Towards a framework for trust-based management of municipal home care. A realist informed evaluation.

Background This study reports on a two-step service innovation project commissioned by the City Council of Oslo, coined as the Trust Model (TM). The aim was to develop a trust-based management model in municipal home care as an alternative to the purchaser-provider split (PPS), which has been the prevalent model in Norwegian and most European health services since the introduction of New Public Management in the 1990s. The TM was developed through a comprehensive participatory process. Methods The objectives were to a) identify important mechanisms facilitating or preventing the development and implementation of the trust model, and b) identify participants perceived outcomes. A realist informed process evaluation using mixed methods was performed in two iterations. The first (autumn 2016) included three districts, three teams and 80 patients. The second (2017) included four districts, eight teams and 160 patients. Results The TM was developed across the four districts. Team members, team leaders and managers found the TM promising, and no one wanted to return to the PPS. Patient satisfaction improved significantly during the first iteration. The TM iterations were complex, involving a series of agents interacting on different levels; interactions which in turn influenced the processes and ongoing interactions. The evaluation displayed a variety of interpretations of the TM and of trust as an organizational value. A series of needs for improvements were identified, including (1) develop a clear description of the model and its managerial principles, (2) develop a culture for trust-based management on all levels, (3) build team leaders’ competence, especially in trust-based management and coaching of team-members (4), clarify roles and responsibilities among team-members, (5) develop common procedures for allocation and distribution of services, and (6) build team-members communication competence in active listening and shared decision-making. Conclusions/implications institutional logics NPM) and an organization’s ability to with institutional hybridization. The topic

slowness of radical service innovation should be further explored.

Background
There is a reported decline in trust in Western health care systems (Abelson, Miller, & Giacomini, 2009;Gille, Smith, & Mays, 2014). This development seems to go parallel with the decline in professional autonomy in health care organizations, as a result of the dominant role of New Public Management (NPM) thinking over the last 20-30 years (Siltala, 2013;Simonet, 2015). In Scandinavia, various trust reforms are initiated on national and municipal levels in order to restore trust in professionals and the public sector in general (Bentzen, 2016a;Otterdal, 2018;SOU, 2018:47).
The purchaser-provider split (PPS) model has been the prevalent model in Norwegian and most European health systems since the introduction of New Public Management (NPM) in the 1990s. The PPS is a service delivery model in which third-party payers are kept organizationally separate from service providers (Tynkkynen, Keskimäki, & Lehto, 2013). In the Nordic countries, the public sector traditionally fills the role of both purchaser and provider, financed by global budgets and without specification of expected volume or quality. Since the early 1990s the need to introduce incentives to manage scarce resources more efficiently gradually emerged as a policy issue (Magnussen, Vrangbaek, Saltman, & Martinussen, 2009:15). In Scandinavia, Sweden was the main innovator, various purchaser-provider split arrangements were introduced in the late 1980s, in an effort to stimulate market-style competition, enhance efficiency within the public sector and widen patient choice (Saltman & Vrangbaek, 2009:79). In Norway, a purchaser-provider separation was first introduced for nursing and care services in the early 1990s (Martinussen & Magnussen, 2009:25).
When different PPS models were established in Norwegian municipal health services in the mid 1990s, the need for enhanced efficiency, quality and equity was the main rationale (NOU 2005:3;RO, 2004).
However, the recommendations to do so were ambiguous. The Government green paper introducing the PPS (NOU 2005:3) pointed to the possible risks of a too strict separation between the purchaser and provider, increased distance between the decision-maker and the service user, unwanted growth of bureaucracy and delay of decision procedures, which especially might disadvantage persons with complex care needs. Further, a national guide for implementation of the PPS in the municipal health and care services, published by the Norwegian Association of Local and Regional Authorities (KS), addressed possible downsides, and lined up 12 positive and 12 negative arguments, leaving it to the municipalities to make the decision (KS, 2004). The research literature on the effects of PPS on the quality of home care services is limited.

The Trust Model innovation project
This study reports from a two-step organizational innovation project in municipal home care, coined as the Trust model (TM). The aim of this project was to develop a trust-based and person-centred alternative to the PPS-model. The term "person-centred" here applies to healthcare practice, and an intention of increasing user involvement and shared decision making concerning care delivery (CCD, 2015).
The TM innovation project was politically initiated by the City Council of Oslo, inspired by the trust reform of the municipality of Copenhagen, Denmark (Bentzen, 2016a)

and the Dutch private
Buurtzorg model, providing home care through self-managed teams of nurses (Monsen & De Blok, 2013). Four of the 15 city districts of Oslo took part in the project, headed by a steering group consisting of senior managers and employee union representatives of the four districts. A key learning from the Copenhagen project was that organizational trust does not follow automatically from strategic decisions from the top (Bentzen, 2016b). Hence, a comprehensive participatory bottom-up process was set in motion across the four districts during 2016 and 2017 in order to develop proposals for a trust model in home care. The steering group decided on a proposal consisting of three core elements: a) organizing in small, self-managed and multi-disciplinary teams (about 12 persons), b) trusting the teams with full responsibility for all persons in need of home care in a limited geographical area (maximum about 150 patients), and c) taking the question "What is important to you?" as point of departure for patient involvement and care decisions. Aligned with the Buurtzorg model concerning self-managed teams, the TM deviates from this model both by its multidisciplinarity (vs uni-disciplinarity/nursing), the scope of team responsibility (about 150 vs about 50 patients per team), and by being an integrated part of a public, municipal health care system responsible for all service users in the area (vs private service delivery responsible for a selection of the service users). The person-centred approach of focusing on patient concerns and priorities ("What is important to you?") was rooted both in the participatory process and in national and local policy documents (NMHCS, 2015;TPSG, 2016). We here apply the term "patient" to all persons receiving home care, i.e. service users or recipients. The term "multi-disciplinary" here implies that the basic staffing of each team should include both the professions of nursing, physiotherapy and occupational therapy as well as home care workers and administrative officers (TPSG, 2017).
According to policy documents expectations with developing and implementing the TM were to achieve a) increased patient safety, satisfaction and participation, b) increased employee motivation and work satisfaction, and c) increased service flexibility, efficiency and quality (CCD, 2015;CCO, 2017;TPSG, 2016).

Trust and complexity
The concept of trust implicit in the terms "trust model" and "trust-based" is complex. On the macro level of the healthcare system, it applies to a political intention of improving the service quality and increasing citizen trust in the municipal services. On the meso level of the homecare organization where the innovation project was situated, it applies to the political intension of increasing healthcare professionals' responsibility, authority and area of discretionary powers. On the micro level of homecare practice it applies to the interpersonal relationships between the patient and healthcare professionals, and the increasing emphasis on person-centredness, involvement and autonomy.
Aware of these three levels of trust, the theoretical backdrop of the study is comprised of three complementary theories of trust, respectively Luhmann's systems theory, conceptualizing trust as a means to reduce social complexity (Luhmann, 1968), Mayer, Davis & Schoorman's model of organizational trust, suggesting ability, benevolence and integrity as the key preconditions of perceived trustworthiness (Mayer, Davis, & Schoorman, 1995;Schoorman, Mayer, & Davis, 2007), and Løgstrup's relational ethics, suggesting a natural human condition of basic trust to be the ontological root of moral responsibility (Løgstrup, 1956).

Methods Aim
The objectives of this study were to a) identify important mechanisms facilitating or preventing the development and implementation of the Trust-model; b) identify participants perceived outcomes, and c) propose a framework for trust-based leadership and organization.

Design and setting
The design was a realist evaluation (Pawson, 2013;Pawson & Tilley, 1997) using multiple qualitative and quantitative methods. The realist evaluation approach seemed particularly appropriate because of the complexity of the innovation project and its context, the tentative nature of the two iterations (named pilots), and the multiple actors, professions, tasks and patients involved. A realist evaluation is not a strict method or procedure, but rather a logic of inquiry guided by the question "What works for whom in what circumstances and in what respects, and how?" (Pawson & Tilley, 2004;Rycroft-Malone et al., 2016;Salter & Kothari, 2014).
According to realist evaluation thinking (Pawson & Tilley, 1997) mechanisms are underlying, often unobservable factors involving the reasoning of participants causing effects in a certain context. The specific context (from Latin, contexere, "weave together") of the trust model intervention was the complexity of organizational elements that are woven into the daily dynamic practice of home care to (mostly) older and vulnerable people, often with complex healthcare needs.
The theory implicit in the political decision of developing a trust model was that trust-based organizational change would increase employee motivation, patient satisfaction and citizens' confidence in the municipal health care. Put in realist evaluation terms, the value of trust was assumed to be a mechanism to create certain effects in the given context of municipal home care. By using a realist evaluation approach, we wanted to uncover what worked for whom, how the mechanism of trust worked in this specific context, and whether or not it produced the expected effects or outcomes. The structure of this realist process evaluation is presented in figure 1.

Figure 1: Process study structure
The study was undertaken in two iterations (TM pilots). The first (autumn 2016) included three districts, six multi-disciplinary self-managed teams and 80 patients. The second (2017) included one more district, four districts in all, eight teams and 160 patients. The districts had varied socio-cultural profiles and between 27,000 and 50,000 inhabitants, which implies that each one of them was the size of a large Norwegian municipality (more than 20,000 inhabitants). All districts were quite autonomous concerning management of their health and care services.
Within the overarching methodology of realist evaluation, we chose an open exploratory design, using both qualitative and quantitative methods. Table 1 gives an overview of the research activities, methods, participants in the specific activities and the timeline through the two iterations of the project.
Insert Table 1  The team member questionnaire To investigate the prerequisites and experiences of transition to trust-based, self-managed teams and new procedures based on what the patient considers important to him or her, we compiled a comprehensive questionnaire: Trust and self-determination were measuredwithfive questions about the degree of trust and selfdetermination at work. The response scale goes from 1 = "to a very small extent" to 5 = "to a very large extent".
Motivation at work was measured with four questions with a response scale from 1-5. person-centred culture. The PCPI is divided into three main areas: (1) prerequisites for person-centredness, (2) care processes and (3) care environment. The answer options range from 1 = "totally disagree" to 5 = "completely agree".

Determinants of innovation:
The team members' evaluation of the prerequisites for implementation were mapped, using the Measuring Instrument for Determinants of Innovations (MIDI) questionnaire (Fleuren, Paulussen, Van Dommelen, & van Buuren, 2013, 2014. This instrument captures four different categories or groups of determinants of innovation, those associated with (a) the sociopolitical context related to legal regulations (1 determinant), (b) the innovation itself (here: the TM; 7 determinants), (c) the users of the innovation, here the healthcare personnel, and the degree to which the personnel believe the innovation is relevant to his or her patients (11 determinants), and (d) the organization where the TM is implemented (10 determinants). The scale range is from 1-5.

Qualitative methods
Multiple qualitative methods were used in order to investigate the background, context and processes of the two iterations, aiming at building a picture of what works, how, why and in what circumstances for the many different agents during the two iterations; Document analysis of white papers, green papers, municipal policy papers, minutes of steering group meetings and process documents concerning development of the TM, Participatory observation in meetings (steering group, project leader group, etc) and workshops.
Workshop process interviews with team members in IGP format (individual-group-plenary). Three IGP process workshops were organized with about half of the team members present each time (about 40-50 persons), each with a specific topic. The workshop interviews were organized in three steps. including open questions about what worked well and less well). Second (group) team members were sharing their answers and reflections with each other. Third (plenary) each team wrote key points on a large piece of paper, put it up on the wall and presented them to the other teams for them to respond to, including their main concerns, ideas for improvements and a narrative illustrating their core experience so far. Data were collected through the questionnaires and observational notes. A reason for using the IGP process as research method was the assumption that it would also stimulate trust building and innovation across teams, which seems to be an outcome of IGP processes in interfirm networks (Gausdal, 2015;Gausdal & Hildrum, 2012). In addition, this enabled collection of rich process data and at the same time strengthening organizational learning and the participatory aspect of the intervention.
Leadership interviews: To investigate managerial-level experience of what works, how, why and in what circumstances, 10 individual in depth interviews were conducted at the end of the second iteration (December 2017) with leaders at different levels (the leader of the steering group, the main project leader, the team leaders, the local project managers, a unit leader and an employee union representative). The interview guide (appendix x) was developed from the preliminary analyses of the observational data, the patient and team member surveys, and the IGP process interviews. The interviews were conducted as open, exploratory conversations, using the interview guide as a checklist. The overall purpose was to explore the leaders' evaluation of the TM and the two-step iteration process, especially with regard to implementation in all districts.

Quantitative analysis
The questionnaire data were analysed with SPSS (IBM, 2019). Descriptive statistics were used to describe the samples. ANOVA-analysis for independent groups were used to determine group differences and possible change over time. Internal consistency of the scales and subscales were assessed by Cronbach's coefficient α (Streiner, 2003). For determining if a MIDI-determinant could be regarded a facilitator or a barrier; MIDI items to which ≥20% of participants responded 'totally disagree/disagree' were regarded as a barriers and items to which ≥80% of participants responded 'agree/totally agree' as facilitators (Kuunders et al., 2017). Cohens D was calculated to determine the degree of change (Lakens, 2013).

Qualitative analysis
The observational notes and transcriptions of the interviews were analysed according to phenomenological-hermeneutical procedures (Lindseth & Norberg, 2004). In addition, the taped interviews were listened to several times, in order to validate the interpretations. Preliminary research findings were presented to and discussed with the teams, the project leader group, the main project leader and the steering group. At the end of both iterations, preliminary research findings were presented to and discussed with the steering group and published in report format, including recommendations for further improvement of the trust model (Eide, Gullslett, Nilsen, Dugstad, & Eide, 2018;. Also the final report was presented to and discussed with managers and healthcare personnel. The results of these discussions were included in the data.
Combining the quantitative and qualitative data to construct a framework To construct the realist evaluation results we iteratively analysed the data in 4 steps: First we identified the possible core positive outcomes for the patients reported by the users and the teammembers. Secondly we identified core qualifications and competences the team-members needed to display in order to co-produce the results together with the patients. Thirdly we identified what the leaders needed to provide for the team-members' in order for them to perform the key tasks related to patient care. Fourth, we extracted the key prerequisites the whole system should afford in order to realize the TM.

Results-making Sense Of The Multitude Of Data Sources
This section is divided into two main sections. First the results from the different data sources; patients, team members and leaders, and the different methods; surveys, interviews (individual and group) and observations will be presented. The second section will suggest a framework for trustbased home care practice.

Part 1-results from different data sources
The patient survey We present the results of the patient surveys in the two iterations together to illuminate the development. All four districts participated in the first survey, but one of the districts (district 4) withdrew from participation in the first iteration (therefore 0 participants March 2017). Table 2 provides an overview of patient participants in the two iterations.

Insert table 2-the participating service users
The participating patients' trust in the services: Patients were largely confident that they would get assistance when needed with what was most important to them. They were also largely satisfied with the home care service. An average summary of the five questions can be viewed as a total assessment of the patients' experienced quality of the services, including confidence and trust. The answers are close to the category "largely agree". During the first iteration there was a statistically significant improvement in patient satisfaction. When the second iteration started, patient satisfaction was on the same level as at the end of the first and remained stable at that level.

Table 3-the users trust
The patients answered open questions about what they liked the most and least. What was most important to the patients were a) «That they come, do a good job, and are caring" and b) "That they are nice, positive and friendly". What they liked least was a) "That they do not keep appointments, are coming late", b) "Too many unknown people coming" and c) "That they have too little time".

The employee survey
Participants: Table 4 provides an overview of the participants in the employee survey in the two iterations. In the first iteration, 200 questionnaires were distributed to employees in the home care services, 50 per district (response 76%). At the beginning and end of the second iteration questionnaires were distributed to the 105 participating team members (response 73%). The respondents represented a broad spectrum of positions, competences and educational backgrounds. Table 4 Trust and self-determination: The employees wanted a high degree of self-determination at work.
They experienced a moderate degree of professional trust. The entire response scale (1-5) was used, indicating large variations on an individual level. (See Table 5) Table 5 degree of perceived trust and self determination At the start of the first iteration, 83% of the employees answered yes and 17 % no to the question whether they believed that TM will give better services for the patients than the PPS model. At the beginning of the second iteration the score was even higher, 96% yes and 4% no.
Motivation at work: The participants found that working with the TM greatly influenced their work positively and to a small extent negatively. The same tendency was found when they were asked about the motivation of the team as a group. From the end of the first until the end of the second iteration, there was a slight decrease in average work motivation, but significant difference between the districts.

Table 6 motivation at work
Person-centredness is central to TM, taking the question "What is important to you?" as a point of departure for shared decision making based on recognition of patient values and capabilities. The employees considered themselves competent and committed, and scored quite high on the composite measures of "Prerequisites for person-centredness" and "Care processes" (see table 7). However, they gave the "Care environment" a lower score, especially the item "supportive organization", which in practice means trust-based management gets the lowest rating. Table 7 PCC and org culture

Determinants of innovation (MIDI):
a) Socio-political context: regulations and legislation: When asked if the TM fits well with current laws and regulations, 64.4% answered agree or completely agree, while 36.6% answered disagree or neither agree nor disagree. This corresponds to findings mentioned above, indicating uncertainty among many employees regarding the TM itself and its legal and administrative aspects, and thus also failure in systematic preparation of the teams.

b) Factors associated with innovation (TM):
Team members in both iterations believed that the TM was relevant to their patients (identified as a facilitator and that the model is professionally sound and knowledge-based. They also found to a varying extent that the trust model has an acceptable complexity, that it fits well with how they are used to working and that it produces visible results. The (in)completeness of the model was identified as a barrier, but indicating that the organization is not well prepared for working according to the Trust Model. (table 8). Table 8-determinants associated with the innovation c) Factors associated with health personnel, patients and relatives: The team members were asked about how the TM works for them and how they think it works for patients and their relatives (table   9). They considered it particularly important that TM makes it possible to provide services according to the patients' needs, which they believe is the case. The outcome expectations were identified as specific facilitators. They also found that the TM complies with their responsibility as professionals, that they use their professional skills better and that the work became more exciting. They did not see major disadvantages of the TM. They also believed that they could count on support from colleagues to a fair extent.
Insert table 9 here d) Factors associated with the organization: Half of the team members found that they had too little background knowledge to work according to the TM when they started. Asked whether management had formalized the TM, 37 answered "yes" (67%), 20 "do not know" (33%) and 3 "no" (5%). Asked if there were other change processes going on at the workplace during the implementation of the TM, such as reorganization, mergers, cost cutting, or other innovations, 40 informants answered "yes" (77%), 12 "no" (23%) while 8 did not responded to this question. This indicates that there was considerable uncertainty among many team members about the TM as a management model, and that other organizational change processes were going on to a relatively large extent parallel to the TM iterations, which usually implies relatively poor conditions for innovation and implementation.
Several factors were identified as barriers; replacement when staff leave, staff capacity, financial resources, time available, material resources and performance feedback. (Table 10). This clearly indicates that employees found that the organization was not sufficiently prepared for working according to the TM intentions. Table 10 Factors associated with the organization What works and does not work for the team-members-qualitative exploration We will here summarize the findings from the three IGP process workshops during the second iteration, where all four districts were represented with members of all eight teams in all at each workshop. Questionnaires were used as the basis for group discussions in teams and the sharing of experiences in plenary sessions.

Workshop 1-Sharing experiences
The main purpose of the first workshop (November 2017) was to share experiences across teams and districts, and to learn from each other's experiences and solutions. The research aim was to explore what works and what does not work for the teams and the patients as perceived by the teams so far.
Forty-nine team members from the four districts took part. Table 11 displays how team members perceive key aspects of the team organization.  concern key aspects of the TM organization that according to a steering group decision should be implemented by all teams, i.e. the answer to the first 13 questions should be "yes". We see that this is not the case. At this point, relatively early in the second iteration, it was not expected that a new service assessment was completed with all patients. Relatively few reported that they involved the patients more than before. The employees appreciated the reassessment of patients' needs [procedures] and multi-disciplinary cooperation, both central to the model. In response to open questions and plenary experience sharing, the team members reported that what worked well both for them and for the patients was the multi-disciplinary co-operation in teams. However, several teams reported minimum staffing, which did not work well neither for the teams, being dependent on substitute staffing, nor for the patients, wanting predictability and reliability.

Insert table 11 task and organization in the team
Leadership support: There were great differences between the teams concerning what worked well and what did not. All agreed that working in multi-disciplinary teams was inspiring and motivating and that leadership support was essential. Some teams expressed clearly that collaborative team decision making did not work for them, that the team leader was absent most of the time, that they were working without clearly defined tasks, and that this was difficult and at times chaotic. Some also reported that they were being trusted with the responsibility of making care decisions, staff rosters and distribution of services without being trained or prepared for it, without sufficient staffing and without receiving the necessary managerial support. Instead they felt that competence was withdrawn from the team. In such circumstances, being trusted with more responsibility did not work for them.
Employee advice to management: The participants responded individually to a question as to what advice they would give to management in order to make the TM work even better for them and for the patients. The most frequent answers were ensuring a) sufficient staffing, b) team-leader being present and participating in the daily work, and c) team competence building when needed. In addition, the needs for d) clarification of tasks and e) sufficient time for administration were frequently addressed.
Workshop 2 -Trust, motivation and cooperation At the second workshop (November 2017) the purpose was to clarify the team members roles in the team and how to work together across competencies in order to enhance the patients' feelings of security, self-determination and coping. Trust was the key concept, related both to co-operating in teams and proving care to vulnerable elderly persons.
All eight teams were represented with 43 team members in all. In the plenary discussion the concept of trust was repeatedly related to relying on others, the team members relying on colleagues, the patients relying on the team members and the service, but also relying on oneself. Emphasis was placed on honesty, openness, communication, searching for shared solutions. Trust was also perceived as having faith that everyone is doing their job and doing their best.
The individual questionnaire focused on trust and co-operation in teams. When reporting from group discussions, there seemed to be a general experience that having less patients to visit made it possible to establish a more trustworthy relationship with each person. A surprising discovery at this workshop was that not all teams worked according to the new, person-centred procedure, in spite of the fact that this was central to the TM and that a new template, based on the "What's important to you?"-question and the principle of shared decision making, was developed during the first TM iteration. Another surprising discovery was that the new procedure was not supported by the digital patient record system, which may be part of the explanation for the first one.

Workshop 3: Roles and communication
The third and final workshop of the second iteration (January 2018) had 35 participants from three districts. Taking the results from the previous workshop as a point of departure, the theme was the initial income interviews with new patients. Key points were how to create trust and good communication in the interview situation, identify what is important for the patient, and create a shared decision-making process. It was reported that following this procedure and showing interest in what was important to the patients, seemed to improve the relationships and had a motivating effect on patients' potential for self-management, working more actively to improve their health condition.
Some also reported radical positive results after some time, like long time service users no longer needing home care. A continuing problem was that the patient record system did not support this person-centred approach, which was considered to have a decelerating effect on the implementation of the TM and the basic idea behind it.

The managerial perspective
Individual interviews were conducted with ten managers half-way into the second iteration, in December 2017. Problems concerning organizational change were directly or indirectly core issues in most of the interviews. The perspectives among the managers were quite diverse. They all identified complexity a central issue, that abolishing the PPS and establishing and formalizing the TM represented a giant organizational step. Most of the leaders pointed out that the two iterations had proven that the transition from PPS to trust-based management and delivery was far more complex than expected.
Insert Table 12 here: leadership complexity issues At the same time, all of the leaders found the TM promising, and no one wanted to return to the PPS.
They gave a series of reasons for this. Multi-professional competence in the teams enhanced the quality of care and of care decisions. Co-operation across professions was motivating for the team members. Having both executive officers, occupational therapists and physiotherapists as permanent members of the team seemed to be a preferred solution, perhaps dividing their time between two teams. A challenge, according to the managers, was finding the balance and distribution of responsibilities. The teams tried out different models, some with more or less collaborative decisionmaking, some with clear distribution of individual roles and responsibilities, and some teams somewhere in between.
Several leaders emphasized that taking "what is important to you?" as a point of departure represented a new mindset both for team leaders and team members. A challenge agreed upon was building competence to develop the new practice of team co-operation, person-centredness and shared decision making further. Also the importance of digital system improvement was addressed i.e. making the patient record system align with the TM priorities and procedures.
The team-leader role was a key topic in most of the interviews. It was a general view that the role was not clearly defined beforehand, except for the idea that trust-based leadership probably would imply a coaching leadership style. The team-leaders were not trained for their new role. In practice, three different roles were tried out during the two iterations, a) a distant, administrative role, b) a coaching role outside the team, or c) a coaching role inside the team. The informants expressed different perspectives on this issue, but most of them communicated that they had picked up the signals from the team members that most of them wanted their team-leader to be present, both mentally and physically at meetings, taking the role as coach and someone to turn to in difficult situations. This is congruent with the results from the workshop process interviews. One team-leader, having chosen the coach-inside-the team-role, explained how during the preparations for the transition from PPS to TM she had changed her way of thinking as a leader:  leadership (Cardiff, 2014;Eide & Cardiff, 2017). The main findings are that patients seem to be more satisfied with care when fewer professionals are involved, and that their trust in the service is enhanced when their basic needs for predictability, reliability and continuity are met. Also, improved health condition and rehabilitation seem to be possible results, as indicated by the narratives. These results seem quite dependent on the degree to which the intended TM elements are implemented.

Team-members prerequisites
The 1 st circle displays the results concerning the circumstances or preconditions under which the TM system seems to work well for the patients, including the visiting team member's behaviours, skills and competences. Important factors reported by the patients themselves are that the helpers are familiar or well known to them, and that they keep appointments, are well prepared and nice, helpful and competent. Compared to the PPS, most of these features are not TM specific. The only element with a distinctive function is fewer professionals involved in individual patient care, due to the team organization, which makes it easier for the patients, with their varying needs, and the helpers to establish a mutual relationship, which seems to be crucial for a positive user experience of the service. We know from previous studies that older persons receiving home care express their concerns indirectly, and that establishing a trusting relationship and listening to narratives and emotions are key to meeting patients' needs (Hafskjold et al., 2016;Lindahl, Lidén, & Lindblad, 2011).
The TM method of organization seems to create favourable circumstances for doing so. From the point of view of the team members, the TM, working with a limited number of patients with the "What is important to you"-question as a point of departure, made it easier to get to know each person's health condition, values and capabilities. Accordingly, they also experienced it easier to make shared decisions and motivate each person to strive to reach their individual goals. This seems to be a consistent finding across the teams, at least when the TM team assumptions were met (figure 2, 2 nd and 3 rd circle). However, patient motivation for rehabilitation and health improvement was not systematically investigated in this study. On the other hand, team member narratives indicate that the TM way of working in small teams, including taking the patients' personal values as a starting point, might stimulate patients' intrinsic capacity for self-care (Carver, 2006), at least in some cases.

Team leader responsibilities
The 2 nd and 3 rd circles display the leadership and system circumstances under which the TM seems to work well for the teams. A main finding was that the teams wanted their team leader to be present in the daily work, and felt less supported when the team leader took a more distant administrative role.
The TM worked well for the teams when the team leader participated in meetings and took a coaching, supportive role, appreciating change proposals and ensuring sufficient staffing and competence building when needed. Conversely, the TM worked less well for the teams when the leader considered team members' concerns and proposals for improvement as criticism, and the teams were left alone to find solutions to questions of patient safety or quality of care. This was especially so when the root causes were perceived to be of a systemic nature, like lack of clear care decision procedures or a well-functioning patient record system. These findings are comparable both with the literature on multi-disciplinary teams (Nancarrow et al., 2013) and on management and organizational risk, pinpointing that well-functioning high risk organizations are characterized by leaders creating a culture where questioning is ok and people feel free to raise concerns (Weick & Sutcliffe, 2007). In order to work well for the teams, the team leader had to assume the quality assurance responsibility concerning administration and care decision routines. In other words, the findings indicate that self-managed teams of the TM need a responsible leader. According to recent research on the transformation to self-managed teams in healthcare, the team-leader role progresses through different phases, from a more active role in the beginning to a more coaching role after a while (Renkema, 2018). Our findings indicate that self-managed teams, at least in the early phase, need a leader both actively clarifying procedures and administrative frameworks, progressing through to coaching the teams in finding their own solutions and making shared decisions. This finding might be explained by insufficient clarity of procedures, lack of training in team co-operation and shard decision making, or an uncertainty about the final responsibility for care decisions made by the teams (which in fact was the team leader's), or a combination of these. There might be a complexity of reasons for the reported need for a coaching team leader, but whatever the reason(s) what works for the teams seems to be the presence of the team leader, the team being trusted, clarity of procedures and socialization of the team members into new ways of working together.

Systems prerequisites and affordances
The 3 rd circle displays the results concerning the system circumstances under which the TM might work well for managers, team leaders and teams, and hence also for the patients. The prerequisites are trust-based leadership and organization, sufficient staff resources, quality system improvement to be in line with the TM and competence building resources. On this system level, the central challenge is that of organizational complexity. This is paradoxical. According to Luhmann (1968) trust is a mechanism for reduction of social complexity. The TM is a concrete way of operationalizing trust in practice in the micro-system of home care service within the macro system of health care and municipal health care services. However, in order for TM to work this way, reduced social complexity and avoiding paving the way for chaos have to be in place, like trust-based management on all levels, sufficient staff and competence building resources and system improvements.
According to the managers themselves, they clearly prefer the TM to the PPS on the service level, but find it hard to deal with the competing institutional logics of management by objectives and trust-based leadership.
The degree to which the TM works for the managers seems to be a question of being able to adapt to the value of trust as a leadership and organizational principle. In addition, balancing the competing organizational logics on a system level of the NPM and management by objectives on the one hand and that of trust-based leadership and organization on the other are important. Coping with the hybridization problem might be key to creating the optimal circumstances for home care services and the individual teams (Andersson & Liff, 2018;Kristiansen, Obstfelder, & Lotherington, 2016).
Introducing the term "trust-based leadership and steering" might seem like a subtle way of institutionalizing hybridity. The question remains how to cope with it, by enhancing focus on steering and control, embracing trust-based management as the main organizational principle, or by some third golden means.
The findings in this study indicate that an organizational coping strategy needs to take a complexity of elements into consideration. The most significant of these being sufficient multi-disciplinary staffing, improvement and simplification of the quality system, including adaption to the TM, and improvement of the digital patient record system in order to make it compatible with the TM. In addition, there is a need for the provision of sufficient competence building resources in order to support the necessary organizational learning and change processes.

Discussion
The overall aim of this study was to evaluate the two-step development of a trust-based organization of home care services. This is to our knowledge the first empirical study both of the complexity of trust-based organizational change in home care, and of developing a trust-based alternative to the purchaser-provider split (PPS).
The Framework for Trust-based Home Care Practice shows the ideal model and what is needed to serve patients in the best possible way and stimulate their own capacities for self-care; which competences the team members need; which qualities the leaders should display and what the larger system should afford. The framework may also be considered a model for organizational trustworthiness (within homecare), characterized by the three factors ability, benevolence and integrity, postulated by Mayer, Davis & Schoorman (2007). Within the TM, ability (team competence, skills, etc) and benevolence (the person-centred approach, etc) are definitely present, even though there was room for improvement. The critical factor is integrity, which here applies to the organization as a whole and the question of consistency of organizational values and principles. As we have seen, there was a divide between competing organizational logics, between trust and control, between strategic intention and implementation practice, perhaps best summarized in the ambiguous phrase of "trust-based leadership and steering". The question of organizational integrity might be the main challenge to successful implementation of the TM.
A series of needs for improvement were identified, including the need to (1)  The assumption that TM might enhance patient motivation for change to a higher degree than the PPS might be explained by Self Determination Theory (SDT) (Deci & Ryan, 2008). This motivation theory postulates three innate psychological needs-competence, autonomy, and relatedness-"which when satisfied yield enhanced self-motivation and mental health and when thwarted lead to diminished motivation and well-being" (Ryan & Deci, 2000). While the PPS method of organizing often implies many different helpers and provision of specific pre-defined tasks determined by others (low degree of relatedness and autonomy), TM organizing in small teams leaves more room for the service provider to attach to the individual person (relatedness), identify the person's resources (competence) and take these and what is important to the person into consideration when making shared decisions at the person's home (autonomy).
The indications that the individual patient's trust in the municipal home care service increases with the TM might be illuminated by Løgstrup's relational ethics (Løgstrup, 1956), postulating an ethical demand of respecting the integrity and autonomy of others, emerging from the basic human condition of vulnerability and basic trust. If this ethical demand is neglected, basic trust will be at risk.
A health care provider working according to the TM, has a lesser number of patients, gets to know each person better and is supposed to work in a more person-centred way than a provider working within a PPS model, and has accordingly a greater opportunity to live up to the Løstrupian ethical demand.
Studying this complexity and the mechanism of trust in the municipal healthcare context has been challenging. There were few models to learn from. Even though the complexity of healthcare management and practice has been recognized since the beginning of this century (Plsek & Wilson, 2001), there are few empirical studies and a lack of theorizing and method development in the field of healthcare (Greenhalgh & Papoutsi, 2018). It is a little bit different in the business and industry literature, where complexity studies (emerging from physics and systems theory in the 1960s) entered the field of organizational studies of business and industry around the millennium shift (Battram, 1998;Stacey, 1996Stacey, , 2001Tsoukas & Chia, 2002).
This study has evoked the question of how to describe the complexity of a healthcare organization, the organizational "chaosmos", as it is termed by Tsoukas (2013). During the two iterations, we observed the emergence of social interaction elements like power play, resistance, conflicts, rivalry, mutual support and professional enthusiasm and co-operation within and across the eight teams and their respective management surroundings which influenced the development of the agents' perception of trust and their evaluations of the TM. It would be far beyond the scope and resources of this project to identify and analyse in depth the complexity of these fluid and dynamic processes. Having systematized the findings and developed the framework for trust-based home care practice, the organizational complexity might seem less complex. We therefore need to add that this framework is limited to key elements that-according to our findings-need to be in place in order to realize trust-based home care practice. There were in most cases significant tensions between the ethical intentionality of the home care service and what Tsoukas calls representation on the one hand, namely the visible, formal and explicit aspects of the organization (like TM procedures decided on by the project group) on the one hand, and practice on the other, namely the hidden and implicit values and practices that characterize the respective organizational cultures.
Strengths and weaknesses of the study As indicated above the complexity of organizational innovation in healthcare goes deep and may seem beyond discursive analysis and description. Possible strengths of this study are the recognition of this complexity and the multi-professional composition of the research team, which has made it possible to discover and see more aspects of this complexity than we otherwise would have done, including scholars with backgrounds in leadership and organization studies, communication and health innovation research, sociology, psychology, nursing, optometry, literature and ethics. A weakness of the study is nevertheless that the complexity of the subject matter goes deeper than we have been able to analyse and display. In fact, most of the elements identified and represented in the proposed framework, like patient outcomes, team co-operation, shared decision-making, team leadership, care decision practices and quality system improvement, might have been made the focus of separate in-depth studies.

Conclusions And Implications
Organizing home care services in self-managed multi-disciplinary teams according to the value of trust seems like a feasible alternative to the PPS. Complex, value-based organizational innovation and change (like TM) challenges existing institutional logics (like NPM) and an organization's ability to cope with institutional hybridization. The complexity of organizational innovation in home care and of researching this topic is understudied. The importance of recognizing the complexity of trust-building and inherent slowness of radical service innovation should be further explored.

Declarations
Ethics approval and consent to participate The Norwegian Data Service for Social Sciences approved the project according to the Personal Data Act (approval no. 50966). All informants consented to participation.

Consent for publication
Not applicable.

Availability of data and materials
Interview guide and questionnaires can be made available on request. The service user and staff questionnaires were distributed and collected by the Municipality of Oslo, and made available to the researchers in anonymised form to analyze within the specific framework of the innovation project.
The qualitative data consist of transcribed interviews and field notes in settings with a limited number of participants, and will not be made available for privacy reasons.

Competing interests
The authors declare that they have no competing interests.

Funding
The evaluation project was funded by the Municipality of Oslo and the University of South-Eastern Norway, Faculty of Health and Social Sciences. The Municipality of Oslo did not have any role in the study design, the writing of the paper or the decision to submit the paper for publication.
# Reversed -high score means not complex * In Bold -reach the threshold to be a facilitator or barrier Abbreviations: TD/D; totally disagree/disagree (barriers), A/TA; agree/totally agree (facilitator). * Reversed Likert scale from 1; totally agree to 5; totally * In Bold -reach the threshold to be a facilitator or barrier Tabell 11: Tasks and organizing in teams  A framework for trust-based home care practice