Selecting outcomes for intimate partner violence intervention trials: Overview and recommendations

Abstract

Intimate partner violence (IPV) is endemic in societies around the world and detrimental to women's wellbeing. Abused women are frequent users of health services. Despite the recent World Health Organization guidelines on IPV and sexual violence, we need more evidence on effective responses to women in health care settings. Developing robust evidence with potential to inform policy and clinical practice requires greater clarity and consistency across studies in the selection and use of outcomes to evaluate interventions. Drawing on systematic reviews and individual trials aimed at reducing abuse and improving women's health, we discuss critical issues in respect of outcomes. We discuss primary, secondary, intermediate and proxy outcomes and measures used to evaluate interventions for women who experience IPV. We offer recommendations about which outcomes to assess and approaches to doing so within the context of trials in health care settings.

Introduction

Intimate partner violence (IPV) is defined as any behavior within an intimate relationship that causes physical, psychological or sexual harm to those in the relationship (Krug, Mercy, Dahlberg, & Zwi, 2002). The latest data from the World Health Organization (WHO) suggest that one in three women experiences physical or sexual violence in their lifetime (World Health Organization, 2013a). Violence against women by male partners is a significant contributor to illness, disability and death of women around the world (Black, 2011, Heise et al., 2002).

Exposure to violence by partners leads to frequent use of health services suggesting that health care settings offer opportunities to support women and intervene in the cycle of violence (Bonomi, Anderson, Rivara, & Thompson, 2009). Although there is some evidence for responding to IPV within health care settings as identified in the recent WHO guidelines (World Health Organization, 2013b), there is still an urgent need for rigorous trials to determine the effectiveness of interventions in low-, middle- and high-income countries. Interventions for IPV in health care settings mainly comprise those that screen women, offer support/advocacy or involve a therapeutic intervention (MacMillan et al., 2001). Generally, these programs aim to safeguard women from harm, manage symptoms, improve health outcomes, encourage safe communication with others when in an abusive relationship, increase women's confidence to seek help, and improve social networks (Ford-Gilboe, Merritt-Gray, Varcoe, & Wuest, 2011).

Synthesis of different studies has been conducted in respect of screening-only interventions (Feder et al., 2009, Nelson et al., 2012, Taft et al., 2013), advocacy (Ramsay et al., 2009) and interventions (with or without screening) to reduce IPV in primary care (Bair-Merritt et al., 2014) and against pregnant women (Jahanfar, Janssen, Howard, & Dowswell, 2013). The limited number of trials overall, heterogeneity of interventions (e.g. treatment intensity) (Ramsay et al., 2009), and barriers to achieving high-quality evidence (e.g. difficulties in minimizing the risk of bias) impede the emergence of better quality evidence of effective IPV interventions (O'Doherty et al., 2014). An additional barrier concerns the inconsistent selection of effect measures that might allow syntheses of evidence if there was some consistency across trials.

The benefits of meta-analysis for improving evidence include increasing power and precision, answering questions not posed by individual studies, and opportunities to settle controversies arising from conflicting findings (Higgins & Green, 2011). However evidence from meta-analyses is particularly lacking in this field, hampering the process of translating research evidence into what can be safely and effectively delivered in clinical practice. Meta-analyses have been particularly challenged by the lack of clarity and consistency in effect measures (O'Doherty et al., 2014, Ramsay et al., 2009). Robustness of evidence can be enhanced where evaluations are based on better informed decisions around selection of outcomes, approaches to measurement and timeframes. The complexity of IPV trials means that evaluation can involve multiple dimensions, most typical being abuse, social, clinical, wellbeing and service use outcomes. The multiplicity of areas of perceived importance leads to significant heterogeneity in outcomes across trials (O'Doherty et al., 2014, Ramsay et al., 2009). Within certain domains, a large number of tools is available which increases difficulties with combining results. Across the ten trials included in a review of advocacy interventions (Ramsay et al., 2009), eight measured abuse using six different scales (or subscales) and a single item question. Another issue is the lack of measurement in certain areas e.g. those associated with children and parenting and harm. In a review of IPV interventions for pregnant women, Jahanfar and colleagues (p. 8) indicated that “a serious problem in this review was the lack of consistency in, and the limited range of outcomes reported, and the varied way that outcomes such as depression or experience of violence were measured.” Ramsay et al. (2009) called for debate over which outcomes should be measured, which outcomes could even be expected to change in response to an intervention and a more standardized approach to measurement.

Thus, there are a number of critical questions about outcome selection. What outcomes are meaningful for survivors of IPV in health care contexts? Which outcomes should be primary outcomes? Which can be nominated as secondary outcomes? What intermediate and proxy outcomes are relevant? We will discuss critical issues in outcome selection, and offer recommendations for future trials in health care settings.