed by the SARA, the DA, the DVSI, and the K-SID (Messing & Thaller, 2013). These results should be interpreted with caution, however, as there is much variation in the methodology and outcome measures of the research studies that have evaluated IPV risk assessments and the risk assessments are not necessarily comparable.

Developed IPV risk assessments are intended to be used in a variety of settings (criminal justice, social service) by different professionals (health professionals, clinicians, police officers, probation officers, social service professionals) gathering data in assorted ways (case files, interviews with victims, interviews with offenders). In addition, they are also intended to predict several different outcomes (reassault, reoffense within the criminal justice system, lethality) and have unique goals (safety planning, criminal justice decision making, investigation; for a review, see Messing & Thaller, 2015). Several of these risk assessments have additionally been revised for various purposes and populations. The DA has been modified for use in female same-sex relationships (DA-R; Glass et al., 2008) and with immigrant women (DA-I; Messing et al., 2013), the SARA has been adapted for use by first responders (B-SAFER, also called the SARA Screening Version; Kropp & Hart, 2004), and the ODARA has been supplemented for use with high-risk offenders (Domestic Violence Risk Appraisal Guide [DVRAG]; Hilton, Harris, Rice, Houghton, & Eke, 2008). Although these instruments and their revisions have been developed and tested primarily in the United States and Canada, there is increasing international attention to the use of IPV risk assessment (e.g., Stockl et al., 2013) and risk assessments, such as the Severe Intimate Violence Partner Risk Prediction Scale (SIVIPAS), have been developed internationally (Echeburúa, Fernandez-Montalvo, Corral, & López-Goñi, 2009).

IPV risk assessment should be utilized within an evidence-based practice framework where the risk assessment instrument is determined to be the best evidence of risk of reassault or homicide and is considered within the context of practitioner expertise and client self-determination (Messing & Thaller, 2015). Risk assessment has been demonstrated to be more accurate than clinical prediction (Ægisdottir et al., 2006; Grove, Zald, Lebow, Snitz, & Nelson, 2000) and some IPV risk assessment instruments have demonstrated higher average predictive validity than survivor prediction (Messing & Thaller, 2013). However, a practitioner’s concern for the safety of their client, or a victim’s concern for her own safety, should not be diminished due to low scores on a risk assessment tool, particularly because victim-survivors of IPV are more likely to underestimate than overestimate their risk (Campbell, 2004; Heckert & Gondolf, 2000). The process of risk assessment can be used to educate IPV victim-survivors about risk and risk factors as part of an empowerment-based safety planning intervention focused on client self-determination (Campbell, 2001, 2004). This type of education may motivate victim-survivors of IPV to take action as recognition that violence is escalating often prompts protective actions and safety concerns (Burke, Denison, Gielen, McDonnell, & O’Campo, 2004; Gondolf & Fisher, 1988; Martin et al., 2000; Pape & Arias, 2000; Short et al., 2000). Practitioner expertise should additionally be used to suggest appropriate risk-informed interventions for both victim-survivors and offenders.

The Lethality Screen

This research study is an evaluation of the predictive and convergent validity of the Lethality Screen, a shortened (11-item) version of the DA that was created for field practitioners or first responders. The scoring system is designed for ease of use with results of “high danger” or “not high danger.” The Lethality Screen is intended to be used in combination with the Lethality Assessment Protocol. Together, the screen and the protocol make up Lethality Assessment Program (LAP; mnadv.org/lethality.html). The LAP combines the criminal justice and social service response to IPV by providing an opportunity for victim-survivors who screen in at high risk for homicide to speak to an advocate on the telephone at the scene of a police-involved IPV incident. The intervention is intended to be brief, to educate the victim-survivor about risk and risk factors, provide some immediate safety planning information, and to encourage the victim-survivor to obtain services. Although previous risk assessments have been designed for use with first responders, the Lethality Screen is the first risk assessment created for first responders that asks questions only of the victim-survivor of violence, is designed to predict severe violence/homicide, and is intended to maximize sensitivity.

The LAP follows the guidelines for use of risk assessment in an evidenced-based practice framework. Incorporating practitioner expertise (or professional judgment), officers are able to screen victim-survivors into the intervention if they believe that they are at high risk, regardless of the score on the Lethality Screen. In addition, the advocate who speaks to the victim-survivor on the telephone is aware of the victim-survivor’s answers on the risk assessment and can tailor their suggested safety strategies accordingly. Self-determination is also important, and victim-survivors are able to refuse to answer any questions on the Lethality Screen and speak to the hotline advocate only if they choose to do so. The Lethality Screen is also considered an educational tool. Whether or not the victim-survivor is at high risk, the officer can use the risk factors listed to educate the victim-survivor about risk of homicide in an intimate relationship. Although the Lethality Screen is intended to be used with male or female victim-suvivors of IPV, this study focuses on female victim-survivors only given their increased risk for IPV, IPV-related injury, and intimate partner homicide (Catalano, 2013).

Method

Sample Recruitment and Data Collection

Participants were recruited into the study at the scene of police-involved IPV incidents in seven participating police jurisdictions in one Southwestern state. It is suggested that the officer at the scene use the Lethality Screen near the end of an investigation involving a past or current intimate relationship, when there is a manifestation of danger, defined by the presence of at least one of the following: (a) the officer believes that an assault or other violent act has occurred whether or not there is probable cause for arrest, (b) the officer is concerned for the safety of the victim-survivor once they leave the incident scene, (c) the officer is responding to a domestic violence call from a victim-survivor or at a location where domestic violence had occurred in the past, or (d) the officer has a “gut feeling” that the victim-survivor is in danger. As such, when police officers responded to an incident of IPV and a manifestation of danger was present, the officer was asked to inquire whether he or she could refer the victim-survivor of violence into the study. If a victim-survivor was willing to speak with researchers, after the police intervention, officers collected one or two safe telephone numbers and a safe time for researchers to call the potential participant.

Contact information for potential participants was emailed or faxed to researchers, typically within 1 to 5 days, depending on the department. Contact information was gathered from 1,137 women between January 2009 and October 2010. Researchers were unable to contact 486 (42.7%) women due to disconnected, never answered, or incorrect telephone numbers. In addition, 47 (4.1%) women who were contacted were not eligible to participate in the study (e.g., below 18 years or not a victim-survivor of IPV). This left 604 eligible referrals who could be contacted by researchers; of these, 440 (72.8%) women participated in a structured interview at baseline and agreed to be contacted 6 months later to participate in a follow-up interview. Women provided many reasons for not wanting to participate in the interview including that they did not want to talk/think about their abuse, it was too emotionally difficult, they did not have time, they had a negative experience/did not trust the police, or they were scared of their abuser.

When possible, researchers obtained safe alternate forms of contact (e.g., email, address, home telephone number, work telephone number, cellular telephone number), and the names and contact information of three people (not their abusive partner) who would know how to contact them if the research team was unable to reach the participant at follow-up. This is a particularly hard to reach population as women may move, hide, or change their contact information to escape an abusive partner. Researchers were able to contact 254 (57.73%) participants at a median time of 7 months to follow-up. This retention rate is similar to that of other research studies using similar methodology (Campbell et al., 2005; Goodman, Dutton, & Bennett, 2000). Researchers attempted to contact women until they had talked to the participant and she had completed the interview or until they had tried to call a disconnected or unanswered number for more than 30 days and exhausted leads provided by alternate contacts if applicable. No participants that researchers were able to contact refused to participate in the second interview.

Interviews lasted approximately 45 minutes to one hour. Interviewers obtained verbal informed consent from participants at the start of the first interview, and were trained to adhere to strict safety protocols because of the potential risk that IPV victim-survivors face if their partner learns that they have participated in research about their violent relationship (Johnson, 2005). The Institutional Review Boards of the University of Oklahoma Health Sciences Center, Oklahoma State Department of Health, Arizona State University, The Johns Hopkins University, Cherokee Nation, and the National Institute of Justice approved this research.

Measures

Demographic and relationship characteristics

To describe the sample, at the baseline interview, participants were asked to report their employment status, educational achievement, racial/ethnic background, and age in years. Although participants were able to report as many racial/ethnic identities as they deemed appropriate, responses to this question were collapsed into five mutually exclusive categories: White, African American, Latina, Native American, and Multiracial/Other. With regard to their intimate relationship, participants were asked to report their legal marital status as single, married, or separated/divorced. Because legal marital status may not describe the state of a participant’s relationship, participants were also asked to report whether they had children with their abusive partner and their current level of involvement with their abusive partner (intimately involved and living in the same household, intimately involved and not living together, on again off again relationship, or no intimate relationship currently). At follow-up, participants were asked to report whether they had been intimately involved with their partner since baseline and whether they were currently intimately involved with their partner.

The Lethality Screen

As described above, the Lethality Screen is an 11-item version of the DA intended to be used by first responders in combination with the LAP. To simulate as closely as possible the experience of administering the Lethality Screen in the field at the scene of a police-involved IPV incident, during the baseline interview, researchers asked participants the Lethality Screen questions first, as soon as possible after the informed consent process. Participants’ partners were classified as “high danger” or “not high danger” using the same scoring procedures that are used by police officers in the field. In practice, if the officer screening the victim-survivor hears or observes something that makes him or her think that the woman is in a high danger situation, then the victim-survivor is also screened in as “high danger.” Because researchers (not officers) were administering the Lethality Screen, this element of professional judgment was not included in the scoring.

The 11-question screen includes 3 questions at the beginning, most associated with risk for homicide or near homicide on the DA (Campbell, Webster, & Glass, 2009), that automatically classify the perpetrator as “high danger” if the victim-survivor responds in the affirmative: use/threats with a weapon, threats to kill, and victim-survivor belief that their partner is capable of killing them. If the woman responds “no” to each of the first three questions, but responds “yes” to four or more of the additional 8 questions, she is also screened in as “high danger”: perpetrator access to guns, strangulation, recent separation, extreme jealousy and/or controlling behaviors, perpetrator suicide threats or attempts, stalking, perpetrator unemployment, or whether the victim-survivor has a child that is not the abuser’s. Each of the risk factors on the lethality screen has been found to increase risk for intimate partner femicide in previous research (Campbell et al., 2007; Campbell et al., 2003; Fox & Zawitz, 2004; Glass et al., 2008; Koziol-McLain et al., 2006; McFarlane et al., 1999; Wilson & Daly, 1993; Wilson, Johnson, & Daly, 1995).

The DA

The questions on the DA were asked at the baseline interview and include an increase in the frequency and severity of abuse (Campbell et al., 2003), having a child that is not the abuser’s (Campbell et al., 2003), recent estrangement (Wilson et al., 1995), controlling behaviors (Campbell et al., 2003), stalking (McFarlane et al., 1999), threats to kill or threats with a weapon (Campbell et al., 2003), strangulation (Glass et al., 2008), partner access to a firearm (Campbell et al., 2007; Campbell et al., 2003; Fox & Zawitz, 2004), perpetrator unemployment (Campbell et al., 2003), partners use of illegal drugs or problem drinking (Sharps, Campbell, Campbell, Gary, & Webster, 2001), avoiding arrest for domestic violence (Campbell et al., 2003), abuse during pregnancy (McFarlane, Campbell, Sharps, & Watson, 2002), forced sex (Campbell et al., 2003), extreme sexual jealousy (Campbell et al., 2003), and suicide threats or attempts (Koziol-McLain et al., 2006). The items on the DA are weighted and summed to produce an overall score that can range up to 37. The scores are placed into four categories of risk: variable danger (0-7), increased danger (8-13), severe danger (14-17), and extreme danger (18 or higher).

Participant assessment of risk

Participants’ assessment of risk for future abuse and future injurious violence were examined with two questions at baseline assessed on a scale of 0 (not likely at all) to 10 (as likely as possible): “How likely do you think it is that your partner will be physically abusive with you in the next year?” and “How likely is it that your partner will seriously hurt you in the next year?” (Cattaneo, Bell, Goodman, & Dutton, 2007; Weisz, Tolman, & Saunders, 2000).

IPV and abuse at follow-up

Experiences of IPV and abuse were assessed at follow-up using an adapted version of the revised Conflict Tactics Scale (CTS-2; Straus, Hamby, Boney-McCoy, & Sugarman, 1996). Additional items examining near lethal violence were added from the Risk Assessment Validation (RAVE) study (Campbell et al., 2005). All IPV and abuse was assessed since the last interview, and are reported as dichotomous variables: “this has happened since the baseline interview” and “this has not happened since the baseline interview.” Near lethal violence was assessed using two questions from the RAVE study that were combined for this analysis: “Since we talked to you last on [date], has your partner done anything that might have killed you or nearly killed you, whether or not he intended to?” and “Since the last time I interviewed you, has your partner tried to kill you?” An affirmative response to either or both of these questions resulted in a “yes” indicator for near lethal violence and a negative response to both questions resulted in a “no” indictor. Severe IPV was defined following the example of the RAVE study and included near lethal violence as described above and/or a “yes” response to at least one of the following: partner “used force (like hitting you, holding you down, or using a weapon) to make you have sex,” “used a knife or gun on you,” “punched you or hit you with something that could hurt,” “choked you,” “beat you up,” “burned or scalded you on purpose,” and/or “kicked you.” Any IPV since the previous interview was defined as an experience of any of the above forms of violence and/or any one or more of the following: partner “threw something at you that could hurt,” “twisted your arm or hair,” “made you have sex without a condom,” “pushed or shoved you,” “slammed you against a wall,” “grabbed you,” and “slapped you.” Intimate partner abuse included any of the previous forms of violence and/or one more of the following forms of verbal or psychological abuse: “insulting and swearing at you,” “shouting and yelling at you,” and “calling you fat or ugly or a lousy lover.”

Data Analysis

Predictive validity, the correct prediction of future events, is the most important measurement of the accuracy of a risk assessment instrument. The function of predictive validity is twofold. First, it is important to correctly identify cases (called sensitivity = the number of true positives / number of true positives + number false negatives); that is, “Is the IPV risk assessment instrument able to correctly identify the people who will be revictimized by near lethal, severe, repeat violence, or abuse on follow-up?” Second, correct classification of non-cases (called specificity = number of true negatives / number of true negatives + number false positives) is also important: “Can an IPV risk assessment correctly identify those people who will not be revictimized?” Similarly, the positive predictive value (= number of true positives / number of true positives + number false positives) examines the proportion of people who were revictimized out of those expected to be revictimized and the negative predictive value (number of true negatives / number of true negatives + number of false negatives) provides information on the proportion of those who were not revictimized out of the number expected not to be revictimized. Predictive validity is generally considered high when the sensitivity, specificity, positive predictive value, and negative predictive value of a risk assessment instrument are maximized and the false positive (people expected to be revictimized who are not revictimized) and false negative (people expected not to be revictimized who were revictimized) rates are minimized.

The Lethality Screen is a shortened version of the DA and it is, therefore, informative to compare the long and short versions of the instrument. Because the Lethality Screen and DA categories are categorical variables, a chi-square analysis was used to make the comparison. We also examined the relationship of the Lethality Screen with participants’ perception of risk from the baseline interview using t tests to examine the relationship between a dichotomous categorical variable (“not high danger”/“high danger”) to a linear scale (0-10).

Results

Participants (n = 254) were primarily White (38.98%), followed by African American (33.46%), Native American (11.81%), Multiracial/Other (11.42%), and Latina (4.33%). The average age of respondents was 33.65 (SD = 9.99). Approximately half (46.46%) of women in the sample were employed either part-time or full-time and the majority (83.07%) had a high school education or higher. Most participants reported that they were single (60.24%), followed by married (21.26%) and separated or divorced (18.11%). While nearly half of the women in this sample (45.67%) reported that they had a child in common with their abusive partner, the majority (68.38%) reported that they had no intimate relationship with their partner at the baseline interview. Smaller proportions of the sample reported that they were intimately involved and living with their partner (16.21%), intimately involved but not living with their partner (10.67%) and in an on again, off again relationship with their partner (4.74%) at the baseline interview. At the follow-up interview, 36.22% of participants reported that they had been in an intimate relationship with their abusive partner since the previous interview and 25.6% of participants reported that they were currently in an intimate relationship with their abusive partner.

The Lethality Screen

Of the 254 women who completed the follow-up interview, 205 (80.71%) of their partners were classified as “high danger” at the baseline interview. Of those who screened in as high danger, 172 (83.9%) answered “yes” to one or more of the first three questions on the Lethality Screen. Table 1 provides the number and proportion of “yes” responses to the 11 questions on the Lethality Screen. Recent separation was a risk factor for the highest proportion of respondents (77.95%), followed by strangulation (71.26%), and jealous or controlling behavior (70.87%). Of the first three questions, perpetrator threats to kill the participant or her children were most prevalent with over half (52.36%) of women responding “yes” to this question.

Table 1.

Affirmative Responses to Questions on the Lethality Screen (n = 254).

	Yes Response
Lethality Screen Questions	n (%)
1. Has he/she ever used a weapon against you or threatened you with a weapon?	92 (36.22)
2. Has he/she threatened to kill you or your children?	133 (52.36)
3. Do you think he/she might try to kill you?	96 (37.80)
4. Does he/she have a gun or can he/she get one easily?	111 (43.70)
5. Has he/she ever tried to choke you?	181 (71.26)
6. Is he/she violently or constantly jealous or does he/she control most of your daily activities?	180 (70.87)
7. Have you left him/her or separated after living together or being married?	198 (77.95)
8. Is he/she unemployed?	146 (57.48)
9. Has he/she ever tried to kill himself/herself?	55 (21.65)
10. Do you have a child that he/she knows is not his/hers?	126 (49.61)
11. Does he/she follow or spy on you or leave threatening messages?	137 (53.94)

Predictive Validity

Near fatal violence

Ten women did not respond to the questions regarding near fatal violence, leaving a final sample size for this analysis of 244; 28 (11.48%) respondents reported experiences of near fatal violence since the baseline interview. The Lethality Screen has 92.86% sensitivity when predicting near fatal violence at follow-up. That is, of the women who experienced near lethal violence at follow-up, nearly 93% were classified as high danger on the Lethality Screen. As shown in Table 2, 2 of the 28 women who experienced near lethal violence were not classified as high danger on the Lethality Screen. The negative predictive value is also high; 95.83% of the women classified as not high danger did not experience near fatal violence. The specificity of the Lethality Screen is 21.30% for near fatal violence and the positive predictive value is 13.27%. This is primarily because of a large number of false positives. The Lethality Screen classified many women as high danger (n = 170) that did not experience subsequent near fatal violence during the follow-up period.

Table 2.

The Lethality Screen and Near Fatal Violence on Follow-Up.

	High Danger on the Lethality Screen
Near Fatal Violence on Follow-Up	No	Yes	Row Totals
No	46	170	216
Yes	2	26	28
Column totals	48	196	244

Severe violence

Of the 234 women who responded to the questions regarding severe violence at follow-up, 44 (18.80%) had experienced at least one form of severe violence since the baseline interview. The sensitivity of the Lethality Screen is approximately the same when predicting severe violence as when predicting near fatal violence, with 93.18% of the women who experienced severe violence at follow-up classified as high danger on the Lethality Screen (Table 3). The negative predictive value is high with 93.18% of the women classified as not high danger not experiencing severe violence at follow-up. Again, the specificity (21.58%) and positive predictive value (21.58%) of the Lethality Screen is low. While the positive predictive value is slightly higher than in the prediction of near fatal violence, the Lethality Screen continues to have a high number (n = 149) of false positives.

Table 3.

The Lethality Screen and Severe Violence on Follow-Up.

	High Danger on the Lethality Screen
Severe Violence on Follow-Up	No	Yes	Row Totals
No	41	149	190
Yes	3	41	44
Column totals	44	190	234

Any IPV

The final sample size for the analysis of the ability of the Lethality Screen to predict any IPV is 233. Of the women who responded to these questions, 69 (29.61%) experienced IPV at follow-up. The sensitivity of the Lethality Screen dips slightly to 86.96% when predicting any IPV. As shown in Table 4, the number of false negatives increased slightly; 9 of the 69 women who experienced IPV at follow-up were not classified as high danger on the Lethality Screen. The negative predictive value is also reduced when examining any IPV; 80% of the women classified as not high danger did not experience IPV at follow-up. The specificity of the Lethality Screen is 21.95% for any IPV and the positive predictive value is 31.91%. The Lethality Screen continues to have a high number (n = 128) of false positives.

Table 4.

The Lethality Screen and Any IPV on Follow-Up.

	High Danger on the Lethality Screen
Any IPV on Follow-Up	No	Yes	Row Totals
No	36	128	164
Yes	9	60	69
Column totals	45	188	233

Intimate partner abuse

The number of women who reported intimate partner abuse is relatively high at follow-up, with 148 of the 240 (61.67%) women included the final sample reporting physical, emotional, or psychological abuse. The sensitivity of the Lethality Screen is 83.78% when predicting intimate partner abuse. As shown in Table 5, there are more false negatives when predicting intimate partner abuse with 24 women who experienced abuse at follow-up not classified as high danger on the Lethality Screen. Because of this increase in false negatives, the negative predictive value decreases somewhat substantially to 47.83% though the positive predictive value increases to 63.91%. The specificity of the Lethality Screen is 23.91% and there are fewer false positives (n = 70) when examining intimate partner abuse as the outcome.

Table 5.

The Lethality Screen and Intimate Partner Abuse on Follow-Up.

	High Danger on the Lethality Screen
Intimate Partner Abuse on Follow-Up	No	Yes	Row Totals
No	22	70	92
Yes	24	124	148
Column totals	46	194	240

Additional Validation Analyses

Comparisons with the DA

Women’s weighted and summed baseline DA scores were categorized into variable danger, increased danger, severe danger, and extreme danger categories. DA category is significantly associated, χ²(3) = 99.14, p < .001, with the Lethality Screen (see Table 6). At the severe and extreme danger levels, the DA improves upon the specificity of the Lethality Screen for predicting near lethal and severe violence and the Lethality Screen improves upon the sensitivity of the DA, as designed.

Table 6.

Danger Assessment and Lethality Screen by Type of Violence Experienced on Follow-Up.

Danger Assessment Category		Experienced Near Fatal IPV		Experienced Severe IPV		Experienced Any IPV		Experienced IPA
	Lethality Screen	Not High Risk	High Risk	Not High Risk	High Risk	Not High Risk	High Risk	Not High Risk	High Risk
Variable danger		1	1	1	1	4	2	11	4
Increased danger		0	5	0	7	2	10	10	26
Severe danger		1	4	2	8	3	14	3	29
Extreme danger		0	16	0	25	0	34	0	65

Note. IPV = intimate partner violence. IPA=intimate partner abuse.

Survivor assessment of risk

We examined the relationship between the Lethality Screen and participants’ predictions of the likelihood that their partner will physically abuse or seriously hurt them in the next year. The Lethality Screen is significantly associated, t = −2.69 (df = 243), p < .01, with participants’ predictions that their partner is likely to physically abuse them in the next year and is significantly, t = −3.15 (df = 239), p < .01, associated with participants’ predictions that their partner is likely to seriously hurt them in the next year.

Discussion

The Lethality Screen was designed to maximize sensitivity and these analyses provide evidence that the Lethality Screen has considerable sensitivity (92%-93%) and a high negative predictive value (93%-96%) for near lethal and severe violence. Of those women who did not screen in as high danger, three women experienced severe violence in the follow-up period and two of those women also experienced near fatal violence. Specificity was low (21%) in these same analyses due to a high number of false positives. When examining experiences of any IPV or intimate partner abuse at follow-up, the sensitivity of the Lethality Screen decreases. This is likely because it was designed to identify the presence of risk factors for lethal and near lethal IPV and not reassault or repeat psychological abuse. The Lethality Screen is adequate at predicting any repeat IPV and intimate partner abuse (84%-87% sensitivity), but is not as sensitive as for severe violence and near lethality.

The Lethality Screen is intended to screen women into a brief advocacy intervention; therefore, sensitivity and the negative predictive value are much more important than specificity and the positive predictive value because false negatives are much more costly than false positives. Screening a woman into a brief advocacy intervention when her partner will not be violent again may cost the woman, the advocate, and the officer some time, but it does not do any harm and may be educational or empowering. On the other hand, depriving a woman of a brief advocacy intervention when her partner is even moderately likely to commit repeat violence, severe violence, or near lethal violence deprives her of an intervention that may educate her about safety options and/or encourage her to seek services from those best trained to provide in-depth assessment and safety planning. This is clearly more costly. Given the high number of false positives, in contexts where a high degree of specificity is desired, the Lethality Screen would not be appropriate. For example, specificity is much more important when a risk assessment is being used to make determinations that limit a perpetrator’s freedom, such as when determining criminal justice sanctions or bail decisions.

In cases where sensitivity is determined to be the most important factor, use of the Lethality Screen may not be necessary. In the case of the LAP, for example, some police departments may explore asking all victim-survivors whether they would like to speak to an advocate over the telephone at the scene of an IPV incident. Although this may be ideal, given budget constraints and high caseloads at many agencies, it may not be realistic. In addition, utilizing the protocol without the Lethality Screen would eliminate the educational function of informing abused women about their risk and risk factors for future severe violence and homicide. Speaking to women about risk factors and their potential for future severe and fatal violence through the use of a risk assessment tool may raise awareness about risk and risk factors and assist women in identifying risk within their intimate relationship, whether or not they screen in as high risk. As previous research has shown that safety concerns often motivate help seeking (Burke et al., 2004; Gondolf & Fisher, 1988; Martin et al., 2000; Pape & Arias, 2000; Short et al., 2000), using the Lethality Screen may encourage women to speak to an advocate or take other steps to become safer while either maintaining or leaving their relationship. This educational component may also be helpful for police officers as officers who have been educated about risk assessment and risk assessment tools may become more aware of the dynamics of abusive relationships and the risk that women face in those relationships.

Three women who experienced severe or near lethal violence at follow-up were not screened in as high danger on the Lethality Screen; two of these women were classified as being in severe danger based on the DA. While this is less than 1% of the women screened, it represents three important missed opportunities for intervention. As such, several analyses were conducted to determine whether the sensitivity of the Lethality Screen could be further increased. The best option consists of including unemployment as a question that automatically screens the victim-survivor in as high risk. This can be done while either moving the question about threats/assaults with a weapon to the following eight questions or making any one of the first four questions sufficient for high risk. In either case, the sensitivity for near lethal violence is 96.43% and the sensitivity for severe violence is 97.73%. This resulted in an increase in the false positive rate as 5% more people were screened in as high danger, two of whom experienced severe violence at follow-up. Even with this change, the Lethality Screen fails to capture one participant who experienced near lethal violence on follow-up.

The Lethality Screen also has good agreement with the DA and with women’s perception of risk. The Lethality Screen is intended to be administered by a first responder and to compliment the use of the DA with an advocate once a woman has chosen to access services. The DA is too lengthy to be used by first responders and has better specificity than the Lethality Screen, particularly at the severe and extreme danger categories. The use of the calendar with the full DA helps both women and advocates to understand the full scope and pattern of an abusive relationship, and is intended to be the starting point of comprehensive safety planning. While the Lethality Screen and subsequent on-scene intervention that make up the LAP are an important starting place for intervention, this should not replace the DA and lengthier advocacy interventions for women who choose to access services.

Study findings should be interpreted in the context of the limitations of this research. First, the sample represents data from women in seven jurisdictions in a single Southwestern state and participants may not be representative of victim-survivors of IPV in other jurisdictions or locales. Having police officers obtain contact information for victim-survivors at the scene of IPV incidents provided access to the population of interest, but may have introduced selection bias. Participants may be different from the women who chose not to participate or who could not be contacted by researchers. Although the response rate was high (72.8%) when a victim-survivor could be contacted, a large proportion of women could not be contacted by researchers (42.7%). Furthermore, only 57.7% of participants interviewed at baseline were retained at follow-up. It is important also to recognize that the time to follow-up was relatively short (approximately 7 months). It may be that women who screened in as high danger, but did not report subsequent violence in the follow-up period (i.e., false positives), experienced violence after the study ended.

Researchers, not first responders, administered the Lethality Screen. While the administration of the risk assessment was as close as possible to the experience of having an officer administer the screen, there were differences—the screen was not administered in person, by a first responder, or at the scene of an IPV incident. Rather, the Lethality Screen was administered as part of a research study after an informed consent procedure. In addition, in the field, the first responder can screen in anyone as “high danger,” regardless of how they answer the questions, lending an element of professional judgment that was not present in this research. Finally, the Receiver Operating Characteristic Area Under the Curve (AUC) is the suggested method of examining predictive validity and takes into account both sensitivity and specificity in a single analysis (Rice & Harris, 2005). However, this analysis was not possible because the AUC should be utilized with a continuous predictor variable and a dichotomous outcome (Douglas, Blanchard, Guy, Reeves, & Weir, 2000).

This study also has notable strengths. It is the first to examine the predictive validity of the Lethality Screen, a tool that is being used as a component of the LAP in 32 states. Women were referred to the study after a police-involved IPV incident, mirroring the situations in which the Lethality Screen is intended for use. Use of the Lethality Screen is generally followed by an intervention. However, because it is part of a larger evaluation of the LAP, this research study was able to administer the Lethality Screen without intervention to more accurately determine predictive validity. The sample was large and included a substantial number of Native American women (11.81%), who are at higher risk for experiencing IPV and intimate partner homicide (Wahab & Olson, 2004). Future research should seek to test the predictive validity of the Lethality Screen in different jurisdictions and should further seek to compare this instrument with other risk assessment instruments intended for use by first responders (e.g., the ODARA, B-SAFER). A research design in which first responders administer the screen and are able to screen in victim-survivors as “high danger” based on professional judgment would most closely match the use of this risk assessment instrument in the field. A longer time to follow-up with additional contact points to increase retention rates would also be ideal. The current research provides evidence that the Lethality Screen is functioning as intended, with a high level of sensitivity and a lack of specificity. While this is ideal for some practice contexts, this imbalance should be considered carefully when determining whether the Lethality Screen is appropriate for other areas of practice with IPV survivors and/or perpetrators.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research study was funded by the National Institute of Justice 2008-WG-BX-0002.

Author Biographies

Jill Theresa Messing, MSW, PhD, is an associate professor in the School of Social Work at Arizona State University. She is the principal investigator on the National Institute of Justice funded Oklahoma Lethality Assessment Study. Her interest areas are intimate partner violence, domestic homicide/femicide, risk assessment, criminal justice-social service collaborations, and evidence-based practice.

Jacquelyn Campbell, PhD, RN, FAAN, is a professor at Johns Hopkins University School of Nursing. She has been principal investigator of 12 NIJ, NIH, DOD, and CDC funded research studies of intimate partner violence and homicide and more than 220 articles and seven books on the subject.

Janet Sullivan Wilson, PhD, RN, is an associate professor in the College of Nursing at the University of Oklahoma Health Sciences Center.

Sheryll Brown, MPH, director of the Injury Prevention Service of the Oklahoma State Department of Health, has worked in injury and violence prevention for 19 years with a focus on violence, intimate partner violence, and disaster epidemiology. Since 2008, she has been a co-investigator for the Oklahoma Lethality Assessment Study.

Beverly Patchell, PhD, APRN, CNS, PMH-BC, is an assistant professor at the University of Utah College of Nursing and program director of the DNP Psychiatric Nurse Practitioner Program. She has an extensive background in working with Native American communities. Her community-based research has addressed substance abuse, mental health, cancer, HIV/AIDS, diabetes, and interpersonal violence.

References

Ægisdottir

White

M. J.

Spengler

P. M.

Maugherman

A. S.

Anderson

L. A.

Cook

R. S.

. . . Rush

J. D.

(2006). The meta-analysis of clinical judgment project: Fifty-six years of accumulated research on clinical versus statistical prediction. The Counseling Psychologist, 34, 383-390.

Bennett

Goodman

Dutton

M. A.

(2000). Risk assessment among batterers arrested for domestic assault: The salience of psychological abuse. Violence Against Women, 6, 1190-1203.

Black

M. C.

Basile

K. C.

Breiding

M. J.

Smith

S. G.

Walters

M. L.

Merrick

M. T.

(2011). National intimate partner and sexual violence survey. Atlanta, GA: Centers for Disease Control and Prevention.

Burke

J. G.

Denison

Gielen

A. C.

McDonnell

O’Campo

(2004). Ending intimate partner violence: An application of the Transtheoretical Model. American Journal of Health Behavior, 28, 122-133.

Campbell

J. C.

(2001). Safety planning based on lethality assessment for partners of batterers in intervention programs. Journal of Aggression, Maltreatment & Trauma, 5, 129-143.

Campbell

J. C.

(2002). Health consequences of intimate partner violence. The Lancet, 359, 1331-1336.

Campbell

J. C.

(2004). Helping women understand their risk in situations of intimate partner violence. Journal of Interpersonal Violence, 19, 1464-1477.

Campbell

J. C.

Glass

Sharps

P. W.

Laughon

Bloom

(2007). Intimate partner homicide: Review and implications of research and policy. Trauma, Violence, & Abuse, 8, 246-269.

Campbell

J. C.

O’Sullivan

Roehl

Webster

D. W.

(2005). Intimate partner violence risk assessment validation study: The RAVE study (Final Report to the National Institute of Justice (NCJ 209731-209732).

10.

Campbell

J. C.

Webster

D. W.

Glass

(2009). The Danger Assessment validation of a lethality risk assessment instrument for intimate partner femicide. Journal of Interpersonal Violence, 24, 653-674.

11.

Campbell

J. C.

Webster

D. W.

Koziol-McLain

Block

C. R.

Campbell

Curry

M. A.

. . . Laughon

(2003). Risk factors for femicide in abusive relationships: Results from a multi-site case control study. American Journal of Public Health, 9, 1089-1097.

12.

Catalano

(2013, November). Intimate partner violence: Attributes of victimization, 1993–2011. Washington, DC: US Department of Jusitce.

13.

Catalano

Smith

Snyder

Rand

(2009). Female victims of violence (Paper 7). Washington DC: U.S. Department of Justice Publications and Materials.

14.

Cattaneo

L. B.

Bell

M. E.

Goodman

L. A.

Dutton

M. A.

(2007). Intimate partner violence victims’ accuracy in assessing their risk of re-abuse. Journal of Family Violence, 22, 429-440.

15.

Devries

K. M.

Mak

Bacchus

L. J.

Child

J. C.

Falder

Petzold

. . . Watts

C. H.

(2013). Intimate partner violence and incident depressive symptoms and suicide attempts: A systematic review of longitudinal studies. PLoS Medicine, 10(5), e1001439. doi:10.1371/journal.pmed.1001439

16.

Douglas

K. S.

Blanchard

A. J. E.

Guy

L. S.

Reeves

K. A.

Weir

(2000). HCR-20 violence risk assessment scheme: Overview and annotated bibliography. Retrieved from http://escholarship.umassmed.edu/cgi/viewcontent.cgi?article=1362&;context=psych_cmhsr

17.

Echeburúa

Fernández-Montalvo

Corral

López-Goñi

J. J.

(2009). Assessing risk markers in intimate partner femicide and severe violence: A new assessment instrument. Journal of Interpersonal Violence, 24, 925-939.

18.

Fox

J. A.

Zawitz

M. W.

(2004). Homicide trends in the US. Washington, DC: Bureau of Justice Statistics.

19.

Glass

Laughon

Campbell

Block

C. R.

Hanson

Sharps

P. W.

Taliaferro

(2008). Non-fatal strangulation is an important risk factor for homicide of women. The Journal of Emergency Medicine, 35, 329-335.

20.

Gondolf

E. W.

Fisher

E. R.

(1988). Battered women as survivors: An alternative to treating learned helplessness. Lexington, MA: Lexington Books.

21.

Goodman

L. A.

Dutton

M. A.

Bennett

(2000). Predicting repeat abuse among arrested batterers: Use of the Danger Assessment Scale in the criminal justice system. Journal of Interpersonal Violence, 15, 63-74.

22.

Grove

W. M.

Zald

D. H.

Lebow

B. S.

Snitz

B. E.

Nelson

(2000). Clinical versus mechanical prediction: A meta-analysis. Psychological Assessment, 12, 19-30.

23.

Heckert

D. A.

Gondolf

E. W.

(2000). Assessing assault self-reports by batterer program participants and their partners. Journal of Family Violence, 15, 181-197.

24.

Hilton

N. Z.

Harris

G. T.

Rice

M. E.

Houghton

R. E.

Eke

A. W.

(2008). An indepth actuarial assessment for wife assault recidivism: The Domestic Violence Risk Appraisal Guide. Law and Human Behavior, 32, 150-163.

25.

Hilton

N. Z.

Harris

G. T.

Rice

M. E.

Lang

Cormier

C. A.

Lines

K. J.

(2004). A brief actuarial assessment for the prediction of wife assault recidivism: The Ontario Domestic Assault Risk Assessment. Psychological Assessment, 16, 267-275.

26.

Johnson

(2005). Assessing the prevalence of violence against women in Canada. Statistical Journal of the United Nations Economic Commission for Europe, 22(3), 225-238.

27.

Koziol-McLain

Webster

McFarlane

Block

C. R.

Ulrich

Glass

Campbell

J. C.

(2006). Risk factors for femicide-suicide in abusive relationships: Results from a multisite case control study. Violence and Victims, 21, 3-21.

28.

Kress

V. E.

Protivnak

J. J.

Sadlak

(2008). Counseling clients involved with violent intimate partners: The mental health counselor’s role in promoting client safety. Journal of Mental Health Counseling, 30, 200-210.

29.

Kropp

P. R.

Hart

S. D.

(2000). The Spousal Assault Risk Assessment (SARA) Guide: Reliability and validity in adult male offenders. Law and Human Behavior, 24, 101-118.

30.

Kropp

P. R.

Hart

S. D.

(2004). B-SAFER (Brief Spousal Assault Form for the Evaluation of Risk): A tool for criminal justice professionals. Ottawa, Canada: Department of Justice (Government of Canada). Retrieved from http://canada.justice.gc.ca/eng/rp-pr/fl-lf/famil/rr05_fv1-rr05_vf1/rr05_fv1.pdf

31.

Martin

A. J.

Berenson

K. R.

Griffing

Sage

R. E.

Madry

Bingham

L. E.

Primm

B. J.

(2000). The process of leaving an abusive relationship: The role of risk assessments and decision-certainty. Journal of Family Violence, 15, 109-122.

32.

McFarlane

J. M.

Campbell

J. C.

Sharps

Watson

(2002). Abuse during pregnancy and femicide: Urgent implications for women’s health. Obstetrics & Gynecology, 100, 27-36.

33.

McFarlane

J. M.

Campbell

J. C.

Wilt

Sachs

C. J.

Ulrich

(1999). Stalking and intimate partner femicide. Homicide Studies, 3, 300-316.

34.

Messing

J. T.

Amanor-Boadu

Cavanaugh

C. E.

Glass

Campbell

J. C.

(2013). Culturally competent intimate partner violence risk assessment: Adapting the Danger Assessment for immigrant women. Social Work Research, 37(3).

35.

Messing

J. T.

Thaller

(2013). The average predictive validity of intimate partner violence risk assessment instruments. Journal of Interpersonal Violence, 28, 1537-1558.

36.

Messing

J. T.

Thaller

(2015). Intimate partner violence risk assessment: A primer for social workers. British Journal of Social Work, 45, 1804-1820. doi:10.1093/bjsw/bcu012

37.

Moracco

K. E.

Runyan

C. W.

Butts

J. D.

(1998). Femicide in North Carolina, 1991-1993: A statewide study of patterns and precursors. Homicide Studies, 2, 422-446.

38.

Pape

K. T.

Arias

(2000). The role of perceptions and attributions in battered women’s intentions to permanently end their violent relationships. Cognitive Therapy and Research, 24, 201-214.

39.

Puzone

C. A.

Saltzman

L. E.

Kresnow

M. J.

Thompson

M. P.

Mercy

J. A.

(2000). National trends in intimate partner homicide: United States, 1976-1995. Violence Against Women, 6, 409-426.

40.

Rennison

C. M.

Welchans

(2000). Intimate partner violence (Special Report). Washington, DC: Bureau of Justice Statistics, Office of Justice Programs, U.S. Department of Justice.

41.

Rice

M. E.

Harris

G. T.

(2005). Comparing effect sizes in follow-up studies: ROC area, Cohen’s d, and r. Law and Human Behavior, 29, 615-620.

42.

Roehl

Guertin

(2000). Intimate partner violence: The current use of risk assessments in sentencing orders. The Justice System Journal, 21, 171-198.

43.

Ruiz-Pérez

Plazaola-Castaño

del Río-Lozano

(2007). Physical health consequences of intimate partner violence in Spanish women. The European Journal of Public Health, 17, 437-443.

44.

Sharps

Campbell

J. C.

Campbell

Gary

Webster

(2001). Risky mix: Drinking, drug use, and homicide. National Institiute of Justice Journal, 250, 9-13.

45.

Short

L. M.

McMahon

P. M.

Chervin

D. D.

Shelley

G. A.

Lezin

Sloop

K. S.

Dawkins

(2000). Survivors’ identification of protective factors and early warning signs for intimate partner violence. Violence Against Women, 6, 272-285.

46.

Stockl

Devries

Rotstein

Abrahams

Campbell

Walls

Garcia-Moreno

(2013). The global prevalence of intimate partner homicide: A systematic review. The Lancet, 382, 859-865.

47.

Straus

M. A.

Hamby

S. L.

Boney-McCoy

Sugarman

D. B.

(1996). The Revised Conflict Tactics Scales (CTS2): Development and preliminary psychometric data. Journal of Family Issues, 17, 283-316.

48.

Tadegge

A. D.

(2008). The mental health consequences of intimate partner violence against women in Agaro Town, southwest Ethiopia. Tropical Doctor, 38, 228-229.

49.

Violence Policy Center. (2012). When men murder women: An analysis of 2010 homicide data. Washington, DC: Author.

50.

Wahab

Olson

(2004). Intimate partner violence and sexual assault in Native American communities. Trauma, Violence, & Abuse, 5, 353-366.

51.

Weisz

A. N.

Tolman

R. M.

Saunders

D. G.

(2000). Assessing the risk of severe domestic violence: The importance of survivors’ predictions. Journal of Interpersonal Violence, 15, 75-90.

52.

Williams

K. R.

Grant

S. R.

(2006). Empirically examining the risk of intimate partner violence: The revised Domestic Violence Screening Instrument (DVSI-R). Public Health Reports, 131, 400–408.

53.

Williams

K. R.

Houghton

A. B.

(2004). Assessing the risk of domestic violence reoffending: A validation study. Law and Human Behavior, 28, 437–455.

54.

Wilson

Daly

(1993). Spousal homicide risk and estrangement. Violence and Victims, 8, 3-16.

55.

Wilson

Johnson

Daly

(1995). Lethal and nonlethal violence against wives. Canadian Journal of Criminology, 37, 331-361.