Agreement with a routine screening policy
Despite growing concern about intimate partner violence as a major public health issue, obstetrician-gynaecologists in our survey largely underestimate the extent of the problem of intimate partner violence, disprove for the most part a universal screening policy, and accordingly, very few of them tend to screen their patients on a regular basis. It may therefore be reiterated that many medical organizations and domestic violence experts do recommend routine screening as the first step in the intervention process for domestic violence because of the prevalence of domestic violence, because routine screening allows for early identification of domestic violence, because knowing about the abuse has potential value in the care of the patient, and because of the low risk of harm in screening [
16]. While most clinicians in our survey only ask their patients for potential exposure to abuse in case of overt laesions and physical injuries, the challenge for health care professionals is therefore to move toward secondary prevention, i.e. routine screening for intimate partner violence whether or not symptoms are immediately apparent. From a public health perspective, physicians involvement in universal screening for domestic violence may fit a broader framework of preventive strategies that encompasses primary as well as secondary and tertiary prevention [
17‐
19]. In Flanders, Belgium obstetrician-gynaecologists not only account for gynaecologic and obstetric pathology but also act as the primary care physicians to the general female population, e.g. in providing primary obstetric care and in offering preventive women's health medicine. Hence, this setting provides obstetrician-gynaecologists with a unique and broad-coverage detection opportunity, which should allow them to be proactively involved in the secondary prevention of intimate partner abuse. At present, no recommendations regarding screening for domestic violence have been made however in this health care setting nor by governmental and public health authorities, neither by professional medical organisations such as the Flemish College of Obstetricians and Gynaecologists.
Barriers to routine screening
In the present study we aimed to assess a wide array of barriers to routine screening for intimate partner violence through a statewide knowledge, attitude, and practice survey among obstetrician-gynaecologists. It was hypothesized that barriers would fit an explanatory model which assumes that physician-targeted information on intimate partner violence first modifies physician's knowledge, then physician's attitude, and eventually physician's behaviours and practice [
13]. Though less than ten percent of the survey participants received or pursued any kind of education or information on intimate partner violence, we established that physician education was indeed the strongest predictor of a positive attitude or willingness to screen each and every patient and of current screening practices for domestic violence.
Major intrinsic barriers to routine screening included a lack of awareness, as obvious from a striking underestimation of the extent of the problem by most participants, and a defined perceived lack of self-efficacy, in particular the inability to skilfully discuss, manage, and refer (potential) cases of domestic violence. Extrinsic barriers on the other hand, including a perceived lack of time and perceived inappropriateness of questioning patients about partner abuse further compromised obstetrician's screening attitude and practice.
These findings are largely in accordance with previous studies indicating that physicians found exploring domestic violence in the clinical setting analogous to "opening Pandora's box" [
20]. In a comprehensive review of published studies on screening for intimate partner violence by health professionals [
21] lack of provider education and lack of time were also among the most commonly cited barriers. Important patient-related external barriers in previous studies were also anticipated patient nondisclosure and anticipated patient fear of repercussions [
21,
22], two issues that were not addressed in our study. Patient nondisclosure much alike the anticipated fear of offending or insulting patients observed in our study may result from a longstanding misconception about patient's beliefs and expectations, as we recently assessed that the vast majority of women from a general obstetric population actually do favour direct questioning about intimate partner violence by their gynaecologists, regardless of a history of partner abuse [
6].
Anticipated patient fear of repercussions is a more intricate issue, as it has not been firmly established from an evidence-base principle that a screening policy does not endanger women and their children at risk of experiencing abusive behaviours. As a matter of fact it has been suggested that screening for violence could be harmful, for example by causing psychological distress or by leading to a further escalation of abuse [
23]. While many others have contradicted this view [
16] it must be acknowledged that, at the very least, future studies are still warranted to assess how patient safety can be ensured when a universal screening policy is to be applied.
Finally our study is less in agreement with the preponderance of previous studies in which lack of effective interventions or poor outcome expectancy was found as a commonly cited barrier [
21,
24]. Gerbert
et al for instance documented in a survey among primary care physicians that interventions for domestic violence were deemed less successful as compared to interventions for other major public health challenges such as tobacco control and HIV/STD risk behaviour interventions [
16]. In our survey, obstetricians generally do not share the fatalistic view that there is no solution for the problem anyway, even though they tend to report a perceived lack of referral services and specialised care facilities for women suffering from domestic violence. Along with their positive outcome expectancy, which is definitely a cornerstone to the implementation of a successful screening policy, it further appears as if medical liability and ethical duty further adds to the screening motivation of respondents in our survey. Though it could be argued that medical liability and ethical duty leaves physicians little room for opting out in this matter, this issue seems to be meagrely addressed in the literature. It is further of note that, in contrast to several previous surveys, obstetricians in our study refuted common misbelieves about partner abuse, in particular they did not consider partner violence as a family affair, for which partners should take responsibility, nor as a phenomenon pertaining to lower social classes or an affliction for which the victim itself is to blame.
Helping physicians in adopting a screening strategy: predisposing, enabling, and reinforcing strategies
As was hypothesized at the outset, we found that physician education was significantly associated with screening attitude and behaviour and interestingly, that having a peer with a history of abuse, which might be viewed as a proxy for familiarity with the issue, also affects physicians' proneness to screen. These findings seem to support the paradigm that increasing knowledge will enhance a positive attitude towards screening and therefore actual screening practices [
13], though this view has been challenged in at least one study [
25].
If anything, it may be acknowledged from intervention studies that instigating 'predisposing' strategies (e.g. education) as such tend to have little effect on eventual physician's behaviours [
21] and hence on their compliance with clinical guidelines. Additionally providing physicians with 'enabling' tools (e.g. screening tools such as the AAS) however, has been found more effective in changing health care provider's behaviours. The use of additional 'reinforcement' strategies such as providing physicians with feedback with regard to their screening practices may further amplify the process of behavioural change [
26‐
28].
Accordingly, it may be inferred that such 'predisposing', 'enabling', and 'reinforcing' strategies may supersede most barriers identified, e.g. the external barrier of time and the internal barrier of perceived self-efficacy, which might no longer be a constraint when provided with an easy-to-handle, time-efficient and acceptable routine screening tool, such as the AAS. 'Enabling' and 'reinforcement' strategies are also likely to help physicians in lowering their threshold for asking questions about partner abuse, to enhance their motivation and to increase their satisfaction with clinical practice.
The crux of intimate partner violence is really that most women who encounter some kind of coercion will not present with overt signs of abuse, but rather with a wide variety of vague and non-specific symptoms, if any. The physician's eye is therefore even in the presence of a high index of clinical suspicion unlikely to grasp most victims and their potential signs in a general obstetric or gynaecologic population. This was also apparent from the present survey. First of all, obstetrician-gynaecologists revealed that clinical detection of violence primarily depends on the presence of physical trauma, whereas psychological or psychosomatic complaints rarely are the impetus to direct questioning about abuse. Secondly, one in three obstetricians stated not to have encountered sexual coercion and two in three not to have confronted physical abuse among their patients over the past five years. Since we previously found an estimated prevalence of physical and or sexual abuse among pregnant women of 3.1% during pregnancy and of 4.4% in the year preceding pregnancy [
6], obstetricians in this survey were actually expected to see at least one patient experiencing partner abuse a month.
Of note is that the optimal mode of administration of screening to detect partner abuse remains uncertain. A number of screening tools have been proposed [
15,
29], but few comparative analyses on these are available. Since gynaecologists in our survey strongly opposed to direct screening with the AAS form, an alternative that might be considered is to provide the woman with the opportunity to self-disclose partner violence using a checklist [
30], such as the Maternal Social Support Scale [
31]. As part of the registration process with the latter, the pregnant women have to complete a Maternity Social Support Scale including two items relating to partner violence: "I feel controlled by my husband/partner" and "There is conflict with my husband/partner" [
31].
Does routine screening result in better patient outcomes?
Whereas poor outcome expectancy was found as a commonly cited barrier in previous studies [
24], it is definitely reassuring that the preponderance of obstetrician-gynaecologists in our survey do believe that screening for intimate partner violence may be an effective means to counteract such abusive behaviours. Glowa
et al previously documented that physicians who identified victims of intimate partner violence after screening indeed reported more often positive patient outcomes (e.g. improved mental health, seeking counselling or services) than negative outcomes (e.g. worsening of violence, substance abuse) and that the physicians believed that these outcomes primarily resulted from disclosure [
32]. Similarly, the adagio that "Recognising the problem of partner violence as such is definitely the first step towards a possible solution" [
16] seems to be a mechanism that has been corroborated by the finding that even brief discussions with a physician, conducted in a concerned and non-judgemental fashion, can help to change the way abused women view their situation, even if they do not disclose the abuse [
33]. It may further be stressed that disclosure of violence improves the patient-provider relationship in terms of communication and of patient and provider satisfaction [
4,
32].
Yet, the effectiveness of a screening policy has been challenged, i.e. it has been put in doubt whether actively identifying partner abuse eventually leads to better patient outcomes. A balanced appraisal of the literature shows that there are actually very few reports on the effectiveness of screening for intimate partner abuse. Though the only systematic review on this subject [
34] discouraged a universal screening policy, it may also be acknowledged that the latter review did not include a single randomised controlled trial on screening for intimate partner violence. While such unbiased studies are therefore definitely warranted, several authoritative publications in the JAMA for instance, have advocated that screening for one of the most imminent health treats to women in our society should not await the proof of evidence-based medicine [
33,
35], as it generally assumed that the benefits of identifying partner abuse are very likely to outweigh its potential drawbacks for both victims and their perpetrators.
Finally, it must be acknowledged that intimate partner violence is strongly associated with other abusive behaviours and child abuse in particular. About half of all children living with mothers suffering from partner abuse will also be exposed to abuse. Intimate partner violence has been associated with severe childhood dysfunction and long-term adverse health effects, e.g. child abuse but also witnessing domestic violence has emerged as a risk for alcoholism, illicit drug use, depressed affect and even suicide [
36,
37]. Accordingly, the American Academy of Paediatrics has recommended intervening on behalf of battered women as an active form of child abuse prevention [
9].
Study limitations
Major study limitations include the use of forced-choice answers, thereby also potentially obliterating barriers yet to be assessed. Though we addressed all six major types of barriers retained in the modified Cabana model [
13] as the conceptual framework of our study, other authors have pinpointed some important potential barriers to screening for domestic violence that we have not accounted for. As discussed above for instance, a limitation to our study is that we did not assess patient nondisclosure, patient safety and patient fear of repercussions. Similarly we failed to account adequately for providers' safety, though we asked if they were anxious to lose control in discussing partner abuse with their patients. Other, yet-to-be-identified determinants undoubtedly may also interfere with physicians' screening attitudes and behaviour. This was also illustrated by the failure of our multivariable analysis in which the nine key determinants included, failed to adequately describe observed screening practices, though there was a strong correlation with physician education. Though the response rate was fairly high in comparison with most similar studies, it should further be acknowledged that the results of our survey should be taken with caution, as almost half of obstetrician-gynaecologists of the sampling frame did not participate in the study. It can therefore not be ruled out that selection bias biased our results, though we established that the sample was representative in terms of gender (sample female/male sex ratio = 65.3 (39.5%,/60.6%) versus 59.0 in the cohort (37.1%/62.9%), Mann-Whitney U test p = 0.522) and age (sample mean = 45.3 years, SD = 10.2 versus cohort mean = 44.2 years, SD = 10.7, independent samples t-test p = 0.18).
Though we designed our study according to the knowledge-attitude-behaviour construct, it must be acknowledged that assessment of knowledge in our survey was limited to some indicators of awareness and familiarity, while the survey did not entail direct questions on risk factors, signs, symptoms, and comorbidity patterns relating to partner violence as an issue of knowledge. Nor did we make an attempt to assess obstetrician's knowledge of screening strategies. Short
et al also emphasized that many standardised intimate partner violence survey tools do not adequately assess actual knowledge [
38]. By accounting for a number of knowledge-related items, Short and colleagues very recently developed a reliable survey instrument – the so-called PREMIS tool (Physician Readiness to Manage Intimate Partner Violence Survey) – that can be used to measure the effectiveness of intimate partner violence educational programs [
38].
Several models other than the knowledge-attitude-behaviour model applied in our survey have indeed been developed to assess health care provider characteristics and training needs in relation to intimate partner violence. Of particular interest are those models constructed through the use of psychometric techniques, which have resulted in some refined tools that may guide future IPV policy interventions and training programs [
38,
39].
It needs to be stressed that in contrast to our study, which was primarily designed as a observational study to identify and quantify a defined set of external and internal barriers to intimate partner violence screening in a health care setting where routine screening is rather an exception to the rule, the aforementioned psychometric models do not only serve as explanatory models, but also allow for monitoring future intervention studies through publicly available instruments derived from these models [
38,
39].