INTRODUCTION

Nearly 2 million women are victims of intimate partner violence (IPV) and between 3 and 10 million children are exposed to IPV each year in the United States.1–3 Numerous national associations have published recommendations detailing the role of physicians in screening women and children for IPV histories.4–7 Dozens of studies have assessed screening procedures whereby women in health care practices are asked about IPV. All of these recommendations and studies pursue a secondary prevention approach to IPV with individuals who are not those perpetrating the violence.

Given the scope and impact of IPV, the pervasive pursuit of secondary prevention is remarkable when framed by the absence of primary prevention recommendations and studies. Identifying and referring male perpetrators for preventive care—such as counseling or batterer programs—is an unexplored step, then, toward a primary prevention strategy that could have the potential to reduce the physical and psychological harm women and children experience as a result of IPV. Screening men for IPV perpetration as they seek primary care may be a possible site wherein this strategy could be pursued.

The aim of the study was to assess this possible strategy, specifically the feasibility and efficiency of primary care providers screening their male patients for IPV perpetration. We were particularly interested in whether men would disclose IPV perpetration when directly asked by a primary care provider, and whether both men and providers would be comfortable with the screening process. Although there was reason initially to consider whether the study should assess the feasibility and efficiency of screening only certain subgroups of men most at risk for perpetrating IPV, the literature indicated that characterizing who these subgroups are would be challenging. Across multiple samples of men, the variables identified as potential predictors of IPV perpetration have been found to be quite broad, and have included: being young; being married to or living with a partner, or both; reporting a high number of sexual partners; having fathered 3 or more children; adhering to traditional gender norms and belief in male dominance; having antisocial personality characteristics; being an ethnic/racial minority; having a drinking or drug use problem, or both; having combat-related posttraumatic stress disorder; and experiencing current depression.8–17 Some investigators suggested that many predictor variables (e.g., age, ethnicity/race) could be indicators for other more-primary variables such as socioeconomic status.17 And there was contradictory evidence about some variables, such as race/ethnicity; for example, the 1992 National Crime Victimization Survey found no statistically significant differences in IPV victimization rates across ethnic groups.18 Thus, we chose to examine routine screening of male patients seeking primary care as a primary IPV prevention strategy.

METHODS

We developed an IPV screening protocol to be integrated into the routine provision of primary care of male patients. We refer to this protocol as “RADAR for Men” (http://www.instituteforsafefamilies.org/pdf/healthcare/MenRADAR.doc [Accessed April 4, 2008], and based on the initial RADAR screen), where “R”=Routine inquiry (of all male patients 14 and older); “A”=Always ask; “D”=Document findings; “A”=Assess safety and lethality; and “R”=Respond.19 Numerous sample questions were provided to providers as part of the protocol to help them ask a male patient directed questions about whether he is in a relationship and, if so, whether he and his partner are ever violent with one another. If, in response to a provider’s asking of these questions, a patient reported current and/or past perpetration and/or victimization, providers were to assess the patient’s lethality and safety, respond to the patient with simple, direct counseling, and refer those men who needed it to specialty care.

Six primary health care providers agreed to participate in screening as many of their male patients with appointments as they were able to screen in a busy practice, and in collecting data on implementation of the screening. We asked providers to screen at least 1 day per week over the course of 5 weeks. The providers included: 1 male medical doctor (MD) at a university clinic (who screened 11 male patients over the course of the study); 2 female MDs at public health clinics (n 1 = 25, n 2 = 25); 1 female doctor of osteopathy (DO) at a federally qualified health center (FQHC; n = 25); 1 male physician assistant (PA) at a FQHC (n = 21); and 1 female nurse practitioner (NP) at a VA health clinic (n = 21). The study was approved by the Institutional Review Board of the University of Pennsylvania, the City of Philadelphia, Philadelphia Veterans Administration (VA) Medical Center, and Temple University. Patients provided verbal informed consent.

After screening patients, providers categorized men as current and/or past IPV perpetrators and/or victims and indicated whether they assessed for lethality/safety, responded directly to this assessment, and made a referral. Providers also indicated their and the patient's comfort during screening (using a 5-point Likert scale from “very comfortable”=1 to “very uncomfortable”=5).

At the end of the visit, each screened man was given a revised Conflict Tactics Scale questionnaire (CTS2) to take home, complete, and mail back (using a stamped, addressed envelope provided to him).20 As the questionnaires had no identifiers, they were anonymous and could not be linked to provider-obtained data.

CTS2 questions assessed “minor” and “major” categories of both psychological aggression and physical assault. Acts of minor versus major psychological aggression and physical assault varied from “insulted or swore” to “threatened to hit or throw something,” and from “grabbed” to “used a knife or gun,” respectively. Men who had committed—in the past year—1 or more minor-category actions more than 5 times, 3 or more minor-category actions 3 to 5 times, or 1 major-category action at least once were classified as currently perpetrating psychological and/or physical IPV; the same criteria were used to classify victimization.

Statistical Analyses

Between-group comparisons were made using Chi square for categorical data and t tests for continuous data. An alpha-level of 0.05 was used for all analyses. Data were managed and analyzed using SPSS 16.0 for Mac ©SPSS Inc., 1989–2007. Given the small sample size and the likelihood that some expected cell values were less than 5, recomputations of categorical data comparisons were done and are reported using the Likelihood Ratio approximation to an exact test from JMP 7.0 ©SAS Institute Inc., 2007.

RESULTS

Participants

Providers screened 128 male patients (numbers per provider noted earlier). All patients who were asked to be in the study participated. The average age of participants was 52 years old (range, 18–84; SD = 13.6). Eighty-six (68%) men had an intimate partner; most of these (66%) were married or similarly committed, and 69% currently lived with the partner.

Screening

Of the 125 and 126 who provided answers to providers’ screening about perpetration and victimization, respectively, 2 (2%) men reported current perpetration and 3 (2%) current victimization. Many more—23 (18%) and 30 (24%)—reported past perpetration and past victimization, respectively. Table 1 collapses these current and past histories into 1 of 4 categories of any perpetration and/or victimization: “Neither Victim Nor Perpetrator” (n = 85); “Victim But Not Perpetrator” (n = 17); “Victim And Perpetrator” (n = 10); and “Not Victim But Perpetrator” (n = 11).

Table 1 Perpetration and Victimization Histories by Patient, Site, and Provider Characteristics
Full size table

As seen in Table 1, there were no differences in age across these subgroups (even when analyses were nested by provider). There were, however, differences in intimate partner status across the subgroups (although these differences were not apparent when analyses were nested by provider). Most of the subgroup differences for intimate partner status (in non-nested analyses) appeared to be across the subgroups other than neither-a-victim-nor-a-perpetrator; those without a partner were more likely to report being a victim-but-not-a-perpetrator or not-a-victim-but-a-perpetrator, whereas those with a partner were more likely to report being both-a-victim-and-a-perpetrator.

There were no differences in provider sex across subgroups. There were, however, differences in clinical site and provider training across the subgroups. (Analyses nested by provider were not possible for the variables of clinical site, provider sex, and provider training as these variables overlapped substantially with the provider variable.) Clinical site and provider training differences indicated that the public health clinic and MDs had the most participants who reported being neither a victim nor a perpetrator. The FQHC and the DO/PA had the most participants who reported not being a victim but being a perpetrator.

Given that the literature suggests age and intimate partner status may be associated with IPV, we explored whether there were differences in age and in intimate partner status by clinical site, provider sex, or provider training that might explain the findings of these latter 3 variables. There were no differences in age across the subgroups of these 3 variables (p = .84, 0.80, and 0.85, respectively). There were differences in intimate partner status across the subgroups of both clinical site (p = .04) and provider training (p = .007). The university clinic and public health sites had ≥50% of participants with an intimate partner, and the VA clinic and FQHC site had <50% of participants with an intimate partner; similarly, the MDs and PA had ≥50% of participants with an intimate partner, and the DO and NP had <50% of participants with an intimate partner.

As a result of the intimate partner differences, we stratified clinic-type analyses by intimate partner status and found that clinic type was only associated with victimization/perpetration for those with an intimate partner (p = .001), with specific findings suggesting that a substantial number (50%) of VA clinic participants with intimate partners were victims not perpetrators of IPV, and many (33%) VA and university clinic participants with intimate partners were both victims and perpetrators of IPV. After stratifying provider-training analyses by intimate partner status, we found that provider training was only associated with victimization/perpetration for those with an intimate partner (p = .004) as well, with these findings being driven by differences for the NP (who practiced in the VA, which led to differences across groups being similar to that described above for clinic type).

Although there were no differences in perpetration/victimization identified by male and female providers when analyzed together, we did find that there were differences in subgroups identified according to provider sex when analyzing MDs versus non-MDs (p = 0.04 and p = 0.07, respectively). The male MD provider identified perpetration and victimization in 27% and 45% of participants, respectively, versus 10% and 10%, respectively, for the female MD providers. The male non-MD provider identified perpetration and victimization in 20% and 5% of participants, respectively, versus 20% and 36%, respectively, for the female non-MD providers.

Providers assessed lethality/safety issues in 14 (58%) of those reporting a perpetration history (including both men reporting current perpetration), responded with direct counseling to 15 (63%; including both men reporting current perpetration), and referred 4 (17%) for services specifically related to the screening (including 1 of the 2 men reporting current perpetration). Providers assessed lethality/safety issues in 22 (67%) of those reporting a victimization history (including the 3 men reporting current victimization), responded with direct counseling after this assessment to 22 (67%; including the 3 men reporting current victimization), and referred 5 (15%) for services specifically related to the screening (including 2 of the 3 men reporting current victimization).

Providers indicated high levels of patients’ and their own comfort with screening—with mean values of 1.42 (SD = 0.74) and 1.40 (SD = 0.72), respectively. The comfort level did not differ by perpetration or victimization status of patients (all p values >0.24).

Questionnaire Assessment

Forty-six (36%) men returned completed CTS2 questionnaires. Thirty-seven and 45 reported enough details to assess rate of psychological aggression of and by a partner in the past year, respectively. Twenty-six (70%) indicated psychological aggression of a partner, and 29 (64%) by a partner. Forty-four reported enough details to assess rate of physical assault of and by a partner in the past year. Nine (21%) indicated physical assault of a partner and 15 (34%) by a partner.

DISCUSSION

This study is the first to evaluate the effectiveness of health care providers verbally screening male patients for current male IPV perpetration as part of a routine clinical encounter. Despite providers’ assessments that the screening was comfortable both for their patients and themselves, only a few patients verbally reported current IPV perpetration. A much larger number—1 in 5—verbally reported past IPV perpetration. Providers appeared to adequately assess lethality and safety issues, as well as to respond with appropriate counseling to patients after this assessment. Referral for specialty services may have been less adequate.

The low rate of identifying current IPV perpetration in face-to-face screening was highlighted by the number of men who subsequently reported current IPV perpetration using a validated questionnaire. This is highlighted by the fact that although only one third of the screened men returned the anonymous questionnaire (clearly an extremely low response rate that limits use of this data to estimate of the rate of perpetration for the sample), nearly three quarters and one quarter of the men who did return it indicated perpetration of IPV-level psychological aggression and physical assault, respectively. The importance of this data is that, despite the low response rate, this approach still led to the identification of 7 more men who were perpetrating physical violence currently and at least 24 more men who were perpetrating psychological violence currently—when compared to face-to-face reports of perpetration. These findings suggest that more perpetrators may be identifiable through a questionnaire-based approach than through face-to-face screening, although it remains to be seen whether these levels of identification would continue if questionnaire-based information were identifiable. Findings also may indicate that written questions describing specific acts of violence may be easier for men to understand and to answer than a general question about “using force” or “hurting.”

Disclosure of past perpetration in face-to-face interactions also was more common. This willingness by men to admit—in person—to past perpetration may indicate an avenue for engaging men in the area of current IPV perpetration through structured discussions that steer toward prevention of violence in a patient’s current relationship(s).

Findings on the men’s verbal reporting of IPV victimization mirrored those described for perpetration: verbal reports of current victimization were infrequent; verbal reports of past victimization were more frequent; and anonymous reports using a validated questionnaire were most frequent (albeit with the aforementioned caveat about the low response rate). Rates of victimization in this primary care sample were similar to those reported previously by men in emergency departments.21,22 Given research that indicates that victimization histories in men are linked to poor health outcomes, these findings indicate another area of opportunity for intervention.23

Given the small numbers of participants in subgroup analyses, results other than those reported above—which are the only ones that specifically focus on and can inform the primary goals of this study, assessing the feasibility and efficiency of primary care provider screening—cannot be used to claim with certainty that clinical site, provider sex, and provider training differences seen across subgroups might be indicative of where, how, and by whom any subsequent studies of primary care screening should be completed. These variables overlapped one another to a great degree, such that differences seen for the nurse practitioner (NP) could just as readily be explained by the fact that the NP was the only provider to have screened patients at a Veterans Administration (VA) site—and thus, differences may have been a result of the VA population not the provider being an NP. What was evident, however, is that victimization was not uncommon in any of the sites, suggesting that screening for victimization may be just as important to do as screening for perpetration.

Additional limitations existed for this study. The study convenience-sampled a small number of men seen by a few health care providers who were IPV experts and located only in urban primary care practices. As providers were able, they were to approach all men on their schedule. Whether to approach the patient was at the discretion of the provider and the deciding factor often was time. Given that the study protocol was comparatively time-consuming for providers in a busy practice, some male patients were skipped; as no data were collected on men not recruited for study participation, we were unable to compute the proportion of male patients not recruited and are unable to characterize what biases other than time concerns may have affected provider choices to screen any patient. Although our results have some issues with generalizability and potential bias, then, issues made more challenging by the small amount of information gathered about participants, we still believe the data remain helpful. These data likely are indicative of the best results one might anticipate from an imposed IPV primary prevention mechanism that employs direct screening of male primary care patients by routine providers in the community.

As noted above, the rate of returning CTS2 questionnaires was extremely low. Thus, the rate of perpetration and victimization reported from these returned questionnaires should not be considered an accurate estimate for the entire sample. Rather, the only point to be drawn from these returned questionnaires is that even with so few questionnaires returned we were able to “identify” more men who were currently perpetrating violence than we had identified in face-to-face screening. These findings suggest that more perpetrators likely can be identified through a questionnaire-based approach than through face-to-face screening, although it does remain to be seen whether the increased rates of identification would continue if questionnaire-based information was actually identifiable.

Despite these limitations, we suspect that use of this group of screeners biased our findings in the direction of higher identification of perpetration. Given the screeners’ low rate of identifying current IPV perpetration, we anticipate the rate would be even lower if less experienced primary care providers did the screening. We also do not suspect that a different population of men or a higher number of patients screened would have increased our identification rate.

This study’s findings indicate, then, that men appear unwilling to disclose information about current IPV perpetration and victimization in a face-to-face interaction with their primary care provider. They may be willing to provide more information if approaches other than face-to-face screenings are employed. Data from Rhodes et al. suggest that one of these other types of approaches may be computer-based screening.24 We recommend that future studies of IPV screening in men use creative approaches that provide alternatives to face-to-face disclosures but still provide identifiable information. These alternative approaches should not depend on patients mailing in the questionnaire after they have left the primary care site and they likely will need to be preceded by an assurance that whatever approach used is intended to provide help to men who may need help and is not intended to criminalize men or otherwise stigmatize them. Given the known and likely health risks to men who are involved in IPV, whether as perpetrators or victims, this study’s findings clearly suggest that future studies are warranted for evaluating different approaches to screening men for involvement in IPV.23