The systematic development of a novel integrated spiral undergraduate course in general practice.

Keele University School of Medicine opened in 
2002, initially delivering the Manchester MB ChB 
curriculum, with the aim that from 2007 it would 
deliver a 'distinctive Keele curriculum' which would 
'graduate excellent clinicians'. This curriculum would 
be spiral and integrated in design, and delivered by 
a 'hybrid' model of learning methods, from small 
group interactive (including problem-based learning) 
through to large group and more didactic methods. 
It would be student-centred and include a strong 
focus on community and primary care contexts, the 
latter reflecting the origins of the university which 
grew out of the Workers' Education Alliance after the 
Second World War as the University College of North 
Staffordshire, becoming the University of Keele in 
1962. The university has always had strong links with 
the Potteries and North Staffordshire, communities 
with long-standing problems of poor health status 
and recruitment of the necessary health workforce. 
While there are strong local drivers for medical 
students learning in and with the community, this 
is in a broader context of primary care becoming 
an increasingly important resource for medical 
undergraduate education arising from policy, health 
service and andragogy. We were charged with the responsibility of 
developing the community element of the curriculum. 
This paper describes the systematic development 
of our general practice curriculum from 2007 to 
the present. We describe the decisions we made 
and general practice's current contribution to the 
curriculum. We will present our evaluation of the 
curriculum in a separate paper.


BACKGROUND
Keele University School of Medicine opened in 2002, initially delivering the Manchester MB ChB curriculum, with the aim that from 2007 it would deliver a 'distinctive Keele curriculum' which would 'graduate excellent clinicians'. 1 This curriculum would be spiral 2 and integrated in design, and delivered by a 'hybrid' model of learning methods, from small group interactive (including problem-based learning) through to large group and more didactic methods. It would be student-centred and include a strong focus on community and primary care contexts, 3 the latter reflecting the origins of the university which grew out of the Workers' Education Alliance after the Second World War as the University College of North Staffordshire, becoming the University of Keele in 1962. The university has always had strong links with the Potteries and North Staffordshire, communities with long-standing problems of poor health status and recruitment of the necessary health workforce. While there are strong local drivers for medical students learning in and with the community, this is in a broader context of primary care becoming an increasingly important resource for medical undergraduate education arising from policy, health service and andragogy. 4 We were charged with the responsibility of developing the community element of the curriculum. This paper describes the systematic development of our general practice curriculum from 2007 to the present. We describe the decisions we made and general practice's current contribution to the curriculum. We will present our evaluation of the curriculum in a separate paper.

THE KEELE CONTEXT
Keele University School of Medicine is a small school which admits 129 students each year the majority of whom are school leavers: the proportion of school leavers varied from 59% to 93% between 2007 and 2014. It has a five-year course and teaches from modern state-of-the-art buildings in the University, its major partner teaching hospitals in Staffordshire and Shropshire and a large body of community partners of which currently 118 are general practices.
The general practice curriculum did not evolve in a vacuum and had to complement the overall curriculum. The initial blueprint for the curriculum is shown in Box 1. This demonstrates its spiral nature, revisiting knowledge and skills but with greater challenges each time, the integration of non-clinical and clinical learning throughout the five years and with increasing focus on clinical skills as the course progresses.  6 ) and teach what general practice can best teach rather than teach general practice as a specialism. This would help address the 'curricular pathologies' of inertia and hypertrophy. 7 • Primary care is generalist and holistic and therefore ideally placed to provide exposure to a wide range of problems and to help students understand the interplay between clinical, psychological and social effects of illness. Nevertheless, because we would deliver core learning required by most doctors for most of their professional lives, learning would be generic rather than vocational. We would not aim to educate students for a career in general practice but for careers in which general practice would play an increasingly important role irrespective of the discipline they choose to join. • Increasingly, the doctor's task is to help patients make difficult decisions which are highly context-dependent in situations of considerable uncertainty and where there is often no single clear 'best answer'. This requires judgement and highly developed cognitive and interpersonal skills. It is these skills rather than the doctor's knowledge which differentiate patients from their doctors. We wanted therefore to focus on the cognitive and communication skills that would facilitate knowledge transfer and decision-making. • The School aims to deliver integrated learning with the medical sciences and clinical medicine being taught together and all clinical disciplines contributing throughout the curriculum. We did not want to create a division between teaching in primary and secondary care so general practice is seen as an integral part of clinical learning. • Sustainability is critical. There is no point in delivering a novel curriculum which exhausts general practices leading to delivery of the curriculum becoming unsustainable in the medium or long term. • While recognising that the community is a resource for the School and its students, we Box 2 Guiding principles for the general practice course The course will: • Teach core learning required by most doctors for most of their professional lives which general practice is best placed to deliver. • Be generic rather than vocational.
• Be predominantly skills-and in particular cognitive skills-based, rather than knowledgebased. • Be an integral part of clinical learning.
• Be socially responsible.
felt it important to highlight that the School and the student body are also a resource for the community. The social compact between medical schools and their communities is increasingly recognised and we believed that we should give something back to the community for its help in teaching our students. 8

Initial considerations
These principles led to an initial set of decisions (Box 3).

Box 3 Initial design considerations
The course would: 1 A focus on consultation skills. The consultation is the core of clinical practice 'and all else in the practice of medicine derives from it'. 9 All graduating doctors must be able to consult and all clinicians consult throughout their clinical careers.
A key element of the development of any skill is sustained deliberate practice 10,11 supported by feedback on performance. 12 General practice offers unparalleled exposure to variety and numbers of patients and repeated opportunities for students to practice their consultation skills. Virtually all general practitioners have, as part of their postgraduate training, been supported to develop their own consultation skills and are wedded to the concept of receiving and giving feedback. They are well placed to provide feedback and support the deliberate practice of these skills. General practice is therefore ideally positioned to support the development of the consultation skills which will be needed by most doctors throughout their professional lives. 2 Consultation skills include clinical reasoning skills. The ability to make difficult judgements in difficult situations is arguably the medical practitioner's single most important skill and 'unique selling point'. Yet the component skills are seldom explicitly taught, their teaching is fraught with difficultly 13 and many feel that they develop through the accretion of experience. 10,13 Whether clinical reasoning skills can be taught or not, general practice offers unparalleled exposure to people with 'new' presentations with no prior sorting by clinical discipline or pathology so for any presentation the possibilities are constrained only by sex and age. This is rich territory in which to learn and, with 'sustained deliberate practice', to start to develop expertise. 3 Include multi-morbidity and complex and continuing conditions. As our population ages and the prevalence of multi-morbidity increases, all clinicians irrespective of their discipline will provide continuing care for people with multiple morbidities. 14 It is therefore essential that tomorrow's doctors develop the skills to manage people with multiple morbidities not only in single episodes of care but to participate in their continuing care, something which needs to be learnt in general practice. 4,14 4 The course would build on the natural strengths of general practices. Successful general practices are unique, independent small businesses highly adapted to the environments in which they work and responsive to external stimuli. We wished to tap into this creativity and adaptability and not to prescribe how practices would teach. We would define key 'deliverables' for the practices but would not specify how they would be achieved. We also offered suggestions as to what other activities would be appropriate. By maximising flexibility of ways in which practices could deliver teaching, we also maximised their ability to teach long term. 5 Enable practices to contribute in different ways.
Not all practices can engage in teaching to the same extent. Some want to make a major investment in teaching, reshaping their practices to enable placement of relatively large numbers of students. Others can only make a modest contribution. We wished to support practices which lay at any point on this spectrum, recognising that across the curriculum there was a need for both brief practice visits and longer, more immersive general practice placements. This inclusive, developmental approach meant that there was 'something for everyone' who wanted to contribute. 6 Put something back. We set out to identify community partners to ensure that students had an opportunity to make a contribution to the community. 7 Establish clear descriptors against which GP tutors can gauge progress. It was essential that tutors knew the School's expectations of students at each stage to gauge progress and to intervene as necessary. RIME 15 (Box 4) is an evaluation framework that offers guidance to GPs on the expected learning outcomes at each year of the course. Unlike other clinical placements, the clinical content of the GPs' caseload is likely to be more consistent from Years 3-5, but what students should master will evolve from being able to describe (report), to understand (interpret) and then to manage and finally to an ability to contribute to patient education.

Box 4 RIME 15
• Reporter: The student can reliably report the patient's history and reliably elicit and describe the findings on physical examination.
-End of year 3.

The curriculum
The general practice curriculum splits into two major phases: phase 1, years 1 and 2, and phase 2, the major spiral in years 3, 4 and 5 ( Figure 1).
• Years 1 and 2: In these early years students have a series of half-day placements in general practices, hospitals and a broad range of third (voluntary) sector providers. The goals of this programme range from vocation testing, to skills practice in support of learning in the skills laboratory, to the Year 2 SSC when every student works on an advocacy project with a third-sector organisation.

• Year 3: Consolidation of Clinical Skills (CCS): this
is a four-week block at the end of the year 3 after students have spent 24 weeks learning in a predominantly hospital environment and have taken their end-of-year examinations. The goal of the block is to enable students to consolidate their skills by providing intensive exposure to consultations with patients: they are expected to be involved in 160 consultations and lead 60 during the placement. They receive frequent informal feedback on their consultation skills and a minimum of three formal workplace-based assessments (WBAs) supported by written feedback. Assessment and feedback were considered critical to the success of the whole curriculum and so were piloted from 2008 in the final three years of the Manchester curriculum. Although the block is 'full time', practice-based learning is supported by one 'cluster session' each week when students from a cluster of neighbouring practices spend half a day learning together, facilitated by a GP teacher from one of the practices. These sessions include a range of activities from peer review of consulting skills using recorded consultations to case discussions to peer teaching. They were not considered a critical element of the CCS block but were considered critical for the year 5 GP Assistantship so we piloted, refined and familiarised students with this learning environment over two iterations of year 3.  (15 weeks) is in general practice. This is an assistantship as described by the GMC in TD2009: 6 students are expected to become part of the practice team and 'to be missed' when they leave. They have a target of conducting 375 consultations seeing patients who have requested 'on the day' consultations as well as people consulting for ongoing care of chronic conditions and health promotion sessions, many on multiple occasions. They work with GPs, nurses, nurse practitioners, healthcare assistants and other healthcare professionals providing procedure-based as well as consultation-based care. They have a series of WBAs throughout the assistantship and their learning is guided by their GP tutor who is their educational supervisor. They spend four days a week in the clinical environment; the fifth day is split between a clinical cluster session and a cluster project session. The clinical cluster session is similar to the cluster sessions in the year 3 CCS block but serves an important social support function as well as its educational one. This was considered to be a critical element of the assistantship and was piloted using the year 3 CCS clusters. In the cluster project the group works with a local community organisation to help address a need it has identified. This innovative community engagement offers our students the opportunity to develop leadership and service development skills and to make a real and lasting contribution to the community in which they have been learning.
Preparation for the implementation of the curriculum We identified a series of challenges which had to be addressed if we were to implement the curriculum successfully: • Developing an agreed consultation skills model. If we were to deliver integrated consultation skills teaching we would have to engage teachers in all clinical disciplines and move past the rhetoric of 'that's the way they do it' (substituting general practice or hospital for 'they' depending on context). A vital step was to construct a simple Venn diagram representing consultation skills as overlapping skills. These included 'traditional' clinical skills, communication skills and less commonly included skills such as procedural skills and information management. The aim was to illustrate how all clinical disciplines could contribute to the teaching of clinical and consultation skills and set each in context ( Figure  2). • Acceptance of a single framework for assessments.
The next task was to develop a single assessment tool which could and eventually would be used for all assessments of consultation skills whether formative or summative in simulated or real clinical environments from year 1 to year 5. The development of this instrument, GeCoS, is described elsewhere 16,17 but key to its acceptance was that it was developed and agreed by a group of hospital and general practice teachers, reducing the likelihood of it being perceived as 'the way they do it' in just one specialty. The School-wide skills teaching group adopted GeCoS as the framework for teaching consultation-based skills from September 2008 and the School's assessment committee adopted it for clinical summative assessment from 2009.  • Development of tools to enable practices to provide written feedback. We decided we needed to support practices to formulate written feedback quickly which contained specific suggestions on how to improve. 17,18 This required the development of specific strategies to improve any of the competences in GeCoS. 17 These were embedded in a web survey and when completed, a summary was automatically sent to the student, tutor and School by email to provide a written record of each WBA. This has been superseded by a mobile 'app' for academic year 2014-2015. • Engage and prepare local general practice to deliver the curriculum. We recognised that the curriculum we were constructing would require a large increase in the number of teaching general practices and that the practices would have to change what and how they taught. We started a systematic conversation with practices in May 2007, three years before the first year 3 students and over four years before the first year 5 students would be placed, in which we discussed what we wished to achieve and why. We also systematically asked practices how we could design the teaching to make it as straightforward as possible for them to deliver and about their perceptions of their development needs. These discussions continued over two years and informed the development of the detail of the curriculum and the practice teacher development programme which would evolve into an integrated spiral three-year programme in parallel with the student curriculum. • Developing and disseminating a minimum set of key quality indicators. Alongside the engagement of general practices in development of the curriculum, we developed the key performance indicators. These were that practices should facilitate a minimum number of appropriately supervised student consultations, perform a minimum number of WBAs and facilitate the appropriate number of cluster sessions (see below) during each placement. • Recruitment. We estimated that we needed to increase the number of teaching practices from 66 in 2006 to approximately 110 by 2011, approximately 50% of the practices in Staffordshire and Shropshire. We set in motion a systematic recruitment programme; team members took responsibility for identifying potential teaching practices in a locality and then approaching them to discuss a potential role in teaching, a form of academic detailing. 19 • Identification and piloting of critical elements of the curriculum. Because general practice was to make a core contribution to the development of our students' clinical skills, the risk associated with the GP placements increased. It became critical that the placements were successful; failure would seriously damage our students' chances of graduating and succeeding in their careers. We therefore identified critical elements of the curriculum and piloted them either within the Manchester course or within the new Keele course before they became critical.

DISCUSSION
We have developed an integrated spiral curriculum which makes a major contribution to the development of students' clinical skills at Keele. Students spend a total of 23 weeks on clinical placements (22% of the clinical placement time in years 3 to 5) in general practice learning core generic clinical skills relevant to medical practice in all specialties. Although the Keele Academic General Practice team has worked to a clearly expressed overall plan, the detailed curriculum was developed in close collaboration with the general practice teaching community; close attention was paid to ensuring it utilised the strengths of general practice overall and of individual practices, minimising barriers to and maximising participation by practices.
There is a long literature concerning the contribution of general practice to undergraduate medical education. 20,21 This has included thought pieces on what general practice could and should contribute, problems with undergraduate medical education to which it could offer solutions, 4,20,22 local regional and national surveys of current teaching, future plans and capacity to teach in general practice and descriptions or evaluations of courses which have been novel for different reasons. 21,[27][28][29] Of these, the most recent development is the longitudinal integrated clerkship (LIC). 30 The drivers for each innovation are usually multiple but include responses to regulatory pressure, andragogic concerns, evolving health economies, aspirations to change healthcare provision, regional and local imperatives and institutional aspirations. Long term the most successful innovation has been moving clinical undergraduate medical education from tertiary and secondary centres into primary care which was almost unknown 40 years ago.
General practice now makes disparate contributions to undergraduate teaching in the UK. At one level this is reflected in the proportion of the clinical curriculum delivered in general practice (3% to 30%, average 13% 31 ). However, total contribution is only one dimension of diversity. Some schools still provide block clinical placements in general practice for one or two students usually in years 3, 4 or 5, others have longitudinal placements with groups of students learning with practices one day a week over several years, usually in years 1 to 4. Often schools combine longitudinal group placements in early years with block placements of one or two students in later years. One graduate entry Irish school has a long placement in general practice, based on its original design at Flinders University in Australia, where more longitudinal, community-based models are more common.
The drivers for development of our course include institutional aspirations, andragogical principles and proactive responses to an evolving local health economy in which hospital capacity has decreased. This reflects expected long-term changes to health provision in the UK to a more primary care-based service which will require both a strong primary care workforce and a hospital workforce which is more aware of the strengths and skills of primary care. The development of the course was informed and influenced by previous work. The overall organising principle of the Keele course is as a spiral curriculum which has a long history in general 32 and in medical education. 33 The staging of learning across the three years is from RIME, a North American evaluation framework. 15 The Higher Consultation Skills course 34 in year 4 is a development of the Clinical Methods Course which has been delivered at Leicester for many years. 32 The original vision for general practice's contribution to the Leicester course was that the clinical methods learning would be consolidated with a final-year general practice placement. Although never realised, this was the precursor of students consolidating their basic clinical skills in general practice in year 3 and, in year 5, their higher consultation skills. The year 5 placement was further developed using the experience of those who had delivered long-term placements in the community and general practice in Cambridge, UK, 28 Australia 29 and North America 35 and given fresh impetus by the GMC's Tomorrows' Doctors requirement to provide students with assistantships in their final year. 6 Our course offers considerable strengths. It has been carefully developed according to, we believe, sound educational principles to provide our students with a future proof undergraduate education which prepares them well for the world of work by providing intensive clinical experience including decision making and patient management. We believe that it prepares our students for a primary-care-oriented health economy and offers them unparalleled opportunities to develop high-order consultation and cognitive skills. Our students will have consulted with at least 525 patients, many under the direct supervision of an experienced clinician before graduation. We are, however, fortunate. Keele is a new school with little 'educational baggage' and we started with a relatively blank canvas on which to plan. Keele is also a small school and we acknowledge that running this curriculum in a school two or three times our size would be challenging. Nevertheless we do not have privileged access to practices: the ratio of first-year medical students at Keele to the population of Shropshire and Staffordshire is 1:8500 while the national ratio of first-year medical students to national population is 1:8210 36,37 suggesting that we do not have access to a much larger population than other UK schools. Many of our practices had never taught before, bringing freshness to teaching.
Although the course was built to be sustainable, it will evolve with time. For example, the fourth year changed from the original one-week 'slices' to a single block of four weeks but is offering the same amount of time in general practice. This is part of a wider change in the architecture of the year to accommodate pressures elsewhere in the curriculum and the essence of the course is unchanged.
We consider that, at Keele, general practice makes a critical and sustainable contribution to undergraduate education. This arises from our systems approach to its design and its horizontal and vertical integration with the rest of the clinical curriculum. Finally it is unique in the UK in the scale of its contribution to clinical learning. We will present the evaluation of the curriculum in a subsequent paper. of the work on the IT underpinning WBA feedback. Finally we would like to thank the practices who helped us with the development of the curriculum and who support our placement programmes and their patients without whom this could not have been accomplished.

Contributions
The overall curriculum was devised by RKM and RBH. RKM led the design and implementation of the curriculum and recruitment of practices and tutor development. RKM and PC initiated the practice liaison and development of the curriculum with practices. RGJ designed and led the year 3 CCS block, SPG and MHB designed, piloted and led the year 4 HCS block. SHG designed and led the year 5 assistantship for its first year before SPG took over this role. PC developed the third-sector projects. MHB, PC, SPG, SHG and RGJ led practice recruitment and development in their respective patches and together with RKM delivered central tutor development. The first draft of the paper was written by RKM with all authors contributing to successive drafts and approving the final version.

Conflicts of interest
None of the authors has any conflict of interest apart from the continued success of the course.

Ethical approval
This was not necessary as we have not used any individual student, practice or teacher data in the paper.

INTRODUCTION
Compared with their counterparts in more affluent areas, patients in deprived areas die younger, are sicker for longer before they die, and present more complex problems to primary care. 1,2 Furthermore, people in areas of socio-economic deprivation are often underserved, as primary care is relatively understaffed and under-resourced and less able to meet patients' needs than primary care in more affluent areas. 3 Important barriers to achieving high-quality primary care in deprived areas are higher levels of complex clinical problems and multi-morbidity, which increase demand for service delivery time. Consequently, time available for service development and quality improvement is also limited, and there are difficulties with training and retaining staff. 4,5 Rhondda Cynon Taff, Merthyr Tydfil and Blaenau Gwent are among the most deprived areas of Wales and the United Kingdom, with large proportions of the region in the lowest quintile of the Welsh Index of Multiple Deprivation. Several health indices are amongst the highest in Wales such as standardised mortality rates and rates of unemployment for 16-64-year-olds. 6 In 2001, Cardiff University and the Welsh Government set up the Academic Fellows' Scheme to address some of the difficulties faced by primary care in deprived areas of Wales, particularly Rhondda Cynon Taff, Merthyr Tydfil and Blaenau Gwent. The Academic Fellows Scheme in South Wales employs recently qualified GPs as academic fellows (AFs). The scheme aims to provide developmental support to general practices in deprived areas of South Wales, whilst promoting the professional development of the AFs in teaching and research. In this paper we describe the background to the scheme and its performance and achievements over the ten years since inception.

CONCEPT AND AIMS OF THE ACADEMIC FELLOWS' SCHEME
The scheme aims to attract enthusiastic newly qualified GPs who are committed to service development, research and teaching. Fellows spend two days a week working as front-line GPs in areas of socio-economic deprivation, freeing up host GPs to undertake a structured programme of service improvement. Fellows spend the remainder of their time at the Cochrane Institute of Primary Care and Public Health at Cardiff University School of Medicine (http://medicine.cf.ac.uk/primary-care-public-health) undertaking further professional qualifications, developing and delivering undergraduate and postgraduate teaching, and participating in research.
Any practice in a deprived area of South Wales can apply to the scheme. Practices submit an initial application form outlining practice structure and function, and confirming that the majority of their patients live in areas of high deprivation. 5 Following acceptance, practices complete a proposed practice development plan (PDP) stating intended improvements, timescales, personnel involved, and outcome measures. Practices meet with the scheme director at the start, at three months, and at six months to review progress and update PDPs. Host GPs are expected to present their practice developments at the annual general meeting to all participants and stakeholders.
Benefits of the scheme are multidirectional. GP practices in deprived areas benefit from time and support from the AFs. Protected time away from front-line service delivery has enabled GPs in these practices to undertake clinically important and innovative service improvement projects that otherwise would not be possible. The AFs, as newly qualified GPs, gain further training and experience in an academic environment at the start of their careers. They are also exposed to working in areas of real need which, without the mutual benefits of the scheme, may not have attracted them previously. Through the work of the Fellows, the university establishes links with historically hard-to-reach communities and can utilise these for research and teaching purposes.

Host practices
To date, the academic fellows scheme has employed 28 Fellows and supported 32 general practices. An evaluation in 2007 by the Welsh Institute of Health and Social Care concluded that the scheme was achieving its aims, with considerable impact on general practice in deprived areas. 7 Reports from supported practices reflect a conscientious and effective use of the time provided, with exciting and innovative service improvement projects completed and implemented on schedule. Projects ranged from addressing service gaps for local populations (such as developing and implementing enhanced contraceptive services), to addressing issues of national importance, such as reducing benzodiazepine and opiate prescribing (Box 1). A GP who has benefited from the scheme expressed: 'When I took over a failing practice eight years ago there was a great deal of work to be done to bring the practice to an acceptable standard, and much of what has been done would not have been possible but for the invaluable support given to me by the academic fellows scheme'.

Fellows
The scheme has succeeded in meeting the career needs of the academic fellows. Fellows have developed novel ways of delivering core undergraduate medical teaching such as collaborating with a theatre company to develop a studentselected component (SSC) for third-year students around the use of drama to deliver health promotion and public health messages. Another exciting initiative was the introduction of gynaecological teaching associates, volunteers trained to teach medical students intimate examination using their own bodies. 8 These innovative approaches to teaching have led to fellows, and students supported by fellows, winning Cardiff University teaching prizes.
Fellows have participated in research with 26 (96%) publishing papers in peer-reviewed journals such as the BMJ, British Journal of General Practice and Cochrane Database of Systematic Reviews. All 28 fellows gained further postgraduate qualifications, ranging from Postgraduate Certificates in Medical Education to Masters in Public Health (Table 1). Several fellows used their research time to 'pumpprime' applications for further research funding with successful awards of project grants and research fellowships from the Medical Research Council and the National Institute of Health Research. For the South Wales' valleys Seventeen (61%) former AFs continue to work in general practices located in deprived areas in South Wales. This is a significant achievement, as these areas have traditionally struggled to recruit and retain GPs. Former AFs have contributed widely to their new practices; eight former AFs teach medical students, four have become GP appraisers and one has become a practice lead for research. Two-thirds of the ex-AFs have continued part-time employment with Cardiff University (six in research and 11 in teaching) maintaining and strengthening the academic links between their practices and the school of medicine.

CHALLENGES
Host practices found the process of completing a practice development plan (PDP) challenging. PDPs would often focus on service delivery rather than service development. The scheme addressed this in several ways: 1 Introducing a structured, outcome-oriented application form which made explicit what changes were intended, who would do it, how long it would take, and how outcomes would be measured. 2 Insisting that projects were for service development not delivery except where limited delivery was required to pilot and assess feasibility of implementing the service into routine practice. 3 Encouraging practices to collaborate with their intended AFs and the scheme director during the PDP phase to ensure all parties were satisfied with the proposed development plan.
Challenges for the AFs mainly centred on settling into the academic environment. In addition to their general practice workload, the AFs needed to identify a research project, learn to design and deliver teaching, and pursue a postgraduate diploma or masters. This demanded the acquisition of time management skills and the ability to identify and address priorities. This was addressed by assigning academic mentors and encouraging AFs to join research teams based on their own interests. Formal courses were organised at the start of the posts targeting research methods and teaching skills. Challenges for the scheme include the organisational and policy changes in general practice such as the introduction of the Quality and Outcomes Framework and changes to out-of-hours' work. The scheme has therefore needed to be fluid and adapt to support the various demands placed on already burdened GPs and to be able to continue to meet their requirements. Projects have changed over time, but a strong feature of the scheme has been to support practices in formulating a development plan that suits their requirements.

DISCUSSION
The academic fellows scheme, from a small but ambitious beginning, continues to grow in strength and in demonstrating positive outcomes. The scheme is meeting the needs of practices and fellows and, through the many novel and exciting completed PDPs, contributes to improving primary care in the South Wales valleys.
Supporting primary care in deprived areas is not a new idea. The Deep End Project is a network of roughly 100 general practices serving the most severely deprived populations in Scotland and works to promote primary care in these areas. This project has been used professionally and politically to lobby for change in policy and investment. 9 The AFs scheme has adopted a more 'bottom-up' approach, showing how standards of care can be improved by supporting GPs to carry out service improvements important to their local population. The scheme has provided academic training to newly qualified GPs, exposed them to work in areas of deprivation and, inadvertently, has been pivotal in many of them continuing to work in these areas. Thus it has the potential to address the recruitment and retention challenges facing the profession. This has led to an increase in the number of practices in these areas offering medical student teaching and participating in research, vastly improving the connection between these practices and Cardiff University School of Medicine. Looking forward, this will surely be beneficial for future recruitment and retention of GPs and the quality of the primary care delivered.

THE FUTURE
The scheme will continue to support primary care in deprived areas of South Wales. A Scheme website (www.pairedpractices.co.uk/) has been set up to allow practices to share knowledge and their experiences of the scheme and, most importantly, their PDPs. GPs who have completed PDPs can offer to mentor practices new to the scheme, providing advice on PDPs and addressing difficulties practices may encounter in achieving their aims. Through this forum, we hope to create a network of practices and GPs with expertise in service improvement in primary care in deprived areas.