|Home | About | Journals | Submit | Contact Us | Français|
Women who are victims of domestic violence frequently seek care in an emergency department. However, it is challenging to hold sensitive conversations in this environment.
To describe communication about domestic violence between emergency providers and female patients.
Analysis of audiotapes made during a randomized, controlled trial of computerized screening for domestic violence.
2 socioeconomically diverse emergency departments: one urban and academic, the other suburban and community-based.
1281 English-speaking women age 16 to 69 years and 80 providers (30 attending physicians, 46 residents, and 4 nurse practitioners).
871 audiotapes, including 293 that included provider screening for domestic violence, were analyzed. Providers typically asked about domestic violence in a perfunctory manner during the social history. Provider communication behaviors associated with women disclosing abuse included probing (defined as asking ≥1 additional topically related question), providing open-ended opportunities to talk, and being generally responsive to patient clues (any mention of a psychosocial issue). Chart documentation of domestic violence was present in one third of cases.
Nonverbal communication was not examined. Providers were aware that they were being audiotaped and may have tried to perform their best.
Although hectic clinical environments present many obstacles to meaningful discussions about domestic violence, several provider communication behaviors seemed to facilitate patient disclosure of experiences with abuse. Illustrative examples highlight common pitfalls and exemplary practices in screening for abuse and response to disclosures of abuse.
Communication about domestic violence has been studied retrospectively through reports of victims and health care providers. Victims frequently report difficulties in disclosing their experiences with domestic violence to health care providers (1– 6). However, with the exception of participant observation work by Kurz in the 1980s (7), research is lacking on what actually transpires when physicians and patients talk about domestic violence. Direct observation can provide useful examples of successful screening and guide educational programs for providers.
We performed a qualitative analysis of audiotaped domestic violence conversations that occurred between female patients and their emergency providers. The parent study, a randomized, controlled trial of a computer screening intervention, found a significant increase in the frequency of domestic violence discussions associated with the intervention (8). The computer-based health risk assessment included questions about abuse and a prompt to the provider to ask about possible risk. We describe here the actual communication between providers and patients and identify common pitfalls and exemplary practices.
From June 2001 to December 2002, we conducted a randomized, controlled trial of a self-administered computer-based health risk assessment tool, which generated health recommendations for patients and alerted physicians to a variety of potential health risks, including domestic violence. The trial took place at 2 socioeconomically diverse emergency departments: an urban academic medical center that serves a predominately publicly insured, inner-city, African-American population, and a suburban community hospital that serves a predominately privately insured, white population. Inclusion criteria were sequential female patients 18 to 65 years of age who were triaged as medically nonemergent and could give consent. The emergency providers (40 attending physicians, 46 residents, and 4 nurse practitioners) involved in the study were aware that the intent was to increase detection of domestic violence. Before the start of data collection, providers received a 1-hour lecture and a 30-minute video and instruction guide about assessing safety and documenting and providing referrals related to domestic violence (9).
Figure 1 shows the study flow. Successfully audiotaped emergency visits (n = 871) were deidentified, and all audible domestic violence discussions (n = 293) between providers and patients were excerpted for transcription and coding. Patients and providers signed written consent, and the institutional review boards of both institutions approved the study. The study ended at patient discharge; there was no subsequent patient follow-up, but all participating patients received information on domestic violence services in the form of a magnet listing a variety of community resources.
A structured domestic violence coding scheme was developed in an iterative manner through group listening and discussion among the authors (10), who have expertise in physician–patient communication, emergency medicine, and domestic violence and psychology. The coding scheme was loosely based on a theoretical framework that patients often hint at their concerns and present their providers with “potential empathic opportunities” when they discuss psychosocial issues (11–13). The coding scheme was designed to identify the range, scope, and frequencies of common practices, as well as key examples of “best” and “worst” practices. Best practices were assessed on the basis of literature about what domestic violence survivors find to be helpful (9, 14, 15) and the existing medical communication literature about sensitive issues (16). A domestic violence discussion was defined as any mention of physical or emotional abuse during the encounter. Domestic violence disclosure was defined as any patient mention of current or past abuse in an intimate or family relationship.
All domestic violence discussions were coded independently by at least 2 authors, who used the final structured coding form while listening to the audiotape and reading the transcript. This was further refined after independent coding of a 15% random sample until agreement for key domestic violence variables was nearly perfect. All coding was examined for discrepancies, which were further reviewed through group listening and discussed until consensus was achieved.
The study was funded by the Agency for Healthcare Research and Policy, which had no involvement in the collection, analysis, or interpretation of the data or in the decision to submit the manuscript for publication. All authors had full access to the data files for this study.
We obtained 293 audiotapes that included a discussion of domestic violence. Patients were predominantly African American (83%) and single (63%), and 18% had less than a high school education. Only 77 of the 293 domestic violence discussions included patient disclosure of domestic violence to the provider. Table 1 compares the demographic characteristics of enrolled patients with and without useable audiotapes and shows that audiotaped patients who did or did not disclose abuse to the provider were similar. Table 2 shows demographic characteristics of providers. Patients were usually seen by more than 1 provider, which is typical at teaching hospitals.
Victims of domestic violence often seek care in emergency departments.
This study examines 293 audiotaped visits to 2 emergency departments during which providers screened adult women for domestic violence. Providers’ queries were often perfunctory. They usually did not include follow-up probing or offer open-ended opportunities to talk. Seventy-seven women disclosed domestic violence during the interviews; however, providers documented only 24 of the disclosures and referred only 19 women for counseling.
Audiotapes were made during a randomized trial that was testing computer screening for domestic violence. Providers knew they were being audiotaped.
Poor communication with victims of domestic violence is probably very common in emergency departments.
Patient disclosure of abuse usually referred to an intimate partner, although 9 patients (12%) reported intra-familial, acquaintance, or other types of domestic violence. Although most of the providers did not directly question patients about when the violence occurred, 27 (35%) patients indicated that the abuse was a current issue in their lives and 34 (44%) disclosed past experiences with abuse. Two patients reported both past and present abuse. In conversations in which the type of abuse was discernible, 12 (16%) patients disclosed emotional abuse and 26 (34%) disclosed physical abuse. Five patients (6%) mentioned both emotional and physical abuse. Only 24 of the 77 patients (32%) who disclosed abuse to the provider had any documentation of this discussion in their chart.
Table 3 shows the context of the domestic violence discussions, which were almost always initiated by providers during the social history, as part of a checklist of risk factors. For example, a provider might ask, “Do you smoke? Do you drink alcohol often or use any street drugs? Do you have any problems with domestic violence?”
Table 4 shows examples of communication strategies that emergency department providers used to discuss domestic violence. Most often, provider inquiry was a variation of the question, “Are you a victim of domestic violence?” Screening questions were frequently (45%) asked in a perfunctory manner and were sometimes (10%) framed in the negative: “He’s never hit you?” Approximately one third of the time, providers probed (defined as asking ≥1 follow-up question after a patient’s initial response) for further information; 10% of the time, this included in-depth questioning or a detailed domestic violence history. Providers mentioned the computer screening 15% of the time.
Table 5 shows provider communication strategies that appeared to be associated with patient disclosure. Patient disclosure of abuse was more likely to be found in audio-tapes in which the provider probed for domestic violence, created open-ended opportunities for discussion, and was generally responsive or expressed empathy when a patient mentioned a psychosocial issue (for example, “stress”). However, disclosures still occurred when the provider hesitated; used broken syntax, such as “um”; or laughed during the course of the domestic violence conversation. In the following excerpt, the provider elicited a traumatic abuse history from a young woman who presented with irregular menses just by mentioning the word “stress” and following up on a clue about a recent change:
Provider: You can have irregular periods and just get plumbing problems. You can just be under a lot of stress.
Patient: That’s what I’m worried about. ‘Cause I haven’t had one since May.
Provider: Has it been worse at home since May?
Patient: (1-second pause) Yeah, it has. I had to leave home because my father— he was real terrible … He came in the room and he took the phone from me and then he threw it. So, at that, uh, point, I was just scared. My momma, she—she was shaking. And he also had a—a knife in his hand. . . . he was trying to stab me.
The opportunity to have a meaningful conversation about abuse was often diminished by provider factors, such as screening the patient in the presence of a third party, failure to acknowledge disclosure of abuse, lack of assessment of safety or level of risk, and failure to link the patient with available resources.
In one encounter, the provider asked during the examination, “Any problems at all with domestic violence? I have to give him the evil eye when I ask that question.” The patient laughed, and the provider then addressed her male partner and asked. “Now, is she givin’ you any trouble?” He responded, “Yep.” This approach minimizes the seriousness of domestic violence and fails to provide the confidentiality needed; patients are unlikely to disclose domestic violence in the presence of an abusive partner. In a similar encounter, the male partner volunteers, “I can leave if you like,” providing evidence that patients and their family members take questions about abuse seriously and expect the topic to be asked about in private.
A few provider responses to patient disclosure of abuse were insensitive. In this example, the provider regards the domestic violence disclosure as something that should be addressed by study personnel, even when specifically asked by the patient about help for abuse:
Patient: It says you know where someone could get help for physical or sexual abuse. Do you have information on how I …
Provider: The lady who gave you this paper will give you this … to the triage area, and she’s gonna give you this. (4-second pause) Okay?
This patient described an assault by someone other than an intimate partner:
Patient: Hit me in the face … that was like … almost a year ago.
Provider: All right. But that’s not really domestic.
Patient: No. Right— Right— N-Not— I don’t—.
Provider: Okay, so that’s (…). Okay. (1-second pause) Any coughing or shortness of breath?
In this excerpt, the provider inquired about stress but changed the topic when the patient disclosed conflict in her relationship, returning the conversation to the biomedical concerns:
Patient: Uh, me and my boyfriend, we fight sometimes. (2-second pause)
Provider: Well, that’s some degree of stress.
Patient: (Laughs) Yeah. (4-second pause).
Provider: Okay. Well, I think this is the problem. Okay. Let’s— Let’s do this … We’ll do the EKG. Okay?
Provider: And we can try to give you some uh, some Mylanta (1-second pause). Maybe you—you might have reflux. Sometimes people with high anxiety, they have high acidity, and this will (…).
In the following example, the provider paused awkwardly for 19 seconds and then abruptly changed the topic of conversation to an unrelated medical issue. This interaction is typical of a missed empathic opportunity (12):
Provider: Have you ever been threatened or hurt by (…) or someone close to you?
Patient: Yeah. (19-second pause)
Provider: Are you allergic to any medicine?
Other providers were more responsive but did not know how to assess safety or match resources appropriate to the level of risk:
Provider: Okay. (1-second pause) Um. (1-second pause) Do you have (1-second pause) a plan to get out if it gets too bad or … ?
Patient: (Overlap) No. I don’t know how to get out. I’m tryin’—I’m tryin’ to go somewhere and get, you know– I don’t know where to go. But they—they said they got the places where they can put (…). (1-second pause)
Provider: I can give you some information. Would that help?
Patient: (Overlap) Okay. I appreciate it. Yeah.
Provider: Here’s some, um, information for ya. Okay.
A number, um, for a domestic violence hotline. Okay? (1-second pause) This has some—
Patient: Don’t they need to go through gettin’ the reports and all that for the domestic violence? You know?
Provider: You know, I don’t know.
We also found examples of positive provider responses to disclosure of abuse. These included allowing the patient to talk about their experiences; checking to be sure the patient was not in any current danger (safety checks); counseling; mentioning available law enforcement and legal recourses, bringing in a social worker, and showing empathy and concern; voicing helpful opinions; and reinforcing the importance of following up with referrals. The following excerpt is exemplary. The provider is generally responsive, listens, validates the patient, and encourages her to get counseling in a way that empowers her to make changes:
Patient: … (…). And I’m really tired. I’m really tired of him taking advantage of me. You know?
Provider: Mm hmm.
Patient: … I might have a little weight on me.
Provider: (Overlap) (…) don’t ever think that you’re not a pretty lady. You know what? We have people here you can talk to. And you may find that just talking to somebody a few times could get you over this.
Provider: You know, you may not need 6 months of therapy. You may just need somebody to tell you what direction to go.
Provider: ’Cause there’s a lot of self-help out there as well. And you’re smart enough to be able to sort it out.
This provider offers a private opportunity for the patient to disclose and leaves the door open for future support:
Provider: [To male third party]: Can you step out for 1 second and just let us talk alone just for 1 second?
Male partner: Sure. Okay. (Laugh)
Provider: I noticed you filled out this questionnaire out in the waiting area. And, uh, the only thing I just wanted to discuss with you is, uh (…) partner-to-partner (…). Does he ever—Has anybody ever threatened you, or do you ever feel threatened?
[long patient explanation edited out]
Patient: Yeah. We’d been talking about (…) controlling thing. But he can’t—Once he’s mad, he can’t just (…). It’s not physical.
Provider: No physical violence.
Patient: (Overlap) (…). No.
Provider: If you need any help, we’re here for you.
Several emergency care providers offered assistance to patients who disclosed past or current abuse; this assistance mainly took the form of discussing safety. In 59% of the 77 domestic violence disclosures, the provider performed a safety check. In 38% of disclosures, the provider expressed empathy or concern for the patient and their circumstances. However, a specific domestic violence referral was discussed in only 19 of 77 (25%) domestic violence disclosures by patients, only 12 (16%) providers mentioned involving the police or legal authorities, and only 3 (4%) patients disclosing domestic violence were seen by a social worker. Consistent with the general failure to document abuse, no provider mentioned that the medical records might be of use, should the patient need to go to court.
Emergency care providers screen for abuse in a perfunctory manner, typically asking a variation of the question “Are you a victim of domestic violence?” during the social history. This communication strategy is not ideal, and most women who indicated that they were at risk for abuse on the computer screening tool did not share this information with the provider. Routine screening for abuse remains a controversial issue (17, 18), but women who experience abuse frequently seek health care in such settings as emergency departments (19–21). Although our study does not directly assess the effectiveness of domestic violence screening, it sheds light on screening behaviors that might increase patient disclosure and open the door to a meaningful discussion about abuse.
Patients were more likely to disclose experiences with abuse when providers used open-ended questions to initiate the topic and probed for abuse by asking at least 1 follow-up question. Use of open-ended questions is recommended as a means for providing patient-centered care, which in turn has been identified by the Institute of Medicine and others as a quality marker (22–24). It is unclear whether a follow-up question was interpreted as true provider interest or whether it gave the patient additional time to reflect and share abuse information. Another successful communication strategy in encounters with disclosure was provider responsiveness to psychosocial clues. Research has found that responsiveness to clues, including use of empathy and creating “windows of opportunity” for sharing highly charged or emotion-laden information, results in patients feeling known and understood (5, 12, 13, 16, 25–27). Practice through role play and the use of mnemonics have been shown to increase medical students’ mastery of communication skills in screening for domestic violence (28). In the context of a hectic clinical environment, creating trust and understanding are critically important in facilitating patient disclosure about abuse experiences (9, 26). Although it may seem counterintuitive, provider responsiveness does not necessarily add substantial time to the visit (29).
Case studies and interviews with survivors of domestic violence have identified the key components of effective domestic violence interventions by health care providers as asking direct questions, being respectful and concerned, being knowledgeable about domestic violence, and providing referrals to services (9, 14, 30, 31). Although courtesy and respect should be present in any provider–patient interaction, it is especially important for victims of abuse, for whom sensitivity to the topic and any disclosure are often a gateway to change (32). We would suggest first normalizing the situation by stating that these questions are asked of all patients who come to the emergency department, then asking a direct question (for example, “Are you in a relationship where you have been hit or threatened?”), slowing down for the screening question, and pausing for a response. If the patient says “no” or hesitates, a follow-up question is appropriate. Although this is a matter of style, some examples might be, “Has anyone ever treated you badly or made you do things you don’t want to do?” “Is there anyone you are afraid of?” “Is there a lot of stress in your relationship?” “Do you and your partner fight a lot—does it ever get physical?”
Provider response to disclosure is just as important as asking the right questions and being sensitive to the patient’s initial response or hesitation. Unfortunately, some providers we observed were more awkward in their responses, sometimes changing the topic or discouraging further conversation, usually by failing to acknowledge abuse or abruptly switching back to biomedical concerns. It is understandable that providers working in very chaotic clinical environments might feel as if they are on uncertain ground when the conversation leaves the biomedical realm. Best responses include the use of empathy (for example, “I’m sorry that happened to you”) and support (for example, let the patient know that the abuse is not her fault and that she does not deserve it). Survivors report that validation of abuse and encouragement by a health care provider can be life-changing if it is done without judgment—that is, when suggestions, not demands, are made (3, 32).
Once a provider detects current domestic violence, appropriate linkage to domestic violence services is critical (14, 30, 31). Although some emergency department providers emphasized the computer-generated referrals, we found very few instances in which counseling or social work services were provided in response to patient disclosure. Likewise, the medical record, which is protected and kept confidential, can be a useful tool for victims who might one day need to use such records in legal proceedings against their partner. Medical record review from our study revealed a general failure of provider documentation: Only one third of patients who disclosed abuse also had it documented in their chart.
Our study has several important limitations. Because the providers were from 1 residency program, our findings are not necessarily generalizable to other emergency departments or to other health care settings. We did not videotape the encounters and probably missed important non-verbal communication. Mainly because of high levels of ambient noise in the emergency department, 410 of 1281 (32%) recordings were unusable for analysis; these may have represented unique encounters. Finally, provider interactions were probably skewed by their awareness of the purpose of the audiorecording study. This “Hawthorne effect” should have influenced emergency providers to be on their best behavior, indicating that our results may be an “upper bound” for the quality of emergency-department domestic violence encounters.
In conclusion, although hectic clinical environments present many obstacles to identifying risk factors for such sensitive topics as domestic violence, several provider communication behaviors seemed to facilitate patient disclosure. Further education should focus on improving provider communication skills and response to abuse disclosures.
The authors thank Mindy Drum, PhD; David Howes, MD; Laura McCloskey, PhD; Melissa Dichter, MSW; and Joanna Bisgaier, BA, for instrumental support and insightful feedback. They also thank the many helpful internal and external reviewers, and the faculty, residents, staff, and patients of the University of Chicago Emergency Medicine Program.
Grant Support: By the Agency for Healthcare Research and Quality (grant RO1 HS 11096-03). Dr. Rhodes is also supported by grant K23 MH64572 from the National Institute of Mental Health. This article was primarily supported by Agency for Healthcare Research and Quality (RO1 HS 11096-03).
From School of Social Policy & Practice, University of Pennsylvania, Philadelphia, Pennsylvania; Indiana University School of Medicine, Regenstrief Institute, and Richard L. Roudebush Veterans Affairs Medical Center, Indianapolis, Indiana; The University of Chicago, Chicago, Illinois; and The University of Toronto, Toronto, Ontario, Canada.
Potential Financial Conflicts of Interest: None disclosed.
Current author addresses and author contributions are available at www.annals.org.
Author Contributions: Conception and design: K.V. Rhodes, R.M. Frankel, W. Levinson.
Analysis and interpretation of the data: K.V. Rhodes, N. Levinthal, R.M. Frankel, E. Prenoveau, W. Levinson.
Drafting of the article: K.V. Rhodes, R.M. Frankel, N. Levinthal, E. Prenoveau.
Critical revision of the article for important intellectual content: K.V. Rhodes, R.M. Frankel, N. Levinthal, W. Levinson.
Final approval of the article: K.V. Rhodes, R.M. Frankel, J. Bailey, W. Levinson.
Provision of study materials or patients: K.V. Rhodes.
Obtaining of funding: K.V. Rhodes, W. Levinson.
Administrative, technical, or logistic support: K.V. Rhodes, E. Prenoveau, J. Bailey.
Collection and assembly of data: K.V. Rhodes, R.M. Frankel, E. Prenoveau, J. Bailey.
Karin V. Rhodes, Division of Health Care Policy Research, Department of Emergency Medicine, School of Social Policy & Practice, University of Pennsylvania, 3815 Walnut Street, Room 201, Philadelphia, PA 19104.
Richard M. Frankel, Indiana University School of Medicine, Indianapolis, IN 46202.
Naomi Levinthal, University of Chicago, 5841 South Maryland Avenue, Chicago, IL 60637.
Elizabeth Prenoveau, University of Chicago, 5841 South Maryland Avenue, Chicago, IL 60637.
Jeannine Bailey, 1655 North Burlington Street, #1, Chicago, IL 60614.
Wendy Levinson, University of Toronto, 190 Elizabeth Street, #3-805, Toronto, Ontario M5G 2C4, Canada.