Implementation and scalability of shared care models for chronic eye disease: a realist assessment informed by health system stakeholders in Finland, the United Kingdom, and Australia

Background/Objectives Several health systems have implemented innovative models of care which share the management of patients with chronic eye diseases between ophthalmologists and optometrists. These models have demonstrated positive outcomes for health systems including increased access for patients, service efficiency and cost-savings. This study aims to understand factors which support successful implementation and scalability of these models of care. Subjects/Methods Semi-structured interviews were conducted with 21 key health system stakeholders (clinicians, managers, administrators, policy-makers) in Finland, United Kingdom and Australia between October 2018 and February 2020. Data were analyzed using a realist framework to identify the contexts, mechanisms of action, and outcomes of sustained and emerging shared care schemes. Results Five key themes relating to successful implementation of shared care were identified as (1) clinician-led solutions, (2) redistributing teams, (3) building inter-disciplinary trust, (4) using evidence for buy-in, and (5) standardized care protocols. Scalability was found to be supported by (6) financial incentives, (7) integrated information systems, (8) local governance, and (9) a need for evidence of longer-term health and economic benefits. Conclusions The themes and program theories presented in this paper should be considered when testing and scaling shared eye care schemes to optimize benefits and promote sustainability.


INTRODUCTION
As the prevalence of chronic eye diseases rises [1] there is a growing demand for eye care professionals to provide examinations for detection and monitoring of disease, and improve access to interventions.Budget and workforce shortages have challenged the ability of health systems to meet demand and provide accessible and timely sight-saving treatments [2].
To improve access, multiple health systems have introduced task-sharing models of care (known as "shared care" or "collaborative care") to manage patients with chronic eye diseases.In shared care, traditional ophthalmologist-led tasks are shared with non-medical teams [3].Often they involve standardized eye examinations being conducted by an optometrist, nurse, or technician who partner with an ophthalmologist to inform a patient's diagnosis and management [4][5][6].However, the mode of delivery and level of task-sharing (including clinical decisionmaking) varies widely.For example, there are referral refinement schemes whereby community-based optometrists support referral triage [7][8][9]; hospital-based optometrist-led/nurse-led schemes [5], or community-based clinics whereby optometrists examine patients with ophthalmologist oversight, either direct supervision [10] or virtual review [5,11,12].
Extensive evaluations of various modalities-both as pilots and ongoing schemes-have demonstrated that shared care is efficient, safe, and effective.Various schemes have improved access to specialist care and increased hospital capacity [5,10,12,13], reduced service duplication (i.e., diagnostic imaging) [8], improved referral targeting [7,9] and reduced service costs [10,12,14].Multiple studies have demonstrated that shared care providers who make independent clinical decisions can achieve high levels of agreement with ophthalmologists [7,8,10,12,13], and deliver reliable clinical assessments [8].
While these benefits are well-documented, fewer studies have investigated the ways these schemes are implemented, the transition from piloting to standard care, or adoption in new settings.This is useful to inform implementation and refinement in new settings, or support scaling-up of existing schemes.The realist theoretical framework can help uncover these program complexities, and identify how different circumstances generate different outcomes [15,16].Realist evaluations aim to understand what works, for whom, in what circumstances, to what extent, how, and why [16].A key assumption is that programs are underpinned by theories which explain the mechanism of change (generative causation) [15].
This study aims to understand the factors which support or impede implementation and scalability of shared care programs in the United Kingdom (UK), Finland and Australia.Using a realist approach [16] the contexts, mechanisms and outcomes of existing and emerging shared care schemes will be identified to inform program theories which can support broader implementation and scalability.

Setting and design
Semi-structured interviews were conducted between October 2018 and February 2020 with key health system stakeholders in the UK, Finland, and Australia.These countries were chosen because they had (1) emerging (Australia), sustained or mature shared care models of care (Finland and UK), and (2) similar health systems (universal health coverage, with public ophthalmology services being free for patients (UK, Australia) or subsidized by regulated co-payments (Finland)).

Subjects
Participants were purposely sampled from various levels of the health system to capture broad perspectives [17]; and because of their (direct or indirect) involvement in the implementation, adaptations, and ongoing delivery of services.Recruitment included: (i) Clinicians: ophthalmologists, optometrists, nurses, ophthalmic technicians; (ii) Managers and administrators: service managers, nurse/optometrist managers; quality improvement and data managers, departments, and hospital leads; (iii) Governance: regulatory or peak clinical organizations; policymakers.
Prior to the interviews, participants were given an invitation (email or inperson), study information, and signed consent forms.Initial sampling targeted clinicians, often ophthalmologists or optometrists, with introductions made through professional networks of the study investigators.Several participants were sought for their expertise or involvement in formally evaluated and published shared care schemes.Additional stakeholders were nominated by enrolled participants during interviews, either by request of the study team or prompted by the participant.Interviews occurred initially in Finland, followed by the UK, and Australia.

Data collection and analysis
A semi-structured interview schedule (Supplementary File 1) was adapted using Pawson and Tilley's "Would it work here?"[18].Interviews were conducted by three investigators (BF, BA, HL) either in person or by phone for 30 min to 1.5 h.During interviews, prompts informed by published shared care evaluations acted as "program theories" that were tested and validated by participants.The interviewers repeated emerging themes to the participant to allow opportunity for validation of emerging program theories through "theory gleaning" [17].All interviews included questions regarding unintended outcomes.Interviews were audio-recorded and notes taken.Audio-files were transcribed verbatim by a professional transcription company or one of the investigators (BF) and checked for accuracy prior to analysis.Several participants were invited to review or respond to selected quotes and themes for further theory refinement [17].
Coding and analyses were conducted using NVIVO (12, QSR International 2018).Transcripts were reviewed iteratively, as available.An initial coding framework of 21 codes was derived from the themes and six program theories reported in Baker et al.'s Systematic realist review [6], and were validated using deductive reasoning.The initial coding framework was used by two investigators (BF, BA) to duplicate code four [4] interviews, representing different stakeholder groups from the UK and Finland.An additional 16 codes were identified from the interviews using inductive reasoning.Three investigators (BA, BF, HL) discussed the coding and discrepancies, and came to a consensus on a final coding framework.
Preliminary themes and relationships between codes was also recorded.One investigator (BF) used the final coding framework to analyze the remaining 17 transcripts, adding 2 additional codes based on emerging ideas.Interviews involving an investigator as a participant were coded by a second investigator (BA) for rigor.

Realist evaluation framework
Emerging themes and selected quotes relating to the implementation and scalability of shared care models were identified by one investigator (BF) and presented to the remainder of the study team (LK, BA, AW, HL) at iterative stages for discussion and further refinement.Once themes were established, coding and quotes were reviewed in-depth to generate the contexts-mechanisms-outcomes (C-M-O) configurations for each theme.

Research team and reflexivity
The research team was multidisciplinary with expertise in ophthalmology (AW), optometry (LK), health economics (HL, BA, BF), public health (all), health administration (BA, BF) and qualitative research (HL, BF, LK).Two members of the study team were considered key stakeholders because they were directly involved in the delivery of a shared care pilot in Australia (AW, BF) and the UK (AW) and were invited to participate.To minimize any bias of being stakeholder-investigators, these interviews were conducted by two external investigators (BA, HL).Four members of the study team had also been involved in evaluations of an emerging collaborative (shared) care model in Australia (BF, BA, LK, AW) [10,12].

Participants
Twenty-one interviews were conducted (Finland-9; UK-5; Australia-7) with participants representing multiple levels of the health system (Table 1).Most participants were clinicians (ophthalmologists, nurses, or optometrists) typically with dual responsibility in management, governance, education, or information technology.Many had initiated shared care, with some having 30-years' experience; while others learned by participation in ongoing schemes.Several participants had been involved in more than one scheme, including movement across care settings and countries.

Types of shared care
The design, funding mechanisms, and the types of chronic eye diseases examined differed across the settings (Table 1).These schemes included hospital-based, hospital-outreach, or community-based clinics; national screening programs; and informal private partnerships.In Finland and the UK, some had tiered approaches whereby multiple schemes were implemented in one hospital.Shared care was mostly implemented as a formal program and funded publicly through the national health insurance, either through direct employment of staff (e.g., hospital clinicians), or reimbursement using a fee-for-service (e.g., community optometrists).Often patients would receive eye care with no or little co-payment.Australia also had reports of informal arrangements between independent optometrists and ophthalmologists, where funding would come from a mix of patient outof-pocket (co-payment) fees and government reimbursements.

Themes
Nine themes relating to the implementation (n = 5) and scalability (n = 4) of shared care models were derived.Key illustrative quotes for themes are presented in Table 2.The realist contextmechanisms-outcome configurations which inform the program theories are presented in Table 3.

Implementation themes
Access blocks, safety concerns, and clinician driven solutions.This depicted the rationale for initiating the innovation (shared care), and how these models gained traction.Across settings, change was often initiated by clinicians, usually ophthalmologists, but also Generating buy-in from decision makers to sustain models of care.Services demonstrated that these models of care were effective, which in turn facilitated teams to get buy-in from decision makers.This included generating evidence on service efficiency, safety, productivity, and acceptability.Participants reported that additional change management could help the transition from pilots to sustained care pathways.For example, regular communication with stakeholders to reinforce the rationale and benefits.(Table 2, quote 4d) Standardized care as equitable care.The use of standardized protocols/processes, care pathways, and proformas streamlined care.This ensured that patients were being managed at a clinically appropriate level.For example, ophthalmologists saw complex and advanced patients, while less-complex patients were assessed by optometrists or nurses.Standardized care was believed to give all patients an equal opportunity to care by removing barriers to access, such as cost or long wait times.

Scalability themes
Health care investment to incentivise and motivate.Incentives are required to motivate and encourage providers to participate and sustain the schemes.Financial incentives were essential to recognize the investments of time, infrastructure, and effort (clinical and administrative) required to deliver services.In the UK and Finland, financial incentives included fee-for-service for community providers (by hospitals), or higher duty payments.Incentives were often justified by the cost-savings accrued through task-sharing.In Australia, the financial model was not clearly developed, and lack of financial incentives available to participating clinicians was a major barrier for sustainability.Participants recognized that funding should be allocated by health departments, and recommended to extend the existing Medicare Benefits Schedule items (telehealth or integrated care) to cover store-and-forward review, or by hospitals introducing an incentive payment for participating private providers.

DISCUSSION
Informed by broad health system stakeholder experiences, this study has drawn evidence from shared eye care interventions at different stages of maturity and within various health system contexts to build an understanding of what factors influence implementation and transition of pilots into scaled and sustained programs.The use of the realist approach meant testing and validation of program theories could evolve our knowledge of how these program work [16].The implementation themes presented in this paper were reported consistently across settings, informing overarching theories to explain why these programs work.Within complex health systems, it has been suggested that innovations with obvious effect will be more readily adopted [19].In this study it was universally recognized that the impetus for change was the high demand for eye care due to the growing population of patients with chronic eye diseases and adverse vision outcomes due to delayed access.This led to health system administrators to seek sustainable and acceptable solutions (task-sharing) to increase capacity and improve access.However, the implementation of these solutions varied greatly across and within countries.By adopting a realist approach, this study has been able to establish the context-mechanism-outcome configurations (Tables 1 and 3) to provide insight into what works and how this is achieved across contexts [16].
When it came to scalability, these themes were informed by stakeholder's experiences of problems/solutions; not only from sustained models of care but also the perceived barriers for broader uptake in systems where these schemes are still emerging.4. Generating buy-in from decision makers to sustain models of care d. "it started off as a pilot scheme, which then just never stopped.We were able to demonstrate that of our new patients, we could discharge 50% of them before they even got to the hospital and that was very powerful in buying-in….I think everyone realizes that it's here to stay now."(Ophthalmologist, UK2) 5. Standardized Care as equitable care e. "Not everybody who gets sent to that kind of system needs to be in that system in the first place….We tended to see a lot of is people who fall through the cracks because they can't afford care and can't afford to wait for care.So, they end up not going anywhere.Some of those cases were actually really quite complex cases that you would need to be seen by an ophthalmologist, but was basically not seeing anybody."(Optometrist, AUS5) 6. Health care investment to incentivise and motivate f. "It's part of the work that doctors did before…it was such an important job, and it made our doctors have more time to do other things when the nurses took part of their job.So that was the reason that they [nurses] got more salary… it's about 10%, and the volume of patients is a lot more."(Hospital Director, FIN9) g. "…all these things require some sort of financial support to get them off the ground, get them working and for people to be remunerated appropriately so they stick with it and it becomes something that's sustainable."(Optometrist, AUS5) 7. Systems to integrate care and link providers h."it's making sure that there's clear communication but it's obviously safe communication, it's two-way, and it talks to all the systems as well.We don't need another standalone system, we need something that's integrated, because nobody has time to write the same thing twice.We need things to be able to go backwards and forwards in a safe, reliable manner, and easily updated when needed to be" (Policy, AUS2)    Participants in all settings strongly believed that health system investment was needed to scale services, notably through remuneration and information technology models.These investments would ensure longevity of outcomes and productivity gains.This study built on theories reported by Baker et al. [6] which describe the success and failures of similar shared eye care in the UK.Several theories were validated.For example, participants across settings validated the effectiveness of task-sharing (Theme 2; Baker et al.Theory 1), the importance of genuine local partnerships for implementation (Theme 3 and 8 Task-sharing in health care is not new and has been used widely in other disciplines to address some of the same obstacles seen in eye care; and can be useful in low income countries where health care resources are scarce [20].In Australian and UK settings, a prominent example of scaled shared care are antenatal programs which involve GPs (or midwives) and obstetricians.These programs are standard practice in most public hospitals and national clinical guidelines ensure fidelity remains.However, these programs can be adapted locally to provide responsive care (e.g., tiered approaches such as midwifery group practice or GPobstetrician programs [21].)In another example, Kemp et al. [22] suggest that scaling (national and international) of a maternal and child health program was achieved by allowing local adaptations while maintaining program fidelity.They liken this to baking a basic cake (core components) and adding recipe variations (local adaptions) to adjust flavor.
Similar core components exist in shared care initiatives used in eye care and other health disciplines.This includes supervision, training or upskilling workforce [20,22]; well-defined scope of practice [20,23]; protocols [20,22,23]; and service integration [21,22,24].Information technology is recognized as an essential component for team communication [23,24].Within eye care, large-scale diabetic retinopathy screening programs in the UK rely on teleophthalmology (store-and-forward) to transfer information across providers [25].In India, electronic medical records that interlace through a three-tiered eyecare network (primary care, secondary clinics and tertiary eye hospitals) are able to capture data and imaging for large populations over time.Furthermore, digitization of health information allows realtime analytics which can support quality control and research to measure longer-term health outcomes [26].In this current study, several participants reported investment into local IT systems.One UK example is a web-based patient record (New Medica) which has facilitated almost 25,000 virtual consultations across multiple clinics.By embedding automated quality checks and real-time feedback for optometrists, Wright et al. were able to demonstrate improvements in patient care [27].However, Sim et al. [25] report that lack of integration with a nationally accessible electronic health system is a known barrier to uptake of teleophthalmology in the UK.In our study, participants were unsatisfied with current IT systems due to lack of integration and inability to extract meaningful data.Thus, longer-term health and economic outcomes cannot be efficiently monitored; and services were unable to make informed decisions on how to allocate resources to achieve the best outcomes [28].
Information technology is not the only challenge.Resistance to change from both organizations and providers is regularly cited as a key concern across disciplines for the successful uptake of shared care [20,22,25].Participants in this study suggested change management strategies were needed, including regular communication and engagement for teams, external clinicians, and stakeholders; motivation (and incentives) for providers; and involving the right people, such as clinicians, to lead the change.Other studies recommend that effective leadership [20,25] or local champions are needed to gain "buy-in" [22] and drive change.However, team culture and power imbalances should be addressed to support uptake [19,24].In this study, interdisciplinary trust evolved organically through daily interactions and training.One Australian study examining interprofessional relationships of shared eye care providers found that trust is established by regular contact or co-location of providers [23], however in another study too much oversight and scrutiny of optometrists work resulted in diminished levels of trust [29].Thus, for broader implementation, pre-planning should consider including change management strategies to facilitate adoption.

Study limitations
This study has several limitations.First, it does not capture the concurrent experiences of piloting and implementation of shared care models, and relied on retrospective accounts from various stakeholders, some of whom had been involved in the initial pilots, program modifications, and ongoing implementation.However, the schemes in Finland and UK were in place for many years which provides valuable insight into sustainability.Second, being a qualitative study, the sample is limited to 21 participants from three countries and does not reflect the experiences of all shared eye care programs, particularly the barriers for programs that have been disbanded when unsuccessful.The contextmechanism-outcomes configurations are based on participant perceptions of events or ability to recall, thus may not cover all aspects of local implementation.However, the sample was purposely selected to reflect a wide range of by experts who have been involved in piloting, program evolution, evaluation and the development of policy, guidelines, and education.

CONCLUSIONS
Using a realist approach, this study has identified nine key factors relating to the implementation and scalability of shared eye care programs in the UK, Finland, and Australia.Overall, implementation is supported by clinician-led solutions, rearranging multidisciplinary teams, building inter-disciplinary trust, generating local buy-in, and using standardized care protocols.Scalability will require investment from broader health systems to support financial incentives to motivate providers and integrated information systems.There was a preference for schemes to be governed locally, allowing for flexible implementation.However, evidence is still needed for longer-term health and economic outcomes when scaled.Shared eye care is necessary to tackle the growing demands for eye care services, and for these programs to be effective, equitable, high quality and sustainable, findings should be systematically addressed.

What was known before
• There is a growing demand for chronic eye care services to detect, monitor, and treat disease.
• Shared care between ophthalmologists and optometrists can increase access for patients, and improve efficiency and costsavings for health systems.

Table 1 .
Characteristics of participants and shared care schemes.bynurses,optometrists,andan endocrinologist.Generally, ophthalmologists had observed negative clinical outcomes (vision loss/blindness) in their patients occurring as a direct consequence of being unable to access overburdened ophthalmology clinics.In several settings, clinical safety was the catalyst for change, prompting administrators to look toward clinicians for solutions.(Table2,quote1a) Redistribution of Health Care resources, to optimize skill sets, with minimal investment.Redistribution of existing healthcare resources by reorganizing (and upskilling) multi-disciplinary teams through task-sharing arrangements promoted efficiency.For example, fundus photographs taken by technicians, clinical decisions by nurses/optometrists, and support staff transferring patient information.Ophthalmologists generally provided supervision through direct oversight or virtual (store-and-forward) review.Ophthalmologists spent less patient-facing time with the lower risk patients, meaning releasing them to spend more time with complex and advanced patients.Participants often reported that existing resources were rearranged, rather than increased.(Table2quote 2b) Interdisciplinary Trust needed to shift clinical responsibility.Motivation and trust were needed to build inter-disciplinary relationships to shift the clinical responsibility from the ophthalmologists to nurses or optometrists.The C-M-O configurations presented in Table3demonstrate the various ways that trust was established.Examples include formal training or more organically through daily interactions.Trusting relationships were thought to lead to better clinical decision-making, with a reinforcing effect whereby improved decision-making meant ophthalmologists were more willing to release responsibility.(Table2,quote 3c) Systems to integrate care and link providers.Health systems need to invest in information technology systems which can transfer patient information between multiple providers.Across settings, there was frustration with current systems which were described as inefficient and clunky.Specific issues were the use of multiple programs and log-ins, and information being transferred via paper and scanned documents.Participants felt that investment in integrated information systems was important for scalability as it could increase efficiency and support continuity of care.
A shift to find evidence of longer-term health and economic outcomes.There is a need to generate evidence of the longerterm health and economic outcomes of these models.Current evidence generally assumed that improved capacity and access leads to better health outcomes.However, longer-term outcomes are not yet understood.Focus should shift to the impact on vision loss and a more wholistic perspective of economic outcomes when patient volume increases; including impacts on access to other eye care services, and health outcomes for patients with more advanced disease, or ocular co-morbidities.(Table2,quote 9l)

Table 2 .
Themes and illustrative quotes for implementation and scalability of shared care for chronic eye diseases in different health care settings."itcomes down to consultants wanting to change things for the better of the patients… the delay in review of patients with glaucoma… every single one of those cases where someone had a reduction in vision because of delay, was essentially, was called a serious incident… that gets reported to the National Patient Safety Agency, etc. It's a very, very big, and rightfully so, a big hullabaloo."(Ophthalmologist/Manager,UK1) 2. Redistribution of Health Care resources, to optimize skill sets, with minimal investment b."youdon't necessarily have to have an ophthalmologist at the top of the tree but essentially the greater the clinical risk of blindness the more involved an ophthalmologist needs to be.[Optometrists] outnumber ophthalmologists six to one.So we need them, just from a workflow point of view.They're not using the most of their training… it'd be really good to get everyone to work at their best capacity." (Ophthalmologist/Manager, AUS4) 3. Interdisciplinary Trust needed to shift clinical responsibility c. "Being clinically competent has to be a key skill of the people who we trust to see our patients.We have to have a level of security that, what they're saying and what they're seeing is correct and accurate…[and] confidence in the findings because especially with things like pressure, you know, if it's a nonsense [unreliable/ inaccurate intraocular] pressure, that makes a big difference to the clinical management of the patients.We do a lot of training with the community optometrists before they're left to do their community glaucoma clinics, and I guess that trust that you mention, and that confidence builds up during that training period."(Hospital Optometrist/Manager, UK4) "…before I did anything, I contacted the Finnish Health Authority and explained what we're planning to do, and ask them if is this okay.And they say, well I'm responsible for everything and if I think this is fine, it's fine.So, I first wanted to make sure that it's legal."(Ophthalmologist/Chief Executive, FIN3) j. "I set up the policies and protocols and sought out the service level agreements through our Commissioning Department here at the hospital… we have named an accredited optometrist at each practise… People who failed to attend the annual accreditation meeting, or if they don't respond to direct feedback about poor referrals or something like that then we'll usually try to find out why.Usually because they've just left the area or something like that… but yeah, we have terminated people from the scheme in the past" (Hospital Optometrist/Manager, UK4) k. "you need to account for local variations… some overarching regulation, particularly more money on AMD and if our budget is not increasing to the same extent, we have to do something else somewhere in some other disease… Otherwise it's very hard to prove that what you are doing makes sense even if you could show that it saves cost here, but if it incurs costs somewhere else." (Ophthalmologist, Chief Executive FIN3)

Table 3 .
Context, mechanism, and outcome configurations for implementation and scalability of shared care for chronic eye diseases in different health care settings.

Table 3 .
continued ; Baker et al.Theory 2), and the benefits of standardizing care (Theme 5; Baker et al.Theory 3).However, Baker et al.'s [6] program theories were founded on studies which report mostly on outcomes and effectiveness of interventions, and were less focussed on implementation processes and contextual factors (Baker et al.Theory 6-Barriers and enablers).Thus, in this study, an important methodological component was inductive analysis to refine and generate new program theories to explain aspects of program implementation.In this study, clinician leadership (Theme 1) and interdisciplinary trust (Theme 3) were mechanisms needed to effectively facilitate task-sharing and shifting of responsibility as reported in Baker et al.'s Theory 1.