Screening women for intimate partner violence in health care settings
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
To assess the effectiveness of screening interventions for intimate partner violence conducted within health care settings on identification, referral and health outcomes for women.
Background
Intimate partner violence
For the purpose of this review, the definition of intimate partner violence (often termed domestic violence) is that defined by the World Health Organisation as any behaviour within an intimate relationship that causes physical, psychological or sexual harm to those in the relationship (Krug 2002). Intimate partner violence often involves various forms of systematic abusive behaviours usually in combination. These include threats of and actual physical violence, sexual violence, emotionally abusive behaviours, economic restrictions and other controlling behaviours. Many survivors of partner violence report that the physical violence is the least damaging suffered: it is the relentless psychological abuse that leaves the woman with long lasting adverse effects (Campbell 2002).
Domestic violence against men is not included in the review because the majority of abuse with serious health and other consequences is that committed by men against their female partners, which is why most screening interventions target women (Taft 2001). We will include same‐sex violence. Abuse perpetrated by ex‐partners is included in the review since women who are separating or have just left partner are at increased risk of violence (Florida 1997; Wilson 1993) .
Prevalence of intimate partner violence
Abuse of women by their partners or ex‐partners is a worldwide phenomenon (Watts 2002). The recent World Health Organisation World Report on Violence and Health (Krug 2002) revealed that in 48 population based surveys around the world, 10%‐69% of women had been physically assaulted by their partners at some stage in their lives. Some comparative examples of prevalence from the WHO studies are outlined in Table 1. Definitions used in prevalence studies range from physical abuse in current relationships to the inclusion of physical, emotional and/or sexual abuse in past relationships (Hegarty 1998). Estimates of the magnitude of intimate partner violence are obtained from community surveys, clinical samples and public records. It is likely that some of the discrepancy in prevalence rates is due to differences in definitions of partner violence, populations sampled, and classification methods used by the researchers. Other differences may reflect cross‐cultural difference.
Impact on women's health and their use of services
Intimate partner violence can have short‐term and long‐term negative health consequences for survivors, even after the abuse has ended. (Campbell 2002) World Development reports (World Bank 2006) and statements from the United Nations (Ingram 2005) emphasise that partner violence is a significant cause of death and disability, e.g. HIV/AIDS, on a world‐wide scale and the World Health Organisation highlights violence against women as a priority health issue (Krug 2002).
Psychosocial health of abused women
The most prevalent mental health sequelae of intimate partner violence are depression and post‐traumatic stress disorder (Herman 1992; Golding 2002) (Hegarty 2004) and often suffer from low self‐esteem and hopelessness. (Kirkwood 1993) Suicide is also associated with intimate partner violence in both industrialised and non‐industrialised countries, as is attempted suicide.(Counts 1987; Golding 2002). Women who have experienced physical or psychological violence are more likely to abuse alcohol and drugs than are non‐abused women (Golding 2002). Any of these effects also impact detrimentally on women's ability to parent and thus consequently also on children.
| Country | Lifetime prevalence | Previous 12 months |
| Switzerland | 21% | 7% |
| Egypt | 34% | ‐ |
| Zimbabwe | 17% | ‐ |
| India | 40% | 14% |
| Cambodia | 16% | ‐ |
| Nicaragua | 28% | 12% |
| Canada | 29% | 3% |
Physical health of abused women
Abused women experience many chronic health problems. The most consistent and largest physical health difference between abused and non‐abused women is the experience of gynaecological symptoms.(Campbell 2002; McCauley 1995). Other conditions include chronic pain and central nervous system symptoms (Campbell 2002; Diaz 1999) self‐reported gastrointestinal symptoms, diagnosed functional gastrointestinal disorders (Coker 2000) and self‐reported cardiac symptoms (Tollestrup 1999).
Intimate partner violence is also one of the most common causes of injury in women (Stark 1996; Richardson 2002) and femicide, with over 50% of all female murders committed by their partners or ex‐partners in the UK and USA (Home Office 1998; Brock 1999). In Australia, a far higher percentage of Indigenous compared with non‐Indigenous women are murdered by their partners (Mouzos 2003) . Intimate partner violence continues when women become pregnant ‐ or may escalate (Gazmararian 2000). Most studies using similar measures of partner violence show a prevalence of between 4%‐8%. (Campbell 2002) The most serious outcome is the death of the mother or the fetus (Parsons 1999; Jejeebhoy 1998) However it is also associated with high rates of pregnancy, miscarriage, abortion, (Taft 2004; Gazmararian 2000) low birth weight, (Murphy 2001) premature birth and fetal injury (Mezey 1997).
Interventions by health practitioners to improve the health consequences for women experiencing domestic violence
Women experiencing intimate partner violence present very frequently to health services and require wide‐ranging medical services (Davidson 2000). Health care services play a central role in abused women's care, but the quality of health care professionals' responses has been a focus of concern since the 1970s (Stark 1996). Over the last few decades there has been a concerted effort by women's and justice organisations and the voluntary sector to respond to the needs of women experiencing or who have experienced violence. In contrast, the response of health services has been slow. While most health professionals believe that intimate partner violence is a health care issue (Richardson 2001), there has been a reluctance to confront the problem. A number of barriers contribute to this ambivalence on the part of practitioners (Hegarty 2001). These include a perceived lack of time and support resources, fear of offending the woman, a lack of knowledge and training about what to do for the woman, and a belief that the woman will not leave the abusive relationship (Waalen 2000) .
Despite these barriers there has been progress and, within the last ten to fifteen years, many health professional associations around the world have published guidelines for clinicians on how to identify women who have been abused. Implicit in many of these recommendations is the assumption that screening or asking routinely about abuse will increase identification of women who are experiencing violence, lead to appropriate interventions and support, and ultimately decrease exposure to violence and its detrimental health consequences, both physical and psychological.
Screening
Screening aims to identify women who have experienced or are experiencing intimate partner violence in order to offer interventions leading to beneficial outcomes. However, within the field of domestic/family violence its immediate and longer term value remains controversial. There is a range of understanding about screening for IPV, a problematic concept when traditional screening criteria are applied (Hegarty 2005).
In a discussion on screening, it is important to distinguish between universal screening (the application of a standardised question to all symptom free women according to a procedure that does not vary from place to place) and case finding (asking questions if indicators are present). For this review, screening is defined as any method that aims for every woman patient in a health care setting to be asked about having experienced or currently experiencing intimate partner violence. This may include the use of screening tools (e.g. Abuse Assessment Screen) (McFarlane 1992), as well as clinicians asking one or a range of questions related to partner violence at one time point or repeated at several time points. It is very unlikely that one single question will address the range of women's experiences of partner abuse. Some women may prefer face to face inquiry whilst others prefer paper based tools (MacMillan 2006). Identifying partner violence is the first step in intervention. Women may have experienced long standing abuse to or it may have commenced recently; they may be unaware of their abuse or actively seeking support for change. Responses may need to differ.
A review of studies addressing the UK National Screening Committee criteria (Ramsay 2002) found that screening leads to a modest increase in the number of abused women being identified following screening by health professionals, but screening was currently not acceptable to the majority of health professionals surveyed. The US and Canadian Task Forces on Preventive Health Care have concluded that there is insufficient evidence to recommend for or against routine screening for violence against women (Wathen 2003; Nelson 2004). They concluded there is no evidence that women experience better outcomes from post‐screening interventions (Ramsay 2002; Wathen 2003). This has not deterred many governments from around the world with proceeding with partner violence screening programs.
At present, it is not clear that there are no short or longer term harms for women who were screened (Jewkes 2002) and with which populations of women and practitioners it is most effective. We will examine sub‐groups of women if possible and the duration of follow‐up in all included studies. Thus, in order to have clear evidence about what health professionals can do safely and effectively to decrease the impact of intimate partner violence on women, to determine what is cost‐effective, and to inform health providers and policy makers about the benefit of screening interventions, there is an urgent need to assess and identify health sector screening interventions for intimate partner violence (Davidson 2000). In particular, this systematic review will examine the most rigorous evidence around health service screening interventions for intimate partner violence to ascertain if they do no harm and promote and enhance women's health.
Objectives
To assess the effectiveness of screening interventions for intimate partner violence conducted within health care settings on identification, referral and health outcomes for women.
Methods
Criteria for considering studies for this review
Types of studies
Any studies that allocate individual participants or clusters of participants by a random or a quasi‐random method (such as alternate allocation, allocation by birth date, etc) to a screening intervention compared with usual or standard care.
Types of participants
Women (aged 16 and above) attending a health care setting.
-
Health care setting is defined as any health setting where health services of any kind are delivered. including the following and home visits by these services:
-
General (family) practice
-
Ante and postnatal services
-
Hospital emergency, inpatient or outpatient services
-
Private specialists (e.g. O&G, GU, Psychiatry)
-
Community health services
-
Drug and Alcohol services
-
Mental health services
Types of interventions
Any partner violence screening intervention implemented by a health professional in a healthcare setting as listed above. Screening intervention is defined as any of a range of methods (face‐to‐face or survey) that aims for all women patients in a health care setting to be asked about having experienced or currently experiencing intimate partner violence, including the use of screening tools (e.g. Abuse Assessment Screening, etc) as well as asking one or a range of questions related to partner violence once or at repeated intervals.
Types of outcome measures
Primary outcome measures are:
A. Identification of abuse by health professionals
B. Information‐giving and referrals to support agencies by health professionals (including take‐up rates when available)
Secondary outcomes include:
C. Partner violence measured by
(i) validated instruments (such as the Composite Abuse Scale, Index of Spouse Abuse etc) to measure actual or threatened partner violence
(ii) self reported partner violence, even if using an unvalidated scale
D. Women's perceived and diagnosed physical health outcomes, using measures of
(i) physical health (such as SF‐36, General Health Questionnaire)
(ii) physical injuries, such as fractures and bruises (self‐reported or documented in medical records)
(iii) chronic health disorders, such as gynaecological problems, chronic pain and gastrointestinal disorders (self‐reported and/or clinical symptoms documented in medical records)
(iv) deaths, all‐cause or partner violence related (documented in medical records or routinely collected data systems)
E. Women's psychosocial health using measures of
(i) depression (such as Beck Depression Inventory (BDI), Center for Epidemiologic Studies Depression Scale (CSED) etc)
(ii) post‐traumatic stress (such as Impact of Events scale, Posttraumatic Stress Disorder Checklist etc )
(iii) anxiety (such as Spielberger's State‐Trait Anxiety Inventory, Beck Anxiety Inventory etc)
(iv) self‐efficacy (such as Generalized Perceived Self‐Efficacy Scale, Sherer's Self Efficacy Scale etc)
(v) self esteem (such as Rosenberg Self Esteem Scale, Coopersmith's Self Esteem Inventory etc)
(vi) quality of life (such as MANSA, SF‐12)
(vii) perceived social support (such as Medical Outcomes Scale, Sarason's Social Support Questionnaire etc)
(viii) alcohol or drug abuse (such as AUDIT, Addiction Severity Index, Alcohol and other Drug Abuse Scale etc)
F. Occurrence of adverse outcomes, such as:
i. increased deaths, all‐cause or partner violence related (documented in medical records or routinely collected data systems)
ii. increase of partner violence as measured by any of the above (validated measures or self report)
iii. increase of physical or psychosocial outcomes as listed above.
iv. False negatives and false positives of screening tests
G. Cost/benefit measures such as:
i. health service use
ii. days out of role
iii. medication use
For the purposes of this review, we define short‐term follow‐up as less than 6 months, medium‐term follow‐up as 6‐12 months, and long‐term follow‐up as more than 12 months.
Search methods for identification of studies
Searches will be made of the international literature for published and unpublished studies (grey literature). There will be no language restrictions imposed. A variety of sources will be used to identify studies:
Electronic databases
Cochrane Central Register of Controlled Trials, (CENTRAL), Cochrane Library
Medline
Embase
Cinahl
Database of Abstracts of Reviews for Effectiveness (DARE), Cochrane Library
Sociological Abstracts
International Trial Registers
Applied Social Sciences Index and Abstracts (ASSIA)
PsycInfo
Searches of these databases will include all studies referenced from the respective start dates of the databases. We will use citation tracking to identify other studies and scrutinise papers in the reference lists of included studies.
Other electronic sources
World Health Organisation/Women's Health and Development website
Other websites related to intimate partner violence, such as the Domestic Violence Data Source (DVDS)
Non‐electronic sources or unpublished research
We will hand search the following four journals for studies of screening for domestic violence since 1980: Journal of Family Violence, Journal of Interpersonal Violence, Violence and Victims, Women's Health. We will also hand search the following six journals starting from the last dates of hand searches already carried out by the Cochrane Collaboration: American Journal of Preventive Medicine, American Journal of Public Health, Annals of Emergency Medicine, Archives of Internal Medicine, Australian & New Zealand Journal of Public Health, Journal of the American Medical Association.
Personal communication with the first authors of all included articles
We will e‐mail the list of primary studies included in the review to relevant researchers to check for omissions (and, in particular omissions of non‐peer reviewed studies (grey literature), and undertake consultation with members of intimate partner violence groups and related organisations. Efforts will be made to make contacts in countries where English is not the first language via the WHO Domestic Violence Programme.
Search Strategy
The following search strategy will be used to search MEDLINE:
1. Battered women/
2. Domestic Violence/
3. Spouse abuse/
4. (abuse$ adj3 wom#n).tw.
5. (abuse$ adj3 spous$).tw.
6. (abuse$ adj3 partner$).tw.
7. ((wife or wives) adj3 abuse$).tw.
8. ((wife or wives) adj3 batter$).tw.
9. (partner$ adj3 violen$).tw.
10. (spous$ adj3 violen$).tw.
11. or/1‐10
12. Mass Screening/
13. screen$.tw.
14. identif$.tw.
15. (routine$ adj3 (ask$ or question$)).tw.
16. detect$.tw.
17. or/12‐16
18. exp women/
19. Adolescent /
20. (wom#n or female$).tw.
21. adolescen$.tw
22. teen$.tw
23. or/18‐23
24. 11 and 17 and 23
This search strategy will be piloted on Medline with and without the Cochrane RCT filter to see if anything is caught which should have been included and then a decision will be made on the value of including the filter. The terms used in this strategy will be modified, where necessary, for use in the other databases listed.
Data collection and analysis
Selection of studies
Two reviewers will independently select studies from titles and abstracts, for potential inclusion in the review. Where possible, disagreement about abstract inclusion between the two reviewers will be resolved by reading the full study followed by discussion. If agreement cannot be reached then a third reviewer will be asked to assess whether the study fulfils the inclusion criteria. If additional information is needed to resolve a disagreement, then this will be sought from the first author of the study in question. Where studies could be considered plausible, but were excluded, the reasons behind those decisions will be given.
Assessment of methodological quality
Methodological quality will be assessed independently by two reviewers. The quality of all included studies will be assessed using criteria outlined below and cross‐checked in accordance with the methodological criteria outlined in Section 6 of The Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2005), 'Assessment of study quality'.
1. Concealment of allocation
DONE ‐ unit of allocation was institution, team or professional and any random process explicitly described, e.g. use of random number tables, OR unit of allocation was the woman or episode of care and some form of centralised randomisation scheme, an on‐site computer system or sealed opaque envelopes used
NOT CLEAR ‐ allocation procedure not described explicitly OR unit of allocation was the woman or episode of care and reported use of `list' or `table', `envelopes' or `sealed envelopes' for allocation
NOT DONE ‐ use of alternation (such as reference to case record numbers, dates of birth, day of the week or any other such approach) OR unit of allocation was the woman or episode of care and reported use of any allocation process that is entirely transparent before assignment, such as an open list of random numbers or assignments OR allocation was altered by investigators, professionals or the women
2. Attrition: Follow‐up of women or follow‐up episodes of care
DONE ‐ outcome measures for > 70% of women randomised or women who entered the trial
NOT CLEAR ‐ not specified
NOT DONE ‐ outcome measures for <70% of women randomised or women who entered the trial
3. Blinded assessment of outcome(s)
DONE ‐ stated explicitly that primary outcome variables were assessed blindly OR outcome variables are objective, e.g. number of presentations to A&E for treatment of physical injuries, levels of abuse assessed by a standard measure (eg Composite Abuse Scale)
NOT CLEAR ‐ not specified
NOT DONE ‐ assessors were not blinded to outcomes.
4. Baseline measurement
DONE ‐ baselines characteristics, e.g. women's IPV, socio‐demographic and health status were measured prior to screening , and no substantial differences were evident across study groups
NOT CLEAR ‐ baseline measures not reported, or unclear whether baseline measures are different across study groups
NOT DONE ‐ differences at baseline in main outcome measures likely to undermine the post intervention differences, e.g. differences between groups before the intervention similar to those found post intervention
5. Outcome measures assessment and inter‐rater reliability
DONE ‐ two or more raters with agreement about the outcomes measures (greater than or equal to 90% or greater than or equal to kappa 0.8) OR outcome assessment is objective, e.g. number of presentations to A&E for treatment of physical injuries, levels of abuse assessed by a standardised test
NOT CLEAR ‐ reliability not reported for outcome measures obtained by file extraction or collected by an individual
NOT DONE ‐ two or more raters with agreement < 90% or kappa < 0.8
6. Protection against contamination
DONE ‐ cluster allocation by community, institution or practice has been effectively undertaken and it is unlikely that control group patients/clients received the intervention
NOT CLEAR ‐ it is considered possible that communication between experimental and control group professionals and patients could have occurred
NOT DONE ‐ likely that control group received the intervention, e.g. cross‐over trials or if women rather than professionals were randomised
In addition, the quality of cluster randomised trials will be assessed in accord with the criteria recommended by Eldridge (Eldridge 2004) below:
In addition, quality criteria for cluster trial designs can be seen in Table 2. These will be assessed as follows:
A= HIGH QUALITY
Study complies with all 5 recommendations
B = MODERATE QUALITY
Study complies with at least two recommendations
C = LOW QUALITY
Study complies with only one recommendation
Data management and extraction
Data will be extracted independently by two reviewers and entered on data collection forms (paper) prior to being independently entered into REVMAN. As before, where possible any disagreements between the two reviewers will be resolved by discussion. If this is not possible then a third reviewer will adjudicate and all such decisions will be documented. If necessary, the first authors of studies will be contacted to assist in resolving the disagreement.
| Recommendation | Order |
| Use of a clustered design is justified | 1 |
| Effects of clustering are allowed for in analysis | 2 |
| Stratification is carried out or justified if not | 3 |
| The number of clusters included is sufficient to enable appropriate analyses and ensure sufficient study power (at least 4 clusters per intervention group) | 4 |
| Analysis allows for effect of confounders or justified if not | 5 |
Incomplete data
Missing data and drop‐out rates will be assessed for each of the included studies and the number of participants who are included in the final analysis will be reported as a proportion of all participants in the study. Reasons given for missing data will be provided in the narrative summary and the extent to which the results are altered by missing data will be ascertained. Best case and worst case scenario analyses will be undertaken to estimate the effect of the missing data on the results of the study Assessment will be made of the extent to which studies have conformed to an intention‐to‐treat analysis.
Data synthesis
If there are sufficient data and it is appropriate to do so, meta‐analysis will be performed using both a fixed effect model and a random effects model.
Binary data
For binary outcomes (for example, woman screened or `woman not screened') a standard estimation of the Odds Ratio with a 95% confidence interval will be calculated.
Continuous data
Continuous data will be analysed if (i) means and standard deviations are available or are obtainable from the authors of studies, and (ii) the data are normally distributed. If the second standard is not met then such data will not be entered on the RevMan software (as this assumes a normal distribution). It will, however, be reported in the 'Other data types' of the results section where they will be analysed using appropriate non‐parametric tests. Where all measurements are comparable and on the same scale (such as, identification rates) then these will be combined to obtain weighted mean differences. Where scales are measuring the same clinical outcomes in different ways (e.g. depression, quality of life), mean differences will be standardised in order to combine results across scales.
Cluster randomized trials
Statistical methods for cluster‐randomised trials to be used in the review are described in section 8.11.2 of the Cochrane Reviewer's Handbook (Higgins 2005). We will attempt to extract from any such trials direct estimates of the required effect measure (for example, an odds ratio with its confidence interval) from an analysis that properly accounts for the cluster design; alternatively, effect estimates and their standard errors will be extracted and the standard errors will be corrected to account for the design effect (Donner 1980). To this end, we need information on intraclass correlation coefficients (ICCs), which are estimates of the relative variability within and between clusters within the studies (Donner 1980). We will extract this information from the articles, and, if not available, we will contact the authors or use external estimates obtained from similar studies. Subsequently, the estimates and their corrected standard errors from the cluster randomized trials can be combined with those from parallel designs using the generic inverse variance method in RevMan. Statistical advice will be sought from editors of the CDPLPG if necessary.
Investigation of heterogeneity
Consistency of results will be assessed visually and by examining I², (Higgins 2002) a quantity which describes approximately the proportion of variation in point estimates that is due to heterogeneity rather than sampling error. We will supplement this with a test of homogeneity to determine the strength of evidence that the heterogeneity is genuine. Where significant statistical heterogeneity is detected differences in the characteristics of the studies or other factors will be explored as possible sources of explanation. Any differences will be summarised in a narrative synthesis.
Sensitivity analyses
Primary analyses will be based on available data from all included studies relevant to the comparison of interest. To assess the robustness of conclusions to quality of data and approaches to analysis, sensitivity analyses will be performed. These analyses will include the following:
-
study quality
-
differential drop‐out
-
intention to treat
-
duration of follow up
Sub‐group analyses
If a sufficient number of studies is identified, we will perform sub‐group analyses for the following:
-
type of screening intervention (use of validated screening tool, single question etc)
-
screening intervention only or embedded within a larger multi‐component intervention
-
type of health care setting
Publication bias
Funnel plots will be drawn to investigate any relationship between effect size and study precision (closely related to sample size) (Egger 1997). Such a relationship could be due to publication or related biases or due to systematic differences between small and large studies. If a relationship is identified, clinical diversity of the studies will be further examined as a possible explanation.
| Country | Lifetime prevalence | Previous 12 months |
| Switzerland | 21% | 7% |
| Egypt | 34% | ‐ |
| Zimbabwe | 17% | ‐ |
| India | 40% | 14% |
| Cambodia | 16% | ‐ |
| Nicaragua | 28% | 12% |
| Canada | 29% | 3% |
| Recommendation | Order |
| Use of a clustered design is justified | 1 |
| Effects of clustering are allowed for in analysis | 2 |
| Stratification is carried out or justified if not | 3 |
| The number of clusters included is sufficient to enable appropriate analyses and ensure sufficient study power (at least 4 clusters per intervention group) | 4 |
| Analysis allows for effect of confounders or justified if not | 5 |
