Alcohol interventions, alcohol policy and intimate partner violence: a systematic review

The World Health Organization (WHO) defines intimate partner violence (IPV) as ‘any behaviour within an intimate relationship that causes physical, psychological or sexual harm’ [1]. WHO recently estimated the global prevalence of physical and/or sexual IPV to be 30% among ever-partnered women [2]. Thus, IPV is a significant global public health and human rights issue that has damaging effects on the health and well-being of women and children, [3] and significant social and economic costs [4].

Alcohol use, especially heavy drinking and drinking large amounts per occasion, is linked to male-to-female partner violence [5]. Across different cultures, violence is more severe when one or both partners (most often the male partner) has been drinking [6]. Meta-analyses suggest that alcohol plays a causal contributing role in aggression generally; [7] however, the extent to which alcohol’s role in IPV is causal, is complex and contested [8]. In addition, across the globe, IPV is a gendered issue, reflecting the unequal power relationships between men and women. Although under experimental conditions alcohol increases aggression in both men and women, the effect is stronger for men [9] and drinking by men has been shown to play a more important role in IPV perpetration than has drinking by women, [10] reflecting the gendered nature of both problem drinking and IPV.

Alcohol is thought to influence aggressive behaviour through detrimental effects on the drinker’s cognitive executive functioning, [11] and problem-solving abilities, [12] narrowing the focus of attention, [13] increasing their willingness to take risks, [14] and increasing concern about personal power among male drinkers [15]. In the context of an intimate couple, when one of the partners has been drinking, he or she will be less able to address conflicts constructively because of (a) the effects of alcohol on cognitive functioning and problem-solving; (b) the drinking partner may have a disproportionate response to a perceived slight, insult or other apparent wrong done by the partner and be less likely to see the partner’s perspective or the situational and environmental factors that may have affected the partner’s behaviour (because of the narrowing of their focus of attention on a specific action of the partner related to their drinking); (c) the drinking partner may engage in highly provocative or aggressive behaviour without thinking about the consequences of his or her actions because of alcohol’s effects on risk-taking; and, (d) for male partners in particular, perceived slights or aggression by the partner may be interpreted as a threat to their masculinity or social identity generally and therefore require an aggressive response to reassert this identity (see [16, 17]). When both partners have been drinking, the role of alcohol may be even greater [18] because of the potential for alcohol to affect the thinking, perceptions and risk-taking of both partners. That is, both partners are more likely to misperceive the other’s behaviour, be less able to resolve the situation without aggression, and be more likely to engage in risky aggression. Social and cultural perceptions of alcohol can also play a role where the acceptance and tolerance of alcohol-related misbehaviour (including aggression), can influence drinkers’ expectations about their behaviour while drinking [19]. This means that, regardless of the effects of alcohol, some people who have been drinking may intentionally engage in aggression or violence toward an intimate partner because they have the expectation that their behaviour will be excused on the basis that they had been drinking at the time.

Although drinking can occur without IPV and IPV without drinking, both are sufficiently linked that the WHO proposed that primary prevention interventions to reduce the harm caused by alcohol could potentially reduce IPV [20]. Further investigation of the effects of alcohol prevention on IPV is important because direct interventions addressing violence against women have been shown to have limited impact [21]. Recognising the multi-dimensional and complex nature of IPV, the WHO recommends an ecological framework for violence prevention wherein factors influence violent behaviour separately and cumulatively at the individual, relationship, community and societal levels (Figure 1) [1, 22]. Although previous reviews of alcohol interventions have focused exclusively on the individual or relationship level [23, 24] (e.g., individual or couple treatment for alcohol dependency), as this model suggests, alcohol interventions relevant to alcohol-related IPV can occur at the community level and the population level, as well as at the individual/relationship level. Community-level interventions are distinguished from population-level interventions in that they apply to a specific community or area, are often developed in response to local issues or concerns and typically involve community stakeholders in their development and management [25]. Population or societal-level interventions, by contrast, are implemented at the population level more broadly (country, state, region) and may be more likely to involve formal mechanisms such as taxation, although similar interventions can occur at both community and population levels.

Because alcohol use is ‘one of the factors most open to intervention and change,’ [26] (p.viii) and broad evidence exists of effective interventions that reduce alcohol consumption and related harms, [27] this review asks the question: Do interventions to reduce alcohol use at the individual, relationship, community and/or population level, reduce intimate partner violence? In this systematic review, we aimed to explore the evidence for the effects of alcohol interventions on IPV, and the extent to which the effects are mediated by changes in alcohol consumption. For this review, we adopted a broad definition of “intervention” to include interventions specifically implemented to reduce alcohol consumption within a target population or community (e.g., alcohol restrictions or addiction treatment) or alcohol policy levers that may affect alcohol consumption indirectly (e.g., alcohol taxes and planning regulations regarding alcohol outlets).

Eleven bibliographic databases were searched systematically for English language peer-reviewed and grey literature studies (such as non peer-reviewed government reports) published between 1 January 1992 and 1 March 2013 including: Medline, CINAHL, EMBASE, PsycINFO, Proquest Central, Cochrane Library, Campbell Collaboration Library, ATSI Health, Drug and Rural Health, and Women’s Studies International. The search strategy combined three concepts of interest: (i) alcohol use, (ii) IPV, and (iii) interventions, using medical subject headings (MeSH), database-specific thesauri search terms, and text-based keywords. Specific terms for alcohol prevention policies were added. Searches were conducted in two stages - the first searches were conducted between 30 October to 1 November 2012 with a second search to update the review conducted on 1 March 2013. A sample search strategy is at (Additional file 1).

A study was included in the review if it investigated whether an intervention or policy to reduce alcohol consumption was directly or indirectly associated with a change in any form of IPV as a primary or secondary outcome. The review included studies of persons 18 years and older and IPV perpetration by either sex within a current heterosexual or homosexual dating, co-habiting or marital relationship, or from a former partner.

The search retrieved 1,810 citations as outlined in the flowchart (Figure 2). IW conducted the initial review of study titles and abstracts with 93 (5%) full text papers retrieved. A further 24 papers were identified from hand searching reference lists and contacting key experts. A total of 117 papers were examined against the initial eligibility criteria. Commentaries, reviews or articles that reported no original data were excluded. Due to questions regarding the integrity of research by Dr. William Fals-Stewart (State of New York v. William Fals-Stewart, 2010), studies in which he was first author or using his data were excluded.

Forty studies (44 papers) met the initial selection criteria. Data for each study was extracted using an agreed standardised template recording details of the study. This included the PICOS criteria - population/sample, intervention, controls or comparisons, outcomes (IPV and alcohol consumption and other measures pertaining to IPV) and study design. We also noted year of study, aims and strengths and limitations. Two additional criteria were applied prior to final study selection. First, we assessed whether the study design and sample size allowed outcomes to be attributed, at least partly, to the intervention or policy being evaluated. Eleven studies met these design criteria (Additional file 2: Table S1). These included: randomized controlled trials, longitudinal studies that measured IPV over multiple time points before and after the intervention, included multiple replications or used an interrupted time series design. Population and community-level ecological studies that used designs able to rule out explanations other than the intervention to account for changes at the community or population level were included, even when they did not include individual measures necessary for making causal attributions regarding the mechanism of change. Excluded studies used cross-sectional and pre-post designs, small pilot samples and had methodological weaknesses that limited interpretation because of potential bias from regression to the mean and other uncontrolled factors [28].

Because only a small number of studies met the design criteria and many did not test the assumption that the effects of the intervention on IPV were mediated by the intervention’s effect on alcohol consumption, we included in our review a second set of ten studies that support the assumption of mediation (Additional file 3: Table S2). That is, while these studies do not meet the design criterion of being able to rule out other explanations for the apparent effects of the intervention, they do provide some evidence that variations in IPV associated with variations in alcohol consumption and that both are associated with the intervention.

Selected studies were categorised by level in the ecological framework – population, community, relationship and individual-level interventions. IW and AT independently reviewed the individual and couple treatment studies for study strengths and limitations (Additional file 2: Table S1), and IW and KG independently reviewed the population and community-level interventions, with agreement reached by consensus.

In the Results section below, we refer to the 40 studies (44 papers) that met the initial selection criteria but discuss in detail only the findings from the 21 studies that met either the design criteria or provided evidence that the effects of interventions on IPV were mediated through alcohol consumption.

The breadth of the review and the heterogeneity in design and quality precluded formal meta-analysis; therefore, findings were synthesized using a critical narrative approach. This involved considering the theoretical basis for how the intervention might work within each level, critically appraising studies for methodological quality and juxtaposing findings within the ecological framework to draw conclusions across the body of evidence [29].

For several decades there has been clear and consistent evidence of an association between alcohol consumption and IPV. There is also evidence that alcohol consumption by one or both partners is associated with increased severity of IPV. Thus, interventions that reduce alcohol consumption may also reduce IPV. These interventions can occur at different levels, from individual clinical treatment to state and federal level taxation. An important contribution of the present review is to bring together the disparate literatures relating to alcohol and IPV to examine the effects of alcohol interventions on IPV from the perspective of all levels of the WHO ecological framework - population, community, relationship and individual levels.

Despite the significance of both alcohol misuse and IPV as public health issues, we found the evidence base for assessing the effectiveness of alcohol interventions on IPV from the last 20 years to be disappointingly small. Our review found few studies that had examined the effect of population-level alcohol measures on IPV despite consistent evidence within the alcohol policy science literature that interventions such as alcohol taxation that reduce demand by increasing the cost of alcohol are effective strategies for reducing alcohol consumption and related harms generally [27, 76]. The small literature available suggested little or weak evidence of an effect of alcohol pricing on IPV, possibly hampered by most studies evaluating very small changes in taxation over time and using a measure of IPV that included both alcohol-related IPV and IPV that was not related to alcohol. While all the reviewed studies were based on the theoretical assumption that changes in price influence consumption, only one study [31] included alcohol consumption measures in their design, and it failed to find a strong enough link to demonstrate that alcohol tax changes can reduce violence against women, with this effect mediated through a reduction in alcohol consumption.

To address the effectiveness of alcohol policy more conclusively, stronger designs are needed that evaluate meaningful pricing changes using appropriate comparison conditions, and that include measures that can distinguish effects on alcohol-related IPV and measures of alcohol consumption for testing mediation. In addition, given that meta-analyses have shown stronger associations of IPV with heavy and binge-drinking patterns of consumption than with other patterns of drinking [5], it is important to assess the extent that those with heavy or binge drinking patterns are sensitive to changes in alcohol price. To enhance the effectiveness of pricing strategies to reduce alcohol-related harm, such policy approaches should be tailored [77] to suit the patterns of consumption of populations highly likely to engage in IPV, for example, younger couples who are most at risk of IPV [78], and binge drinkers who are most prevalent among adolescents and young adults in many countries [27]. From a public health perspective, population approaches have the opportunity to have the greatest impact; thus, further investigation is warranted regarding how such pricing mechanisms can influence IPV where alcohol use is implicated.

At the community-level, we found policy interventions that restricted the availability of alcohol through reduced trading hours were introduced into communities or areas as a response to significant problems with alcohol and violence. The evidence of an impact on IPV was inconclusive based on the one study in Brazil [35] that met design criteria but which measured violence against women generally, not alcohol-related IPV. The remaining studies of alcohol restrictions were implemented in remote Indigenous Australian communities. Although these studies did not meet design criteria, the comprehensive community approaches used in these studies, with multiple interventions directed toward restricting access to alcohol, provide a model for undertaking better controlled evaluation studies in the future to address alcohol-related IPV, which remains a significant problem in many Indigenous communities worldwide [79].

Although a relatively strong body of research has linked alcohol outlet density to violence, [27] research relating specifically to IPV is inconsistent with regard to outlet type. This finding may reflect a complex and variable relationship between outlet density and IPV. In particular, given that IPV is much more likely to occur in the home than in other locations [80], one might expect a stronger link with drinking in the home and therefore IPV would be more strongly associated with off-premises sales. On the other hand, to the extent that IPV is associated with heavier consumption [78, 81] and heavier consumption is more likely to occur in licensed premises, [82–84] a stronger association might be expected for on-premise drinking. Further, whether the link is with on or off-premise drinking may vary by culture. For example, in some cultures, it may be common for couples to drink together at home with the associated increased risk of aggression, while in other cultures it may be common for the male partner to drink large amounts in licensed premises and become violent after coming home or for conflict to arise over his drinking on his return. Thus, despite this inconsistency in findings related to type of outlet, there is sufficient evidence to suggest that the association between alcohol outlet density and IPV is worth further investigation. Mixed methodology research is needed to better understand the relationship between alcohol-related IPV and drinking location and how this might be linked to outlet density, including the extent to which this association is affected by cultural, social and individual factors not just availability [85, 86].

Over half of the studies we reviewed were treatment studies at the relationship and individual levels of the WHO ecological framework. These studies were all conducted in the USA, the majority of these in clinical settings with older, treatment-seeking alcohol dependent men amongst whom IPV is significantly more prevalent than in the general population [57]. None of the studies of alcohol treatment only - using couple or individual approaches - met design criteria. Thus, although these pre-post studies reported some evidence of reduced IPV after individual or couples-based treatment associated with reduced drinking, excessive drinking and related behaviours such as alcohol-related violence can decrease naturally over time (e.g., natural recovery, regression to the mean) [27]. Well-controlled trials with men in various age groups are needed to confirm that alcohol treatment alone can have an impact on IPV. In addition, while the majority of couples-based alcohol treatment studies measured bi-directional violence, partner substance use was a common exclusion criteria thus the extent to which problematic alcohol use by both partner contributes to IPV, remains untested from the current evidence base.

The more recent treatment studies that combined IPV and alcohol interventions used stronger designs though several had methodological limitations. These studies focused on the effects of the addition of an IPV component to addictions treatment and/or addictions component to IPV treatment to examine the combined effect of addressing both alcohol and IPV. These studies were able to demonstrate significant reductions in alcohol consumption and IPV (compared to the control condition) but these effects were not sustained and none of the studies tested the relationships between reductions in drinking and reductions in IPV. The study of dating couples, [63] though unable to show a link between reductions in alcohol use and physical aggression, suggests that brief interventions with younger, non-dependent adult populations are worthy of further study using larger samples because these are low cost interventions and address the population most at risk in many countries. The randomized controlled trial in which a brief alcohol intervention was added to a batterer program [66] is also important despite changes not being sustained because it is the first of its kind illustrating the potential impact of addressing alcohol within the context of addressing IPV perpetration, an area that has received little attention from the IPV prevention field.

This is the first systematic review to examine studies that have addressed the effects of alcohol interventions on IPV at the population, community, relationship and individual levels. These studies included a variety of research methods from across different disciplines. Despite the importance of this comprehensive approach to examining effects of alcohol interventions on IPV, the existing research on which the review is based has some significant limitations. First, the relatively small literature and the heterogeneity of the study designs precluded meta-analysis or even comparison of effects by characteristics of participants or type of intervention. Second, not all studies included alcohol consumption measures that could be used to test the central assumption that alcohol interventions affect IPV by changing alcohol consumption. Third, most population and community studies used police or hospital statistics to measure IPV which represent the most severe cases of IPV, and none of these studies was able to separate estimates of alcohol-related and non alcohol-related IPV. Thus, one reason for the general null findings of the population and community studies we reviewed could be that a substantial proportion of incidents of IPV do not involve alcohol, with this proportion varying considerably for different countries [6]. Finally, the search strategy focussed on English language studies and interventions found were predominately from middle/high income countries and all studies of couple and individual interventions were from the USA.

Employing strategies to reduce problematic alcohol use integrated at all levels of the ecological framework (the population, community, relationship and the individual) and combining alcohol and IPV interventions could have the potential to reduce the incidence of IPV and enhance the safety of victims where alcohol use is intertwined with patterns of IPV perpetration.