Intimate partner violence (IPV) has been defined by the World Health Organization as ‘any behaviour within an intimate relationship that causes physical, psychological or sexual harm to those in the relationship’ (Krug Reference Krug, Dahlberg and Mercy2002: p. 89).
One in three women globally report violence from an intimate partner at some stage in their lives. The effects of such violence on women are widespread and IPV is estimated to be the leading cause of death, disability and illness for women of childbearing age, owing to mental health problems that are associated with it (Ayre Reference Ayre, Lum On and Webster2016).
This month's Cochrane Corner review (Hameed Reference Hameed, O'Doherty and Gilchrist2020) clearly outlines the current understanding of IPV, its prevalence and effects on women. Its authors do so by referring to research that has been published in the field that enables the reader to clearly contextualise IPV experienced by women. The need for this review is demonstrated by the limited research published on the subject.
The review mentions that there is an increasing awareness of the negative effects of IPV on victims’ long-term mental health and that there is a growing understanding of the term ‘complex trauma’ (Hermann Reference Hermann1992; Courtois Reference Courtois and Ford2012; Baird Reference Baird, Alaggia and Jenney2019). However, it does not elaborate or define this term. It is also unclear how factors such as frequency, recency, severity and type of trauma influence clinical presentations, which may, in turn, affect the treatment that victims receive. These issues are not explicitly discussed in the review.
The Blue Knot Foundation at the National Centre of Excellence for Complex Trauma in Australia describes complex trauma as exposure to multiple traumas that have severe, persistent and cumulative effects on the individual. These include difficulties with shame, trust, self-esteem, identity, emotion regulation and feelings of being trapped. Maladaptive coping strategies to manage complex trauma include substance misuse, self-harm and disordered eating, among others (Blue Knot Foundation 2021).
Clinical question and outcome measurement
This review aimed to explore the effectiveness and safety of psychological therapies delivered to women who experienced IPV at any time in their adult life. Primary and secondary outcome measures were used to assess the effectiveness of the therapies using a variety of psychometric scales.
The primary outcomes measured were depression, self-efficacy and drop-out rate from therapy, and the secondary outcomes included anxiety, post-traumatic stress disorder (PTSD), quality of life, re-exposure to IPV and safety planning behaviour. Outcome measurement was assessed in the short term (under 6 months), medium term (6–12 months) and long term (over 12 months).
It is a positive feature that the studies included in this review appear to have measured a wide variety of outcomes, as the effects of IPV are widespread across several domains of victims’ lives. However, the measures used fail to capture the fundamental psychological processes that arise from complex trauma, which may be underlying clinical presentations. Additionally, there seem to be inconsistencies with regard to which specific outcomes studies report. This could reflect a lack of consensus in the research community on how to measure the effects of IPV.
The review authors used drop-out from therapy as an indicator of harm, but they did not include reasons for drop-out. Conversely, some studies included in the review reported on adverse events from therapy. These were included in the results of the review, which, like reason for drop-out, give a better representation of the potential harm of therapy.
Although all psychometric measures used in the trials were appropriate measures, not all studies included information about whether patients met the threshold for clinical diagnosis or not. This may have repercussions for the quality of the results as, although a therapy might have resulted in a significant reduction in a score on a scale, it is unclear whether this correlated with clinical improvement or not.
Study methodology
The review included randomised controlled trials (RCT), quasi-RCTs (Box 1), cluster RCTs and cross-over trials. Trials that did not include a suitable comparator intervention and trials that used convenience sampling (Box 2) with no randomisation were excluded.
Quasi-RCTs
Quasi-experimental studies (also called quasi-RCTs or quasi-randomised trials) are studies in which, for logistical or ethical reasons, the method of allocation is not truly random. A common scenario in a quasi-RCT is the comparison of outcomes measured before and after the interventions. The lack of randomisation used in this method should be kept in mind when interpreting results, as there may be significant barriers to interpreting intervention and comparator populations (Harris Reference Harris, McGregor and Perencevich2006).
Convenience sampling
Convenience sampling is a non-probability sampling method in which the sample is taken from a population that is readily available to study. This method of sampling is quick, easy and cost-effective. However, when used in isolation this technique has several disadvantages, such as high bias and low power, resulting in the inability to draw conclusions about differences between subgroup populations. Using homogeneous convenience sampling as opposed to conventional convenience sampling can result in narrower but clearer generalisability (Jager Reference Jager, Putnick and Bornstein2017).
Overall, the methodology of this review is robust and comprehensive. The researchers carried out a thorough electronic search on several online databases. Of note, it is highly commendable that the researchers put in the effort to contact authors for information on unpublished or ongoing trials. The selection process was clearly recorded in a PRISMA flow diagram. Two review authors independently assessed for trial inclusion and bias. Any disagreements were resolved by discussion or by consulting the senior review author. Risk of bias was systematically and extensively documented and presented in the review.
The statistical analysis used pooled odds ratio (OR) and 95% confidence intervals (CI) to measure dichotomous outcomes and mean difference (m.d.) and standardised mean difference (s.m.d.) for continuous outcomes (Box 3). GRADEpro GDT software was used to classify the certainty of the overall body of evidence on the basis of trial design, risk of bias, inconsistency, imprecision, indirectness and publication bias (Box 4).
Odds ratios and standardised mean differences
An odds ratio (OR) refers to the level of association between two events (e.g. exposure and outcome) and is used to measure dichotomous outcomes. It is a measure of correlation (not causation) and describes the ratio between two events occurring together and one event occurring in the absence of the other. Two events are unrelated and independent of each other if the OR is 1 (Szumilas Reference Szumilas2010).
The standardised mean difference (s.m.d.) is used in meta-analyses when different studies measure the same outcome using different measures. For example, in this review depression was assessed using a variety of scales across the different trials. The aim of using the s.m.d. is to standardise the results from different studies on a common scale, so that they can be compared more accurately. The s.m.d. is used when analysing continuous outcomes (Higgins Reference Higgins and Green2011).
The GRADEpro GDT Guideline Development Tool
GRADEpro GDT (Cochrane Methods GRADEing 2021) is an application used to create ‘summary of findings’ tables for Cochrane systematic reviews. It also helps in the development of clinical practice guidelines and recommendations for healthcare services.
The overall certainty of the body of evidence was ranked as low to moderate, owing to clinical and statistical heterogeneity, inconsistency, imprecision and risk of bias. Indeed, only five trials (15%) had a low risk of bias across all criteria that were considered.
Participant characteristics
The review included 33 studies involving 5517 women randomly assigned to either experimental (2798 women, 51%) or comparator interventions (2719 women, 49%) from a variety of healthcare and community settings. The study included all participants identifying as women aged 16 years or older who self-reported recent or past exposure to IPV. Women with coexisting mental health diagnoses and/or substance use problems were also included.
Studies that included victims of violence not perpetrated by an intimate partner and women who were sexually abused and traumatised as children were excluded from this review. Although the review is specifically about women who have experienced IPV, it would have been interesting to compare women with and without a history of childhood abuse to see whether there were any significant differences in both clinical presentation and response to therapy.
The review included studies from all over the world, but most took place in high-income English-speaking countries (USA, Australia, UK). In most studies, participants came from diverse ethnic backgrounds. In total, 28 of the included trials (85%) had been published in the past decade, which is reflective of how young this field of research is. This is worth bearing in mind when analysing reviews like this, as the evidence base appears to still be in its early stages.
It is generally beneficial to compare research from different parts of the world. However, IPV is a complex problem that is heavily influenced by social factors that vary greatly between countries. These include social welfare, income, class, education, female empowerment and mobility. This review does not analyse these variables and does not discuss whether studies from different countries are comparable or not.
The review mentioned that participants’ socioeconomic status was generally low. However, it was unable to explore the complex relationship between IPV and social class, ethnicity, racial discrimination, disability, social vulnerability and discrimination based on gender identity or sexual orientation, any of which could augment the negative effects of IPV. These factors could also influence the probability of a woman being a victim of IPV and may have implications for their access to treatment.
Furthermore, information about participants and the relationship between the participants and their abusers was lacking. For example, no trials reported any information about participants’ financial dependence or economic support from the partner. Not having enough information about the intimate and the relational context in which the violence occurred limits our understanding of IPV and its treatment. Additionally, the review does not mention maternal, perinatal and postnatal factors, which seem to be pertinent when discussing IPV experienced by women.
Intervention
Classification
The review authors classified the experimental interventions, which were delivered by both healthcare and non-healthcare workers, according to the Cochrane Collaboration Depression, Anxiety and Neurosis Group (CCDAN) classification of psychological interventions and were delivered by both healthcare and non-healthcare workers. They also proposed mechanisms by which each type of therapy might work, which is helpful in enabling the reader to understand how therapy might help victims of IPV.
There were no studies that looked at psychodynamic therapies. This is unfortunate, given that IPV is complex, interpersonal and may affect victims differently depending on previous relationships and experiences. It would be interesting to see studies looking at psychodynamic therapy for victims of IPV in the future. Furthermore, the review mentions that therapy was delivered individually (16 trials, 55%), in a group (11 trials, 38%) or both in a group and individually (2 trials, 7%). However, it does not compare individual and group interventions, which would have added another dimension to the analysis.
Comparator interventions included usual care, no treatment, delayed provision of psychological interventions or waiting-list conditions, and minimal interventions. It can be appreciated that there is significant heterogeneity within the comparator intervention arm of the study, which is a challenge when conducting systematic reviews of this kind. It is unclear from the review how the risks of this heterogeneity were mediated.
Duration
Trials were included if the therapy delivered was of one or more sessions. The minimum duration for the sessions that were included was 30 min. Within those restrictions, treatments of any duration or frequency were included; however, the intensity of intervention was divided into <4 sessions and ≥5 sessions.
National Institute for Health and Care Excellence (NICE) guidelines recommend at least 5–12 sessions of therapy for mild to moderate mental health symptoms and 16–20 sessions for more severe symptoms (NICE 2009). It appears that the inclusion criteria were broadened to the point where a substantial number of participants included in this study received therapy for a duration significantly shorter than is advised in the guidelines. The review authors do not clearly comment on this discrepancy, but it might have been done to include more studies in the review, which would improve its power. However, this may have had repercussions on the quality of the results obtained. It would have been interesting for the duration of therapy to have been divided further into short-, medium- and long-term therapy, which may be more reflective of clinical practice and guidelines.
Excluded interventions
Systemic therapies (e.g. couples’ and family therapy) were excluded on the basis that the focus of the review was solely women who experienced IPV. This might be an overly individualistic viewpoint, as it is inevitable that victims of IPV exist within social contexts. It would have been interesting to have compared systemic with non-systemic approaches to therapy.
Trials whose interventions included advocacy, empowerment or safety planning interventions were also excluded. This is understandable, given the review's focus on psychological therapy. The review authors recommend a more holistic, multidisciplinary approach to treating IPV in the future. It would be interesting to compare a combination of psychological therapies and advocacy interventions with therapy alone in the future.
Intervention results
There was no evidence to suggest better outcomes for one specific psychological therapy over another, and there was no evidence to suggest that outcomes differed depending on who delivered the therapy. This was the case across all outcome measures. A greater improvement in outcome measures was observed when participants were given ≥5 sessions, which is in keeping with current clinical guidelines.
The 19 studies that evaluated depression as a primary outcome found that psychological therapies probably reduced depression at medium- and short-term follow-up. However, the evidence for a reduction in depression at long-term follow-up is less certain. Four trials evaluated self-efficacy. There was no evidence to suggest whether this was affected by therapy or not. All trials provided information about drop-out rates. There was no statistical difference between the drop-out rates from experimental and comparator interventions at any follow-up point.
It was found that psychological therapy may improve general mental health in the short and long term. It was also found that therapy probably reduces anxiety in the short term. No difference between the experimental and comparator groups was evident with regard to other secondary outcomes.
Twelve trials explored the participants’ experience of undergoing psychological therapies. In general, women said that they were satisfied with therapy, glad to have participated in the trials (Hegarty Reference Hegarty, O'Doherty and Taft2013; Cheung Reference Cheung, Deng and Tsao2019) and felt significantly more supported by the experimental intervention than the comparator intervention (Hegarty Reference Hegarty, O'Doherty and Taft2013).
It was reported that uptake of therapy was associated with favourable doctor communication (O'Doherty Reference O'Doherty, Taket and Valpied2016) and the therapist's belief that the partner's behaviour was problematic (Evans Reference Evans, Malpass and Agnew-Davies2018). Women also subsequently reported that therapy resulted in an increase in awareness of abuse, their openness to seeking help and formation of help-seeking cognitions.
There seems to be a discrepancy when comparing outcome measure scores and subjective feedback from participants. It would be interesting for this discrepancy to be explored using qualitative means in the future, which might give us a better understanding of how psychological therapies may be beneficial to people who have suffered IPV.
Out of the 12 trials that reported adverse events or harms, no study could specifically link therapy to adverse events. Only one study measured benefits and harms using a validated measure, which did not detect any significant harms (Valpied Reference Valpied, Hegarty and Brown2020). This might be a more robust way of measuring harm (as opposed to drop-out rate) if it were more widely used.
Although all interventions were pragmatic and tailor-made to suit the needs of the participants, only one trial had a specific trauma focus. It would be interesting to see more research looking at trauma-focused therapy and comparing this with more general forms of therapy.
The effectiveness of cognitive–behavioural interventions in the treatment of depression, anxiety and PTSD in general populations has been shown in several Cochrane reviews (Butler Reference Butler, Chapman and Forman2006; Hunot Reference Hunot, Churchill and Teixeira2007; Bisson Reference Bisson, Roberts and Andrew2013). It appears that this review has not brought to light new evidence regarding the measured outcomes that psychological therapies have on women who experience IPV. Instead, it reinforces what has already been concluded in other studies.
Finally, although this review talks about treating the victims of IPV, it does not discuss ways of reducing or preventing IPV from happening. This stance on discussing the treatment as opposed to the prevention of IPV may echo a culture in which the emphasis of tackling a problem is on dealing with victims in the aftermath of a crime, rather than preventing perpetrators from committing the crime in the first place. Although it is essential to offer victims of IPV (usually women) support and treatment, it is equally, if not more, important to research and implement interventions to prevent IPV, which is most commonly perpetrated by men, from occurring at all.
Conclusions
Owing to the overall low quality of evidence, high heterogeneity and limited results, it is unclear whether this review will translate into meaningful changes in clinical practice. However, the review clearly highlighted the need for more research in the field on an issue that is so prevalent worldwide.
A more holistic approach to evaluating complex trauma is needed to assess how IPV affects women and how interventions may improve different domains in women's lives. This may be achieved by including both qualitative and quantitative outcome measures and by exploring a combination of interventions delivered in a multidisciplinary, trauma-focused approach.
Data availability
Data availability is not applicable to this article as no new data were created or analysed in this study.
Acknowledgements
I thank Dr Riccardo De Giorgi, Dr Daniel Maughan and Dr Dan Joyce for their advice on this project.
Funding
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Declaration of interest
None. The views expressed here are the views of the author and not necessarily those of the National Health Service.
An ICMJE form is in the supplementary material, available online at https://doi.org/10.1192/bja.2021.35.
eLetters
No eLetters have been published for this article.