Trauma-Informed Approaches in Primary Healthcare and Community Mental Healthcare: A Mixed Methods Systematic Review of Organisational Change Interventions

,


Introduction
Psychological trauma has a devastating impact on the health of individuals, communities, and societies [1].Traumatic experiences can be caused by single events (e.g., sexual assault and unexpected family death) or chronic phenomena (e.g., adverse childhood experiences (ACEs), domestic abuse, community violence, and historical trauma) [2].Structural inequalities (e.g., healthcare, economic, gender, and racial disparities) may exacerbate efects of these traumatic experiences [3].
Lifetime traumatic events are associated with risk-taking behaviours, poor health, adverse socioeconomic outcomes, and increased use of primary care and mental health services [1].Coercive practices and invasive procedures within healthcare services (e.g., removal of choice regarding treatment, judgemental attitudes following a disclosure of abuse, and lack of accessible services) can retrigger or retraumatise both patients and healthcare staf [4].As a result of empathetic engagement with trauma survivors, health-care providers can experience secondary traumatisation and/or vicarious trauma [5].
Over the last two decades, a trauma-informed approach has gained momentum as a framework for organisational (synonym system) change interventions that address the high prevalence and impact of trauma among health-care providers and users.Te approach difers from standard "trauma blind" service delivery by integrating 4 Rs throughout healthcare organisation: realising and recognising the impacts of trauma on patients and staf, responding by integrating knowledge about trauma into policies and practices, and creating environments and relationships that prevent retraumatisation and promote physical and emotional safety for all [6].Te framework of a trauma-informed approach is not a protocol but high-level guidance for organisational change interventions that can be adapted to any health service.Although diferent authors used difering terminology and defnitions, they mostly aligned with the philosophy and principles of the trauma-informed approach proposed by Harris and Fallot [7] and developed further by the US Substance Abuse and Mental Health Services Administration (SAMHSA) [8].Subsequent framework developments drew attention to the intersection of individual and interpersonal trauma and structural inequalities [4,9,10], universal applicability of the trauma-informed approach [11], benefts to patients and staf [12], and application to services other than mental health and addiction [9,13] (Supplementary materials S1).
Its proponents consistently highlighted the organisational level of a trauma-informed approach, requiring changes in the structure and culture of the organisation (organisational domain).Tese organisational changes should precede changes in clinical practices (clinical domain) [7].Becoming a trauma-informed organisation is described as a transformation process rather than a one-of activity.Te transformation work is guided by the SAMHSA six key principles of safety, trust, collaboration, choice, empowerment, and cultural sensitivity.Tese principles can be implemented through varied intervention components and activities tailored to organisational needs, abilities, and preferences and to the wider contexts [8].One contested component is screening for a history of traumatic events in adult healthcare settings [14].Most authors consider it as an essential component [7,8,10], while some think that disclosure of violence and trauma is not the goal of a trauma-informed approach, and service providers do not necessarily need to know about peoples' lived experiences to provide appropriate healthcare [12].Te conceptual mutability of a trauma-informed approach framework and lack of empirical evidence for efectiveness has been challenged [15].Tese and the various defnitions and applications might have contributed towards misconceptions about trauma-informed approaches at the organisational level, for example, confusion between universal trauma-informed organisational change interventions for all staf and patients and trauma-specifc treatments for people with consequences of trauma [16].
A growing body of literature, policies, and guidelines recommends trauma-informed approaches in healthcare organisations and health systems; however, the evidence base for the efectiveness is still being assessed [17][18][19][20].Our pilot searches and consultations with experts found extensive literature on articulating trauma-informed approaches, and how and why we should embrace and evaluate them.We identifed a growing market of training and certifcation on trauma-informed approaches.In contrast, we found a small number of evaluations of the efectiveness of traumainformed organisational change interventions within healthcare.Currently, studies of standalone training interventions about trauma-informed care without any changes at the organisational or wider system level dominate the evaluation literature [21].While a few evaluations of trauma-informed organisational change interventions were conducted in secondary mental healthcare [22] and services for children [23,24], we found no systematic reviews of the trauma-informed approach in adult primary care and community mental healthcare.Tese services are a patient's frst point of contact with a health system [25].
Tis systematic review is part of a programme of research on trauma-informed health systems (TAP CARE study).We aimed to systematically identify, appraise, and synthesise the empirical evidence on trauma-informed organisational change interventions in primary healthcare and community mental healthcare to understand: (1) What models of trauma-informed organisational change interventions have been applied?
(2) What are the efects of these interventions on psychological, behavioural, and health outcomes in health-care providers and adult patients?

Materials and Methods
2.1.Design.We registered study protocol with PROSPERO [26] and have published it elsewhere [27].In brief, we conducted a mixed methods systematic review with a results-based convergent synthesis [28].Te authors' positionality within the critical realism paradigm [29] infuenced decision to treat quantitative and qualitative fndings equally and do not undertake data transformation.Tis report follows the PRISMA 2020 statement [30,31].

Patient and Public Involvement.
In line with the key principles of a trauma-informed approach [8], we involved people with lived experience of trauma in each stage of the review.Te public advisory group included eight people with lived experience of trauma.Te professional advisory group included ten people who plan, fund, and deliver health services.Both groups discussed research questions and listed outcomes that they viewed as meaningful to patients, service providers, managers, and funders of services.Te professional group also developed a list of UK's primary and community mental health services.We met with the advisory groups every six months to consult on data extraction, logic model refnement, interpretation, and dissemination of study fndings.

Development of a Logic
Model.We used the measurement model for trauma-informed primary care [32] as a foundation for our logic model and incorporated data from background literature in version 1 [27].Version 2 incorporated fndings at the data extraction stage.Version 3 incorporated revisions at the data synthesis stage and included the following constructs: (i) A component of the trauma-informed organisational change intervention categorised by the SAMHSA ten implementation domains [8]. (ii) An intermediate psychological (cognitive or afective) or behavioural outcome regarding traumainformed care that the components might infuence (e.g., provider readiness or practices) categorised by the four-level framework of the healthcare system (individual patient, care team, organisation, and political and economic environment) [33].(iii) A long-term health-related outcome/phenomenon of interest that the intermediate outcomes/phenomenon of interest might infuence (e.g., patient or provider mental health) categorised by the fourlevel framework of the healthcare system [33].(iv) A moderator-a factor that could afect either positively or negatively, the link between a component and any outcome (Figure 1).

Search Strategy and Selection Criteria.
Based on previous systematic reviews [34,35] and the expertise of the research team and advisory groups, the frst reviewer (SD) developed a search strategy combining MeSH and free-text terms.SD conducted scoping exercises in diferent databases to maximise the search strategy's sensitivity and specifcity.Te search terms were modifed and tailored for fve electronic bibliographic databases: Cochrane Library, MEDLINE, EMBASE, Cumulative Index to Nursing and Allied Health Literature (CINAHL), and PsycINFO.We limited the search to primary studies published between January 1990 and February 2020, updated in June 2021 (Supplementary materials S2).SD searched the PROSPERO database for relevant systematic reviews in progress, and the ETOS library and ProQuest for dissertations.Additionally, SD conducted a grey literature search on websites of organisations involved in development and application of trauma-informed approaches.SD and NVL checked references and citations of included papers.NVL contacted corresponding authors, subject experts in trauma-informed approach, and study advisory groups for additional reports.We included primary studies of any design that evaluated a trauma-informed organisational change intervention in primary care or community mental healthcare (Table 1).

Study Selection.
We used Rayyan software [37] to combine, export, and screen the results of the database searches.Te frst reviewer (SD) and a second reviewer (AB or NVL) independently screened titles and abstracts and full reports against study inclusion criteria (Table 1).Te reviewers met and resolved discrepancies through discussion.Where they could not reach consensus, senior team members (GF and JM) acted as third reviewers.We included multiple reports of the same study if they contained new information and collated multiple reports so that each study was the unit of analysis.Te earliest most detailed report was used as study ID.
2.6.Data Extraction.SD adapted a data extraction proforma from previous systematic reviews.For each quantitative outcome, we extracted type of measure and efect estimates as reported in the primary study.If a follow-up measure was reported repeatedly, we extracted all results.If a study recruited a mixed sample or had multiple sites, we only extracted data relevant to adult primary care or community mental healthcare.SD extracted data and NVL checked and reconciled the forms and asked all corresponding authors to check.Five of the six authors responded.
We treated included reports as primary qualitative data and used NVivo 12 to simultaneously extract and code data on intervention characteristics and qualitative phenomena of interest (i.e., perceptions of intervention efects or factors that might infuence intervention efects).We used the framework synthesis method recommended for addressing applied policy questions [38].Our initial coding frame included constructs from the SAMHSA framework of traumainformed approach [8], our logic model, and the four-level healthcare system model [33].Two reviewers (NVL, KT) deductively coded intervention description, participants' quotes, and authors' interpretations relevant to our research questions.First, NVL and KT independently manually coded two reports and met to discuss the codes.Ten, NVL imported the framework into NVivo and completed the coding, refning the framework throughout this process, and grouping codes into themes.2.7.Quality Appraisal.We conducted quality appraisal as part of data extraction to indicate methodological limitations in each included study.Since we included studies of multiple designs, we used the Mixed Methods Appraisal Tool (MMAT) [39].SD completed the MMAT checklist for each study, NVL checked and reconciled through consensus.

Data Synthesis.
We conducted a results-based convergent synthesis [28] at three stages: (i) concurrent quantitative and qualitative syntheses, (ii) integration of fndings from the two syntheses through a line of argument, and (iii) mapping onto a logic model [40].At stage one, SD synthesised quantitative results in tables and descriptive summaries.Due to the variation of intervention models, measures, and outcomes, we could not conduct a metaanalysis.NVL grouped deductive codes into themes and wrote descriptive accounts with illustrative quotes.
At stage two, NVL and SD displayed quantitative and qualitative syntheses in tables and developed lines of argument for integrated outcome domains, intervention mechanisms, and moderators.We judged intervention efects by change in any quantitative outcome and/or participant perception of change reported in the primary studies.In the quantitative synthesis, we used authors' interpretation of their results based on p values, 95% confdence intervals (CI), or point estimates.In the qualitative synthesis, we summarised participants' quotes and authors' interpretations of primary data about perceived intervention efects.We categorised measured and perceived efects as improvement, mixed efect, nil efect, and negative efect/ harm.We ascribed a mixed efect when one or more, but not all measures of the same outcome changed under the same intervention.If diferent studies reported contradicting fndings on the same outcome, we categorised such evidence as conficting.
At stage three, NVL mapped the integrated lines of argument onto the constructs of the logic model.Te fnal logic model only included constructs and items that were supported by evidence from the included studies (Figure 1).

Study Characteristics.
Te primary studies used nonrandomised quantitative and qualitative designs.Of six studies, three were from the US [48,50,52] and one each from the UK [41], Canada [45], and Australia [42].Tree    Although the seminal paper introducing the philosophy and principles of trauma-informed mental healthcare was published in 2001 [7], professional stakeholders advised to extend searches to the preceding 10 years to capture early discussions of trauma-informed approach principles

Language
No language restrictions provided an English language abstract is available

Paper without abstract in English
Note.SAMHSA, the Substance Abuse and Mental Health Services Administration and United States Department of Health and Human Services.
Trauma-informed organisational change interventions were applied in public primary care clinics that served populations with high rates of trauma [45,48,50] and public [42,52] and third sector [41] specialist organisations that served women with a history of interpersonal violence.Two studies were single-site evaluations [41,42] and four were multisite studies [45,48,50,52].Te Equipping Primary Health Care for Equity (EQUIP) study took place in four Canadian primary care clinics that used the same intervention model [45].From the US Women Co-occurring Disorders and Violence Study (WCDVS), we included the Washington DC site with four community mental health centres [52].From the Dubay et al. report [48], we included three primary care settings that used diferent intervention models: Women's HIV Clinic San Francisco, Montefore Medical Group of 22 primary care practices New York, and Family Health Clinic Philadelphia.From the US Aspire to Realize Improved Safety and Equity (ARISE) evaluation [50] and Australian Young Women's Clinic [42], we extracted data for the patient group aged 18 and above.
Te total number of participants in the included studies was 117,447 patients and 137 health-care providers.Te number of patients ranged from 6 in qualitative service evaluation [48] to 116,871 in analysis of routine data [50].

Models of Trauma-Informed Organisational Change
Interventions.Te six studies evaluated eight diferent models of trauma-informed organisational change interventions.Although each model was tailored to the patient population, organisation, and wider contexts, all the models had sufcient common features for cross-study comparison.Our framework synthesis confrmed that each model aligned with the 4 Rs of the SAMHSA framework.Te models varied by their level of theoretical development, formalisation, and activities within each component (Supplementary materials S3).Tree of eight interventions used existing models of trauma-informed organisational change interventions.Te UK One-stop-shop Women's Centre had been using the Trauma-informed Service Systems model [7] for less than a year [41].Te Family Health Clinic Philadelphia had been adopting the Sanctuary Model [54], while the Women's HIV Clinic San Francisco had used the Trauma-informed Primary Care framework [55] for more than two years [48].Te Canadian team developed and evaluated the new EQUIP intervention for 24 months [45].Te other four sites (WCDVS Washington site, Montefore Medical Group, Young Women's Clinic, San Francisco Health Network Primary Care) applied tailored organisational change interventions up to 12 months [52], 17 months [50], more than 24 months [48] to 13 years [42].Te Sanctuary Model by Bloom has been operationalised as a certifed business model [54].(1) Quality improvement team; (2) Staf education; (3) Trauma-informed team-based clinical practice; (  (2) Organisational resources.
Health & Social Care in the Community Te intervention components varied in the extent to which they mapped onto the SAMHSA ten implementation domains [8] (Supplementary materials S3).Only EQUIP [45] and Trauma-informed Young Women's Clinic [42] included components from all ten implementation domains; the Montefore Medical Group intervention covered six domains [48].Te four common domains across all eight models were as follows: (i) budget, (ii) workforce development, (iii) identifcation and/or response to violence and trauma, and (iv) evaluation of change.All interventions were funded through project grants or joint fnancing.Te budgets covered training and ongoing support for all staf, trauma-informed practices, and changes in the physical environment.Te content, format, and duration of the training varied; the common features were delivery by external experts, tailoring to the organisational context and patient population, and booster sessions.Similarly, tailored self-care activities included mindfulness sessions, well-being days, and trauma-informed supervision.All models included on-site or external trauma-specifc treatments tailored to the population served.

Methodological Quality of Included Studies. Te quan-
titative nonrandomised studies and mixed methods study components were of moderate quality.Te methodological quality of qualitative study components was high (Table 2).Most studies had clearly defned research questions, which were addressed by the data collected [41,42,45,48,52].One study did not pose a clear research question [50].All four qualitative studies/component showed coherence between data sources, collection, analysis, and interpretation [41,42,45,48].Te three quantitative studies/component used appropriate sampling techniques and measures [45,50,52]; two had a good completion rate [45,52].Only one study considered possible confounding factors and confrmed that the intervention was administered as intended [45].Te only mixed methods study provided a design rationale and adequately integrated the qualitative and quantitative components [45].However, it did not address the divergence or inconsistency between components, nor did it fulfl the methodological quality criteria (Supplementary materials S4 and S5).

Efects of Trauma-Informed Organisational Change
Interventions on Patient or Health-Care Provider Psychological, Behavioural, and Health Outcomes.We found limited, mixed, and conficting evidence for the efects (or perceived efects) of trauma-informed organisational change interventions on 11 outcome domains, with an overall direction towards some improvement.Most evidence came from the controlled before-after WCDVS study [52], mixed methods EQUIP study [45], and three qualitative service evaluations [41,42,48].Te evidence for each intervention model was based on a single study.None used the same measures (Table 3, Figure 1).
Te studies reported some improvements in fve out of seven psychological and behaviour outcomes and in two out of four health outcomes.Only two studies reported both psychological and health outcomes [42,45].Although most interventions ofered training and self-care activities for staf, no studies measured provider health and well-being.No studies reported adverse events/harm among patients or staf.No studies evaluated cost-efectiveness.

Intermediate Psychological and Behavioural Outcomes.
We found limited evidence suggesting that some interventions may change organisational culture and create safe environments for patients and staf, potentially leading to improved perceived safety, patient disease management, and access to services.Tese changes were measured through assessing organisational readiness to provide trauma-informed care, provider sense of community, patient readiness for disease management and access to services, and patient and provider perceived safety.However, the evidence for the efect on provider behaviour and patient satisfaction was conficting.
(2) Provider Behaviour.In contrast, the evidence for change in collective behaviour and practices was conficting.While the ARISE programme reported a 22% increase in screening rates for depression, substance use, and interpersonal violence [50], Dubay et al. [48] found that in three US sites, health-care providers "had rich, nuanced views on the benefts and drawbacks of screening for trauma, which did not always neatly align with their organisations' ofcial policies."Several interviewees called screening for trauma "controversial," said they were "conficted" or had "mixed" feelings about it or said "there are pros and cons (p.24)".
Similarly, while Bradley [41] found that staf used and appreciated well-being activities, Dubay et al. [48] reported that "many interviewees said they did not adopt new selfcare techniques after trainings, but appreciated grantees" eforts to promote self-care (p.VII)".
(3) Patient Psychological Readiness for Disease Management.Two qualitative studies reported that patients felt in control of treatment [42,48].Four studies consistently reported improvement in patients' confdence in managing health conditions and self-confdence [41,42,45,48].
(4) Patient Satisfaction with Services.Two studies found qualitative evidence for improved satisfaction with services [41,42], while Dubay et al. [48] reported that "some patients felt frustrated by high staf turnover and by the lack of staf diversity in some practices (p.33)".
(5) Patient Access to Services.Tree qualitative studies reported improved access to care through on-site provision or referrals to external organisations [41,42,48].
(6) Provider and Patient Safety.Two qualitative studies reported improvement in perceived safety among patients and staf [41,42] suggesting that the interventions created safe environments.

Long-Term Patient Health
Outcomes.We found limited, mixed, and conficting evidence with regard to change in some patient health outcomes with an overall direction towards some improvement.Tree studies reported conficting evidence for four health outcome domains with improvement in two, nil efect in one, and mixed efects in one.Te EQUIP study [45,49] found strong evidence for the improvement in patient quality of life, chronic pain, depression, and PTSD symptoms at 18and 24-monthfollow-ups.Brooks et al. [42] found qualitative evidence for improvement in depressive symptoms.In contrast, the WCDVS reported mixed efect on mental health symptoms and severity of alcohol and drug problems; in this study, PTSD symptoms and alcohol remained unchanged at 6 months then improved at 12 months [47,52,53].
Intervention worked as a package of components in four studies [41,42,45,48].Two studies reported qualitative evidence explaining the challenging process of changing organisational readiness to provide trauma-informed care [45,48,51].In the EQUIP study [45], this shift happened through "surfaced tensions that mirrored those in the wider community, including those related to racism, the impacts of violence and trauma, and substance use issues.Surfacing these tensions was disruptive but led to focused organisational strategies (p.1)."Similarly, Dubay et al. [48] described tensions due to difering values and attitudes among staf groups.Furthermore, the EQUIP study reported evidence linking engagement with the intervention and increased awareness and confdence among staf [45,51].Bradley [41] found qualitative evidence for the link between traumainformed service and patients' satisfaction and provider sense of community.
Brooks et al. [42][43][44] reported some qualitative evidence suggesting that all components of the Trauma-informed Young Women's Clinic contributed to patient satisfaction with services through improving access to healthcare for young women from deprived communities.
Two studies reported some evidence for the link between intervention as a whole and outcomes at the level of individual patient.Te WCDVS study reported that "intervention condition and programme elements" led to improvement in women's trauma symptoms at 12 months (0.414 (95% CI 0.081 to 0.747)) [53].After using the Trauma-informed Women's Clinic, some participants reported increased self-confdence, confdence in care, perceived safety and support, and improved mental health [42].
Women-only space, when services are delivered by female providers to female patients, increased patient satisfaction with services through improving access to care, perceived safety and support, and self-confdence [41,42].However, Brooks et al. [42] reported that "sometimes people get the wrong idea of the place, they think it is for people that are man-haters (p.15)".
EQUIP care dose was proposed as an indirect mechanism to patient health outcomes [49].When patients perceived their care as more equity-oriented and trauma-informed, they felt more comfortable and confdent in that care.Tis led to patients feeling more confdent in their own ability to manage health problems.Over time, these psychological changes translated into better quality of life and less depression, trauma symptoms, and chronic pain.
Staf education was proposed as a mechanism for change in provider readiness.Two qualitative evaluations found that educating all staf about trauma-informed approach and self-care led to improvement in provider knowledge, skills [41,48], and relationships [48].
Tailoring staf education to organisational and wider context was the EQUIP mechanism.Educational content and format were adapted to address clinic-specifc needs, capacities, and priorities.Such tailoring contributed to staf feeling unifed and to increased readiness for providing trauma-informed care [51].
Staf self-care activities contributed to staf perceived safety [41].
Safe environment created by the women-only policy, staf nonjudgemental attitudes, and confdential services were reported as a link to patient safety and satisfaction through building trusting patient-provider relationships [41] and increasing patient perception of safety [42].
Shared decision making led to patient safety and satisfaction through education, feeling of control over treatment, and empowerment [42].

Moderators.
We developed lines of argument for the two overarching themes with seven subthemes summarising factors that facilitated or hindered intervention efects: (i) contextual factors at the levels of wider political and economic environments, organisation (culture, resources), and individual patient (social determinants of health) and (ii) intervention factors at the organisation level (intervention components and implementation process) (Figure 1, Supplementary material S6).
(1) Contextual Moderators.Tree qualitative studies identifed political and economic conditions that could afect intervention efects [41,45,48].Tese were relevant to values, regulatory, and fnancial regimes within the health system [41,45,48,51].Rigid policies, governance, and proft-driven business models made it difcult for healthcare providers to fnd time for training participation, selfcare, and other organisational change activities [45,48,51].Canadian and US providers acknowledged difering values between their organisations and the wider health system that acted as a barrier [45,48].Bradley [41] reported similar conficting regulations and values in the UK third sector.EQUIP "participants noted the amplifying infuence of other trauma-informed initiatives in the community" [51] (p.6).
Two studies described negative moderators at the organisation level: unsupportive culture with high pressure environment, disconnected leaders, hierarchical structure, difering values, and power imbalances [45,48,51].In contrast, a supportive work environment and organisational values aligned with the principles of trauma-informed approach were described as positive moderators [51].One study reported that their facility had limited capacity for changes in the physical environment [41].
One study described how wider contextual moderators had negative efect at the individual patient outcomes.Path analysis suggested that the EQUIP intervention was less efective for patients with experiences of intersecting structural violence (i.e., fnancial strain and discrimination) [49].
(2) Intervention Moderators.Barriers to the intervention implementation were reported most frequently.Four studies described poor engagement of some members of staf in intervention activities [48,51], inadequate funding, and dependence on project grants [41,42,51] as major barriers to sustainable organisational change.In contrast, two studies reported factors enabling successful workforce development leading to changes in the organisational culture.First, health-care providers thought that collective learning through interprofessional conversations worked better than didactic methods [51].Second, they highlighted the importance of the leadership buy-in [48].Tird, providers emphasised the importance of involving all staf in educational activities [48].
Two studies of services for women with a history of interpersonal violence found some evidence that components of the organisational change intervention can have a modifying efect.Te hierarchical linear modelling in the WCDVS produced conficting results.While receiving integrated counselling for trauma, mental health, and substance use resulted in better patient health outcomes, receiving more study services resulted in less improvement [53].Bradley [41] quoted "one member of staf emphasised the importance of prioritising staf well-being as equal to the support provided to women (p.16)."

Discussion
4.1.Principle Findings.Tis mixed methods systematic review of six nonrandomised studies which assessed eight models of trauma-informed organisational change interventions in primary care and community mental healthcare found limited, mixed, and conficting evidence for their efects on patient and health-care provider psychological, behavioural, and health outcomes with an overall direction towards some improvement.Healthcare organisations tailored diferent models of trauma-informed organisational change to their needs, abilities, and preferences.Te most common components included an allocated budget, ongoing training and support for all staf, identifcation and treatment for trauma, and evaluation.Four studies reported improvement in provider readiness to deliver trauma-informed care and improvement in their sense of community.However, two studies reported that only some providers used self-care activities and screened for traumatic experiences.Four studies reported some improvement in patient readiness for disease management and access to services; however, the evidence for patient satisfaction was conficting.Two studies found that patients and providers felt safe.While one study reported some improvements in patient quality of life and chronic pain, three studies reported mixed and conficting fndings regarding efect on mental health symptoms and alcohol use, and one found no efect on drug problem severity.No studies measured adverse events/harm, cost-efectiveness, or staf health and well-being.Te limited evidence for programme' mechanisms suggested that interventions may work either as a whole or through separate components-staf education tailored to the local context, self-care activities for staf, safe environments, and shared treatment decision making.We identifed contextual and intervention factors that may moderate intervention efects.Contextual moderators included health system values, policies, governance, and business models, wider trauma-informed programmes, organisational culture, and patient social determinants of health.Intervention moderators included buy-in and engagement from all staf, collective learning through interprofessional conversations, equal attention to well-being of staf and patients, and sustainable funding.
Our frst important fnding is that the empirical evidence base for the efectiveness of trauma-informed organisational change interventions in primary care and community mental healthcare is very limited.Despite exhaustive searches, we only identifed three nonrandomised quantitative studies and three qualitative service evaluations of diferent intervention models.One of the reasons for the evidence gap could be the methodological challenges of evaluating organisational change interventions within 12 Health & Social Care in the Community Health & Social Care in the Community 14 Health & Social Care in the Community complex health systems.However, the literature ofers varied tools and guidance on how to evaluate trauma-informed organisational change interventions [32,35,56,57].It is possible that some evaluation reports have not been made public.
Our second important fnding is that despite heterogeneity in the included models and evaluation designs, we found comparable domains for intervention components, outcomes, mechanisms, and moderators.By mapping intervention components on the SAMHSA trauma-informed approach framework [8], we showed that all eight intervention models were built on the 4 Rs assumptions and all included components within the same domains: (i) budget allocation, (ii) training and workforce development, (iii) identifcation and/or response to violence and trauma, and (iv) evaluation of the organisational change.Such similarities can explain convergence of efects when these were detected.Tey may work through the same mechanisms of changing provider readiness, sense of community, and safety to changes in patient readiness, satisfaction, safety, and health.Our fndings on outcome domains and possible mechanisms are in line with recent systematic reviews of trauma-informed interventions at the organisational level that did not include any of our studies [6,21].Both reviews found conficting efects on provider readiness and practices regarding provision of trauma-informed care and service user perception of care.None of our studies used validated measures for evaluating organisational readiness, culture, and performance identifed by Melz et al. [6].Te overlapping mechanism for increased provider readiness was staf education and ongoing support.
Our third important fnding about increased perceived safety and sense of community among health-care providers indicates positive changes in organisational environments, relationships, and culture which may facilitate and support subsequent changes in clinical practices.Tis fnding supports the recommendation for the organisation domain of the trauma-informed approach to be the precondition which enables and helps sustain traumainformed changes in clinical practices by individual health-care providers [7,8].
Our fourth fnding supports the proposition that a universal trauma-informed approach does not have to include screening component to improve patients' experiences and outcomes.In our review, the fve models which included screening for trauma (WCDVS, ARISE programme, Women's HIV Clinic, Montefore Medical Group, and One-stop-shop Women's Clinic) and the three models that did not include screening for trauma (EQUIP, Traumainformed Young Women's Clinic, and Sanctuary Model) reported improvement in some patient outcomes (readiness for disease management, safety, and health).No studies explored the mechanisms linking screening to health outcomes or harm among patients and staf.Providers had conficting views on the acceptability and feasibility of screening all patients for traumatic experiences.Tis fnding on the uncertain evidence for the efectiveness and safety of screening for traumatic experiences is in line with the recent systematic reviews [6,21].

Strengths and Limitations.
We conducted a methodologically robust systematic review with two reviewers working in parallel at each stage.Our rigorous search strategy included both peer-reviewed and grey literature without language restrictions other than inclusion of an English abstract.Tis resulted in a global view of traumainformed organisational change interventions in primary and community mental healthcare.Additionally, we contacted and received responses from study authors to identify other relevant studies and to clarify information regarding data extraction and quality appraisal.A methodological limitation is that we used search terms based on trauma-informed terminology introduced in early 2000, which meant that earlier studies meeting the inclusion criteria may have been excluded, as they were not labelled as "trauma-informed'.We addressed this limitation through seeking input from our public and professional advisory groups when designing the search strategy.We involved people with lived experience and professionals in diferent stages of the review to ensure that our fndings are relevant and benefcial to them.By using a logic model to map review fndings, we produced fndings that are understandable to health-care providers and policy makers.Exclusion of papers without an English abstract might have resulted in missing relevant studies reported in other languages.
Te evidence we found is very limited and uncertain due to the small number and nonrandomised designs of the primary studies.Tat said, the included studies were generally characterised by good sample size.Nonrandomised studies provide weaker evidence of causal efects of interventions on outcomes.Although we tried to hypothesise causal links through mapping onto our logic model, these are assumptions supported by six studies at the bottom of the hierarchy of evidence and further high-quality research in this area is warranted.

4.3.
Implications for Policy, Practice, and Research.Any generic framework for trauma-informed approach should be contextually tailored for organisational needs, abilities, and preferences.If primary care and community mental healthcare organisations integrate trauma-informed assumptions and principles across at least six implementation domains, they may change organisational culture and create safe environments for staf and patients potentially leading to improvement in patient disease management and satisfaction, access to services, quality of life, and chronic pain.Every trauma-informed organisational change intervention should have funding for an embedded evaluation and agreement on target outcomes and measures that are evidence based and theoretically informed.Future research exploring trauma-informed approaches in primary care and community mental healthcare should include randomised designs and validated measures to capture changes across individual, team, organisation, wider system levels, and enable meta-analysis.Studies should also evaluate adverse events/harm, provider health, and cost-efectiveness.
Health & Social Care in the Community

Conclusions
Trauma-informed organisational change interventions in primary care and community mental healthcare may improve provider readiness and sense of community, patient readiness for disease management and access to services, provider and patient safety, and some patient health outcomes, but the evidence is very limited and conficting.
Training for all staff (ii) On-going support for all staff (iii) Organisational change activities for all staff (committee, working group, meetings) (iv) Trauma-informed supervision (v) Self-care for all staff (vi) Hiring trauma-aware staff 8. PROGRESS MONITORING (n=5) (i) System for monitoring (ii) Feedback from staff and patients (iii) Evaluation of staff experiences (iv) Quality improvement with feedback loop INDIVIDUAL PATIENT CONTEXT (n=1) (i) Financial strain (ii) Discrimination 4. SURVIVOR ENGAGEMENT IN (n=6) (i) Organisational change activities (ii) Decision making (iii) Power sharing (iv) Treatment choice 5. CROSS SECTOR COLLABORATION (n=6) (i) Partnerships with agencies (ii) Referrals to trauma-specific services (iii) Cross-sector training 9. BUDGET FOR (n=6) (i) Workforce development (ii) Cross-sector training (iii) Peer specialists (iv) Changes in physical environment 10.EVALUATION OF TI CHANGE (n=6)

Figure 1 :
Figure 1: Logic model for trauma-informed organisational change interventions in primary and community mental healthcare.Note: (n � ), number of studies that provided evidence.↑, improvement.0 nil efect.↑↓, mixed or conficting evidence.

Table 1 :
[25]y inclusion and exclusion criteria.primarycareand/or community mental healthcare services in public, private, and third sector[25]Organisations that do not provide primary care and/or community mental healthcare services Studies with mixed samples if they reported outcomes for the primary care and/or community mental healthcare subsample irrespective of the proportion of the subsample Studies with mixed samples that did not reported separately outcomes for the primary care and/or community mental healthcare subsample [8]address the heterogeneity in terminology and defnitions, we used the SAMHSA's framework for trauma-informed approach that is grounded in a set of four assumptions, six key principles, and ten implementation domains (supplementary materials S1)[8]Comparator No trauma-informed organisational change intervention We included studies without a control group Quantitative outcomes Main outcomes: any psychological, behaviour, health outcomes at the organisational level No outcomes reported Additional outcomes: any psychological, behavioural, and health outcomes at the individual level Psychological outcomes can be measured through cognitive (e.g., knowledge and skills), afective (e.g., attitudes), behavioural (e.g., clinical practices) outcomes

Table 3 :
Evidence for efects and perceived efects of trauma-informed organisational change interventions on provider and patient outcomes.

Table 3 :
Continued.Mixed efect, when one or more, but not all measures of the same outcome changed under the same intervention.PTSD, post-traumatic stress disorder.CI, confdence interval.