Heidi D. Nelson, MD, MPH; Peggy Nygren, MA; Yasmin McInerney, MD; Jonathan Klein, MD, MPH
Disclaimer: The authors of this article are responsible for its contents, including clinical or treatment recommendations. No statement in this article should be construed as an official position of the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services.
Acknowledgments: The authors thank members of the USPSTF and reviewers of the full evidence report for their contributions to this project; Patty Davis for conducting the library search; and Miranda Norbraten for helping prepare the manuscript.
Grant Support: This study was conducted by the Oregon Evidence-based Practice Center under contract to the Agency for Healthcare Research and Quality (contract no. 290-97-0018, Task Order Number 2, Rockville, Maryland). Dr. McInerney was supported by the Veterans Affairs special fellowship in Health Issues of Women Veterans.
Potential Financial Conflicts of Interest: None disclosed.
Requests for Single Reprints: Reprints are available from the Agency for Healthcare Research and Quality Web site (http://www.preventiveservices.ahrq.gov) and through the Agency for Healthcare Research and Quality Publications Clearinghouse (telephone, 800-358-9295).
Current Author Addresses: Dr. Nelson and Ms. Nygren: Oregon Health & Science University, Mail Code BICC 504, 3181 SW Sam Jackson Park Road, Portland, OR 97201.
Dr. McInerney: 13107 Mindanao Way, #1, Marina del Rey, CA 90292.
Dr. Klein: Department of Pediatrics, University of Rochester, 601 Elmwood Avenue, Rochester, NY 14642.
Nelson H., Nygren P., McInerney Y., Klein J.; Screening Women and Elderly Adults for Family and Intimate Partner Violence: A Review of the Evidence for the U.S. Preventive Services Task Force. Ann Intern Med. 2004;140:387-396. doi: 10.7326/0003-4819-140-5-200403020-00015
Download citation file:
Published: Ann Intern Med. 2004;140(5):387-396.
Family and intimate partner violence is common in the United States and is often associated with acute and chronic health problems. Although the clinician's role in identification and intervention is considered a professional, ethical, and sometimes legal responsibility, the effectiveness of screening is uncertain.
To examine evidence on the benefits and harms of screening women and elderly adults in health care settings for family and intimate partner violence.
MEDLINE, PsycINFO, CINAHL, Health & Psychosocial Instruments, AARP Ageline, Cochrane Controlled Trials Register, reference lists, and experts.
The authors selected English-language studies that included original data focusing on the performance of screening instruments (14 studies for women, 3 for elderly persons) and the effectiveness of interventions based in health care settings (2 studies for women, none for elderly persons).
Study design, patient samples and settings, methods of assessment or intervention, and outcome measures were extracted, and a set of criteria was applied to evaluate study quality.
No trials of the effectiveness of screening in a health care setting for reducing harm have been published. Several screening instruments have been developed; some have demonstrated fair to good internal consistency and some have been validated with longer instruments, but none have been evaluated against measurable violence or health outcomes. Few intervention studies have been conducted. Existing intervention studies focused on pregnant women, and study limitations restrict their interpretation.
Although the literature on family and intimate partner violence is extensive, few studies provide data on detection and management to guide clinicians.
As many as 1 to 4 million women are physically, sexually, or emotionally abused by their intimate partners each year in the United States (1, 2), and 31% of all women report abuse in their lifetime (3). Prevalence rates of abuse in clinical samples range from 4% to 44% within the past year and from 21% to 55% over a lifetime (4-14). The incidence of acute cases in emergency care settings ranges from 2% to 7% (15). Approximately 20% of female teenage survey respondents reported being physically or sexually abused by a dating partner (16). Although women also commit violence against men, women are 7 to 14 times more likely to sustain severe physical injury from an assault by an intimate partner (17).
Approximately 551 000 older adults in domestic settings were abused or neglected in 1996 (18). A random-sample survey of a community population indicated a prevalence rate of 32 per 1000 for physical violence, verbal aggression, and neglect (19). Complicating these estimates, however, is the difficulty in defining and quantifying elder abuse. Abuse of elderly persons takes many forms, including physical, sexual, psychological, and financial exploitation as well as neglect (20). Available data indicate that the highest rates of elder abuse are among women and those 80 years of age and older (18). In 90% of cases, the perpetrator is a family member, most often an adult child or spouse (18).
Many health problems are associated with abuse and neglect at all ages. These include repercussions of acute trauma, including death and unwanted pregnancy, as well as long-term physical and mental problems, such as depression, post-traumatic stress disorder, somatization, suicide, and substance abuse (16, 21-30). Children who witness intimate partner violence are at risk for developmental delay, school failure, psychiatric disorders (31, 32), and violence against others (33).
Physician and nursing organizations consider the clinician's role in identification and intervention to be a professional responsibility (34, 35). Reporting child and elder abuse to protective services is mandatory in almost all states; 4 states (California, Colorado, Rhode Island, and Kentucky) have laws requiring mandatory reporting of intimate partner violence. Hospitals are also required to address abuse in order to maintain accreditation (36).
Whether screening leads to a decline in abuse is unknown. In the mid-1990s, after several medical organizations recommended screening for intimate partner abuse, rates of abuse decreased (37). A systematic review reported that most studies of screening for intimate partner violence in health care settings found that screening detected more abused women than no screening (38). Surveys indicate that 43% to 85% of female respondents consider screening in health care settings acceptable, although only one third of physicians and half of emergency department nurses favored screening (38). The evidence on how to screen and effectively intervene once problems are identified is limited, and few clinicians routinely screen patients who do not have apparent injuries (39-44).
In 1996, the U.S. Preventive Services Task Force (USPSTF) concluded that there was insufficient evidence to recommend for or against the use of specific screening instruments to detect family or intimate partner violence, although including questions about abuse in the routine history could be recommended on the basis of prevalence of abuse among adult women and the potential value of the information to clinicians (45). This report is an update on the current literature on family and intimate partner violence. It focuses on studies of the performance of screening instruments designed for the clinical setting and the effectiveness of clinical-based interventions for women and elderly adults. A separate report on screening for family violence in children is available elsewhere (46).
The analytic framework and key questions guiding this review are detailed in the Figure. Relevant studies were identified from multiple searches of MEDLINE (1966 to December 2002), PsycINFO (1984 to December 2002), CINAHL (1982 to December 2002), Health & Psychosocial Instruments (1985 to December 2002), AARP Ageline (1978 to December 2002), and the Cochrane Controlled Trials Register (Appendix). Additional articles were obtained by reviewing 2 recent systematic reviews (38, 47), by reviewing reference lists of pertinent studies, and by consulting experts.
KQ 1: Does screening for family and intimate partner violence reduce harm and premature death and disability? KQ 2: How well does screening identify current harm or risk for harm from family and intimate partner violence? KQ 3: What are the adverse effects of screening? KQ 4: How well do interventions reduce harm from family and intimate partner violence? KQ 5: What are the adverse effects of intervention? *Including physical trauma (such as fractures, dislocations, brain injury); unwanted pregnancy and sexually transmitted diseases; mental trauma; and social isolation and its repercussions, such as depression, anxiety, and nightmares.
We defined screening as an assessment of current harm or risk for harm from family and intimate partner violence in asymptomatic persons in a health care setting. Universal screening assesses everyone; selective screening assesses only those who meet specific criteria. The target populations for this review were women and elderly victims of abuse from family members, intimate partners, caretakers, or others with similar relationships. The USPSTF focused this review on these populations because they are the largest at-risk groups in general primary care settings.
Studies included in this review had English-language abstracts; were applicable to U.S. clinical practice; described abuse and violence against women or elderly adults; were conducted in or linked to primary care (for example, family practice or general internal medicine), obstetrics and gynecology, or emergency department settings; and included a physician or other health care provider in the process of assessment or intervention. We excluded studies about patients presenting with trauma. All eligible studies were reviewed, including those published before the 1996 USPSTF recommendation.
Assessment studies were included if they evaluated the performance of verbal or written questionnaires or other assessment procedures, such as physical examinations, that were brief and applicable to the primary care setting. Included studies described the study sample, the screening instrument or procedure, the abuse or neglect outcome, and the collection of data. Outcomes included indicators of physical abuse, neglect, emotional abuse, or sexual abuse and any reported related health outcomes (for example, depression).
Intervention studies were included if they measured the effectiveness of an intervention in reducing harm from family and intimate partner violence compared with nonintervention or usual care groups. We excluded studies that tested the effectiveness of interventions to educate health care professionals about family violence or to increase screening rates in institutions. We also excluded studies about mandatory reporting laws, descriptions of programs, the accuracy of physician diagnosis and reporting of abuse, and physician factors related to reporting.
From each included study, we abstracted the study design, number of participants, setting, length and type of interventions, length of follow-up, outcomes, methods of outcome measurement, and study duration, among other variables. Two reviewers independently rated each study's quality using criteria specific to different study designs developed by the USPSTF (Appendix) (48). When reviewers disagreed, a final score was reached through consensus.
This research was funded by the Agency for Healthcare Research and Quality under a contract to support the work of the USPSTF. Agency staff and Task Force members participated in the initial design of the study and reviewed interim analyses and the final manuscript. Additional reports were distributed for review to content experts and revised accordingly before preparation of this manuscript. The authors are responsible for the content of the manuscript and the decision to submit it for publication.
Of 806 abstracts identified by database searches, 14 met inclusion criteria. These included 6 studies that compared one instrument with another, 3 that compared an instrument with a directed interview, 2 that measured interrater reliability or internal consistency, and 3 that compared methods of administration. None evaluated the performance of a screening instrument or procedure by using verified abuse outcomes. Screening instruments are described in Appendix Figures 1 and 2 and Appendix Tables 1 and 2(49-61).
Six studies compared brief screening instruments with previously validated instruments and were rated as good or fair in quality (Table 1) (15, 53, 54, 56, 57, 62). Brief instruments were generally correlated with longer instruments and in some cases performed better.
The Hurt, Insulted, Threatened, or Screamed at (HITS) instrument includes 4 questions (54). When administered to 259 women in family practice clinics, it demonstrated fair internal consistency (Cronbach α statistic = 0.80), and its results correlated with the previously validated 19-item Conflict Tactics Scales (CTS) (r= 0.85). In urban emergency department settings, the Partner Violence Screen (PVS), consisting of 3 questions, was compared with the 30-item Index of Spouse Abuse (ISA) (sensitivity, 64.5%; specificity, 80.3%) and the Conflict Tactics Scales (sensitivity, 71.4%; specificity, 84.4%) (53). However, the validity of the Conflict Tactics Scales may not have been tested sufficiently to qualify it as a gold standard in these studies.
A study of 1152 predominantly African-American women presenting for care at university-affiliated family practice clinics found that the 10-item Women's Experience with Battering (WEB) Scale had a higher detection rate (16%) than the 15-item Index of Spouse Abuse—Physical Scale (10%) (56). Another trial studying predominantly white women in family practice clinics found that the 8-item Woman Abuse Screening Tool (WAST) was correlated with the 25-item Abuse Risk Inventory (r= 0.69) (57). A study of pregnant women in public prenatal clinics tested the 3-item Abuse Assessment Screen (AAS) against the Index of Spouse Abuse (62). Women identified as abused on the Abuse Assessment Screen also scored significantly higher on the Index of Spouse Abuse than nonabused women.
The previously validated Abuse Assessment Screen was modified for use in the emergency department setting to detect ongoing abuse rather than abuse within the previous 12 months and was renamed the Ongoing Abuse Screen (OAS) (15). Women presenting to an emergency department were screened with both instruments as well as with a single question about present abuse. The Abuse Assessment Screen yielded positive results in 59% of women screened, and the Ongoing Abuse Screen yielded positive results in 16%. The single question “Are you presently a victim of intimate partner violence?” yielded positive results in 3% of women.
Three studies comparing a screening instrument with an interview were rated as poor quality (51, 52, 55). The major limitation of these studies was that they did not identify a protocol for the directed interview. These studies reported higher detection rates with questionnaires than with interviews.
Two fair-quality studies measured the internal consistency of screening instruments. The Partner Abuse Interview, an 11-item questionnaire modified from the Conflict Tactics Scales, showed fair internal consistency (Cronbach α statistic = 0.82) when tested in 90 women at a suburban family practice clinic and university hospital (49). The WEB Scale, which was tested in primary care clinics and community groups, showed good internal consistency (Cronbach α statistic = 0.99) (63).
Three fair-quality studies compared methods of administration of screening instruments (42, 50, 58). A study of 4641 women presenting to 11 community emergency departments found that the prevalence of past-year and lifetime violence was significantly higher when a questionnaire containing items from the Abuse Assessment Screen was self-administered than when it was administered by a nurse (42). In another study conducted in an emergency department (58), reports of abuse were similar when a questionnaire was given as part of a face-to-face-interview (16%) and when it was administered by tape recorder with a written self-reported answer sheet (15%). In a study at a Planned Parenthood clinic using 4 questions, rates of reported abuse were higher on a nurse-conducted interview (29%) than by self-report (7%) (50).
From 667 abstracts identified by database searches, only 2 studies met inclusion criteria (Table 2). These fair-quality studies evaluated interventions for abused, pregnant women and reported less violence after delivery even when a minimal or “brief” intervention was performed (64, 65). Neither study had a nonintervention control group.
In one study in a prenatal clinic (64), 329 pregnant Hispanic women who had positive results for abuse on the Abuse Assessment Screen were randomly assigned to one of 3 groups: brief, in which they were given a wallet-sized card listing community resources; counseling, in which they received unlimited access to a counselor in the clinic; or outreach, in which they received counseling plus a “mentor mother” in the community. At 2-month follow-up, violence scores measured by using the Severity of Violence against Women Scale were significantly lower in the outreach group than in the counseling group but not in the brief group. However, at 6-, 12-, and 18-month follow-up, violence scores were lower in all groups without statistically significant differences between groups.
In another study of pregnant women in prenatal clinics who had positive results on the Abuse Assessment Screen (65), 132 received 3 counseling sessions and 67 were offered wallet-sized cards listing community resources. At 6 and 12 months after delivery, less violence occurred in the intervention group, as measured by the Index of Spouse Abuse (P= 0.007) and Severity of Violence against Women Scale (P= 0.052). However, differences were not statistically significant.
Of 1045 abstracts identified by database searches, 3 studies of screening instruments for elder abuse met modified inclusion criteria (60, 61, 66)(Table 3). None were developed or tested in traditional clinical settings. However, because the care of elderly adults occurs largely outside these settings, studies were included if it appeared that they could be adapted to clinical settings.
Appendix Table 1.
Appendix Table 2.
In one study (61), a screening instrument for caregivers was tested in 3 groups: abusive caregivers from a social service agency, nonabusive caregivers from a social service agency, and nonabusive caregivers from the community. The Caregiver Abuse Screen (CASE) is based on “yes” or “no” responses to 8 items. Scores on the Caregiver Abuse Screen distinguished abusers from nonabusers (Cronbach α statistic = 0.71) and correlated with the previously validated Indicator of Abuse (IOA) (r= 0.41; P< 0.001) and the Hwalek–Sengstock Elder Abuse Screening Test (HSEAST) (r= 0.26; P< 0.025).
Two studies described screening elderly adults (60, 66). One study (60) evaluated 3 groups: victims of abuse, individuals who were referred to adult protective services and were found not to be abused, and nonabused elderly adults from a family practice clinic. The 15-item HSEAST was administered to all groups and correctly classified 67% to 74% of cases (P< 0.001). The HSEAST was also evaluated in a study of elderly adults living in public housing in Florida (66). Abuse status (past abuse or none) was reported by participants and verified by a social worker who reviewed their records at the housing authority. Scores for abused and nonabused persons were significantly different (mean total score, 4.01 vs. 3.01; P= 0.049). This study also indicated that a 9-item model performed as well as the longer 15-item version, correctly identifying 71.4% of abused persons with 17% false-positive and 12% false-negative rates.
From 1084 abstracts identified by database searches, 72 articles were retrieved for further review. However, none provided data about effective interventions. Some papers described individual elder abuse programs, but none included comparison groups or health outcome measures.
No studies were identified that provided data about the adverse effects of screening or interventions. No screening instrument demonstrated 100% sensitivity and specificity. False-negative test results may hinder identification of those who are truly at risk. False-positive test results, most common in low-risk populations, can lead to inappropriate labeling and punitive attitudes. Additional possible adverse effects of screening and interventions include psychological distress, escalation of abuse and family tension, loss of personal residence and financial resources, erosion of family structure, loss of autonomy for the victim, and lost time from work. Women who leave an abuser can become the target of retaliation, which can lead to homicide (67).
There has been concern that patients may feel uncomfortable or threatened if asked questions about family and intimate partner violence. Most women in a study of screening in prenatal clinics believed it was a good idea (98%) and felt “OK” during the process (96%) when asked at a subsequent visit (68). In another study, only 3% of women found 3 screenings with the Abuse Assessment Screen, during and after pregnancy, unacceptable (69). Although most women presenting with their children to a pediatric emergency department believed screening for intimate partner violence was appropriate, many indicated that their willingness to disclose information might be affected by fear of being reported to child protective services (70). This concern was validated by clinicians in the study, who indicated that they would feel obligated to file a report if violence was present in the home.
A telephone survey of abused and nonabused women in 11 U.S. cities indicated that abused women were less likely to support mandatory reporting than nonabused women (59% vs. 73%; P< 0.01). Respondents believed that victims would be less likely to disclose abuse, that victims would resent someone else having control of the situation, and that reporting would increase the risk for perpetrator retaliation (71, 72).
We identified no studies that directly addressed the effectiveness of screening in a health care setting for reducing harm from family and intimate partner violence or the adverse effects of screening and interventions. Several instruments have been developed for intimate partner violence screening. Some have demonstrated fair to good internal consistency, and some have been validated with longer instruments, although none have been evaluated against measurable violence or health outcomes. The optimal methods of administration have not been determined. Few intervention studies have been conducted, and these focused on pregnant women. Outcomes were based on scores on questionnaires and suggest benefit; however, study limitations restrict interpretation.
Few screening instruments have been developed to identify potential elderly victims of abuse or perpetrators of such abuse. These instruments performed fairly well when administered in studies but have not been tested in health care settings. We found no studies of interventions in elderly persons.
Other systematic reviews of interventions for victims of intimate partner violence found few studies with outcomes other than the health outcomes we sought (38, 47). Referrals to community resources, shelters, social workers, and police often increased when abused women were identified. However, it is not known whether these interventions improved violence or health outcomes because the studies were inadequately designed to answer these questions and provided inconsistent results (38, 47).
The prevalence of abuse and the sensitivity and specificity of screening instruments depend on definitions of abuse (physical, sexual, emotional, and combinations) and acuity (current, past, and any). These definitions are not standardized across instruments. Performance characteristics of screening instruments are difficult to determine because comparisons of scores from instruments and actual episodes of abuse are lacking and the accuracy of self-report varies widely. The effectiveness of specific screening methods and interventions could also vary by setting, delivery, culture, and population.
Self-reported abuse by elderly persons may be compromised by cognitive impairment and overshadowed by other medical problems addressed in health care settings. A more comprehensive approach, including physical examination and caretaker and home evaluations, as well as direct questioning, may be more effective.
There are many gaps in the evidence (73). Definitions and measures of abuse, neglect, severity, and chronicity need to be standardized across studies. Existing screening instruments require more testing and validation in medical settings and in languages other than English (74). Little is known about the course of violence during pregnancy and postpartum periods, health implications for the mother and child, the role of violence in reproductive decision making, and what screening and intervention strategies are most effective for pregnant women. Studies of the effectiveness of treatment programs for abused victims, as well as for perpetrators (75-77), would provide needed evidence that identification and intervention can lead to improved health outcomes. These outcomes should include not only measures of reduced violence but also improved quality of life, mental health, social support, self-esteem, and productivity. The feasibility of screening procedures and interventions in health care settings requires evaluations that consider costs, time, resources, clinician consistency, barriers, and patient adherence. Strategies enlisting and evaluating health systems and community programs are needed (78).
Although the literature on family and intimate partner violence is extensive, few studies provide data on detection and management to guide clinicians. As a result, clinicians confront difficulties in fulfilling their role in prevention and treatment of the adverse health effects of violence.
MEDLINE (1966 to 2002), PsycINFO (1984 to 2002), and Health & Psychosocial Instruments (1985 to 2002) were used for the following search strategy.
spouse abuse/or domestic violence.mp. or battered woman.mp. [mp = title, abstract, heading word, table of contents, key phrase identifiers]
(screening or identity or early detection).mp.
physicians, family/ or “family physicians”.mp.
primary health care/ or “primary care”.mp.
family practice/or “family practice”.mp.
2 or 3 or 4 or 5 or 6
1 or 7
From limit 8 to.
MEDLINE (1966 to 2002) and CINAHL (1982 to 2002) were used for the following search strategy.
spouse abuse/ or domestic violence.mp. or battered women.mp.
((intimate partner or life partner or partner or wife or husband) and (violence or abuse)).mp.
1 or 2
Physicians, Family/ or family physicians.mp.
exp Primary Health Care/ or primary care.mp.
Family Practice/ or family practice.mp.
EMERGENCIES/ or emergency.mp.
exp emergency service, hospital/ or emergency department$.mp.
OBSTETRICS/ or “OBSTETRICS AND GYNECOLOGY DEPARTMENT, HOSPITAL”/ or obstetrics.mp.
4 or 5 or 6 or 7 or 8 or 9 or 10
3 and 11
pc.fs. or prevent$.mp. or intervention.mp. or assessment.mp.
exp counseling/ or counsel$.mp
(patient education or questionnaire$).mp.
interviews/ or interview$.mp.
exp clinical trials/ or clinical trial$.mp.
13 or 14 or 15 or 16 or 17 or 18
12 and 19
From limit 20 to.
from 21 keep 1–151
PsycINFO (1984 to 2002) was used for the following search strategy.
exp Partner Abuse/ or spouse abuse.mp.
exp Battered Females/
exp Family Violence/ or exp Victimization/ or exp Emotional Abuse/ or battered women.mp.
3 and (women or females).mp.
((intimate partner$ or life partner$ or partner or wife or husband) and (violence or abuse)).mp.
1 or 2 or 4 or 5
exp Family Physicians/ or family physicians.mp.
exp Family Physicians/ or family practice.mp.
exp emergency services/ or emergenc$.mp.
exp OBSTETRICS/ or obstetrics.mp.
exp GYNECOLOGY/ or gynecology.mp.
7 or 8 or 9 or 10 or 11 or 12 or 13 15. 6 and 14
(prevent$ or intervention or assessment).mp.
exp counseling/ or counsel$.mp.
exp Client Education/ or patient education.mp.
questionnaires/ or questionnaire$.mp.
exp interviews/ or interview$.mp.
exp at risk populations/ or cohort study$.mp.
(domestic violence or family violence).mp.
(elder$ or aged or old or ageing).mp.
(vulnerable or disabled or handicapped).mp.
(2 or 4) and 3
1 or 5
mass screening/ or screening.mp.
questionnaires/ or questionnaire$.tw.
interview/ or interview$.tw.
7 or 8 or 9
6 and 10
From limit 11 to.
from 12 keep 1–1009
AARP Ageline (1978 to 2002) was used for the following search strategy.
((family or domestic) and (abuse or violence)).mp.
(elder$ or old or ageing or aging or aged or geriatric).mp.
2 and 3
1 or 4
(internal medicine or geriatrics or family physicians or family practice).mp.
(primary care or emergency or emergency services).mp.
1 or 4 or 5
GERIATRICS/ or geriatrics.mp.
Internal Medicine/ or internal medicine.mp.
Family Practice/ or family practice.mp
exp Emergency Service, Hospital/ or emergency department.mp.
7 or 8 or 9 or 10 or 11 or 12 or 13
6 and 14
From limit 15 to.
exp COUNSELING/ or counseling.mp.
INTERVIEWS/ or interviews.mp.
17 or 18 or 19 or 20 or 21 or 22 or 23
16 and 24
from 25 keep 1–129
(elder$ or aged or aging or ageing or old or geriatric).mp.
3 or 4
2 and 5
1 or 6
exp GERIATRICS/ or geriatrics.mp.
internal medicine.mp. or exp Physicians/
exp Family Physicians/ or exp General Practitioners/ or family practice.mp.
exp emergency services/ or emergency$.mp.
8 or 9 or 10 or 11 or 12 or 13
7 and 14
prevention/ or prevent$.mp. or intervention.mp. or assessment.mp.
exp counseling/ or counsel$.mp. or assess$.mp.
exp at risk populations/ or exp cohort analysis/ or cohort stud$.mp.
from 25 keep 1–36
6 or 7
5 and 8
from 9 keep 1–75
The quality rating criteria for diagnostic accuracy studies were as follows: screening test relevant, available for primary care, and adequately described; study uses a credible reference standard, performed regardless of test results; reference standard interpreted independently of the screening test; study handles indeterminate results in a reasonable manner; spectrum of patients included in study; adequate sample size; and administration of reliable screening test. The definition of ratings are as follows, based on these criteria.
A study with a rating of “good” evaluates a relevant, available screening test; uses a credible reference standard; interprets the reference standard independently of the screening test; assesses the reliability of the test; has few indeterminate results or handles them in a reasonable manner, and includes a large number (>100) of broad-spectrum patients with and without disease.
A study with a rating of “fair” evaluates a relevant available screening test; uses a reasonable standard, although not the best; interprets the reference standard independently of the screening test; has a moderate sample size (50 to 100 patients); and includes a “medium” spectrum of patients.
A study with a rating of “poor” has important limitations, such as an inappropriate reference standard, an improperly administered screening test, a biased ascertainment of the reference standard, and a small sample size involving a narrow selected spectrum of patients.
For initial assembly of comparable groups, the following criteria apply. For randomized, controlled trials, adequate randomization is required, including concealment and whether potential confounders were distributed equally among groups. For cohort studies, inception cohorts and potential confounders with restriction or measurement for adjustment in the analysis should be considered.
Additional criteria include the following: maintenance of comparable groups (includes attrition, crossovers, adherence, contamination); important differential loss to follow-up or overall high rates of loss to follow-up; equal, reliable, and valid measurements (includes masking of outcome assessment); clear definition of interventions; important outcomes considered; and adjustment for potential confounders for cohort studies, or intention-to-treat analysis for randomized, controlled trials. Definitions of ratings are as follows, based on these criteria.
A study with a rating of “good” meets all the criteria. Comparable groups are assembled initially and maintained throughout the study (follow-up ≥ 80%). Reliable and valid measurement instruments are used and applied equally to the groups, interventions are spelled out clearly, important outcomes are considered, and appropriate attention is given to confounders in the analysis.
Studies will be graded “fair” if any or all of the following problems occur, without the important limitations noted in the “poor” category below: Generally comparable groups are assembled initially, but some question remains about whether some (although not major) differences occurred in follow-up; measurement instruments are acceptable (although not the best) and generally applied equally; some but not all important outcomes are considered; and researchers account for some but not all potential confounders.
Studies will be graded “poor” if any of the following major limitations exist: Groups assembled initially are not close to being comparable or are not maintained throughout the study, unreliable or invalid measurement instruments are used or are not applied at all equally among groups (including not masking outcome assessment), and key confounders are given little or no attention.
For quality rating of case–control studies, the following criteria apply: accurate ascertainment of cases, nonbiased selection of cases and controls with exclusion criteria applied equally to both, adequate response rate, diagnostic testing procedures applied equally to each group, measurement of exposure accurate and applied equally to each group, and appropriate attention to potential confounding variable. Definition of ratings are as follows, based on these criteria.
A study with a rating of “good” involves appropriate ascertainment of cases and nonbiased selection of case and control participants, equal application of the exclusion criteria to cases and controls, a response rate equal to or greater than 80%, accurate diagnostic procedures and measurements that are applied equally to cases and controls, and appropriate attention to confounding variables.
A study with a rating of “fair” is recent and relevant, without major apparent selection or diagnostic work-up bias, but with a response rate less than 80% or attention to some but not all important confounding variables.
A study with a rating of “poor” has major selection or diagnostic work-up biases, a response rate less than 50%, or inattention to confounding variables.
The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.
Adrian A Boyle
August 13, 2004
Screening women for domestic violence
The article by Nelson et al (1) concludes that intervention studies fail to show any benefit for women suffering domestic violence. They have only considered interventions based in health care and have ignored an extensive literature from psychologists, sociologists and criminologists. There are well conducted controlled trials which show some benefit. (2) (3) Women suffering domestic violence do not only present to physicians but are likely to involve the police and voluntary agencies. Considering interventions based only in healthcare alone is artificial, since this is unlikely to represent the experience of the victim. (4) Though including these studies would probably not have changed the overall conclusions of this review, it is important that researchers in this field consider the wider sources of information before making recommendations.
(1) Nelson HD, Nygren P, McInerney Y, Klein J. Screening Women and Elderly Adults for Family and Intimate Partner Violence: A Review of the Evidence for the U.S. Preventive Services Task Force. Annals of Internal Medicine 2004; 140(5):387-396.
(2) Sullivan CM, Bybee D, I. Reducing violence using community-based advocacy for women with abusive partners. J Consult Clin Psychol 1999; 67(1):43-53.
(3) Sherman L, Berk RA. The specific deterrent effects of arrest for domestic assault. American Sociological Review 1984; 49(2):262-272.
(4) Mirrlees-Black C. Domestic violence: findings from a new British crime survey self completion questionnaire. 1-136. 1999. Home Office Publications.
Results provided by:
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only