- Email: [email protected]
Development and Validation of an Emergency Department Screening and Referral Protocol for Victims of Domestic Violence
CONCEPTS
D evelopment and Validation of an Emergency Department Screening and Referral Protocol for Victims of Domestic Violence From the University of North CarolinaSchool of Medicine, Department of EmergencyMedicine*, and the EmergencyDepartment, University of North Carolina Hospitals~, ChapelHill, North Carolina; and the Orange-Durham Coalitionfor Battered Women, Durham, North Carolina.~ Receivedfor publication March 25, i996. Acceptedfor publication March 26, I996. Presented by PJ Shah at the University of North CarolinaSchool of Medicine Student ResearchDay, January 31, 1996. Funded in part by a faculty small grant awardfrom the Injury Prevention Research Centerof the University of North Carolinaat Chapel Hill. Copyright © by the American College of Emergency Physicians.
Anna E Waller, ScD* Susan M Hohenhaus, RN * Poorvi J Shah* Elizabeth A Stern, MPH §
Study objective: To describe the development, design, and validation of an emergency department protocol for the identification, documentation, and referral of victims of domestic violence. Methods: We based protocol development and design on a departmental needs assessment. The validation component involved the screening of women 16 years and older treated in the ED during a 2-week period at both triage (stage 1) and nursing assessment (stage 2). Sensitivity and specificity of the triage screen were determined. Results: The departmental needs assessment revealed several important limiting factors that motivated the design of the protocol. In response,the protocol design included a two-stage screening process, stage 1 taking place at triage and stage 2 as part of the nursing assessment. During the 2-week validation study, 595 women 16 years and older were treated in the ED, but complete two-stage screening data were obtained for only 114 (19%). Of the patients who were appropriately screened, eight screened positive at stage 1 and two of the eight were confirmed at stage 2. Two additional cases were identified at stage 2 in whom violence had not been suspected at stage 1. Triage screen sensitivity was 50%, specificity 95%. Of the women properly screened at both stages, 3.5% were identified as victims of domestic violence. Conclusion: We identified many obstacles to implementation of an ED domestic violence screening and referral protocol, demonstrating that evaluation is imperative in determining actual clinical impact. [Waller AE, HohenhausSM, Shah PJ, Stern EA: Development and validation of an emergency department screening and referral protocol for victims of domestic violence. Ann EmergMed June 1996;27:754-760.]
754
ANNALS OF EMERGENCY MEDICINE
27:6
JUNE 1996
ED P R O T O C O L Waller et all
INTRODUCTION
RESULTS
Domestic violence is a major health problem among women. More women are treated for injuries resulting from domestic violence than for any other cause of injury. 1,2 It has been estimated that 17% to 30% of all women treated in hospital EDs are victims of domestic violence. ~,3 The Joint Commission for the Accreditation of Healthcare Organizations (]CAHO) requires EDs to institute a protocol for identifying and treating victims of domestic violence. The American Medical Association has published guidelines to assist medical practitioners in this process) In response to these needs, we developed a domestic violence screening and referral protocol in the ED of our Level I tertiary care hospital (annual census, 36,000). In this article we describe two distinct components of this process: (1) the development and design of the protocol (March-June 1995) and (2) the initial efforts to validate the screening procedure that was implemented (September 1995).
The interdisciplinary committee included representation from the hospital's emergency medicine, nursing, social work, and police departments and the local domestic violence agency. In addition, the committee occasionally sought input from the hospital's trauma service and patient-registration and legal departments, as well as others in the schools of medicine, nursing, and public health. The committee met almost weekly between March and June 1995. Our review of existing literature about screening for domestic violence in the ED, as well as discussions with researchers in this field, revealed that the recommended practice is to ask all female patients a few key screening questions. Our search for standard, written ED protocols revealed no other area hospitals with such a protocol, although everyone we contacted told us that they "know what to do" to identify and deal with victims of domestic violence. Several serious limitations were identified in our ED indicating that the routine screening of all female patients with standard questions would not be feasible. The most fundamental problem identified was the lack of privacy during the questioning of patients in our ED. Triage is conducted in an open area just inside the ED entrance; people accompanying the patient, and strangers, wait within a few feet and are easily in earshot. Most examining areas in the ED are separated only by thin cotton curtains. Lack of privacy means that there is often no safe place to ask a woman sensitive personal questions. Another issue we identified was that the current staffing levels of the ED were not adequate to handle the identification rate suggested by previous studies: We also explored resources such as the hospital social work services and the local domestic violence agency. At the time the protocol development process was begun, our ED did not employ a social worker. During the development process, a social worker, shared with several other hospital clinics, was assigned to the ED for limited hours. Although the social workers themselves were eager to provide services for identified Victims of domestic violence, coverage was not available 24 hours a day, 7 days a week. The local domestic violence agency, which does provide this level of on-call coverage, agreed to be the first line of referral when a social worker was not available in the hospital but expressed concern about how the agency would handle the increased demand for services if routine screening and referral produced the expected number of additional referrals. Investigation of existing policies in the ED revealed that no clear, routinely followed policy existed for the referral
PROTOCOL DEVELOPMENT AND DESIGN
The specific needs and unique requirements of our ED had to be determined and incorporated into the protocol design. All parties who would be affected by its implementation had to be involved in the protocol-development process. In addition, the implementation and use of the protocol had to be evaluated to determine its efficacy and the original protocol modified in response to this evaluation. A committee with representation from the ED, the hospital, and the community was established to develop the protocol. Existing literature about ED screening for domestic Violence was reviewed. Other EDs were informally surveyed by means of phone calls and Internet postings to determine how they addressed the problem of domestic violence and what written protocols existed that we could consult as models. In addition, we assessed the physical constraints of our ED with regard to privacy during screening and referral. The existing resources of the ED and the hospital for dealing with identified victims of domestic violence were ascertained, as well as any existing policies that would have to be accommodated as part of the new protocol. The project was approved by the University of North Carolina School of Medicine Committee on the Protection of the Rights of Human Subjects and was exempted from a full committee review.
JUNE 1996
27:6
ANNALS OF EMERGENCY MEDICINE
755
ED PROTOCOL Waller et al
of patients identified as victims of domestic violence. There was a history of difficulties experienced by ED personnel in trying to gain access to community services for patients, as well as failed attempts to implement screening for domestic violence in the ED. These efforts had been organized by people outside the ED and used external personnel to conduct screening and counseling. Our findings reinforced the conclusion that the protocol under development would have to function within the existing ED structure. The design of the protocol was also affected by the ED's visitor policy, under which a pass is automatically issued to each patient, allowing one family member or friend to accompany the patient during the ED visit. Obviously this precluded any possibility of private questioning of the patient. We included in the protocol an automatic override of this policy in cases of suspected domestic violence. The role of the hospital police was also identified as a point of confusion in the ED. ED personnel were unclear about legal reporting requirements and often did not realize that our hospital police are a commissioned police force, present in the ED 24 hours a day and able to take patient reports and required reports from ED personnel. The hospital police are also available to photograph injuries for the patient's record. Their involvement in this process establishes a legal "chain of evidence" that does not exist if ED staff themselves take the photos. Our protocol includes clearly defined roles for the hospital police.
The protocol we developed addresses domestic violence and eider abuse and neglect. The overlap between the two issues is extensive and the indicators for identification similar. The protocol was developed to cover all patients who present to the ED, regardless of sex, aged 16 years or older. Patients younger than 16 years are covered by the hospital child abuse protocol, regardless of the identity of the abuser. We have limited this article and the evaluation of the protocol, however, to female domestic violence victims, recognizing that this population is most likely to be affected by this protocol. The key feature of the protocol is a two-stage screening process. Stage 1 is triage and stage 2 nursing assessment. Stage 1 is conducted silently by the triage nurse, who notes the presence of any of several indicators of domestic violence (Figure 1) and must indicate on the nursing form that abuse is or is not suspected by simply checking a box, yes or no. This stage of screening requires no verbal exchange between the nurse and the patient about violence or battering, thereby accommodating the lack of privacy at triage in our department. The triage screen is designed to be overly sensitive, identifying many false positives but minimizing false negatives. If the triage nurse indicates that abuse is suspected, the hospital police in the ED are notified as a precaution. When the patient moves on to registration, no visitor's pass is issued, ensuring that the patient is initially treated
Figure 1.
Matrix of domestic violence indicators used at stage 1 (triage) screen. Physical Indicators
Emotional Indicators
Other Indicators
Injuries consistent with abuse (eg, burns, welts, bruises, bites}
Fearful of caregivers (including hospital staff and visitors)
Restrained or locked in/out of house
Fractures, dislocations, injuries not consistent with trauma described
Feels threatened with violence, institutioealization, abandonment, guardianship
Is kept socially isolated
Unusual injury marks, as if from belt, rope, hairbrush injury
Is withdrawn, fearful, depressed and has physical
Has frequent, unscheduled use of ED care
Dried blood, semen
Process of emotional abuse is observable (eg, visitor yells/threatens/swears)
Sleeping or eating disorders
Site of injury: face, throat, breasts, abdomen, genitals, bilateral extremities
Suicide attempt in context of "relationship problems" and seeking treatment
Prolonged interval between onset of illness or injury
Drug toxicity/overdose; alcohol intoxication
Vague, nonspecific complaint of ill or failing health in context of "can't seem to do what I'm supposed to do"
Single-car crash, either as a driver or passenger
Problem pregnancies; preterm bleeding/miscarriage; self-induced abortion
History of or current self-mutilation
7 56
ANNALS OF EMERGENCY MEDICINE 27:6 JUNE 1996
ED P R O T O C O L
Wailer et al
in private. Efforts are also made to place the patient in one of the few private examining rooms. If stage 1 results are positive, stage 2 screening is performed during the nursing assessment. The examining nurse asks the patient the following direct questions to ascertain whether abuse is present: Are you currently in a relationship with a
(spouse, partner, etc) in which you have felt afraid? been threatened? been physically hurt? If the answer to any of these questions is yes, the nurse asks the following questions: Is something like that happening now? Is that what happened this time? Who did this to you? What is this person's name? How do you see the situation now? Is it
Figure 2.
Domestic violence protocol patient flow chart.
All women age 16 years or older Stage 1 screen:triage assessment
I Stage 1 screen negative I
I Stage 1 screen positive I
I Abusenot suspected I
[
[
Protocolends
]
No visitor's pass
Abusesuspected
I
1
I Hospital police informed I ]
k
Stage 2 screen: nurse assessment ]
I
[ Stage 2 screen negative 1
Stage 2 screen positive Mandatory reporting: 1. To hospital police (any serious injury) 2. To hospital social worker {domestic violence victim who is elderly or disabled)
Patient offered referral to: 1. Hospital social worker (MF, 9:30 AM-6PM) OR 2. Domesticviolence agency THEN 3. Hospital police take patient report
,
I
Protocolends
]
[ Visitor's pass issued ]
T
Patient accepts referral Referral made
Police informed of referral status
Patient refuses referral Complete medical record 1 documentation
Completemedicalrecord documentation
JUNE 1996
27:6
ANNALS OF EMERGENCY MEDICINE
7 57
ED PROTOCOL Wailer et al
getting worse? Do you feel safe in returning home? If there are children in the home, the nurse asks the patient whether she feels the children are safe. The wording of the questions is only suggested in the protocol. ED nurses wanted the flexibility to use their own words so they could adapt the questioning as the situation required. These questions are only asked of patients in whom abuse is suspected at triage. If the answer to any of these questions is yes, the referral component of the protocol is implemented. The examining nurse discusses various options with the patient, including talking with the ED social worker or with an advocate from the local domestic violence agency; filing a police report, even if the patient does not elect to press charges at the time; taking written information about the services available through the local domestic violence agency or perhaps only the phone number for the agency hotline, which is given on a slip of paper with the number only and no identifying information. If the patient wishes to talk with someone, the appropriate calls are made and further efforts are made to arrange a private place in which to talk. The protocol requires documentation of both its implementation and outcomes in the patient medical record. This includes the presence of any indicators noted at triage; the answers to direct questioning about current abuse, including verbatim quotes from the victim and the name and relationship of the perpetrator; referral offers made and what was accepted or rejected by the patient; and the final opinion of the examining nurse regarding whether abuse was confirmed, suspected but ruled out on further examination, or continued to be suspected although the patient denied its presence. Nurses are also asked to document that the attending physician was notified of the confirmed or suspected abuse case. Figure 2 presents an overview of the protocol and patient flow and ED communication within it. Protocol approval from the hospital was obtained in July 1995. In July and early August, training about domestic violence and the new ED protocol was provided for ED emergency care providers. The local domestic violence agency provided a health educator with experience in community education and professional training to conduct these sessions. All emergency care providers were required to attend at least one of these sessions. Nonetheless, attendance was poor. These training sessions included information about the dynamics of domestic violence, indicators of abuse, a values clarification exercise, and specific information about the ED protocol and its
758
planned evaluation. Those who attended the 1-hour sessions rated them very highly and reported that they found them very informative and useful. In addition, all nursing staff in the ED received a self-learning packet about the protocol and domestic violence as an issue for emergency care providers. This packet included a required °'test" that they completed and returned to the department's clinical nurse educator, who monitors nursing credentials. As part of the implementation of this protocol, an evaluation component was designed and incorporated. It was recognized by all involved in its development that the protocol did not directly follow guidelines established by previous research. It was also recognized that a protocol is merely a piece of paper unless it is understood and used appropriately by the staff. Evaluation is necessary to determine whether the presence of a written proto-col, and the training and education that accompanied its implementation, actually improve the identification and referral of domestic violence victims in our ED. We also want to be able to revise the protocol as needed to make it as effective as possible. A 2-week validation study of the two-stage screening process was one component of this evaluation. Details of this component are presented in the section below. VALIDATION OF THE SCREENING PROCEDURE
The two-stage screening process used in this protocol differs from screening protocols previously described in the literature. Therefore we designed a validation project to determine whether the two-stage screening was effective. Shortly after the September 1, 1995, implementation of the protocol in the ED, for a 2-week period all female patients older than 15 years treated in the ED were screened at both stages of the protocol, regardless of their status on the stage 1 screen. We compared the presence of first-stage screening indicators with answers given to specific screening questions at the second stage. With screening all women treated in the ED at both stages, an estimate of false negatives and false positives identified at the first stage of screening was possible, and sensitivity and specificity were calculated. Finally, we compared the rate of positive identification of domestic violence cases using the two screening stages with previous identification in our ED. During the 2-week study period, 595 women aged 16 years or older were treated in the ED. Of these women, complete screening data from both stage 1 and stage 2 were collected for only 114 (19%). Of the patients who were appropriately screened, eight screened positive
ANNALS OF EMERGENCY MEDICINE
27:6
JUNE 1996
ED PROTOCOL Waller et al
(abuse suspected) at stage 1 (triage). Two of the eight were confirmed to be victims of domestic violence at stage 2 (nurse assessment). Another two patients were identified at stage 2 who had been screened as negative (abuse not suspected) at stage 1. This yielded a total identification of four confirmed cases of domestic violence during the 2week study period, or only 3.5% of the women appropriately screened at both stages. Six false positives were identified through the stage 1 screening at triage and two false negatives were identified by the stage 2 screening, yielding a triage screen sensitivity of 50% and a specificity of 95% (Table). Presumably, the estimate of false negatives is a minimum because the second stage screening cannot be expected to be 100% effective, thus resulting in false negatives at both screening levels. Although only four confirmed cases of domestic violence were identified during the 2-week study period, the previous identification of domestic violence victims in our ED had generally been two or three cases per month. DISCUSSION
Although much effort, expertise, and consideration went into the design and development of the domestic violence screening and referral protocol now in use in our ED, it is clear that we are not yet meeting the needs of our patient population for consistent and accurate identification of victims of domestic violence followed by appropriate referral to services. A protocol is only paper and ink unless it can be successfully implemented in the actual ED setting. ED staff must believe in its utility and must be willing and able to use it as part of their normal routine of providing patient care. Research involving comprehensive screening and referral is often conducted with external funding to provide additional personnel to train the ED staff and provide identification and referral services. The disadvantage of Table.
Distribution of positive and negative cases by level of screening. Screen 2: Nurse Assessment Screen 1: Triage Assessment
Negative Screen
Positive Screen
Total
Negative screen Positive screen
104 6
2 2
106 8
Total
110
4
114
Triage sensitivity: 2 of 4 {50%); triage specificity: 104 ef ~10 (95%).
JUNE 1996
27:6
ANNALS OF EMERGENCY MEDECINE
such programs is that when the research funding ends, the comprehensive screening and referral program often ceases. We believed our protocol had to become a part of the normal everyday workings of the ED or it would not survive or function well in the long term. Without appropriate evaluation, it is often impossible to know whether a protocol accomplishes what it was designed to do or to identify why it does or does not work. The development process, actual protocol, and validation component presented in this paper were designed to optimize the opportunities for constructing a working protocol that meets the needs of both the patients and the ED staff. Although the success of this protocol has been limited, we believe the approach we have taken is appropriate and will lead to eventual success. Clearly the validation component of our protocol evaluation was unsuccessful on many fronts. Compliance with the study methods by ED staff was extremely poor. The rate of identification, even for the minority of women who were appropriately screened, although greater than that previously seen in our ED, was still far below estimates of true prevalence of domestic violence in our patient population. The stage 1 screen at triage appeared to be highly specific rather than highly sensitive, the opposite of our intent. Although it would be dangerous to draw any solid conclusions from such a poor data-collection effort, certainly the indication is that the two-stage screening mechanism is not working as intended. Several explanations are available for the poor data collection in the validation study. Qualitative interviews with ED personnel, in response to this failure, indicated that staff were confused by the data collection methods used in the validation study and by the silent screening for indicators used during stage 1, at triage. Although both these issues were addressed during staff training, through selflearning packets about the protocol and through one-onone communication with ED nursing staff, it is obvious that we failed to clarify these important issues. The most prominent concern about the protocol expressed in the interviews with ED staff was that of time constraints in processing patients through the ED. This issue had an obvious effect on the validation study and is likely to be an ongoing barrier to optimizing the efficacy of the protocol. Shortly after the implementation of the 2-week validation study (but unrelated to it), operational problems in time for triage and ED length of stay were addressed by the adoption of an ED policy with the goal of reducing triage time to 2 minutes and ED disposition time to 2 hours. Because our protocol was designed with the explicit goal of working within the existing routine of
759
ED PROTOCOL Waller et al
the ED, no time allowance was made for data collection for the validation study. This lack of time was a key factor in the low data collection rate of 19%, but it underscores the importance of efficient integration of the screening protocol into ED operations. It is obvious that further education of the ED staff is needed before the validation study can be repeated in the ED. This supplementary training is being designed to include actual role-playing using the various stages of the screening and referral protocol. We are exploring other avenues of communication with the nursing staff to try to identify more efficient ways of sharing information and feedback. Meanwhile, we continue with the other parts of our evaluation, which focus on changes in attitudes and knowledge, as well as medical record documentation. Domestic violence is a serious public health problem in our society. The hospital ED has an important role in identifying and breaking the cycle of violence experienced by many of our patients. Simply having a written protocol is not enough. Our experience indicates that, even with careful planning and design, a protocol may not function as designed or may take much time and effort to implement. Evaluation of the implementation process and the effect of the protocol on actual identification of domestic violence victims in the ED is imperative. Deficits in staff understanding of their roles and confusion about the protocol may not be readily apparent or easily communicated. Changing the culture of an ED is not easy or straightforward. To address this important issue, we must take a long-term approach to the problem. Progress must be reinforced at every step, and our approach must be constantly reassessed. Furthermore, we must be willing to redesign our protocol as needed to make it effective. Finally, unless an institution is willing to commit longterm additional personnel and other resources to address the issue of domestic violence, it is essential that an ED protocol for identification and referral be designed to be a part of existing ED clinical practice.
The authors thank the following people for their contributions to the design and development of the domestic violence screening and referral protocol and the validation procedure described in this article: Sandra Martin, Department of Maternal and Child Health, University of North Carolina School of Public Health; and Greg Johnson, Department of Nursing; Dexter Morris, Department of Emergency Medicine; Lee Marazas, Connie Culbreth, and Kathy Kalanyos, Department of Social Work; and Tom Smith and Richard Brickey, Hospital Police, University of North Carolina Hospitals. The authors also thank Alison Hilton of the University of North Carolina Department of Health Behavior and Health Education for her work interviewing ED staff about their experiences.
Reprint no. 47/1/73922
Address for reprints: Anna EWaller, ScD Department of EmergencyMedicine Universityof North Carolina CB7594 Chapel Hill, North Carolina27599-7594
REFERENCES 1. ResenburgM, FinleyMA: Violencein America: A Public Health Approach. New York: Oxford UniversityPress,1991. 2. The NationalCommitteeon InjuryPreventionand Control:Injury Prevention: Meeting the Challenge. New York: OxfordUniversityPress,1989. 3. RandallT: Domesticviolenceinterventioncalls for morethan treating injuries.JAMA 1990;264:939-940. 4. AmericanMedicalAssociation:Diagnosticand treatmentguidelineson domesticviolence. Arch Fam Med 1992;I:39-47.
760
ANNALS OF EMERCENCY MEDICINE 27:6
JUNE 1996
D evelopment and Validation of an Emergency Department Screening and Referral Protocol for Victims of Domestic Violence From the University of North CarolinaSchool of Medicine, Department of EmergencyMedicine*, and the EmergencyDepartment, University of North Carolina Hospitals~, ChapelHill, North Carolina; and the Orange-Durham Coalitionfor Battered Women, Durham, North Carolina.~ Receivedfor publication March 25, i996. Acceptedfor publication March 26, I996. Presented by PJ Shah at the University of North CarolinaSchool of Medicine Student ResearchDay, January 31, 1996. Funded in part by a faculty small grant awardfrom the Injury Prevention Research Centerof the University of North Carolinaat Chapel Hill. Copyright © by the American College of Emergency Physicians.
Anna E Waller, ScD* Susan M Hohenhaus, RN * Poorvi J Shah* Elizabeth A Stern, MPH §
Study objective: To describe the development, design, and validation of an emergency department protocol for the identification, documentation, and referral of victims of domestic violence. Methods: We based protocol development and design on a departmental needs assessment. The validation component involved the screening of women 16 years and older treated in the ED during a 2-week period at both triage (stage 1) and nursing assessment (stage 2). Sensitivity and specificity of the triage screen were determined. Results: The departmental needs assessment revealed several important limiting factors that motivated the design of the protocol. In response,the protocol design included a two-stage screening process, stage 1 taking place at triage and stage 2 as part of the nursing assessment. During the 2-week validation study, 595 women 16 years and older were treated in the ED, but complete two-stage screening data were obtained for only 114 (19%). Of the patients who were appropriately screened, eight screened positive at stage 1 and two of the eight were confirmed at stage 2. Two additional cases were identified at stage 2 in whom violence had not been suspected at stage 1. Triage screen sensitivity was 50%, specificity 95%. Of the women properly screened at both stages, 3.5% were identified as victims of domestic violence. Conclusion: We identified many obstacles to implementation of an ED domestic violence screening and referral protocol, demonstrating that evaluation is imperative in determining actual clinical impact. [Waller AE, HohenhausSM, Shah PJ, Stern EA: Development and validation of an emergency department screening and referral protocol for victims of domestic violence. Ann EmergMed June 1996;27:754-760.]
754
ANNALS OF EMERGENCY MEDICINE
27:6
JUNE 1996
ED P R O T O C O L Waller et all
INTRODUCTION
RESULTS
Domestic violence is a major health problem among women. More women are treated for injuries resulting from domestic violence than for any other cause of injury. 1,2 It has been estimated that 17% to 30% of all women treated in hospital EDs are victims of domestic violence. ~,3 The Joint Commission for the Accreditation of Healthcare Organizations (]CAHO) requires EDs to institute a protocol for identifying and treating victims of domestic violence. The American Medical Association has published guidelines to assist medical practitioners in this process) In response to these needs, we developed a domestic violence screening and referral protocol in the ED of our Level I tertiary care hospital (annual census, 36,000). In this article we describe two distinct components of this process: (1) the development and design of the protocol (March-June 1995) and (2) the initial efforts to validate the screening procedure that was implemented (September 1995).
The interdisciplinary committee included representation from the hospital's emergency medicine, nursing, social work, and police departments and the local domestic violence agency. In addition, the committee occasionally sought input from the hospital's trauma service and patient-registration and legal departments, as well as others in the schools of medicine, nursing, and public health. The committee met almost weekly between March and June 1995. Our review of existing literature about screening for domestic violence in the ED, as well as discussions with researchers in this field, revealed that the recommended practice is to ask all female patients a few key screening questions. Our search for standard, written ED protocols revealed no other area hospitals with such a protocol, although everyone we contacted told us that they "know what to do" to identify and deal with victims of domestic violence. Several serious limitations were identified in our ED indicating that the routine screening of all female patients with standard questions would not be feasible. The most fundamental problem identified was the lack of privacy during the questioning of patients in our ED. Triage is conducted in an open area just inside the ED entrance; people accompanying the patient, and strangers, wait within a few feet and are easily in earshot. Most examining areas in the ED are separated only by thin cotton curtains. Lack of privacy means that there is often no safe place to ask a woman sensitive personal questions. Another issue we identified was that the current staffing levels of the ED were not adequate to handle the identification rate suggested by previous studies: We also explored resources such as the hospital social work services and the local domestic violence agency. At the time the protocol development process was begun, our ED did not employ a social worker. During the development process, a social worker, shared with several other hospital clinics, was assigned to the ED for limited hours. Although the social workers themselves were eager to provide services for identified Victims of domestic violence, coverage was not available 24 hours a day, 7 days a week. The local domestic violence agency, which does provide this level of on-call coverage, agreed to be the first line of referral when a social worker was not available in the hospital but expressed concern about how the agency would handle the increased demand for services if routine screening and referral produced the expected number of additional referrals. Investigation of existing policies in the ED revealed that no clear, routinely followed policy existed for the referral
PROTOCOL DEVELOPMENT AND DESIGN
The specific needs and unique requirements of our ED had to be determined and incorporated into the protocol design. All parties who would be affected by its implementation had to be involved in the protocol-development process. In addition, the implementation and use of the protocol had to be evaluated to determine its efficacy and the original protocol modified in response to this evaluation. A committee with representation from the ED, the hospital, and the community was established to develop the protocol. Existing literature about ED screening for domestic Violence was reviewed. Other EDs were informally surveyed by means of phone calls and Internet postings to determine how they addressed the problem of domestic violence and what written protocols existed that we could consult as models. In addition, we assessed the physical constraints of our ED with regard to privacy during screening and referral. The existing resources of the ED and the hospital for dealing with identified victims of domestic violence were ascertained, as well as any existing policies that would have to be accommodated as part of the new protocol. The project was approved by the University of North Carolina School of Medicine Committee on the Protection of the Rights of Human Subjects and was exempted from a full committee review.
JUNE 1996
27:6
ANNALS OF EMERGENCY MEDICINE
755
ED PROTOCOL Waller et al
of patients identified as victims of domestic violence. There was a history of difficulties experienced by ED personnel in trying to gain access to community services for patients, as well as failed attempts to implement screening for domestic violence in the ED. These efforts had been organized by people outside the ED and used external personnel to conduct screening and counseling. Our findings reinforced the conclusion that the protocol under development would have to function within the existing ED structure. The design of the protocol was also affected by the ED's visitor policy, under which a pass is automatically issued to each patient, allowing one family member or friend to accompany the patient during the ED visit. Obviously this precluded any possibility of private questioning of the patient. We included in the protocol an automatic override of this policy in cases of suspected domestic violence. The role of the hospital police was also identified as a point of confusion in the ED. ED personnel were unclear about legal reporting requirements and often did not realize that our hospital police are a commissioned police force, present in the ED 24 hours a day and able to take patient reports and required reports from ED personnel. The hospital police are also available to photograph injuries for the patient's record. Their involvement in this process establishes a legal "chain of evidence" that does not exist if ED staff themselves take the photos. Our protocol includes clearly defined roles for the hospital police.
The protocol we developed addresses domestic violence and eider abuse and neglect. The overlap between the two issues is extensive and the indicators for identification similar. The protocol was developed to cover all patients who present to the ED, regardless of sex, aged 16 years or older. Patients younger than 16 years are covered by the hospital child abuse protocol, regardless of the identity of the abuser. We have limited this article and the evaluation of the protocol, however, to female domestic violence victims, recognizing that this population is most likely to be affected by this protocol. The key feature of the protocol is a two-stage screening process. Stage 1 is triage and stage 2 nursing assessment. Stage 1 is conducted silently by the triage nurse, who notes the presence of any of several indicators of domestic violence (Figure 1) and must indicate on the nursing form that abuse is or is not suspected by simply checking a box, yes or no. This stage of screening requires no verbal exchange between the nurse and the patient about violence or battering, thereby accommodating the lack of privacy at triage in our department. The triage screen is designed to be overly sensitive, identifying many false positives but minimizing false negatives. If the triage nurse indicates that abuse is suspected, the hospital police in the ED are notified as a precaution. When the patient moves on to registration, no visitor's pass is issued, ensuring that the patient is initially treated
Figure 1.
Matrix of domestic violence indicators used at stage 1 (triage) screen. Physical Indicators
Emotional Indicators
Other Indicators
Injuries consistent with abuse (eg, burns, welts, bruises, bites}
Fearful of caregivers (including hospital staff and visitors)
Restrained or locked in/out of house
Fractures, dislocations, injuries not consistent with trauma described
Feels threatened with violence, institutioealization, abandonment, guardianship
Is kept socially isolated
Unusual injury marks, as if from belt, rope, hairbrush injury
Is withdrawn, fearful, depressed and has physical
Has frequent, unscheduled use of ED care
Dried blood, semen
Process of emotional abuse is observable (eg, visitor yells/threatens/swears)
Sleeping or eating disorders
Site of injury: face, throat, breasts, abdomen, genitals, bilateral extremities
Suicide attempt in context of "relationship problems" and seeking treatment
Prolonged interval between onset of illness or injury
Drug toxicity/overdose; alcohol intoxication
Vague, nonspecific complaint of ill or failing health in context of "can't seem to do what I'm supposed to do"
Single-car crash, either as a driver or passenger
Problem pregnancies; preterm bleeding/miscarriage; self-induced abortion
History of or current self-mutilation
7 56
ANNALS OF EMERGENCY MEDICINE 27:6 JUNE 1996
ED P R O T O C O L
Wailer et al
in private. Efforts are also made to place the patient in one of the few private examining rooms. If stage 1 results are positive, stage 2 screening is performed during the nursing assessment. The examining nurse asks the patient the following direct questions to ascertain whether abuse is present: Are you currently in a relationship with a
(spouse, partner, etc) in which you have felt afraid? been threatened? been physically hurt? If the answer to any of these questions is yes, the nurse asks the following questions: Is something like that happening now? Is that what happened this time? Who did this to you? What is this person's name? How do you see the situation now? Is it
Figure 2.
Domestic violence protocol patient flow chart.
All women age 16 years or older Stage 1 screen:triage assessment
I Stage 1 screen negative I
I Stage 1 screen positive I
I Abusenot suspected I
[
[
Protocolends
]
No visitor's pass
Abusesuspected
I
1
I Hospital police informed I ]
k
Stage 2 screen: nurse assessment ]
I
[ Stage 2 screen negative 1
Stage 2 screen positive Mandatory reporting: 1. To hospital police (any serious injury) 2. To hospital social worker {domestic violence victim who is elderly or disabled)
Patient offered referral to: 1. Hospital social worker (MF, 9:30 AM-6PM) OR 2. Domesticviolence agency THEN 3. Hospital police take patient report
,
I
Protocolends
]
[ Visitor's pass issued ]
T
Patient accepts referral Referral made
Police informed of referral status
Patient refuses referral Complete medical record 1 documentation
Completemedicalrecord documentation
JUNE 1996
27:6
ANNALS OF EMERGENCY MEDICINE
7 57
ED PROTOCOL Wailer et al
getting worse? Do you feel safe in returning home? If there are children in the home, the nurse asks the patient whether she feels the children are safe. The wording of the questions is only suggested in the protocol. ED nurses wanted the flexibility to use their own words so they could adapt the questioning as the situation required. These questions are only asked of patients in whom abuse is suspected at triage. If the answer to any of these questions is yes, the referral component of the protocol is implemented. The examining nurse discusses various options with the patient, including talking with the ED social worker or with an advocate from the local domestic violence agency; filing a police report, even if the patient does not elect to press charges at the time; taking written information about the services available through the local domestic violence agency or perhaps only the phone number for the agency hotline, which is given on a slip of paper with the number only and no identifying information. If the patient wishes to talk with someone, the appropriate calls are made and further efforts are made to arrange a private place in which to talk. The protocol requires documentation of both its implementation and outcomes in the patient medical record. This includes the presence of any indicators noted at triage; the answers to direct questioning about current abuse, including verbatim quotes from the victim and the name and relationship of the perpetrator; referral offers made and what was accepted or rejected by the patient; and the final opinion of the examining nurse regarding whether abuse was confirmed, suspected but ruled out on further examination, or continued to be suspected although the patient denied its presence. Nurses are also asked to document that the attending physician was notified of the confirmed or suspected abuse case. Figure 2 presents an overview of the protocol and patient flow and ED communication within it. Protocol approval from the hospital was obtained in July 1995. In July and early August, training about domestic violence and the new ED protocol was provided for ED emergency care providers. The local domestic violence agency provided a health educator with experience in community education and professional training to conduct these sessions. All emergency care providers were required to attend at least one of these sessions. Nonetheless, attendance was poor. These training sessions included information about the dynamics of domestic violence, indicators of abuse, a values clarification exercise, and specific information about the ED protocol and its
758
planned evaluation. Those who attended the 1-hour sessions rated them very highly and reported that they found them very informative and useful. In addition, all nursing staff in the ED received a self-learning packet about the protocol and domestic violence as an issue for emergency care providers. This packet included a required °'test" that they completed and returned to the department's clinical nurse educator, who monitors nursing credentials. As part of the implementation of this protocol, an evaluation component was designed and incorporated. It was recognized by all involved in its development that the protocol did not directly follow guidelines established by previous research. It was also recognized that a protocol is merely a piece of paper unless it is understood and used appropriately by the staff. Evaluation is necessary to determine whether the presence of a written proto-col, and the training and education that accompanied its implementation, actually improve the identification and referral of domestic violence victims in our ED. We also want to be able to revise the protocol as needed to make it as effective as possible. A 2-week validation study of the two-stage screening process was one component of this evaluation. Details of this component are presented in the section below. VALIDATION OF THE SCREENING PROCEDURE
The two-stage screening process used in this protocol differs from screening protocols previously described in the literature. Therefore we designed a validation project to determine whether the two-stage screening was effective. Shortly after the September 1, 1995, implementation of the protocol in the ED, for a 2-week period all female patients older than 15 years treated in the ED were screened at both stages of the protocol, regardless of their status on the stage 1 screen. We compared the presence of first-stage screening indicators with answers given to specific screening questions at the second stage. With screening all women treated in the ED at both stages, an estimate of false negatives and false positives identified at the first stage of screening was possible, and sensitivity and specificity were calculated. Finally, we compared the rate of positive identification of domestic violence cases using the two screening stages with previous identification in our ED. During the 2-week study period, 595 women aged 16 years or older were treated in the ED. Of these women, complete screening data from both stage 1 and stage 2 were collected for only 114 (19%). Of the patients who were appropriately screened, eight screened positive
ANNALS OF EMERGENCY MEDICINE
27:6
JUNE 1996
ED PROTOCOL Waller et al
(abuse suspected) at stage 1 (triage). Two of the eight were confirmed to be victims of domestic violence at stage 2 (nurse assessment). Another two patients were identified at stage 2 who had been screened as negative (abuse not suspected) at stage 1. This yielded a total identification of four confirmed cases of domestic violence during the 2week study period, or only 3.5% of the women appropriately screened at both stages. Six false positives were identified through the stage 1 screening at triage and two false negatives were identified by the stage 2 screening, yielding a triage screen sensitivity of 50% and a specificity of 95% (Table). Presumably, the estimate of false negatives is a minimum because the second stage screening cannot be expected to be 100% effective, thus resulting in false negatives at both screening levels. Although only four confirmed cases of domestic violence were identified during the 2-week study period, the previous identification of domestic violence victims in our ED had generally been two or three cases per month. DISCUSSION
Although much effort, expertise, and consideration went into the design and development of the domestic violence screening and referral protocol now in use in our ED, it is clear that we are not yet meeting the needs of our patient population for consistent and accurate identification of victims of domestic violence followed by appropriate referral to services. A protocol is only paper and ink unless it can be successfully implemented in the actual ED setting. ED staff must believe in its utility and must be willing and able to use it as part of their normal routine of providing patient care. Research involving comprehensive screening and referral is often conducted with external funding to provide additional personnel to train the ED staff and provide identification and referral services. The disadvantage of Table.
Distribution of positive and negative cases by level of screening. Screen 2: Nurse Assessment Screen 1: Triage Assessment
Negative Screen
Positive Screen
Total
Negative screen Positive screen
104 6
2 2
106 8
Total
110
4
114
Triage sensitivity: 2 of 4 {50%); triage specificity: 104 ef ~10 (95%).
JUNE 1996
27:6
ANNALS OF EMERGENCY MEDECINE
such programs is that when the research funding ends, the comprehensive screening and referral program often ceases. We believed our protocol had to become a part of the normal everyday workings of the ED or it would not survive or function well in the long term. Without appropriate evaluation, it is often impossible to know whether a protocol accomplishes what it was designed to do or to identify why it does or does not work. The development process, actual protocol, and validation component presented in this paper were designed to optimize the opportunities for constructing a working protocol that meets the needs of both the patients and the ED staff. Although the success of this protocol has been limited, we believe the approach we have taken is appropriate and will lead to eventual success. Clearly the validation component of our protocol evaluation was unsuccessful on many fronts. Compliance with the study methods by ED staff was extremely poor. The rate of identification, even for the minority of women who were appropriately screened, although greater than that previously seen in our ED, was still far below estimates of true prevalence of domestic violence in our patient population. The stage 1 screen at triage appeared to be highly specific rather than highly sensitive, the opposite of our intent. Although it would be dangerous to draw any solid conclusions from such a poor data-collection effort, certainly the indication is that the two-stage screening mechanism is not working as intended. Several explanations are available for the poor data collection in the validation study. Qualitative interviews with ED personnel, in response to this failure, indicated that staff were confused by the data collection methods used in the validation study and by the silent screening for indicators used during stage 1, at triage. Although both these issues were addressed during staff training, through selflearning packets about the protocol and through one-onone communication with ED nursing staff, it is obvious that we failed to clarify these important issues. The most prominent concern about the protocol expressed in the interviews with ED staff was that of time constraints in processing patients through the ED. This issue had an obvious effect on the validation study and is likely to be an ongoing barrier to optimizing the efficacy of the protocol. Shortly after the implementation of the 2-week validation study (but unrelated to it), operational problems in time for triage and ED length of stay were addressed by the adoption of an ED policy with the goal of reducing triage time to 2 minutes and ED disposition time to 2 hours. Because our protocol was designed with the explicit goal of working within the existing routine of
759
ED PROTOCOL Waller et al
the ED, no time allowance was made for data collection for the validation study. This lack of time was a key factor in the low data collection rate of 19%, but it underscores the importance of efficient integration of the screening protocol into ED operations. It is obvious that further education of the ED staff is needed before the validation study can be repeated in the ED. This supplementary training is being designed to include actual role-playing using the various stages of the screening and referral protocol. We are exploring other avenues of communication with the nursing staff to try to identify more efficient ways of sharing information and feedback. Meanwhile, we continue with the other parts of our evaluation, which focus on changes in attitudes and knowledge, as well as medical record documentation. Domestic violence is a serious public health problem in our society. The hospital ED has an important role in identifying and breaking the cycle of violence experienced by many of our patients. Simply having a written protocol is not enough. Our experience indicates that, even with careful planning and design, a protocol may not function as designed or may take much time and effort to implement. Evaluation of the implementation process and the effect of the protocol on actual identification of domestic violence victims in the ED is imperative. Deficits in staff understanding of their roles and confusion about the protocol may not be readily apparent or easily communicated. Changing the culture of an ED is not easy or straightforward. To address this important issue, we must take a long-term approach to the problem. Progress must be reinforced at every step, and our approach must be constantly reassessed. Furthermore, we must be willing to redesign our protocol as needed to make it effective. Finally, unless an institution is willing to commit longterm additional personnel and other resources to address the issue of domestic violence, it is essential that an ED protocol for identification and referral be designed to be a part of existing ED clinical practice.
The authors thank the following people for their contributions to the design and development of the domestic violence screening and referral protocol and the validation procedure described in this article: Sandra Martin, Department of Maternal and Child Health, University of North Carolina School of Public Health; and Greg Johnson, Department of Nursing; Dexter Morris, Department of Emergency Medicine; Lee Marazas, Connie Culbreth, and Kathy Kalanyos, Department of Social Work; and Tom Smith and Richard Brickey, Hospital Police, University of North Carolina Hospitals. The authors also thank Alison Hilton of the University of North Carolina Department of Health Behavior and Health Education for her work interviewing ED staff about their experiences.
Reprint no. 47/1/73922
Address for reprints: Anna EWaller, ScD Department of EmergencyMedicine Universityof North Carolina CB7594 Chapel Hill, North Carolina27599-7594
REFERENCES 1. ResenburgM, FinleyMA: Violencein America: A Public Health Approach. New York: Oxford UniversityPress,1991. 2. The NationalCommitteeon InjuryPreventionand Control:Injury Prevention: Meeting the Challenge. New York: OxfordUniversityPress,1989. 3. RandallT: Domesticviolenceinterventioncalls for morethan treating injuries.JAMA 1990;264:939-940. 4. AmericanMedicalAssociation:Diagnosticand treatmentguidelineson domesticviolence. Arch Fam Med 1992;I:39-47.
760
ANNALS OF EMERCENCY MEDICINE 27:6
JUNE 1996