Education and knowledge are seen as a key to the obliteration of social ills. This belief is evidenced by the encouragement of social work and medicine to teach content on domestic violence (DV) in professional schools (LCME, 2000; NASW, 2000). Additionally in most states, continuing education units (CEUs) are required for maintenance of licensing (American Association of Social Work Boards, 2003). The extent to which (DV) education influences screening behavior among medical social workers, family practitioners (FPs) and obstetrician–gynecologists (OB–GYNs) is the focus of this study.

Statement of the problem

The documented effects of DV on the well-being of women have prompted various U.S. Surgeon Generals to identify violence in American as a “public health emergency” (Koop & Lundberg, 1992). Numerous empirical studies have demonstrated that battered heterosexual women have more physical (i.e., trauma and somatic diagnoses) and mental health diagnoses than nonbattered women. Mental health problems associated with battering include depression, attempted suicide, substance abuse; sleep problems, and anxiety (Campbell, Kub, & Rose, 1996; Hathaway et al., 2000; McCauley et al., 1995; Plichta, 1996; Plichta & Weisman, 1995). Common physical problems associated with battery include: gynecological problems; gastrointestinal complaints, shortness of breath, chest pain, and breast pain (Drossman, Talley, Leserman, Olden, & Barreiro, 1995; Gupta, Wells, O’Connor, & Horwitz, 1998; McCauley et al., 1995). Additionally, abuse during pregnancy has been associated with miscarriage, pregnancy termination, neonatal death, low birth weight, low maternal weight gain, infection, anemia, smoking, and use of alcohol or drugs (Parker, McFarlane, & Soeken, 1994; Webster, Chandler, & Battistutta, 1996).

These various symptoms should peak the interest of health care professionals, especially medical social workers and physicians who are in key positions for DV screening and intervention. Early detection and intervention could serve to reduce disease, injury, and death related to DV. Koziol-McLain, Coates, Lowenstein (2001) demonstrated that screening predicts future violence. They found that women who responded positively on a three-question screen for DV were 46.5 times more likely than the women who screened negatively to experience severe physical violence four months subsequent to the administration screen. Hence, screening is an effective method to detect DV.

It is important to recognize that improper DV screening may put battered women in danger and women who leave their partner are at increased danger. It is critical to ask about DV in private, that is, no friends, relatives, and preferably no children over 2-years-old; use professional interpreters when needed. Also, it has been well documented that the most dangerous time for a battered women is within 1 year of leaving (Ganley, 1996; Koziol-McLain, Coates, & Lowenstein, 2001). Even so, the Council on Ethical and Judicial Affairs (1992) suggests that intervening in DV is in the patient's best interest because only treating injuries and symptoms do not address the root of the patient's health problems, and both DV and the symptoms associated with it will continue and escalate.

Healthcare professionals clearly have the ability to play a significant role in the DV crisis by simply screening for it and offering help. Yet, this may not be happening. This lack of screening behavior appears to be related to barriers that prevent healthcare professionals from reaching out to potential victims. The link between these barriers and the screening behaviors of physicians and medical social workers is the focus of this study.

Literature review

The existence of screening barriers has been examined and reported in the professional literature. However, much of the research has focused on physicians. To lay the groundwork for the conceptual framework, it is important to briefly review selective research studies that informed and continue to inform this area of research. The Governor's Task Force (1994) surveyed six groups of licensed health professionals in Florida (n=2702)—social workers, mental health counselors, marriage and family counselors, psychologists, physicians, and nurses. A brief survey of eight questions asked professionals about their exposure to DV education, screening behaviors, and treatment for DV. Findings were reported for the entire sample; the physician group was often contrasted with the sample: (a) 30% of the sample and 16% of physicians had been exposed to DV education during their professional education; (b) 50% of social workers and 27% of physicians had taken CEUs in DV; (c) 85% of the sample and 63% of physicians indicated that they see victims of DV in their practice; and (d) 53% of the sample and 17% of physicians routinely screened for DV.

Snugg and Inui (1992) conducted an ethnographic study to determine barriers that prevented 38 HMO primary care physicians from intervening in DV cases. Barriers included: lack of time, lack of training, fear of offending the patient, powerlessness in helping, no control over patient behavior, and close identification with the victim. Barriers identified from this research were used as a basis to develop quantitative surveys reported in (Parsons, Zaccaro, Wells, & Stovall, 1995; Rodriguez, Bauer, McLoughlin, & Grumbach, 1999; Sugg, Thompson, Thompson, Maiuro, & Rivara, 1999).

Parsons et al. (1995) identified barriers OB–GYNs had in screening for DV and the effect of training on screening behavior. Only 18% (n=168) of physicians routinely screen, 58% (n=542) selectively screen, and 12% (n=115) do not screen. Gender (p=.004) and age (p=.002) were associated with screening behavior; that is, women and younger physicians were more likely to screen than men and older physicians. Rodriguez et al. (1999) surveyed California primary care physicians (n=400), reporting that while 79% of primary care providers screened injured patients, only 10% routinely screened new patients for DV, 9% screened during periodic checkups, and 15% screened during prenatal checkups. Furthermore, obstetrician–gynecologists (OB–GYNs) (17%) were more likely than internists (6%) to routinely screen for DV.

Sugg et al. (1999) assessed the knowledge, attitudes, beliefs and screening behaviors of primary care provider teams (physicians, physician assistants, nurses, and medical assistants) in five primary care clinics (n=206). Providers seldom or never asked about DV for injuries 45.2 %(n=38), depression or anxiety 60.5% (n=52), chronic pelvic pain 61.8% (n=47), headache 73.3% (n=63), and irritable bowel syndrome 81.9% (n=68). Clinicians reported that inquiring about DV was not an invasion of privacy (85%) and were not concerned about offending patients with screening (65%). Garimella et al. (2000) used the same instrument to survey emergency medicine, family practice, obstetrics–gynecology, and psychiatry physicians at a large, urban hospital (n=76). Ninety-seven percent of the physicians believed assessing DV was part of their role; however, 30% hold victim-blaming attitudes toward victims and 70% believe they have insufficient resources to assist.

Gremillion and Kanof (1996) conceptualized four categories of screening barriers. These barriers are identified as: (a) societal and cultural barriers—implicit and explicit social norms regarding gender, inequalities in relationships, societal tolerance of violence, and desensitization through exposure to social issues; (b) personal barriers—prejudicial attitudes including, class attitudes and behaviors, racism, sexism, ageism, homophobia, and clinician's problematic identification with patients; (c) institutional and legal barriers—lack of time, perceived powerlessness to help, and marginalization by colleagues and institutions; and (d) professional barriers— lack of clarity about the proper role of medicine in DV, patient–physician relationship, and education. These conceptual categories of barriers to screening for DV were used in formulating this research.

Societal and cultural barriers

Societal characteristics influence the perception of DV by health professionals. Sassetti (1993) suggests that “…battering is an expression of firmly entrenched societal beliefs and attitudes about the roles of women, men, and violence in our culture (p. 292). Societal beliefs hold the importance of preserving the family, without regard to the costs of the individual members. Battered women report that (a) traditional beliefs (e.g., violence is a spouse's right and adherence to gender roles), and (b) fear of reaction from friends, family, or medical professionals, are barriers to disclosure of abuse (McCauley, Yurk, Jenckes, & Ford, 1998; Pinn & Chunko, 1997; Rodriguez et al., 1999).

Violence within the family may be framed as a private event. Violence, as a private event isolates and silences the victims. Battered women reported fear of further retaliatory or escalating violence as a barrier to disclosing DV (Gerbert, Johnston, Caspers, Bleecker, Woods, & Rosenbaum, 1996; McCauley, Yurk, Jenckes, & Ford, 1998; Pinn & Chunko, 1997; Rodriguez, Quiroga, & Bauer, 1996). DV screening is often foregone, citing the private nature of violence in the home (McGrath, Betacchi, Duffy, Peipert, Becker, & St Angelo, 1997). However, it should be noted that physicians physically and verbally probe into other personal areas (Gremillion & Kanof, 1996; Jecker, 1993). Burge (1989) suggests that the privacy ideology may be overcome by reframing DV as a public or criminal behavior.

Societal expectations of what a “true victim” is, and how she ought to behave, influence healthcare responses toward battered women. In four large metropolitan hospitals, Kurtz (1987), observed interactions between battered women and staff, conducted informal interviews of staff, and reviewed medical records, finding that emergency room staff report being more responsive to battered women who have pleasant personalities, are polite, have no discrediting or stigmatizing attributes (e.g., alcohol on breathe, using drugs, or acting “crazy”), are in immediate danger, and intend to take action to leave the relationship. Victims are often viewed as deserving, provoking, or desiring their victimization because it is difficult for health professionals to believe that “normal” men engage in violence (Sassetti, 1993). Health care providers may perceive battered women responsible for their abuse because they do not readily accept help, do not leave the relationship, or even provoke it (Kurtz, 1987).

Battered women also report their low level of readiness to change the abusive relationship, as a barrier to disclosure (McCauley et al., 1998). From a random sample of 3676 women's medical records in an emergency department, Kurtz and Stark (1990) reported that 86% of battered women were labeled, for example, “hysteric,” “neurotic female,” or “crock,” in medical records compared to 4% of nonbattered women. Battered women are a source of frustration or seen as “difficult” because they may appear hostile, dependent, helpless, or noncompliant (Kurtz, 1987; Sassetti, 1993). Burge (1989) suggests that these behaviors may be survival strategies. Kurtz (1987) found that healthcare providers who positively responded to battered women focused on making a referral and viewed helping as “honorable work,” rather than focusing on a patients’ demeanor.

Personal barriers

Personal attitudes and experiences may impede social workers and physicians from screening for DV. Clinicians’ differences and prejudicial attitudes including, class attitudes and behaviors, racism, sexism, ageism, homophobia, and clinician's problematic identification with patients, are also barriers to screening for DV. Beliefs that abuse is a problem restricted to people in poverty; abuse is not an issue in same sex relationships or higher educated women; and abuse among older people is not a priority, may affect who is screened for abuse (Council on Ethical and Judicial Affairs, 1992; Durant, Gilbert, Slatzman, Johnson, & PRAMS Working Group, 2000). Rodriguez et al. (1999) surveyed California primary care physicians (n=400), reporting that cultural and language differences were significant barriers. Middle-class European American physicians with no experience of DV may assume patients with similar backgrounds also do not experience abuse.

Personal experiences may also affect attitudes toward screening. Researchers have found high rates of family violence among medical students (Cullinane, Alpert, & Freund, 1997; DeLahunta & Tulsky, 1996). Cullinane et al. (1997), for example, reported that 38% (n=139) of first-year medical students reported personal histories of abuse: 13% reported child abuse and 32% reported DV or sexual assault. Further, Cullinane et al., found that students with a history of abuse compared to students with no history of abuse were more likely to endorse the need for DV education for physicians (9.6±2.3 versus 9.0±1.9, p < .05) and advocacy by physicians (23±5.2, p < .001). Alternately, Sugg and Inui (1992) suggest that discovery of abuse by women physicians may evoke feelings of vulnerability or loss of control, which may negatively affect screening for DV. The effect of personal experience with DV on health care professionals’ screening behavior is unclear.

Health professionals’ concerns for their safety may impact attitudes toward screening. In the past decade, violence has been increasing against helping professionals (Rey, 1996). Yet, violence often goes unreported by professionals (Macdonald & Sirotich, 2001). Although the number of physicians accosted by patients is relatively small, the number of physicians accosted by patients’ family members is even smaller; moreover, studies do not indicate whether family members who perpetrate violence are batterers (Hobbs & Keane, 1996; Paola, Malik, & Qureshi, 1994). Because concern appears to be increasing among helping professionals and their professional associations (Grant, 1995) safety concerns may be a significant barrier to screening for DV.

Institutional and legal barriers

Institutional and legal barriers deter DV screening. Institutional barriers include: lack of time, perceived powerlessness to help, and marginalization by colleagues and institutions (Cohen, DeVos, & Newberger, 1997; Sugg & Inui, 1992). Cohen et al. (1997) conducted a qualitative study of barriers to identifying, treating, and referring victims of DV by physicians and other healthcare workers (e.g., nurses, sociologists, and mental health professionals), in five diverse healthcare systems (i.e., Atlanta, Duluth, Providence, Riverside, and Roswell).

Health care settings are reportedly busy by both physicians and patients. Health care providers report lack of time as a significant barrier to responding to battered women (Ferris, 1994; Gerbert, Caspers, Brownstone, Moe, & Abercrombie, 1999; Kurtz, 1987; McGrath, et al., 1997). For example, Ferris (1994) surveyed a Canadian national sample of 963 family physicians and general practitioners about knowledge, attitudes, treatment, and continuing medical education in DV. Battered women reported that providers’ appearance of being busy or rushed was a barrier to disclosure (Gerbert et al., 1996; McCauley et al., 1998).

Health care providers reported that due to the complexity of DV they did not have the tools they needed to help, including sufficient community resources (Gerbert et al., 1996; Kurtz, 1987). Surveying a convenience sample of 207 staff (e.g., ED and OB–GYN physicians; ED nurses, and social workers) about perceived barriers, training, protocols, and screening behaviors for DV Mcgrath et al. (1997) reported that lack of 24-h social service support in an emergency department was a barrier to screening. In addition, providers felt that their interventions were useless if patients did not leave the abusive relationship (McGrath et al., 1997). Providers perceived that they were not effective in producing “results” (Kurtz, 1987). In other words, physicians were discouraged because they were not able to control patients’ behaviors and patients were not able to control their situations.

Providers who redefined how they viewed their ability to help had less of a sense of powerlessness, for example, redefining success to mean compassionately asking about abuse rather than a positive affirmation of abuse (Gerbert et al., 1999). Rodriguez et al. (1996) reported that abused women believed that healthcare providers should repeatedly offer assistance, understanding that change takes time. From a case-control study of 202 abused women and 240 randomly selected non-abused women in an HMO, Gielen et al. (2000) reported that 95.6% of abused and non-abused women “would be glad someone took an interest” by screening and 86.1% believed “it would be easier for abused women to get help.”

Further, resources, for example, shelters and use of hospital admittance, are limited (Gremillion & Kanof, 1996). Legal barriers include possible civil action by the abuser or abused or time commitment connected to being a witness. Battered women report legal barriers, including, possible police involvement, immigration or legal status, and discrimination by insurance providers (Pinn & Chunko, 1997; Rodriguez, Quiroga, & Bauer, 1996).

Professional barriers

Professional barriers include the proper role of medicine in DV, the patient–physician relationship, and education. The perceived role of medicine in DV interferes with healthcare professionals screening behavior for DV. Fletcher (1994) suggests that DV is not a medical problem, but a social problem derived from the decaying morals of society. Moreover, he argues, DV has no medical treatments; however, the resulting injuries are treatable. Others argue that because DV has medical consequences, it is a medical problem (Fullin & Cosgrove, 1992). Filtcraft (1992) has remarked, “Violence epidemiology is in its infancy” (p. 3194). In brief, providers who view addressing abuse as within their professional role are more likely to screen patients for DV (Kurtz, 1987).

Treating only the injuries and symptoms of abuse does not address the ongoing violence that is at the root of its victims’ health problems (Council on Ethical and Judicial Affairs, 1992). Filtcraft (1992) suggests “… violence is more than the sum total of injuries, and optimal clinical intervention must address the social context in which violence occurs, not simply the pattern or severity of the resulting injuries” (p. 3194). Clinical interventions, for injuries and especially somatic complaints, are often limited to the biological causes. Warshaw (1989) reported that physicians and nurses failed to connect injury with how it happened, the relationship of the perpetrator to the victim, the impact of the violence to the victim, and implications and meanings for the victim outside of the emergency department. For nontraumatic problems, physicians do not ask about abuse because they do not make the connection of violence to nontraumatic problems (Sassetti, 1993). The result is extensive medical examinations to identify organic causes or medications to alleviate pain, anxiety, and other symptoms.

Misunderstanding within patient–physician relationship has also created barriers for DV screening. Although Sugg and Inui (1992) have reported physicians’ fear of offending patients for inquiring about DV, other studies indicate that patients favor inquiries about abuse and believe physicians can help (Friedman, Samet, Roberts, Hudlin, & Hans, 1992; Hamberger, Ambuel, Marbella, & Donze, 1998). From the Commonwealth Fund Survey of Women's Health, a telephone survey of a geographically stratified sample of 2525 women and 1000 men ages 18 and over, Plichta (1996) reported that abused women reported poorer quality communication with their physicians than non-abused women. Abused women reported more difficultly talking to a doctor (34.2 versus 15.5%), physicians not listening well (17.1 versus 5.4%), being told that the problem was “all in their head” (28.3 versus 20.3%), and sexual harassment by physicians (12.7 versus 6.2%) than nonbattered women.

Patients’ sense of guilt, shame, fear of their batterers’ retribution, or fear of physicians’ reactions preclude women from spontaneously reporting that they are battered (Chescheir, 1996; Sassetti, 1993). Patients’ “unresponsiveness to questions” or “lack of disclosure” have been reported by physicians as barriers to screening (Brown, Lent, & Sas, 1993; Ferris, 1994; Ferris & Tudiver, 1992; Kurtz, 1987; Rodriguez et al., 1999). Such responses indicate a significant disjuncture between patients’ and doctors’ views of their experiences surrounding inquiries into DV. Researchers have also reported communication skills that increase and decrease the likelihood of abuse disclosure. Battered women reported physicians’ lack of empathy, discomfort with DV, or appearing not to listen as barriers to disclosing DV (McCauley et al., 1998). Patients are more likely to disclose abuse if the physician demonstrates confidentiality (e.g., the clinician directs, rather than asks the partner to leave the room); is direct in asking about abuse, but not interrogative; and is supportive, compassionate, and nonjudgmental (Hamberger et al., 1998; Rodriguez et al., 1996).

Burge (1989) suggests that DV falls outside the scope of biological systems and, therefore, does not correspond with the medical educational system. Because physicians are not specifically trained to deal with psychological trauma, they rely on their own capacities to address painful and potentially overwhelming issues, including DV (Warshaw, 1996). Professional socialization may mask capacities they already have. Lack of training, knowledge, or education have been identified by healthcare clinicians as barriers to identifying, treating and referring victims of DV (Cohen et al., 1997; Ferris, 1994). From a sample of 962 obstetrician–gynecologists, Parsons et al. (1995) report that 49% (n=429) of physicians felt inadequate in dealing with abuse due to lack of training. Inadequacy due to insufficient training was found to be significantly correlated with lower incidence of screening for abuse. Cohen et al. concluded that “…knowledge and especially the attitudes and skills of practicing physicians constituted a major impediment to the identification and treatment of victims of family violence” (p. S23). In particular, myths and assumptions about DV that interact with prejudicial attitudes about race, class, sex, and sexual orientation, impact care (e.g., family violence is confined to poor people; it is not a concern for homosexual people). In another study, physicians, physician assistants, nurses, and medical assistants identified a lack of confidence in their ability to ask about abuse and make appropriate referrals (Sugg et al., 1999). Even when compared to other health promotion practices, physicians reported feeling unprepared to counsel patients about DV (Williams, Chinnis, & Gutman, 2000).

The socialization process of social workers likely does not serve as a barrier to identifying battered women, as social work education is grounded in helping with psychosocial issues. Social workers are trained to assess the person-in-environment and DV is such a component. Social workers’ professional barriers, their attitudes, and perceptions that affect the extent of their screening for DV are unknown.

It is well documented that battered women present themselves for care in the medical arena; yet, medical social workers, FPs, and OB–GYNs appear to be missing opportunities to identify such women and initiate the helping process, something these women desire. Several barriers to screening for DV have been identified in the literature. Education is a critical barrier, and if addressed has the ability to reduce other barriers to screening. Understanding the connection between barriers, and medical social workers and physicians’ perceptions of their education about DV, is important to increasing the screening behavior of these professionals.

Method

Sampling

All social workers listed in Florida NASW's database who: (a) held an MSW, (b) identified their work setting as “health outpatient” and “health inpatient,” and (c) identified “clinical/direct practice” as their major function were asked to participate in the research. Members meeting these criteria included: health outpatient and health inpatient. Board certified, medical doctors specializing in FP and OB–GYN, listed in The Official American Board of Medical Specialists (ABMS) Directory of Board Certified Medical Specialists who had complete and current information and were listed as having full-time clinical interaction with patients, were invited to participate.

A package consisting of an insert, cover letter, survey, and self-addressed and stamped envelope was mailed to the participants. The package mailed to physicians contained an additional letter from, a medical doctor, encouraging physicians to participate. Return envelopes were addressed to and from the researcher, ensuring anonymity of the participant. Three weeks after the surveys were mailed, all participants were mailed a follow-up post card thanking or reminding them to participate.

One thousand seven hundred and seventy-three clinicians had an opportunity to respond (508 social workers, 676 FPs, and 589 OB–GYNs). The overall response rate was 24%: 37% for social workers, 19% for family practitioners, and 14% for OB–GYNs. Of the 432 responses, 44 surveys were excluded from analysis (e.g., incomplete). Physician response rates for mail surveys are generally very low. Studies in this area report between 15% (n=962) (Parsons, Zaccaro, Wells, & Stovall, 1995) to 61% (n=963) (Ferris, 1994). The high rate for the latter may be explained by the costly and time-consuming data collection method that included mailing, reminder letter, second mailing, and phone calls to nonrespondents. Furthermore, this study consisted of Canadian physicians, whom may differ in a way that may influence their response rate from U.S. physicians.

Measures

The instrument, Providers’ Knowledge, Attitudes, and Beliefs about Domestic Violence Scale (KAB), developed by the Group Health Cooperative and Harborview Injury Prevention and Research Center (1997), was utilized. The KAB measures screening barriers and screening behavior. It was chosen because it is the only multidimensional, psychometrically tested scale specific to health care providers. Both the validity and reliability of the instrument have been explored with promising results (Maiuro, Vitaliano, Sugg, Thompson, Rivara, & Thompson, 2000; Thompson, Rivara, Thompson, Barlow, Sugg, Maiuro, et al., 2000).

Maiuro et al. (2000) established content validity, whether a measure reflects a specific content domain, in the development of the initial items for the scale. Derived from the literature, they used a conceptual framework consisting of processes (knowledge, attitudes, and behavior) and content (system, provider, victim, and batterer) about DV to develop 104 initial items with eight domains. Two independent pilot tests (n=129 and n=246) of physicians, nursing staff, physician assistants, and intake and clinical support staff were conducted. These data were analyzed using factor analysis (Principal Component Analysis with oblique rotations) until six distinct subscales, consisting of 4–8 items, with each domain explaining unique and incremental variance.

Thompson et al. (2000) documented the scale's convergent validity, whether a measure corresponds to the results of other methods for measuring the same construct. A group-randomized controlled trial in five HMO clinics was conducted. Participants included physicians, physician assistants, nurse practitioners, nurses, medical assistants, and receptionists (Thompson et al., 1998). An intensive 1-year intervention consisted of two half-day training sessions; extra training for designated leaders; bi-monthly newsletters; four clinic education rounds; posters, cue cards, and questionnaires; and feedback of results. Improvements were reported for the intervention clinics on both the documentation of screening in the chart review (n=3795) and four of the six KAB domains (n=128).

For the data presented herein, the overall scale Cronbach alpha=.9567. The subscale alphas were as follows: Perceived Self-Efficacy (α=.8290), Perceived System Support (α=.7955),Blaming the Victim (α=.7697), Fear of Offending (α=.8349), Safety Concerns (α=.8192), and the original Frequency of Inquiry (α=.9165).

Domestic violence education

Education was conceptualized to include formalized DV content in the context of course work in an MSW program, CEUs on DV post MSW, agency-sponsored in-service training on DV, and other additional training hours. Although education is a component of professional barriers, as conceptualized by Gremillion and Kanof (1996), education is measured as separate variables in this study, as the scale and its subscales do not measure education. In other words, education is conceptualized as a potential influencer of screening barriers, and lack of education is conceptualized as an additional barrier to screening.

Screening barriers

Five of the six KAB subscales, measured screening barriers. The “Blaming the Victim” Subscale included questions about provider's perception that victims cause their abuse (societal and cultural barriers). The “Safety Concerns” Subscale included questions about the provider's safety concerns for him or herself or for the victim (personal barriers). The “Fear of Offending [the patient]” Subscale included questions about the provider's perception of appropriateness of screening for DV within the professional role (professional barriers). The items in these subscales are reverse-scored so all scales point in the same direction. Institutional and legal barriers were measured using the following two subscales. The “Perceived Self-Efficacy” Subscale included questions about the provider's perceived ability to help victims or batterers; the “Perceived System Support” Subscale solicited responses about accessibility and helpfulness of social services for DV patients (institutional and legal barriers).

Table 1  
Full size table
Table 2 Frequencies of health care professionals screening behaviors
Full size table

Screening behavior

The “Frequency of Domestic Violence Inquiry” KAB Subscale measured self-reported screening behavior. It asked social workers if they screened “at risk” clients or when clients presented with certain symptoms, for example, injuries, chronic pelvic pain, and headaches. A list of physical and psychosocial signs of DV was added to more fully measure screening behavior for at risk patients of DV. The list included: unable to make eye contact; withdraws from touch; wears long sleeves, scarf, or a high collar; and wears heavy make-up or sunglasses indoors (Women in Distress, 1997). This modification did not alter the instrument's reliability; the alpha of the Frequency of Inquiry Subscale increased from the original subscale (α=.9165) to the modified subscale (α=9623).

Findings

Participants

The sample consisted of 186 (47.9%) medical social workers; 119 (30.7%) FPs, and 81 (20.9%) OB–GYNs. Study participants were mostly European American (75.0%) and female (52.6%). Social workers were primarily female (82.8%) and physicians were primarily male (76.5% and 72.8%). Please see Table 1.

Forty percent (n=155) of the participants reported no exposure to DV in their professional education. DV content occurred as a separate course for 38.76% (n=150) of the participants with 7.2% (n=28) taking it as an elective course and 27.1% (n=105) taking it as a required course. Less than 4.4% (n=17) identified the course as both an elective and required or did not specify which. Participants reported a mean of 12.48 CEUs/CMEs in DV and 5.93 in-service hours. Less than 16% of participants (n=61) reported that they had obtained additional training in DV; the most common type was conferences or workshops.

Participants reported institutionalized barriers and supports for DV screening. Only 28.5%, (n=109) participants reported that their institution had screening guidelines and 14.4% (n=56) reported that they were unsure if guidelines existed. Checklists or other paperwork reminders to screen for DV in facilities were reported by 66.8% (n=255) of participants.

Despite 78.9% of participants identifying at least one victim of DV in their practice in the past year, only 20.8% nearly always or always routinely screened for DV in their practice in the last three months; 24.0% reported that routine screening did not apply to their role. Please refer to Table 2 for frequencies of selected screening behaviors.

Participants were asked about their perceptions of women at risk of being a DV victim. Participants reported that they strongly agreed or agreed that women of all socio-economic statuses are at risk of DV (95.9%, n=371), women of all races are at risk (92.5%, n=358), and lesbian women are at risk (73.0%, n=279).

Table 3 Means and standard deviations of subscales and graduate education for entire sample using 7 one-way ANOVA’s
Full size table
Table 4 Correlations of barriers and screening behaviors for health care professionals
Full size table

Bivariate analyses

Education

Health professionals who reported exposure to DV content during their professional education perceived fewer barriers—Perceived Self-Efficacy (F=4.769, df=1, p=.030), Perceived System Support (F=6.193, df=1, p=.013), Blaming the Victim (F=12.289, df=1, p=.001), and Fear of Offending (F=10.676, df=1, p=001)—and reported more screening behavior (F=7.316, df=1, p=.007) than participants who reported no exposure to DV content. Please see Table 3.

Moderate bivariate correlations were found between Frequency in Inquiry and the barriers subscales: Perceived Self-Efficacy (r=.525, p < .01), Perceived System Support (r=.396, p < .01), Blaming the Victim (r=.319, p < .01), Fear of Offending (r=.486, p < .01), and Safety Concerns (r=.353, p < .01). Please see Table 4.

Participants with additional training were significantly different from participants who did not report additional training on the six subscales: Frequency of Inquiry (F=23.890, df=1, p=.000), Perceived Self-Efficacy (F=19.227, df=1, p=.000), Perceived System Support (F=13.780, df=1, p=.000), Blaming the Victim (F=8.413, df=1, p=.004), Fear of Offending (F=20.642, df=1, p=.000), and Safety Concerns (F=9.990, df=1, p=.002). Participants who reported additional training reported more screening behaviors and perceived fewer barriers to screening, as indicated by higher mean scores on the subscales than participants who did not report additional training.

Subsamples

A statistically significant difference was found for exposure to DV during professional education between the social work, FP, and OB–GYN groups (χ 2=10.909, df=2, p=.004), with social workers 64.29% (n=117), FPs 62.07% (n=72), and OB–GYNs 43.21% (n=35) reporting exposure to DV content during their professional education. A chi-square showed a difference between specialty and CEUs/CMEs (χ2=42.109, df=2, p=.000). Social workers more often had 11 or more CEUs 41.85% (n=77) than FPs 13.56% (n=16) and OB–GYNs 11.11% (n=9). A chi-square showed a difference between specialty and in-service hours (χ2=73.061, df=4, p=.000). Fewer social workers had 0 in-service hours 27.32% (n=50) than FPs 65.22% (n=75), and OB–GYNs 77.03% (n=57). For specialty and additional training, the chi-square did not meet the assumption of at least five cases per cell.

Table 5 Multiple regression of screening behavior
Full size table

Statistically significant differences were found between social workers, FPs, and OB–GYNs for five barriers subscales and screening behavior subscale: Perceived Self-Efficacy (F=54.940, df=2, p=.000), Perceived System Support (F=21.843, df=2, p=.000), Blaming the Victim (F=12.213, df=2, p=.000), Fear of Offending (F=38.853, df=2, p=.000), Safety Concerns (F=26.383, df=2, p=.000), and Frequency of Inquiry Subscale (F=31.522, df=2, p=.000). For each of the subscales, social work participants perceived fewer barriers toward screening for DV and screened more.

Demographics

Female participants screened more than male participants (F=20.426, df=1, p=.000), and perceived fewer barriers to screening—Perceived Self-Efficacy (F=15.076, df=1, p=.000), Perceived System Support (F=11.509, df=1, p=.001), Blaming the Victim (F=20.655, df=1, p=.000), Fear of Offending (F=44.773, df=1, p=.000), and Safety Concerns (F=7.786, df=1, p=.006)—than male participants on all of the subscales. ANOVA's were run to determine if there were significant differences between barriers to screening and race. No significant differences were found.

ANOVA's were run to determine if there were significant differences between barriers to screening and age. A significant difference was found on the Blaming the Victim Subscale (F=5.544, df=2, p=.004). Participants 44–55 blamed the victims slightly less (X=27.62, SD=3.97) than participants 26–43 (X=27.28, SD=4.06) and participants 56–84 (X=25.88, SD=4.85).

Table 6 Multiple regression of victims identified
Full size table

Institutional supports

Participants who reported that their institution had screening guidelines (28.5%) perceived fewer barriers to screening on the Perceived Self-Efficacy (F=32.978, df=1, p=.000) and Perceived System Support (F=46.350, df=1, p=.000), Blaming the Victim (F=10.436, df=1, p=.001), Fear of Offending (F=12.129, df=1, p=.001), and Safety Concerns (F=7.428, df=1, p=.007) Subscales. In addition, participants who reported that their institution had screening guidelines screened more (F=30.457, df=1, p=.000) and identified more victims in the past year (F=35.620, df=1, p=.000) than participants who reported that their institution did not have screening guidelines or were unsure if guidelines existed.

Participants who reported that their institution had a checklist item or paperwork reminder to screen for DV (66.8%) perceived fewer barriers on the Perceived Self-Efficacy (F=23.769, df=1, p=.000), Perceived System Support (F=37.271, df=1, p=.000), Blaming the Victim (F=5.811, df=1, p=.016), Fear of Offending (F=11.743, df=1, p=.001), and Safety Concerns (F=5.306, p=.022) Subscales. In addition they screened more (F=49.813, df=1, p=.000) and identified more victims of DV in the past year (F=43.790, df=1, p=.000) than participants who reported that their institution did not have a checklist item or paperwork reminder to screen or were unsure if paperwork exists.

Multivariate analyses

Hierarchical linear regressions were performed to see if the bivariate relationships established between the dependent variable and the independent variables could be used to predict a model for the occurrence of the dependent variable. The model discarded variables when the level of significance was p > .05. Five of the independent variables contributed significantly to prediction of screening behaviors for the entire sample: Perceived Self-Efficacy (β=.467) Fear of Offending (β=.232), Safety Concerns (β=−.205), CEUs/CMEs (β=.208), and In-service hours (β=.154). Altogether 50.3% (48.4% adjusted) of the variability in Frequency of Inquiry was predicted by knowing scores on these five independent variables. These findings suggest that the Perceived Self-Efficacy is the strongest predictor of screening behavior with Fear of Offending, Safety Concerns, CEUs/CMEs, and in-service hours with approximately equal contribution to predicting screening behavior. Table 5 displays the multiple regression findings for the sample and sub-samples.

Four of the independent variables contributed significantly to prediction of the number of victims identified for the entire sample: Frequency of Inquiry (β=.415), Blaming the Victim (β=.222), Additional Training (β=.174), and Professional Education (β=.139). Altogether 38.5% (36.7% adjusted) of the variability in Frequency of Inquiry was predicted by knowing scores on these four independent variables. These findings suggest that Frequency of Inquiry is the strongest predictor of the number of victims identified with Blaming the Victim, Additional Training, and Professional Education also contributing to the prediction of the number of victims identified. Table 6 displays the multiple regression findings for the sample and sub-samples.

Discussion

This study found that, for health care professionals, namely, social workers, FPs and OB–GYNs, DV education was associated with fewer barriers to DV screening, fewer barriers to DV screening was associated with more screening behavior, and more screening behavior was associated with more victims identified. This study made important preliminary connections between these areas of study, thereby beginning to fill a gap in the literature.

This study offers more information than provider's barriers to screening, but how much each barrier contributes to behavior. Institutional and legal barriers (i.e., perceived self-efficacy) followed by professional barriers (i.e., fear of offending clients, CEUs and in-service hours), and a personal barrier (i.e., safety concerns), influenced screening behaviors. Screening for DV was the most important predictor of the number of victims identified; followed by blaming the victim, a societal and cultural barrier; and professional education and additional training, professional barriers. Although three of the four categories of barriers—professional, institutional and legal, and personal barriers—predicted screening behavior, the fourth category—societal and cultural barriers—helped predict victims being identified. This study offers some empirical support for the importance of the four categories of barriers in the literature, as classified by (Gremillion & Kanof, 1996).

Moreover, this study provides some evidence that screening behavior is necessary, but not sufficient to identifying victims. In other words, even when professionals screen, it appears that barriers to identification still exist. Barriers to disclosure, which may or may not be the same as barriers to identification, recognized in the literature include: (a) traditional beliefs (e.g., violence is a spouse's right and adherence to gender roles); and (b) fear of reaction from friends, family, or medical professionals (McCauley et al., 1998; Pinn & Chunko, 1997; Rodriguez et al., 1996). Barriers to identification, from the patient's perspective also identified in the literature relate to the provider's disposition—communication of empathy, and respect for the patient's need to make decisions (McCauley et al., 1995; McCauley et al., 1998; Rodriguez et al., 1999).

Limitations of the study

There are several limitations to the study. First, there are limitations in the sampling. The 24% response rate may limit generalizability of the findings. Only those Florida social workers who were NASW members and Florida board certified FPs and OB–GYNs listed in the ABMS with complete and current contact information had an opportunity to participate. The social workers in this study are demographically similar to Florida NASW members, suggesting that this sample may be representative of Florida NASW members. Although not all ABMS members had an opportunity to be sampled, the sample appears to be representative of the population.

Implications

While recognizing the study limitations, the results suggest that education is related to screening barriers; screening barriers increase screening behavior; and screening behavior increases the number of victims identified. Education needs to be integrated throughout professional education, reinforced through CEUs/CMEs, in-service hours, and additional training for screening to become salient to professional's thinking during diagnosis and treatment.

This study showed how influential the institutionalization of DV education is to practice. Florida Statute (456.031), requiring DV education, thereby aligning itself with requirements of other health care agencies and professionals. Hence, the Statute raises DV to a higher priority among multiple stakeholders. Although providers have come a long way since the Governor's Task surveyed Florida health care professionals, there is still a way to go. Further alignment of local communities’ objectives and services are needed. Policy needs to influence professionals so that DV is always at the forefront of their thinking. Florida providers may be learning how to ask about DV, addressing barriers to screening studied here; however, it leaves an important component of institutional barriers out.

It is likely that DV education includes how to sensitively ask about and respond to DV. But, are practitioners and administrators being educated on assessment tools that would combat an important institutional barrier, namely, “lack of time”? Researchers have found that the addition of various tools containing 1–5 questions, increase the detection of abuse in primary care settings (Feldhaus et al., 1997; Freund, Bak, & Blackhall, 1996; McFarlane, Greenberg, Weltge, & Watson, 1995; Norton, Peipert, Zierler, Lima, & Hume, 1995). An additional innovative tool successfully included in a health system's response to DV was displaying brochure holders in bathrooms with help cards and safety plans (Kheder & VandenBosch, 2001). Organizations such as the AMA and the National Health Resource Center on Domestic Violence offer said materials and more at no charge.

These tools need to be offered to practitioners. Such tools are even more important to private practitioners, as private physician offices are not regulated by JACHO, thus not needing to develop guidelines. Education, then, needs to help private and group practitioners develop and institute screening guidelines; include a checklist item for DV; or both, as this study showed that these institutional supports decreased barriers to screening, increased screening behavior and increased the number of victims identified. CEU/CME curricula need to include education and suggestions for creating these supports. Kheder and VandenBost (2001) suggest that these strategies would not have had as great an impact without continued educational efforts.

It is imperative that managers are responsive to the complexity of the needs professionals have in helping victims of DV. Although tools support professional in the inquiry of DV, institutional supports need to make available practical things to offer women. Pease (2001) suggests that at times cab fare, food, personal hygiene materials, and a safe place may be needed. How can institutions begin to provide these supports to professionals who identify DV?

A significant amount of work has been conducted to understand barriers to fostering a response to DV. This work is often limited to a single component of this complicated problem, namely, providers’ barriers to screening. Results are temporary because efforts generally include a single educational session that the research is beginning to demonstrate the effects are temporary. The response to DV does not become integrated in the system at any level.

Because of the complexity of DV, a true successful response is a system wide response, both within local health care systems and the larger community system (Kheder & VandenBosch, 2001). Finally institutional supports and continuing education need to be placed in a community-based response to battering. Such a community-based response would include a task force law enforcement and legal assistance; medical services; shelters; and social and mental health services (Hamlin, 1991).

This study is in need of replication. As the first research to link education, barriers, screening behaviors, and victims identified, at times groups were too homogenous or too small to draw reliable conclusions. In addition, further research is needed to help explain the following unexpected findings. Are barriers to disclosure tantamount to barriers to identification? Research is needed to paint a more complete picture of what the major contributors to identification and disclosure, within the patient-professional interaction. It is not enough to ask each party's perspective. What happens when these perspectives interact within the relationship? Future research may also reveal if distinct groupings across professions exist.

Conclusion

This study has brought together areas that have previously been studied separately. This study suggests that education impacts DV barriers; DV barriers impact screening behavior; and screening behavior impacts victims identified. The study also offered detailed information, including the types of physical and psychosocial symptoms that prompt provider screening. Detail about when and how much DV education in professional and post-professional settings were discussed. Also this detail was offered for the understudied profession of medical social work, as well as FPs and OB–GYNs. This study showed how important institutional and legal barriers are to influencing screening behavior for DV. Florida Statute (456.031) on mandating DV education also appeared influential in affecting practice by raising DV as a concern of priority to multiple stakeholders.