Introduction

Children who experience parental domestic violence and abuse (DVA), mental ill-health (MH) and substance misuse (SU) are at greater risk of various adverse outcomes during childhood and later in life (Cleaver et al., 2011; Hughes et al., 2017; Kelley et al., 2015; Rodriguez, 2006). These adverse childhood experiences (ACEs) are prevalent public health problems in the UK (Baker, 2021; Office for National Statistics, 2022a, 2022b) and worldwide (United Nations Office on Drugs and Crime, 2021; WHO, 2018, 2021, 2022). In the UK, approximately 5% of the population have experienced DVA victimisation (Office for National Statistics, 2022b), 17% have experienced MH (Office for Health Improvement and Disparities, 2022), 11% have experienced increasing risk or higher risk alcohol drinking behaviour (NHS Digital, 2024), and 9.2% have used illict drugs in the past year (Office for National Statistics, 2022b). Notably, an estimated 3.6% of UK families are likely to experience these ACEs in combination (Chowdry, 2018). Families experiencing this co-occurrence or clustering of ACEs are likely to be particularly vulnerable because services are ill-equipped to address the complexity of multiple needs (Allen et al., 2024; Kedzior et al., 2024; Mason & O’Rinn, 2014). Increasing our understanding of how best to support these families is essential (HM Government, 2017, 2019; Safe Lives, 2019).

There is some evidence about how to address parental DVA, MH and SU when considered individually. Previous systematic reviews have identified that interventions focused on intensive advocacy (British Columbia Centre of Excellence for Women’s Health, 2013; Rivas et al., 2015) or empowerment (Kiani et al., 2021) can be beneficial for women experiencing DVA, and there is some evidence that interventions involving motivational interviewing or behavioural couples therapy might reduce perpetration of abuse (British Columbia Centre of Excellence for Women’s Health, 2013). Those that include components targeting emotional regulation skills, through addressing thought-emotion-behaviour cycles (e.g., cognitive behavioural therapy) or improving interpersonal relationships (e.g., interpersonal therapy), are well known to be effective in treating MH problems such as depression (Cuijpers et al., 2013; Dennis & Dowswell, 2013) and anxiety (Bandelow et al., 2017). Support that includes intensive case management or involves family members or a parenting component can help reduce parental SU (McGovern et al., 2018, 2021), as well as psychoeducation, social support, developing action plans and goal setting (Gomez et al., 2021).

In contrast, our understanding of what works to support families with multiple, co-occurring needs related to DVA, MH and SU are limited because services and research mostly exclude those with more than one of these difficulties. Recent systematic reviews found few evidence-based interventions addressing such needs. In a large review of reviews, Barrett et al. (2024) identified only two reviews that synthesised evidence on interventions aiming to provide support for children exposed to multiple ACEs including parental DVA, MH and SU (Courtin et al., 2019; Marie-Mitchell & Kostolansky, 2019), and no reviews of interventions aiming to address all three. Our recent systematic review addressed this gap, reviewing family-focused interventions addressing parental DVA, MH and SU. However, we found only one study reporting benefits for two of these outcomes and no interventions reporting benefits for all three (Allen et al., 2022).

In the absence of effective evidence-based interventions, we need to consider what an intervention for parental DVA, MH and SU should look like. Alongside engaging with families, service providers and senior leadership (Allen et al., 2024; Muir et al., 2024), one way to do this is to identify common intervention components across interventions addressing parental DVA, MH and SU, and consider which components might have common impacts across parental DVA, MH and SU (Allen et al., 2022; Barrett et al., 2024). Highlighting key components that are hypothesised to contribute to changes in parental outcomes is essential to identify which aspects of support are most helpful for co-occurring parental DVA, MH and SU. Furthermore, understanding which common components have demonstrated success in addressing parental DVA, MH and SU will help guide the development of future interventions, and provision of current services. To our knowledge, no previous studies have explored this despite an identified need within the literature (Allen et al., 2022; Barrett et al., 2024).

This study aims to identify the ‘best-bet packages’ for parents at risk of, or experiencing, a combination of DVA, MH and SU, with a focus distinct elements of intervention content or design described and hypothesised to contribute to change in key outcomes. This deeper analytical approach will help inform the development of new interventions targeting these clustered outcomes and guide provision of services. It will also highlight any adverse effects of specific components or those that may warrant additional investigation.

Materials and methods

This intervention components analysis (ICA) extends a systematic review we conducted to examine the effectiveness of family-focused interventions for parental DVA, MH and/or SU (Allen et al., 2022). While some of the methods have been previously reported (Allen et al., 2022), we briefly summarize these below and expand to outline the methods unique to the ICA. The protocol for both the ICA and systematic review are available on PROSPERO (CRD42020210350).

Eligibility Criteria

We included studies that: employed a randomised controlled trial (RCT), cluster RCT or pilot RCT; focused on parents/carers at risk of, or experiencing, one or more of parental DVA, MH and SU or children in their care; involved a psychosocial family-focused intervention; and measured one or more of DVA (restricted to victimisation/perpetration (HM Government, 2019), MH (restricted to common mental health disorders including depression, anxiety, post-traumatic stress disorder (PTSD), panic disorder, obsessive compulsive disorder (OCD) and general mental health of parents/caregivers (NICE, 2011) or SU (restricted to alcohol or drug use) of parents or caregivers. We focused on common mental health disorders as these commonly co-occur with DVA and MH and allowed us to ensure the search results were manageable. We excluded studies focused on postnatal depression as this is restricted to a specific time period and disorders such as bipolar depression, psychosis, schizophrenia and borderline personality disorder as these are considered severe mental-health disorders.

Search Strategy

Our search was developed with support from an information specialist (Alison Bethel, University of Exeter) and was conducted by KA. We searched the following databases from inception to March 2020 and updated searches in July 2021: MEDLINE, PsycINFO, Embase, CINAHL, Education Research Information Centre (ERIC), Sociological Abstracts, Applied Social Sciences Index & Abstracts (ASSIA), ProQuest Dissertations and Theses Global, Web of Science Core Collection and Cochrane Central Register of Controlled Trials (CENTRAL). Our search terms fell into five main categories, combined as follows: [1) DVA OR 2) MH OR 3) SU] AND 4) parents/family AND 5) RCTs. Our search included free-text and controlled search terms, adapted for each database. All searches were limited to ‘English Language Only’.

Study Selection

We imported search results to EndNote V9 (The EndNote Team, 2013) and removed duplicates. We used EPPI-Reviewer 4 RCT classifier (EPPI-Reviewer, 2015) to categorise the results based on likelihood of pertaining to an RCT. Studies classified as < 20% likelihood of pertaining to an RCT were title-and-abstract screened by KA alone. Studies classified as ≥ 20% likelihood of pertaining to an RCT were title-and-abstract screened by KA and a second independent reviewer. All full-text screening was conducted by KA and 10% were screened by a second independent reviewer. Agreement levels were > 90% and therefore no further second screening was conducted. Any disagreements were resolved through consultation with a third reviewer (VB).

Data Extraction

TM, AB and KA extracted information from included studies using a standardised form piloted prior to use (Supplementary File 1) including study design, participants, intervention details, outcome measures and intervention/control group raw scores (including N, mean, SD, SE, CI, and p-value) at baseline, post-intervention and the latest possible follow-up time point. Outcome measures included self-report or validated measures of: DVA (including physical, sexual, psychological/emotional, controlling/coercive behaviour and economic abuse (HM Government 2019)) victimisation, perpetration and victimisation/perpetration; depression, anxiety, panic disorder, PTSD, OCD and general MH; and alcohol use, drug use (including specific types of drug use) and general SU. We did not extract information on proxy measures of these outcomes.

We extracted all descriptive text outlining the intervention from the included papers and any relevant, referenced supplementary material. This included text from introductions, methods and discussions. We did not conduct forward or backward citation searches for additional information on the interventions or contact study authors.

Quality Appraisal

TM, AB and KA quality-appraised studies using the Risk of Bias Tool 2 (RoB2) (Higgins et al., 2019; Eldridge et al., 2020). The RoB2 examines potential sources of bias related to the randomisation process, participant identification and recruitment, intervention assignment, handling of missing outcome data, outcome measurement and the reporting of results (Higgins et al., 2019; Eldridge et al., 2020).

Data Analysis

We used Sutcliffe et al.’s (2015) approach to undertake our ICA. This involved developing an understanding of the differences between interventions, and using this information to assess differences in outcomes both within and across conditions (Sutcliffe et al., 2015).

Developing an Understanding of the Differences between Interventions

TM and KA open-coded the same random selection of 30 studies to generate an initial intervention-components framework. In our ICA, we define ‘intervention component’ as a distinct element of intervention content or design that was described and hypothesised to contribute to change in key outcomes—namely, parental experiences of DVA, MH and/or SU. These may pertain to intervention content, delivery processes, relational or contextual adaptations, or support mechanisms. As such, we focused on coding intervention components that told us something about how the intervention might work (e.g., intervention strategies, frameworks or overarching theories) rather than intervention descriptors (e.g., length of the intervention, number of sessions etc.), which were extracted elsewhere. This initial framework was then used by TM, AB, AG and KA to continue coding other included studies while allowing for flexibility to generate new codes as needed. The team met regularly to discuss coding, agree additional codes and adapt existing codes based on the literature. Once all coding was completed and the framework finalised, coding of the included studies was checked by AB and KA to ensure that any new codes had been applied consistently across studies. All coding was conducted in NVivo and, at the end of the process, an Excel table was produced to highlight the intervention components present across each study. We descriptively analysed the components commonly present across DVA, MH, SU and clustered (targeting two or more of DVA, MH and/or SU) interventions.

Assessing Differences in Outcomes

We conducted random-effects meta-regressions (Borenstein et al., 2007) with robust variance estimation (Tanner-Smith & Tipton, 2014; Tipton, 2015) to examine the effect of family-focused interventions and intervention components on parental DVA, MH and SU at post-intervention and later follow-up. First, we used meta-regressions to examine differences in outcomes at post-intervention and later follow-up based on the identified intervention components, examining patterns across outcomes to highlight which components have common beneficial or harmful effects across outcomes. Second, we examined the effect of single-issue interventions (targeting one of DVA, MH or SU) and clustered interventions (targeting two or more of DVA, MH and SU) on parental DVA, MH and SU outcomes at post-intervention and later follow-up. Here, we grouped dyadic (targeting two of DVA, MH and SU) and clustered interventions (targeting all three of DVA, MH and SU) together for simplicity. We did not examine how contextual factors – such as fidelity, participant engagement or setting – may shape the effectiveness of the identified components, nor potential synergistic or interactive effects between components. However, all trials were subject to an RCT (typically with high fidelity). While contextual considerations are important, they were beyond the scope of the current analysis.

Patient and Public Engagement and Involvement

Patient and public engagement and involvement (PPIE) with commissioners and service providers shaped the design and focus of our ICA, informing our focus on family-focused interventions and our decision to include prevention and treatment interventions to ensure our findings would be relevant for providers and commissioners (Allen et al., 2022). We also held two PPIE workshops with an established Lived Experience Advisory Group to help inform our interpretation and presentation of the ICA findings. This group included five people with experience of violence and abuse, MH and SU (https://www.vamhn.co.uk/lived-experience-advisory-group.html). The first workshop focused on familiarising participants with the ICA and reviewing the preliminary intervention components table. The second workshop involved discussing the key ICA findings in greater depth. Both workshops lasted two hours and were structured around guided prompts to elicit the groups reflections on: the naming and descriptions of components, common/uncommon components, key findings and implications. We also consulted five service providers about the results of the study in a one-off workshop which involved reflecting on the findings and key points raised by the Lived Experience Advisory Group workshops. We have acknowledged where these groups have contributed to our presentation/understanding of the findings throughout.

Results

Figure 1 reports the number of records identified, screened and included in the ICA. Our searches retrieved 88,550 unique references. Of these, 1657 were screened at full text. Full-text screening resulted in 169 records included in the ICA: 12 records represented studies that measured parental DVA, 73 measured parental MH, 39 measured parental SU 27 measured two of parental DVA, MH and/or SU and 18 measured all three of parental DVA, MH and/or SU.

Of the 169 records included in the ICA, 152 unique studies were identified (some records referred to the same study). These studies reported on 164 interventions which were included in our analyses (some studies reported on two or more interventions). Using the RoB2, the majority of studies were rated as either ‘some concerns’ or ‘high risk’ of bias (see Supplementary File 2). From this point onwards, we refer solely to the 164 interventions included in our analysis. We refer to studies measuring DVA only as ‘DVA interventions’, MH only as ‘MH interventions’, SU only as ‘SU interventions’, two of DVA, MH and SU as ‘dyadic interventions’, and all three of DVA, MH and SU as ‘clustered interventions’.

Characteristics of Included Interventions

All interventions aimed to improve the outcomes of parents and families broadly, but interventions primarily targeted either mothers, fathers, the caregiver, the child or the whole family. Of the 164 interventions reported, 71% focused on the mother, 16% on the family, 7% on the caregiver, 4% on the father and 2% on the child. In terms of level of prevention, 43% of the interventions intended to treat participants, 35% had multiple levels of prevention, 12% were secondary prevention and 9% were tertiary prevention.

Of all the interventions, 9% targeted DVA, 43% targeted MH, 24% targeted SU, 16% were dyadic and 8% were clustered. Although the review included interventions tested across 23 countries, interventions were most commonly undertaken in the U.S. (65%), the U.K (6%) and Australia (5%). Of the U.S. interventions, 34% targeted MH, 28% targeted SU and another 29% targeted dyadic or clustered risks. Only 9% of U.S. interventions primarily targeted DVA. A similar distribution was observed for U.K and Australian interventions, with most targeting MH, followed by SU and DVA. Interventions were also undertaken in Bosnia and Herzegovina (0.6%), Canada (4.3%), China (0.6%), Columbia (1.2%), Germany (0.6%), Greece (1.2%), Guatemala (0.6%), Iran (1.8%), Jamacia (0.6%), Jordan (0.6%), Nepal (0.6%), Netherlands (2.4%), New Zealand (0.6%), Pakistan (2.4%), Peru (0.6%), Romania (0.6%), Rwanda (1.2%), South Africa (1.8%), Sweden (0.6%) and Thailand (0.6%). Full details of intervention characteristics are provided in Supplementary File 3.

Intervention Components Analysis

Our initial intervention-components framework involved 99 codes (Supplementary File 4). Of these, 12 codes fell under the broad category of intervention ‘activities’ and 87 codes under ‘pathways to change’ i.e., how interventions were attempting to reduce or prevent parental DVA, MH and/or SU. Continued coding of intervention descriptions and regular team discussions refined the initial framework into 26 codes. This revised framework was applied to the remaining intervention descriptions and further refined following extensive discussion to produce a final framework of 20 codes (Table 1). Four of these contained additional sub-codes describing additional intervention details. Supplementary File 5 describes the most common components identified.

Addressing Additional Needs

Around a fifth of interventions (18%) included a component addressing additional needs, such as those concerning economic, housing or employment needs. These included employment skills training, parental life-skills education (Donohue et al., 2014; Jacobs et al., 2016) and directing families to services supporting basic needs, such as food pantries (Guo et al., 2016). This component did not relate to stabilising an individual prior to the intervention but was related to components which addressed additional needs during the intervention. Addressing additional needs was common in clustered interventions (46%).

Changing Cognitive Processes

Changing cognitive processes involved intervention elements aimed at changing maladaptive thought processes and perceptions using approaches such as cognitive behavioural therapy or cognitive restructuring. This component was common in dyadic (48%) and MH interventions (37%), with a third of interventions including this component (32%).

Community Referrals

Around one fifth of interventions (18%) included a community referral component which involved linking participants with local organisations providing support for DVA, MH or SU. Parents were referred to the relevant service dependent upon the detected problem (Bartu et al., 2006). For example, parents or caregivers were referred for mental-health services when deemed appropriate (Boyd et al., 2017). Home visiting interventions commonly included this component, signposting parents to support (Duggan et al., 1999, 2004, 2007; Jack et al., 2019; LeCroy & Krysik, 2011). Community referrals were commonly present in DVA (36%) and clustered interventions (31%).

Cultural and Community Adaptations

Several interventions (17%) included a component modifying the intervention to reflect different perspectives, needs and practices of specific cultural, ethnic geographic and/or religious communities. For example, intervention materials or content, such as psychoeducation modules, were modified to reflect the intervention population (Jones et al., 2014) or specific practices and customs (Walkup et al., 2009). This component was most prevalent in MH interventions (28%).

Developing a Safe Intervention Space

Intervention Environment

Several interventions (17%) described ways in which providers made considered decisions about the intervention environment to develop a safe and comfortable space for participants. For some this involved in-home delivery of the intervention (Beeber et al., 2013; Black et al., 2003) or group sessions involving those with similar experiences (Rosenblum et al., 2017). This component was most common in DVA interventions (50%).

Relationship between Participant and Provider

A third of interventions (34%) described ways in which they had supported the development of a strong and trusting relationship between participant and provider. This included supporting the development of a therapeutic relationship (Graham-Bermann et al., 2019), positive partnerships between family-support workers and participants (Fergusson et al., 2006), or relationships between paraprofessional or peer mentors and participants (Cupples et al., 2011). This component was most commonly present in clustered interventions (54%) followed by dyadic (52%) and DVA interventions (43%).

Developing Healthy Relationships

Family Relationships

Several interventions (23%) included elements aiming to improve relationships within the family within or outside the home. This included identifying the priorities and goals of the family and introducing concepts related to family relationships (Barnhart et al., 2020) or improving family functioning (Duggan et al., 2007; Ondersma et al., 2017). This component was most common in clustered interventions (54%).

Parent-Child Relationships

Almost half of interventions (44%) included activities aiming to improve parent-child relationships. These included modelling effect parent-child interaction (Duggan et al., 1999), targeting parenting behaviours related to parent-child bonding (Silovsky et al., 2011) or using restorative parenting to rebuild father-child relationships (Stover, 2015). This component was most common in MH interventions (61%) followed by clustered (54%) and dyadic interventions (48%).

Parent Relationships

A few interventions (13%) included activities to improve interparental relationships. For example, interventions provided support for parents to help them address relationship problems (Guo et al., 2016), support to improve interparental communication for co-parenting (Stover, 2015) or broader communication skills (Barlow et al., 2019), as well as psychoeducation to reduce conflict, manage stress and co-parent (Wadsworth et al., 2011). Parent relationships was common in DVA interventions (36%).

Empowerment

Empowerment described any component focused on supporting an individual to recognise their strengths and capabilities, and take control of their actions and lives (Jones et al., 2014). The empowerment component was most common in DVA interventions (36%) with models such as the empowerment process model (Clark et al., 2018) being used and brief empowerment interventions being added to existing support (Jack et al., 2019).

Enhancing Motivation To Change

Around a quarter of interventions (26%) involved techniques to encourage and promote motivation to change behaviour, with the aim of reducing or eliminating maladaptive behaviours. Osterman (Osterman, 2009) provides an example of this component within their motivational interviewing intervention: “The importance and confidence of the woman for changing drinking behaviours was assessed, followed by further exploration of the importance of change for the woman and the building of her confidence for change. During this process, information regarding strategies for behaviour change was exchanged while resistance for behaviour change was reduced” (p. 47). This component was most common in SU interventions (51%).

Enhancing Skills

Communication Skills

Around a fifth of interventions (21%) supported participants in developing communication skills, aiming to enable them to talk, share and express their ideas and emotions (e.g., Cicchetti et al., 2000; Feder et al., 2018; Slesnick & Erdem, 2013). This component was most common in DVA interventions (36%).

Conflict Resolution

A few interventions (10%) involved a component aiming to help participants understand and undergo the process of finding peaceful solutions to conflict (e.g., Barnhart et al., 2020; Clark et al., 2018; Feder et al., 2018). As above, this component was most common in DVA interventions (50%).

Coping Skills

Several interventions (18%) supported participants to learn how to cope with stressful situations. For example, McWhirter et al. (McWhirter, 2011) talk about discussing and working on coping strategies: “the women were presented with information regarding adaptive and nonadaptive coping strategies. Participants were then encouraged to identify a nonadaptive coping strategy to work on changing as part of the treatment process” (p. 2464). This component was most common in DVA (21%), MH (25%) and dyadic (19%) interventions.

Decision-Making

Only a few interventions included a decision-making skills component (7%) aiming to support participants to gather information and assess potential resolutions, to identify a solution to their problems (e.g., Sharps et al., 2016). This was most common in DVA (14%) and MH (10%) interventions.

Emotional Regulation

Around a fifth of interventions (21%) included a component helping participants to develop healthy strategies to regulate negative emotions (e.g., Barlow et al., 2019). This was most common in DVA (21%), MH (30%) and dyadic (22%) interventions.

Goal Setting

A fifth of interventions (20%) included activities enabling participants to take steps to set and achieve desired goals (e.g., Catalano et al., 1997; Verduyn et al., 2003). This was most common in SU interventions (41%).

Parenting Skills

Over half of interventions (54%) supported parents to develop skills to provide their child with physical care, psychological support and protection. This involved addressing parenting needs (e.g., LeCroy & Krysik, 2011), supporting parent-child interactions through role modelling or promoting parental empathy (e.g., Duggan et al., 2007). Over two-thirds of all clustered interventions (69%) included this component. Parenting skills was also a common component in dyadic (63%) and MH interventions (66%).

Problem-Solving Skills

Over a quarter of interventions (28%) included activities aiming to enable participants to handle unexpected situation and challenges, and identify solutions. In some instances, this involved role-modelling problem-solving (e.g., Duggan et al., 1999) or practising problem-solving (e.g., Slesnick & Zhang, 2016). Problem-solving skills was most common in MH (38%) and clustered (38%) interventions.

Stress Management

The stress-management component was coded for all interventions that involved supporting the development of methods to overcome feelings of stress. Around a fifth of interventions included this component (22%). Often, interventions focused on supporting parents to manage the stresses of parenting and family life (e.g., McWhirter, 2011). This component most often featured in DVA (29%) and MH (30%) interventions.

System Navigation

A few interventions (12%) included a component supporting participants to understand and navigate health and social care services and/or community resources. An example of this can be found in Jack et al. (2019): “System navigation involves identifying, referring, and actively facilitating participant access to external domestic violence, legal, housing, or other health or social care services” (p. 1578). This component was most common in DVA (36%), dyadic (19%) and clustered (15%) interventions.

Financial Skills

Six interventions (4%) had a component focused on financial skills, which involved supporting parents to develop skills to manage finances, solve financial problems and understand what financial services and support are available to them. This component was most common in DVA interventions (14%).

Facilitators to Engagement

Interventions were coded as including a component facilitating engagement if they referred to an implementation process or measure which encouraged participant engagement for example via transportation assistance (Mullins et al., 2004; Olds et al., 2010), childcare (Volpicelli et al., 2000) or flexibility in intervention delivery site (Handley et al., 2017). Just over a quarter of interventions (27%) included this component, being most common in MH (31%) and SU interventions (33%).

Increasing Knowledge

DVA

These interventions aimed to increase knowledge and understanding of different types of DVA, the violence cycle, attributes of healthy/unhealthy relationships and safety in intimate relationships (e.g., Feder et al., 2018; Jack et al., 2019; Sharps et al., 2016). This component was most common in DVA interventions (71%).

Health Promotion

Interventions were coded thus if they aimed to increase knowledge and understanding of health in general, including the delivery of health-related information through home visits (e.g., Cupples et al., 2011), manuals or specific modules (e.g., Black et al., 2003). This component was most prevalent in SU interventions (38%).

MH

Interventions were coded under this component if they provided education or information on MH, including signs and symptoms of mental illness or skills to promote good mental health. Education on specific mental illnesses such as depression, anxiety, post-natal depression and “baby blues” were common (e.g., Zlotnick et al., 2011). This was most common in MH interventions (39%).

SU

Interventions were coded thus if they aimed to increase knowledge and understanding of addiction, the effects of SU on parents and children (including behaviour effects of substances) and of relapse prevention and abstinence (e.g., Lander et al., 2015) This component was most common in SU interventions (64%).

Managing Symptoms and Relapse Prevention

Over a third of interventions (37%) included a component focused on managing symptoms of MH (i.e. anxiety or depression) or SU disorders, or techniques for relapse prevention. This could involve educational modules on remaining abstinent (e.g., Barlow et al., 2019), managing relapses or practicing refusal skills (e.g., Slesnick & Erdem, 2013). This was most common in SU interventions (44%).

Post-Intervention and Long-Term Support

Around a quarter of interventions (26%) provided long-term support to participants including booster sessions to maintain positive effects. Some interventions offered family intervention support throughout the pre-school years (Fergusson et al., 2013) while others offered discrete booster sessions (Zlotnick et al., 2019). Post-intervention and long-term support was most common in clustered (69%) and DVA (43%) interventions.

Promoting Health-Related Behaviours and Development

Over a third of interventions (37%) targeted promoting health-related behaviours for parents and children (child health and development and maternal health including pre- and post-natal health). This included promoting immunisation, vaccinations, safer sex practices and medical check-ups (e.g., Bartu et al., 2006; Donohue et al., 2014; Duggan et al., 2007). This was most common in clustered (62%) and SU (51%) interventions.

Reflective Practice and Active Feedback

Just under a third of interventions (31%) included a reflective practice and active feedback component encouraging participants to reflect on previous intervention sessions or behaviours to identify hidden patterns and thought processes, as well as understand the participant’s motivation for joining the intervention. For example, one intervention involved aimed to enhance a “mother’s capacity for mentalization or reflective functioning in the parenting role” (p. 618, Suchman et al., 2017), while others provided opportunities for participants to reflect on previous sessions to reinforce content (Zlotnick et al., 2011). This component was most often identified in SU (36%) and dyadic (30%) interventions.

Reinforcing Positive Behaviours

A few interventions (7%) aimed to reinforce healthy/positive behaviours through praise or incentives. These included monetary vouchers for abstinence (Jones et al., 2011) or attending treatment (Morgenstern et al., 2006) or participants receiving verbal recognition, stickers on graphs or certificates for negative urine screening results (Jones et al., 2014). This component was most often present in SU interventions (13%).

Sense of Self, Self-Esteem, and Self-Efficacy

A fifth of interventions (20%) aimed to develop healthy self-concept, self-efficacy and/or self-esteem as an individual or parent. This component aimed to increase women’s self-efficacy and reduce self-blame (Clark et al., 2018), supporting improvements in mother’s knowledge of child development and child rearing practices (Baker-Henningham et al., 2005). This component was most common in dyadic (37%) and DVA (29%) interventions.

Social Support

Just under a third of interventions (30%) included social support to help participants through the intervention. This included group therapy and community-based interventions. Clark et al. (2018) provides an example of social support within the Mom’s Empowerment Program (MEP): “The MEP is community-based in that it was developed in partnership with community agencies and domestic violence shelters… It aims to increase women’s self-efficacy and reduce their self-blame through participation in group discussions and providing social support” (p. 260, Clark et al., 2018). This component was most common in dyadic (30%) and DVA (29%) interventions.

Stabilisation

Several interventions (15%) included a component aiming to provide greater stability to participants lives, including ensuring families have appropriate housing and living conditions or safety planning (Graham-Bermann et al., 2019). This component was often associated with ‘crisis management’ and supporting families in times of need (e.g., Duggan et al., 2007; Morgenstern et al., 2006; Slesnick & Erdem, 2013; Suchman et al., 2010). Stabilisation was most prevalent in clustered (38%) and DVA (36%) interventions.

Tailoring and Personalising

Many interventions (40%) modified and/or tailored intervention components to the needs of individuals, families, groups or communities participating in the intervention. Interventions could be tailored to consider a mother and child “needs, strengths, and circumstances” (p. 920, Sullivan et al., 2002) as well as their individual experiences in relation to DVA, MH, and SU. This component was most common in SU interventions (54%) followed by MH (41%) and clustered (31%) interventions.

Effectiveness of Intervention Components

Robust variance meta-regressions did not reveal any effective components, that is no single component was significantly associated with improved DVA/MH/SU outcomes. However, there were several individual components that were associated with adverse DVA/MH/SU outcomes. At post-intervention, addressing additional needs (coefficient=-0.26, 95% CIs=-0.43 to -0.08, p = 0.008), post-intervention and long-term support (coefficient-0.22, 95% CIs=-0.43 to -0.004, p = 0.046), and stabilisation (coefficient=-0.22, 95% CIs=-0.39 to -0.06, p = 0.0112) components were associated with an adverse effect on MH, and empowerment had a negative impact on SU (coefficient=-0.18, 95% CIs=-0.35 to -0.02, p = 0.0278). At follow-up, developing healthy relationships components were associated with an adverse effect on MH (coefficient=-0.23, 95% CIs=-0.45 to -0.01, p = 0.039) and empowerment was associated with an adverse effect on DVA (coefficient=-0.15, 95% CIs=-0.19 to -0.12, p < 0.001) and SU (coefficient=-0.30, 95% CIs=-0.37 to -0.22, p < 0.001). Full results are tabulated in Supplementary File 6. Caution is needed interpreting these results given the heterogeneity across study designs and populations.

Effectiveness of Intervention Type

We also explored whether intervention type might be associated with intervention effectiveness. Taking each type of intervention in turn: five (36%) DVA interventions were coded as having a beneficial effect on DVA outcomes at post-intervention and one (7%) had a beneficial effect on follow-up outcomes; 18 (25%) MH interventions were associated with beneficial effects on at least one MH outcome at post-intervention and seven (10%) of these went on to have longer-term effects on follow-up outcomes; eight (21%) SU interventions demonstrated effects on post-intervention SU outcomes of which two (5%) had beneficial effects at follow-up, and one (3%) intervention was coded as having mixed impacts on SU outcomes; eight (29%) dyadic interventions were coded as having impacts on at least one outcome and only one (4%) had impacts on two target outcomes at post-intervention (MH and SU); and four (31%) clustered interventions resulted in positive impacts on outcomes, spread across domains (one on DVA, two on MH and one on SU). No clustered interventions had effects across multiple or all outcomes.

Robust variance meta-regression analyses examining the efficacy of interventions targeting two or more issues (i.e., dyadic and clustered interventions) versus single-issue interventions (i.e., DVA, MH or SU interventions) are reported in Table 2. These suggest that interventions targeting two or more issues are less likely to be helpful for MH and SU outcomes at post-intervention (co-efficient=-0.23, 95% CIs =-0.42 to -0.03, p = 0.02; co-efficient=-0.22, 95% CIs=-0.37 to -0.06, p = 0.008) and less likely to be helpful for MH at follow-up (co-efficient=-0.22, 95% CIs=-0.43 to -0.003, p < 0.05) as compared to single-issue interventions.

Patient and Public Involvement and Engagement

Stakeholders contributed to our presentation of the findings and our interpretation of the results, providing us with recommendations for our discussion and future work. The Lived Experience Advisory Group suggested ordering our intervention components in terms of their perceived importance/priority level, using a traffic light system (see Table 1). While they argued all the components were important, they emphasised that certain components must be in place before others (e.g., ensuring the intervention starts with ‘developing a safe intervention space’ before moving onto components like ‘managing symptoms and relapse prevention’). The Lived Experience Advisory Group identified five key service features as particularly critical to effective support: (1) meeting the family where they are at; (2) personalising the support offered to the families; (3) providing a safe space for the family; (4) ensuring long-term support; and (5) adapting provision so that it is culturally appropriate. Service providers agreed that these components were essential, emphasising the need for personalisation, addressing underlying needs, and providing a safe space. While both groups were pleased to see interventions attempting to address families’ multiple needs, they were disappointed with the limited evidence for their effectiveness. The Lived Experience Advisory Group suggested that factors such as the timing and quality of intervention components may contribute to this lack of impact – factors that are often overlooked in current service provision.

Discussion

To our knowledge, this is the first ICA of common components across DVA, MH and SU interventions, and the first to explore the components of clustered interventions in this field. These findings are critical for policy and service development. Our systematic review (Allen et al., 2022) found no current ‘best-bet’ interventions to support families experiencing these multiple adversities and signalled a need for interventions to address co-occurring risk. We have built on this review to examine the nature and frequency of different change components included in interventions across domains, to explore the potential for integrated services.

The ICA identified 20 components common to interventions aiming to support families with parental DVA, MH and/or SU. There were no components that were unique to singular focus or clustered interventions, suggesting that different interventions are targeting shared or common risks across domains. Certain components were more frequently employed depending on the need. For example, empowerment was typical in DVA interventions, while interventions focused on changing cognitive processes and managing symptoms and relapse prevention featured regularly in MH and SU interventions, respectively. These components might be considered core or necessary building blocks for services for families with these particular issues, but they are not exclusive to any one concern. In contrast, clustered interventions were more likely to include components addressing additional needs and offering post-intervention/long-term support, which is unsurprising given the complex profiles of families experiencing multiple adversities (Early Intervention Foundation, 2021; HM Government, 2022a).

The effectiveness data are concerning for this sector. While reviews of interventions for these ACEs as singular risks have identified effective responses (e.g., Barlow et al., 2015; McGovern et al., 2018; Nylen et al., 2006; Rizo et al., 2011), there are currently no effective interventions for families experiencing a combination of DVA, MH and SU. While there is no evidence currently to suggest that single-focus interventions are ineffective (or harmful) for families with complex/multiple needs, it is likely that families experiencing DVA in the context of SU, for example, may require a different response to those experiencing DVA alone. In addition, the study did not identify any ‘magic bullets’, that is individual effective components. The data suggest that interventions should respond on multiple levels and that the combination of components may be a critical consideration. Indeed, these data suggest that current approaches to clustered risk have the potential to undermine the effects of some components on MH and SU, although this may in part be due to the severity of cases included in these studies. An alternative conclusion is that provision may be best targeted at a primary intervention focus, while not excluding individuals with comorbid conditions – a common approach in specialist sectors.

Our engagement with those with lived experience has been critical to considering the next steps for service development in the context of these disappointing findings. Stakeholders asserted both the need for interventions addressing multiple needs and the inadequacy of current approaches. This engagement also revealed that it is the quality of delivery and timing of interventions, rather than the content per se, that requires attention. Our experts agreed on the value of the identified intervention components but provided examples of how these have been poorly implemented in current services. They emphasised the importance of five key service features: meeting the family where they are at; personalising the support offered to the families; providing a safe space for the family; ensuring long-term support; and adapting provision so that it is culturally appropriate. Service providers agreed on this importance of these five components. Some of these suggestions (e.g., recognition that support needs to be personalised for each family), chime with UK Government recommendations (HM Government, 2022c).

This work indicates the need for further (re)design and development of services to support families with multiple needs, with a clear ethos of working with families at the intersection of needs, and considering support for parents and children. Moreover, there is a dearth of interventions addressing social and structural determinants, such as poverty, lack of education and discrimination, or wider systemic issues in the aetiology and impact of these adverse experiences. Despite agreement that these factors are central to prevention (e.g., Marmot et al., 2020; The Lancet Public Health, 2021), almost all the programmes we identified focused on individual behaviour change or relational support. More could be done to develop existing components. For example, only one identified intervention (Skar et al., 2021) aimed to influence wider community values or interactions, and two studies (Silverstein et al., 2017; Skar et al., 2021) referred to the potential for system change but these did not feature as formal change components in the interventions. There is scope for integrated care approaches (Baxter et al., 2018; Ross & Greenberg, 2020) to inform the design of new interventions for families with complex needs.

Limitations and Future Research

This ICA explored the components of interventions identified in our earlier systematic review (Allen et al., 2022). The scope of that review on family-focused interventions prevented conclusions about the nature and effectiveness of intervention components targeting DVA, MH or SU in adults alone (i.e., not as parents), which may be informative for service development in this sector. In the UK and elsewhere, MH and SU services are commonly commissioned and delivered separately for adults and children/young people. Despite calls for more integration (HM Government, 2022b; Stolper et al., 2024), the evidence base largely reflects this siloed provision. Future research could explore whether the components identified in family-focused interventions are mirrored in adult-only programmes and the relative effectiveness of those components.

Our findings are limited by the quality of reporting of intervention descriptions in evaluation papers. Improved reporting standards introduced over the past decade, due to guidelines like the TIDieR checklist (Hoffmann et al., 2014), mean that components included in more recently evaluated interventions are likely to be over-represented in our ICA. Furthermore, poor reporting of intervention programme theories and informing theoretical frameworks limit the extent to which shared mechanisms of change across domains can be inferred. Adopting a realist perspective to future systematic reviews of clustered interventions may provide more useful information about the drivers of change in particular contexts (see Bonell et al., 2023).

Most ICAs, ours included, are limited to assessing the presence/absence of components in interventions and relating this to effectiveness. This type of analysis does not account for variation in how components are delivered, the quality of delivery, or key contextual factors (e.g., practitioner skill, participant engagement or readiness). As a result, the ICA may obscure complex mechanisms of change that rely on synergistic or additive combinations of multiple components. The impact of interventions, and their constituent components, is dependent on myriad implementation and contextual factors (Bauer & Kirchner, 2020). While this study has demonstrated that the presence of any singular component was not related to effects on DVA, MH or SU outcomes, it is plausible that positive impacts for families with complex needs may be achieved via attention to those contextual factors and/or a combination of two or more components. Indeed, the lack of identifiable ‘effective’ components is perhaps unsurprising given that impact is more likely to arise from the interaction between multiple components – a point raised by others (EIF, 2022). Future research could undertake a qualitative comparative analysis to explore this hypothesis (Sutcliffe et al., 2022).

Conclusions

This study is the first to explore common components of family-focused interventions targeting DVA, MH and/or SU. While it has been possible to identify shared components across target domains, suggesting the potential to offer integrated services or packages of support to families with complex/multiple needs, findings on the impact of these common elements are not encouraging. Policymakers and service providers developing responses to support families will require more rigorous information about the implementation of these interventions and the contextual factors influencing their impact. In addition, there is a critical need for service design innovation to address the social determinants of DVA, MH and SU in local communities, as well as the systemic issues, such as funding/resourcing models, workforce education and awareness, communication and data sharing, which can hinder effective delivery and sustain service siloes.