Testing Stratied Care for Musculoskeletal Pain in General Practice: Integrating Clinician Support and Intervention Design

Stratied care involves subgrouping patients based on key characteristics, e.g. prognostic risk, and matching these subgroups to early treatment options. The STarT-MSK programme developed and tested a new stratied primary care intervention for patients with common musculoskeletal (MSK) conditions in general practice. Stratied care involves changing General Practitioners’ (GPs) behaviour, away from the current ‘stepped’ care approach to identifying early treatment options matched to patients’ risk of persistent pain. Changing healthcare practice is challenging, and to aid the successful delivery of stratied care, education and support for GPs was required from the initial stages of the programme. This paper details the steps in integrating the development of a clinician support package throughout the 6-year programme, to support GP engagement in delivering the STarT MSK intervention. Practical recommendations are made for future general practice interventions.


Abstract
Background Strati ed care involves subgrouping patients based on key characteristics, e.g. prognostic risk, and matching these subgroups to early treatment options. The STarT-MSK programme developed and tested a new strati ed primary care intervention for patients with common musculoskeletal (MSK) conditions in general practice. Strati ed care involves changing General Practitioners' (GPs) behaviour, away from the current 'stepped' care approach to identifying early treatment options matched to patients' risk of persistent pain. Changing healthcare practice is challenging, and to aid the successful delivery of strati ed care, education and support for GPs was required from the initial stages of the programme. This paper details the steps in integrating the development of a clinician support package throughout the 6year programme, to support GP engagement in delivering the STarT MSK intervention. Practical recommendations are made for future general practice interventions.

Methods
Clinician support was developed through an iterative, mixed methods approach. Qualitative research with patients and GPs identi ed barriers and facilitators to the adoption of strati ed care, which were mapped onto the Theoretical Domains Framework (TDF) and Behaviour Change Technique (BCT) taxonomy.
Identi ed domains/BCTs were 'translated' into an educational paradigm, and an initial version of the support package developed. This was further re ned following a feasibility and pilot RCT, and a nalised support package was developed for the main RCT.

Results
The clinician support package comprised face-to-face sessions combining adult-learning principles with behaviour change theory in a multimethod approach, which included group discussion, simulated consultations, patient vignettes and model consultation videos. Structured support for GPs was crucial in enabling delity and, ultimately, a successful trial. Results highlighted that clinician support is a two-way process-the study team can learn from and adapt to speci c local factors and issues not previously identi ed. The support from senior clinicians was also required to ensure 'buy in', and results indicated the importance of monitoring GP performance and providing regular feedback.

Conclusion
Designing effective clinician support from the onset of trial intervention design, in an evidence-based, theory-informed manner, is crucial to encourage active engagement and intervention delity within the trial, enabling the delivery of a robust and reliable proof of principle trial.

Page 3/26
Background The STarT Back model of strati ed care for low back pain (LBP) (stratifying care according to each patient's risk of persistent disabling pain) has been adopted widely, following a proof-of-principle randomised trial in community physiotherapy clinics [1] and an implementation study in general practice [2]. A further programme of work, STarT MSK (Subgrouping for Targeted Treatment in MuSculosKetal conditions), has been developed to test strati ed care for a broader range of patients with musculoskeletal (MSK) pain in general practice. Conducting a proof-of-principle trial in general practice has the advantage that the intervention is simultaneously "road tested" in the target clinical setting, with elements designed in an "implementation-ready" manner, enabling rapid translation into clinical practice.
General practice is a complex environment and, currently in the UK (before and since the COVID-19 pandemic), is under unprecedented pressure. In part, this is due to a shortage of general practitioners (GPs) and an ageing population, with high levels of multi-morbidity, where management is not simply a sum of the parts [3]. Changing healthcare practice in such circumstances is challenging [4,5,6]. Important considerations are clinicians' preference for established frameworks, whether they consider the intervention meaningful and relevant [7] and perceived threats to clinical autonomy [8]. Failure to address such concerns in clinical trials that are focused on innovation can result in poor recruitment to trials, compounding poor or misleading data, and a risk that the intervention will not be adopted [9].
In clinical practice, clinical decision support systems are most effective when combined with education for the professionals using them [10] and the perceived usefulness of the interventions is a decisive factor in their acceptance by clinicians [11]. In the research domain, there is a risk of focusing exclusively on research design and delivery, with clinician education and support an afterthought. In this research programme, it was clear from the start that a thorough design process would be needed for the strati ed care intervention for MSK pain to be used effectively and without disrupting the consultation, and that the simultaneous design of a fully integrated clinician support package would be required. The multidisciplinary research team brought expertise from general practice, physiotherapy, clinical research, epidemiology, social sciences, health informatics and medical education.
The aim of this paper is to detail the steps in integrating the development of a clinician support package to support GP engagement in delivering the STarT MSK intervention.

The STarT MSK research programme
The development of the clinician support package described here was conducted within the context of the six-year STarT MSK programme consisting of four work packages (see Fig. 1, below). Several of the studies from these work packages have previously been published in this journal [4,12,13].
The strati ed care intervention is described in detail elsewhere [14,17] and comprises two components: the use of the prognostic risk tool (Keele STarT MSK Tool) and the selection of an appropriate clinical treatment, matched to the patient's prognostic risk group. In brief, the key components for intervention practices were: A computer template within the electronic health record (EHR), triggered automatically on entering a relevant MSK diagnosis or symptom into the patient's EHR, asking the GP to complete the Keele STarT MSK Tool, based on patients' responses to ten prognostic questions Automatic calculation of the patient score as being of high, medium or low risk of persistent pain Presentation to the GP of recommended matched treatment options for the patient, based on pain site (e.g. back, neck, knee, shoulder or multi-site) and risk subgroup Integration of self-management information resources to be shared with the patient Within the MSK consultation, the key objectives were: To ensure that the GP engages with the EHR to trigger the strati ed care template whilst the patient is still present For the GP to use the risk strati cation tool and to share decisions about the matched treatment options with the patient For the GP to feel con dent in being able to integrate the strati ed care intervention with their usual clinical history taking, examination, reasoning and diagnosis.
GPs in control practices, within the cluster randomised controlled trial, were required to continue "usual care", after completing a brief template to identify eligible control participants to be invited into the trial. This paper describes how the clinician support package was developed and conceptualised from the outset of the research programme, building on the evidence from work packages 1 and 2 and tested and re ned in work package 3, before being used in the main trial in work package 4.

Methods
Designing the clinical platform and clinician support package Clinician support was developed through an iterative, mixed methods approach. The pilot version of the clinician support package was initially developed throughout work packages 1 and 2, as summarised in Table 1, below.
This development began with exploratory qualitative research with GPs and patients, which identi ed barriers and facilitators to the adoption of strati ed care, giving rich insights into the environment, pressures, motivations, beliefs and expectations of both groups, and providing an understanding of the acceptability of using strati ed care and how it might t within GP consultations [4].
A theoretically underpinned approach to analysis enabled a robust and coherent explanation of the ndings. The Theoretical Domains Framework (TDF) [18] was employed, a framework that synthesises 112 psychological constructs determining behaviour change into fourteen domains including knowledge, skills, social/professional role and identity, memory, attention and decision processes. It facilitates identi cation of barriers and facilitators of clinical behaviour change at an individual and organisational level. The output of the analysis using the TDF was a series of strategies to address barriers and facilitators to GP behaviour change, which informed development of the intervention format and provided a starting point for developing the clinician support package.
In parallel to this work, we began to design the platform to deliver the Keele STarT MSK tool and matched treatment recommendations within the EMIS Web clinical EHR, used by many UK GPs and all participating practices. EMIS allows bespoke protocols and data entry templates to be designed then implemented in target practices. We designed a version of the tool to be embedded within the system for use during face-to-face consultations. This had to meet both the needs of the research and requirements of the user, i.e. complement the consultation, be easy to use in a time pressured environment and record clinical information in a meaningful way for future use.
The strategies for encouraging GP behaviour change identi ed from the qualitative research fell short of specifying an educational and support package. This required "translation" of these strategies into an educational paradigm, using well-recognised adult learning principles [19,20]. A modi ed form of Bloom's taxonomy of learning [21] suggests division into three domains of learning: cognitive, affective and psychomotor, commonly represented in the form of knowledge, skills and attitudes. The importance of this approach is that educational activities can be aligned [22] to the learners' needs (barriers and facilitators identi ed in the TDF) and to intended outcomes (use of the tool and matched treatments).
In designing educational interventions, there is a risk of privileging "delivering" knowledge -what the teacher/researcher needs to "tell", rather than what the learner/GP needs to know. Using the three domains as a framework ensures that affective components, e.g. beliefs, motivation, doubts and di culties are addressed adequately and that skills components are taught appropriately.
We realised that, for intervention practices, this study would have a signi cant impact on GP consultations (i.e. completing the prognostic tool and accessing recommended matched treatment options) which may differ from GPs' usual practice. The qualitative ndings indicated that GPs' beliefs about the validity, worth and feasibility of the approach would be important elements in the affective domain. This suggested that a comprehensive educational package was needed, ideally split into two sessions, beginning with discussion amongst GPs about how they consult and make decisions and including an opportunity to try the tool and re ect on its use.
GPs would also need some knowledge of the principles of strati ed care and how it differs from the usual 'stepped' care approach, the Keele STarT MSK tool and its derivation, the matched treatments and essential elements of the trial design and conduct. Inclusion of background information about previous studies and the study team might also contribute positively. This cognitive component could be provided by a didactic/interactive group session with slides and printed material.
The psychomotor or skills domain includes the use of the tool within the EHR and some complex consultation skills elements: interacting with a computer earlier in the consultation than is usual for most GPs, in order to launch the tool explaining to patients the use of the tool and decision support element integrating these elements into the consultation at appropriate stages managing the consultation within allocated time slots Skills development requires active participation and we designed a simple consultation simulation, using a vignette of a typical patient, for GPs in each practice to experiment and discuss approaches to integrating the tool. The design team had ideas about the "best t" of the tool and matched treatments in a consultation but were not prescriptive and learned from observation of, and discussion with, GPs in training sessions. This allowed more guidance to be provided in the main trial. Performance monitoring and feedback are important elements for encouraging engagement and we included a system for collating basic performance data at individual GP, practice and trial arm level for anonymised feedback during an early review visit to each practice.

Results
Clinician support in the pilot trial The iterative development work throughout the initial phases of the programme led to an initial pilot clinician support package being developed to be used to support GPs in delivering strati ed care as part of the pilot RCT. The pilot RCT tested the STarT MSK intervention in eight GP practices (four strati ed care practices; four control practices); full details of the pilot RCT have been published previously in this journal [12,13]. Figure 2, below gives an overview of the clinician support package; Table 2 provides full details of the structure and content of the training delivered to GPs as part of the pilot trial: We had already established that affective elements are crucially important to the success of the trial. Due to the complexity of the task, training for intervention practices was delivered by two members of the study team: one GP and one clinical researcher. Most or all GPs in intervention practices attended the training sessions, though a few attended only one of the two linked sessions. Overall, the training was delivered to time and to plan, and GPs participated actively in discussions and practical sessions. Facilitators from the local Clinical Research Network undertook the brief training for control practices in the pilot trial.
The Keele STarT MSK tool was designed so that all selections made on the template were coded within the patient's medical record. Anonymised data were extracted from each participating practice on a regular basis to facilitate analysis of the tool's usage at three levels: individual GP, single practice and trial arm. These data were an essential part of evaluating and encouraging GP engagement with the strati ed care template. We could identify any under-performing practices or individual GPs at an early stage, intervene and work with them to identify and remedy potential barriers. Examples included GPs not entering clinical codes and thus not triggering the tool, and information about using the tool and matched treatments not being disseminated to new members of staff. On a monthly basis each practice received an audit report on their tool usage at individual GP level.
A key element of the tool was to enable GPs to opt out of using it whilst discouraging the use of the "Esc" key which would have left no further engagement data in the EHR. Instead, GPs opting-out selected reasons; for instance, 'patient not appropriate' or 'no time to complete', enabling the team to assess the feasibility of using the tool in consultations.
A nested qualitative study was conducted as part of the pilot trial, in which consultations were video or audio recorded and used to stimulate recall in post-consultation interviews with matched pairs of GPs and patients, to explore the acceptability and feasibility of delivering the strati ed care intervention [13]. We found di culties in integrating the intervention in consultations within the standard 10-minute timeframe. However, GPs did report nding this easier with practice throughout the course of the pilot.
A frequent comment was that four hours of GP training was excessive and di cult to accommodate. GPs valued the skills component but felt that less was required on the trial background and recommended that the training be reduced to a single two-hour session for the main trial. They also felt that the team should focus the training on how best to integrate the strati ed care approach within a routine MSK consultation. These recommendations were incorporated into the support package and shortened training for the main trial. See Table 5, below, for a summary of the re nements made to the clinician support packages based on pilot trial ndings. As a result of direct observation of training simulations, discussion with participating GPs and the qualitative work [13], we also felt able to be more prescriptive about the best t of the intervention within a consultation and produced a short video of a simulated consultation to illustrate this. Re nements made to the clinician support package are summarised in Table 3, below: The revised clinician support package included printed and laminated prompt sheets for GPs, a training log to ensure that no GPs were excluded, and a plan agreed with each practice to cascade the training to any new recruits or locum GPs.
Whilst both GPs and patients felt the Keele STarT MSK tool added value to the consultation, some items, derived from the original self-administered tool in WP1, were judged to be "cumbersome" [See 13 for full detals]. This led to development and validation of a clinical version of the tool, with more conversationally styled questions, including a statement of the construct underpinning each item to help clinicians effectively communicate these to patients [23]. Support materials for the main trial were updated to include the revised tool and the training programme was adjusted to focus on the item constructs and the need to adhere to the exact wording of questions.
From outcome measures and feedback, it was clear that the brief training for control practices was not engaging many GPs and posed a risk of attrition bias. For the main trial we decided to use the same GP and clinical researcher pairs for clinician support in both trial arms, with a one-hour session for control practices focused on the recruitment template and the purpose and importance of the research.

Clinician support in the main trial
The nalised clinician support package comprised face-to-face sessions combining adult-learning principles with behaviour change theory in a multimethod approach, which included group discussion, simulated consultations, patient vignettes and model consultation videos. Clinician support was delivered by the same team at all 24 practices (intervention and control) in the main trial. Figure 3, below, gives an overview of the nalised clinician support package and Table 4 provides full details of the structure and content of the training delivered to GPs as part of the main trial: Almost all participants attended the single clinician support session and the level of understanding and engagement appeared strong, despite the shortened session. A major change from the pilot trial was the availability of performance data for practices and individual clinicians at monthly intervals throughout the main trial. Each month, the same study team member emailed the lead GP and practice manager, including the performance table (Figure 4), as feedback and for motivation. Besides being an integral part of educational practice, a recent paper [24] had demonstrated the effectiveness of feedback in reinforcing and encouraging behaviour change in clinicians.
Through weekly data extraction from all practices and monthly analysis and reporting, the study team monitored activity and performance of practices and individual GPs in the trial. This demonstrated a high level of engagement with the tool and use of recommended matched treatment options according to risk strati cation of individual patients.

Recommendations for future interventions in general practice
In describing the design, re nement and delivery of the intervention and clinician support package, this paper emphasises the importance of an integrated approach and of incorporating robust educational principles and practice (See Table 5).
We make the following recommendations integrating systematic clinician support into future complex research interventions in general practice: 1. General practice is a complex and pressured environment, so the potential impact of an intervention must be anticipated and explored thoroughly.
2. Structured support is crucial to enabling delity and, ultimately, a successful clinical trial.
3. Clinician support is a two-way process -the study team can learn from and adapt to speci c local factors and to issues they have not previously identi ed.
4. Professional identities are important -support needs to be from senior clinicians, perceived as understanding the task and pressures involved.

Monitoring of performance matters, as does early intervention if problems appear.
. Feedback on performance is a key element in support.
In summary, this paper describes and demonstrates the importance of designing clinician support right from the onset of trial intervention design, in an evidence based, theory informed manner drawing on the principles of action research [25]. Effective clinician support enables active engagement and intervention delity within the trial, enabling the delivery of a robust and reliable proof of principle trial. Patient recruitment to the trial reached its target in July 2019, after 14 months. Follow-up measures were completed in Feb 2020 and the trial data is currently being analysed. The results of the trial are anticipated during the autumn of 2020.   Design and validation of speci c clinical version, with constructs stated for GPs.

Sub-optimal treatment recommendations
Rationalisation and re nement Excessive length of clinician support sessions and requests to re-focus some parts.
Reduced to one 2 hour session with less background information.
Trainers reluctant to specify best t of intervention within consultation.
Application of experience gained to be more directive, including production of video of simulated consultation for training.
Some GPs missed clinician support sessions.
Training logs and prompt sheets introduced for each practice.
Delays in detecting problems and taking remedial action with practices.
Early monitoring and re-visit to practice.
Monthly feedback and personal contact by same trainer.
Poor engagement and performance by control practices.
Control and intervention practices to have training visit from GP and clinical researcher from study team.
More focused clinician support sessions for control practices.  Selection of appropriate methods, tools and content to address domains, particularly clinical and IT skills.
Session planning. Length of training, balance of components, methods and resources. Selection and preparation of trainers.
Delivery of training. Sessions booked and delivered at practices.
Records and "safety net".
Training log and plan to train any who miss session.
Monitoring of performance.
Regular data extraction and analysis.

Feedback.
Early intervention if problems identi ed. Monthly email feedback with performance data and encouragement. Evaluation.
Qualitative interviews with a sample of GPs.   Outline of the nalised clinician support package Sample of feedback data to intervention practices

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.