• 
• 

Three Part Question

For [women suffering domestic abuse attending an emergency department], how effective are [emergency department-based interventions] in [reducing domestic abuse-related morbidity]?

Clinical Scenario

A young woman presents to the emergency department (ED) with a number of vague, non-specific complaints. The team suspect she may be suffering domestic abuse. Upon enquiry, the patient discloses a recent history of abuse by her current partner. What ED-based interventions will be most effective in improving the management of such a patient to reduce her risk of further abuse?

Search Strategy

Titles and Abstracts of online databases Medline, EMBASE, CINAHL and PsycINFO were searched from their respective start dates to May 2015 using the following key terms and Medical Subject Headings:
["domestic violence" OR "intimate partner violence" OR "partner abuse" OR "spouse abuse" OR "abusive partners" OR "battered women" OR abused] AND ["accident and emergency" OR "emergency department" OR exp EMERGENCY SERVICE, HOSPITAL/] AND intervention*

Search Outcome

The primary search identified 271 unique articles. Based on screening titles and abstracts for relevance, 22 full-text articles were deemed as potentially suitable and assessed in detail. Of these, 16 articles did not meet the selection criteria and 3 were of insufficient quality in terms of the outcomes measured. Manually searching reference and citation lists of key articles revealed an additional 3 articles, providing a total of 6 studies for review.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Fanslow et al. 1999 New Zealand	Two EDs, one of which received training in a protocol for the identification and management of abused women. In the control ED, no training was received and no new protocol implemented.	2 group quasi-experimental before/after study	1) Identification of women experiencing partner abuse including 'confirmed ' and 'suspected' cases 2) Acute management of those identified 3) Appropriate interventions offered by staff Outcomes measured at baseline, 1 month post-intervention and at 1 year follow up	1) No significant changes in overall proportion of women identified as victims. Significant increase in 'confirmed' identification of cases at the intervention ED compared with control at 1 month post-intervention but no difference between classification of victims as 'confirmed' or 'suspected' cases at 1 year post-intervention ( χ²=12.2, p=0.007). 2) Significant increase in interventions offered and documentation at the intervention ED at 1 month but not time at 1 year post-intervention.	 ED personnel were not blinded to their group allocation  Use of record review to assess staff performance
Koziol-McLain et al. 2010 New Zealand	English-speaking women aged 16 or older randomized to a brief intervention (n=199) or usual care (n=200). The brief intervention involved screening for intimate partner violence, receiving accompanying supportive messages about the unacceptability of violence, risk assessment and referral.	Randomized control trial	1) Primary outcome: re-victimization by any current or past intimate partner in the interval between the ED index visit and follow-up, measured by the Composite Abuse Scale (CAS) 2) Secondary outcome: safety behaviours and resource utilization Outcomes measured 3 months post-intervention.	1) 12.0% (20 of 167) of women in the treatment group and 13.6% (24 of 177) of women in the usual care group reported partner violence. The absolute risk difference was –1.6% (95% CI –8.7% to 5.5%). This was not significant. 2) Median number of new safety behaviours instituted during follow-up was 1 (interquartile range 0-2) in both groups. 3) No significant difference in median number of resources used between intervention (informal 1.5, formal 0.5) and usual care groups (informal 2, formal 1).	 Single hospital study with high proportion of patients excluded. Certain exclusion groups known to have increased rates of abuse.  High rate of abuse not necessarily representative  Research assistants who delivered the intervention were not blinded to patient group allocation  Baseline differences in age and ethnicity between intervention and usual care groups  Possible contamination of usual care group as 30% at follow-up reported being asked about domestic abuse during index visit  14% lost to follow up with differences in age and ethnicity between those lost and those successfully contacted  16% in intervention group compared with 11.5% in usual care group lost to follow-up  Short follow-up period (3 months) despite outcome measure (CAS) using a 12-month referent period  Reliance of outcome measures on self-report
Ritchie et al. 2013 New Zealand	Random selection of 80 clinical records of women aged 16 or older presenting to the ED with a coded diagnosis of domestic violence. Records were selected before the introduction of training in assessment of domestic violence (n=20), after training but before the introduction of a Family Violence Identification Form for recording history, assessment and referral (FVIF) (n=20), after the FVIF was introduced (n=20) and finally nine years later to assess change over time (n=20).	Before/after interventional study	1) Standard of documentation including history, risk assessment and examination. Consisted of 17 items including whether a body map was used, the patient was pregnant, the abuser was present, the patient was afraid of her partner, the patient was afraid to go home, the abuse had increased in severity, children had been abused, children had witnessed violence, abuser threatened homicide, perpetrator and/or victim threatened suicide, a gun was in the house, alcohol or drug abuse by the perpetrator and/or victim, a safety plan was discussed, an appropriate referral made, and presence of a legible signature by the staff member.	1) No significant difference between group 1 and group 2 (following training). 2) Higher scores for group 3 (post-training and documentation) compared to group 2 (training alone) (p=0.09) with these higher scores achieved in records using FVIF and a group of low scores where FVIF not used Comparing group 3 with groups 1 and 2 combined approached significance (p=0.055) 3) Group 4 also had two groups of scores but fewer low scores and higher high scores. Comparing group 1 and 4 showed a significant change (p=0.009)	 Single hospital study. Small rural hospital with limited DA-related presentations  Small sample size  Randomization method not described  Documentation used as indicator of patient assessment and management
Basu et al. 2014 UK	All referrals made to the IDVA service by ED staff from July 2010 to July 2011 and after the intervention from July 2011 to July 2012. The intervention involved introduction of a standardized form (for assessment, flow pathways and supporting contacts), presence of two on-site IDVA's, training for all senior staff on a rolling basis and development of an electronic coding system so staff were aware of previous domestic violence presentations.	Before/after interventional study	1) Referrals made into the service 2) Staff satisfaction Referrals measured in the year preceding and following the intervention.	1) Increase in number of referrals post-intervention (172 post-intervention compared with only 1 referral made to social services in the one-year period pre-intervention) 2) IDVA's highlighted satisfaction in the way their service had been received and utilised. ED staff reported greater clarity regarding what to do when abuse was suspected. Shift in attitude with many doctors feeling it was their duty as well as the nurses duty to enquire about abuse. Increased awareness and confidence in dealing with abuse.	 Single hospital study  No indication of sample sizes
Houry et al. 2011 USA	African-American women aged 21-55 were randomized to receiving a targeted brochure with useful contacts and explanation of this information by a research assistant or enhanced usual care.	Randomized control trial	1) IPV measure: Index of spouse abuse (also alcohol/drug/smoking abuse measures for the relevant women - these results not included here) Outcome measured at baseline and 3 months post-intervention.	1) Strong trend but to increased resource contact but not statistically significant 2) Women in intervention group more likely to take harm reduction actions than enhanced usual care group, but when broken down according to health-risk, no significant changes in the intimate partner violence group	 Single hospital study  Included only African-American women so may not be representative of general target population  Patients had to consent to participation. There may have been systematic differences between those who chose to participate and those who did not.  Research assistants who delivered intervention and follow-up staff were not blinded to patient allocation  Significant differences in age between intervention and usual care group  Control group were given 'enhanced usual care' which may be an invalid control  High attrition rate (22.3%)  Outcome measures reliant on self-report
Muelleman et al. 1999 USA	Women aged 18 or older identified in the ED as injured by a current or former partner before the intervention (n=105) or after the intervention (n=117). The intervention involved on-site advocacy.	Before/after interventional study	1) Use of police resources 2) Orders of protection 3) Shelter use and shelter-sponsored counselling use 4) Repeat visits to the same hospital ED after the index visit Outcomes measured at baseline and after advocacy.	1) No difference in numbers between baseline and after-advocacy group who filed police reports after their index visit (25% vs. 35%, 95% CI difference of -3% to 24%). However, of the women who did not make a police call in ED, more in the after group called police at a later date (18% vs. 39%, 95% CI difference of 1% to 40%). 2) No difference in number of women who obtained full orders of protection 3) Statistically significant increase in number of women who sought shelter (11% vs. 28%, 95% CI difference of 6% to 27%) and counselling (1% vs. 15%, 95% CI difference of 7% to 21%) after-advocacy. 4) No difference in number of women who returned to ED after index visit for any injury, any assaultive injury or for domestic violence injuries during the observation period.	 Single hospital study  No randomization of patients to before/after group  Participants and outcome assessors were not blinded to group allocation  Differences in ethnicity between before/after group  In the after-group, only 57% took up the offer to meet with the advocate so even if the intervention increased resource utilization, large proportion did not take up the intervention in the first place

Comment(s)

Training of ED health providers has a limited effect on clinical practice, as indicated by outcomes such as identification and referral of patients suffering domestic abuse. Moreover, where improvements are noted, they are not maintained over time. However, where staff training is implemented in conjunction with supporting system changes, particularly standardized documentation forms for assessment and referral of victims as well as on-site advocacy, clinically relevant improvements are seen that are sustained over time. There is a lack of evidence that looks directly at the effect of interventions on abuse-related morbidity, largely due to the difficulties in assessing this outcome. It remains uncertain whether surrogate outcomes such as standard of assessment, referral rate and community resource utilization translate to changes in domestic abuse-related morbidity; the ultimate goal. Of note, studies that did evaluate effects on more direct outcomes such as patient re-victimization found no significant benefits.

Clinical Bottom Line

Staff training alone is not sufficient to bring about improvements in the management and in turn, outcome of patients attending an ED due to domestic abuse. However, in combination with support processes to protocolize the changes emphasized during training, sustainable improvements are possible.

References

1. Fanslow JL, Norton RN, and Robinson EM. One year follow-up of an emergency department protocol for abused women Australian and New Zealand Journal of Public Health 1999; 418-20
2. Koziol-McLain J, Garrett N, Fanslow J, et al. A randomized controlled trial of a brief emergency department intimate partner violence screening intervention Annals of Emergency Medicine 2010; 413-23
3. Ritchie M, Nelson K, Wills R, et al. Does training and documentation improve emergency department assessments of domestic violence victims? Journal of Family Violence 2013; 471-7
4. Basu S, and Ratcliffe G. Developing a multidisciplinary approach within the ED towards domestic violence presentations Emergency Medicine Journal 2014; 192-5
5. Houry D, Hankin A, Daugherty J, et al. Effect of a targeted women's health intervention in an inner-city emergency department Emergency Medicine International 2011; 543493
6. Muelleman RL and Feighny KM. Effects of an emergency department-based advocacy program for battered women on community resource utilization Annals of Emergency Medicine 1999; 62-6