It’s estimated that 1 in 4 women in the US have suffered intimate partner violence (IPV). Victims suffer from poorer health outcomes (such as higher rates of cardiac disease and problem drinking) and also use health services more.
To take advantage of this contact with health services, and attempt to address the health needs of people who experience IPV, policy makers in the US in 2013 upgraded their recommendations regarding screening for IPV in primary care from “insufficient evidence to recommend routine screening” to recommending that clinicians screen all women of child bearing age.
An assumption within is that we have sufficient evidence that effective interventions are available for those who screen positive. However, the evidence on interventions for IPV in primary care had not, previously, been systematically reviewed and synthesised, and that’s what this paper aimed to do.
The authors aimed to review any peer-reviewed research examining the impact of IPV interventions associated with visits to a primary care provider. The intervention itself had to include at least a component that was delivered within primary care. The intervention also had to be ‘patient focused’ (rather than for example aimed only at professionals) and measure ‘patient specific outcomes’, which is the kind of unnecessarily vague term that gets my critical skills Spidey-Sense tingling. The authors included any study with quantitative data, rather than for example limiting the review to randomised controlled trials.
- 944 articles were found in the initial search, with 80 subject to full-text review and 17 included in the final set. 11 of these were randomised, two were pre-post tests, two were prospective cohorts and two were descriptive
- All the studies focused on women; there were none of men (remember that the policy recommendation was for “women of child bearing age”)
- Interventions were delivered by IPV advocates, social workers, counsellors or nurses. Only 2 studies included the family physician or GP themselves and this was at screening stage rather than in delivering a response
The authors report findings for multiple outcomes, both related to violence/safety and to health. In the results, they comment that “It should be noted that 76% (13/17) of included studies demonstrated at least one intervention related benefit for patients.” This kind of statement makes me less Spider-Man and more Hulk.
There’s a reason that we focus on primary outcomes rather than any outcome, and why we consider whether multiple studies show us something consistent about an agreed outcome: because if you run several studies each with several outcomes then something is bound to come out as significant, but it’s probably not very reliable.
What I found notable was that health outcomes seemed to rarely be impacted, despite the initial rationale about increased health use and poorer health outcomes for this group. (In the abstract, this is helpfully summarised as “Some studies also documented health improvements”).
It might seem like a no-brainer to involve primary care in this kind of work, but remember that every intervention has a cost, whether that’s financial, time (for example, taking up the time that could be spent on other, perhaps more effective interventions) and also possible harm to the patient themselves, particularly for complex and sensitive issues such as this. This means we need a sound evidence base to make the judgement of whether these interventions are effective and whether their benefit outweighs the cost. Unfortunately, this review provides us with very little information to help us make this assessment. Take it away Dr Banner…
ARRRGH! HULK NO FIND SUPPORTED CONCLUSION! HULK SKIP STRAIGHT TO LIMITATIONS, WHICH ARE MORE TELLING.
- A huge limitation for me is that the authors don’t report the quality of the included studies. One of the primary functions of a review is to examine the quality of an evidence base, and I find the absence of any quality assessment quite shocking.
- The authors also don’t stratify the findings according to the type of study – for example, picking out any trends or consistencies in those studies with more rigorous designs (the randomised trials) or those with bigger sample sizes (the range was apparently 18 to 2708 which seems worthy of comment!). There’s also no consideration of sensitivity to other factors, for example age or ethnicity.
- Considering they limited the included studies to those with quantitative data, there’s no attempt at pooling results or comparing effect sizes.
- We also don’t hear any report about drop outs and follow up rates (which can act as a proxy measure of how acceptable an intervention is, something that I suspect would be extremely important for an issue as sensitive as this.)
- “Although the authors did not use a specific systematic review protocol like PRISMA, the methods met most of PRISMA’s 27-item checklist.” To me, this reads as “we didn’t use the PRISMA, one of the peer reviewers told us off, so we’ve added a line saying it’s probably PRISMA-esque.” I disagree. PRISMA includes reference to quality, bias, sample size, follow up rates and so on – all very important components and all of which are unfortunately missing from this review.
Bair-Merritt MH. Primary care-based interventions for intimate partner violence: a systematic review. Am J Prev Med. 2014 Feb;46(2):188-94. doi: 10.1016/j.amepre.2013.10.001.