Intimate partner violence (IPV) is at epidemic proportions in the United States, with the vast majority perpetrated by men on women. Among U.S. women, there is an estimated 25% to 30% lifetime prevalence of rape or physical assault by a current or former spouse, cohabiting partner, or date.1 Current estimates suggest that physical violence occurs in four to six million intimate partner relationships each year.2 Numerous medical organizations now offer practice guidelines for IPV screening,3 many of which explicitly recommend that physicians routinely screen all female patients for IPV. Whether women actually are screened for IPV is likely to be influenced by many factors, including patient attributes as well as physician training, competence, and comfort with this issue. Although most physicians receive some formal training on IPV during medical school, more than 25% of U.S. graduating physicians in 2002 believed that the curriculum time devoted to IPV was inadequate.4
Inadequate educational preparation for IPV screening is likely to translate into low-frequency IPV screening among practicing physicians. Indeed, among physicians practicing in family medicine (FM), internal medicine (IM), or obstetrics and gynecology, only about 10% perform IPV screening routinely or for new patients, whereas nearly 80% screen for IPV if there is evidence of injury.2 Chart reviews and anonymous patient surveys suggest that the rate of IPV screening by physicians is actually much lower.5
Because it is difficult to ascertain the extent to which IPV screening takes place in routine medical encounters, it remains a challenge to identify those barriers that preclude IPV screening. Using a series of paper-and-pencil “trigger scenarios,” the current study aimed to identify potential barriers for IPV screening among new residents at the beginning of their training.
A one-page questionnaire depicted one of four different scenarios describing a divorced woman with two children who presented with nonlocalized abdominal pain of two months duration. Case scenarios were identical except for patient age (either 22 or 45 years old) and presence or absence of abdominal bruising.
With institutional review board approval, the questionnaire was administered to three successive groups of incoming residents in 2000, 2001, and 2002, representing anesthesiology (n = 7), emergency medicine (n = 32), FM (n = 56), IM (n = 93), medical pediatrics (n = 5), obstetrics and gynecology (n = 14), orthopedics (n = 6), neurology (n = 2), pediatrics (n = 19), psychiatry (n = 12), radiology (n = 3), surgery (n = 33), or “other” (n = 12) (total n = 294). Residents randomly received one of the four questionnaires and used a seven-point Likert scale to indicate the likelihood that they would perform each of eight tasks during this patient’s examination: screen for IPV or drinking; take a sexual, social, dietary, or smoking history; or conduct a pelvic exam or pregnancy test. A score of 1 indicated that the resident definitely would not perform the task, whereas a score of 7 indicated that the resident would definitely perform the specified task. Residents self-assessed their competence in performing each of these tasks using six-point Likert scales. Residents also identified their gender and area of medical specialization. The survey took less than ten minutes to complete.
We used linear regression models to examine the extent to which patient characteristics (age and presence of bruising) and resident characteristics (gender and self-assessed competence) predicts the likelihood that residents will perform each task. Modeling was done sequentially. Model 1 included patient bruising, patient age, and the interaction term. Model 2 included resident gender, resident self-assessed competence, and the interaction term. Model 3 predicted task performance from both patient and resident characteristics, excluding interaction terms because none was significant in the earlier analyses. Model 3 focused only on the tasks for which both Model 1 and Model 2 predicted performance (i.e., R2 for both models was significant, with p < .05). For all regression analyses, all variables were forced to enter the equation.
To investigate further the relationship between residents’ self-assessed competence in IPV screening and their likelihood of screening for IPV, respondents were grouped into four categories. The lowest three levels (ratings of 1, 2, and 3 on the original six-point Likert scale) were condensed into one group, which yielded four levels of self-assessed competence. This grouping resulted in residents being distributed roughly into quartiles (1–4 = lowest to highest self-assessed competence). A secondary analysiswas performed to evaluate whether patient age or patient bruising predicted IPV screening, with separate regression models constructed for each of the four self-assessed competence levels.
Residents’ likelihood to screen for IPV varied with each case. For the older patient without bruising, the mean likelihood of screening ± standard deviation (SD) was 5.09 ± 1.76 (n = 78); presence of bruising increased screening likelihood to 6.36 ± 1.26 (n = 66). For the younger patient with no bruising, IPV screening likelihood was 5.60 ± 1.56 (n = 77); again, concomitant bruising increased the likelihood of screening to 6.77 ± 0.54 (n = 73).
Table 1 summarizes the regression analyses for all history-taking and screening tasks. In Model 1, the presence of bruising increased the likelihood of performing both IPV screening and a pelvic exam. In contrast, older patient age decreased the likelihood that residents would screen for IPV; take a sexual or social history; or perform a pregnancy test. Patient attributes had modest explanatory power for IPV screening, accounting for 18% of the variance in self-reported likelihood to conduct IPV screening (R2 = 0.18) but for less than 5% of the variance in likelihood to perform any of the other tasks.
Model 2 revealed that male residents were less likely to screen patients for IPV or to conduct a pelvic exam. Overall, resident gender or self-assessed competence significantly influenced the likelihood of screening for IPV or for drinking, taking a sexual or smoking history, or conducting a pelvic exam, although the percentage of variance accounted for by resident attributes was small to nonsignificant for all dependent variables.
Model 3 focused on the three tasks for which both Model 1 and Model 2 predicted performance. In this model, patient bruising increased, whereas older patient age or male resident gender decreased the likelihood for screening for IPV or performing a pelvic exam. Older patient age or male resident gender also decreased the likelihood of sexual history taking. Higher self-assessed competence in IPV screening or sexual history taking increased the likelihood that these tasks would be performed. Overall, patient and resident attributes had modest explanatory power for IPV screening, accounting for 22.6% of the variance in self-reported likelihood to conduct IPV screening (R2 = 0.226), considerably more than for taking a sexual history (R2 = .055) or for conducting a pelvic exam (R2 = .074).
Table 2 presents the secondary regression analyses that investigated the influence of residents’ self-assessed competence on the likelihood of screening the patient in each case scenario for IPV. Higher self-assessed competence increased the likelihood that residents would screen patients for IPV. For all residents but those expressing the highest self-assessed competence in IPV screening (Level 4), patient bruising, younger patient age, or both increased the likelihood of IPV screening. In contrast, residents with the highest self-assessed competence in IPV screening were likely to conduct IPV screening regardless of patient age or bruising.
Patient attributes influenced the likelihood that first-year residents would conduct IPV screening in a woman with nonlocalized abdominal pain. Perhaps not surprisingly, presence of bruising was the strongest positive predictor. Previous studies2 suggest that a majority of physicians would screen an injured female patient for IPV but would be unlikely to screen routinely for IPV. Without routine screening, however, IPV in women without overt signs of physical injury would be likely to go undetected.
The 45-year-old patient was less likely to be screened for IPV than her 22-year-old counterpart, suggesting that first-year residents may hold the view that increased patient age reduces risk for IPV, although there is little evidence that patient age per se influences IPV. In a case-control study of women with acute injuries resulting from a physical assault by a male partner, neither patient age nor other patient attributes, such as race, education, or alcohol use, increased the likelihood that she would sustain physical abuse by her male partner.6 However, perpetrators of IPV are more likely to be young men.7 Because young women are more likely to have young male partners, their risk for IPV may be higher, suggesting that an effective IPV screen should include questions about the patient’s partner.
Our findings suggest that male resident gender decreases the likelihood that patients would be screened for IPV. Similar gender differences were observed in a recent cross-sectional study of practicing physicians8 designed to identify IPV screening barriers, although gender differences were not observed in an earlier study of similar design.2 Physician gender influences performance of many screening practices, both related and unrelated to patient gender.9 Whether female physician gender is associated with enhanced awareness of the impact of IPV, greater empathy for patients who have experienced IPV, greater comfort in discussing relationship issues, or other factors warrants further study.
Not surprisingly, self-assessed competence is a strong predictor for IPV screening. Residents with the highest self-assessed IPV screening competence were most likely to screen for IPV regardless of patient age or bruising. In contrast, residents expressing lower competence levels were less likely to conduct routine IPV screening, although patient bruising would trigger IPV screening by these residents. In a cross-sectional survey of practicing physicians, self-confidence and recent training in IPV screening enhanced self-reported IPV screening practice.8 Training and other opportunities that enhance competence clearly play key roles in increasing IPV screening.
There is still a large gap between estimated IPV prevalence and the actual rate of IPV screening in medical settings. For reasons that include patient safety and privacy, it is difficult to corroborate the true rate of IPV screening by chart review or by patient surveys.Thus, it is important to develop reliable proxy measures for actual clinical practice. In an elegant study, Peabody et al.10 reported that paper-and-pencil case scenarios could effectively quantify clinical competence in many areas of medical practice. This methodology was more accurate than chart abstraction in estimating provider skills, and it closely approximated the “gold-standard” measurement of clinical acumen: highly trained standardized patients.
Although responses to paper-and-pencil trigger scenarios may not be perfect predictors of real-life provider practice, they offer an opportunity to test hypotheses about patient and provider attributes that could potentially influence—positively or negatively—medical responses to IPV. The current study was limited to assessing impact of patient age, patient bruising, resident gender, and residents’ self-assessed competence on IPV screening. Physicians identify many other barriers to IPV screening,8,11 including lack of time or privacy, fear of “opening Pandora’s box,” and inadequate training. Physicians’ preconceptions about IPV demographics may represent yet another set of barriers that increases the chance that vulnerable patients will not be screened. The methodology described here could readily be used to probe other physician beliefs that affect IPV screening. It also could be applied to determine whether barriers to IPV screening, evident in new residents, disappear (or become more fully entrenched) over time. Such information will be helpful in designing curricula that more effectively teach medical practitioners how to screen for IPV.
The authors express their appreciation to the Meyers Primary Care Institute, Worcester, Massachusetts, for funding, to Laura Sefton for able assistance with data collection and management, and to Lisa Keller for expert statistical advice.
1.Tjaden P, Thoennes N. Extent, nature, and consequences of intimate partner violence. 〈http://www.ncjrs.org/pdffiles1/nij/181867.pdf
〉. Accessed 13 June 2003. National Institute of Justice and the Centers for Disease Control and Prevention. U.S. Department of Justice, Washington, DC, 2000.
2.Rodriguez MA, Bauer HM, McLoughlin S, Grumbach K. Screening and intervention for intimate partner abuse. Practices and attitudes of primary care physicians. JAMA. 1999;282:468–74.
3.Rhodes, KV, Levinson W. Interventions for intimate partner violence against women. Clinical applications. JAMA. 2003;289:601–5.
5.Hamberger LK, Saunders DG, Hovey M. Prevalence of domestic violence in community practice and rate of physician inquiry. Fam Med. 1992;24:283–7.
6.Kyriacou DN, Anglin D, Taliaferro E, Stone S, Tubb T, Linden JA, Muelleman R, Barton E, Kraus JF. Risk factors for injury to women from domestic violence against women. N Engl J Med. 1999;341:1892–8.
7.Fagan J, Browne A. Violence between spouses and intimates: physical aggression between women and men in intimate relationships. In: Reiss AJ Jr., Roth JA (eds). Understanding and Preventing Violence, Volume 3: Social Influences, Washington, DC: National Academy Press, 1994;115–292.
8.Elliott L, Nemey M, Jones T, Friedmann PD. Barriers to screening for domestic violence. J Gen Intern Med. 2002;17:112–6.
9.Kreuter MW, Strecher VJ, Harris R, Kobrin SC, Skinner CS. Are patients of women physicians screened more aggressively? A prospective study of physician gender and screening. J Gen Intern Med. 1995;10:119–25.
10.Peabody JW; Luck J, Glassman P, Dresselhaus TR, Lee M. Comparison of vignettes, standardized patients, and chart abstraction. A prospective validation study of three methods for measuring quality. JAMA. 2000;283:1715–22.
11.Lapaidus G, Cooke MB, Gelven E, Sherman K, Duncan M, Banco L. A statewide survey of domestic violence screening behaviors among pediatricians and family physicians. Arch Pediatr Adolesc Med. 2002;156:332–6.