One important element in developing the health care response to intimate partner violence (IPV) is to help physicians and nurses understand and meet the needs of the individual battered woman when she presents for health care. Racial and ethnic background is one contextual factor that may influence the physician–patient relationship. Prior research suggests that race may affect battered women’s experiences with the health care system (McNutt et al. 2000). In addition, a substantial body of literature has demonstrated racial differences, and disparities, in the health care system response to people with various other medical conditions. Understanding racial differences in battered women’s experiences will help physicians anticipate individual patients’ unique needs. This study explores White and African American battered women’s experiences with the health care system, and treatment preferences.

The significance and impact of impact of IPV on the health care system and the health of individual women is no longer debated. In a comprehensive review, Hamberger and Phelan (2004) reported IPV prevalence ranges of 25 to 44% in family medicine settings, 5.5 to 14% in internal medicine, 3.8 to 16% in obstetrics during pregnancy and 10 to 15% during the pregnancy year. In emergency medicine, prevalence of IPV ranges from 12 to 52%. Hamberger and Phelan (2004) also reported that high prevalence of battered women has been documented in ophthalmology clinics, infectious disease clinics, pediatric clinics, and physical medicine/rehabilitation settings.

Partner violence is a significant cause of morbidity and mortality, resulting in reduced health status (Wagner and Mongan 1998), more injury (Kyriacou et al. 1999), infectious diseases (Coker et al. 2000), and depression and other emotional problems (Gleason 1993). Victims of IPV use the health care system more often (Bergman and Brismar 1991) and have higher health care costs (Wisner et al. 1999).

Several initiatives have focused on training physicians to identify and respond effectively to IPV victims. In general, it has been found that the most effective programs combine IPV training with systems interventions such as the establishment of clinical protocols and use of chart reviews to monitor compliance, troubleshoot and resolve problems (Waalen et al. 2000). To establish an empirical basis for training that is responsive to the needs of battered women, Hamberger et al. (1998) and Rodriguez et al. (1996) have called for including the voices of consumers. Research from these two laboratories found that battered women value being (a) asked privately about domestic violence, (b) provided with emotional support when they disclose violence and abuse, (c) given thorough psychosocial and physical examinations of their injuries, (d) provided with community resource information, and (e) given help developing a safety plan. Hamberger et al. (1998) found that such patient preferences were similar for battered women regardless of their past history of seeking help in the health care system for abuse-related problems, and regardless of educational level, occupational status, cohabitation status, or age.

A significant unresolved issue in this research is whether there are racial differences in patient preferences. Hamberger et al. (1998) found that African American women rated some physician behaviors differently than did White women. In particular, African American women rated as neutral (compared to high negative ratings by White women) such physician behaviors as sympathizing with the abuser, suggesting the patient change her behavior to stop the abuse, and that women are at least partially responsible for abuse against them. Hamberger et al. (1998) suggested that these racial differences were due to the fact that more African American women reported having received health care from emergency departments than White women, who reported mostly receiving medical services from private physicians. Racial differences were not predicted, however, and the study design was limited for a number of reasons, including Type I error, no control for potential confounding variables such as socioeconomic status, insurance status, and access to care. Finally, racial differences were observed for relatively few items (10 out of 94), and the differences were one of degree not direction, as noted above. In no instance did African American women rate a physician behavior as positive and White women rate the item as negative. Hamberger, et al. concluded that additional research was necessary to explore racial differences in the preferences of battered women.

Race impacts the physician–patient relationship in a number of ways. African American and other ethnic minority patients, compared to White patients, seem to have different patterns of health care access. Racial or ethnic minority patients and patients who are poor are less likely to obtain health care from a consistent primary care physician and more likely to receive care in an urgent or emergency care setting (Grumbach et al. 1997; Komaromy et al. 1996). African American patients are less likely to experience continuity of care from a doctor who provides services more closely aligned to patient expectations (Cornelius 1997). In addition, Fitchenbaum and Gyimah-Brempong (1997) present evidence that the observed racial differences are due to differential site of care, cultural differences in use of health care services, and possible differential treatment from health care providers.

Having a doctor of the same racial or ethnic background as the patient may enhance quality of medical care due to increased trust from a shared cultural identity and experience, as well as shared communication and expectations for care (Rhee et al. 1979). Rhee et al. (1979) found clear evidence for racial mutual selection between patients and physicians among Asian American patients, though no difference in treatment and quality of care. Sleath et al. (2000) found doctors were equally likely to make empathic expressions both within and across Hispanic-Caucasian doctor–patient combinations, but physicians made more positive expressions to non-Hispanic than to Hispanic patients.

Regarding doctor–patient communication with African American battered women, Bent-Goodley (2004) suggests that a physician assigning some responsibility for battering to the woman patient might resonate with her cultural experience of shouldering major responsibility to maintain the family and protect her partner, regardless of the cost to herself out of “racial loyalty” (p. 309). Likewise, physician expressions of even a modicum of sympathy with the batterer might touch the woman’s awareness of societal oppression of African American men, and her own sympathy for and loyalty to him. The cost of such loyalty may be increased and continuing violence, as well as less sensitive and competent medical services, as providers respond to the stereotype of the “strong African American woman who can sustain anything, has no fear, and can easily protect herself” (Bent-Goodley 2004, p. 309).

The above review suggests that experiences of African American and other ethnic minority women with the health care system is complex but in some ways different than that of White women, particularly in the types of health care settings available and to which they turn for services. Minority women may experience different interpersonal interactions with their physician providers than White women in the form of fewer positive comments, even though overall quality of care received may, objectively, be similar across racial and ethnic groups. Moreover, subtle differences in doctor–patient communication across racial groups may be experienced as less satisfactory and result in lowered expectations for quality of care for minority women.

In the present study African American and White battered women were surveyed about their interactions with physicians when seeking services for abuse related problems, and rated specific physician behaviors as either positive or negative. This is a conceptual replication of Hamberger et al. (1998). However, the present study includes a number of changes in research design to more carefully evaluate the potential impact of race on patient ratings of physician behaviors. First, to facilitate a more in-depth analysis of racial differences, African American women were intentionally oversampled. Second, to evaluate the impact of potential confounding variables, indices of socioeconomic status, including educational attainment, employment status, and family income were obtained. Third, to more carefully assess the role of access to care, questions were included that assessed the respondent’s source of care, including emergency department, urgent care clinic, and private physician’s clinic. Fourth, to assess racial and gender concordance, respondents were asked about the perceived racial identity and gender of their physician provider and their comfort level with that doctor–patient racial and gender combination.

Three hypotheses regarding racial differences were studied. Compared to White women, African American women were predicted to:

  1. 1.

    Report receiving more care for abuse related problems from urgent (walk-in) care and emergency department sources.

  2. 2.

    Respond differently to physician behaviors that express some sympathy with the woman’s partner or attribute responsibility for some aspects of the violence or for ending it to the woman.

  3. 3.

    Report more comfort with a physician identified as similar in racial and gender identity.

Materials and Methods

Participants

Participants were 132 women who had been abused by an intimate partner and were obtaining services from one of five battered women’s programs in southeastern Wisconsin. These agencies are the sole providers of battered women’s services in their respective locales, with the exception of one located in a major city that has several agencies. The participating agencies serve a diverse population of women encompassing urban, suburban and rural residence, as well as varied socioeconomic status and racial and ethnic identity. To assess racial differences with groups of roughly equal numbers, African American participants were intentionally oversampled. Although some women who were initially recruited declined to participate in the study, it was not possible to maintain records of participation rates, or compare participants and nonparticipants.

Two participants who chose both African American and another racial category were classified as African American and one individual who chose both white and American Indian was classified as American Indian. Surveys completed by two participants were discarded as these respondents showed no variation in ratings for 94 physician behaviors, suggesting that their responses were not valid.

Instruments Used

The three instruments used in the present study were nearly identical to those used by Hamberger et al. (1998). The first, a background questionnaire, queried age, race and ethnicity, employment status, educational attainment, employment status, and family income.

The second questionnaire, the Conflict Tactics Scale (CTS, Straus 1979, 1990), was used to assess violence experienced by the respondent in intimate relationships, both throughout her life and in the past 12 months.

The third questionnaire, the Physician Assessment and Treatment of Abuse Inventory is a validated survey (PATAI, Hamberger et al. 1998) that contains 94 descriptions of helpful and unhelpful behaviors that a physician may engage in while treating a woman for IPV. Participants rate the behaviors on a 5-point Likert-type scale indicating the desirability of the behavior, with 1 denoting highly undesirable, 3 indicating neither desirable nor undesirable, and 5 indicating highly desirable physician behavior. Respondents also indicated, yes or no, whether they had encountered each behavior the most recent time they sought medical care for abuse related problems.

Procedure

Procedures followed closely those reported by Hamberger et al. (1998). The research assistant collaborated with representatives of the women’s programs to recruit survey respondents. All prospective participants were read a description of the study. Participants were assured that their participation was voluntary, and anonymous, and that participation was independent of the agency’s program, and that declining to participate would not affect the services they received. Those agreeing signed a written consent form. No respondent received compensation for participation.

Participants then completed the instruments. Because surveys about violence can cause a strong emotional reaction, participants were offered assistance if they were to experience distress from survey completion. No participants requested assistance.

Data Analysis

Percentages were calculated for categorical survey items such as demographics, abuse experience, location of treatment, and sex and racial concordance in the doctor–patient relationship. Means and medians were computed for the ratings of physician behaviors and univariate comparisons between White and African American women were tested using the Wilcoxon rank-sum test. Fisher’s exact test and an exact version of the Kruskal–Wallis test were used for categorical data analyses. Multivariate analyses of ratings of physician behaviors for African Americans versus Whites, controlling for education, income, care for physical injuries and exclusive use of the ER for physical injuries, was done by linear regression. In order to remove the influence of response style and to normalize ratings to meet linear regression assumptions, ratings were standardized, a constant was added to ensure that all values were positive and then these values were transformed by the Box–Cox method. Analyses were done using Stata and StatXact statistical software (StataCorp 2003; CYTEL Software Corporation 2001).

Results

Demographics

One hundred thirty-two women participated in the study. Their mean age was 32 years with a range of 18 to 56 years. Thirty-nine percent reported their race as African American, 51% as White and 10% as other or unknown. Just over half of the participants (59%) had received a high school diploma or less education and 36% came from households with an income less than $10,000 per year. The vast majority of respondents were not currently married, and not living with an intimate partner—18% were currently married and 10% were cohabiting, 38% were separated or divorced, and 30% living alone and never married. Slightly over half (53%) report currently working outside of the home.

Violence History

Many of the women reported quite severe violence, as measured by their maximum on the CTS. Twenty-eight percent reported having been strangled, 35% threatened with a knife or gun, and 16% assaulted with a knife or gun. There was no significant difference between African American and White women in the severity of abuse experienced (Kruskal–Wallis test, p = 0.08), though twice as many African American women experienced assault with a knife or gun (23 vs. 10%).

Source of Care

The majority of women reported that they had never sought care for abuse related physical injuries (59%), emotional support (78%) or abuse related stress (66%). There was no significant difference in the proportion of women who sought care by race. Among those women who reported a source of care for physical injuries, African American women were significantly more likely to report ever having gone to an emergency room for care for physical injuries than White women (91% = 21/23 vs 60% = 15/25, Fisher’s exact test, p = 0.02). We also compared the African American and White women in their exclusive use of emergency departments or walk-in clinics for abuse-related care. Seventy-four percent (23/31) of African American and 45% (18/40) of White women reporting a care location have utilized only an ER or a walk-in clinic for abuse-related care stemming from physical abuse, emotional abuse or stress related to living in an abusive relationship (Fisher’s exact test, p = 0.02).

Physician Behaviors

To test Hypothesis 2, we identified nine items from the PATAI that describe physician behaviors that express sympathy for the woman’s partner or attribute responsibility for the violence, or for ending it, to the woman (see Table 1). Separate multivariate linear regression analyses were conducted, controlling for education, household income, seeking care for physical injuries and seeking care only through emergency rooms, using the ratings for these nine items as the outcome variable. There was a significant difference between African American and White women for only two items, “The doctor acted as if you were to blame for being assaulted,” and “The doctor offered to talk to your partner/abuser about his reaction.” For both items, African American women were less definitive than White women in condemning such physician behaviors, although neither group of women, on average, endorsed these physician behaviors.

Table 1 Physician behaviors that express sympathy with the woman’s partner or attribute responsibility for the violence or for ending it to the woman
Full size table

Preference for Physician Characteristics

Table 2 shows the sex and race/ethnicity preferences for the African American and White women. Although a majority of women reported no preference in the sex of their physician, a significant minority (48% of Whites, 16 of 33; 40% of African Americans, 8 of 20) preferred a female physician, and very few preferred a male physician (6% of Whites; 5% of African Americans).

Table 2 Racial/gender concordance of patient and physician
Full size table

A majority of women also reported no preference for the race of their physician. However, a significantly higher proportion of White women reported a preference for someone similar to them in race/ethnicity (49% of whites; 15% of African Americans), and a significantly higher proportion of African American women reported no preference in the race/ethnicity of their physician (51% of whites; 80% of African Americans; Fisher’s exact test p = 0.01).

Discussion

The present study investigated racial differences in battered women’s treatment experiences and preferences when seeking medical care for abuse. The study tested three hypotheses—that, compared to White women, African American women would (a) report seeking more health care services for abuse related problems from emergency and urgent care settings than from primary care, (b) respond less negatively to physician behaviors that express some sympathy with the woman’s partner, or attribute responsibility for some aspect of the violence or for ending it to the woman, and (c) report more comfort with a physician identified as more similar to the patient in race and gender.

Overall, the hypotheses received mixed support. Hypothesis 1 received support. Among women who sought care for physical injuries, emotional support or stress, African American women were more likely than White women to access an emergency or urgent care setting. This finding is consistent with prior research, generally, which has shown that African Americans are more likely than Whites to seek care from emergency and urgent care settings (Grumbach et al. 1997; Komaromy et al. 1996). This finding has important implications for the development and implementation of health care based IPV prevention programs. Specifically, health care systems need to implement such interventions at multiple points of entry, primary care, urgent care, and emergency departments, or they will miss significant patient groups.

Hypothesis 2 was not supported. Using multivariate analyses to control for potential confounders such as education, income, and source of care, we identified two physician behaviors for which there was a statistical difference between White and African American women. However, this statistical difference is not clinically significant because African American and White women, on average, gave negative ratings to these physician behaviors.

Just as domestic violence crosses all racial, ethnic, and socio-economic boundaries, the values that battered women place on physician behaviors also cross racial boundaries. Battered women, regardless of racial background, want physicians to provide gentle, but competent physical exams and treatment, referral to community resources, and a supportive, nonjudgmental environment (Hamberger et al. 1998). Battered women, regardless of racial background, also want physicians to avoid victim blaming, sympathizing with the abuser, minimizing the abuse, or joking about the abuse (Hamberger et al. 1998).

On the other hand, the observed statistical difference in the present study may be consistent with prior research that has demonstrated racial differences in provider-patient continuity and communication patterns (Sleath et al. 2000). In particular, when we consider the findings in the context of the body of literature that explores the experience of women of color in the health care system, the observed statistical difference suggests that the experience of African American and White women may be somewhat different. We think that African American and White women ultimately want the same things from their physicians. However, the present findings also suggest that services need to be delivered in a culturally competent manner, centered on the needs of the individual patient, and based upon her cultural identity and lived experiences.

Hypothesis 3, that African American women would prefer physicians of the same or similar race and gender, was not supported. In fact, we found the opposite. Compared to White women, African American women, in general, were unconcerned about their physician’s race. Further research is needed to explore attitudes and beliefs of patients and providers about race to understand its impact on doctor–patient communication, patient comfort and satisfaction, and willingness to disclose IPV and seek help.

Very few African American and White women expressed a preference for a male physician, the vast majority of both groups expressing either no preference or preference for a female physician. These gender preferences are in line with prior research which has shown that battered women prefer female providers (Hayden et al. 1997). Other research has shown that medical practices with female providers identify more victims of domestic violence (Larkin et al. 1999). Because these are emerging as consistent findings in domestic violence and health care, more research is needed to understand how the presence or absence of female health care professionals affects identification of battered women.

Implications

Findings from the present study and from Hamberger et al. (1998), as well as other researchers (e.g., Rodriguez et al. 1996) have a number of implications for training and practice. First, it appears that battered women, regardless of race, place high value upon a number of physician behaviors in the treatment of abuse-related problems. These include, within a confidential setting, taking a complete history; thorough, careful and gentle physical exam; explanation of findings; treatment of all injuries; providing emotional support; and validation of her right to be safe. Further, battered women also value being provided with community resource information, assisted with development of a safety plan, and provided with follow-up visits (Hamberger et al. 1998). These “universals” should be built into training programs for health care providers.

Existence of such universals does not imply a cookie-cutter approach to service delivery. Although the present study found few racial differences, there was evidence that women of different racial and ethnic groups experience the health care system and interaction in subtly different ways. It will be necessary to develop training and practice guidelines that facilitate delivery of universally valued services in ways that are responsive to the needs of different cultural and racial groups. Our knowledge of this area is relatively rudimentary and further research is needed to clarify the content and process of delivery of such services to different patient groups.

A further implication of the present study relates to where health care services are delivered and received. We observed that African American women were more likely to receive their abuse-related services from emergency and urgent care settings. This suggests that different populations use different points of access to the health care system. Therefore, health systems should implement screening and intervention programs at multiple points of entry. In that regard, Minsky et al. (2005) found that different points of care face different challenges and opportunities for screening and helping abuse victims. Thus, it will be important to tailor both professional in-service and continuing education programs to the unique culture of individual points of access. Such educational interventions must address both system issues that facilitate and/or obstruct competent and consistent patient intervention, as well as cultural competence for delivery of abuse-related services.

Study Limitations

Although we improved on the methodology of Hamberger et al. (1998) by collecting data on potential confounding variables to assess racial differences, the present study has a number of limitations. First, the study sample was fairly small and one of convenience. Although we oversampled African American participants, they still comprised less than 50% of the total sample. In addition, participants were not recruited directly from within a health care setting. The latter critique is mitigated by the findings of Hamberger et al. (1998) that showed no difference in physician behavior valuations between battered women who had and had not actually sought abuse-related health care services. Still, more research is needed with larger samples of women directly involved in the health care system as both inpatients and outpatients. Third, questionnaire length and the omission of two items in an initial version contributed to a higher level of missing data than is desirable. Many statistical tests were done on a relatively small sample, increasing the probability of Type I error and reducing power to show differences.

Battered women of all races and backgrounds deserve technically and culturally competent care. The present study, along with Hamberger et al. (1998) and Rodriguez et al. (1996) provide rich information from patient consumers about how battered women define such care. Programs of training and clinical care must hear the voices of such consumers to inform curricular design and protocols of clinical care.