This article proposes an analytic framework for knowledge-building on IPS sustainment and scale-up and, tentatively, evidence-based CPIs more broadly. Existing frameworks serve different purposes: EPIS addresses phased implementation processes and contexts [12], RE-AIM/PRISM focuses on planning and evaluation across implementation outcomes and contextual influences [13], ExpandNet on scale-up strategy [14], the Dynamic Sustainability Framework on continuing intervention–context fit and perceiving sustainment as a process of continuous adaptation [15], and FRAME on intervention adaptations [16]. Schell et al.’s nine-domain Sustainability-capacity framework identifies political support, funding stability, partnerships, organizational capacity, program evaluation, adaptation, communication, public health impacts, and strategic planning as important for sustaining public health programs [17]. LEL therefore contributes at a different analytical level by asking which policy-level domains are most useful for analyzing external context’s role in sustainment and scale-up.

This article draws on policy implementation research and welfare-state scholarship as adjacent interpretive resources. They are therefore reported in the Results, where they help interpret the findings and support theory-building.

Sustainment and scale-up are discussed together because they pose related but distinct implementation challenges: sustainment as a process refers to creating conditions that allow an intervention to remain a routine, high-fidelity practice over time, and scale-up usually refers to expanding an intervention to more sites and service users under broadly similar conditions. Successful scale-up may therefore be understood as a function of the penetration and reach of sustained services. In contrast, scale-out extends an intervention to new populations and/or delivery systems. In IPS, scale-out includes youth, people with substance use disorders, and other groups whose support is organized across different service systems.

The research questions were:

What recurring empirical patterns can be identified in the external-context conditions shaping sustainment and scale-up across empirical IPS implementation studies?

This study used a theory-building integrative review design to synthesize heterogeneous empirical evidence on external context conditions shaping IPS sustainment and scale-up. Integrative reviews aim to synthesize evidence across heterogeneous study designs and to generate conceptual insights that can inform future research and practice [18, 19]. An integrative rather than a scoping review was chosen to permit interpretive engagement with implementation science, policy implementation research, and welfare policy analysis. Empirical findings were grouped inductively and interpreted abductively to develop plausible explanations and formalize the LEL framework [20, 21].

The empirical evidence base was built in two steps. First, empirical IPS implementation studies were drawn from a recent scoping review; 32 of the 59 reported studies were included as they reported relevant implementation outcomes: sustainment, penetration, or reach [22]. Second, the evidence base was updated through a Web of Science search from the scoping review end date through November 2025. The updated search yielded 125 records; 28 full-text articles were screened, and nine additional articles met the inclusion criteria and were added to the evidence base. Included studies are presented in Supplementary File 1. They spanned legislative contexts, welfare sectors (health, social care, and employment services), and governance levels (local, regional, and national), and used qualitative, quantitative, and mixed-method designs. No studies from low- or middle-income countries were identified; relevance in such settings therefore requires separate examination. Because the aim was conceptual synthesis, no formal quality appraisal was conducted; instead, recurring themes reported across independent studies and settings were prioritized.

Theoretical manuscripts were selected purposively for their explanatory relevance to inductively identified patterns. This approach was also appropriate because the included IPS implementation literature was not anchored in a single research paradigm and most studies addressed pragmatic implementation questions. A smaller subset engaged more explicitly with implementation structures, institutional logics, or welfare-service perspectives.

Text excerpts reporting on sustainment, penetration, or reach, together with the external-context conditions discussed in relation to these outcomes were extracted. The analytical process followed Braun and Clarke’s six-phase thematic analysis [23]. The extracted material was read for familiarization, coded for external-context conditions linked to these outcomes, and grouped into candidate themes. These themes were reviewed against the extracted data and theory sources, then refined, defined, and named as broader analytical domains. The process was iterative and recursive; the final framework structure was cross-checked against empirical studies to ensure that each domain was grounded in recurring empirical patterns. This check supported the three-domain structure. The final domains were written up as an integrative framework and are presented in a summary table. Empirical examples from the included studies are provided to illustrate the domain structure.

Analyses were conducted in ATLAS.ti (version 26) for thematic coding and categorization. ChatGPT (version 5.5) and Grammarly were used for ideation and language editing/proofreading, not for data coding, analysis, interpretation, or generating results.

The findings across the literature were grouped into three policy-level domains associated with sustainment and scale-up: Policy Legitimacy, Policy Embedding, and Policy Learning (Table 1). Policy Legitimacy concerns how policy paradigms and legitimization processes shape the political-administrative decision chains through which IPS is mandated or contested across organizations. Policy Embedding captures multi-organizational implementation structures as systems in which core implementation functions are connected or fragmented across levels and sectors. Policy Learning focuses on structures through which implementation experience informs practice, policy decisions, and model refinement. The domains are considered distinct but mutually reinforcing: legitimacy shapes whether embedding is authorized, embedding creates organizational conditions for learning, and learning can renew legitimacy and guide further embedding.

Policy Legitimacy refers to whether IPS’s principles and objectives are regarded as normatively appropriate, institutionally compatible, and politically authorizable within a given welfare and labor-market regime. Its unit of analysis is the decision chain through which IPS is authorized, contested, or marginalized. Policy implementation research conceptualizes implementation as a multi-step chain of decisions linking high-level policy directives to frontline practice [28]. The unit of analysis is defined as a process that begins with ‘policy formulation’ and ‘policy design’ within legal and administrative frameworks and culminates in the explanation of variation in outputs [29]. Scaling EBPs may be regarded as an institutional logic competing for legitimization and institutional mandates with other policy logics: action may involve designing institutions as well as managing implementation [36]. In Hall’s terms, paradigms are shared frameworks of ideas and norms that define legitimate goals, instruments, and target populations [26]. The institutional logics perspective shifts attention from the policy legitimacy to congruence among policy design, intervention principles, organizational logics, and accountability mechanisms [31].

Labor-market participation among marginalized populations is a long-standing welfare-state question concerning how social citizenship, social rights, and responsibilities are defined [24]. IPS offers a distinctive evidence-based response to this question that aligns with “social investment” approaches that emphasize participation and capability-building as routes to inclusion [37]. However, in public policy, deservingness criteria vary across policy sectors, and public attitudes toward social security and the welfare state differ across electorates, whose preferences ultimately shape political agendas through electoral accountability [25]. Over recent decades, welfare-system reforms have more often followed a ‘workfarist’ than a social-investment orientation [38], despite evidence that stringent benefit-sanction systems can adversely affect vulnerable groups, including through financial hardship, housing instability, and worsened mental health [39, 40]. Pierson’s account of incremental welfare-state change suggests that reforms often become politically viable through recalibration of existing institutions rather than through overt welfare expansion or retrenchment [41]. Evidence-based practices such as IPS should therefore be understood as one reform agenda among others.

Policy Embedding refers to the formation and stabilization of an implementation structure: the cluster of organizational units, actors, mandates, resources, accountability mechanisms, and support routines through which IPS is delivered and maintained beyond the pilot implementation. Policy implementation research (PIR) explains variation in outputs by tracing how policy intent is translated through a multiorganizational implementation structure and enacted through professional routines and frontline discretion [29]. ‘Top-down’ PIR accounts emphasize reducing “slippage” by aligning incentives and responsibilities, and delegating implementation to agencies whose mandates align with the policy objectives [56]. For example, implementation has been conceptualized as a chain containing multiple decision and clearance points, at each of which the original policy purpose may be delayed, reinterpreted, weakened, or displaced [28]. “Bottom-up” accounts emphasize that policy processes are indeterministic and shaped by frontline discretion and fragmented jurisdictional, organizational, and collaboration arrangements that do not derive from the implementer’s regulatory authority [57, 58]. Hybrid implementation models have also been described, in which clear mandates and adequate resourcing create feasibility, while local ownership and problem-solving capacity enable adaptation and routinization [29]. The concept of the implementation structure provides a rationale for defining the relevant unit of analysis as the network of organizational actors assembled around a policy problem rather than any single formal organization [27].

Policy Learning refers to sustained implementation evidence infrastructure through which implementation experience informs IPS practice and policy over time. The definition includes governance-level feedback loops in which implementation results inform system-level decision making and model refinement. Frameworks such as EPIS and PRISM highlight implementation-support and learning functions as important for sustainment, and they recognize the interaction between inner and outer contexts [7, 12]. The sustainability capacity perspective acknowledges the importance of system-level structures for sustainability [17]. However, these accounts less explicitly conceptualize policy-level learning structures as units of analysis. The capacity-building function is central to implementation-support models, in which implementation support practitioners build provider implementation capacity [65] and support systems strengthen delivery systems’ capacity for implementation [6]. The learning-collaborative model adapted by the originators of IPS may be more effective than low-intensity strategies for improving provider practices, although the evidence remains inconclusive [34, 66].

Policy implementation research adds that such learning structures are not only mechanisms for maintaining implementation quality but also a mechanism through which operational experience can recalibrate administrative and policy decisions. The integrated implementation model [29] links monitoring and evaluation back into subsequent policy design and implementation management through a feedback mechanism. The concept of implementation structures directs attention to the network of organizational actors assembled around learning functions [27]. Research on public-sector coordination adds that such learning depends on how information, accountability, and joint action are organized across administrative levels and sectors [30].

This study has several limitations. It is theory-building and integrative rather than a full systematic review or meta-analysis. By spanning three broad theoretical traditions, the synthesis prioritizes conceptual integration and does not comprehensively represent any one tradition; the proposed framework’s domains should therefore be treated as integrative and exploratory. Moreover, the review did not systematically compare welfare-state models, national policy regimes, or implementation settings; country and jurisdictional examples were used to ground the interpretation rather than as objects of systematic comparison. The empirical literature was drawn entirely from high-income welfare-state contexts, and the framework requires separate empirical examination before being applied to low- and middle-income settings.

This integrative review identified three policy-level domains for analyzing the sustainment and scale-up of the Individual Placement and Support model: Policy Legitimacy, Policy Embedding, and Policy Learning. LEL can be used as a framework for comparing policy legitimization and decision chains, implementation structure, and policy learning shape sustainment and scale-up. As an IPS-derived framework, its relevance for other complex evidence-based psychosocial interventions should be tested through comparative research.