Fact-checked by Grok 2 weeks ago

Open Dialogue

Open Dialogue is a psychosocial treatment approach for acute mental health crises, particularly psychosis, developed in the 1980s in Western Lapland, Finland, that prioritizes immediate, collaborative meetings among the affected individual, their family and social network, and multidisciplinary clinicians to generate shared understanding and need-adapted responses without predetermined protocols or routine antipsychotic medication. The method draws on dialogic principles inspired by Bakhtin, emphasizing relational continuity, tolerance for uncertainty in early phases, and generating multiple voices in treatment planning to foster crisis resolution through social connection rather than isolated symptom suppression. Key principles include rapid response within 24 hours of referral, network meetings open to all involved parties, and a team-based structure where at least one consistent ensures continuity across inpatient and outpatient care. Medications, if used, are introduced collaboratively and at low doses only after fails to alleviate distress, contrasting with standard practices that often prioritize pharmacological . This approach has been implemented primarily in systems, integrating with existing services like employment support while avoiding hospitalization unless necessary, leading to reported reductions in involuntary admissions and restraint use. Empirical outcomes from Finnish longitudinal studies indicate superior results compared to treatment as usual, including over 80% of patients with first-episode psychosis achieving symptom remission and returning to work or study within two years, with schizophrenia diagnoses dropping to near zero in treated cohorts and lower long-term disability rates. A 10-year follow-up confirmed sustained cost-effectiveness, with reduced healthcare expenditures due to shorter hospital stays and fewer relapses. However, while qualitative and observational data support these findings, the evidence base relies heavily on non-randomized studies from the originating region, prompting calls for randomized controlled trials to assess generalizability and rule out selection effects or systemic factors in Finland's integrated care model. International adaptations show mixed implementation fidelity, with promising but preliminary results in enhancing social cognition and recovery experiences.

Origins and Development

Historical Context in Finland

The Open Dialogue (OD) approach originated in Western Lapland, , during the early 1980s as a response to the limitations of conventional psychiatric treatment, which often prioritized medication, seclusion, and prolonged hospitalization over social and relational factors. Developed at Keropudas Hospital in by clinical Jaakko Seikkula, psychiatric nurse Markku Sutela, and a multidisciplinary team trained in systemic , OD emphasized immediate mobilization of the patient's social network for collaborative crisis response. This model evolved from the Need-Adapted Treatment (NAT) framework established by psychiatrist Yrjö Alanen and colleagues in starting in 1975, which integrated psychodynamic, systemic, and elements to customize care based on individual needs rather than standardized protocols. In 1984, Keropudas staff initiated changes to inpatient practices for acute , replacing isolated assessments with open network meetings involving , friends, and professionals to foster and reduce coercive interventions. By the early 1990s, had expanded into a region-wide system through initiatives like the Acute Integrated Project (1992–1993), becoming the standard psychiatric service in Western Lapland and supplanting traditional sector-based models with a focus on early intervention and community integration. Early outcome tracking from 1990 onward in this area documented lower rates of residual symptoms and compared to national averages, informing further refinements.

Key Contributors and Evolution

Jaakko Seikkula, a clinical , is the primary developer of the Open Dialogue approach, initiating its formulation in the early 1980s at Keropudas Hospital in , Western Lapland, Finland, as a response to treating acute through network meetings involving patients, families, and professionals. In collaboration with Markku Sutela and a multidisciplinary team including psychiatrists like Birgitta Alakare and Jukka Aaltonen, Seikkula integrated elements of systemic family therapy and dialogic principles, emphasizing immediate, open conversations without predetermined expert agendas. The approach evolved from experimental practices in the mid-1980s, where the Western Lapland district shifted from traditional inpatient care to community-based network interventions, achieving notable reductions in hospitalization rates—dropping to under 20% for first-episode psychosis by the early 1990s—and lower antipsychotic use compared to national averages. By the 1990s, longitudinal observational studies by Seikkula and colleagues documented sustained outcomes, such as 80% of patients returning to work or study within two years, prompting formalization into a comprehensive system of care across the region. Internationally, Open Dialogue gained traction in the early through Seikkula's publications and training, influencing adaptations in the United States via programs like for Dialogic Practice, established in 2010, and a launched in the UK in 2017 to assess its efficacy against standard care. Despite its expansion, evolution has included critiques of over-reliance on cultural context, leading to refinements like " Practice" to broaden applicability beyond to general crises.

Theoretical Foundations

Influences from Systemic and Dialogic Therapies

Open Dialogue incorporates foundational elements from systemic family therapy, which conceptualizes crises, such as , as emerging from relational patterns within family and s rather than isolated individual deficits. This influence manifests in practices like mobilizing the patient's for immediate meetings, typically within 24 hours of referral, to address symptoms contextually and collaboratively. Developers in Western Lapland, including Jaakko Seikkula, drew from systemic approaches and family therapy traditions, adapting circular questioning and hypothesis-testing to foster systemic understanding without pathologizing the . Complementing systemic roots, Open Dialogue adopts dialogic therapy principles, emphasizing co-constructed meaning through polyphonic, non-monologic interactions that prioritize responsiveness over expert interpretation. A key inspiration is Tom Andersen's reflecting team method, introduced in the , which shifts power dynamics by allowing family members to overhear and respond to professionals' reflections, thereby generating multiple perspectives and reducing hierarchical authority. In practice, this translates to therapists speaking in the second person to , generating utterances that invite continuation rather than , and tolerating to enable emergent dialogue. These dialogic elements, integrated since the approach's formulation in the early , aim to create "open" conversational spaces that mitigate and promote shared sense-making.

Core Philosophical Assumptions

The Open Dialogue approach posits that psychotic crises emerge from breakdowns in shared relational , rather than isolated biological pathologies, emphasizing the restoration of dialogue within social networks to generate new understandings. This view draws directly from Mikhail Bakhtin's philosophy of dialogism, which conceives and as inherently social and polyphonic, where meaning arises through the interplay of multiple, non-hierarchical voices without a single authoritative interpretation. In practice, this assumes that even seemingly incoherent psychotic utterances carry relational significance and require responsive listening to elicit coherence, rejecting monological expert dominance in favor of co-constructed narratives. Central to these assumptions is a relational ontology, wherein the self and are not autonomous entities but emerge from interactions within family and social systems, influenced by systemic traditions that highlight circular causality over linear biomedical models. Proponents argue that crises reflect unmet relational needs or traumas manifesting linguistically, treatable by mobilizing the patient's network to foster "aliveness" through immediate, flexible responses that prioritize emotional connection over premature diagnostic closure. This entails enduring uncertainty in therapeutic encounters, avoiding fixed categorizations like diagnoses during acute phases to prevent of the person. Epistemologically, Open Dialogue assumes knowledge about the crisis is not pre-held by clinicians but generated collaboratively, aligning with Vygotsky-inspired views of through zones of proximal understanding, extended to treatment. Causally, it privileges multifaceted etiologies—including social disruptions and communicative failures—over exclusive explanations, positing that sustained can realign relational dynamics and reduce symptom persistence without relying on hospitalization or antipsychotics as first-line interventions. These tenets, articulated by developer Jaakko Seikkula since the , underpin the method's shift from individual symptom management to network-centered sense-making, though their empirical validation remains tied to observational outcomes rather than universal causation.

Core Principles and Practices

Network Meetings and Immediate Response

In Open Dialogue, the immediate response principle mandates convening the first treatment meeting within 24 hours of the initial contact regarding a psychiatric crisis, such as acute , to provide rapid engagement and reduce escalation risks. This protocol prioritizes accessibility over preliminary assessments, ensuring the person's is involved from the outset to foster collective problem-solving rather than isolated expert . Network meetings form the core mechanism for this response, gathering the individual experiencing distress, family members, close friends or natural supports, and at least two professionals in an open, circular typically lasting 90 minutes. Organized by the receiving the initial referral, these meetings emphasize , with no separate staff discussions; all reflections, decisions on hospitalization or , and therapeutic dialogues occur in the presence of the full network. Sessions begin with two standardized open-ended questions—"What is the of the idea of coming here today?" and "How would you like to use this meeting?"—to elicit the network's perspectives and collaboratively. The conduct of network meetings prioritizes generating polyphonic dialogue over directive advice, with therapists facilitating by reflecting aloud on each other's statements to model uncertainty tolerance and validate multiple viewpoints, allowing up to one-third of the conversation to remain monological if needed for coherence. Subsequent meetings occur as frequently as required during the —often weekly initially—to maintain continuity and adapt to evolving crises, shifting focus from symptom labeling to relational narratives and shared . This approach aims to empower the network's agency, reportedly enhancing trust and perceived mattering among participants in implementations like those in outpatient settings.

Guidelines on Medication and Hospitalization

In the Open Dialogue approach, hospitalization is minimized through rapid mobilization of a crisis response team that convenes the first meeting within 24 hours of the initial contact, prioritizing home-based or community interventions to address acute without inpatient admission whenever feasible. This practice stems from the principle of immediate, need-adapted help, where case-specific teams maintain continuity of care across settings, resorting to only if safety risks necessitate it, resulting in significantly shorter stays compared to standard care—averaging 14.3 days in Open Dialogue versus 116.9 days in historical comparison groups for first-episode . Involuntary commitments, when unavoidable, involve team involvement from the outset to integrate the process into the environment. Regarding , neuroleptics are deliberately deferred and not introduced during the initial meeting; decisions on their use require discussion across at least three meetings to allow to influence symptom trajectories first. In the acute phase, particularly the first three weeks, benzodiazepines may be employed temporarily for or sleep if pharmacological support is deemed necessary, but neuroleptics are reserved for cases showing no improvement in psychotic symptoms or functioning despite engagement. This restrained approach aligns with observed outcomes, where 64% of first-episode patients in implementations avoided neuroleptics altogether, correlating with 79% achieving no or mild residual symptoms at two-year follow-up. Overall, the model de-emphasizes routine pharmaceutical in favor of generating shared meaning through , viewing as a secondary tool rather than a primary .

Evidence of Outcomes

Observational Studies from

Observational studies evaluating Open Dialogue (OD) in primarily draw from its system-wide implementation in the Western Lapland health district, where comprehensive training and protocol adherence transformed routine care for first-episode starting in the late . These studies, often prospective and register-linked, compared OD-treated cohorts to national benchmarks or contemporaneous non-OD regions, reporting metrics such as symptom remission, , hospitalization duration, and reliance. A foundational prospective tracked 75 patients with first-episode nonaffective treated via OD in Western Lapland from 1992 to 1997. At five-year follow-up, 81% exhibited no residual positive psychotic symptoms, 86% had returned to or studies, and the mean duration of hospitalization was 85 days in the first year, declining thereafter. Only 27% remained on pensions, contrasting with national rates exceeding 50% for similar cohorts. use was minimal long-term, with 71% discontinuing within two years. Longer-term analyses reinforced stability of these outcomes. In a cohort from the early followed up to five years post-acute phase, relapse rates remained low at 22%, with sustained functional recovery and reduced service dependence compared to Finland's overall figures, where residual symptoms persisted in approximately 50% of cases. Hospital bed days per patient in Western Lapland fell to 38 annually by the mid-, versus a national average of 153 for psychosis episodes. More recent register-based observational data from databases examined first psychiatric contacts among adolescents (aged 13–17) from 2006 to 2013, distinguishing OD-implemented regions like Western Lapland. At 10-year follow-up, 34.5% of the OD group required ongoing psychiatric treatment or disability allowances (or had died), compared to 41.3% in non-OD areas; adjusted odds for adverse outcomes were 28% lower in OD (aOR 0.72). Mortality, including suicides, was also lower in OD cohorts during this period. These findings, while uncontrolled for confounders like socioeconomic factors, highlight consistent patterns of reduced chronicity and resource use attributable to OD's network-oriented, low-medication framework in settings.

Recent Controlled Trials and International Data

A 2023 scoping survey identified Open Dialogue implementations in over 20 countries, including the , , , and , but noted a predominance of qualitative evaluations, pilot programs, and service descriptions rather than controlled trials assessing . International adaptations often report feasibility and practitioner satisfaction, such as reduced reliance on hospitalization in U.S. early programs, yet lack or comparison groups to isolate Open Dialogue's causal effects. These data highlight systemic challenges in exporting the model, including training fidelity and cultural differences in family involvement, without establishing superior outcomes over standard care. The largest controlled evaluation to date is the ODDESSI trial, a multi-site cluster-randomized controlled trial launched in 2021 across trusts, randomizing teams to deliver Open Dialogue or treatment as usual for adults in acute crises. Enrolling approximately 400 participants, the measures primary outcomes including time to all-cause service disengagement and secondary metrics like symptom reduction via the Brief Psychiatric Rating Scale at 12 and 24 months, alongside cost-effectiveness. As of October 2025, full results remain unpublished, though a 2025 qualitative process evaluation within the trial described peer-supported elements as enhancing without adverse effects. Protocol analyses projected potential reductions in hospital bed days, echoing observational patterns, but await confirmation. In , a 2025 unblinded multi-center across 12 sites tested a community-based explicitly informed by Open Dialogue, emphasizing dialogical network meetings with service users, relatives, and professionals for severe mental illness over 24 months. , involving 314 participants, aimed to evaluate effectiveness on functioning and hospitalization rates compared to standard outpatient care. Preliminary reporting indicated no significant differences in primary outcomes like scores, attributing this to implementation variability and baseline severity, though subgroup analyses suggested benefits for early engagement. This study underscores methodological hurdles in controlled settings, such as ensuring adherence to core practices like immediate response, which deviate from real-world applications. Overall, international controlled data as of 2025 provide tentative support for feasibility but insufficient evidence of consistent superiority, prompting calls for larger, blinded trials to address biases in non-randomized international reports.

Criticisms and Limitations

Methodological Weaknesses in Research

The foundational research on Open Dialogue (OD), primarily conducted in Western Lapland, Finland, consists of naturalistic, observational studies comparing outcomes before and after OD implementation in the 1990s, rather than randomized controlled trials (RCTs), which precludes definitive causal attribution of benefits to the approach. These studies, such as those by Seikkula and colleagues, reported lower rates of long-term hospitalization and antipsychotic use post-OD (e.g., from 78% to 22% hospitalization in one cohort), but employed historical controls spanning decades (1980s versus 1990s), introducing confounders like evolving diagnostic criteria, broader societal changes in mental health services, and non-contemporaneous populations. Sample sizes were small (typically N=50–100 per group), with retrospective diagnoses and inconsistent fidelity monitoring, exacerbating selection bias and limiting generalizability. Systematic reviews have consistently rated this evidence as low quality due to high risk of bias across multiple domains. For instance, a 2018 review of eight quantitative studies identified universal issues with unblinded assessments, lack of , and researcher allegiance—where developers evaluated their own intervention—potentially inflating positive outcomes. Only one study included a group (N=14), and none featured or blinding of outcome evaluators, violating key standards for establishing in psychosocial interventions. Qualitative studies (N=16 reviewed) fared similarly, with small, non-representative samples and opaque methods, yielding anecdotal rather than robust insights into mechanisms or acceptability. International adaptations have attempted controlled designs but reveal persistent flaws. Early non-randomized trials, such as a study (N=16), lacked controls and reported implementation variability, undermining comparability to results. Ongoing or completed RCTs, like the 's ODDESSI trial initiated in 2017, face challenges in achieving full fidelity outside resource-intensive contexts, with preliminary data indicating higher dropout rates and mixed symptom outcomes compared to as usual. Critics argue that without standardized adherence measures—only recently developed—these trials diluting elements like immediate network meetings, further confounding results. Overall, the absence of high-quality RCTs as of 2023 has led reviewers to conclude that 's evidence base does not yet support widespread adoption over established models, necessitating rigorous, blinded trials to isolate effects from non-specific factors like intensified support.

Potential Risks and Ethical Concerns

One potential risk of the Open Dialogue approach lies in its guideline of initially postponing antipsychotic , which may prolong acute psychotic symptoms in cases driven primarily by neurobiological factors, potentially increasing the likelihood of , aggression toward others, or chronicity. Early-phase carries a lifetime risk of approximately 7%, with elevated danger in untreated states, and randomized controlled trials for standard care demonstrate that prompt antipsychotics reduce rates by 20-30% over compared to . Although cohort studies report no increased poorer outcomes from —such as 82% recovery at two years without initial neuroleptics—these lack and controls, introducing high risk that precludes firm causal conclusions on . Critics, including systematic reviews, contend this selective strategy assumes factors dominate , potentially under-treating biologically rooted episodes where delays correlate with worse long-term functioning in broader meta-analyses. Ethical concerns arise from the heavy reliance on involvement, which can expose vulnerable individuals to dynamics or peer influences during crises, undermining therapeutic neutrality and . Network meetings, while promoting shared decision-making, risk power imbalances where dominant voices—familial or professional—marginalize patient autonomy, particularly in acute states impairing judgment; qualitative reports indicate some participants experience confusion or mistrust when sessions devolve into non-collaborative exchanges. Furthermore, implementing Open Dialogue outside resource-intensive systems raises equity issues, as incomplete fidelity (e.g., inadequate training) may yield suboptimal outcomes without the safeguards of evidence-based alternatives, potentially violating principles of non-maleficence amid unproven global adaptations. Proponents align it with via dignity-focused dialogue, yet absent robust trials, this positions it as an experimental paradigm with uncertain risk-benefit for diverse populations.

Comparisons and Alternatives

Relation to Biomedical and Other Psychosocial Models

Open Dialogue (OD) fundamentally contrasts with the of , which attributes symptoms primarily to neurobiological dysfunctions amenable to and institutionalization, by framing psychotic crises as relational and contextual phenomena embedded in social networks rather than isolated brain disorders. In OD, treatment prioritizes immediate, home-based dialogic engagement with the patient's and connections to generate shared meaning, deferring diagnostic labeling and reserving psychotropic s for later use only if dialogue proves insufficient, typically at low doses and short durations as adjuncts to processes. This approach critiques the biomedical emphasis on rapid symptom suppression through neuroleptics and hospitalization, which can exacerbate and , instead aiming to reduce overall —evidenced in implementations where initial neuroleptic use ranged from 28-50% but dropped to 11-29% over treatment—while promoting de-medicalized, everyday language to avoid psychiatrization. Despite these divergences, does not wholly reject biomedical elements; it integrates and crisis interventions within a flexible "treatment web" when needed, subordinating them to the core to maintain relational and patient agency, as opposed to the biomedical model's hierarchical, expert-driven . Proponents argue this hybridity aligns with recovery-oriented care, de-emphasizing biomedical individualism in favor of systemic understanding, though critics note potential risks in delaying medications for biologically driven psychoses. Relative to other psychosocial models, OD shares foundational influences with systemic family therapy—such as circular questioning—but diverges by emphasizing polyphonic to foster emergent meaning rather than directive behavioral change or family restructuring. Unlike cognitive-behavioral therapy for psychosis, which employs structured techniques to challenge delusions, or psychoeducational programs that deliver diagnostic information to prevent relapse, OD tolerates uncertainty and client-led narratives without predefined goals, viewing network involvement itself as the therapeutic process rather than a means to an expert-defined end. It also aligns partially with narrative therapy's constructivist ethos but prioritizes responsive, multi-voiced exchanges over therapist-guided re-authoring, rendering OD less structured and more improvisationally oriented than many established interventions. This relational primacy allows OD to encompass elements of various therapies while maintaining distinctiveness through its organizational mandate for rapid, inclusive meetings.

Cost-Effectiveness and Systemic Impacts

Evaluations of () in indicate potential cost savings primarily through reduced and use. A nationwide register-based study of adolescents treated under OD from to found average 10-year per capita treatment costs of €439, compared to €539 in a matched control group receiving standard care, with the difference driven by lower treatment expenses (€87 vs. €201) and costs (€3 vs. €8). Hospitalization averaged 7 days in the OD group versus 20 days in controls, contributing to these savings despite OD serving a larger volume. Disability-related expenditures also favored OD, with 7% of participants receiving allowances versus 11% in controls, and mean allowance costs of €3,627 versus €5,522 per recipient over 10 years. For adults, observational data from Western Lapland's implementation show analogous reductions in long-term service use, including fewer psychiatric bed-days and lower neuroleptic prescriptions, correlating with decreased disability pension reliance compared to national averages. However, comprehensive cost-benefit analyses remain limited, with no significant differences emerging after statistical adjustments in some models, and ongoing randomized trials in the UK aim to clarify net economic effects against treatment as usual. Systemically, OD promotes a shift from hospital-centric to community-network models, reducing overall psychiatric demand in implementing regions like Keropudas, where average episodes shortened without compensatory increases in acute admissions. This has lowered chronic incidence and risks, as seen in Danish adaptations with 26% reduced and fewer visits at 10-year follow-up. Implementation demands upfront investments in staff training and co-therapy, potentially straining resource-limited systems, but yields broader efficiencies by integrating social networks and minimizing coercive interventions, fostering sustainable care continuity. scaling faces barriers in biomedical-dominant frameworks, yet data suggest net resource reallocation toward outpatient and preventive services.

Global Adoption and Future Directions

Implementation Challenges Outside Finland

Efforts to implement Open Dialogue () outside have encountered substantial barriers, primarily stemming from differences in healthcare , professional requirements, and organizational cultures that contrast with the cohesive, community-based in Western Lapland. A global scoping survey of 142 OD teams across 24 countries identified that only 42% had more than 50% of staff trained in the approach, while 22% reported no ongoing , limiting to core principles such as immediate response and network meetings. These implementations often occur in outpatient settings (64% of teams) and alongside biomedical or other models, which professionals noted could impose additional workload burdens without achieving the systemic integration seen in . Training demands represent a core challenge, as OD requires extensive interdisciplinary preparation to foster dialogic practices, yet high staff turnover and resource limitations hinder sustained competence. In , where 77 trainings have been conducted since 2007 involving approximately 30 participants each, fewer than half of staff in 45% of facilities received full , compounded by turnover that disrupts team continuity. Similarly, in countries like , and , adaptations for local contexts—such as integrating OD into or refugee services—have not replicated all seven Finnish principles, with staff reporting anxiety over role shifts, self-disclosure, and interdisciplinary collaboration. No system outside Western Lapland has fully enacted these principles, highlighting fidelity issues in scaling the model. Systemic and financial constraints further impede adoption, particularly in fragmented healthcare environments where rapid, 24/7 access to multiprofessional teams is not feasible. German teams, for instance, reported that 71.1% could not provide help within 24 hours due to sectoral divisions, relying instead on partial adaptations like treatment conferences under temporary contracts. In the UK, clinicians in peer-supported OD pilots described organizational fragmentation, hierarchical resistance, and insufficient resources as key obstacles, with non-trained staff often undermining the flattened power dynamics essential to the approach. Service users echoed these difficulties, with some experiencing network meetings as distressing or lacking structure, such as one participant noting family members' hesitation to speak openly in the patient's presence. Funding models, including billing structures in systems, exacerbate these issues by not accommodating the time-intensive nature of continuous meetings. Cultural and practice-related hurdles, including tolerance for and shifts from expert-driven to collaborative , add to implementation difficulties. UK clinicians found reflecting aloud and containing therapeutically valuable yet challenging, calling for broader and buy-in to counter . Internationally, power dynamics in psychiatric hierarchies and professionals' discomfort with have been cited as persistent barriers, often necessitating models like peer-supported variants that dilute original structural elements. These adaptations, while enabling partial uptake, underscore the causal role of contextual mismatches in preventing outcomes comparable to Finland's, where low caseloads and cultural homogeneity facilitate adherence.

Ongoing Research and Adaptations

Recent randomized controlled trials, such as the ODDESSI study in the , are evaluating 's efficacy compared to treatment as usual for adults experiencing first-episode , with preliminary qualitative data indicating improvements in dialogical processes but awaiting full outcome results on rates and service use. The HOPEnDialogue project, launched to standardize across implementations, connects ongoing in multiple countries to assess fidelity and outcomes, emphasizing the need for consistent metrics in non-Finnish settings. Adaptations outside include integration with coordinated specialty care for early in the United States, where a five-year pilot in a demonstrated feasibility, reduced hospital stays, and high satisfaction, though long-term randomized data remain limited. In , efforts face challenges like constraints and cultural differences in multidisciplinary , with qualitative studies highlighting obstacles in achieving full network meetings but noting potential for . A global scoping survey of over 50 services identified variations in practice, such as shorter crisis meetings and less emphasis on antipsychotics, underscoring the need for tailored adaptations to local healthcare systems. Emerging research explores extensions beyond , including peer-supported models where trained peers facilitate dialogues, reporting enhanced relational dynamics in qualitative evaluations from . Studies also investigate Open Dialogue's role in improving in families, proposing mechanisms like shared polyphony to foster and reduce . Fidelity assessments emphasize need-adapted modifications, cautioning that deviations from core principles—like immediate response and tolerance for uncertainty—may dilute effects, as evidenced in a 2025 discussion paper. programs worldwide, evaluated through practitioner experiences, show shifts toward practices but require sustained supervision to embed changes.