Professional Documents
Culture Documents
http://jama.ama-assn.org/cgi/content/full/296/5/530
Topic collections Violence and Human Rights; Violence and Human Rights, Other; Randomized
Controlled Trial; Screening; Author in the Room
Contact me when new articles are published in these topic areas.
Permissions Reprints/E-prints
permissions@ama-assn.org reprints@ama-assn.org
http://pubs.ama-assn.org/misc/permissions.dtl
A
S INTIMATE PARTNER VIO - written self-completed questionnaire, and computer-based self-completed question-
lence (IPV) has gained recog- naire. Two screening instruments—the Partner Violence Screen (PVS) and the Woman
nition as a major public health Abuse Screening Tool (WAST)—were administered and compared with the Compos-
ite Abuse Scale (CAS) as the criterion standard.
problem,1,2 research efforts
have focused on the development of uni- Main Outcome Measures The approaches were evaluated on prevalence, extent
versal screening instruments and proto- of missing data, and participant preference. Agreement between the screening instru-
cols for use in health care settings to iden- ments and the CAS was examined.
tify women exposed to IPV.3 Many Results The 12-month prevalence of IPV ranged from 4.1% to 17.7%, depending on
national medical organizations, govern- screening method, instrument, and health care setting. Although no statistically signifi-
mental agencies, and advocacy groups cant main effects on prevalence were found for method or screening instrument, a sig-
nificant interaction between method and instrument was found: prevalence was lower
have recommended universal or rou-
on the written WAST vs other combinations. The face-to-face approach was least pre-
tine IPV screening,4 although there is a ferred by participants. The WAST and the written format yielded significantly less miss-
lack of research examining its effective- ing data than the PVS and other methods. The PVS and WAST had similar sensitivities
ness on health outcomes for women.5-7 (49.2% and 47.0%, respectively) and specificities (93.7% and 95.6%, respectively).
An ongoing question in the field is Conclusions In screening for IPV, women preferred self-completed approaches over
whether health care professionals should face-to-face questioning; computer-based screening did not increase prevalence; and
routinely screen their female patients for written screens had fewest missing data. These are important considerations for both
exposure to IPV.8 clinical and research efforts in IPV screening.
Previous studies have demonstrated Trial Registration clinicaltrials.gov Identifier: NCT00336297
that women will disclose experiences of JAMA. 2006;296:530-536 www.jama.com
violence in response to screening3-7; how-
ever, few studies have compared meth-
Author Affiliations and a list of the members of the
ods of administration. In a review of IPV ods, including written or computer- McMaster Violence Against Women Research Group
screening in the primary care setting, based questionnaires. The limited appear at the end of this article.
Corresponding Author: Harriet L. MacMillan, MD,
Chuang and Liebschutz9 identified 2 amount of research to date has gener- MSc, FRCP(C), Offord Centre for Child Studies, De-
main approaches to screening: (1) ver- ally compared IPV prevalence on face- partment of Psychiatry and Behavioural Neuro-
sciences, McMaster University, Patterson Building, Che-
bal methods (questions asked by a cli- to-face questioning with a written self- doke Site, 1200 Main St W, Hamilton, Ontario, Canada
nician) and (2) self-administered meth- completed questionnaire. Anderst and L8N 3Z5 (macmilnh@mcmaster.ca).
530 JAMA, August 2, 2006—Vol 296, No. 5 (Reprinted) ©2006 American Medical Association. All rights reserved.
Health care providers received special- and the Woman Abuse Screening Tool tablet computer and asked to complete
ized training in responding to IPV. (WAST). 19 These 2 measures were the screening instruments (PVS and
The study was approved by the re- selected following a systematic review WAST, randomly ordered), followed
search ethics boards of McMaster Uni- of screening instruments based on by the evaluation questions. If a par-
versity/Hamilton Health Sciences, The their psychometric properties and use ticipant did not respond to a question,
University of Western Ontario/ in settings comparable with those a reminder window appeared; she
London Health Sciences Centre, Cam- in this study. 3 The PVS (3 items) could then answer the question or
bridge Memorial Hospital, and Nor- addresses physical abuse and feelings continue without answering. Once
folk General Hospital’s Medical of safety; when compared with the done, the participant exited the ques-
Advisory Committee. Conflict Tactics Scales (CTS), the sen- tionnaire program and returned the
sitivity was 71.4% and the specificity tablet to the study recruiter. She then
Sample Size and Randomization was 84.4%. 18 The WAST (8 items) completed the demographic questions
Sample size was calculated based on the includes multiple forms of abuse and CAS on paper.
null hypothesis of no differences in 12- (physical, sexual, and emotional) and Written Self-Completed Method.
month IPV prevalence across meth- has good internal consistency (Cron- The participant was given a paper ver-
ods, with ␣ set at .05 (2-tailed test) and bach ␣ coefficient of 0.75), and more sion of the demographic question-
power set at 1−=80%. Based on the than 90% of women reported being naire and screening instruments (PVS
literature,16,17 we expected an overall “comfortable” or “very comfortable” and WAST, randomly ordered), fol-
prevalence of IPV of 15%. It was hy- when administered the WAST in a lowed by the evaluation questions; the
pothesized that prevalence across previous study. 1 9 Both screening questionnaire closed with the CAS.
methods of administration would be: instruments ask about experiences Completed questionnaires were re-
face-to-face, 10%; written, 15%; and within the last 12 months. The instru- turned to the recruiter in a sealed en-
computer, 20%; this required a sample ments and information about scoring velope.
size of 246 per group per care type. are shown in the BOX. Face-to-Face Method With Verbal
Randomization was by day (or shift TheCompositeAbuseScale(CAS),20,21 Questioning by the Health Care
for sites with regular hours longer than a 30-item validated research instrument, Provider. After obtaining consent, the
8 hours) in 6-week (for sites with no was selected as the criterion standard recruiter informed the health care pro-
shifts or 2 shifts) or 9-week (for sites for its comprehensiveness and strong vider of the patient’s participation by
with 3 shifts) periods. A table for each psychometric properties: the Cronbach inserting a pink slip of paper into the
day of the week was created, and a ran- ␣ for each of 4 subscales is greater patient’s chart. Participants were ver-
dom number table was used to assign than 0.85, and they correlate highly bally screened by their health care pro-
clinic shifts to 1 of 3 methods. For with corresponding subscales of the vider with one of the 2 screening
example, the table for a Sunday in an CTS. The CAS was administered to instruments, randomly determined.
emergency department with two determine the agreement of the WAST Due to the nature of the screening
12-hour shifts would have 3 columns and PVS with this instrument. It was method, disclosures became part of
(computerized, written, and face-to- scored as recommended by summing the clinical encounter; women who
face) and 2 rows (day, night). The ran- the frequency scores for the 30 items; disclosed abuse were offered the usual
dom number table determined the order a score of 7 or more was the criterion care provided by that site. Following
of the numbers 1 through 6 in the cells. for exposure to IPV. the screen, the participant completed a
So, for example, the Sunday of week 1 To evaluate participant preference of written version of the demographic
was allocated to written, night shift; the screening approach, women were asked questions, the evaluation items, and
Sunday of week 2 was allocated to com- 3 questions about their method of the CAS.
puterized, day shift, and so on, for the screening: (1) Was it “easy”?; (2) Did
6-week period. This ensured balance you “like answering” [in that way]?; and Statistical Analysis
across shifts and days of the week. The (3) Was it “private enough”? The re- Descriptive statistics were run for
research coordinator created calen- sponses were scaled 1 (“not at all”) to sample characteristics by group. Be-
dars that informed site coordinators of 5 (“very easy” or “a lot”). cause the participant evaluation items
the assignments. The order in which the showed a skewed distribution, items
screening instruments were com- Procedures were dichotomized as less than 5 or 5.
pleted was also randomly varied. After obtaining consent, the on-site Data were considered missing if
study recruiter provided participants abuse status could not be determined.
Measures with 1 of 3 methods, according to the For the abuse instruments, if a partici-
In addition to standard demographic randomization schedule. pant provided sufficient data to score
questions, participants completed Computer-Based Self-Completed positive, she was deemed “positive,”
the Partner Violence Screen (PVS)18 Method. The participant was given a regardless of the number of missing
532 JAMA, August 2, 2006—Vol 296, No. 5 (Reprinted) ©2006 American Medical Association. All rights reserved.
(ease: computerized [12=21.5; P⬍.001] dictive values were almost the same est prevalence. We were surprised by
and written [ 21 = 92.1; P⬍.001]; (PVS, 94.2%; WAST, 94.0%), leading this finding, although the evidence re-
preference: computerized [12 = 121.1; to very similar accuracies (PVS, 89.2%; garding verbal vs written disclosure is
P⬍.001] and written [ 21 = 107.0; WAST, 90.6%). mixed.10-13 It is noteworthy that lower
P⬍.001]; and privacy: computerized written disclosure was specific to the
[ 21 = 36.7; P⬍.001] and written COMMENT WAST.
[12=46.4; P⬍.001]). There were no sta- This randomized trial compared 3 Use of written questionnaires led to
tistically significant interactions be- methods of IPV screening using 2 in- significantly fewer missing data, in con-
tween instrument and method or be- struments on IPV detection, extent of trast with the findings of Anderst et al.10
tween setting and method. missing data, and acceptability of On all 3 measures of acceptability
screening approach, yielding some in- (ease of responding, likeability, and pri-
Test Characteristics teresting findings. Although some lit- vacy), the face-to-face method was least
of Screening Instruments erature suggests that use of computer- preferred by participants. These find-
The estimated test characteristics of the based questionnaires may lead to higher ings have some recent support in the
PVS and WAST screens were com- disclosures of sensitive issues than other literature. A study comparing audio-
pared with the CAS. The sensitivities approaches,24,25 we did not find that taped screening with written screen-
(PVS, 49.2%; WAST, 47.0%) as well as computer-based screening increased the ing in a pediatric emergency depart-
the specificities (PVS, 93.7%; WAST, detection of IPV relative to other screen- ment26 found no statistically significant
95.6%) were very similar. The posi- ing methods. difference in IPV disclosures between
tive predictive value of the WAST We found that there was an interac- the 2 methods but several patterns in
(55.3%) was minimally higher than for tion between method and instrument, women’s preferences. Specifically,
the PVS (47.0%), and the negative pre- with the written WAST having the low- women found the audiotaped method
to be less risky and more private than
the written approach, and among both
Table 1. Sample Characteristics by Group
the entire sample and the subgroup of
No. (%)
women disclosing abuse, the written
Computerized Face-to-Face Written Total and audiotaped methods were signifi-
(n = 769) (n = 853) (n = 839) (N = 2461)
cantly preferred to the idea of disclos-
Married 396 (52.7) 455 (56.9) 469 (57.5) 1320 (55.8)
ing IPV directly to a health care pro-
1 or 2 children at home 342 (45.6) 378 (47.8) 374 (46.2) 1094 (46.6)
vider. Coupled with our findings, and
Born in Canada 643 (85.7) 711 (88.8) 714 (87.5) 2068 (87.4)
those of Glass et al,13 there seems to be
Education ⬎14 y 391 (52.6) 413 (51.7) 425 (52.3) 1229 (52.2)
emerging evidence that direct ques-
Working full- or part-time 343 (45.9) 373 (46.7) 392 (48.1) 1108 (46.9)
tioning by clinicians is less favored by
Main source of income wages or salary 416 (55.7) 467 (58.5) 479 (59.0) 1362 (57.8)
Household income in lowest quintile 157 (21.5) 121 (15.5) 125 (15.9) 403 (17.6)
women compared with self-report ver-
(⬍$24 000)* sions, whether delivered by com-
Age, mean (SD), y 36.7 (11.6) 37.6 (12.1) 36.9 (12.0) 37.1 (11.9) puter, audiotape, or written question-
*Data from Income Statistics Division.23 naire.
Table 2. Observed Prevalence, Missing Data, and Participant Preference by Screening Method
Computerized, % (95% CI) Face-to-Face, % (95% CI) Written, % (95% CI)
534 JAMA, August 2, 2006—Vol 296, No. 5 (Reprinted) ©2006 American Medical Association. All rights reserved.
The estimated sensitivities and speci- In summary, the findings from this sciences and Clinical Epidemiology and Biostatistics,
McMaster University, Hamilton, Ontario; Marilyn Ford-
ficities of both instruments in relation study examining 3 approaches to IPV Gilboe, RN, PhD (coinvestigator), School of Nursing,
to the CAS were remarkably similar; the screening in health care settings sug- The University of Western Ontario, London; Susan Jack,
RN, PhD (coinvestigator), School of Nursing, McMaster
low sensitivity means that a sizeable gest that the face-to-face approach is the University, Hamilton, Ontario; Clare Freeman, CYW,
proportion of women who disclosed ex- least preferred by women, irrespective BA, MSW (coinvestigator), Interval House of Hamil-
posure to IPV on the CAS were not of instrument. With regard to selec- ton Women’s Shelter, Hamilton, Ontario; Amiram
Gafni, PhD (coinvestigator), Department of Clinical Epi-
identified on either the WAST or the tion of method based on prevalence, demiology and Biostatistics and Centre for Health Eco-
PVS. This is likely because the CAS in- however, there was an interaction be- nomics and Policy Analysis, McMaster University,
Hamilton, Ontario; Iris Gutmanis, BSc (PT), PhD (co-
cludes many more questions covering tween method and instrument: it ap- investigator), Department of Epidemiology and Bio-
a broad range of abusive behaviors in pears that the written format of the statistics, Schulich School of Medicine and Dentistry,
The University of Western Ontario, London; Ellen
several domains, including harass- WAST may lead to some underestima- Jamieson, MEd, and Nadine Wathen, PhD (coinves-
ment. tion of disclosure. In theory, sensitiv- tigators), Department of Psychiatry and Behavioural
This study has limitations that need ity of the WAST could be improved by Neurosciences, McMaster University, Hamilton, On-
tario; Barbara Lent, MA, MD, CCFP, FCFP (coinves-
to be considered in interpreting the re- changing the scoring criteria to in- tigator), Department of Family Medicine, Schulich
sults. First, review of the sample char- clude more items (see Box). The re- School of Medicine & Dentistry, The University of
Western Ontario, London; Joyce Lock, MD, CCFP
acteristics by method shows that the sult of least missing data by written self- (EM), FRCP(C) (coinvestigator), Department of Emer-
women completing the computer- completed questionnaire is worth gency Medicine, McMaster University, Hamilton, On-
based screen had a higher proportion of noting, especially for research applica- tario, and Nina’s Place, the Halton Regional Domestic
Violence/Sexual Assault Care Centre, Burlington, On-
participants in the lowest income quin- tions, but also when considering clini- tario; Daina Mueller, RN, BScN, MSc (coinvestiga-
tile compared with those administered cal policies for IPV detection and in- tor), Hamilton Social and Public Health Services De-
partment, and School of Nursing at McMaster
the other 2 approaches. Women of lower tervention. Prevalence, missing data, University, Hamilton, Ontario; Rosana Pellizzari, MD,
socioeconomic status might respond dif- and preference are all important con- MSc, CCFP (coinvestigator), Department of Family and
Community Medicine, University of Toronto and Medi-
ferently to a computer-based approach; siderations for both clinical and re- cal Officer of Health, Perth District Health Unit, Strat-
however, there was no difference in other search efforts in IPV screening. ford, Ontario; Anna Marie Pietrantonio, MSW, RSW
related variables, including education and (coinvestigator), Child Advocacy and Assessment Pro-
Author Affiliations: Department of Psychiatry and Be-
havioural Neurosciences (Drs MacMillan, Wathen, gram, McMaster Children’s Hospital, and Depart-
work status, reducing the likelihood of ments of Psychiatry and Behavioural Neurosciences,
Boyle and Mss Jamieson and Webb), Department of
a bias regarding acceptability of the com- Pediatrics (Dr MacMillan), and Department of Emer- and School of Social Work, McMaster University,
Hamilton, Ontario; Rachelle Sender (Beauchamp) MD,
puter screen. gency Medicine (Dr Worster), McMaster University,
PhD, CCFP, and Diana Tikasz, MSW, RSW (coinves-
Hamilton, Ontario; Department of Epidemiology and
Second, although we attempted to Biostatistics, University at Albany—State University of tigators), Sexual Assault/Domestic Violence Care Cen-
keep all other aspects of the protocol New York (Dr McNutt); and Department of Family tre, Hamilton Health Sciences, Hamilton, Ontario; Helen
Medicine, The University of Western Ontario, Lon- Thomas, RN, MSc (coinvestigator), School of Nurs-
consistent across methods, the re- don (Dr Lent). ing, McMaster University and Hamilton Social and Pub-
sponses of women who underwent the Author Contributions: Drs MacMillan and Boyle and lic Health Services Department, Hamilton, Ontario;
Ms Jamieson had full access to all of the data in the Jackie Thomas, MD, MSc, FRCS(C) (coinvestigator),
face-to-face approach were known to Department of Obstetrics and Gynecology, Univer-
study and take responsibility for the integrity of the
the health care providers asking them. data and the accuracy of the data analysis. sity of Toronto and Mount Sinai Hospital, Toronto, On-
Study concept and design: MacMillan, Wathen, tario; Leslie Tutty, PhD (coinvestigator), School of So-
This aspect could have influenced wom- cial Work, University of Calgary, Calgary, Alberta;
Jamieson, McNutt, Worster, Lent.
en’s willingness to disclose, although in- Acquisition of data: MacMillan, Worster, Lent, Webb. Margo I. Wilson, PhD, MSL (coinvestigator), Depart-
terestingly, there were no consistent Analysis and interpretation of data: MacMillan, ment of Psychology, McMaster University, Hamil-
Wathen, Jamieson, Boyle, McNutt. ton, Ontario; Andrew Worster, MD, MSc (coinvesti-
patterns in disclosure by method. In ad- Drafting of the manuscript: MacMillan, Wathen, gator), Department of Medicine, McMaster University
dition, they were asked questions from Jamieson. and Emergency Medicine at Hamilton Health Sci-
Critical revision of the manuscript for important in- ences, Hamilton, Ontario.
only 1 instrument, either the WAST or tellectual content: MacMillan, Wathen, Jamieson, Consultants: Louise-Anne McNutt, PhD, Depart-
the PVS, to reduce burden on both the Boyle, McNutt, Worster, Lent, Webb. ment of Epidemiology, School of Public Health, Uni-
clinician in having to administer more Statistical analysis: MacMillan, Jamieson, Boyle, versity at Albany, State University of New York; Jef-
McNutt. frey Coben, MD, Department of Emergency Medicine
than 1 questionnaire verbally and the Obtained funding: MacMillan, Wathen. and Community Medicine, West Virginia University
participant in having to verbally re- Administrative, technical, or material support: Worster, School of Medicine, Morgantown; Jacquelyn C. Camp-
Lent, Webb. bell, PhD, School of Nursing and Associate Dean for
spond to 2 instruments with similar Study supervision: MacMillan, Worster, Lent. Faculty Affairs, Johns Hopkins University, Baltimore,
questions. Financial Disclosures: None reported. Md. The consultants were compensated from the grant
Members of the McMaster University Violence funds for their time attending meetings and partici-
Last, although the CAS is a useful Against Women Research Group: Harriet L. Mac- pating in phone calls.
standard for comparison, it is not free Millan, MD, MSc, FRCP(C) (principal investigator), De- Funding/Support: This study was funded by the On-
of error: this error, in conjunction with partments of Psychiatry and Behavioural Neuro- tario Women’s Health Council, Ontario Ministry of
sciences and Pediatrics, McMaster University, Hamilton, Health and Long-term Care. Dr MacMillan holds a Ca-
the error associated with the screen- Ontario; Tom Abernathy, PhD, and Kathryn Ben- nadian Institutes of Health Research (CIHR) New
ing tests, will serve to attenuate the es- nett, PhD (coinvestigators), McMaster University, Emerging Team grant from the Institutes of Gender
Hamilton, Ontario; Charlene Beynon, RN, MScN (co- and Health; Aging; Human Development, Child and
timation of sensitivities and specifici- investigator), Middlesex-London Public Health Re- Youth Health; Neurosciences, Mental Health and Ad-
ties. The CAS was chosen because it was search Education & Development (PHRED) Program, diction; and Population and Public Health. Dr Wa-
and School of Nursing, The University of Western On- then holds a CIHR-Ontario Women’s Health Council
the most comprehensive measure of the tario, London; Michael Boyle, PhD (coinvestigator), De- Fellowship. Dr Boyle holds a Canada Research Chair
IPV experience. partments of Psychiatry and Behavioural Neuro- in the Social Determinants of Child Health.
©2006 American Medical Association. All rights reserved. (Reprinted) JAMA, August 2, 2006—Vol 296, No. 5 535
Role of the Sponsor: The Ontario Women’s Health 6. Nelson HD, Nygren P, McInerney Y, Klein J. Screen- community hospital emergency departments. JAMA.
Council had no role in the design or conduct of the ing women and elderly adults for family and intimate 1998;280:433-438.
study; in the collection, management, analysis, or in- partner violence: a review of the evidence for the U.S. 18. Feldhaus KM, Koziol-McLain J, Amsbury HL,
terpretation of the data; or in the preparation, re- Preventive Services Task Force. Ann Intern Med. 2004; Norton IM, Lowenstein SR, Abbott JT. Accuracy of 3
view, or approval of the manuscript. 140:387-396. brief screening questions for detecting partner vio-
Acknowledgment: We thank Harry S. Shannon, PhD, 7. Wathen CN, MacMillan HL. Interventions for vio- lence in the emergency department. JAMA. 1997;277:
Department of Clinical Epidemiology and Biostatis- lence against women: scientific review. JAMA. 2003; 1357-1361.
tics, McMaster University, Hamilton, Ontario, for his 289:589-600. 19. Brown JB, Lent B, Schmidt G, Sas G. Application
consultation regarding sample size determination. Dr 8. Taket A, Wathen CN, MacMillan HL. Should health of the Woman Abuse Screening Tool (WAST) and
Shannon received a small honorarium for his work. We professionals screen all women for domestic violence? WAST-Short in the family practice setting. J Fam Pract.
are very grateful to the participants, clinical staff, and PLoS Med. 2004;1:e4. 2000;49:896-903.
administrators of the participating sites: Cambridge Me- 9. Chuang CH, Liebschutz JM. Screening for inti- 20. Hegarty K, Sheehan M, Schonfeld C. A multidi-
morial Hospital, Carlisle Family Practice, Hamilton mate partner violence in the primary care setting: a mensional definition of partner abuse: development
Health Sciences, London Health Sciences Centre critical review. J Clin Outcomes Mgt. 2002;9:565-573. and preliminary validation of the Composite Abuse
(LHSC), Norfolk General Hospital, and Victoria Fam- 10. Anderst J, Hill TD, Siegel RM. A comparison of Scale. J Fam Violence. 1999;14:399-415.
ily Medical Centre of LHSC. domestic violence screening methods in a pediatric 21. Hegarty K, Bush R, Sheehan M. The Composite
office. Clin Pediatr (Phila). 2004;43:103-105. Abuse Scale: further development and assessment of
11. McFarlane J, Christoffel K, Bateman L, Miller V, reliability and validity of a multidimensional partner
REFERENCES Bullock L. Assessing for abuse: self-report versus nurse abuse measure in clinical settings. Violence Vict. 2005;
interview. Public Health Nurs. 1991;8:245-250. 20:529-547.
1. Tjaden P, Thoennes N. Full Report of the Preva- 12. Webster J, Holt V. Screening for partner vio- 22. Rasbash J, Steele F, Browne W, Prosser BA. Us-
lence, Incidence and Consequences of Violence Against lence: direct questioning or self-report? Obstet er’s Guide to MLwiN, Version 2.0. London, England:
Women: Research Report. Washington, DC: Na- Gynecol. 2004;103:299-303. University of London, Institute of Education; 2004.
tional Institute of Justice; 2000. NCJ 183781. 13. Glass N, Dearwater S, Campbell J. Intimate part- 23. Table 202-0405: Upper income limits and in-
2. Statistics Canada. Family Violence in Canada: A ner violence screening and intervention: data from come shares of total income quintiles, by economic
Statistical Profile 2002. Ottawa, Ontario: Canadian eleven Pennsylvania and California community hos- family type, 2004 constant dollars. Income Statistics
Centre for Justice Statistics; 2002. Catalogue No. 85- pital emergency departments. J Emerg Nurs. 2001;27: Division. Statistics Canada Web site. http://www
224-XIE. 141-149. .statcan.ca/english/freepub/75-202-XIE/2004000
3. MacMillan HL, Wathen CN; Canadian Task Force 14. Rhodes KV, Lauderdale DS, He T, Howes DS, /related.htm. Accessibility verified June 23, 2006.
on Preventive Health Care. Prevention and Treat- Levinson W. “Between me and the computer”: in- 24. Turner CF, Ku L, Rogers SM, Lindberg LD, Pleck
ment of Violence Against Women: Systematic Re- creased detection of intimate partner violence using JH, Sonenstein FL. Adolescent sexual behavior,
view and Recommendations. 2001. http://www.ctfphc a computer questionnaire. Ann Emerg Med. 2002;40: drug use, and violence: increased reporting with
.org/Full_Text/CTF_DV_TR_final.pdf. Accessed June 476-484. computer survey technology. Science. 1998;280:867-
28, 2006. 15. Rhodes KV, Drum M, Anliker E, Frankel RM, 873.
4. Datner EM, O’Malley M, Schears RM, Shofer FS, Howes DS, Levinson W. Lowering the threshold for 25. Hasley S. A comparison of computer-based and
Baren J, Hollander JE. Universal screening for inter- discussions of domestic violence: a randomized con- personal interviews for the gynecologic history update.
personal violence: inability to prove universal screen- trolled trial of computer screening. Arch Intern Med. Obstet Gynecol. 1995;85:494-498.
ing improves provision of services. Eur J Emerg Med. 2006;166:1107-1114. 26. Bair-Merritt MH, Feudtner C, Mollen CJ, Win-
2004;11:35-38. 16. Clark JP, Du Mont J. Intimate partner violence and ters S, Blackstone M, Fein JA. Screening for intimate
5. Ramsay J, Richardson J, Carter YH, Davidson LL, health: a critique of Canadian prevalence studies. Can partner violence using an audiotape questionnaire: a
Feder G. Should health professionals screen women J Public Health. 2003;94:52-58. randomized clinical trial in a pediatric emergency
for domestic violence? systematic review. BMJ. 2002; 17. Dearwater SR, Coben JH, Campbell JC, et al. Preva- department. Arch Pediatr Adolesc Med. 2006;160:311-
325:314. lence of intimate partner abuse in women treated at 316.
536 JAMA, August 2, 2006—Vol 296, No. 5 (Reprinted) ©2006 American Medical Association. All rights reserved.