More than 25% of women living in the United States report a history of intimate partner violence, the occurrence of which is associated with adverse health consequences as well as increased health care use.1–5 Within obstetrics and gynecology, 35% of patients report intimate partner violence exposure.6 Because women of all age groups, races, and socioeconomic backgrounds are at risk for intimate partner violence,1,7 most major medical organizations, including the American Medical Association and the American College of Obstetricians and Gynecologists, recommend routine screening of adult women for partner abuse.8,9 Despite established clinical guidelines, however, a minority of patients report ever receiving intimate partner violence screening by a health care professional.10 Although most obstetrician-gynecologists appear to be familiar with the nature and consequences of intimate partner violence, they routinely screen for intimate partner violence only 20% of the time, an estimate that, although low, exceeds that of health care providers in other specialties.11–13
Prior research has shown that obstetrician–gynecologists may selectively screen patients for intimate partner violence based on certain patient characteristics. In one survey of 962 practicing obstetrician–gynecologists in the United States, 46% indicated that the type of patient influenced their decision to ask about partner abuse.14 This finding suggests that health care providers may use some form of risk assessment to prompt screening or may anticipate abuse in specific subgroups of patients. Increased insight into these screening behaviors could direct systems-based interventions aimed at achieving routine intimate partner violence screening. In addition, previous studies suggest that standardizing intimate partner violence screening in the clinical setting, for example by medical record prompts or formal protocols, may improve rates of intimate partner violence inquiry. Research to support such initiatives within the ambulatory gynecologic setting is, however, currently lacking.15,16
The goal of this cross-sectional study was to estimate the rate of routine interpersonal violence screening in an ambulatory gynecologic clinic and to identify patient and health care provider characteristics that may be predictive of screening. We hypothesized that the use of a standardized documentation form would result in higher screening rates, that certain patient characteristics such as younger age or history of mental illness might prompt health care providers to screen selectively for violence, and that certain health care provider characteristics such as gender would be associated with increased violence screening.
For the planned cross-sectional study, medical records were reviewed for all women who presented to the hospital's outpatient ambulatory clinic (the Women's Primary Care Center) for annual health care visits between January 1 and December 31, 2007. This clinic serves as the clinical care site for the Brown University/Women & Infants Hospital National Center of Excellence in Women's Health. The Women's Primary Care Center provides comprehensive obstetric, gynecologic, and primary care services as well as psychiatry, nutrition, family planning, and specialty care to an ethnically diverse and economically underserved population. Nurse practitioners and resident physicians staff the Women's Primary Care Center; attending physicians provide clinical supervision for these health care providers. Approval was obtained from the Women and Infants' Institutional Review Board (Project 08-0064, approved June 10, 2008) before study initiation.
Before data extraction, a sample size calculation was performed using the rate of interpersonal violence screening as the primary outcome variable. As previously noted, published screening rates among obstetricians and gynecologists for intimate partner violence approximate 20%. We conservatively projected the screening rate in our ambulatory population to be 30%. To detect an absolute difference in screening rates of 10%, a sample size of 262 women would have been necessary to achieve a two-sided alpha of 0.01 and a power of 90%. Assuming missing data in 15% of charts, the final sample size was increased to 300.
During the study period, 1,857 nonpregnant women presented for annual health care visits. Women were excluded from the study if greater than 50% of their health care record was incomplete, if health care provider documentation was illegible (handwritten patient charts are used in the Women's Primary Care Center), or if the type of patient visit had been miscoded. For the purposes of the analysis, we selected 300 patients from the eligible population using a random numbers table. Study investigators then abstracted data from a health history form designed by an interdisciplinary group of physicians (including representatives from gynecology, internal medicine, and psychiatry) in 2004 as part of the Center of Excellence's initiative to improve health screening and preventive care services. The health history instrument prompted health care providers to document a detailed medical, surgical, psychological, contraceptive, sexual, family, and social history. Embedded within the social history was a line item stating “Physical/Sexual/Emotional Abuse.” Data on patient demographics (including age, race and ethnicity, disability, and insurance status), patient health characteristics (eg, gravidity and parity, body mass index, sexual and sexually transmitted infection history, contraceptive use, alcohol and substance abuse, mental illness, and somatic complaints), and health care provider characteristics (including type and gender) were placed onto data collection forms and entered into a password-protected computerized database. Somatic complaints recorded included current abdominal pain, constipation, diarrhea, vaginal concerns, fatigue, weight gain or loss, frequent or severe headaches, dizziness, chest pain, palpitations, back pain, or other muscular pain. Ten percent of the charts were randomly selected and rereviewed by study authors to ensure data accuracy.
The primary study outcome, screening for interpersonal violence, was considered to have occurred when documentation of a positive or negative screen existed or when specific types of abuse had been recorded. If the form's line item “Physical/Sexual/Emotional Abuse” had been left blank, screening was judged not to have occurred. In the secondary analysis, patient and health care provider variables associated with screening were assessed. For the purposes of the study, two composite variables were created to assess for specific patterns that might be associated with screening. A first variable, somatic pain, was defined as documentation of at least one complaint of current abdominal, chest, back, or other muscular pain or headache. The second variable, health care provider comprehensiveness, consisted of documentation of screening for three other preventive care variables (exercise, sleep, and presence of guns in the home). These variables were located in close proximity on the health history form to the interpersonal abuse screening prompt. When a health care provider documented screening for all three of these preventive care variables, the patient encounter was classified as a comprehensive preventive health screen.
All P values were two-sided with P≤.05 considered statistically significant. Continuous variables were compared between groups by t test (means) or the Wilcoxon rank sum test (medians). Associations between patient and health care provider variables and interpersonal violence screening were estimated using unconditional logistic regression models constructed to obtain odds ratios (ORs) and 95% confidence intervals (CIs). Variables associated with screening in which P<.1 in the univariable analysis were selected for a multivariable logistic regression model. The fit of the final multivariable model was acceptable by the Hosmer-Lemeshow test (P=.3). SAS 9.1 (SAS Institute, Cary, NC) was used for all data analyses.
Data were abstracted from 300 patient annual health care encounters that met eligibility criteria. Of note, 322 charts were initially examined, and 22 were excluded for the following reasons: 1) patient seen for a problem visit only (17 instances); and (2) patient seen for annual women's cancer screening program visit only (five instances). The overall screening rate for interpersonal violence among the study population was 81% (95% CI 77–85%). Of the 243 women screened, 48 (20%) gave a history of abuse and four (2%) reported current abuse. Physical abuse was most commonly reported (10% of the population screened) followed by sexual abuse (8%) and emotional abuse (7%).
Compared with the population that was not screened for domestic violence, women in the screened cohort were younger (mean age 31.7 compared with 34.7 years, P=.07). No significant differences in terms of race, gravidity, parity, martial status, insurance status, presence of disability, or the presence of children living at home emerged between the two populations (Table 1). In examining patients' medical and gynecologic histories, several significant differences became apparent (Tables 2 and 3). In the univariable analysis, the likelihood of screening was greater for women presenting with any pain (OR 2.46, 95% CI 1.13–5.34). Although screening rates were high in women with normal body mass indices, rates significantly dropped among overweight women (74% compared with 87%; OR 0.44, 95% CI 0.21–0.91). A similar, but nonsignificant, association was also seen among obese women (78% screened compared with 87% among normal weight women; OR 0.55, 95% CI 0.26–1.15). Health care provider characteristics, including both health care provider type and gender, were not associated with differences in screening behaviors, although the majority of health care providers were female (Table 3).
Health care provider documentation of other preventive care inquiries was predictive of screening for interpersonal violence. For example, although no significant differences were seen in screening rates in women reporting a history of smoking, alcohol abuse, or drug abuse compared with women without such histories, rates of screening fell dramatically if health care providers did not document the patient's history of these three behaviors (Table 3). Such failures of screening, however, were rare. Comprehensiveness in preventive care screening was assessed using a composite variable consisting of screening documentation for three preventive care variables (exercise, sleep, and presence of guns in the home) located directly underneath the interpersonal abuse screening prompt on the health history form. Although health care providers were less likely to document screening for these three variables compared with the variables for smoking, alcohol, and drug use, the association between screening for exercise, sleep, and the presence of guns at home and screening for violence persisted in the univariable analysis (OR 2.80, 95% CI 1.55–5.05.
In the final model, both the presence of a complaint of pain (adjusted OR 2.55, 95% CI (1.12–5.83) and comprehensiveness of health care provider screening (adjusted OR 2.50, 95% CI 1.26–4.99) were significantly associated with documentation of screening of interpersonal violence (Table 4). Conversely, overweight women (body mass index more than 25 or less than 30 kg/m2) were less likely to be screened for partner violence (adjusted OR 0.44, 95% CI 0.20–0.99). Similar findings were not, however, observed among obese women (adjusted OR 0.86, 95% CI 0.35–2.13).
Historically, identified barriers to screening for intimate partner violence have included inadequate health care provider training, time constraints, health care providers' feelings of powerlessness, and lack of resources for identified victims.12–14,17 Standardized documentation tools have been offered as a means to overcome such barriers and increase detection of intimate partner violence.15,16 Our findings support previous data that suggest embedding inquiries about partner abuse in the medical record may improve intimate partner violence screening rates. The observed screening rate of more than 80% in our gynecologic population may be attributable in part to the standardized health history form, which provided prompts for preventive care screening (including interpersonal violence) and documentation. Such systems-based initiatives can be easily adopted to increase routine comprehensiveness in any ambulatory setting and are integral components of model programs addressing partner violence.18
The results of this study also imply that one of the strongest influences on a health care provider's screening behavior may be the tendency to perform comprehensive preventive health screening. Knowledge about intimate partner violence screening is a competency expected of all medical students completing their obstetrics and gynecology clerkships, and many residency programs (eg, obstetrics and gynecology, internal medicine, family medicine, and psychiatry) are required to include training on partner abuse to receive accreditation by the Accreditation Council for Graduate Medical Education.19,20 Such efforts are undoubtedly aimed to advance the health care response to partner violence. Although many models of intimate partner violence curricula exist,21 our findings indicate that encouraging health care providers to address overall preventive health care may also improve screening for this important public health issue. Demarginalizing intimate partner violence training and incorporating it as part of preventive health education may positively influence physicians' internal framework for routine inquiry about partner abuse.
Clinicians' tendency toward case selection merits further exploration as it relates to screening for interpersonal violence. Certainly, physicians' traditional cognitive framework, which relies on pattern recognition for the development of differential diagnosis, may influence clinician screening patterns. Our finding that a somatic pain complaint at the time of the health care visit conferred an increased likelihood of interpersonal violence screening may represent a form of health care provider prescreen risk assessment derived from knowledge of the adverse health associations of partner abuse. The finding of decreased screening among overweight women as compared with normal weight women is of uncertain, although of potentially important significance pertaining to health care provider bias. National prevalence data do not demonstrate an association between increased body mass index and a lifetime risk of intimate partner violence.1 Although little is known about health care providers' biases in domestic violence screening with regard to obesity, disparities in other areas of preventive screening and care have emerged. For example, overweight and obese women are less likely to undergo both routine mammography and cervical cancer screening, practices that have been linked to significantly higher mortality rates for both breast and cervical cancer in these populations.22,23 Although health care provider antiobesity bias has been documented,24 other possible explanations for decreased intimate partner violence screening in overweight patients may exist. For instance, obesity-related comorbidities may necessitate additional health care provider counseling time, thus limiting opportunities for routine preventive screening. Because our study was not specifically designed to evaluate the association between weight and violence screening, further research is warranted to elucidate these and other health care provider biases that may influence partner violence screening.
The main strengths of this study included its large sample size and use of a standardized form, which facilitated consistent acquisition of data on many variables under consideration. The retrospective design of the study, however, limited complete data collection. Furthermore, although information on patients' race and ethnicity is now collected comprehensively, during the timeframe of this study, patients identified as anything other than white or African American were placed into a single category. Thus, our ability to determine differences in health care provider screening based on patient race and ethnicity was restricted. Lastly, our rates of exposure for interpersonal violence are lower than those reported in other obstetrics and gynecology settings6 and perhaps relate to the lack of standardized, effective screening questions used by health care providers in this study. Behaviorally specific inquiries (for example, “Does your partner ever hit, slap, kick, choke, or punch you?”) are critical to capturing intimate partner violence in clinical settings. Because the chart prompt embedded in our health history form was simply a line item stating “Physical/Sexual/Emotional Abuse,” health care providers may have underidentified victims of past or current abuse. Of note, before the implementation of the new annual health history forms in 2004, all health care providers were oriented to the form components and a brief update on intimate partner violence screening was provided. Since 2007, didactic sessions on intimate partner violence have been added to the residents' core curriculum, new postgraduate year 1's are now introduced during orientation to the hospital's standardized intimate partner violence screening and interventional protocol, the Women's Primary Care Center nurse practitioners now participate in a didactic session on intimate partner violence, and monthly audits of intimate partner violence screening in the Women's Primary Care Center have begun as part of a hospitalwide initiative to improve preventive care. Finally, evidence-based screening questions have been directly incorporated into the annual examination forms, replacing the more general prompt of “Physical/Sexual/Emotional Abuse.”
Although a recently published randomized trial failed to demonstrate benefit from intimate partner violence screening,25 most major medical societies continue to recommend routine screening of women for partner abuse. The need exists for further study of the effects of screening and the development of effective interventions. Until that time arrives, however, clinicians should continue to screen all their adult female patients for partner violence not only for compliance with national guidelines, but because its high prevalence and extensive health effects warrant routine inquiry. Global efforts to improve overall preventive health screening may in part drive increased partner violence screening, as was seen in this study, and could have a significant impact on both medical education and systems-based efforts to promote screening for interpersonal violence.
1. Centers for Disease Control and Prevention (CDC). Adverse health conditions and health risk behaviors associated with intimate partner violence—United States, 2005. MMWR Morb Mortal Wkly Rep 2008;57:113–7.
2. Bonomi AE, Anderson ML, Reid RJ, Rivara FP, Carrell D, Thompson RS. Medical and psychosocial diagnoses in women with a history of intimate partner violence. Arch Intern Med 2009;169:1692–7.
3. Kovac SH, Klapow JC, Kroenke K, Spitzer RL, Williams JB. Differing symptoms of abused versus nonabused women in obstetric-gynecology settings. Am J Obstet Gynecol 2003;188:707–13.
4. Eberhard-Gran M, Schei B, Eskild A. Somatic symptoms and diseases are more common in women exposed to violence. J Gen Intern Med 2007;22:1668–73.
5. Rivara FP, Anderson ML, Fishman P. Healthcare utilization and costs for women with a history of intimate partner violence. Am J Prev Med 2007;32:89–96.
6. McCloskey LA, Lichter E, Ganz M, Williams CM, Gerber MR, Sege R, et al. Intimate partner violence and patient screening across medical specialties. Acad Emerg Med 2005;12:712–22.
7. Thompson RS, Bonomi AE, Anderson M, Reid RJ, Dimer JA, Carrell D, et al. Intimate partner violence: prevalence, types, and chronicity in adult women. Am J Prev Med 2006;30:447–57.
8. American Medical Association Diagnostic and Treatment Guidelines on Domestic Violence. Arch Fam Med 1992;1:39–47.
9. American College of Obstetricians and Gynecologists. Screening Tools—Domestic Violence. 2006. Available at: http://www.acog.org/departments/dept_notice.cfm?recno=17&bulletin=585
. Cited August 3, 2006. Retrieved November 9, 2009.
10. Klap R, Tang L, Wells K, Starks SL, Rodriguez M. Screening for domestic violence among adult women in the United States. J Gen Intern Med 2007;22:579–84.
11. Horan DL, Chapin J, Klein L, Schmidt LA, Schulkin J. Domestic violence screening practices of obstetrician-gynecologists. Obstet Gynecol 1998;92:785–9.
12. Rodriguez MA, Bauer HM, McLoughlin E, Grumbach K. Screening and intervention for intimate partner abuse: practices and attitudes of primary care physicians. JAMA 1999;282:468–74.
13. Elliott L, Nerney M, Jones T, Friedman PD. Barriers to screening for domestic violence. J Gen Intern Med 2002;17:112–6.
14. Parsons LH, Zaccaro D, Wells B, Stovall TG. Methods of and attitudes toward screening obstetrics and gynecology patients for domestic violence. Am J Obstet Gynecol 1995;173:381–6; discussion 386–7.
15. Freund KM, Bak SM, Blackhall L. Identifying domestic violence in primary care practice. J Gen Intern Med 1996;11:44–6.
16. Olson L, Anctil C, Fullerton L, Brillman J, Arbuckle Sklar D. Increasing emergency physician recognition of domestic violence. Ann Emerg Med 1996;27:741–6.
17. Sugg NK, Inui T. Primary care physicians' response to domestic violence. Opening Pandora's box. JAMA 1992;267:3157–60.
18. McCaw B, Kotz K. Family violence prevention program: another way to save a life. The Permanente Journal 2005;9:65–8.
19. Association of Professors of Gynecology and Obstetrics. Women's Health Care Competencies for Medical Students. Crofton (MD): APGO; 2004. Available at: http://wheocomp.apgo.org/area-of-competency.cfm?compid=5
. Retrieved November 10, 2009.
20. Accreditation Council for Graduate Medical Education. Residency Review Committees—Program Requirements. Chicago (IL): ACGME; 2008. Available at: http://www.acgme.org/acWebsite/navPages/nav_comRRC.asp
. Retrieved November 10, 2009.
21. Hamberger LK. Preparing the next generation of physicians: medical school and residency-based intimate partner violence curriculum and evaluation. Trauma Violence Abuse 2007;8:214–25.
22. Maruther NM, Bolen S, Brancati FL, Clark JM. Obesity and mammography: a systematic review and meta-analysis. J Gen Intern Med 2009;24:665–77.
23. Maruther NM, Bolen SD, Brancati FL, Clark JM. The association of obesity and cervical cancer screening: a systematic review and meta-analysis. Obesity (Silver Spring) 2009;17:375–81.
24. Jay M, Kalet A, Ark T, McMacken M, Messito MJ, Richter R, et al. Physicians' attitudes about obesity and their associations with competency and specialty: a cross-sectional study. BMC Health Serv Res 2009;9:106.
25. MacMillan HL, Wathen CN, Jamieson E, Boyle MH, Shannon HS, Ford-Gilboe M, et al. Screening for intimate partner violence in health care settings: a randomized trial. JAMA 2009;302:493–501.