Understanding integrated care at the frontline using organisational learning theory: A participatory evaluation of multi-professional teams in East London

Integrated care has been proposed as an organising principle to address the challenges of the rising demand for care services and limited resources. There is limited understanding of the role of learning in integrated care systems. Organisational Learning (OL) theory in the guise of ‘Learning Practice’ can offer a lens to study service integration and reflect on some of the challenges faced by multi-professional teams in developing a learning culture. The study presents findings from two qualitative evaluations of integrated care initiatives in three East London boroughs, England, undertaken between 2017 and 2018. The evaluations employed a participatory approach, the researcher-in-residence model, to coproduce findings with frontline staff working in multiprofessional teams in community care. Thematic analysis was undertaken using an adapted version of the ‘Learning Practice’ framework. The majority of learning in the teams was single loop i.e. learning was mainly reactive to issues that arise. Developing a learning culture in the three boroughs was hindered by the differences in the professional and organisational cultures of health and social care and challenges in developing effective structures for learning. Individual organisational priorities and pressures inhibited both the embedding of learning and effective integration of care services at the frontline. Currently, learning is not inherent in integrated care planning. The adoption of the principles of OL optimising learning opportunities, support of innovation, managed risk taking and capitalising on the will of staff to work in multidisciplinary teams might positively contribute to the development of service integration.


Background
Health and social care systems in England are facing the unprecedented pressures of increasing needs from an ageing population, rising workload for an overburdened workforce and limited financial resources (Ham et al., 2011;Armitage et al., 2009). There is a growing consensus that better integration of care is a key part of the approach to tackling these challenges with some evidence that doing so may improve patients' satisfaction, but more mixed evidence that it reduces costs (Baxter et al., 2018;Humphries, 2015).
Since the introduction of the 2012 Health and Social Care Act in England (Timmins, 2012), there has been significant investment in integrated care initiatives. In 2014, an arm's length body of the Department of Health and Social Care, NHS England published the Five Year Forward View which called for the funding of 'Vanguard' sites to test 'New Care Models.' (Nhse, 2014) One of these models, the multi-speciality community provider, advocated for GP practices to form networks and federations while working collaboratively with other health and social care providers, with the primary aims of reducing hospital admissions and moving care closer to the home (Turner et al., 2016). Integrated care in England continued to evolve in the form of Sustainability and Transformation Plans in 2015, Accountable Care Organisations in 2017 and more recently Integrated Care Systems. Each of these developments were underpinned by a premise of transferring care away from hospitals (thought to be costlier) to supposedly less expensive community settings, as well as a more collaborative approach with planning of individual institutions complemented by place-based planning for local populations (Hammond et al., 2019).
These changes have been implemented on less than robust evidence. Firstly, greater integration including the expansion of community services may not release cash savings or deliver less costly care (Ruane, 2019). Secondly, integrated care has yet to deliver improvements in health service outcomes, whereby it has been shown to both increase and decrease use of community services with limited evidence to suggest that community-based initiatives reduce unplanned hospital admissions (Baxter et al., 2018;Purdy, 2010). Nonetheless, at the patient level, integrated care approaches have led to greater patient satisfaction, increases in the perceptions of the quality of care and improved access to services (Mason et al., 2015). Given the somewhat indeterminate picture of the benefits and outcomes of integrated care it is perhaps not surprising that commentators have since suggested that the Health and Social care Act (2012) paved the way for fragmentation of the health service, increasing competition between providers rather than promoting local collaboration and partnership working, as it was intended to (Ruane, 2019). In the end, "a rapidly changing policy context, significant central control and the emergence of other single agency priorities over time have all made it difficult to join services up in practice." (Glasby, 2016) In community care, integration envisages the creation of multiprofessional teams from across health and social care (Roland et al., 2012;Hamilton et al., 2015) which were initially expected to focus on the segment of the population with the most complex health and social care needs. More recently, there has been a shift to a whole population health approach which aims to improve the physical and mental health and wellbeing within and across a defined population, in an effort to reduce health inequalities (Buck, 2018). In addition to preventing unnecessary hospital admissions, multi-professional teams also aspire to provide patient-centred and holistic care, reduce fragmentation of care delivery and promote self-care. (Nhse, 2017) The basis for an integrated care system is established at the strategic level between organisations through the pooling of budgets and aligning of governance, managerial and administrative systems. At the service delivery level, within multi-professional teams, health and social care professionals are required to work in partnership (Ham, 2018). Effective partnership working requires culture change at organisational and professional levels, sharing of data, effective communication, learning for improvement, trust and an understanding of mutual responsibility and accountability (D'amour and Oandasan, 2005). These factors must be considered in light of the well-documented structural and cultural divides between health and social care, as well as the limited investment of resources to support genuine organisational development (Leutz, 1999;Miller, 2016;Stein, 2016).

Organisational learning
In this study, we use Organisation learning (OL) theory as a lens to study service integration. OL is the process by which organisations improve and build knowledge capacity through experiential or planned learning activities (Carroll and Edmondson, 2002). The term 'organisational learning' originally emerged from the business management literature, with several seminal publications shaping its understanding, in particular the work of Argyis (1977) and Schon (1983) who introduced learning as an action theory. Senge et al. (1997) suggested OL could be a means of understanding the relationships between different organisational components, identifying the importance of leadership, and in particular the decentralisation of leadership, to empower staff at all levels and facilitate the development of a learning culture.
The OL literature is dominated by descriptions of various models and approaches, prescriptive advice, and anecdotal accounts of organisational change. However, while organisational change can be facilitated by insightful planning and analysis, performance will often depend on situational variables (Dunphy and Stace, 1993), including power and politics (Buchanan and Badham, 1999). Project management, action research and organisational development are among the other main approaches to organisational change in complex systems. Whereas project management is about driving a defined change process by developing tools to help structure and implement change, action research uses research in an interventionist way (Iles and Sutherland, 2001). Kurt Lewin, who first developed action research as a methodological approach, also promoted democratic values and participatory engagement in order to encourage change and address social conflict (Lewin, 1946). In action research, organisational change is understood as a cyclical process where theory guides practice and practice in turn informs theory. Organisational development (OD) is based on behavioural science knowledge and practice (e.g. leadership, group dynamics, and work design), "where the aim is to help members of an organisation gain relevant skills to address the challenges entailed by a change process through involving them directly and transferring knowledge across the system."  In this respect, there is some overlap with OL, which is characterised by a continuous cycle of learning and change. OL fosters adaptation of structures, promotion of innovative and empowering leadership behaviours and practice, supportive organisation cultures and shared information systems, as integral elements to facilitate whole systems change (Iles and Sutherland, 2001). Learning may also generate real-time insights into implementation processes. This is particular pertinent to healthcare initiatives, where learning about the change process is often superseded by a focus on improvements in outcomes (Barry et al., 2018).
Within health systems, there is a high degree of interdependence between practitioners, and between practitioners and processes, which, combined with continuing technical and organisational advances, means these systems are dynamic as well as complex and highly regulated (Iles and Sutherland, 2001). Working practices tend to evolve slowly, often amid patterns of resistance to change, through new training, developments in technology, policy change and influential goal oriented leaders focusing on the improvement of performance (Carroll and Edmondson, 2002). Within this context, learning must be understood as a cyclical process of actions and reflection which may become part of everyday working practices (Argyris, 1977).
In healthcare, organisational performance is often characterised by outcomes associated with quality and safety. Yet, learning in healthcare is seldom ubiquitous in an organisation and may vary among wards, teams, groups and individuals (Carlile, 2002). In healthcare, learning is often reactive, in response to incidents (e.g. patient safety failures) or, less frequently, as a consequence of leaders keen to change the organisational culture (Smith and Valenta, 2018;Senge, 1997). Organisations that are committed to a long term ambition to improve performance might prioritise a learning culture, using a combination of disciplines, skills, values and behaviours to support systemic learning (Edmondson, 1999). Whereas systemic learning can be facilitated through use of audits, surveys and performance evaluations, investment in supportive structures and information systems, training and meetings are of equal importance to provide learning opportunities. Use of staff and patient feedback, as well as their involvement in service reorganisation, also fosters learning (Edmondson, 1999). Nonetheless, increased demand and reduced capacity mean that these organisations have limited time to learn, adapt and develop.
Three types of OL (Argyris, 1977) have been identified and can be applied to the context of healthcare organisations. Single-loop learning refers to actions that respond to shortcomings emerging for instance, from a clinical audit assessing a service against national standards, with minimal impact on organisational objectives (Davies and Nutley, 2000). Double-loop learning is a more sophisticated approach connecting knowledge for understanding, by challenging existing values, assumptions and behaviours of organisations and the individuals within them. Organisations committed to triple-loop learning have an innate understanding of learning and focus on learning how to learn; they use learning to develop and test new learning strategies by understanding the relationship between actions and results, demonstrating a capacity to adapt. Features of triple loop learning may have a pivotal role in developing care integration given the ever changing landscape of the commissioning and restructuring of services in integrated care systems (Nuño-Solinís, 2017).
Despite OL being extensively described in the context of healthcare organisations, with a few exceptions (Shortell, 2016;Nembhard and Tucker, 2016) less attention has been given to the role of learning in integrated care and OL theory has tended to focus on healthcare organisations at the strategic level. This paper addresses this gap in the literature by exploring how change occurs when frontline teams adapt their working practices. We apply the 'Learning Practice' framework developed by Rushmer et al., which adapts organisation learning theory to the characteristics of frontline care delivery, providing a framework to examine the ways in which frontline care teams can develop their own regime of learning, innovation and change through their day to day work (Rushmer, 2004).
The paper contributes to our understanding of learning in integrated care teams and assesses types of learning that can strengthen partnership working and greater integration of care, based on the experience of three cases in East London, generating new insights that can inform both policy and practice.

Subjects and settings
The study presents findings from two qualitative evaluations of integrated care initiatives in East London. In 2013, three city boroughs (referred to throughout as A, B and C) came together to form an Integrated Care Programme which comprised local health and social care organisations selected by NHS England to act as pioneers in the development of innovative approaches to deliver integrated care (Eyre and Marshall, 2015). This programme was subsumed into a tri-borough East London transformation programme in 2015, which aimed to improve the local health and social care economy in line with the challenges set out in the NHS Five Year Forward View. (Nhse, 2014) The health and social care systems in the three boroughs are described in Table 1.

Study design
Both studies were participatory evaluations of integrated care delivery. One of the researchers (ML) was embedded in the integrated care programme in borough A from June 2017 to November 2018 The other researcher (SB) undertook a comparative study of the delivery of integrated care programmes across the three boroughs and was embedded in multi-professional teams involved in Admission Avoidance, Discharge from Hospital and End of Life Care, from May 2017 to May 2018. In this paper we draw on the findings from field notes of observations and interviews (semi-structured and group interviews) with stakeholders from the multi-professional teams operating in community care. Interviews were conducted by ML, a researcher with experience of conducting health services research using qualitative methods and SB, a qualitative researcher, with expertise in participatory research and a social science background.
We used the Researcher in Residence model, a participatory approach to evaluation. In response to a recognised concern that 'established approaches to getting health services research into practice are not radically changing the extent to which management decisions are influenced by scientific evidence,' the Researcher in Residence model embraces the concept of 'co-creating' knowledge between researchers and practitioners (Marshall et al., 2014). The model places the researchers as key members of the delivery teams within the organisations under study, as opposed to external observers of change. ML and SB co-created knowledge with participants in the study; an evaluation steering group was set up involving stakeholders from health and social care organisations to co-design the research protocols, and workshops were organised with frontline staff to interpret findings and coproduce recommendations. The participatory approach facilitated the mobilisation of existing knowledge (from the academic and policy literature) and newly created evidence (generated by the research) across the localities and, to an extent, influenced implementation and development of community care service provision locally. A participatory approach inevitably raises several new ethical issues on power dynamics and relationships between academic and non-academic researchers, while problematising traditional ethicsi.e. anonymity, consent. We examine these issues in detail elsewhere, based on our experience as researchers-in-residence in East London .

Data collection
This paper is based on findings from 35 semi-structured individual and five group interviews (total n = 15 participants) with multiprofessional community teams, and participant observation of relevant meetings, amounting to approximately 170 h. Interviews were Table 1 Site descriptions of local health and social care systems.

Borough A
In 2015, a partnership of multi-speciality community provider organisations was awarded 'Vanguard' status (support and funding to develop innovative models of care which other parts of the country can learn from) by NHS England. The partnership comprised a collaboration of health and social care commissioners and providers (including voluntary services). The Vanguard sites were awarded substantial funding to further develop local integrated care approaches with a primary focus on complex care provision. The borough is comprised of four localities (population of 60000-80000) and each locality has a multi-professional community care team known as an Extended Primary Care team (EPCT) which provides community nursing and therapies for residents aged over 18. The teams provide care coordination and case management for patients whose needs are most appropriately met by co-located community care professionals; community/district nurses, health care assistants, occupational therapists, physiotherapists, mental health nurses, rehabilitation support workers and care navigators. * At the time of the study some of the EPCTs were supported by a social worker from the Local Authority although this was sporadic. * Care navigators take on many non-clinical responsibilities pertaining to a wider variety of aspects of health and social care characterised by supporting patients and their families. (Hee, 2016) Borough B At the time of the study, the borough was establishing a provider organisation board to support the creation of a provider partnership made up of commissioners and providers of acute, community, mental health, social care and primary health services. However, the governance and accountability structure had not yet been formalised.
In terms of community care provision, the site has a similar model to borough A with EPCTs covering four localities, with a population of approximately 80,000 each, and incorporating eight General Practice clusters. The professionals working within the EPCTs in this borough are the same as in borough A. The EPCTs work directly with and support local GP networks, provide care coordination and case management, and deal with referrals from GPs, hospitals, care homes and social services. There are no dedicated social workers co-located with the EPCTs. Borough C This borough has been working toward the formation of an Accountable Care System with a focus on developing three key elements; 1) a strategic commissioning function bringing together the local clinical commissioning group and the local authority; 2) an outcomes framework linked to population-based contracts and 3) an integrated, place based service delivery model that provides gateways into redesigned services and pathways. The system has four priorities: Community Care, Integrated Urgent Care, Leaving Hospital Pathways, to identify the most appropriate pathway depending on patient needs and potential for rehabilitation and Reablement, and End of Life Care, integrating provision across professional boundaries. Community health services are based on three Integrated care teams (ICTs), North, Central and South. The teams are multi-professional and provide adult community health services. They are similar in set up to the EPCTs in borough A and B, with the addition of community matrons (also providing care navigation). The ICTs deliver nursing, case management and therapy, End of Life and incontinence care. Initially, each team had dedicated social workers based with them, but a lack of capacity resulted in the scheme ending.
undertaken with service managers, EPCT/ICT team leads, various health professionals from the teams as well as social workers aligned to the EPCT/ICT. We used a purposive sampling strategy to identify relevant service managers from both health and social care (see Table 2). We purposefully selected a range of EPCT/ICT staff for interview based on their level of experience and qualification. We interviewed staff on permanent contracts with a provider organisation and agency workers. Given the embedded approach to the evaluation most participants were known to the researchers. Interview guides were formulated using relevant themes from the literature on models of integrated care and were also informed by participant observation data, as well as discussions with participants and the members of the steering group. An inductive approach was taken with emerging themes from initial interviews used as a basis for further iterations of the interview guide. Interviews with staff were held at the participant's workplace in a private meeting room. Interviews lasted between 45 and 90 min.

Data analysis
Interviews were audio-recorded and transcribed verbatim. Data were managed using NVivo version 11.0. ML and SB conducted qualitative analysis using a thematic framework approach to code the data and identify patterns and themes (Green and Thorogood, 2018). A sample of transcripts were coded independently by ML and SB and the resulting themes and sub-themes were discussed to create a thematic framework. The framework was developed from the existing theoretical frameworks on Learning Practice in the context of integrated care with some iterative adaption to capture emerging themes. Data was also informed by field notes from participant observations. Components of the analysis plan, including co-interpretation of the findings, was undertaken by all three authors.

Ethics
Ethics and governance approvals were provided by the NHS Research Ethics Committee (REC ref. 154 17/SC/0687) and the Health Regulatory Authority. All interview participants were approached by email or telephone by one of the researchers who outlined the purpose of the study and interview process where appropriate. Written informed consent was obtained from each participant prior to the interview. Participants agreeing to interview returned their signed consent forms at the time of the interview. Participants were assured of confidentiality and anonymity and that participation was voluntary, and that they were free to withdraw from the study. No participants withdrew their consent.

Findings
The thematic framework comprised two main components of the Learning Practice: shared values and structural characteristics, under which the data were categorised into themes and sub-themes. For each theme or sub-theme, we provide examples from the data and show the extent to which multi-professional teams in each borough are practicing the principles of OL. We also outline the stage at which the teams are in terms of learning: single, double or triple loop.

Shared values
Organisations and the individuals within them are responsible for cultivating a learning culture, supporting and empowering staff to test, innovate, learn and share. Sub-themes related to shared values are described below.

Supportive leadership
The literature defines supportive local leadership (Yukl, 2013) as creating an environment within which staff are recognised for their achievements, can operate without fear of blame and with tolerance for mistakes, and are supported to undertake professional and career development. Overall, we observed the existence of a blame culture especially between district nurses and social workers around several issues. Most of these issues were defined by gaps in care provision as a result of differing organisational priorities. For example, social workers suggested nurses too often recommended care packages that did not align with social care provision due to Local Authority pressures. Nonetheless, health and social care organisations endeavoured to work in partnership around issues, such as safeguarding and the sharing of information and subsequent learning from safety incidents and near misses.
Both health and social care professionals mentioned a lack of protected time to undertake professional development and limited time to reflect on practice, so as to enable learning. These issues were compounded by perceived workforce pressures such as problems with retention and recruitment, a reliance upon agency staff and a pervasive view of having to constantly 'firefight' against a backdrop of limited resources. Participants highlighted a lack of supportive leadership and acknowledgement from management of the pressures experienced on the frontline. This left staff feeling overworked with some expressing concerns about the impact of such working conditions on their mental health.
'They (management) are there to train us, they are telling us they are providing flexibility, you're given opportunities. But how are we going to manage the patients day-to-day if we are going to spend time on training? With a lot of training, so much training…if you release people to do those things, how to manage the staffing? …… I'm on a course, its a few days every few weeks… if I'm not here two days, what I'm supposed to do in five I have to do in three days, so it's just how to manage that. It puts stress on people.' District nurse

Shared learning
Within Learning Practice, shared learning may arise from professionals either within the same discipline or from different professional groups (Skinner, 2007). Learning 'windows' are formal or informal opportunities that enable the sharing of experiences and knowledge. We observed a few examples of learning opportunities across the three boroughs. In borough A, the community health care provider organisation offered reflective practice sessions facilitated by a clinical psychologist and 'health coaching' sessions that centred on reduction in task-orientated care in favour of more holistic care practices. Additionally, borough wide staff engagement events provided a platform for sharing experiences and networking between different professional groups. These were well attended, although scepticism remained as to their lasting impact in terms of fostering relationships across the different care sectors. Indeed, staff suggested they were dedicated to management contributions and were seldom led by frontline staff.
'On the face of it these staff engagement events are a great opportunity to bring together the different services across the borough… but it tends to be the same faces; senior, middle and service managers, rarely frontline staff. I wonder if it's also an opportunity for senior managers to showcase their own work ongoing at the strategic level.' Field notes ML In comparison, there were some examples of 'bottom-up' efforts that enabled multi-disciplinary learning. Interviewees in borough A mentioned a discharge forum organised by local hospital teams that included EPCT staff. Meeting discussions were centred upon the discharge of patients with complex health and social care issues back to the community. These meetings were almost entirely led by frontline staff and service managers.
Interviewees from both health and social care shared the ethos of multi-disciplinary working to reduce duplication, share workload and deliver joined up care. However, this was hard to deliver in practice, in a context of continuous service reorganisation, a lack of clarity about new services and high turnover of stuff. Staff suggested rotations between sectors to develop inter-professional relationships and enhance the understanding of the different dynamics of care provision. However, it was clear from the interview data that multi-professional learning through joint training or rotations across health and social care was not prioritised by the provider organisations.

Understanding roles and responsibilities
Across the three boroughs there were gaps in the understanding of the roles and responsibilities of the different professionals, which was true for both established roles such as social workers and new, 'extended' roles e.g. care navigators. Social workers expressed a frustration at the lack of understanding of the parameters of their role from the perspective of district nurses, in particular the nurses' understanding of the Care Act (2014) (an assessment of people's needs along with their eligibility for publicly funded care and support). This gap in understanding was thought to result in nurses recommending care packages that set unrealistic expectations for patients as they seldom met the threshold of funding for homecare set by the Local Authority.
'You know our health colleagues have expectations in terms of what we (social workers) should be doing and that causes conflict as it impacts on our time and our ability to fulfil those expectations. They think we will just drop everything and sort out an issue, it's not realistic…. if the care is not there, in terms of how they want it then it's our fault. So, some of the social workers feel that they have to defend themselves.' Social worker, team manager In borough A and B the role of care navigator was introduced in 2015. At the time of the study a comprehensive framework for the role did not exist, but broadly the care navigator supported patients, providing a wide range of assistance from grocery shopping to applying for welfare support. Care navigators were thought to have an overview of health, social care and voluntary care provision locally. They were perceived to be an effective conduit between health and social care professionals, with the latter describing them as their key contact for the EPCTs.

Outward looking and innovative
A fundamental component of Learning Practice requires teams to challenge the engrained culture of psychological safety in healthcare, taking managed risks, so as to enhance their skills and knowledge while adopting quality improvement approaches as part of the learning process (Edmondson et al., 2016). Across the three boroughs, social workers perceived health care staff as risk averse, with a preference for task-orientated care. Conversely, social care professionals saw their own role as promoting user independence, also relying on family members to support care delivery and carry out everyday tasks. At the time of the study, health service managers spoke of a changing approach of the EPCT/ICT staff toward more holistic and less task-orientated care, but it was recognised that this would require a cultural shift. Moreover, such a transition was perceived as challenging given that temporary healthcare agency staff demonstrated a preference for task-orientated care.
EPCT staff are being trained up to take a more holistic approach …. What can you do for yourself? Who else can assist you? Family, neighbours? What can the professionals give? So, say if it was a new patient on insulin, we will be setting up a training package for you. Have you got attendance allowance? Who is your carer?' Service manager (EPCTs) The major challenge to a more holistic approach remained the limited capacity and human resources within health and social care, visà-vis rising demand for complex care, resulting in daily heavy caseloads, particularly for district nurses.
In sites A and B, the main community and mental health provider organisation was regarded nationally as a quality improvement pioneer.. We observed several successful quality improvement initiatives but engagement with these initiatives was affected by the limited time and resources available to staff. Indeed, some staff viewed quality improvement as 'additional work' rather than a component of their everyday work practice. Furthermore, quality improvement was largely practiced by the EPCT health professionals with less involvement from social workers.

Structural characteristics
Overall, the findings revealed that borough A was the most matured integrated care system, as a result of a history of partnership working between health and social care organisations. Furthermore, recent alignment of governance, managerial and administrative functions, such as a partnership board comprising senior managers, joint commissioning and several jointly funded middle managers roles positioned at the interface of both sectors, strengthened relationships across organisations at the senior and middle management level. These changes have been facilitated by considerable financial investment in integrated care services (significantly more than the other two boroughs) and an emphasis on quality improvement in many aspects of system design and service delivery. However, the adoption of the principles of OL in borough A was no more advanced than the other two sites and this translated into similar challenges at the point of delivery across all three boroughs. Embedding OL principles in an organisation or team requires infrastructure that enables communication and information sharing, as discussed further below (Rushmer, 2004).

Flatter hierarchies
Learning Practice promotes the development of non-hierarchical structures within teams. In boroughs A and B some of the locality teams were led by therapists, a change to the traditional structure of community care which is dominated by district nurses (Lalani et al., 2019). The team lead provided a managerial function for the EPCT with clinical supervision for team members provided by a senior clinician within their own professional group.
Overall, interviewees expressed dissatisfaction with their respective organisations primarily due to the lack of involvement for frontline staff in the planning and implementation of major system restructures. System and service changes were often top down, which for many participants suggested an inability of the organisation to empower staff and create a permissive environment where they could self-manage service care delivery.
'The middle management have to be giving permission for frontline staff to pursue opportunities…unless you address that hierarchy, that's not going to shift easily. You're always going to have a degree of command and control, especially if you've got issues… … there is that nuance of then how do you shift it where you empower people to get involved in decision making? When you listen to what's being said by those that actually provide the service.' Team lead, EPCT

Teamwork structures
Co-location and care coordination are often presented as important facilitators of service integration (Kaehne and Catherall, 2012). In all three boroughs, EPCT/ICT staff were co-located, sharing office space and facilities. Interviewees suggested that co-location had fostered more effective communication and had provided more informal opportunities to share information, knowledge and experiences relating to patient care. Indeed, 'corridor conversations' about patient cases were important in informing care delivery. For example, in site B, EPCT staff mentioned that being co-located with the Rapid Response team was crucial to effective care coordination for their shared cohort of patients. Even so, social workers were not co-located and when they did visit healthcare colleagues they remarked that the quality of space was challenging and they struggled to access their own data system. In all three sites, while the EPCT/ICT teams were on the same floor, they were segregated by professional group in separate offices which had implications for effective partnership working, reducing opportunities to share and learn.
'…the problem is the communication isn't too great, because we (care navigators) sit in a separate office to the nurses. We thought we would all be together…. so we form our own social group and relationships based on where we sit.' Care Navigator An important component of care coordination were multidisciplinary team meetings which were held monthly within each General Practice surgery. The membership comprised GPs, EPCT/ICT professionals and social workers. The effectiveness of these meetings was determined by the continuity of attendance from each of the professional groups. Overall, they were seen as useful opportunities to share knowledge and to develop strategies for dealing with patients with complex care needs. Even so, social workers struggled to attend meetings and when they did they were not always able to provide relevant information, citing a lack of capacity and high staff turnover as a prominent challenge. This caused frustration among other professionals. Furthermore, effective care coordination was thought to be impacted by a reliance across the sector on agency staff. Several interviewees also mentioned that they seldom undertook joint visits especially between health and social care professionals because of misaligned organisational procedures and differing standards.

Communication and information networks
Effective communication and information sharing rely upon access to patient data and records. Several participants expressed frustration at not being able to access relevant patient information. While the EPCTs in boroughs A and B had access to secondary care and primary care patient records, in borough C the electronic systems for community care were different to the other sectors. Across the three boroughs neither health nor social care professionals had access to the same patient/user records due to incompatible IT systems. All interviewees thought this limited the sharing of information, hindering care coordination.

Types of organisational learning
Examples of triple loop learning were not identified from our findings. Overall, across the three boroughs, the EPCTs/ICTs were mainly practicing single loop learning i.e. they were quick to recognise problems and rectify them. Staff only occasionally challenged assumptions or questioned their behaviours (double loop) and hence learning was undertaken in a reactive manner with few cases of learning being applied to the planning and development of services. This resulted in missed opportunities for embedding learning increasing the risk of the recurrence of previous problems.
However, we did identify examples of double loop learning initiatives in all three boroughs, especially where frontline staff had recognised the opportunity for learning and embedded quality improvement practices. The palliative champion's model employed in Borough A was conceived by an EPCT lead (district nurse) who identified a gap in the understanding of the needs of patients, families and carers in receipt of palliative care, as well as recognising an opportunity for more collaborative working with the local hospice. The aim of the initiative was to build capacity among the EPCTs to enable the delivery of tailored care to palliative patients in the community. The initiative involved formal and informal meetings organised by palliative champions (EPCT nurses) to raise awareness about palliative care and end of life pathways, while providing opportunities to learn through collaboration with 'specialist' staff at the local hospice. The designated palliative champions are also responsible for training colleagues within the EPCTs on relevant aspects of palliative care including working with families and carers.
'What it demonstrates is the importance of retaining staff and developing them. I've worked in the organisation for more than a decade, and only now I have been really able to develop my role in palliative care, through the training that I've had. And through the fact that I have developed the links with the hospice and... When you retain staff, you have got that organisational memory and the connections.' Team lead, EPCT

Discussion
This paper has used OL theory in the form of 'Learning Practice' to unpick how limited emphasis on learning is affecting implementation of service integration. The relationship between learning and service integration has received minimal attention in the literature and this work provides an original contribution on how the principles of Learning Practice could support future policy and practice of integrated care. Findings from across the three boroughs suggest that professional and organisational cultures play a crucial role in developing or hindering effective structures for learning. Differing organisational priorities raise huge barriers to cooperation between organisations. Therefore, it is unsurprising that teams tend to practice single loop learning but find it difficult to exercise more sophisticated models of learning. Nonetheless, participants from across health and social care expressed their ambition to work collaboratively and maximise opportunities for formal and informal learning.
The community provider organisations implemented several initiatives to promote care integration. The introduction of 'health coaching' for EPCT professionals was designed to promote the adoption of holistic care approaches while reducing task-orientated practice. Joint coaching and training were designed to optimise internal collaboration (between employees of the same organisation), and promote the notion of the EPCT as a collective, transcending professional boundaries (Nuño-Solinís, 2017). Reducing task-orientated care may challenge the norms, behaviours and working practices of healthcare staff resulting in a double loop approach to learning. However, it should be emphasised that one major obstacle to holistic care remains the limited capacity of existing teams.
Co-location is often described as the cornerstone of service integration and pivotal to internal collaboration (Cameron et al., 2014). The co-location of community health professionals has provided opportunities for informal learning, promoting peer-peer learning through sharing of information, knowledge and experiences, corridor conversations and staff room discussions (Liberati et al., 2019) However, our findings show that the effectiveness of co-location in fostering learning across the care sector was limited because social workers were not co-located with the EPCTs/ICTs. In some cases, even though social workers spent extended periods of time in EPCT offices, a lack of quality of space, different recording systems, differing organisational priorities and their accountability to and managerial responsibility within, the Local Authority hindered any positive impact. Co-location does not guarantee successful integration but merely provides a basis for joint working and learning.
The risk averse attitudes of healthcare professionals emanating from a traditional biomedical model approach to care provision (Wade and Halligan, 2004), as much as from the culture of blame within competing organisations within the internal market produced by the purchaser/provider split, are compounded by a lack of understanding of routine social care practice, the parameters of care packages and a knowledge gap with regard to the Care Act (2014). Risk aversion coupled with resistance to change has been previously identified as a barrier to innovation among middle and senior managers in the NHS (Dixon--Woods et al., 2012). We find that risk aversion and knowledge gaps also affect the way staff approach learning of collaborative practices, as they often feel they do not have permission to take risks and make decisions, within highly hierarchical contexts (Lalani et al., 2019).
A key principle of Learning Practice, supportive leadership, was a defining feature of the integrated care system in borough A. In principle, senior managers supported the notion of creating a permissive environment in which frontline staff could innovate, taking managed risks and learn from mistakes without fear of blame. Yet, in practice, the approach to communication, sharing information and knowledge transfer was top down, thereby failing to engage frontline staff meaningfully. Changing this hierarchical culture is a significant challenge for senior managers who will struggle to institute a learning culture unless they engage frontline staff in designing initiatives for learning (Ogunlayi and Britton, 2017).
Frontline staff in this study have demonstrated a capacity for change and innovation. In part, this is due to the support provided by middle managers, but it is also a result of frontline staff identifying gaps in service provision as opportunities for improvement, on their own initiative and despite the organisational and contextual pressures of an overburdened workforce, high vacancy rates and an overreliance on agency staff. These innovators ought to be supported and nurtured with successful initiatives celebrated as best practice and shared across systems and services (Birken et al., 2012)..
These findings have implications for policy and research in integrated care. They unpick the gap between the vision and rhetoric of integrated care initiatives and the reality of largely underwhelming health service outcomes. There is a clear role for learning in addressing this persisting chasm. In light of the development of partnerships at the neighbourhood level in England in the guise of Primary Care Networks, which represent the building blocks of integrated care systems (Baird, 2019), policymakers may want to consider that efforts to integrate care such as those that were evaluated in this study are floundering at the frontline (Nao, 2017) and further progress might be difficult in the absence of a learning culture. The success of new networks is dependent upon effective multi-disciplinary working and cross-sectoral collaboration. How can local integrated care systems better enable cross-organisation learning? The approach of borough A in this study in aligning certain governance functions provided a foundation for cross-organisational learning, but there is a requirement to go further. Endeavours such as training, networking events and social activities will address some of the relational issues but structural reconfiguration such as mechanisms for information sharing, access to data, formalised inter-agency committees and teams, and shared management lines are also needed. Greater support for distributed leadership and embracing risks as part of learning are necessary but difficult to envisage within a context driven by performance targets.
Further research is required into the effect of relational aspects on integrated care. What are the key mechanisms that could foster the greying of boundaries between health and social care? We have shown co-location alone is not enough and nor is good intention at the strategic or frontline level. Finally, there is a question as to whether integration in itself is a thankless endeavour. The increasingly specialised roles in medicine and nursing both in the hospital and community sector are less compatible with the principles of integration (Ferrer et al., 2005), especially when contrasted with the more extended and generalist roles of professionals working in integrated care teams. Perhaps that is why the hopes for Integrated Care Systems have been pinned on GPs who have a purview of the primary and community care landscape.
The use of our conceptual framework in future research in integrated care may enable the assessment of the role of learning in services and teams and in the development of care pathways. Up to now research in integrated care has focussed on establishing the impact at the system level using metrics such as elective and non-elective admissions to hospital, or at the service level, assessing outcomes such as patient satisfaction. However, there is a need for reframing research priorities in integrated care, moving away from health system and service outcomes to focus on learning and its association with the relational aspects of integration such as partnership working and professional culture.
A strength of this study is its novel approach to the application of OL theory to integrated care, demonstrating the need for developing a learning culture to optimise the effectiveness of multi-disciplinary teams in delivering coordinated care. Additionally, the in-depth participatory approach to research enabled us to explore the nuances of learning within service integration in community care which was achieved through co-interpretation of findings and co-production of recommendations with the research participants. The study is limited by the focus on service integration in community care in three demographically similar boroughs in London, which means that the findings may be less representative of other integrated care systems in the UK and elsewhere.

Conclusion
The integration of health and social care systems in the UK continues to accelerate (Baird, 2019). Even so, there is much scepticism on the success of integrated care initiatives (Glasby, 2016). Currently, the infrastructure for learning is absent in integrated care planning and service design. Adoption of the principles of OL through optimising learning opportunities, support of innovation, managed risk taking and capitalising on the will of staff to work in multidisciplinary teams might positively contribute to the development of service integration.
The structures associated with integrated care such as co-location and care coordination provide a framework to develop a learning culture but this requires senior managers to equip teams with the necessary human resources and financial capacity to embrace learning. Effective initiatives are often conceived by frontline staff and senior managers ought to ensure they support such endeavours by fostering an empowering and innovative culture.

Credit author statement
All authors conceived and designed the study. ML and SB undertook data collection. ML and SB undertook coding and developed the thematic analysis. All authors were involved in the process of interpreting the data. ML prepared the manuscript. All authors have read and approved the content of the final version of the manuscript.

Declarations of competing interest
None.