520574
research-article2014
VAWXXX10.1177/1077801213520574Violence Against WomenSprague et al.
Research Note
Prevalence of Intimate Partner Violence Across Medical and Surgical Health Care Settings: A Systematic Review
Violence Against Women 2014, Vol. 20(1) 118–136 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1077801213520574 vaw.sagepub.com
Sheila Sprague1, J. Carel Goslings2, Celine Hogentoren2, Simone de Milliano2, Nicole Simunovic1, Kim Madden1, and Mohit Bhandari1
Abstract Intimate partner violence (IPV) is a serious health problem and a leading cause of nonfatal injury in North American females. Prevalence of IPV has ranged from less than 20% to more than 50% across primary care, emergency medicine, and family medicine. We conducted a systematic review and meta-analysis of the literature to examine best estimates of IPV prevalence as opportunities for targeted interventions in health care specialties. We included 37 articles in this study. Based on our pooled data, best estimates of the lifetime prevalence of any type of IPV were 38% in family medicine and 40% in emergency medicine. Keywords intimate partner violence, prevalence, systematic review
Introduction Intimate partner violence (IPV) against women has been identified as a serious health problem and is a leading cause of nonfatal injury in females in North America (Dannenberg, Baker, & Li, 1994; Kyriacou et al., 1999). IPV is described by the American Medical Association as “a pattern of coercive behaviors that may include 1McMaster 2University
University, Hamilton, Ontario, Canada of Amsterdam, The Netherlands
Corresponding Author: Sheila Sprague, Department of Clinical Epidemiology and Biostatistics, McMaster University, 293 Wellington St. North, Suite 110, Hamilton, Ontario, Canada L8L 8E7. Email: [email protected]
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
119
Sprague et al.
repeated battering and injury, psychological abuse, sexual assault, progressive social isolation, deprivation, and intimidation” (McCloskey et al., 2007; p. 1067). These behaviors are perpetrated by someone who is in a private and personal relationship with another individual, such as a victim’s spouse, common-law partner, sexual partner, or dating partner. Some studies have reported that as high as 54% of all women have experienced some form of IPV during the course of their life (Reisenhofer & Seibold, 2007). Women who are victims of IPV have been known to utilize health care services at higher rates than women who have not been abused (Campbell, 2002). They may present to a number of different medical specialists including emergency room physicians, orthopedic or trauma surgeons, family physicians, and specialists in obstetrics and gynecology, during routine appointments or for appointments specific to their injuries. Researchers and advocates have argued that health care providers can play a vital role in detecting IPV (Davidson et al., 2001). Despite these recommendations for universal screening, the response from the medical community to address the issue of domestic violence has been characterized as “slow and inconsistent” (Davis, 2008). IPV is an underrecognized, recurrent part of trauma, present in at least one in five women who present to the Emergency Department, and physicians frequently fail to make this diagnosis (Davis, 2008). The failure to detect IPV contributes to recidivism and long-term health problems (Glass, Dearwater, & Campbell, 2001). In addition, when injuries resulting from IPV are not recorded in medical records, abused women are denied documentation of their injuries for future references in court cases (Glass et al., 2001). In addition, opportunities to provide education about prevention, lethality assessment, safety planning, and options for escaping an abusive situation and referrals to resources within the community may also be missed (Glass et al., 2001). The reported rates of IPV vary across different subspecialties and may be underreported. While several studies have reported the prevalence rates of IPV in the health care setting, these estimates vary substantially. Best estimates of rates across various subspecialties are lacking. We therefore undertook a systematic review and meta-analysis of the published literature to determine the prevalence rates of IPV reported to different medical subspecialties including family medicine, emergency medicine, obstetrics and gynecology, internal medicine, addiction recovery clinics, and adults with child patients in pediatric clinics. The study has important implications for health care professionals in various settings. It will direct attention to medical practitioners in subspecialties where IPV is overlooked and will affect their interaction with and treatment of IPV victims as well as victims’ willingness to disclose. In addition, identification of health care settings with the highest prevalence rates will allow targeted research and focus toward improvement of the quality of life of IPV victims.
Method This systematic review adheres to the reporting guidelines of the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) Statement (Stroup et al., 2000).
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
120
Violence Against Women 20(1)
Eligibility Criteria We identified articles in all languages that met the following eligibility criteria: (a) The study was published between January 1995 and July 2009, (b) the target population consisted of adult women between the ages of 16 and 65 years presenting to physicians of any medical specialty, and (c) the primary objective of the study was to determine the prevalence rates of IPV. We included all study designs including cross-sectional designs, surveys, and chart or medical record reviews. We excluded studies focusing on children, males, and special populations such as pregnant women, elderly patients, and patients suffering from a specific illness (i.e., depression, chronic gastroenterological disease). We also excluded review articles, letters, comments, case reports, and guidelines.
Identification of Studies Two reviewers (C.H. and S. dM.) independently conducted a search of the electronic database PubMed for relevant articles published from January 1995 to July 2009 in all languages with the help of a professional librarian. Additional strategies used to identify studies included consultation with experts, a review of reference lists from articles that fulfilled our eligibility criteria, and use of the “related articles” feature in PubMed for all studies meeting our entry criteria.
Assessment of Study Eligibility Two of the authors (C.H. and S. dM.) independently assessed all studies identified for full evaluation and resolved disagreements through discussion toward consensus. The eligibility was verified by two additional authors (S.S. and N.S.). Agreement between observers for study eligibility with regard to abstract and full-text review was examined using the κ (kappa) statistic. The interobserver agreement in methodological quality scores was evaluated using the intraclass correlation coefficient. We chose an a priori criterion of κ ≥ .65 for adequate agreement. The 95% confidence intervals were calculated using the generic formula: Estimate ± 1.96 × (Standard Error).
Assessment of Methodological Quality Two reviewers (N.S. and S.S.), both with methodological expertise and one with content expertise (S.S.), independently graded the methodological quality of each included study using questions adapted from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement designed for cross-sectional studies (von Elm et al., 2008). The STROBE statement is a checklist of 22 items that are considered for good reporting of observational studies. The STROBE statement was developed to assist authors when writing analytic observational studies, to support editors and reviewers when considering such articles for publication, and to help readers critically appraise published research (von Elm et al., 2008).
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
121
Sprague et al.
The original authors emphasized that the STROBE statement was not developed as a tool for assessing the quality of published observational research (von Elm et al., 2008). Unfortunately, very few checklists exist for critically appraising cross-sectional studies (Katrak, Bialocerkowski, Massy-Westropp, Kumar, & Grimmer, 2004; Kelley, Clark, Brown, & Sitzia, 2003; Sanderson, Tatt, & Higgins, 2007), and we did not find that any of the previously developed checklists were appropriate for this study. Therefore, we adapted the STROBE statement to assess study quality. A priori we determined that studies that met at least 10 of the 13 reporting criteria within our modified STROBE criteria would be considered high quality; studies that met 7 to 9 reporting criteria would be considered medium quality; and studies that met 6 or fewer criteria would be classified as low quality.
Data Extraction A structured data extraction form was developed and relevant data were extracted from each eligible study in duplicate (S.S. and N.S.) to ensure accuracy. Pertinent data included study characteristics (year and location), patient characteristics (age), type of medical subspeciality (emergency medicine, family medicine, obstetrics and gynecology, internal medicine, addiction recovery, pediatrics, and public health), method of data collection (survey, interview, chart review, other design), time frame (lifetime, 1 year, 2 years, and other), sample size, and response rate. We documented how the authors defined overall IPV as well as physical, emotional, and sexual abuse. Finally, we recorded the reported rates of overall IPV and of physical, emotional, and sexual abuse as reported by the authors.
Data Analysis We provide pooled rates of lifetime and 1-year physical abuse, emotional abuse, and sexual abuse for emergency medicine, family medicine, and other specialties. Analyses were performed with Predictive Analytics SoftWare (PASW) Statistics version 18.0 (SPSS, Inc., Chicago, IL).
Evaluation of Heterogeneity Before analyzing the data, we hypothesized that there would be a large degree of heterogeneity across the studies. Differences such as study methodology (e.g., survey vs. interview vs. other study designs), timelines of reported IPV (e.g., lifetime vs. 1 year), differences in screening tools used (e.g., Abuse Assessment Screen (AAS), Conflict Tactics Scale (CTS), Partner Violence Screen (PVS)), variability within and between populations (e.g., socioeconomic status, age, jurisdiction), differences in the definition of IPV perpetrator (e.g., male perpetrator vs. perpetrator of any gender, or spouse vs. any intimate partner), or variable study quality could contribute to the heterogeneity.
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
122
Violence Against Women 20(1)
Results Study Identification Our literature search identified 894 potentially relevant citations, of which 32 were considered for inclusion after full-text review (Figure 1). Eighteen of these 32 studies were excluded after verification by additional reviewers (N.S. and S.S.) because of insufficient data on IPV prevalence, or they did not meet the eligibility criteria. Of 80 studies considered for inclusion based on review of bibliographies, 20 were included in this study. Three studies were identified and included after consultation with a content expert. In total, 37 studies were included in this review. The weighted kappa on overall agreement between reviewers when choosing articles for the full-text review was .70 (95% CI = [0.59, 0.82]), and for final study inclusion there was only one disagreement.
Study Characteristics Details of the selected studies are shown in Table 1. The mean age of participants in the included studies was between 28 and 65 years, with 17 studies not reporting mean age. Response rates ranged from 9% to 100% with most studies (24/37, 64.9%) reporting a response rate of 70% or greater. Twenty-five studies (67.6%) took place in the United States, 3 studies took place in each of Australia and the United Kingdom, and other locations (Canada, Spain, Iran, Israel, South Africa, and China) each having 1 study. Most studies took place in family medicine clinics (15/37, 40.5%) and emergency departments (12/37, 32.4%). Of the remaining studies, 3 took place in obstetrics and gynecology clinics, three took place in internal medicine clinics, and 4 studies had more than one specialty. Nineteen studies (51.4%) used written questionnaires to assess rates of IPV, 14 studies (37.8%) used an in-person interview, 1 study (2.7%) used a telephone survey, 1 study (2.7%) used a computer-based survey, and 2 studies (5.4%) used more than one method.
Study Quality We judged 26 studies to be of high methodological quality, 10 studies to be of moderate quality, and the remaining 1 study to be of low quality (Table 1). The reviewers of methodological quality achieved moderate agreement (intraclass correlation coefficient, 0.65; 95% CI = [0.57, 0.71]).
Definition of IPV The definition of IPV varied greatly among studies. The two most commonly used screening tools among the studies were the Abuse Assessment Screen (13 studies, 35.1%) and the Conflict Tactics Scales (5 studies, 13.5%), or modified versions of these tools. Many studies used gender-neutral language, such as “partner,” but others
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
123
Sprague et al.
PubMed search: Citations identified based on title (n=894) Excluded based on title (n=471)
Further review of abstracts (n=423) Excluded based on abstract review (n=343) Further review of full-text (n=80) Excluded based on full-text review (n=48) Additional verification of full-text (n=32)
Hand search of bibliography (n=80) Excluded based on full-text (n=60): Specialized population (n=20) Prevalence is not primary focus (n=19) Review (n=7) Duplicate (n=4) Not presenting to health professional (n=3) Not IPV (n=2) Physicians surveyed only (n=2) Pre-1995 (n=1) Does not look at specific subspecialty (n=1) Doesn’t separate specialties (n=1)
Excluded based on full-text review (n=18): Specialized population (n=6) Duplicate (n=5) Prevalence is not primary focus (n=4) Not IPV (n=1) Not presenting to health professional (n=1) Doesn’t separate specialties (n=1) Consultation with content expert (n=3)
Excluded based on fulltext review (n=0)
Included from reference search (n=20)
Included from literature search (n=14)
Included from expert consultation (n=3)
Articles included in systematic review (n=37)
Figure 1. Flowchart of study process. Source. Adapted from Guyatt et al. (2007).
specified that the abuser must be male (8 studies, 21.6%). Most studies included only female victims, but a small number (3 studies, 8.1%) also included male victims.
IPV Prevalence Twenty-nine of the included studies (78.4%) presented an overall rate of IPV. Overall pooled lifetime prevalence rates were similar across family medicine and emergency medicine (38% and 40%, respectively). Rates reported for other specialties, however,
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
124
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
Year
Location
Family
2000 United States
2007 United States
2004 Canada
1998 United States
2007 United States
1995 United States
1997 United States
Coker, Smith, McKeown, and King Coker et al.
Cox et al.
Dearwater et al.
El-Bassel et al.
Elliott and Johnson
Ernst, Nick, Weiss, Houry, and Mills
18-65 English/Spanish speaking
Emergency
ObGyn
18+, inner city, English speaking, excluded if they never had a partner, men and women Iranian women, married
18-65, English speaking
18-65, insured, in relationship for at least 6 months Lower income women in rural clinics 16+, English or French speaking, able to communicate 18+, English and Spanish speaking, able to communicate 18+
Emergency
Family
Emergency
Emergency
Emergency
Family
2008 United States
Faramarzi, Esmailzadeh, 2005 Iran and Mosavi Glass, Dearwater, and 2001 United States Campbell
English, Spanish, 18-86, 7% pregnant
Population
Family, ObGyn, 18-46, 31% African American, Internal 36% Latina, 31% White Internal Female veteran patients, English speaking Internal 18-60, not pregnant, English/ Spanish speaking
Emergency
Medical specialty
Cleary, Keniston, Havranek, and Albert
1995 United States Abbott, Johnson, Koziol-McLain, and Lowenstein Bauer, Rodriguez, and 2000 United States Perez-Stable Caralis and Musialowski 1997 United States
Study
Table 1. Characteristics of Included Studies.
NR
28.2 (6.6)
35.0 ± 13
NR
Pennsylvania: 48 California: 41 37.1
41.0
39.5 ± 12.6
37.6
History of IPV: 44.0 ± 10 No history of IPV: 45.0 ± 12
50.4 ± 16
33.5 (7.5)
34.0
Written survey Lifetime, 1 year, acute, 1 month
Timelines presented
2,400 In-person 1 year interview Lifetime, 1 year 3,455 In-person interview, Chart review, Written survey
Phone Lifetime, 1 year interview 406 In-person Lifetime, 1 year, interview current 72 Written survey Lifetime, 1 year, current, most recent relationship, past relationship 1,401 In-person Lifetime, current, interview past 5 years 3,664 In-person Lifetime, 1 year, interview current, acute 983 Written survey Lifetime, 1 year, current, acute 3,455 In-person Lifetime, 1 year, interview acute 799 In-person Lifetime interview 42 In-person Lifetime, past interview relationship 207 Written survey Current, past relationship
734
648
M age of Sample Method of data participants (years) size collection
High
73.7
(continued)
Moderate
High
Moderate
High
High
High
High
NR
NR
65.6
63.9
74.4
80.4
74.1
High
High
92.3
87.3
High
High
High
Quality scorea
78.7
74.0
77.8
Response rate (%)
125
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
Family
Family, Emergency Emergency
Family
Family
Internal
1997 United States
2004 United States
1999 South Africa
1998 Australia
1995 United States
2005 United States
M. Johnson and Elliott
Kramer, Lorenzon, and Mueller Krishnan, Hilbert, and Pase Marais, de Villiers, Moller, and Stein Mazza, Dennerstein, and Ryan McCauley et al.
McCloskey et al.
2001 United States
Family
Family, Emergency, ObGyn, Pediatrics, Addictions
ObGyn
Emergency
Family
2002 Australia
Houry, Kemball, 2006 United States Rhodes, and Kaslow John, Johnson, Kukreja, 2004 United Found, and Lindow Kingdom D. Johnson 1997 United States
Family
2001 Israel
Medical specialty
Grynbaum, Biderman, Levy, and PetasneWeinstock Hegarty
Location
Year
Study
Table 1. (continued)
English-speaking adults, able to read, no mental diminishment, not acutely ill English, Spanish, Chinese or Russian speaking, 18+, some pregnant women
18+
18+, English and Spanish speaking 18+, English and Spanish speaking 18+
18+, English speaking, no dementia
18-94 years, rural
16+, English speaking, women presenting with child patients also included African American, 18-55, spoke English at grade 5 level All women
Hebrew and Russian speaking, 18-60 years
Population
Computer 1 year survey Written survey Lifetime, 1 year
1,952 Written survey Lifetime, 1 year
2,650 Written survey Lifetime, 1 year
NR NR
35.0 ± 13.4
NR
87
Written survey Lifetime, 1 year, current 127 In-person Lifetime, interview current, past relationship 1,268 Written survey Lifetime, 1 year
280
825
461
In-person Lifetime, current interview 1,050 In-person Lifetime interview 2,181 Written survey 1 year
NR
NR
65.0
NR
NR
32.4
1,836 Written survey Lifetime, 1 year, 5 years
NR
Written survey Lifetime, 1 year
133
Timelines presented
39.2
M age of Sample Method of data participants (years) size collection
62.4
81.6
72.1
99
70
(continued)
High
High
High
Moderate
Low
High
Moderate
63.5
9
Moderate
Moderate
High
High
Moderate
Quality scorea
NR
89.7
75.8
78.5
95.7
Response rate (%)
126
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
Pediatric Emergency Family Family Family
1998 United States
2005 United States
Muelleman, Lenaghan, and Pakieser Newman, Sheehan, and Powell Paul, Smith, and Long Peralta and Fleming Richardson et al.
aThe
2005
China
ObGyn
Family Family
Emergency
2007 United States
2006 Spain 1995 United States
Emergency
2006 Ireland 2003 United States 2002 United Kingdom 1996 Australia
Emergency
18-60, Mandarin speaking
18-65, Spanish speaking, literate 18+, urban and rural setting, White and African American
18+, English/Spanish speaking, admitted to hospital
English- and Spanish-speaking women with children 18-65, women and men 18-36, English speaking 16+, English, Turkish or Bengali speaking 16+, English speaking, men and women
19-65 years, English speaking
18-44, African American, White and Hispanic (predominant), English/Spanish speaking 18-50 years, English speaking
Population
31.3 (7.5)
38.8 (11.2) NR
37.0
45.1
NR NR NR
32.0 ± 10
NR
NR
NR
1 year
Timelines presented
Written survey 1 year
600
In-person interview
Lifetime, 1 year
87.6
86.0 86.2
75.1
Lifetime Phone interview, Written survey 1,402 Written survey Lifetime 407 In-person Lifetime interview 321
67.7
77.2 NR 55.1
97.8
73.3
100
NR
Response rate (%)
Written survey Lifetime
533
139 Written survey Lifetime 399 Written survey Past 3 months 1,207 Written survey Lifetime, 1 year
451
In-person Lifetime, 1 year interview 4,501 Written survey Lifetime, current
421
7,443 In-person interview
M age of Sample Method of data participants (years) size collection
quality score is based on the STROBE Statement for cross-sectional studies. NR = not reported.
Ruiz-Perez et al. Wagner, Mongan, Hamrick, and Hendrick Xu et al.
Roberts, O’Toole, Raphael, Lawrence, and Ashby Roche, Moracco, Dixon, Stern, and Bowling
Emergency
1999 United States
McLaughlin et al.
Family
2005 United States
Medical specialty
McFarlane, Groff, O’Brien, and Watson
Location
Year
Study
Table 1. (continued)
High
High High
High
High
High High High
High
Moderate
Moderate
Moderate
Quality scorea
127
Sprague et al.
were considerably higher (59%; Table 4). Prevalence rates of IPV in the past year were also similar across studies in family practice and emergency clinics (19.9% vs. 19.5%, respectively; Tables 2 and 3). Between 2% and 4.3% of female patients presenting to emergency departments were reported to have injuries caused by a current or former intimate partner (Table 2). Based on a single study in each of pediatric emergency medicine and public health clinics, these specialties have lower rates of physical IPV than other specialties in the reviewed articles (Table 4). Please see Figure 2 for pooled prevalence rates of IPV across specialties.
Discussion Our systematic review and meta-analysis found that the pooled prevalence rates of IPV in emergency medicine and family medicine are similar for both lifetime and 1-year timelines. Pediatric emergency medicine and public health clinics had lower rates of physical IPV than other specialties, although there were fewer studies in these specialties than others. Physical IPV was more commonly reported than both emotional IPV and sexual IPV, and the rates of sexual abuse tend to be lower than physical and emotional abuse, with emotional abuse being the most common type of IPV. Several strengths contributed to the quality of this study. A comprehensive search of the literature and a data abstraction from eligible studies were conducted by two independent reviewers to ensure accuracy, and the articles were screened thoroughly and systematically for inclusion criteria. There was a high degree of agreement between the two reviewers, and many studies of general high quality are included in the review. Also, the results can be highly generalized due to the broad eligibility criteria. Despite these strengths, our study is limited by a large degree of heterogeneity across the included studies. Heterogeneity was likely a product of many factors including different methodology, populations, and definitions of IPV utilized across the eligible studies. The included studies had varying population characteristics and utilized several different screening tools, which likely contributed to the wide range in prevalence rates reported. Despite the large variations, our pooled data show that the prevalence of IPV in emergency medicine and family medicine are similar. Another limitation is the moderately low intraclass correlation coefficient for agreement on methodological quality. This is likely due to the chance for greater variability with an extensive 13-item checklist. The definition of IPV used is essential to compare studies of IPV prevalence. Differences in definitions include the timeline reported, screening tool used, type(s) of IPV included, and definition of perpetrator and victim. Nine different timelines were presented by included studies (lifetime, 1 year, acute, 1 month, 3 months, most recent relationship, current relationship, past relationship, and past 5 years). The included studies also used a variety of screening tools that have been validated to differing degrees, or adaptations of screening tools that have not undergone validation studies. Rabin, Jennings, Campbell, and Bair-Merritt (2009) reported that even the most commonly studied screening tools are only validated in a small number of studies. Since
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
128
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
aRates
are adjusted
Lifetime Past year Past 6 months Past month Current relationship Most recent relationship Acute Past Relationship Cumulative
Timeline
Abbott, Johnson, Koziol-McLain, and Lowenstein (1995) Roberts, O’Toole, Raphael, Lawrence, and Ashby (1996) Muelleman, Lenaghan, and Pakieser (1998) McLaughlin et al. (1999) Krishnan, Hilbert, and Pase (2001) Cox et al. (2004) Kramer, Lorenzon, and Mueller (2004) McCloskey et al. (2005) El-Bassel et al. (2007) Roche, Moracco, Dixon, Stern, and Bowling (2007) Abbott, Johnson, Koziol-McLain, and Lowenstein (1995) Dearwater et al. (1998) McLaughlin et al. (1999) Cox et al. (2004) Kramer et al. (2004) McCloskey et al. (2005) Houry, Kemball, Rhodes, and Kaslow (2006) El-Bassel et al. (2007) Abbott et al. (1995) Ernst, Nick, Weiss, Houry, and Mills (1997) Muelleman et al. (1998) Krishnan et al. (2001) Cox et al. (2004) Krishnan et al. (2001) Abbott et al. (1995) Dearwater et al. (1998) Glass, Dearwater, and Campbell (2001) Cox et al. (2004) Ernst et al. (1997) Abbott et al. (1995)
Study
Table 2. Prevalence in Emergency Medicine.
27.7 (23.4-32.2) 16.1 NR 47.3 NR 51 (49-53) NR 41a 49.6 33.3 15.3 (12.6-18.3) 14.4 12.1 25 (23-39) NR 16.5a 36.0 11.8 11.9 (9.5-14.6) NR NR 18.2 9 (7-12) 32.3 NR 2.2 4.3 2 NR 54.2
Overall IPV rate (%) 23.1 NR 37 NR 20.7 NR 58.1 NR 44.2 27.1 NR 12.6 NR NR 20.4 NR 22.0 8.9 6.5 18.8 5.9 NR NR NR 2.2 NR NR NR 32.2 NR
Physical IPV rate (%) NR NR NR NR 5.7 NR 33.9 NR 24.9 5.6 NR 4.6 (3.8-5.2) NR NR 6.5 NR 8.9 6.4 NR NR NR NR NR NR NR NR NR NR NR NR
Sexual IPV rate (%) NR NR NR NR 23.0 NR 67.7 NR NR NR NR NR NR NR 37.8 NR 32.1 NR NR 15.5 NR NR NR NR NR NR NR NR 22.3 NR
Emotional IPV rate (%)
129
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
aRates
are adjusted.
Lifetime Past year Past 3 months Past 5 years Over 5 years ago Current relationship Most recent relationship Past relationship
Timeline
Study
Elliott and Johnson (1995) Wagner, Mongan, Hamrick, and Hendrick (1995) D. Johnson (1997) M. Johnson and Elliott (1997) Marais, de Villiers, Moller, and Stein (1999) Coker, Smith, McKeown, and King (2000) Grynbaum, Biderman, Levy, and Petasne-Weinstock (2001) Richardson et al. (2002) Hegarty (2004) Kramer, Lorenzon, and Mueller (2004) Paul, Smith, and Long (2006) Ruiz-Perez et al. (2006) D. Johnson (1997) Mazza, Dennerstein, and Ryan (1998) Bauer, Rodriguez, and Perez-Stable (2000) Grynbaum et al. (2001) Hegarty (2002) Richardson et al. (2002) Kramer et al. (2004) McCloskey et al. (2005) Peralta and Fleming (2003) Coker et al. (2007) Hegarty (2002) D. Johnson (1997) M. Johnson and Elliott (1997) Coker et al. (2000) Coker et al. (2007) Coker et al. (2000) M. Johnson and Elliott (1997) Coker et al. (2000)
Table 3. Prevalence in Family Medicine.
45 66.1 NR 44.9 21.5 55.1 30.8 27.7 (25-30) 37 (31-42) NR NR 31.7 NR 28.9 (24-34) 13 NR 27.6 NR NR 8.6a 44.3 25.6 41.5 NR 20.5 14.8 13.3 33.0 27.6 46.0
Overall IPV rate (%) 31 38.8 13.9 29.1 14.6 37.6 NR 41.1 (38-44) 23.3 53.8 43.2 14.1 8.2 22.4 NR 9.8 5.0 16.8 (14-19) 9.6 NR 10.3 10.1 NR NR NR 14.8 16.3 18.9 NR 32.0
Physical IPV rate (%) 19 24 2.1 15.0 1.6 23.1 NR NR 10.6 24.2 NR 8.8 NR NR NR NR 1.9 NR 3.7 NR NR 24.3 NR NR NR 5.1 13.3 14.4 NR 17.3
Sexual IPV rate (%)
31 55 NR 29.1 NR NR NR NR 33.9 54.2 76 30.7 NR 20.1 (16-24) NR NR 7.6 NR 25.4 NR 40.6 NR NR 5.0 NR NR NR NR NR 10.1
Emotional IPV rate (%)
130
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
Past year
John, Johnson, Kukreja, Found, and Lindow (2004) Xu et al. (2005) Bauer, Rodriguez, and Perez-Stable (2000) John et al. (2004) Faramarzi, Esmailzadeh, and Mosavi (2005) Xu et al. (2005) Cleary, Keniston, Havranek, and Albert (2008) McCauley et al. (1995) Caralis and Musialowski (1997) Bauer et al. (2000) McCloskey et al. (2005) Cleary et al. (2008) Caralis and Musialowski (1997)
Study
aRates
are adjusted.
Current relationship Addiction recovery Lifetime McCloskey et al. (2005) Past year McCloskey et al. (2005) Pediatrics (adults accompanying Past year McCloskey et al. (2005) pediatric patients) Pediatric emergency (adults Past year Newman, Sheehan, and Powell (2005) accompanying pediatric patients) Public health Past year McFarlane, Groff, O’Brien, and Watson (2005)
Internal medicine
Lifetime
Past year
Lifetime
Timeline
Obstetrics/gynecology
Specialty
Table 4. Prevalence in Other Specialties.
research-article2014
VAWXXX10.1177/1077801213520574Violence Against WomenSprague et al.
Research Note
Prevalence of Intimate Partner Violence Across Medical and Surgical Health Care Settings: A Systematic Review
Violence Against Women 2014, Vol. 20(1) 118–136 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1077801213520574 vaw.sagepub.com
Sheila Sprague1, J. Carel Goslings2, Celine Hogentoren2, Simone de Milliano2, Nicole Simunovic1, Kim Madden1, and Mohit Bhandari1
Abstract Intimate partner violence (IPV) is a serious health problem and a leading cause of nonfatal injury in North American females. Prevalence of IPV has ranged from less than 20% to more than 50% across primary care, emergency medicine, and family medicine. We conducted a systematic review and meta-analysis of the literature to examine best estimates of IPV prevalence as opportunities for targeted interventions in health care specialties. We included 37 articles in this study. Based on our pooled data, best estimates of the lifetime prevalence of any type of IPV were 38% in family medicine and 40% in emergency medicine. Keywords intimate partner violence, prevalence, systematic review
Introduction Intimate partner violence (IPV) against women has been identified as a serious health problem and is a leading cause of nonfatal injury in females in North America (Dannenberg, Baker, & Li, 1994; Kyriacou et al., 1999). IPV is described by the American Medical Association as “a pattern of coercive behaviors that may include 1McMaster 2University
University, Hamilton, Ontario, Canada of Amsterdam, The Netherlands
Corresponding Author: Sheila Sprague, Department of Clinical Epidemiology and Biostatistics, McMaster University, 293 Wellington St. North, Suite 110, Hamilton, Ontario, Canada L8L 8E7. Email: [email protected]
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
119
Sprague et al.
repeated battering and injury, psychological abuse, sexual assault, progressive social isolation, deprivation, and intimidation” (McCloskey et al., 2007; p. 1067). These behaviors are perpetrated by someone who is in a private and personal relationship with another individual, such as a victim’s spouse, common-law partner, sexual partner, or dating partner. Some studies have reported that as high as 54% of all women have experienced some form of IPV during the course of their life (Reisenhofer & Seibold, 2007). Women who are victims of IPV have been known to utilize health care services at higher rates than women who have not been abused (Campbell, 2002). They may present to a number of different medical specialists including emergency room physicians, orthopedic or trauma surgeons, family physicians, and specialists in obstetrics and gynecology, during routine appointments or for appointments specific to their injuries. Researchers and advocates have argued that health care providers can play a vital role in detecting IPV (Davidson et al., 2001). Despite these recommendations for universal screening, the response from the medical community to address the issue of domestic violence has been characterized as “slow and inconsistent” (Davis, 2008). IPV is an underrecognized, recurrent part of trauma, present in at least one in five women who present to the Emergency Department, and physicians frequently fail to make this diagnosis (Davis, 2008). The failure to detect IPV contributes to recidivism and long-term health problems (Glass, Dearwater, & Campbell, 2001). In addition, when injuries resulting from IPV are not recorded in medical records, abused women are denied documentation of their injuries for future references in court cases (Glass et al., 2001). In addition, opportunities to provide education about prevention, lethality assessment, safety planning, and options for escaping an abusive situation and referrals to resources within the community may also be missed (Glass et al., 2001). The reported rates of IPV vary across different subspecialties and may be underreported. While several studies have reported the prevalence rates of IPV in the health care setting, these estimates vary substantially. Best estimates of rates across various subspecialties are lacking. We therefore undertook a systematic review and meta-analysis of the published literature to determine the prevalence rates of IPV reported to different medical subspecialties including family medicine, emergency medicine, obstetrics and gynecology, internal medicine, addiction recovery clinics, and adults with child patients in pediatric clinics. The study has important implications for health care professionals in various settings. It will direct attention to medical practitioners in subspecialties where IPV is overlooked and will affect their interaction with and treatment of IPV victims as well as victims’ willingness to disclose. In addition, identification of health care settings with the highest prevalence rates will allow targeted research and focus toward improvement of the quality of life of IPV victims.
Method This systematic review adheres to the reporting guidelines of the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) Statement (Stroup et al., 2000).
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
120
Violence Against Women 20(1)
Eligibility Criteria We identified articles in all languages that met the following eligibility criteria: (a) The study was published between January 1995 and July 2009, (b) the target population consisted of adult women between the ages of 16 and 65 years presenting to physicians of any medical specialty, and (c) the primary objective of the study was to determine the prevalence rates of IPV. We included all study designs including cross-sectional designs, surveys, and chart or medical record reviews. We excluded studies focusing on children, males, and special populations such as pregnant women, elderly patients, and patients suffering from a specific illness (i.e., depression, chronic gastroenterological disease). We also excluded review articles, letters, comments, case reports, and guidelines.
Identification of Studies Two reviewers (C.H. and S. dM.) independently conducted a search of the electronic database PubMed for relevant articles published from January 1995 to July 2009 in all languages with the help of a professional librarian. Additional strategies used to identify studies included consultation with experts, a review of reference lists from articles that fulfilled our eligibility criteria, and use of the “related articles” feature in PubMed for all studies meeting our entry criteria.
Assessment of Study Eligibility Two of the authors (C.H. and S. dM.) independently assessed all studies identified for full evaluation and resolved disagreements through discussion toward consensus. The eligibility was verified by two additional authors (S.S. and N.S.). Agreement between observers for study eligibility with regard to abstract and full-text review was examined using the κ (kappa) statistic. The interobserver agreement in methodological quality scores was evaluated using the intraclass correlation coefficient. We chose an a priori criterion of κ ≥ .65 for adequate agreement. The 95% confidence intervals were calculated using the generic formula: Estimate ± 1.96 × (Standard Error).
Assessment of Methodological Quality Two reviewers (N.S. and S.S.), both with methodological expertise and one with content expertise (S.S.), independently graded the methodological quality of each included study using questions adapted from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement designed for cross-sectional studies (von Elm et al., 2008). The STROBE statement is a checklist of 22 items that are considered for good reporting of observational studies. The STROBE statement was developed to assist authors when writing analytic observational studies, to support editors and reviewers when considering such articles for publication, and to help readers critically appraise published research (von Elm et al., 2008).
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
121
Sprague et al.
The original authors emphasized that the STROBE statement was not developed as a tool for assessing the quality of published observational research (von Elm et al., 2008). Unfortunately, very few checklists exist for critically appraising cross-sectional studies (Katrak, Bialocerkowski, Massy-Westropp, Kumar, & Grimmer, 2004; Kelley, Clark, Brown, & Sitzia, 2003; Sanderson, Tatt, & Higgins, 2007), and we did not find that any of the previously developed checklists were appropriate for this study. Therefore, we adapted the STROBE statement to assess study quality. A priori we determined that studies that met at least 10 of the 13 reporting criteria within our modified STROBE criteria would be considered high quality; studies that met 7 to 9 reporting criteria would be considered medium quality; and studies that met 6 or fewer criteria would be classified as low quality.
Data Extraction A structured data extraction form was developed and relevant data were extracted from each eligible study in duplicate (S.S. and N.S.) to ensure accuracy. Pertinent data included study characteristics (year and location), patient characteristics (age), type of medical subspeciality (emergency medicine, family medicine, obstetrics and gynecology, internal medicine, addiction recovery, pediatrics, and public health), method of data collection (survey, interview, chart review, other design), time frame (lifetime, 1 year, 2 years, and other), sample size, and response rate. We documented how the authors defined overall IPV as well as physical, emotional, and sexual abuse. Finally, we recorded the reported rates of overall IPV and of physical, emotional, and sexual abuse as reported by the authors.
Data Analysis We provide pooled rates of lifetime and 1-year physical abuse, emotional abuse, and sexual abuse for emergency medicine, family medicine, and other specialties. Analyses were performed with Predictive Analytics SoftWare (PASW) Statistics version 18.0 (SPSS, Inc., Chicago, IL).
Evaluation of Heterogeneity Before analyzing the data, we hypothesized that there would be a large degree of heterogeneity across the studies. Differences such as study methodology (e.g., survey vs. interview vs. other study designs), timelines of reported IPV (e.g., lifetime vs. 1 year), differences in screening tools used (e.g., Abuse Assessment Screen (AAS), Conflict Tactics Scale (CTS), Partner Violence Screen (PVS)), variability within and between populations (e.g., socioeconomic status, age, jurisdiction), differences in the definition of IPV perpetrator (e.g., male perpetrator vs. perpetrator of any gender, or spouse vs. any intimate partner), or variable study quality could contribute to the heterogeneity.
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
122
Violence Against Women 20(1)
Results Study Identification Our literature search identified 894 potentially relevant citations, of which 32 were considered for inclusion after full-text review (Figure 1). Eighteen of these 32 studies were excluded after verification by additional reviewers (N.S. and S.S.) because of insufficient data on IPV prevalence, or they did not meet the eligibility criteria. Of 80 studies considered for inclusion based on review of bibliographies, 20 were included in this study. Three studies were identified and included after consultation with a content expert. In total, 37 studies were included in this review. The weighted kappa on overall agreement between reviewers when choosing articles for the full-text review was .70 (95% CI = [0.59, 0.82]), and for final study inclusion there was only one disagreement.
Study Characteristics Details of the selected studies are shown in Table 1. The mean age of participants in the included studies was between 28 and 65 years, with 17 studies not reporting mean age. Response rates ranged from 9% to 100% with most studies (24/37, 64.9%) reporting a response rate of 70% or greater. Twenty-five studies (67.6%) took place in the United States, 3 studies took place in each of Australia and the United Kingdom, and other locations (Canada, Spain, Iran, Israel, South Africa, and China) each having 1 study. Most studies took place in family medicine clinics (15/37, 40.5%) and emergency departments (12/37, 32.4%). Of the remaining studies, 3 took place in obstetrics and gynecology clinics, three took place in internal medicine clinics, and 4 studies had more than one specialty. Nineteen studies (51.4%) used written questionnaires to assess rates of IPV, 14 studies (37.8%) used an in-person interview, 1 study (2.7%) used a telephone survey, 1 study (2.7%) used a computer-based survey, and 2 studies (5.4%) used more than one method.
Study Quality We judged 26 studies to be of high methodological quality, 10 studies to be of moderate quality, and the remaining 1 study to be of low quality (Table 1). The reviewers of methodological quality achieved moderate agreement (intraclass correlation coefficient, 0.65; 95% CI = [0.57, 0.71]).
Definition of IPV The definition of IPV varied greatly among studies. The two most commonly used screening tools among the studies were the Abuse Assessment Screen (13 studies, 35.1%) and the Conflict Tactics Scales (5 studies, 13.5%), or modified versions of these tools. Many studies used gender-neutral language, such as “partner,” but others
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
123
Sprague et al.
PubMed search: Citations identified based on title (n=894) Excluded based on title (n=471)
Further review of abstracts (n=423) Excluded based on abstract review (n=343) Further review of full-text (n=80) Excluded based on full-text review (n=48) Additional verification of full-text (n=32)
Hand search of bibliography (n=80) Excluded based on full-text (n=60): Specialized population (n=20) Prevalence is not primary focus (n=19) Review (n=7) Duplicate (n=4) Not presenting to health professional (n=3) Not IPV (n=2) Physicians surveyed only (n=2) Pre-1995 (n=1) Does not look at specific subspecialty (n=1) Doesn’t separate specialties (n=1)
Excluded based on full-text review (n=18): Specialized population (n=6) Duplicate (n=5) Prevalence is not primary focus (n=4) Not IPV (n=1) Not presenting to health professional (n=1) Doesn’t separate specialties (n=1) Consultation with content expert (n=3)
Excluded based on fulltext review (n=0)
Included from reference search (n=20)
Included from literature search (n=14)
Included from expert consultation (n=3)
Articles included in systematic review (n=37)
Figure 1. Flowchart of study process. Source. Adapted from Guyatt et al. (2007).
specified that the abuser must be male (8 studies, 21.6%). Most studies included only female victims, but a small number (3 studies, 8.1%) also included male victims.
IPV Prevalence Twenty-nine of the included studies (78.4%) presented an overall rate of IPV. Overall pooled lifetime prevalence rates were similar across family medicine and emergency medicine (38% and 40%, respectively). Rates reported for other specialties, however,
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
124
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
Year
Location
Family
2000 United States
2007 United States
2004 Canada
1998 United States
2007 United States
1995 United States
1997 United States
Coker, Smith, McKeown, and King Coker et al.
Cox et al.
Dearwater et al.
El-Bassel et al.
Elliott and Johnson
Ernst, Nick, Weiss, Houry, and Mills
18-65 English/Spanish speaking
Emergency
ObGyn
18+, inner city, English speaking, excluded if they never had a partner, men and women Iranian women, married
18-65, English speaking
18-65, insured, in relationship for at least 6 months Lower income women in rural clinics 16+, English or French speaking, able to communicate 18+, English and Spanish speaking, able to communicate 18+
Emergency
Family
Emergency
Emergency
Emergency
Family
2008 United States
Faramarzi, Esmailzadeh, 2005 Iran and Mosavi Glass, Dearwater, and 2001 United States Campbell
English, Spanish, 18-86, 7% pregnant
Population
Family, ObGyn, 18-46, 31% African American, Internal 36% Latina, 31% White Internal Female veteran patients, English speaking Internal 18-60, not pregnant, English/ Spanish speaking
Emergency
Medical specialty
Cleary, Keniston, Havranek, and Albert
1995 United States Abbott, Johnson, Koziol-McLain, and Lowenstein Bauer, Rodriguez, and 2000 United States Perez-Stable Caralis and Musialowski 1997 United States
Study
Table 1. Characteristics of Included Studies.
NR
28.2 (6.6)
35.0 ± 13
NR
Pennsylvania: 48 California: 41 37.1
41.0
39.5 ± 12.6
37.6
History of IPV: 44.0 ± 10 No history of IPV: 45.0 ± 12
50.4 ± 16
33.5 (7.5)
34.0
Written survey Lifetime, 1 year, acute, 1 month
Timelines presented
2,400 In-person 1 year interview Lifetime, 1 year 3,455 In-person interview, Chart review, Written survey
Phone Lifetime, 1 year interview 406 In-person Lifetime, 1 year, interview current 72 Written survey Lifetime, 1 year, current, most recent relationship, past relationship 1,401 In-person Lifetime, current, interview past 5 years 3,664 In-person Lifetime, 1 year, interview current, acute 983 Written survey Lifetime, 1 year, current, acute 3,455 In-person Lifetime, 1 year, interview acute 799 In-person Lifetime interview 42 In-person Lifetime, past interview relationship 207 Written survey Current, past relationship
734
648
M age of Sample Method of data participants (years) size collection
High
73.7
(continued)
Moderate
High
Moderate
High
High
High
High
NR
NR
65.6
63.9
74.4
80.4
74.1
High
High
92.3
87.3
High
High
High
Quality scorea
78.7
74.0
77.8
Response rate (%)
125
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
Family
Family, Emergency Emergency
Family
Family
Internal
1997 United States
2004 United States
1999 South Africa
1998 Australia
1995 United States
2005 United States
M. Johnson and Elliott
Kramer, Lorenzon, and Mueller Krishnan, Hilbert, and Pase Marais, de Villiers, Moller, and Stein Mazza, Dennerstein, and Ryan McCauley et al.
McCloskey et al.
2001 United States
Family
Family, Emergency, ObGyn, Pediatrics, Addictions
ObGyn
Emergency
Family
2002 Australia
Houry, Kemball, 2006 United States Rhodes, and Kaslow John, Johnson, Kukreja, 2004 United Found, and Lindow Kingdom D. Johnson 1997 United States
Family
2001 Israel
Medical specialty
Grynbaum, Biderman, Levy, and PetasneWeinstock Hegarty
Location
Year
Study
Table 1. (continued)
English-speaking adults, able to read, no mental diminishment, not acutely ill English, Spanish, Chinese or Russian speaking, 18+, some pregnant women
18+
18+, English and Spanish speaking 18+, English and Spanish speaking 18+
18+, English speaking, no dementia
18-94 years, rural
16+, English speaking, women presenting with child patients also included African American, 18-55, spoke English at grade 5 level All women
Hebrew and Russian speaking, 18-60 years
Population
Computer 1 year survey Written survey Lifetime, 1 year
1,952 Written survey Lifetime, 1 year
2,650 Written survey Lifetime, 1 year
NR NR
35.0 ± 13.4
NR
87
Written survey Lifetime, 1 year, current 127 In-person Lifetime, interview current, past relationship 1,268 Written survey Lifetime, 1 year
280
825
461
In-person Lifetime, current interview 1,050 In-person Lifetime interview 2,181 Written survey 1 year
NR
NR
65.0
NR
NR
32.4
1,836 Written survey Lifetime, 1 year, 5 years
NR
Written survey Lifetime, 1 year
133
Timelines presented
39.2
M age of Sample Method of data participants (years) size collection
62.4
81.6
72.1
99
70
(continued)
High
High
High
Moderate
Low
High
Moderate
63.5
9
Moderate
Moderate
High
High
Moderate
Quality scorea
NR
89.7
75.8
78.5
95.7
Response rate (%)
126
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
Pediatric Emergency Family Family Family
1998 United States
2005 United States
Muelleman, Lenaghan, and Pakieser Newman, Sheehan, and Powell Paul, Smith, and Long Peralta and Fleming Richardson et al.
aThe
2005
China
ObGyn
Family Family
Emergency
2007 United States
2006 Spain 1995 United States
Emergency
2006 Ireland 2003 United States 2002 United Kingdom 1996 Australia
Emergency
18-60, Mandarin speaking
18-65, Spanish speaking, literate 18+, urban and rural setting, White and African American
18+, English/Spanish speaking, admitted to hospital
English- and Spanish-speaking women with children 18-65, women and men 18-36, English speaking 16+, English, Turkish or Bengali speaking 16+, English speaking, men and women
19-65 years, English speaking
18-44, African American, White and Hispanic (predominant), English/Spanish speaking 18-50 years, English speaking
Population
31.3 (7.5)
38.8 (11.2) NR
37.0
45.1
NR NR NR
32.0 ± 10
NR
NR
NR
1 year
Timelines presented
Written survey 1 year
600
In-person interview
Lifetime, 1 year
87.6
86.0 86.2
75.1
Lifetime Phone interview, Written survey 1,402 Written survey Lifetime 407 In-person Lifetime interview 321
67.7
77.2 NR 55.1
97.8
73.3
100
NR
Response rate (%)
Written survey Lifetime
533
139 Written survey Lifetime 399 Written survey Past 3 months 1,207 Written survey Lifetime, 1 year
451
In-person Lifetime, 1 year interview 4,501 Written survey Lifetime, current
421
7,443 In-person interview
M age of Sample Method of data participants (years) size collection
quality score is based on the STROBE Statement for cross-sectional studies. NR = not reported.
Ruiz-Perez et al. Wagner, Mongan, Hamrick, and Hendrick Xu et al.
Roberts, O’Toole, Raphael, Lawrence, and Ashby Roche, Moracco, Dixon, Stern, and Bowling
Emergency
1999 United States
McLaughlin et al.
Family
2005 United States
Medical specialty
McFarlane, Groff, O’Brien, and Watson
Location
Year
Study
Table 1. (continued)
High
High High
High
High
High High High
High
Moderate
Moderate
Moderate
Quality scorea
127
Sprague et al.
were considerably higher (59%; Table 4). Prevalence rates of IPV in the past year were also similar across studies in family practice and emergency clinics (19.9% vs. 19.5%, respectively; Tables 2 and 3). Between 2% and 4.3% of female patients presenting to emergency departments were reported to have injuries caused by a current or former intimate partner (Table 2). Based on a single study in each of pediatric emergency medicine and public health clinics, these specialties have lower rates of physical IPV than other specialties in the reviewed articles (Table 4). Please see Figure 2 for pooled prevalence rates of IPV across specialties.
Discussion Our systematic review and meta-analysis found that the pooled prevalence rates of IPV in emergency medicine and family medicine are similar for both lifetime and 1-year timelines. Pediatric emergency medicine and public health clinics had lower rates of physical IPV than other specialties, although there were fewer studies in these specialties than others. Physical IPV was more commonly reported than both emotional IPV and sexual IPV, and the rates of sexual abuse tend to be lower than physical and emotional abuse, with emotional abuse being the most common type of IPV. Several strengths contributed to the quality of this study. A comprehensive search of the literature and a data abstraction from eligible studies were conducted by two independent reviewers to ensure accuracy, and the articles were screened thoroughly and systematically for inclusion criteria. There was a high degree of agreement between the two reviewers, and many studies of general high quality are included in the review. Also, the results can be highly generalized due to the broad eligibility criteria. Despite these strengths, our study is limited by a large degree of heterogeneity across the included studies. Heterogeneity was likely a product of many factors including different methodology, populations, and definitions of IPV utilized across the eligible studies. The included studies had varying population characteristics and utilized several different screening tools, which likely contributed to the wide range in prevalence rates reported. Despite the large variations, our pooled data show that the prevalence of IPV in emergency medicine and family medicine are similar. Another limitation is the moderately low intraclass correlation coefficient for agreement on methodological quality. This is likely due to the chance for greater variability with an extensive 13-item checklist. The definition of IPV used is essential to compare studies of IPV prevalence. Differences in definitions include the timeline reported, screening tool used, type(s) of IPV included, and definition of perpetrator and victim. Nine different timelines were presented by included studies (lifetime, 1 year, acute, 1 month, 3 months, most recent relationship, current relationship, past relationship, and past 5 years). The included studies also used a variety of screening tools that have been validated to differing degrees, or adaptations of screening tools that have not undergone validation studies. Rabin, Jennings, Campbell, and Bair-Merritt (2009) reported that even the most commonly studied screening tools are only validated in a small number of studies. Since
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
128
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
aRates
are adjusted
Lifetime Past year Past 6 months Past month Current relationship Most recent relationship Acute Past Relationship Cumulative
Timeline
Abbott, Johnson, Koziol-McLain, and Lowenstein (1995) Roberts, O’Toole, Raphael, Lawrence, and Ashby (1996) Muelleman, Lenaghan, and Pakieser (1998) McLaughlin et al. (1999) Krishnan, Hilbert, and Pase (2001) Cox et al. (2004) Kramer, Lorenzon, and Mueller (2004) McCloskey et al. (2005) El-Bassel et al. (2007) Roche, Moracco, Dixon, Stern, and Bowling (2007) Abbott, Johnson, Koziol-McLain, and Lowenstein (1995) Dearwater et al. (1998) McLaughlin et al. (1999) Cox et al. (2004) Kramer et al. (2004) McCloskey et al. (2005) Houry, Kemball, Rhodes, and Kaslow (2006) El-Bassel et al. (2007) Abbott et al. (1995) Ernst, Nick, Weiss, Houry, and Mills (1997) Muelleman et al. (1998) Krishnan et al. (2001) Cox et al. (2004) Krishnan et al. (2001) Abbott et al. (1995) Dearwater et al. (1998) Glass, Dearwater, and Campbell (2001) Cox et al. (2004) Ernst et al. (1997) Abbott et al. (1995)
Study
Table 2. Prevalence in Emergency Medicine.
27.7 (23.4-32.2) 16.1 NR 47.3 NR 51 (49-53) NR 41a 49.6 33.3 15.3 (12.6-18.3) 14.4 12.1 25 (23-39) NR 16.5a 36.0 11.8 11.9 (9.5-14.6) NR NR 18.2 9 (7-12) 32.3 NR 2.2 4.3 2 NR 54.2
Overall IPV rate (%) 23.1 NR 37 NR 20.7 NR 58.1 NR 44.2 27.1 NR 12.6 NR NR 20.4 NR 22.0 8.9 6.5 18.8 5.9 NR NR NR 2.2 NR NR NR 32.2 NR
Physical IPV rate (%) NR NR NR NR 5.7 NR 33.9 NR 24.9 5.6 NR 4.6 (3.8-5.2) NR NR 6.5 NR 8.9 6.4 NR NR NR NR NR NR NR NR NR NR NR NR
Sexual IPV rate (%) NR NR NR NR 23.0 NR 67.7 NR NR NR NR NR NR NR 37.8 NR 32.1 NR NR 15.5 NR NR NR NR NR NR NR NR 22.3 NR
Emotional IPV rate (%)
129
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
aRates
are adjusted.
Lifetime Past year Past 3 months Past 5 years Over 5 years ago Current relationship Most recent relationship Past relationship
Timeline
Study
Elliott and Johnson (1995) Wagner, Mongan, Hamrick, and Hendrick (1995) D. Johnson (1997) M. Johnson and Elliott (1997) Marais, de Villiers, Moller, and Stein (1999) Coker, Smith, McKeown, and King (2000) Grynbaum, Biderman, Levy, and Petasne-Weinstock (2001) Richardson et al. (2002) Hegarty (2004) Kramer, Lorenzon, and Mueller (2004) Paul, Smith, and Long (2006) Ruiz-Perez et al. (2006) D. Johnson (1997) Mazza, Dennerstein, and Ryan (1998) Bauer, Rodriguez, and Perez-Stable (2000) Grynbaum et al. (2001) Hegarty (2002) Richardson et al. (2002) Kramer et al. (2004) McCloskey et al. (2005) Peralta and Fleming (2003) Coker et al. (2007) Hegarty (2002) D. Johnson (1997) M. Johnson and Elliott (1997) Coker et al. (2000) Coker et al. (2007) Coker et al. (2000) M. Johnson and Elliott (1997) Coker et al. (2000)
Table 3. Prevalence in Family Medicine.
45 66.1 NR 44.9 21.5 55.1 30.8 27.7 (25-30) 37 (31-42) NR NR 31.7 NR 28.9 (24-34) 13 NR 27.6 NR NR 8.6a 44.3 25.6 41.5 NR 20.5 14.8 13.3 33.0 27.6 46.0
Overall IPV rate (%) 31 38.8 13.9 29.1 14.6 37.6 NR 41.1 (38-44) 23.3 53.8 43.2 14.1 8.2 22.4 NR 9.8 5.0 16.8 (14-19) 9.6 NR 10.3 10.1 NR NR NR 14.8 16.3 18.9 NR 32.0
Physical IPV rate (%) 19 24 2.1 15.0 1.6 23.1 NR NR 10.6 24.2 NR 8.8 NR NR NR NR 1.9 NR 3.7 NR NR 24.3 NR NR NR 5.1 13.3 14.4 NR 17.3
Sexual IPV rate (%)
31 55 NR 29.1 NR NR NR NR 33.9 54.2 76 30.7 NR 20.1 (16-24) NR NR 7.6 NR 25.4 NR 40.6 NR NR 5.0 NR NR NR NR NR 10.1
Emotional IPV rate (%)
130
Downloaded from vaw.sagepub.com at WESTERN OREGON UNIVERSITY on May 31, 2015
Past year
John, Johnson, Kukreja, Found, and Lindow (2004) Xu et al. (2005) Bauer, Rodriguez, and Perez-Stable (2000) John et al. (2004) Faramarzi, Esmailzadeh, and Mosavi (2005) Xu et al. (2005) Cleary, Keniston, Havranek, and Albert (2008) McCauley et al. (1995) Caralis and Musialowski (1997) Bauer et al. (2000) McCloskey et al. (2005) Cleary et al. (2008) Caralis and Musialowski (1997)
Study
aRates
are adjusted.
Current relationship Addiction recovery Lifetime McCloskey et al. (2005) Past year McCloskey et al. (2005) Pediatrics (adults accompanying Past year McCloskey et al. (2005) pediatric patients) Pediatric emergency (adults Past year Newman, Sheehan, and Powell (2005) accompanying pediatric patients) Public health Past year McFarlane, Groff, O’Brien, and Watson (2005)
Internal medicine
Lifetime
Past year
Lifetime
Timeline
Obstetrics/gynecology
Specialty
Table 4. Prevalence in Other Specialties.
