Annals of Emergency Medicine, Journal of the American College of Emergency Physicians
August 1997, Volume 30, Number 2, p. 190-197


Amy A Ernst, MD
Department of Emergency Medicine, Vanderbilt University, Nashville, TN
Todd G. Nick
School of Health Related Professions, University of Mississippi Medical Center, Jackson
Steven J Weiss, MD
Department of Emergency Medicine, Vanderbilt University, Nashville, TN
Debra Houry
Tulane University School of Medicine
Trevor Mills, MD
Section of Emergency Medicine, Louisiana State University, New Orleans, LA



Study objective: To determine the prevalence of domestic violence (DV) for male and female ED patients and to determine the demographics of DV.
Methods: The study design was a descriptive written survey of adults. We used the Index of Spouse Abuse (ISA), a previously validated survey tool. The study was set in an inner-city ED with approximately 75,000 patients annually, most of them indigent. Patients 18 years or older who were able to give consent were included. Patients were excluded if they had a language barrier, were a prisoner, or had never had a partner. All patients presenting during 31 randomly selected 4-hour shifts during July 1995 were considered for the study. DV was defined as either physical or nonphysical on the basis of ISA scoring. The prevalence was determined for present (in the preceding year) and past (more than 1 year ago) abuse. Four violence parameters were calculated for patients who had a partner at the time of presentation: present physical, present nonphysical, past physical, and past nonphysical. Only the past parameters were calculated for patients who had had a partner in the past but had no partner at the time of presentation. We used the 2 test to determine individually significant predictors of the four parameters. Logistic-regression models were constructed to determine the significant predictors of DV. Associations among the present physical, present nonphysical, past physical, and past nonphysical abuse categories were determined with McNemars test.
Results: We enrolled 516 patients, 233 men and 283 women. On the basis of ISA scoring, 14% of men and 22% of women had experienced past nonphysical violence (P=.02, men versus women), and 28% of men and 33% of women had experienced past physical violence (P=.35). Of the 157 men and 207 women with partners at the time of presentation, 11% of men and 15% of women reported present nonphysical violence (P=.20), and 20% of men and 19% of women reported physical violence (P=.71). Using logistic-regression models, we determined that women experienced significantly more past and present nonphysical violence but not physical violence than men. For all four parameters, the victim's suicidal ideation and alcohol use were independently associated with DV. The victim's family history was strongly associated with past abuse. Using McNemar's test, we found that physical and nonphysical abuse were correlated in the past and present.
Conclusion: Using a validated scale, we found that the prevalences of physical DV for men and women are high and that they are not statistically different in this population. Using 2 testing, we found that women had experienced significantly more past nonphysical violence than men; using logistic regression we found that they experienced significantly more nonphysical violence (both past and present) than men. DV was frequently associated with suicidal ideation, alcohol use, and family history of violence.
[Ernst AA, Nick TG, Weiss SJ, Houry D, Mills T: Domestic violence in an inner-city ED. Ann Emerg Med August 1997;30:190-197.]


Domestic violence (DV) is defined as abuse involving two adults involved in an intimate relationship. The perpetrator may use physical, verbal, emotional, psychologic, economic, or sexual means to manipulate the victim. The abusive incidents are rarely isolated, often following a pattern of increasing frequency and severity.(1,2)

During the last three decades, public awareness of DV has increased dramatically. Child, spouse, and elder abuse have all become major public health issues, and spouse abuse has now become synonymous with DV.

One of the barriers to identifying victims of DV is the paucity of known associations. Sex, race, education, economic status, disability, substance abuse, history of abuse, and pregnancy have all been examined.(3,4) Probable associations include history of family violence; alcohol use by the male batterer, the female victim, or both; and a current relationship involving abuse.(4) Although earlier studies indicated that most of the perpetrators of DV were male,(3) some data suggest that the frequencies of male and female victims of domestic DV are equal.(5,6) The existence of male victims of DV also is being acknowledged.(7-9)

The purpose of this study was to evaluate the scope and characteristics of DV, to compare men and women presenting to an inner-city ED, and to determine the importance of selected associations in the patient. Our null hypothesis was that men and women in this population are equally likely to have experienced abuse. We also believed that previously determined historical correlations of DV would be generalized in this population.


A confidential written survey was conducted in New Orleans at the ED of Charity Hospital, an urban Level I trauma center with an annual census of approximately 75,000. The hospital is an inner-city institution, mainly serving the indigent and medically underserved. The questionnaire was designed to screen for a history of DV and several demographic associations. The study was approved by the Louisiana State University Institutional Review Board.

In randomly selected 4-hour blocks on each day during July 1995, the questionnaire was offered to all men and women older than 17 years who were able to give consent. The time blocks were midnight to 4 AM, 4 to 8 AM, 8 AM to noon, noon to 4 PM, 4 to 8 PM, and 8 PM to midnight. A total of 124 hours of surveillance were conducted in the ED. We excluded patients on the basis of the following criteria: (1) if the patient was in extremis; (2) if the patient was unable to give consent because of psychiatric, medical, or traumatic illness; (3) if the patient had taken a drug overdose or was alcohol intoxicated; (4) if the patient was incarcerated; (5) if the patient had a language barrier; (6) if the patient had participated in this survey during a previous visit; if the patient refused to participate with, had deserted, or had never had a partner. A log was kept of patients who presented to the ED, those patients eligible for the study, whether they participated or declined to participate, and reasons for not participating.

Two researchers with no other duties at the time of surveying consistently administered the questionnaire. All adult patients presenting to the ED during the designated time blocks were invited by one of the researchers to participate in a "study about domestic violence" as part of the routine triage system. The area in which the patients were asked about the survey was strictly limited to patients unless the patient was younger than 18 years. Visitors and relatives were excused from the room, the study was explained by the researcher, and written informed consent was obtained. Patients were assured that the questionnaire was confidential. Referrals to social services were offered to everyone and given on completion of the questionnaire if the patient requested assistance. The reasons for presentation were determined according to the patient's chief complaint. The diagnosis of DV was determined as the reason for presentation if the patient agreed that he or she was in the ED because of DV. This definition was not explained to the patient before participation in the study.

Each subject was asked to complete the 30-item ISA questionnaire and the 24-item demographic survey with the assistance of the study coordinator. The ISA is a validated survey tool with questions concerning both acute and non-acute DV.(10) The scale was validated by the authors, who used discriminant and construct validity tools. It was validated for both physical and nonphysical parts of the scale. Individual items in the ISA are weighted to produce a final score between 0 and 100 for both physical and nonphysical violence. Cutoff values of 10 for physical abuse and 25 for nonphysical abuse have been shown to have a sensitivity of 92.2% and a specificity of 90.7%.(10) All 30 questions were asked twice, once about physical and nonphysical DV occurring in the preceding year and once about physical and nonphysical DV occurring before the past year. If the subject had no present or past partner, this was documented, and corresponding sections of the survey were excluded from analysis.

For marriage, we grouped the patients into one of the following three categories: (1) single, (2) married, and (3) separated or divorced. For race we had two categories (1) black and (2) white or other. Insurance status was classified on the basis of any versus none and sex as male versus female. The other parameters were based on the presence or absence of the variable in question, such as employment, suicidal ideation, and alcohol use. The study results were reviewed and calculated after the patient left the ED.

Past nonphysical violence, past physical violence, present nonphysical violence, and present physical violence were qualitatively determined as positive or negative with the use of the ISA cutoffs. We used the 2 test to determine whether the rates of a characteristic's presence, such as alcohol use, were the same for men and women and to determine whether univariate comparisons of the proportions of patients who experienced DV were the same for men and women. McNemar's test was used to compare past and present abuse. We used the -statistic for percent agreement adjusted for chance.

To allow for several explanatory factors, logistic-regression models were used for each of the four abuse responses. Logistic-regression models quantify the effect of a candidate predictor in terms of an odds ratio (OR), which is a natural description of an effect in a probability model because an OR can be constant. We used multiple logistic-regression models with 10 candidate predictor variables for the present DV models, with one additional candidate predictor for the past DV models, whether the patient had had a partner in the preceding year. ORs in logistic-regression describe the effect of a treatment independent of covariables that affect the risk of DV. Predictor variables were selected carefully on the basis of the literature. The following variables were included: sex, race, suicidal ideation, family history of DV, marriage, alcohol use, drug use, insurance status, disability, tranquilizer use, and having had a partner during the preceding year.

A two-group 2 test with a two-sided significance level of .05 had at least 80% power to detect the difference between an OR of 2.5 for abuse in men versus women. For past violence, the sample sizes were larger; hence the power to detect a significant OR of 2.5 was at least 90%.


During 31 randomized 4-hour shifts, 549 adults who were able to give consent presented to the ED. Of these, 28 refused to participate, 4 had no past or present partner, and 1 left before completing the survey.

Of the 516 included in the study, 283 (54.8%) were women and 233 (45.2%) were men. Of the participants, 207 women and 157 men had a current partner so that present violence data could be obtained, and 76 women and 76 men had no current partner so that only past parameters could be determined. The mean age of both women and men was 35 years, with an SD of 13 for women and 11 for men. A description of the group's racial mix, marital status, reason for presentation, insurance status, and number of children is presented in Table 1.

Table 1. Population demographics.
Demographic DataNo. (%)
Marital status
Married100 (19.5)
Divorced57 (11.0)
Separated31 (6.0)
Widowed18 (3.5)
Single310 (60.0)
Ethnic origin
Black424 (82.2)
Hispanic6 (1.2)
White85 (16.5)
Other1 (.2)
Reason for presentation
DV22 (4.3)
Medical301 (58.3)
Trauma134 (26.0)
Surgical27 (5.2)
Obstetrics/gynecology27 (5.2)
Psychiatric4 (.8)
Unknown1 (.2)
Insurance status
None419 (81.8)
Medicaid36 (7.0)
Medicare10 (2.0)
Private42 (8.2)
Veterans5 (1.0)
No. of children
None158 (30.7)
One101 (19.5)
Two101 (19.5)
More than two156 (30.3)

Alcohol use, past suicidal ideation, family history of violence, disability, tranquilizer use, drug use, and unemployment were associations compared in men versus women. These results are summarized in Table 2.

Table 2. Frequency and percentage of patients with historical associations.
Historical AssociationsFemale % [n=283]Male % [n=233]P*
Alcohol use17 (6)48 (20).001
Suicidal ideation in past31 (11)36 (15).13
Family history of DV35 (12)26 (11).66
Disability3 (1)14 (6).002
Tranquilizer use22 (8)11 (5).16
Drug use13 (5)38 (16).001
Unemployment139 (49)143 (61).006
*2 test.

We found a significant difference in the number of women versus men who reported past abuse to the police (19% for women and 6% for men; P=.001, 2). Patients were asked where they would seek assistance if they were victims of DV. One hundred sixty-one (31.2%) reported that they would seek help from the police, 143 (27.7%) said they did not know, 76 (14.7%) said they would go to a hospital, 55 (10.7%) said they would approach a family member, 55 (10.7%) said they would go to a shelter, 16 (3.1%) said they would forego assistance and simply retaliate, and 10 (2%) gave other responses.

We compared the occurrence of past nonphysical and physical violence, determined with the ISA, in men versus women with the use of the 2 test. There was a significant difference in the amount of past nonphysical violence in women versus men (P=.02). There was no difference in past physical violence in men versus women (P=.35). These results are summarized in Table 3.

Table 3. Percentage of men and women reporting past and present violence.*
Violence HistoryMen (%)Women (%)P
Past nonphysical33/233 (14)63/283 (22).02
Past physical66/233 (28)91/283 (32).35
Present nonphysical17/157 (11)32/207 (15).2
Present physical32/157 (20)39/207 (19).71
*From population of 233 men and 283 women who had ever had partners and 157 men and 207 women with current partners.
2 test.

McNemar's test for paired categorical data revealed an association between past nonphysical violence and past physical violence (P=.001). These responses were in agreement 62% of the time (adjusted for chance using the statistic).

Present nonphysical and physical violence, determined with the ISA, was also compared in men versus women. We found statistically significant differences in the amount of present nonphysical violence (P=.2) or present physical violence (P=.71) for men versus women. These results also are summarized in Table 3. McNemar's test for paired categorical data revealed an association between present non-physical violence and present physical violence (P=.001). These responses were in agreement 66% of the time (adjusted for change with the -statistic).

Past physical violence was associated with present physical violence (McNemar's P value=.02, =.85), with 85% agreement adjusted for chance. McNemar's test revealed an association between past nonphysical violence and present nonphysical violence (P=.007). These responses were in agreement 88% of the time (adjusted for chance with the -statistic).

In the logistic-regression model one patient was excluded because of incomplete data, so the models were based on a total of 515 patients. Using multiple logistic-regression models, we found significant associations between past non-physical violence and female sex, suicidal ideation, family history of violence, marital status (separated and divorced), alcohol use, having had a partner in the last year, and white race (Table 4).

Table 4. Significant predictors of past nonphysical violence as determined with multiple logistic regression.
VariablesOR95% CIP
Sex2.781.55, 4.98<.001
Suicidal ideation3.471.83, 6.59<.001
Family history2.871.57, 5.61<.01
Marriage2.301.27, 4.15<.001
Alcohol2.591.27, 5.28<.001
Partner in preceding year.55.32, .95.03
Race1.961.06, 3.62.03
Drug use, insurance status, disability, and tranquilizer use were not significantly associated with past nonphysical violence.

Disability, drug use, insurance status, and tranquilizer use were not associated with past nonphysical violence.

Using a multiple logistic-regression model, we found that significant predictors of past physical violence were suicidal ideation, family history of DV, alcohol use, and disability (see Table 5) for a listing of the ORs, confidence intervals, and P values). There were no significant associations between past physical violence and sex, drug use, race, tranquilizer use, insurance status, marriage, or having had a partner in the last year.

Table 5. Significant predictors of past physical violence as determined with multiple logistic regression.
VariablesOR95% CIP
Suicidal ideation4.492.42, 8.32<.001
Family history2.601.35, 5.02<.01
Alcohol2.591.27, 5.26<.01
Disability4.111.15, 14.72.03
Sex, race, drug use, insurance status, tranquilizer use, having a partner in the preceding year, and marriage were not significantly associated with past physical violence.

Using a multiple logistic-regression model, we found that significant predictors for present nonphysical violence were female sex, suicidal ideation, alcohol use, and disability (Table 6).

Table 6. Significant predictors of present nonphysical violence as determined with multiple logistic regression.
VariablesOR95% CIP
Sex2.711.17, 6.23.02
Suicidal ideation5.432.35, 12.86<.001
Alcohol2.861.09, 7.48.03
Disability5.761.14, 29.2.03
Race, drug use, insurance status, tranquilizer use, family history, and marriage were not significantly associated with present nonphysical violence.

There was no association of present nonphysical violence with family history, tranquilizer use, race, drug use, insurance status, or marriage.

Using a multiple logistic-regression model, we found that alcohol use and suicidal ideation were significant predictors for present physical violence. Sex was not significantly associated with present physical violence (P=.36). These results are summarized in Table 7.

Table 7. Significant predictors of present physical violence as determined with multiple logistic regression.
VariablesOR95% CIP
Suicidal ideation5.472.57, 11.66<.001
Alcohol4.061.81, 9.14<.001
Sex, race, drug use, insurance status, tranquilizer use, marriage, family history, and disability were not significantly associated with present physical violence.

We found no associations for present physical violence and disability, tranquilizer use, family history, marriage, insurance status, or drug use. A summary of all the predictors for DV is presented in Table 8.

Table 8. Summary of significant predictors of past and present violence as determined with multiple logistic regression.

VariablesPast NonphysicalPast PhysicalPresent PhysicalPresent Nonphysical
Suicidal ideation<.001<.001<.001<.001
Family history<.01<.01NSNS
Tranquilizer useNSNSNSNS

Many patients had combinations of past and present, physical and nonphysical violence. According to ISA scoring, 42 patients had experienced all four types. The overlaps among the four categories are summarized in Table 9.

Table 9. Overlap of physical and nonphysical violence, past and present.
Type of ViolencePresent Nonphysical (%)Past Physical (%)Past Nonphysical (%)
Present physical43 (8.3)68 (13.2)45 (8.7)
Present nonphysical 43 (8.3)48 (9.3)
Past physical  89 (17.2)

Pregnant women enrolled in the study experienced 2.5 times more incidents of nonphysical and physical violence than nonpregnant patients (95% confidence interval [CI], .506 to .414), but this finding was not statistically significant. Six men had pregnant partners; 50% of the men reported abuse.

Thirteen homosexual patients were enrolled in the study, 4 women and 9 men. Three of the four women had experienced past nonphysical and physical violence and one experienced present physical violence. Four of the men had experienced past abuse and two experienced present physical violence. More than 50% of the homosexual patients experienced some form of abuse, but the numbers were too small for statistical analysis.

Of the 22 patients who presented for what was described by the patient as a DV-related incident, 13 had positive findings for all four parameters, 2 had present nonphysical and past physical DV, 2 had past and present physical violence, 1 had present physical violence, 1 had present physical and nonphysical violence, and 3 had no DV according to ISA scoring. These three involved an isolated incident of being struck.

All patients were told that a referral to social services could be arranged if the patient desired. Twelve patients requested referrals. Because of the anonymous nature of this study and consent issues, follow-up could not be conducted in the patients who obtained referrals.


Despite the magnitude of the problem of DV, identification of victims of abuse is still a complex diagnostic dilemma. Through retrospective analysis, it is estimated that only 5% to 10% of all female victims of abuse are detected in the ED.(2,4) In our study only 4.3% of patients admitted that they had presented to the ED for DV-related incidents. This figure is most likely an underrepresentation. Sex-related differences in reporting may be a factor in the numbers of patients presenting for DV. There may be differences between men and women in their perceptions of DV. It is known that female victims of DV may not define pushing, shoving, threats, or verbal abuse as DV; however, men may differ in this interpretation. The ISA, a standardized questionnaire, should correct for any varying opinions about the occurrence of DV on the part of the subject or the person administering the survey. We did not ask patients whether they believed they were victims of DV in the past or present. The results were based solely on ISA scoring.

This study is unique because of its use of the ISA in determining abuse among ED patients.(10) Whether victims of DV are aware of the victimization in their lives is uncertain. A survey such as the ISA, which is standardized and previously validated, addresses the question of whether abuse exists without requiring a direct opinion of the patient as to whether violence is present in his or her home. The original survey was validated in 64 known victims of spouse abuse and compared with results from 43 known not to be involved in an abusive relationship. The ISA was compared with other scales that had been validated and were available at the time.(10)

Studies that simply ask direct questions of patients may underestimate or overestimate the prevalence of DV.(5,11) In fact, in our study 3 of the 22 patients who presented for DV-related incidents were not victims, according to the ISA. Looking at the three cases individually, each could be considered an isolated incident and was not determined by the ISA to be ongoing abuse. DV entails repeated physical or nonphysical abuse.

Another unique aspect of our study is that both male and female ED patients were surveyed. The authors of many previous studies have surveyed only women.(11-13) Other studies including men have not used validated entry criteria(5) or have used nonrandom inclusion criteria.(14) In an Australian ED, when questions were directed toward the patients understood occurrence of DV, 23.3% of women and 6.3% of men claimed they were victims of DV. This study was recently repeated by the same group in the same setting, and similar results were obtained.(15) This difference may represent the lack of validity in the measurement tool or a difference in cultures.(14,15) It might also represent a difference in study design, which may more closely reflect the prevalence of DV than the prevalence of physical assault episodes. We tried to eliminate this problem with the use of the ISA.

In our study, rates of physical DV were high and nearly equal between men and women. This is supported by prior studies of DV toward men, which demonstrated that DV by women directed against men is a major problem. Among ED patients, Goldberg et al (5) found no statistically significant differences between male and female victims of DV. The authors of other studies noted decreases in rates of male DV directed against women, with a much smaller decrease in female DV directed against men, suggesting awareness of only half the problem between 1975 and 1985.(8,16)

In this study, women reported DV to police more often than men, even though the ISA determined that rates were similar for men and women. This may be explained by the more potentially injurious nature of male versus female violence, or it may be because women have been better educated than men about the importance of reporting violence. There is an expressed fear among some that recognition of DV by women directed against men will deemphasize the importance of providing services to women.(8,9) Recognition of the global nature of violence may be more realistic than assuming that only women are victims.

Using logistic-regression analysis, we found that demographic factors related to DV included drug, alcohol, and tranquilizer use, family history of DV, suicidal ideation, and psychiatric history. These have been positively reported associations with DV in previous studies.(2-4)

According to our logistic-regression analyses, sex was related in past and present nonphysical violence; however, according to our 2 data, sex was not a factor in present nonphysical violence. This difference may have arisen because multivariate analysis determines factors independently associated with the outcome of interest adjusting for correlations with other independent variables. In the case of present nonphysical violence, other positive associations were suicidal ideation, alcohol use, and disability of the victim. One or all three of these associations may have affected the results attained for sex. A higher occurrence of nonphysical violence in women is supported by other ED DV surveys.(14)

Using both 2 testing and logistic-regression analysis data, we found that sex was not associated with past or present physical violence. The reasons for this finding in our patient population are unknown.

Many patients were victims of more than one aspect of domestic violence. Out of the total 516 patients in our study, 42 (8%) were victims of past and present physical and nonphysical violence, 68 (13%) were victims of past and present physical violence, and 48 (9%) were victims of past and present nonphysical violence. This demonstrates that because violence appears to be perpetuated over a long period in many cases, more needs to be done to assist and educate the victims.

The ED may be an important place for such education. In a previous Canadian study of the ED as a site for DV intervention, it was found that follow-up could be made in only 50% of cases of known DV.(29) Many battered people and their families use the ED as their sole source of medical care, providing ED professionals the opportunity to assist and educate the victims, especially with regard to sources of assistance.

One limitation of this study is that it was performed at a single institution with a large, indigent black population. It may not be possible to extrapolate these results to other areas. Another limitation is that the study is a survey, subject to patient interpretation, although the survey was administered by two members of the study group for consistency to minimize this effect. Another limitation is that those who could not give consent because of severe illness or age less than 17 years were not included, many of whom may have been victims. Because we used the ISA, which requires approximately 10 to 20 minutes of a patients time, many patients with severe illness were excluded. Approximately one third of those excluded had mental illness or alcohol/drug intoxication, rendering them unable to answer an extensive survey, and the other two thirds had extreme medical or surgical illnesses. It is not known whether the patients who refused to participate were actually victims of DV. However, the numbers were relatively small compared with the group surveyed.

In conclusion, DV rates were high in our population, with nearly equal rates of past and present physical violence for men and women. Alcohol, drug use, and suicidal ideation are important factors associated with all forms of DV. Further education about where to find assistance when DV is present is needed.


1.  Walker LE: The Battered Woman. New York: Harper Perennial, 1979.

2.  Salber PR, Taliaferro E: What do we know about domestic violence?, in Brown Z (ed): The Physicians Guide to Domestic Violence. Volcano City, CA: Volcano Press, 1995.

3.  Council on Ethical and Judicial Affairs, American Medical Association: Physicians and domestic violence. JAMA 1992;267:3190-3193. [MEDLINE]

4.  American Medical Association: American medical association diagnostic and treatment guidelines on domestic violence. Arch Fam Med 1992;1:39-47. [MEDLINE]

5.  Goldberg WG, Tomlanovich MC: Domestic violence victims in the emergency department. JAMA 1984;251:3259-3264. [MEDLINE]

6.  Gelles RJ: Violence in the family: A review of research in the seventies. J Marr Fam 1980;42:873-885.

7.  Steinmetz SK: The battered husband syndrome. Victimology 1977-1978;2:499-509.

8.  Straus MA, Gelles RJ: Societal change and change in family violence from 1975 to 1985 as revealed by two national surveys. J Marr Fam 1986;48:465-479.

9.  Hamberger LK, Potente T: Counseling heterosexual women arrested for domestic violence: Implications for theory and practice. Violence Vict 1994;9:125-137. [MEDLINE]

10.  Hudson WW, McIntosh SR: The assessment of spouse abuse: Two quantifiable dimensions. J Marr Fam 1981;43:873-885.

11.  Abbott J, Johnson R, Koziol-McLain J: Domestic violence against women: Incidence and prevalence in an emergency department population. JAMA 1995;273:1763-1777. [MEDLINE]

12.  Appleton W: The battered woman syndrome. Ann Emerg Med 1980;9:84-91. [MEDLINE]

13.  McLeer SV, Anwar R: A study of battered women presenting in an emergency department. Am J Public Health 1989;79:65-66. [MEDLINE]

14.  Roberts GL, OToole BI, Lawrence JM: Domestic violence victims in a hospital emergency department. Med J Aust 1993;159:307-310. [MEDLINE]

15.  Roberts GL, OToole BI, Raphael B: Prevalence study of domestic violence victims in an emergency department. Ann Emerg Med 1996;27:747-753.

16.  Straus MA. Sexual inequality, cultural norms, and wife-beating. Victimology 1976;1:54-76.

17.  American Medical Association Council on Scientific Affairs. Violence against women: Relevance for medical practitioners. JAMA 1992;267:3184-3189. [MEDLINE]

18.  Saunders DG: When battered women use violence: Husband abuse or self-defense? Violence Vict 1986;1:47-60. [MEDLINE]

19.  Randall T: American Medical Association joint commission urge physicians to become part of solution to family violence epidemic. JAMA 1991;266:2524-2527. [MEDLINE]

20.  Marwick C: Health and justice professionals set goals to lessen domestic violence. JAMA 1994;271:1147-1148. [MEDLINE]

21.  CDC: Emergency department response to domestic violence: California 1992. JAMA 1993;270:1296-1297. [MEDLINE]

22.  Dickstein LJ: Spouse abuse and other domestic violence. Psychiatr Clin North Am 1988;11:611-628. [MEDLINE]

23.  Fitch F, Papantonio A: Men who batter: Some pertinent characteristics. J Nerv Ment Dis 1983;171:190-192. [MEDLINE]

24.  Straus MA: Measuring intrafamily conflict and violence: The conflict tactics (CT) scales. J Marr Fam 1979;41:75-88.

25.  McFarlane J, Parker B, Soeken K: Assessing for abuse during pregnancy. JAMA 1992;267:3176-3178. [MEDLINE]

26.  Helton AS, McFarlane J, Anderson ET: Battered and pregnant: A prevalence study. Am J Public Health 1987;77:1337-1339. [MEDLINE]

27.  Hamberger LK: Domestic partner abuse: Expanding paradigms for understanding and intervention. Violence Vict 1994;9:91-94. [MEDLINE]

28.  Letellier P: Gay and bisexual male domestic violence victimization: Challenges to feminist theory and responses to violence. Violence Vict 1994;9:95-106. [MEDLINE]

29.  Hotch D, Grunfeld AF, Mackay K: An emergency department-based domestic violence intervention program: Findings after one year. J Emerg Med 1996;14:111-117. [MEDLINE]