Case management interventions seeking to counter radicalisation to violence and related forms of violence: A systematic review

Part I – Countering radicalisation to violence

No eligible studies spoke to Objective 1.

Forty-seven studies related to Objective 2. Research on Objective 2a (n = 7 studies) focused on approaches. Research on Objective 2b focused on implementation factors pertaining to stages of the case management process (n = 43); and implementation factors (n = 41) and moderators (n = 28) relevant to the full process.

Part II – Countering other forms of violence

Eight reviews were included. Five examined the effectiveness of case management approaches (n = 2) and tools (n = 3) (Objective 3); two examined how tools were implemented (Objective 4a); and seven considered implementation factors and moderators (Objective 4b).

Are case management tools and approaches effective in countering radicalisation to violence?

It is not possible to draw conclusions about effectiveness as no eligible studies were identified.

Are case management tools and approaches implemented as intended?

Four studies concluded that the assumptions underpinning three interventions were sound and aligned with academic research. Four studies reported mixed results as to whether three interventions were implemented in line with their internal logic. Two studies highlighted weaknesses in programme logic, including misalignment between activities and intended outcomes.

What explains how tools and approaches are implemented?

Are case management tools and approaches effective at countering interpersonal and collective forms of violence?

The effectiveness of case management in countering other forms of violence remains unclear. Two reviews examining the effectiveness of interventions did not find conclusive evidence that they effectively countered violence. Three reviews on risk assessment tools (n = 2) and polygraphs (n = 1) reported mixed results. However, use of these tools alone would not be expected to directly reduce violence as any positive impact would be indirect.

How are case management tools and approaches implemented in the context of countering collective and interpersonal forms of violence?

Evidence focused almost entirely on risk assessment tools (n = 5). Two reviews found that risk management is not always directly informed by structured risk assessment. The extent to which practitioners use risk assessment tools to inform case planning is informed by their willingness to take risk assessments into account when making decisions, and their ability to offer services that can effectively target needs or risks. Feedback on the perceived utility of these tools was therefore found to be mixed. Whilst feedback on the use of polygraphs was positive, this feedback was drawn from a limited evidence base (n = 1).

Facilitators of risk assessment include the ability to adapt tools to local needs; training and guidance; opportunities to pilot tools; professional ownership; positive relationships with clients; and multi-disciplinary working. Barriers included uncertainty about the utility of tools; insufficient room for clinical judgement; the perceived complexity and resource intensity of assessment; lack of experience and perceived self-efficacy; subjective interpretations of risk; and uncertainty around translating assessments into practical action. Expertise, training, and time spent with clients facilitated the implementation of mentoring programmes.

Are case management tools and approaches used to counter other forms of violence transferable to counter-radicalisation work?

The research in Part II was considered transferable to counter-radicalisation interventions. Risk assessment tools and mentoring approaches are already widely used within counter-radicalisation interventions. The utility of using polygraphs has also been considered, however evidence for their effectiveness is insufficient to recommend their implementation. Evidence relating to the use of Multi-Systemic Therapy (MST) for countering radicalisation to violence  was not identified in the literature included in Part I, however its adherence to socio-ecological models of violence prevention suggests it is potentially transferrable.

Identification

The case management process begins once potential clients in need of support have been identified. In secondary interventions, this identification process typically involves identifying particular patterns of risks and needs considered likely to indicate an elevated risk of involvement in terrorism. In tertiary interventions, an individual's involvement in terrorism or violent extremism has typically already been recognised, for example through a terrorism conviction.

A range of methods may be used to refer clients into interventions. Costa et al. (2021) set out a typology of three different mechanisms: (1) Active; (2) Passive; and (3) Mediated. The active approach involves clients being referred to interventions by front-line professionals from different sectors (e.g., police, education, healthcare, etc.). This is seen in the UK's approach to secondary interventions (Pettinger, 2020a, 2020b; Weeks, 2018), although similar methods are used in secondary and tertiary interventions in Australia (e.g., Cherney & Belton, 2021a, 2021b; Harris-Hogan, 2020) and the Netherlands (AEF, 2018; Eijkman & Roodnat, 2017). This type of active approach can include mechanisms through which family members, friends, or community members refer individuals they are concerned about into programmes, or raise concerns about potential radicalisation to relevant front-line professionals or agencies (e.g., Thomas et al., 2020). Secondary and tertiary interventions may also use a more passive approach, whereby individuals self-refer into interventions (e.g., Christensen, 2015). And finally, whilst less common, secondary interventions might also incorporate a mediated approach, where families can provide individuals with information about a relevant programme, or even physically take them to an intervention provider (e.g., Costa et al., 2021).

A significant proportion of individuals identified as being potentially in need of counter-radicalisation support through the active or mediated approach never formally enter the case management process. For example, statistics from the UK's Channel programme indicate that 23% of the 6406 referrals made in the year ending 31st March 2022 were formally assessed by a multi-agency ‘Channel panel’ (HM Government, 2023). As a result, research that focused on the methods by which members of the public and frontline professionals identify perceived indicators of risk, and decide when to refer into interventions was excluded from this review. Instead, we define the client identification stage as the period during which counter-radicalisation practitioners make preliminary assessments relating to the potential eligibility of individual clients, and determine whether cases should progress to the client assessment stage. In secondary interventions these assessments may relate to screening out obviously inappropriate, misguided, or ‘spurious’ referrals that do not warrant further action (Lewis, 2021), as well as identifying referrals relating to individuals who are already the focus of a criminal investigation or other, harder forms of intervention (Cherney & Belton, 2021b).

Client assessment

The identification process is typically followed by a formal assessment process. In secondary interventions, the assessment aims to differentiate between those who pose a radicalisation risk, those who do not, and others who might have needs unconnected to terrorism and who could benefit from signposting to alternative forms of support, such as mental health provision. Those ineligible for secondary interventions may fall below the threshold for radicalisation risk (Pettinger, 2020a, 2020b), and may be referred to other agencies or forms of support to address other needs identified through the assessment, or be considered too high risk and in need of harder forms of intervention (Cherney & Belton, 2021b). In tertiary interventions, the assessment process is primarily used to identify client-specific intervention goals and the relevant forms of support needed to support disengagement from violent extremism, and reduce risk (Cherney & Belton, 2021b). This aspect of case management is examined in detail below.

Eligibility screening can be conducted in different ways and involve different actors. This may include those who are involved in delivering the intervention – on the basis this has the potential to enable them to ‘build rapport and assess their clients’ (van der Heide & Schuurman, 2018, p. 205) – as well as those who will not go on to work with the individual (AEF, 2018; Christensen, 2015). Assessments can be conducted in person, by one or two individuals (AEF, 2018; Christensen, 2015), or a specific team to assess individual cases before their discussion at a multi-agency case conference (Inspectorate of Justice & Security, 2017, p. 24; also Eijkman & Roodnat, 2017; Vandaele et al., 2022a). However, it is more common for eligibility to be determined by multi-agency case conferences before intervention staff engage with the client.

Case conferences draw on information from different partners and assess this information to determine whether clients are eligible for counter-radicalisation support; identify their specific needs; and tailor intervention plans. They are a common feature of interventions across the world, including in the UK (Pettinger, 2020a, 2020b; Weeks, 2018); the Netherlands (AEF, 2018; Hardyns et al., 2022); Canada (Thompson & Leroux, 2022); and Australia (Cherney, 2022). The way case conferences operate varies across different programmes (Vandaele et al., 2022a). For example, some interventions may discuss deidentified cases when assessing individuals (Thompson & Leroux, 2022), in contrast to many interventions, which do not appear to de-identify potential clients. In some contexts, individuals may be aware they are being discussed at conferences, but in others, they may not (Vandaele et al., 2022a).

Multi-agency interventions often appoint a dedicated case manager, or in some interventions, multiple case managers (Cherney et al., 2022; van der Heide & Schuurman, 2018). Case managers are typically assigned overall responsibility for managing clients throughout the case management process. They help develop the intervention plan and identify relevant services and organisations able to meet the client's needs, as well as monitoring their progress (AEF, 2018). Some interventions will appoint a dedicated mentor (or mentors) who deliver support to clients alongside facilitating other aspects of the case management process, such as client assessment and case planning (e.g., Christensen, 2015; Fisher et al., 2020).

Risk and needs assessment tools can help inform the assessment process (Lloyd & Dean, 2015), although these tools are not always used in this way (Barracosa & March, 2022; Costa et al., 2021). These tools can be general or specialised for use in terrorism cases. Specialised tools include the Vulnerability Assessment Framework (VAF) (Pettinger, 2020a, 2020b); the Violent Extremism Risk Assessment, Version 2 Revised (VERA-2R) (Raets, 2022; van der Heide & Schuurman, 2018); and the Terrorist Radicalization Assessment Protocol (TRAP-18) (Corner & Pyszora, 2022; Raets, 2022). The majority of specialist tools use a Structured Professional Judgement (SPJ) approach, which ‘relies on the discretion of the assessor whilst providing a basic, empirically informed structure to help guide their decision-making’ (Copeland & Marsden, 2020, p. 7). Specialised tools that use a more structured, actuarial approach are also in use internationally (e.g., Raets, 2022), as are general, non-specialist tools, such as the Level of Service Inventory Revised (LSI-R) (Cherney, 2021; Inspector of Custodial Services NSW, 2018). A range of bespoke specialist tools may also be used. This includes tools that are specific to individual interventions, or that are used only in specific regions such as the ‘Radix’ tool developed by one Belgian municipality (Costa et al., 2021; Raets, 2022), and tools that are used to assess specific cohorts, such as youth (Barracosa & March, 2022). All of these tools seek to accurately assess the risk an individual poses and inform decisions about whether and what kind of support they should receive.

Case planning

Case management interventions are operationalised through the delivery of a case plan. The development of the case plan is informed by the assessment process. Some interventions will use risk and needs assessment tools to inform intervention planning (Lloyd, 2019). The basic approach to developing client-specific case plans typically involves multi-agency partners discussing the support needs of each client and designing a tailored intervention plan that is designed to target each client's needs, and deliver specific intervention goals (Cherney & Belton, 2021b), through provision of a tailored set of services (Ellis et al., 2022).

Case managers or dedicated mentors often play an important role in developing tailored intervention plans (e.g., AEF, 2018; Inspectorate of Justice & Security, 2017) and monitoring the individual's progress to determine if the case plan is adequately meeting their needs (Harris-Hogan, 2020). However, the specific approach used may vary across interventions. For example, in some settings, the client and coach/mentor will co-design an action plan, which will then be presented to, and assessed by, a multi-agency case conference (AEF, 2018).

In many cases, the case plan structures the process by which services are delivered, for example, sequencing activities in ways which take account of the individual's learning style or needs (Cherney & Belton, 2021a). Case plans developed at the outset of the case management process are unlikely to remain static and may be reviewed and updated regularly to account for an individual's changing circumstances and needs (Disley et al., 2016; Thompson & Leroux, 2022), and respond to any emerging challenges (Cherney, 2022; Vandaele et al., 2022a).

Delivery and implementation

Case management interventions involve the delivery of tailored intervention plans which deploy services considered likely to meet the individual's needs and reduce their risks (Cherney & Belton, 2021a). A diversity of services are typically available including education; employment; lifestyle; psychological help; family provision; and religious or ideological advice, as well as more specific types of support such as music programmes; childcare services; speech therapy; or referral to mental health services (Cherney & Belton, 2021a; Raets, 2022). In some settings, such as the Netherlands, the range of services is much broader, with up to 50 different types of intervention available to clients (AEF, 2018), which are under constant review and expansion (Eijkman & Roodnat, 2017).

The quality and delivery of interventions are supported in a variety of ways. Some intervention programmes – such as secondary and tertiary interventions in the UK – have developed mentor selection and accreditation processes for intervention providers (Pettinger, 2020a, 2020b; Thornton & Bouhana, 2019; Weeks, 2018). In other settings, competency is developed through an ongoing process. For example, in EXIT Sweden, former clients may become ‘client-coaches’, who support clients whilst continuing to work on their own rehabilitation (Christensen, 2015). Whereas the Team TER reintegration intervention specifically set out to develop a group of specialists from the Dutch Probation Service who would apply their knowledge of working with other types of offenders to this programme, whilst gaining specialist expertise working with terrorist and violent extremist offenders (Schuurman & Bakker, 2016). A similar emphasis on ‘learning by doing’ is used in other interventions in the Netherlands (AEF, 2018; Eijkman & Roodnat, 2017). The aim of these methods is to develop a body of skilled professionals able to deliver and support the case management process.

Monitoring and evaluation

A range of approaches are available to monitor and evaluate an individual's progress through the case management process. These include the use of multi-agency case conferences which meet to review ongoing cases (e.g., Thompson & Leroux, 2022) and formalised assessment tools. For example, several interventions in Australia use information drawn from the Radar tool and from other sources such as case reviews to assess and track client progress (Cherney, 2022; Cherney & Belton, 2021b; Harris-Hogan, 2020). Assessments are typically made against their original intervention plan and can be undertaken independently of the client, or as is the case in the Netherlands, collaboratively, so the coach and client assess progress against their action plan, and identify areas that might need additional attention (e.g., AEF, 2018).

As part of a range of different kinds of qualitative data, case files and case notes are used to assess progress and can be used to record a mix of factual information, such as recording that a client attended a session, as well as more subjective client feedback and assessments (Cherney & Belton, 2021a, p. 15; 2021b, p. 630). This process can involve the collation of a range of data collected at different points during the case management process, from different sources, and using different methods, including:

[…] qualitative inputs relating to client background information, risk and needs assessments, client intervention goals, dated observations about intervention staff/service provider engagements with clients, service provider and family members correspondence relating to client appointments, activities and participation, psychologist/counsellor feedback, nature and reason for police contact, court documents and forms of open-source data.

Transition and exit

The decision of when an individual exits a case management intervention is largely determined by their circumstances and differs between secondary and tertiary interventions. There are few specific timeframes for secondary interventions (AEF, 2018; Costa et al., 2021), whereas when someone has been convicted, their involvement in an intervention is likely to be shaped by the length of their sentence or conditions of their parole. For example, Cherney (2020) notes that PRISM staff may begin the process of engaging clients 2 years before their earliest possible release date. When the individual's sentence has been served, they will typically exit the programme, although in some cases there are opportunities for individuals to continue to receive support after this point should they wish to, or the parole service can request assistance from the original intervention provider when managing a client in the community (Cherney, 2018; Marsden, 2017).

Where the intervention ends before a prisoner's sentence has been completed, or when someone is engaged in a secondary intervention, they are typically assessed to determine if their risk has reduced, and their needs have been met in a way which is consistent with their individualised intervention goals. For example, Khalil et al. (2019) note that exit from the Serendi rehabilitation centre in Somalia is dependent on the individual meeting ‘agreed and personalised rehabilitation objectives relating to family connections, education, vocational training, security issues in the locations of reintegration, and so on’ (p. 4). However, some criteria are more generic. For example, Vandaele et al. (2022a) note that cases in Germany, the Netherlands, and Belgium are closed ‘if no new events of concern occurred’ (p. 71).

The level of aftercare differs across contexts (Costa et al., 2021). Some interventions, such as Forsa in the Netherlands, Serendi in Somalia, and PRISM in Australia, provide ongoing support (AEF, 2018; Cherney, 2020; Khalil et al., 2019). Others including Team TER (van der Heide & Schuurman, 2018), and EXIT Sweden (Christensen, 2015) do not, although intervention staff may choose to stay in contact with former clients when no formal aftercare is offered. The approach to aftercare also varies, some contact former clients periodically or when information on them needs to be updated; others refer clients to other organisations/partners; and some have a structured follow-up strategy (Costa et al., 2021). These processes aim to provide ongoing support for the individual's reintegration, and to monitor their progress outside the formal case management process.

Criteria for considering studies for Part I

Search methods for identification of studies

The search process for English language material involved six stages that sought to identify relevant academic and grey literature research. 1.

Identification and piloting of search terms.

Data collection and analysis

Deviations from the protocol

The review made four deviations from the protocol. Two related to searching for and screening the literature in languages other than English, and two related to the methods of analysis used.

The strategy used to search for studies in languages other than English had to be adapted. We had intended to use the translated search terms to search each of the platforms listed in the protocol (reproduced in Table 1). However, a number of these platforms did not have the functionality to search non-English characters (see Supporting Information: Appendix I for specific databases). We were therefore only able to search these databases using the English search terms, applying filters to filter out English studies. For consistency, we conducted two sets of searches in those databases that did allow us to use the translated search terms: (1) a search of the translated search terms in line with the approach outlined in the protocol; and (2) a separate search of the English search terms, with language filters applied.

A large number of the records returned by the searches in languages other than English had an English title and/or abstract. To remain consistent with the English language screening process, rather than ask the language experts to review these in the first instance, the primary reviewer (JL) conducted the initial relevancy screening of any English titles/abstracts returned by these searches. All titles and abstracts that were in languages other than English were then screened by the relevant language specialists.

Data analysis relating to Question 2b (on implementation) was structured around two elements of an intervention's theory of change as outlined in Section 1.2.6 of the protocol (Lewis et al., 2023); this covered implementation factors, and moderators. The structure used in the final review aligned with the protocol, however we refined the approach to interpreting implementation factors by examining two aspects of implementation: facilitators and barriers. There were three reasons for this: (1) it allowed for a more consistent, fine-grained approach to synthesising the findings; (2) it is a distinction informed by our data extraction tool (see Supporting Information: Appendix IV); and (3) it was informed by the combination of deductive and inductive approaches to analysing the data we proposed to take in the protocol (see Section 4.8 of the protocol), which saw facilitators and barriers emerge as important codes from our analysis and which were then used to structure the results.

Finally, we did not review the transferability of findings to other contexts as originally outlined in Section 3.9 of the protocol, for two reasons. First, as we were unable to identify any eligible studies relating to Objective 1 (on effectiveness), it was not possible to assess whether effective interventions were transferable to different contexts. Second, the findings included in Part I were drawn from a wide range of countries and settings, which meant that the findings of Part I themselves reflected how different tools and approaches might be used in different contexts and served to highlight how contextual factors might influence the implementation (and thus the potential transferability) of tools.

Description of studies

Results of the search

The results of the search and screening process are set out in Figure 2. The initial English database searches identified 49,410 records. An additional 14,838 records were identified from the searches in Languages other than English, which included research in French, German, Norwegian, Swedish and Danish, and Russian. These were combined with material identified through the search of grey literature sources, hand searches of relevant journals and consultation with experts (n = 3,018). A total of 67,266 records made up the initial corpus. After de-duplication, 45,658 records went forward for title and abstract screening using the Covidence platform.

OPEN IN VIEWER

Figure 2

PRISMA flow diagram.

The title and abstract screening process led to the removal of 45,296 records, leaving 362 which went forward for full text review in Covidence. The full text of 14 records were unavailable in institutional repositories. We assessed the title and abstract of these records using the full-text screening criteria and these were subsequently excluded. The remaining records were reviewed by two members of the research team and any conflicts adjudicated through discussion with a third team member.

From a total of 348 records that underwent full-text screening, 50 went forward for inclusion in the review. No records met the inclusion criteria for Objective 1 on the effectiveness of case management interventions. All of the records therefore related to whether case management interventions were implemented as intended (Objective 2a), and/or discussed what influenced how interventions were implemented, focusing on factors which facilitated, generated barriers, or acted as moderators to implementation processes (Objective 2b). A further three eligible records were identified through forward and backward citation searches of the included records, leaving a total of 53 eligible records that were included in Part I. These records contained data relating to 47 different empirical studies. In total, seven studies contained data relevant to Objective 2a, and 47 contained data relevant to Objective 2b.

Risk of bias in included studies

The CASP tool described in Section 4.1.3 was used to assess elements of the 45 qualitative or mixed methods studies. The quality appraisals for each study, and each domain, are shown in Table 10. Whilst the authors of the CASP tool do not recommend a scoring system, it is worth noting that only three of the included studies were assessed as having no limitations across the ten different domains covered in CASP (Cherney, 2018; Jukschat et al., 2020; Orban, 2019). We only included studies that did not have a critical weaknesses across seven of the ten domains, and which therefore scored positively on a majority of the domains assessed by the tool. Only studies meeting the below criteria were included:

Well-defined research question

Table 10 highlights that the answers to the questions relating to the other three domains were ‘No’ or ‘Can't Tell’ for several studies. Whilst these domains are no less important than the seven listed above, we did not exclude studies based on these domains in isolation as such answers did not necessarily highlight a critical weakness. It is not unusual for published and non-published studies to fail to specifically reference ethical issues or the relationship between researcher and participants, and so we did not exclude studies based on these domains. Studies that had obviously failed to consider these issues would have been excluded, including studies that were clearly conducted in an unethical way, or which failed to discuss an obvious power imbalance between researchers and participants. However, these issues did not appear to be present in the included studies. Similarly, whilst studies that clearly lacked a rigorous approach to data analysis would have been excluded, studies that did not specify a specific approach to data analysis were included where the authors presented evidence in support of their findings and conclusions.

The quantitative components of the eight mixed methods studies and the two studies that only presented a quantitative analysis of qualitative data were assessed using the EPHPP tool described in Section 4.1.3. However, as shown in Table 11, domains relating to confounders, blinding, and withdrawals/dropouts were not relevant to assessing the specific quantitative research designs included in the review (i.e., one-time surveys and quantitative analysis of programme documentation), and so were not used to assess study quality.

Whilst it was not possible to assess for publication bias, a comparison of the themes identified in published and non-published studies highlighted that the results of both were comparable.

Synthesis of results

The following discussion of the results is made up of three parts: the first reviews the research on case management approaches, addressing Objectives 1 (on effectiveness) and 2 (on implementation); the second covers the same objectives for case management tools split according to the different stages of the case management process; whilst the third discusses the research relating to both objectives as it relates to case management as an overarching process.

Case management approaches

The analysis of approaches is split into four sections: Identifying case management approaches; Assessing the effectiveness of case management approaches; Examining the implementation of case management approaches; and Identifying implementation factors and moderators that influence how approaches are delivered.

Case management tools

The analysis of tools is split into three sections: Assessing the effectiveness of case management tools; Examining the implementation of case management tools; and Identifying implementation factors and moderators that influence how approaches are delivered.

The case management process

Summary of main results

Part I of the review had two objectives: to understand the effectiveness of tools and approaches used in case management interventions seeking to counter radicalisation to violence, and to examine those factors and moderators which impact their implementation. One of the main findings from Part I is that very little is known about the effectiveness of tools and approaches used in this context (Objective 1). Whilst a number of innovative methodologies for assessing the effectiveness of case management were identified (e.g., Cherney & Belton, 2021a, 2021b; Cherney, 2022), no quasi-experimental or experimental evaluations of case management tools and/or approaches were identified. As a result, we are unable to offer any conclusions relating to Objective 1 on effectiveness.

However, we identified a relatively large, and growing, body of research (n = 47 studies) relating to Objective 2 on implementation. The key findings relating to the facilitators, barriers, and moderators impacting the implementation of tools and approaches are summarised below.

Another important finding is the utility of the case management model in supporting the analysis of secondary and tertiary interventions. Although the language of case management has only recently begun to be applied to counter-radicalisation research and practice, it offers a valuable way of organising the array of tools and approaches that are used in this context.

Implementing case management approaches

We had hoped to understand whether particular approaches, such as the RNR model or strengths-based approaches, were more or less effective at informing case management interventions. However, it was not possible to develop a typology of different approaches based on the information provided in the eligible studies, and in turn to assess the implementation of approaches in the way conceptualised in the background section above, or the original protocol.

However, it was possible to assess the specific theories of change or programme logics (i.e., the approach) underpinning individual interventions. Seven studies assessed the assumptions underpinning an intervention and/or examined whether a programme was being delivered in line with a clearly defined programme logic, theory of change, or approach (Objective 2a). In general, these studies reported positive, albeit somewhat mixed results. Four studies that used academic research as a benchmark concluded that the assumptions underpinning interventions were sound, but caveated this observation by noting that the academic literature was limited. Four studies that assessed whether interventions were being implemented in ways which aligned with their own underlying logic reported that programmes generally adhered to their underlying logic, but that specific, practical challenges might undermine their ability to do so.

Two additional studies highlighted the challenges created by weaknesses in an intervention's theory of change. These included a lack of clarity over goals informed by differences between the ambitions of policymakers and practitioners' perceptions about what was feasible; and a disconnect between an intervention's activities and its intended outcomes.

Caution is needed when seeking to transfer the findings of these studies to other contexts, as they focus on specific interventions delivered in specific settings. The findings cannot therefore be assumed to be relevant to all case management interventions. However, the broader finding pointing to the importance of interventions having a clearly defined theory of change is likely to be relevant to all interventions, regardless of the context in which they are delivered.

Implementing case management tools

The review identified a reasonable amount of research on those factors that shaped the implementation of case management programmes and associated tools (Objective 2b). Although the strength of the evidence may not be robust, a body of primarily qualitative research has developed that provides insights into what facilitates and inhibits the implementation of case management tools and approaches, and has begun to identify moderators or contextual factors that inform this process. Relevant facilitators and barriers are set out in Table 12 and are examined in more detail in the discussion below.

Stage 1: Client identification

Two barriers were identified in two studies that examined the processes by which individuals are identified and referred to case management interventions. Both barriers speak to the difficulties of finding appropriate actors to undertake this work. First, the difficulties posed by working with external actors on the basis of short-term contracts, which limited their long-term commitment to the intervention, and by institutional mistrust of those external to the state agencies that managed this work. Second, the challenges of using police officers to fulfil this function due to the differing commitments of care and control that the police are subject to.

Stage 2: Client assessment

Twenty-six studies looked at client assessment processes, with an emphasis on multi-agency client assessment (n = 14 studies) and risk and needs analysis tools (n = 12), alongside a smaller number of studies that looked at screening tools (n = 3). Factors which facilitated screening tools included a shared, clearly defined, often standardised set of criteria to determine who is eligible for an intervention, and the perceived benefits of having more than one person involved in the screening process. Overall, these processes were believed to reduce the potential of inappropriate and/or subjective referral decisions, and identify those most at risk of radicalisation or terrorist recidivism, and/or best able to benefit from the intervention.

Fourteen studies identified factors which facilitated or generated barriers to multi-agency collaboration and client assessment. All of these studies considered multi-agency working to be a valuable tool to support client assessment and reduce the potential for mistakes. Factors which supported multi-agency collaboration included effective communication processes; the availability of risk assessment tools; and trusted relationships between those representing different agencies. These relationships were enabled by strong leadership, a clear mandate, and regular meetings which allowed positive, reciprocal relationships to develop over time.

Identified barriers to effective multi-agency working included inefficient or limited information sharing between partners, particularly when working with the police or security agencies, and inconsistency and subjectivity in assessment processes. Power differentials between stakeholders were also considered able to undermine relationships between partners, and lead to authority bias where those with greatest power are assumed to be correct. More practical barriers included excessive bureaucracy, and time and scheduling difficulties. Finally, unconscious bias was considered to potentially lead to certain identity groups – particularly Muslims – being disproportionately considered at greater risk of radicalisation, whilst confirmation bias may see stakeholders foreground evidence that supports pre-existing assumptions.

Risk and needs assessments (RNA) were examined by twelve studies. Factors which facilitated the use of RNA tools included the use of tools designed specifically for countering radicalisation to violence, as standard measures were considered less able to adequately capture factors relevant to this context. Having knowledgeable and experienced practitioners helped to facilitate assessments, as well as having the potential to identify disguised compliance. Training was therefore considered valuable, as was the use of multiple assessors, and the provision of formal and informal support for assessors to help mitigate some of the pressures associated with this role. Finally, an awareness of differing levels of knowledge and experience across stakeholders was considered important in facilitating assessments.

Barriers to client assessment processes included the inconsistent use of RNA tools, and an associated uncertainty as to how valuable they were. Conceptual barriers were also identified, such as a lack of clarity over how risk factors were defined, and uncertainty as to whether the most appropriate factors were covered in RNA tools, with particular concerns about the neglect of protective factors. A lack of guidance around how to interpret patterns of risk factors as well as limited support to help practitioners understand how RNA might be able to support case planning and case management were also considered potential implementation barriers.

There were some areas of debate in the literature on RNA tools. Several studies suggested that subjectivity in how clients were assessed acted as a barrier to effective risk assessment, whilst one study presented the views of some practitioners who suggested that it was useful to allow for more subjective processes that took greater account of professional judgement.

Stage 3: Case planning

Case planning received less attention in the literature: a total of five studies looked at tools used to develop individualised case plans for clients. Three of these examined intervention planning tools, and found that it was important to ensure the outcome of the client assessment stage fed into the case planning process. The second tool used to inform case planning was case conferences. Many of the same themes identified in the work on multi-agency working in the assessment stage were identified in research on case planning, including the benefits of trust, motivation, expertise, appropriate information sharing arrangements, structured regular meetings informed by a clear mandate and overseen by a neutral chair, as well as equality amongst stakeholders, good interpersonal relationships developed over time, and a positive working environment. Conversely, the lack of commonly agreed goals, a shortage of resources, difficulties sharing information, hierarchical struggles and overly bureaucratic processes all had the potential to act as barriers to case conferences used to inform case planning.

Stage 4: Delivery and implementation

A total of twenty-eight studies looked at the delivery stage, including research that examined the tailoring of intervention services and goals (n = 19 studies); practitioner characteristics and approaches (n = 20) and practitioner supervision and quality assurance processes (n = 13).

Factors facilitating the delivery and implementation of case management interventions include structured methods of recruiting mentors to ensure the most appropriate individuals are selected for this work, and experience and expertise of working with radicalised individuals. The importance of adequately supporting practitioners who deliver this work was identified across the different tools that helped deliver interventions. This support included organisational, managerial and, in particular, peer support to help manage the emotional toll of working with what can be a challenging population. In contrast, poor working relationships with management, ineffective management structures, poor communication between stakeholders, and the absence of a competency framework were identified as barriers to delivery.

Stage 5: Monitoring and evaluation

Sixteen studies considered monitoring and evaluation tools, of which nine looked at client assessment tools; seven focused on case file and case note data; five on case conferences; and five on other, less structured forms of qualitative data.

Among those factors with the potential to facilitate monitoring and evaluation were the availability and use of structured assessment tools able to help monitor change, inform evaluations, and support the delivery of interventions. The detection of false compliance may be aided by triangulating different sources of monitoring data. Case notes and case files able to capture multiple forms of data, including practitioner feedback, information on the quality of the client's engagement, and their behaviour and attitudes can also support delivery. As can input from multiple practitioners which helps reduce the potential of biased reports from individual stakeholders. In addition to those benefits associated with their use during the assessment and case planning stages as outlined above, case conferences can support monitoring and evaluation by enabling plausibility checking over client progress, and supporting case reviews and light-touch internal audits.

Potential barriers to monitoring and evaluation include subjective assessment processes, biases, inconsistency in the quality and content of information recorded, uneven use of tools to monitor client progress, and failing to complete risk assessments before an intervention has started which means there is no baseline against which to assess change. The absence of longitudinal monitoring limits assessments about long-term outcomes such as recidivism. A lack of agreed measures of client progress can create a barrier to monitoring and evaluation processes, as can the absence of systematic approaches and the use of subjective criteria to interpret progress which in turn can lead to disagreements between practitioners over levels of risk and progress.

Stage 6: Transition and exit

Ten studies examined the transition and exit aspect of interventions and identified two factors which facilitated this process: inter-agency coordination, and continuity of support during the pre-release and post-release periods. Potential barriers included fears over closing cases too early and concern over the consequences of making the wrong decision, as well as the challenges associated with ending the relationship between practitioners and clients smoothly. Finally, in the post-exit context, difficulties monitoring clients after their release from prison generated practical barriers, as did a lack of data on the position of former clients.

Implementation factors across the case management process

The review also identified factors that affected the full case management process. These factors included multi-agency working; risk-oriented logics; public and policy pressure; the intensity of intervention work; resourcing; staff expertise; voluntary and mandatory interventions; and broader legislation. This aspect of the review also considered the moderators, or contextual factors, that were relevant to case management interventions including delivery context; local context; standalone interventions; and client challenges.

Multi-agency working. All of the interventions covered in the review incorporated some form of multi-agency working, and 34 studies examined how this operated in practice. In some cases, multi-agency working structured one or a number of stages of the case management process. In others, they were central to the entire process of managing interventions. Twenty-three of the studies that looked at multi-agency working suggested it helped to facilitate interventions.

Implementation factors considered to facilitate multi-agency working included visibility, so that interventions were known to internal and external stakeholders and were considered an appropriate source of expertise and support; and efficient information sharing between partners, which was supported by the development of codified rules, secure data transfer systems, trust between stakeholders, and knowledge about the rules associated with sharing information.

Having a shared understanding of clearly defined intervention goals and of the intended outcomes of multi-agency working were considered facilitators, and their absence a barrier. Reviewing and clarifying outcomes and goals through logic models was one means of addressing this issue. Differing organisational cultures, where practitioners are focused on different types of goals – for example public protection and rehabilitation – had the potential to create barriers to effective multi-agency working. Positive working relationships facilitated case management processes and were supported by training around the working practices and mandates of other organisations, as well as efforts to develop trust between practitioners from different agencies.

Barriers to implementing multi-agency working were comparable to those identified at individual stages of the case management process, most notably issues associated with information sharing, where the less transparent processes used by the police and security agencies inhibited information sharing, as well as confidentiality rules which had the same effect, but typically involved those from healthcare and social work. Having large numbers of actors involved in multi-agency meetings was identified as a possible barrier to sharing relevant information, whilst ensuring the right partners with the relevant expertise and capacity to address specific case management functions were involved in multi-agency processes facilitated this process. Power differentials and hierarchical struggles, often between the police and other stakeholders, again had the potential to create barriers to implementation, as could a lack of clarity over the responsibilities and jurisdictional boundaries of different partners.

Risk-oriented logics. Fifteen studies examining tertiary interventions discussed the use of risk-oriented approaches. A third of these studies identified potential barriers generated by a preoccupation with short-term efforts to manage and control risk, noting how this focus on risk can come into tension with longer-term rehabilitative goals. Four studies discussed risk-oriented approaches to secondary interventions, and again highlighted how risk logics can contribute to risk aversion which can bump up against rehabilitative goals, particularly where professionals from non-security fields were concerned. Client-centred approaches using strengths-based and desistance-informed approaches had the potential to mitigate these implementation barriers.

Public and political pressure. The particular public attention paid to counter-radicalisation interventions was identified as a potential barrier to implementation by ten studies. Public scrutiny has the potential to make practitioners adopt a more risk averse attitude, whilst the high-profile nature of these interventions could negatively impact practitioners' willingness and confidence to engage in this work. Public debates around the risk of disguised compliance were also considered potentially disempowering. Risk aversion may lead to interventions diverging from their underlying logic, for example, by not reducing supervision where levels of risk are seen to dip because of concerns over the repercussions should an individual reoffend.

Political and media scrutiny has the potential to create a barrier to constructive relationships between practitioners and clients who may be sceptical of efforts to work with them, particularly if the client feels that decisions are being made due to political considerations rather than because of a genuine assessment of their progress. The stigma associated with a terrorism conviction has the potential to undermine reintegration and rehabilitation options.

Resourcing. Seventeen studies drew attention to the barriers that a shortage of resources, time, staff or adequate support for staff can create. Resource-related barriers were described as particularly challenging in contexts with less well-developed CVE infrastructures; for newer interventions; and where practitioners' roles in interventions were less well embedded in existing systems. The potential for overwork and stress were also identified as barriers.

Time pressures can lead to inadequate preparation and guidance, and a poor understanding of intervention goals. A reduction or lack of sustainable funding was also identified as a challenge, particularly where this meant practitioners were employed on short-term contracts and did not have clarity over their expected caseloads or salary. These issues were described as a particular challenge given the perception that radicalised clients need more supervision and support.

Staff expertise. Staff expertise was considered an important facilitator by twenty-three studies. The diversity of knowledge held in multi-disciplinary teams was considered helpful, as was transferable expertise gained from working with other kinds of clients, alongside experience of engaging with at risk or radicalised individuals. Relevant language skills were identified as a facilitator (and their absence a barrier), in enabling practitioners to make informed risk assessments and develop a positive relationship with the client.

Sixteen studies identified training as an important facilitator, including specialised training on countering radicalisation to violence and knowledge of other relevant techniques such as motivational interviewing, multi-agency collaboration, or broader skills such as counselling. Barriers associated with staffing included challenges recruiting and retaining staff with the necessary expertise, and a lack of specialist knowledge, in particular in relation to ideology. Drawing on external experts has the potential to address this barrier.

Voluntary and mandatory interventions. The barriers and opportunities of voluntary and mandated interventions were considered by eleven studies. This research suggested practitioners had a preference for voluntary approaches as these were considered better able to elicit the motivation needed to change. However, practitioners delivering voluntary interventions also described barriers, including an inability to compel individuals to participate in programmes, and difficulties encouraging people to engage with interventions once enroled.

Broader legislation. Eight studies considered the impact of broader counter-terrorism legislation. This research drew attention to the barriers counter-terrorism powers can generate when trying to pursue rehabilitative aims, including where clients are subject to harder forms of counter-terrorism intervention which may negatively impact their motivation to engage with efforts to support their reintegration. Where clients are not adequately prepared, the imposition, but also the lifting of sanctions can create a barrier to rehabilitation efforts. Although in some cases these sanctions were considered able to facilitate case management processes, particularly if used selectively and as part of a well-coordinated multi-agency approach.

Moderators across the case management process

Moderators, or those contextual conditions that influence how case management interventions are implemented, included the delivery context, local context, whether an intervention was standalone, and client challenges that impacted their ability to engage with the intervention.

Delivery context. The role delivery contexts played in case management interventions was considered by eleven studies. The specific features of correctional settings had the potential to generate barriers to delivering interventions in this context, including the impact controls placed on individuals can have on the ability of practitioners to develop constructive relationships with clients, and the ability of prisoners to take part in rehabilitative activities.

The negative impact that prison conditions can have on prisoners can also act as a potential barrier. Providing expedited access to support services can help mitigate this challenge. Whilst those on probation do not face these challenges, the greater freedom individuals have can make monitoring and assessment harder. Institutional contexts able to facilitate interventions are characterised by good conditions and clearly communicated values and principles which guide decision making. Finally, operating in conflict affected contexts where violent extremist groups were still active has the potential to undermine efforts to promote long-term disengagement.

Local context. The influence of local contexts was considered by ten studies which described the benefits of tailoring interventions to local conditions or the CVE infrastructure in a given country in ways which take account of differing levels of resources, expertise and risk. Being sensitive to the relative availability of employment opportunities or the services available in local neighbourhoods was considered helpful, as was the presence of an effective local coordinator. Tensions between different regional authorities and the quality of cooperation between agencies had the potential to create barriers to implementation.

Standalone interventions. Four studies considered the role of standalone interventions, as distinct from those integrated into existing organisational structures offering broader support. Delivering an intervention through an organisation that was already well-established in a local area was considered able to facilitate delivery. In the absence of this sort of organisation, standalone interventions may overcome this barrier by becoming known in the area, nurturing a good reputation, and employing a positive public relations approach. However, a lack of CVE-specific support can act as a barrier to supporting those convicted of terrorism offences.

Client challenges. Challenges in clients' lives can create barriers through the impact they have on an individuals' ability or willingness to engage with an intervention. Challenges described in the research include mental health problems; addiction and substance abuse; and a breakdown or absence of supportive relationships. The dynamic nature of these challenges means interventions benefit from taking a responsive approach that aims to accommodate and address challenges as they emerge rather than seeing them as permanent setbacks.

Overall completeness and applicability of evidence

The studies identified through the review only allowed us to respond to Objective 2 on the factors that shaped the implementation of case management interventions. No studies were identified that enabled us to speak to the effectiveness of different tools and approaches (Objective 1). A modest number of studies (n = 7) allowed us to draw some insights into the assumptions underpinning the implementation of case management approaches (Objective 2a), but no research was identified that enabled the review to draw conclusions as to what influenced the implementation of different approaches (Objective 2b). Research relating to Objective 2b instead examined the implementation of specific case management tools and specific stages of the case management process (n = 43) and those factors which facilitated, generated barriers, or acted as moderators across the full case management process (n = 41).

The research included in the review was international in nature. Studies reported on interventions in seventeen countries, and six examined more than one country context. Most of these studies were based in the Global North so whilst the review is not representative of CVE initiatives internationally, it has gone some way to develop a broader evidence base on secondary and tertiary interventions operating across the world. The inclusion of languages other than English supported this ambition. Secondary and tertiary interventions were almost equally represented in the review (n = 12 and n = 14 respectively), whilst the remaining 21 studies examined the use of tools and/or approaches that spanned secondary or tertiary prevention.

This review has demonstrated that much more attention has been paid to client assessment (specifically risk assessment tools and methods); aspects of the delivery and implementation process; and to a lesser extent, tools to support monitoring and evaluation. Client identification, case planning, and exit and transition processes have received less attention. An additional finding is the limited efforts to develop and deploy theories of change or logic models, and the often implicit and hybrid nature of underpinning frameworks such as RNR or strengths-based models. Together these issues speak to the organic way the CVE field has evolved, and the relative absence of agreed systems, processes, or measures of success, all of which present challenges when trying to integrate a wide body of work into a structure that may not have been considered when the tools and approaches discussed in this review were developed.

One notable limitation is the lack of research examining the potential unintended consequences of case management tools and approaches. Broader research examining policy and practice has pointed to the potentially negative impacts that counter-radicalisation work might have on individuals and communities (e.g., Heath-Kelly, 2013; Abbas, 2019). Although these studies provide initial data pointing to the potential for case management interventions to generate these effects, robust empirical research examining whether case management tools and approaches are producing these effects is lacking. None of the eligible studies examined these potential impacts in any detail, which means we are unable to comment on whether and how these unintended consequences might play out. It is crucial not to overlook the concerns that have been raised, making more empirical research necessary to better understand the scope of these issues, and how they might be addressed.

Quality of the evidence

The quality of evidence in the review was uneven. In comparison to more typical Campbell systematic reviews, this review included a broader range of research designs, including quantitative and, predominantly, qualitative designs that are traditionally considered unsuitable for inclusion in systematic reviews of intervention effectiveness, and we did not identify any eligible studies that used experimental or stronger quasi-experimental designs. In the context of a traditional review of intervention effectiveness, these types of designs can be understood as having important methodological weaknesses, and/or elevated risk of bias. The results of our analysis should therefore be read with these potential weaknesses in mind, however these research designs provide strong evidence in relation to implementation, and are therefore not considered ‘weak’ in the context of evaluating implementation, or of our specific, objectives.

Overall, the analysis of implementation (Objective 2) was based on a robust body of qualitative and mixed methods research. As discussed in Section 4.2.2, all of the studies included in this analysis were assessed as being of good quality by the review team, and as having no critical weakness when assessed using the relevant quality assessment tool (i.e., the CASP or EPHPP tools). However, the evidence relating to Objective 2b was more robust than for Objective 2a, with the evidence-base for the former comprising of 47 studies, compared to only seven for the latter. More research that examines the programme logics underpinning case management interventions, and which assesses the implementation of these logics is therefore badly needed. More research including ‘stronger’ quantitative methods is also needed to explore those themes identified in qualitative studies.

Future research exploring effectiveness will also be important, as the research designs identified through this review are not able to assess whether case management tools and approaches are effective at reducing the risk of radicalisation, recidivism, or terrorism. Similarly, the number of different tools, approaches, and stages of the case management process makes it difficult, on the basis of the current evidence, to determine which aspects might be more or less helpful. Further, although the case management framework seems able to support the process of structuring and delivering interventions, there is little evidence to determine if this is a reasonable assumption to make.

These challenges with the evidence base are due to the lack of an evaluation culture in this field (Baruch et al., 2018), and the significant methodological and conceptual challenges facing efforts to evaluate interventions (Lewis et al., 2020). An additional challenge in trying to understand the wide range of tools, approaches, activities and actors operating in this field is the relatively recent incorporation of the case management framework. Although the language of case management is beginning to be integrated into policy and research, it has not reached a stage where this is the dominant framework used in this space. This means that research is unevenly distributed across different aspects of the case management process and, with some exceptions (e.g., Cherney & Belton, 2021a, 2021b; Cherney, 2022), little effort has been made to use the case management framework to look holistically at CVE programmes, or to integrate policy or research across the different processes that make up interventions.

Limitations and potential biases in the evidence

Part I was designed to capture the broadest body of evidence possible. To do so, it included research in languages other than English and incorporated qualitative and quantitative research designs. Given the aim of looking across the case management process, and the wide number of search terms that sought to capture the range of tools and approaches used in this area of practice, we feel we reduced the chances of missing relevant evidence as far as possible.

Including a broader body of languages did however introduce the potential for bias. There are no agreed guidelines on how to carry out searches in languages other than English (Walpole, 2019). The nature and visibility of databases, grey literature sources and search functionality in non-English language contexts may vary in ways which are harder to control for. Similarly, there are challenges identifying the most appropriate translations of search terms which may mean some relevant research in the searches in languages other than English could have been missed. Collaborating with subject matter experts with relevant language skills sought to mitigate some of these challenges, however the additional resources needed to enable the identification, translation and analysis of material in languages other than English has the potential to impact how comprehensive these searches might be (Walpole, 2019).

The inclusion of qualitative research sought to overcome some of the limitations of the evidence base. The outcome of the search demonstrated the benefits of this strategy as no methodologically robust quantitative studies were identified. However, identifying, screening, and synthesising qualitative and weaker quantitative research designs carries a number of challenges (see Soilemezi & Linceviciute, 2018) including an increased risk of subjectivity at different stages of the process. The use of double, and sometimes triple coding; working with a team of subject matter experts; and employing a widely used quality assessment tool sought to mitigate some of these challenges, however it is possible that other research teams may have reached different conclusions as to what to include and exclude.

A broader challenge was that that the term ‘case management’ was rarely used in the identified literature, which meant that the research team had to determine what was (and what was not) a case management tool or approach based on the descriptions of the tool or approach provided by the original authors in conjunction with the definition of case management used in this review. This may have led to some bias in our inclusion or exclusion decisions. Whilst we used double, and sometimes triple coding to minimise this risk, the identification of case management tools and approaches remains somewhat subjective. To this end, we hope that the conceptual framework that we presented in the protocol (Lewis et al., 2023), and in Section 2 of this review, provides a foundation for future synthesis.

Agreements and disagreements with other reviews

This is the first review to examine tools and approaches used in case management interventions for countering radicalisation to violence. There are therefore no reviews to assess levels of agreement or disagreement against. However, two systematic reviews examined related questions that are helpful to consider in relation to the findings of our review. Hassan et al. (2021b) carried out a systematic review of tertiary intervention programmes. This primarily focused on the overall impact of programmes seeking to reduce violent radicalisation risk, as opposed to ours which looked at the tools and approaches used to support the delivery of these programmes. However, Hassan et al. (2021b) did identify facilitators and challenges to the implementation of tertiary interventions. Our review largely agrees with their findings.

Challenges identified in Hassan et al.'s (2021b) review included inadequate training; uncertainty and a lack of clarity over programme objectives; insufficient human and financial resources; expensive external experts; tensions between staff members; competition between stakeholders in multi-agency partnerships; short-term interventions; overwork; concerns over safety; and challenges supporting the reintegration of clients due to the stigma of the offence. Facilitators included the benefits of trust between practitioners and clients; strong working relationships; tailoring interventions to client needs; and engaging with family members. All of these factors were identified in our review of the case management process.

Conclusions from a systematic review of research on multi-agency programmes with police as a partner for reducing radicalisation to violence (Mazerolle et al., 2021) also aligned with the findings of our analysis around the importance and role of multi-agency working. Factors found to facilitate multi-agency working in this context included trusting relationships between partners; a shared understanding of goals; reducing the bureaucratic load practitioners are required to carry; appropriate means of dealing with information and intelligence sharing; and the availability of adequate support and training for practitioners. Our review also found all of these factors relevant to the implementation of multi-agency working.

Criteria for considering reviews for inclusion in Part II

Types of interventions

Reviews examining case management interventions or tools and approaches used at different stages of the case management process that are designed to prevent engagement in, or promote disengagement from, collective or interpersonal violence were eligible for inclusion.

As set out in more detail in the protocol (Lewis et al., 2023), violence is understood as ‘[t]he intentional use of physical force or power, threatened or actual’ that ‘either results in or has a high likelihood of resulting in injury, death, psychological harm, mal-development, or deprivation’ (Dahlberg & Krug, 2002, p. 5). It is common to distinguish between collective, interpersonal, or self-directed forms of violence (Dahlberg & Krug, 2002). This review focuses on collective or interpersonal violence as these are the most relevant to counter-radicalisation interventions. These are defined as: a.

Collective Violence: Physical, psychological or sexual violence perpetrated by those acting as part of a collective such as gang-related violence (e.g., Randhawa-Horne et al., 2019) or larger-scale militancy (e.g., USAID, 2021).

Interventions designed to address these kinds of violence are increasingly seen as holding lessons for efforts to counter radicalisation to violence. Research has begun to explicitly draw the lessons from, for example, interventions to address larger-scale militancy or gang violence (Ris & Ernstorfer, 2017; Davies et al., 2017), as well as sexual offending (Cherney et al., 2021) for CVE work. Recognising these potential synergies, reviews in Part II were eligible if they focused on interventions that: 1.

Were designed for individuals rather than communities or collectives.

Search methods for identification of reviews

The search strategy for reviews in English and languages other than English followed a similar process to that set out for Part I and involved the following stages: 1.

Identification of search terms.

Data collection and analysis

Deviations from the protocol

Four of the deviations identified in Section 4.1.4 in relation to Part I also applied to Part II. First, the search strategy for literature in languages other than English was adapted to include searches for the translated search terms, and searches for the English language search terms filtered on languages other than English. Second, it was not possible to conduct searches using the translated search terms in the Ovid and ProQuest databases, which meant that only English language search term searches, filtered on languages other than English, were conducted on these platforms. Third, studies returned by the searches in languages other than English that had English language titles and abstracts were initially screened by the primary reviewer (JL), and not one of the language specialists. Finally, the discussion of evidence relating to implementation (i.e., relating to Objectives 4a and 4b) was structured around a discussion of facilitators, barriers and moderators in the same way as was done in Part I of the review.

One further deviation from the protocol was specific to the Part II analysis. It was not possible to construct a typology of different tools and approaches as outlined in the original protocol due to the small number of eligible reviews identified, and the fact that over half of the reviews identified (n = 5) focused on a single type of tool: risk assessment tools.

Description of included reviews

Results of the search

The results of the Part II search and screening process are set out in Figure 3. The initial English database searches identified 55,872 studies. An additional 4104 studies were identified from the languages other than English searches, which included research in French, German, Norwegian, Swedish and Danish, and Russian. These were combined with material identified through the search of grey literature sources, hand searches of relevant journals and consultation with experts (n = 998). A total of 60,974 references made up the initial corpus. After de-duplication, a total of 36,626 references went forward for title and abstract screening using Covidence.

OPEN IN VIEWER

Figure 3

PRISMA diagram of included reviews.

The titles and abstracts of English language material were screened by two members of the research team, conflicts were assessed by a third team member and a decision reached through discussion. Non-English language material was similarly reviewed by team members with relevant language expertise. This initial screening process led to the removal of 36,169 reviews, leaving 457 which went forward for full text review in Covidence. The full text of four reviews were unavailable in institutional repositories. After assessing the title and abstract of these reviews using the full-text screening criteria, these were subsequently excluded. The remaining 453 reviews were reviewed by two members of the research team and conflicts adjudicated through discussion with a third team member.

From a total of 453 reviews that underwent full text screening, eight went forward for inclusion in the review, all of which were published in English. No eligible reviews were identified in German, Norwegian, Danish, Swedish, or Russian. Five reviews met the inclusion criteria for Objective 3 on the effectiveness of case management tools (n = 3) and approaches (n = 2). Seven of the reviews presented data related to whether case management interventions were being implemented as intended (Objective 4a), and/or discussed what influenced how interventions were implemented, focusing on factors which facilitated, generated barriers, or acted as moderators to implementation processes (Objective 4b).

Quality of included reviews

Synthesis of results

The following discussion of the Part II results is made up of three parts: the first part reviews the research on case management approaches, addressing Objectives 3 (effectiveness) and 4 (implementation); the second covers the same objectives for different case management tools; and the third discusses the transferability of these tools and approaches to the field of countering radicalisation to violence, addressing Objective 5 (transferability).

Following Cochrane guidelines (Pollock et al., 2021), the analysis that follows summarises relevant data reported within the included systematic reviews. Where possible, we summarise results drawn from meta-analyses of outcomes. However, only one review presented a relevant meta-analysis (van der Stouwe et al., 2014). This meant that data from the remaining seven reviews could only be extracted narratively. When presenting findings in narrative form, we quantify the number of studies reporting on a specific theme, and where relevant, cite the original study or studies that the review refers to in their findings. However, we did not undertake a separate analysis of the individual studies themselves.

Transferability to interventions seeking to counter radicalisation

The discussion of transferability is split into two sections. The first section examines the extent to which the research covered in the reviews cited in Part II are transferable to case management interventions seeking to counter radicalisation to violence. Transferability is interpreted using the domains set out by Munthe-Kaas et al. (2019; 2020). The second discusses the implications of the insights set out in Part II for counter-radicalisation interventions. It describes a series of themes identified in Part II which relate to the objectives of the review, considering effectiveness and implementation of case management interventions in the context of radicalisation to violence. It uses the same transferability framework to look at the one intervention that was eligible for inclusion in Part II that is not already a feature of CVE practice according to the literature reviewed in Part I.

Summary of main results

Part II of the review had three objectives: to understand the effectiveness of tools, approaches and interventions used in case management interventions seeking to counter violence; to examine those factors and moderators which impact how they are implemented; and to consider the transferability of these tools and approaches to counter-radicalisation work.

The analysis in Part II illustrates that case management tools and approaches are being used in a variety of different settings (i.e., community, correctional, clinical), and to prevent different forms of interpersonal violence. However, their effectiveness remains unclear: only five systematic reviews examined the impact of different tools, approaches or interventions on outcomes relating to the prevention of violence and they reported mixed findings.

The presence of these reviews highlights that research examining the use of case management tools and approaches to counter other forms of violence is more mature than research relating to countering radicalisation to violence where no studies, and therefore no systematic reviews of research on the effectiveness of these measures, were identified. The pattern of research identified in Part II was comparable to that of Part I; both illustrate a heavy emphasis on risk assessment tools, and offer more comprehensive analyses of implementation facilitators and barriers than effectiveness and outcomes.

Overall completeness and applicability of evidence

The reviews identified through the literature searches enabled us to respond to all of the research objectives for Part II. However, the amount and quality of evidence identified varied across the research questions. In total, five reviews presented evidence relating to Objective 3 on the effectiveness of case management tools and approaches. Two reviews examined the effectiveness of different interventions, which limits the conclusions that can be drawn about the effectiveness of case management more broadly; particularly as neither of the interventions were specifically defined as ‘case management’ in the reviews. It was also not possible to examine the effectiveness of different approaches to case management as outlined in the protocol (Lewis et al., 2023) due to the limited number of reviews identified, and the lack of detail provided about their underlying assumptions.

Three reviews allowed us to draw some conclusions about the potential effectiveness of risk assessment tools and the polygraph in contributing to violence reduction. However, because any relationship between the use of such tools and intervention outcomes is likely to be indirect, it was not possible to comment on how these tools might contribute to a reduction in violence. No eligible reviews examined the effectiveness of other case management tools identified in Part I, which limits the conclusions that can be drawn about the effectiveness of relevant tools.

Whilst the number of reviews that examined their implementation remained small, it was possible to explore whether risk assessment tools were being delivered in the ways originally outlined in Section 2 (Objective 4a) using data from two studies. However, no relevant evidence for other case management tools was identified. Moreover, all but one of the six reviews that pointed to relevant implementation factors and moderators affecting the use of case management tools (Objective 4b) focused on risk assessment tools and their use in informing risk management. The evidence relating to implementation was therefore heavily weighted towards one case management tool, and one stage of the case management process.

The reviews in Part II were international in nature, and included studies conducted in ten countries. Similarly to Part I, the vast majority of the evidence was drawn from the Global North, which limits the representativeness of our findings. Whilst we searched for relevant research in languages other than English, no eligible reviews were identified in French, Russian, German, Norwegian, Danish, or Swedish. The included reviews spanned secondary and tertiary prevention work delivered in a range of settings (i.e., clinical, community, and correctional), and included research relating to violent and sexual offending. Taken together, Part II captured evidence relating to a diverse range of populations and problems, albeit with some limitations.

Part II attempted to cover a huge body of research that was not restricted to any one type of violence, or any one field of violence prevention. Whilst our search strategy was comprehensive and systematic, there are inevitably challenges in trying to identify research from across multiple fields that may use different terminology to describe relevant tools and approaches. This is particularly the case given that Part II of the review was designed to identify transferable lessons for counter-radicalisation work. As a result, the search terms developed to search this wider literature were based on those search terms used for Part I, which was informed by the literature on countering radicalisation to violence. Whilst we took steps to minimise any potential challenges created by this approach – such as piloting our search terms and benefiting from the input of an information retrieval specialist with experience in conducting Campbell systematic reviews relating to violence prevention (EE) – it is important to recognise this issue.

Quality of the evidence

Only two of the reviews were assessed as being high quality using the AMSTAR II tool. Every review had at least one methodological weakness, and only one review included a meta-analysis of relevant outcomes. In addition, the methodological inclusion criteria used in most of the reviews was less stringent than for Campbell systematic reviews; only two set strict restrictions on the types of quantitative designs that could be included. However, the original methodology as outlined in the protocol (Lewis et al., 2023) did not set any restrictions on the types of studies that could be included in Part II, and every review was assessed as being of sufficient quality to be included using the AMSTAR II tool. Moreover, as noted in Section 4.3.3, data drawn from weaker quantitative and qualitative studies (as cited in included reviews) provided relevant insights relating to implementation.

Limitations and potential biases in the evidence

As discussed in Section 4.3.4 in relation to Part I, the inclusion of multiple languages, and the use of search terms that related to different stages of the case management process aimed to reduce the chances of missing relevant evidence. In addition, feedback from the Campbell Crime and Justice editorial board on our original search terms led us to include a larger set of search terms in the ‘problem’ domain of our search strategy for Part II (see Supporting Information: Appendix I) at the protocol stage so as to capture a broader range of synonyms for different types of violence. However, although some members of the team (particularly AC ad EE), and the advisory board also had experience in conducting research on topics relating to Part II, the research team had specific expertise in countering radicalisation to violence, and therefore had a stronger understanding of that literature than the wider field of violence prevention.

Whilst we used an appropriate tool for assessing the quality of included reviews (AMSTAR II), such assessments are subject to potential bias. The use of double coding, and in some instances detailed discussions between team members on individual reviews, helped to mitigate this issue. However, as noted in Section 4.3.4, it is possible that other research teams may have reached different conclusions as to what to include and exclude based on quality. Furthermore, the decision not to exclude studies based on the quality of the studies that they cited potentially introduced an increased risk of bias into our analysis of Part II. To mitigate this, as far as possible, we considered the potential biases that might have impacted the analysis presented in each review and reported on this where relevant.

No eligible reviews focused specifically on ‘case management’ interventions. Instead, the reviews included in Part II used a range of different terms to describe tools and approaches that the research team assessed as being relevant to this review. This may have led to some bias in our inclusion or exclusion decisions, as this process relied on the research team assessing whether the tools and approaches adhered to the core assumptions of our conceptual framework outlined in Section 2. Whilst we used double, and sometimes triple coding to minimise this risk, the identification of case management tools and approaches remains subject to bias. Although we are confident in the methods used to screen studies, we recognise that other research teams may have made different inclusion and exclusion decisions.

A related challenge is that a number of systematic reviews which examined studies relating to relevant case management interventions were identified, but were not included in the review on the basis that they only presented outcome data at an aggregate level, or did not conduct sub-group analysis of case management specifically. Whilst these studies would be captured by a systematic review of primary research studies, they are not covered here.

Agreements and disagreements with other reviews

Due to the broad focus of Part II, no comparable overviews of reviews were identified.