Does Stage of Change Predict Improved Intimate Partner Violence Outcomes Following an Emergency Department Intervention?


  • Funding support was provided from the Avon Foundation.

  • The authors have no relevant financial information or potential conflicts of interest to disclose.

Address for correspondence and reprints: Justin D. Schrager, MD, MPH; e-mail:



The objective was to assess the effect of an emergency department (ED)-based computer screening and referral intervention on the safety-seeking behaviors of female intimate partner violence (IPV) victims at differing stages of change. The study also aimed to determine which personal and behavioral characteristics were associated with a positive change in safety-seeking behavior. The hypothesis was that women who were in contemplation or action stages of change would be more likely to endorse safety behaviors during follow-up.


This was a prospective cohort study of female IPV victims at three urban EDs, using a computer kiosk to deliver targeted education about IPV to provide referrals to local resources. All noncritically ill adult English-speaking women triaged to the ED waiting room during study hours were eligible to participate. Women were screened for IPV using the validated Universal Violence Prevention Screening Protocol (UVPSP), and all IPV-positive women further responded to validated questionnaires for alcohol and drug abuse, depression, and IPV severity. The women were assigned a baseline stage of change using the University of Rhode Island Change Assessment (URICA) scale for readiness to change their IPV behaviors. Study participants were contacted at 1 week and 3 months to assess a variety of predetermined safety behaviors to prevent further IPV during that period. Descriptive analyses were performed to determine if stage of change at enrollment and a variety of specific sociodemographic characteristics were associated with taking protective action during follow-up.


A total of 1,474 women were screened for IPV; 154 (10.4%) disclosed IPV and completed the full survey. Approximately half (47.4%) of the IPV victims were in the precontemplation stage of change, and 50.0% were in the contemplation stage. A total of 110 women returned at 1 week of follow-up (71.4%), and 63 (40.9%) women returned for the 3-month follow-up. Fifty-five percent of those who returned at 1 week and 73% of those who returned at 3 months took protective action against further IPV. Stage of change at enrollment was not significantly associated with taking protective action during follow-up. There was no association between demographic characteristics and taking protective action at 1 week or 3 months.


Emergency department–based kiosk screening and health information delivery is a feasible method of health information dissemination for women experiencing IPV and was associated with a high proportion of study participants taking protective action. Stage of change was not associated with actual IPV protective measures.


¿Puede la Etapa del Cambio Predecir la Mejoría de los Resultados de Violencia de Pareja tras una Intervención en el Servicio de Urgencias?


Los objetivos fueron evaluar el impacto de un despistaje computarizado en un servicio de urgencias (SU) y la intervención de derivación en los comportamientos que intentan conseguir la seguridad de las mujeres víctima de violencia de pareja (VP) en las diferentes etapas del cambio. También se determinó qué características personales y de comportamiento se asociaron con un cambio positivo en la actitud de búsqueda de seguridad. La hipótesis fue que las mujeres que estaban en la etapa de contemplación o activa tenían más probabilidad de buscar comportamientos seguros durante el seguimiento.


Se trata de un estudio de cohorte prospectivo de mujeres víctimas de VP de tres SU urbanos, que utilizó una cabina computarizada para entregar formación dirigida sobre VP y proporcionar derivaciones a recursos locales. Se eligió para participar a toda mujer de habla inglesa clasificada como no críticamente enferma en la sala de espera del SU durante las horas del estudio. Se realizó el despistaje de las mujeres víctima de VP mediante el protocolo validado de despistaje de prevención de violencia universal Universal Violence Prevention Screening Protocol (UVPSP), y todas las mujeres positivas para VP posteriormente respondieron a unos cuestionarios validados para el abuso de drogas y alcohol, la depresión y la gravedad de la VP. Las mujeres se asignaron a una etapa de cambio basal mediante la escala University of Rhode Island Change Assessment (URICA) para prepararlas para cambiar sus comportamientos respecto a la VP. Se contactó con las participantes del estudio a la semana y a los tres meses para valorar una variedad de comportamientos de seguridad predeterminados para prevenir una futura VP durante ese periodo. Se realizó un análisis descriptivo para determinar si la etapa de cambio en la inclusión y una variedad de características sociodemográficas específicas se asociaban con el tomar acciones de protección durante el seguimiento.


Se realizó el despistaje para VP a 1.474 mujeres, 154 (10,4%) revelaron VP y completaron la encuesta totalmente. Aproximadamente la mitad (47,4%) de las víctimas de VP estaban en la etapa de cambio de precontemplación y el 50,0% en la etapa de contemplación. Ciento diez mujeres volvieron a la semana de seguimiento (71,4%), y 63 (40,9%) a los tres meses de seguimiento. El 55% de aquéllas que volvieron a la semana, y el 73% de aquéllas que volvieron a los tres meses tomaron acciones protectoras contra una futura VP. La etapa de cambio a la inclusión no se asoció de forma estadísticamente significativa con el tomar acciones de protección durante el seguimiento. No hubo asociación entre las características demográficas y el tomar acciones protectoras a la semana o a los tres meses.


El despistaje mediante una cabina en el SU y el proporcionar información de salud es un método viable para distribuir información sanitaria a las mujeres que sufren VP, y se asoció con una alta proporción de participantes del estudio que tomaron acciones de protección. La etapa del cambio no se asoció con las medidas de protección frente a la VP.

With an estimated 28.9% of women experiencing intimate partner violence (IPV) during their lifetime,[1] at an annual medical and mental health care cost of $4.1 billion,[2] IPV represents a significant problem with personal and societal consequences. The hospital emergency department (ED) is an ideal location to study methods of screening and intervening in IPV. Abbott et al.[3] showed that although 12% of women visiting the ED were in abusive relationships, only 13% of those women told ED practitioners, or were asked, about IPV. The prevalence of ongoing IPV appears to be much higher (36%) among urban, low-income, black or African American females, as does concomitant mental illness and addiction.[4] The best strategy to guide screening and subsequent counseling for IPV has not been determined.

The objective of our study was to assess the effect of an ED-based computer screening and referral intervention for IPV victims and to determine what characteristics—stage of change, behavioral (e.g., depression and addiction), socioeconomic, and demographic—resulted in positive changes in their safety seeking behavior. We hypothesized that women who were experiencing IPV and were in contemplation or action stages would be more likely to take protective action following the intervention. Additionally, we hypothesized that women with concomitant addictive, mental health, and socioeconomic difficulties would be less likely to endorse taking protective action.

The Transtheoretical Model (TTM), or Stages of Change model, has increasingly been applied to IPV screening and prevention strategies,[5-8] as well as the treatment of IPV perpetrators.[9] The TTM, originally applied to smoking cessation, weight loss, and alcoholism, theorizes that most people move in a predictable fashion through stages of change (precontemplation, contemplation, action, and maintenance) as they progress toward ending a deleterious behavior.[10-12] The TTM represents a promising theory to measure and provide stage-matched interventions for women at different levels of readiness to change with regard to their experience with IPV. Socioeconomic and demographic differences likely influence a woman's ability to benefit from screening and interventions based on the TTM, but this has not thoroughly been explored.


Study Design

We conducted a prospective cohort study for a 1-year period. This study was approved by the university institutional review board as well as the hospital research oversight committee.

Study Setting and Population

This study was conducted at three hospitals: Hospital 1 is a large public teaching hospital affiliated with two medical schools with a volume of >100,000 ED visits. It is the city's only public hospital and the state's largest level one trauma center. The hospital serves a predominantly black or African American population (92%). Hospital 2 is a community-based hospital with approximately 50,000 ED visits. Hospital 3 is a university-based tertiary/quaternary care facility specializing in the care of the acutely ill adult, with nearly 30,000 ED patient visits and a 30% admission rate. Physician staffing in these EDs is provided by 59 emergency medicine residents and supervised by 80 faculty physicians.

All female adult patients over 18 years of age, regardless of chief complaint, who presented to one of the three EDs and screened positive for IPV were eligible for participation. Participants completed a 20 minute, computer-based questionnaire and returned or were contacted by phone at one week and three months to complete follow-up assessments.

Study Protocol

Research assistants obtained a list of all patients in the ED waiting area from the nursing or triage areas. They then approached all female patients between the hours of 11:00 a.m. and 7:00 p.m., Monday–Friday, between June 1, 2008, and December 1, 2009, and asked if they were “Interested in taking a computerized survey about mental and physical health, including your use of tobacco, alcohol, and drugs, as well as any experiences you may have had with domestic violence.” Those who agreed were led to a touchscreen computer kiosk to participate in the survey. Participants responded to questions from validated risk assessment scales and other general health questions on their own. The survey had several skip patterns and took on average 15 minutes to complete. The survey assessed various behavioral and social risks to personal health, including but not focusing on IPV. All patients who screened positive for IPV received printouts with general IPV information, as well as targeted referrals to local and national social service agencies such as the state and national domestic violence hotline, the Legal Aid Society, the hospital's domestic violence support group, a women's shelter hotline, and the local women's resource center. Patients also received educational information about health risk behaviors for which they screened positive, including drug and alcohol abuse, smoking, seatbelt use, HIV risk behavior, depression, and exercise.

Female patients who screened positive for IPV at the kiosk were then asked to participate in a research study that entailed the completion of additional paper surveys. Patients who agreed were consented for the study at this time. Following completion of the surveys, the participants were given two follow-up appointments (at 1 week and 3 months) for which they received monetary compensation of $20 at 1 week and $40 at 3 months. These follow-up assessments included the same measures as the initial ED visit plus information on specific IPV safety measures and resource utilization. Follow-up visits took place at the original enrollment hospital and consisted of retaking slightly modified versions of the paper surveys at 1 week and slightly modified 3-month versions of the kiosk survey and the paper surveys administered by research assistants. All women who participated in the study were given the same intervention: a printout with IPV information and referral phone numbers and addresses and two follow-up appointments at 1 week and 3 months. Women who screened positive for IPV but who declined enrollment were given the same general IPV information and referrals.


The initial assessment of IPV victimization was determined with the Universal Violence Prevention Screening Protocol (UVPSP), a six-item screening tool developed by the George Washington University ED to identify IPV victimization.[13] The scale measures physical, sexual, and verbal violence in a yes/no format. A positive answer to any of the questions indicates victimization. The UVPSP has been effectively studied and validated by Heron et al.[14] in the same low-income, largely African American, ED population. Participants who screened positive on the UVPSP were then asked to complete the danger assessment (DA) instrument at the kiosk. The 20-item DA uses a weighted scoring system to count yes/no responses of risk factors associated with intimate partner homicide.[15] The DA stratifies response scores into categories of personal danger. A score of 0 to 7 is described as variable danger, a score of 8 to 13 is described as increased danger, a score of 14 to 17 is described as severe danger, and a score of 18 and above is described as extreme danger. The language was designed to have meaningful significance and tested among female survivors of abuse. Variable danger, specifically, is described as a nonnegligible risk of lethal violence that can change quickly depending on a person's situation.

Stage of change at enrollment was the primary exposure variable in this study. Women who disclosed IPV and were enrolled in the study were asked to complete the University of Rhode Island Change Assessment (URICA), a flexible 32-item self-report instrument that includes four subscales measuring the stages of change: precontemplation (PC), contemplation (C), action (A), and maintenance (M).[10-12] The subscales were then combined arithmetically (C + A + M – PC) to yield a second-order continuous readiness to change score used to assess readiness to change at entrance to treatment. Whereas the URICA is primarily designed to assess addictive and deleterious behaviors, such as smoking, we employed a modified URICA to help determine how each participant felt about making changes with regard to her violent intimate relationship. Changes to the standard URICA were limited to replacing instances of a deleterious behavior (such as “smoking”) with “intimate partner violence.” The URICA is a flexible questionnaire that has been successfully adapted to a variety of behaviors, including IPV.[5-8]

For alcohol abuse, the Tolerance, Worried, Eye openers, Amnesia, K(C)ut Down (TWEAK) scale,[16] consisting of five questions, was used to assess alcohol abuse behavior at the computerized kiosk. The questions are answered yes/no and it is scored on a seven-point scale. A score of ≥2 was used to identify alcohol abusers in our study. This cut point has been suggested by several studies as being the most appropriate cut point to use for females[17] and for patients in the ED.[18] The TWEAK scale was shown to perform well among African American women, with high sensitivity, using these cut points in comparison to other measurement tools, such as the CAGE.[17]

The Drug Abuse Screening Test (DAST-10)[19, 20] was used to assess whether participants were abusing drugs and was administered at the computerized kiosk. The DAST was designed to provide a brief instrument for clinical and nonclinical screening to detect drug abuse or dependence disorders. Patients respond either yes or no to each of 10 questions about drug use in the past 12 months. A “yes” response equals a positive response to that question. A score of ≥3 was considered a positive screen for drug abuse. The DAST has been shown to have high internal validity and reliability and has been successfully employed in screening minority female patients in an inner-city ED setting.[21]

Study participants completed a brief, standardized interview to collect demographic information such as age, race, sex, educational attainment, and relationship status. Information was collected concerning the female participant and her partner, as well as her desired relationship outcomes and expectations. We collected general health information from each participant that included the Beck Depression Inventory (BDI; a 21-item questionnaire designed to identify patients with moderate to severe depression[22]) during the kiosk screening survey.

At 1-week and 3-month follow-up, study participants were asked to complete the same instruments as they did during the initial interview, as well as additional interview questions regarding IPV resource use and protective action. Protective action, the primary outcome of the study, was defined by a number of predetermined patient actions, including ending the relationship, moving out, changing or unlisting her telephone number, making a safety plan with a neighbor or friend, carrying a knife or firearm, obtaining a restraining order, seeking counseling, and contacting legal organizations and shelters. In addition to these protective actions, patients were also allowed to provide the research assistant with their own protective actions.

Data Analysis

Baseline demographic characteristics were analyzed individually and stratified by stage of change. Stage of change was ascertained at the index interview, conducted immediately after screening and enrollment into the study. The readiness-to-change score was used to assign each patient a specific stage of change using the cutoffs of 8 and 12 for precontemplation and contemplation, respectively. These cutoffs are the accepted numbers chosen when conducting URICA-based research on the general population.

After assigning each study participant to a specific stage of change, we performed descriptive statistical analysis with chi-square and exact tests of significance for all categorical variables to examine any associations between baseline stage of change and the measured covariates. Exact tests of significance were employed when the expected cell counts were fewer than five in the stratified tables. We analyzed each participant's score on the various validated scales for severity of abuse (DA), depression (BDI), drug abuse (DAST), and alcohol abuse (TWEAK). We also examined various demographic characteristics including study site, whether or not the participant was in the ED as a result of domestic violence, race (African American, white, or other), age as both a continuous and a categorical variable, marital status, education, employment, income, and whether or not the participant has children. Age categories were chosen based on predicted meaningful quantiles derived from prior research in the same source population.[14] All statistical analysis was performed with SAS v9.2 software (SAS Institute Inc., Cary, NC). We then assessed the same demographic characteristics stratified by the outcome of interest in the study, namely, taking protective action. Again, we employed chi-square and exact tests of significance to compare the strata of participants who did or did not take protective action during follow-up.

A nonrespondent analysis of women who were enrolled but lost to follow-up was conducted to determine if they differed significantly by the various baseline characteristics. We employed chi-square and exact tests of significance to compare patients based on return or nonreturn.


Study Population Screening and Participation

A total of 3,381 women were approached to participate in the computer kiosk screening in three EDs. A total of 1,474 (43.6%) agreed to be screened at the kiosks. The screening tool, the UVPSP, identified IPV victimization in 260 women (17.8%) who used the kiosks. Of those 260 women, 154 (59.2%) opted to enroll in the study.

At enrollment, baseline URICA questionnaire responses revealed that 73 women (47.4%) were in the precontemplation stage of change concerning taking action to prevent further IPV. An additional 77 women (50.0%) were in the contemplation stage. Four women (2.6%) were considered to be in the action stage of change. Over half (54.5%) of the women enrolled in the study were characterized as being in “variable danger” according to their DA questionnaire, compared to 37.5% who were at “increased danger” and 8% who were in “severe danger.” The DA results did not differ significantly by baseline stage of change.

The majority of women enrolled in the study were enrolled at Hospital 1 (76.6%), compared to 21.4% at Hospital 2 and 2% at Hospital 3. Their baseline stage of change by enrollment location was not statistically different. The majority of the women enrolled (92.7%) were not in the ED as a result of IPV, and being in the ED because of IPV did not differ significantly by the stage of change the woman was in at the time of enrollment.

Approximately half of the women enrolled in the study had not completed high school, and the education level of the study participants varied significantly (p = 0.0357) by baseline stage of change. Eighty-five percent of the women enrolled in the study self-assigned black as their race. The mean (±SD) age at enrollment was 34.6 (±12.0) years. Most of the women enrolled in the study were unmarried (92.2%), 31.2% held jobs outside of the home, 79.9% had a household income less than $20,000 per year, and 63.6% had children. Race, age, marital status, employment, household income, and parenthood were not significantly associated with stage of change at baseline (Table 1).

Table 1. Baseline Clinical and Demographic Patient Characteristics by Stage of Change for Study Population
Study VariableCategoriesStudy Population, N = 154Precontemplation, n = 73 (47.4%)Contemplation, n = 77 (50.0%)Action, n = 4 (2.6%)Chi- square Exact p valuea
  1. a

    Exact tests of significance were used when expected cell counts were lower than five.

  2. b

    Three missing reason for visit to ED is IPV related.

  3. c

    Four people missing BDI.

  4. d

    Four people did not complete TWEAK.

  5. e

    Forty-two people missing the danger assessment form.

  6. BDI = Beck Depression Inventory; DA = danger assessment; DAST = Drug Abuse Screening Test; HS = high school; IPV = intimate partner violence; TWEAK Tolerance, Worried, Eye openers, Amnesia, K(C)ut Down.

Study/interview locationHospital 1, n (%)118 (76.6)59 (80.8)55 (71.4)4 (100)0.5579
Hospital 2, n (%) 33 (21.4)13 (17.8)20 (26.0)0 (0)
Hospital 3, n (%)3 (2.0)1 (1.4)2 (2.6)0 (0)
In ED as a result of a domestic violence incidentbYes, n (%)11 (7.3)5 (7.0)6 (7.9)00.9999
No, n (%)140 (92.7)66 (93.0)70 (92.1)4 (100)
RaceBlack or African American131 (85.1)64 (87.7)64 (83.1)3 (75)0.6510
White 18 (11.7)7 (9.6)10 (13.0)1 (25)
Other5 (3.2)2 (2.7)3 (3.9)0 (0)
Age (yr)Mean (±SD)34.6 (±12.0)32.8 (±11.6)36.2 (±12.4)36.5 (±7.5) 
Age category (Quartiles)Age 18–24 yr, n (%) Mean 20.8 (SD±2.0)41 (26.6)24 (32.9)17 (22.1)0 (0)0.3494
Age 25–32 yr, n (%) Mean 28.7(SD±2.3) 38 (24.7)16 (21.9)20 (26.0)2 (50)
Age 33–46 yr, n (%) Mean 40.0 (SD ± 4.0)44 (28.6)21 (28.8)21 (27.3)2 (50)
Age 47+ yr, n (%) Mean 52.3 (SD ± 5.4)31 (20.2)12 (16.4)19 (24.7)0 (0)
Marital status categoryUnmarried, n (%)142 (92.2)63 (86.3)75 (97.4)4 (100)0.0559
Married, n (%)12 (7.8)10 (13.7)2 (2.6)0 (0)
Education levelSome or no HS, n (%)80 (52.0)38 (52.1)40 (51.9)2 (50)0.0357
HS graduate, n (%)49 (31.8)17 (23.3)30 (39.0)2 (50)
Any college, college graduate, or graduate school, n (%)25 (16.2)18 (24.6)7 (9.1)0 (0)
Job outside the homeYes, n (%)48 (31.2)23 (31.5)24 (31.2)1 (25)0.9999
No, n (%)106 (68.8)50 (68.5)53 (68.8)3 (75)
Household income category<$20,000, n (%)123 (79.9)56 (76.7)63 (81.8)4 (100)0.5733
>$20,000, n (%)31 (20.1)17 (23.3)14 (18.2)0 (0)
BDI positive, n (%)cYes75 (50)37 (50.7)36 (49.3)2 (50)0.9999
No75 (50)36 (49.3)37 (50.7)2 (50)
DAST positive, n (%)Yes40 (26.0)13 (17.8)26 (33.8)1 (25)0.0592
No114 (74.0)60 (88.2)51 (66.2)3 (75)
TWEAK + (alcohol), n (%)dYes97 (64.7)45 (61.6)49 (67.1)3 (75)0.7615
No53 (35.3)28 (38.4)24 (32.9)1 (25)
Are you currently with your partner? n (%) Yes85 (55.6)41 (56.2)42 (55.3)2 (50)0.9999
No68 (44.4)32 (43.8)34 (44.7)2 (50)
Do you have children, n (%)Yes98 (63.6)46 (63.0)48 (62.3)4 (100)0.4404
No56 (36.4)27 (37.0)29 (37.7)0 (0)
Severity of abuse (DA), n (%)eVariable danger61 (54.5)31 (60.3)28 (47.5)2 (50)0.4549
Increased danger42 (37.5)14 (28.6)26 (44.1)2 (50)
Severe danger9 (8.0)4 (8.2)5 (8.5)0 (0)

Half of the women enrolled in the study had a positive BDI screen, although this did not vary significantly by their stage of change at baseline. Twenty-six percent of the participants screened positive for drug abuse on their baseline DAST questionnaire, and 65.7% screened positive on the TWEAK alcohol dependence questionnaire. Neither the DAST nor the TWEAK was significantly associated with the baseline stage of change (Table 1).

Taking Protective Action

Among the 154 women who screened positive for IPV and were enrolled in the study, 110 returned at 1 week (71.4%), and 63 (40.9%) returned at three months. With respect to the outcome of interest, taking protective action against IPV, 55.5% of those who returned at 1 week and 73% of those who returned at 3 months had taken protective action (Table 2). The type of protective action sought by each woman varied. Among the 61 women who took protective action (or a combination of protective actions) by the 1-week follow-up, 25% reported ending the relationship, 10% moved out, 10% changed or unlisted their phone number, 27.5% made a safety plan, 8% started carrying a weapon, 3% obtained a restraining order, and 15% sought counseling or support services for IPV. Seven percent of the women enrolled and who returned at 1 week contacted one of the support services described at enrollment. Of the women who took protective action during the first week of follow-up, 92% had taken more than one protective action. Among the 46 women who had taken protective action and returned for follow-up at 3 months, 70% had ended the relationship, 33% had moved out, 41% had changed or unlisted their phone number, 50% had made a safety plan, 26% started carrying a weapon, 13% obtained a restraining order, and 33% sought counseling or support services. At 3 months, 75% of the women who had taken protective action reported participating in more than one protective action during that period.

Table 2. Sociodemographic Characteristics at One Week and Three Months of Follow-up by Whether or Not the Person Has Taken Protective Action to Prevent Further IPV
CharacteristicWeek 1 Follow-up3-Month Follow-up
Protective Action, n = 61 (55.5%)No Protective Action, n = 49 (44.6%)Protective Action, = 46 (73%)No Protective Action, = 17 (27%)
  1. a

    Three patients missing BDI.

  2. b

    Three patients missing TWEAK questionnaire at 1 week and one patient missing TWEAK at 3 months.

  3. c

    Thirty patients missing severity of abuse at Week 1 and 19 missing at 3 months.

  4. BDI = Beck Depression Inventory; DAST = Drug Abuse Screening Test; HS = high school; IPV = intimate partner violence; TWEAK = Tolerance, Worried, Eye openers, Amnesia, K(C)ut Down.

Stage of change at initial visit to ED
Precontemplation24 (39.3)26 (53.1)26 (56.5)7 (41.2)
Contemplation37 (60.7)21 (42.9)18 (39.1)9 (52.9)
Action0 (0)2 (4.0)2 (4.4)1 (5.9)
African American or black53 (86.9)41 (83.7)40 (87.0)16 (94.1)
White5 (8.2)7 (14.3)4 (8.7)1 (5.9)
Other3 (4.9)1 (2.0)2 (4.35)0 (0)
Age category, yr
<2413 (21.3)11 (22.5)9 (19.6)4 (23.5)
25–32 14 (23.0)10 (20.4)9 (19.6)1 (5.9)
32–46 19 (31.2)16 (32.7)17 (37.0)7 (41.2)
>47 15 (24.6)12 (24.5)11 (23.9)5 (29.4)
Less than HS education32 (52.5)25 (51.0)22 (47.8)12 (70.6)
HS graduate18 (29.5)17 (34.7)15 (32.6)3 (17.7)
Some college, completed college, or has graduate degree11 (18.0)7 (14.3)9 (19.6)2 (11.8)
Has job outside of the home 18 (29.5)11 (22.5)14 (30.4)4 (23.5)
BDI positivea31 (53.5)25 (51.0)26 (57.8)6 (35.3)
DAST positive21 (34.4)10 (20.4)11 (23.9)5 (29.4)
TWEAK positiveb36 (62.1)32 (65.3)27 (60.0)11 (64.7)
Has children45 (73.8)34 (69.4)34 (73.9)11 (64.7)
Children live with respondent 18 (29.5)13 (26.5)16 (34.8)3 (17.7)
Unmarried55 (90.2)46 (93.9)44 (95.7)15 (88.2)
Household income <$20,00052 (85.3)38 (77.6)38 (82.6)14 (82.4)
In ED for domestic violence at enrollment8 (13.6)1 (2.1)4 (8.9)0 (0)
Severity of abusec
Variable24 (52.2)20 (58.2)18 (56.2)6 (50)
Increased17 (36.9)11 (32.4)12 (37.5)3 (25)
Severe5 (10.9)3 (8.8)2 (6.3)3 (25)

Of the 61 women who had taken protective action at 1 week, 47.5% were in the precontemplation stage at enrollment, 45.9% were in the contemplation stage at enrollment, and 6.6% were in the action stage. Seventy-three percent of the women who returned at 3 months and had taken protective action to prevent further IPV, and 56.5% of these women were initially in the precontemplation stage, 39.1% were in the contemplation stage, and 4.4% were in the action stage. Women who took protective action at 1 week and 3 months did not vary significantly (p < 0.05) by race, education level, employment status, age, marital status, or a positive screen on the BDI, DAST, TWEAK, or DA (Table 2).

Analysis of Loss to Follow-up

Forty-four women (28.6%) were lost to follow-up 1 week after enrollment, and 91 of 154 (59%) did not return at the 3-month follow-up. Women who did not return for follow-up did not differ by baseline stage of change. Fifty-eight percent were in the precontemplation stage at enrollment, 39.5% were in the contemplation stage, and 2.8% were in the action stage. This was also true for the women who did not return at 3 months, where 53.8% were in the precontemplation stage, 38.5% were in the contemplation stage, and 7.7% were in the action stage at baseline. The women who were lost to follow-up did not differ significantly at 1 week or 3 months by baseline race, educational attainment, BDI screen, DAST screen, TWEAK screen, having children in the home, marital status, or having an income of less than $20,000 per year.

The women who were lost to follow-up during the study period tended to be younger, with 40.9% of those lost to follow-up being in the youngest age group (<24 years old) compared to 20.9% of those in the youngest age group who returned at 1 week (p = 0.0150). This was also noted at 3 months, with 30.8% in the youngest age group (<24 years) lost to follow-up compared to 20.6% of those who returned who were in the youngest age group (p = 0.0220). Women who were lost to follow-up at 1 week were significantly more likely to be employed outside of the home (43.2 employed vs. 26.4 not employed; p = 0.0425). This was not true for women at 3 months of follow-up. Women with children were significantly more likely to return for follow-up at 1 week (71.8% with children and 43.2% without children; p = 0.0008), but not significantly at 3 months (71.4% with children vs. 58.2% without children; p = 0.094).


This study aimed to determine the effect, through follow-up, of an ED-based computer screening and referral intervention on IPV victims at various stages of change with regard to taking protective action in their abusive relationship. We hypothesized that women who were experiencing IPV and were in the contemplation or action stage of change would be more likely to endorse safety behaviors following this intervention. We found, however, that the ED-based kiosk tool was associated with safety behaviors in women in all stages of change and among women of varying sociodemographic backgrounds. At follow-up, more than half of the participants had taken protective action 1 week after interacting with the kiosk, and nearly three-quarters had taken protective action by 3 months. The broad implications for this kind of intervention suggest that kiosk-based screening and referral tools appear to be an effective means of reaching a varied population of at-risk women in the ED, and that these tools result in women taking protective action over a relatively short time period, regardless of their initial stage of change.

Computer-based health screening assessments have been shown to be an effective means of identifying victims of IPV in the ED.[23] These tools are successful in bringing about safety-related change in the lives of victims of IPV.[4] This study agrees with prior research in that regard, and was able to demonstrate a high proportion of women taking protective action following the computer-based screening and intervention.

The TTM has increasingly been applied to IPV research.[5, 6, 8, 24] However, the use of this model necessitates defining specific behavioral endpoints or goals, for which entering the “action” stage of change results in a concrete safety benefit to the abused woman. Prior research has shown that women often participate in multiple safety actions before leaving their abusive relationship.[7] This has made it difficult to define what exactly constitutes a safety action. Furthermore, it has been shown that women in abusive relationships often progress or regress in their readiness to change depending on external factors, such as increasing levels of partner violence.[25] In our study, 92% of the women who reported making a protective action during the first week of follow-up performed more than one protective action—the most common safety behaviors being the construction of a safety plan and self-reported ending of the relationship. The variety and quantity of protective actions taken by the women in our study supports findings by Chang et al.[26] that women in violent relationships desire direct access to multiple safety resources that preserve their autonomy and anonymity. The rapidity with which many of the participants in our study took protective action supports the theory that IPV victims often progress in a nonlinear or rapid fashion through the stages of change depending on the severity of abuse they are experiencing.

To what extent the participants' presence in an ED can act as a marker of that severity is unknown, as the majority of the participants in our study claimed not to have come to the ED as a direct result of IPV. Nonetheless, the rapidity with which the majority of the women in our study took protective action(s) points to the importance of having IPV resources readily available. Our study did not show the measurement of stage of change in the clinical setting to be predictive of whether a woman took protective actions over the next week or 3 months.

Studies have demonstrated difficulties in screening for IPV in various health care settings, including the ED.[27-31] Kiosk screening and intervention in the ED has been shown to be a safe and effective way to disseminate IPV information.[23, 32] This is especially important among vulnerable minority and low-income populations, like the women in our study.[14] Another advantage of the kiosk screening and information dissemination model is that it allows for nondisclosure to a person on the part of the woman, while raising awareness and presenting a variety of safety resources. Despite the fact that women may progress in a nonlinear or rapid fashion through the stages of change with regard to IPV, it is important that they have the resources on hand when the time to take a protective action presents itself.

Another important finding in our study is that we were unable to identify specific sociodemographic groups that benefited the most from the kiosk screening and intervention. Screening positive for depression, for example, was not significantly associated with a specific stage of change, nor was it associated with whether or not the study participant took a protective action—despite the knowledge that various studies have shown a link between IPV and depressive symptoms.[33, 34] It may be that because the majority of the women in our study had several major sociologic correlates with IPV, such as depression, substance abuse, or low household income, we were unable to effectively determine which specific sociodemographic characteristics served as barriers to change. And while we attempted to use these measures to gather a more complete contextual understanding of the lives of the women in our study, our intervention did not address these problems specifically. Therefore, we were unable to assess the effectiveness of an ecologic approach to IPV as it has been discussed in recent literature.[35, 36] Nonetheless, our intervention was followed by a high level of safety actions.

Whether or not the TTM is directly applicable to the IPV population remains unknown. More work is needed to address the vagaries of what the ideal endpoint is for each individual woman's experience with IPV. Additionally, more research into the timing, contextual, and ecologic framework of IPV interventions may aide in determining which interventions should be employed in specific populations. Nonetheless, broad-based screening and a variety of patient safety information should be part of future IPV interventions, as it appears to benefit women in all stages of change and from various sociodemographic backgrounds.


Our findings must be considered within the context of several study limitations. First, this study is not a clinical trial testing the efficacy of an intervention, but rather an epidemiologic cohort study comparing the stage of change at enrollment with the outcome of interest, taking protective action. As such, the study suffers from several of the limitations common in observational research. The participants in this study represent a random sample of women in the ED triage areas who were willing to take a women's health survey at a kiosk. Therefore, this study likely does not capture the entire scope of the population affected by IPV in this setting. All women in the ED waiting areas were approached in a sequential manner and screened. However, due to the time constraints of the enrollment process and available number of kiosks, not all eligible women could be screened. Second, as mentioned in the introduction, two studies have found wide differences in the prevalence of IPV among women seeking care in the ED: between 12 and 36%. In our study, 17.8% of women screened in the ED were positive for IPV, but only 58.4% of those women agreed to participate in the study. This likely represents selection bias in that our sample was probably an underestimate of the prevalence of IPV in our source population. Nonetheless, our tool likely captures most of the population who would be amenable to this type of intervention. Third, our study population represents a predominantly urban, low-income, minority black or African American population, which may not be generalizable to women of other socioeconomic backgrounds, including non-English speakers. Fourth, within the population of people who experience IPV, the URICA has been shown to yield answers that are artificially skewed toward the action and maintenance ends of the scale, wherein respondents consciously or unconsciously amplify their readiness to change.[37, 38] This effect is not well described in the literature for this study population, however, and may be a small source of misclassification bias. Finally, 28.6% of the study participants were lost to follow-up at 1 week and 59% at three months. We have attempted to minimize this potential source of bias by conducting an analysis of this group.


These results elucidate the importance of developing and implementing alternative screening strategies for women victimized by interpersonal violence. Within our largely minority, inner-city, study population of ED patients, stage of change was not associated with whether or not a woman took protective action to prevent further intimate partner violence, nor were stage of change and protective action associated with specific socioeconomic, health, or demographic characteristics within the population. The overall success of participants in this intervention, regardless of the stage of change, may point to the effectiveness of screening with information dissemination and follow-up for women at any stage of change; however, the utility of the stages of change model for this population is not supported by the results of this study.