Health CommitteeWritten evidence from Professor Alan Maynard, University of York (ETWP 115)

Alan Maynard is a Professor of Health Economics in the Department of Health Sciences and the Hull-York Medical School at the University of York. He was involved in a non-executive capacity in NHS management for 27 years, and from 1997–2010 he was Chairman of York Hospitals NHS Foundation Trust. He has worked as a consultant for the World Bank, the World Health Organisation, the European Union and the UK Department for International Development in over 20 countries.

He has been a critic of NHS workforce planning for over 30 years. In a series of publications with Dr Arthur Walker he has advocated, inter alia, more sophisticated methods of forecasting and investment in the evaluation of skill mix options (eg nurses replacing GPs in primary care). Some relevant publications by Maynard and Walker are: “Doctor Manpower 1975–2000” (Merrison Royal Commission on the NHS, HMSO, 1978);”Too many doctors?” (Lloyds Bank Review, 1977); “Managing the medical workforce: time for improvement? (Health Policy, 1995).

More recently he has worked with Dr Karen Bloor on a programme of work on NHS consultant productivity. The results of the work she has led are discussed below.

The lack of impact of such work on practices and policies in “Whitehall Village” may be indicative of the old American saying “what’s regular ain’t stupid”! Or that improved productivity in the use of the NHS labour force cannot enter the policy agenda as it is both too great a threat to existing restrictive practices and yet another demonstration of the failure of research to impact on policy making in the “London bubble”!


Planning the NHS workforce is a crucial task. Ensuring that the right numbers of the right staff are educated, trained and available to work in the right areas of the NHS at the right time is extremely challenging. Forecasting the medical workforce is perhaps the most difficult task, due to the long time lag between entry to medical school and emergence of a fully trained GP or consultant. Historically, medical workforce planning has been undertaken separately from planning other NHS roles, but this is increasingly inappropriate given changes over time in skill mix and role development.

In this note I approach NHS workforce planning mainly from the supply side, raising a number of persistent issues relevant to planning the supply of doctors, nurses and other health care professionals. It is however essential to consider the demand for health care in order to model and forecast an appropriate workforce for the future.

1. Supply Side Issues

1.1 Changing skill mix

Over recent decades there has been considerable discussion of and some development of an evidence base for, changes in the skill mix of health care provision. This is particularly evident in primary care, where nurses have taken on tasks such as immunisation, screening and health promotion, and are increasingly developing roles further with nurse prescribing and case management in chronic disease. In hospital care there has perhaps been less change in skill mix, and there are fewer research studies to inform change. Overall, it is essential to remember that as workforce roles develop, this tends to be accompanied by grade inflation and wage increases, so that simplistic assumptions of ‘cost savings’ from skill mix change rarely apply.

1.1.1 Skill mix in primary care

The Burlington experiment was a Canadian randomised clinical trial of the comparative effectiveness of nurse practitioners and physicians in the delivery of primary care. Its authors found that mortality was similar between the two groups of providers and there was no difference in the physical, social and emotional quality of life of patients in the two groups. Patient satisfaction with nurse practitioners and physicians was similar (NEJM, 1974).

A systematic review of this and subsequent research found that nurses tended to offer patients more advice and often achieved higher levels of patient satisfaction, but the depressing fact is the relative absence of well-designed studies. The Cochrane reviewers (Laurent, Reeves et al (2004)) reviewed the literature systematically. They initially found 4,253 studies but after applying tests of scientific robustness they included only 25 papers in their review and these were products of only 16 studies. As with most Cochrane reviews, their work reveals lots of opinion and poorly designed studies: a nice example of the fact that academic publications are often of poor quality and R&D funds are often squandered!

These reviewers concluded that nurses had the potential to deliver high quality care with good outcomes for patients, but the few existing studies did raise concerns about nurses tending to operate more slowly than physicians and to order more tests. Furthermore the short follow-up periods associated with the available studies (usually less than 12 months) and methodological problems in the design of many studies means that their conclusions, though optimistic, were also cautious.

An important policy issue is whether nurses are complements or substitutes for physicians. Does the employment of nurses extend the scope of care with little or no effect on physician workload? Or can nurses take over physicians’ tasks freeing them to extend services or be made redundant?

The Burlington authors (1974) noted that altering skill mix was not financially profitable for doctors and thus the development of nurse practitioners was likely to be limited. So it was and is in Canada. The way in which Canadian physicians are funded (payment by fees for service) ensure that potential nurse competition is emasculated. Despite these obstacles there is evidence of increased use of nurse practitioners in Canada and of patient satisfaction with the services they offer. With often inadequate access to physicians, Canadians appear to be voting with their feet (see Mythbusters: seeing a nurse practitioner instead of a doctor is second class service, Canadian Health Services Research Foundation, 2010).

In the UK as a result of the GP Quality and Outcomes Framework (QOF) nurses have been used to ensure practice success in achieving activity targets. The resultant rewards, of course, accrue to GPs as their employers. This may not always facilitate the development of nursing services in primary care eg a recent RCT demonstrated that diabetes care could be transferred to practice nurses but such substitution remains limited (Houweling et al, Journal of Clinical Nursing, 2011)

Thus nurses in primary care, particularly the 30,000 who have full prescribing rights, have the potential to replace GPs in the delivery of much of primary care. This has been emphasised by clinical analysts (eg Fry (1977) in C I Philips and J N Wolfe 2007); and the current author for decades (eg Richardson, Maynard et al, Health Policy, 1998).

Why hasn’t there been more substitution of nurses for doctors in primary care? The incentive and contract structures and conservatism and relative power of the professions is such that change in roles and real substitution has been slight, but investment in nurses as complements has been substantial. Now, in an age of austerity, can a policy of nurse substitution be implemented? As in Canada, increasing frustration with access to GPs may drive change. My understanding of NHS primary care practice design is that at least one qualified medical practitioner has to be involved in service delivery by a practice. If this is correct, will it need the NHS and/or the private sector to establish a primary care practice with, for example, one GP and 10 nurses serving a population of 30,000 or more? If such change were to emerge, the National Commissioning Board and any CCGs involved should ensure the conduct of a well designed research study to identify clearly the costs and benefits of change.

Similar arguments apply to primary care dentistry where dental assistants may have the skills and expertise to carry out many tasks currently monopolised by dental practitioners. This is even more reminiscent of the Canadian situation of fee for service payment, as the ability to replace dental practitioners is again restricted by contracts and payment mechanisms in the UK, as in North America.

1.1.2 Skill mix in secondary care

The paucity of nursing skill mix evidence in the hospital sector is remarkable, and many research questions remain. For instance:

(a)What is the efficient staffing level for wards in a hospital? An American literature initiated by Aiken and further developed by Needleman has asserted a positive relationship between higher levels of qualified nurse staffing and improved outcomes for patients. This literature does not identify either where diminishing returns in effectiveness set in or the level of staffing that is cost effective. However, it has been influential eg it resulted in the Californian legislature mandated minimum nurse staffing level (recently abandoned due to their cost). This work used cross section data and as is often found when this method is compared with time series information, the results alter. The few time series analyses carried out show weaker relationships between nurse staffing volumes and patient outcomes (see Lankshear, Sheldon and Maynard, Advances in Nursing, 2005).

(b)Nurse staffing and staff and patient outcomes. A recent Cochrane review again shows the paucity of research (Butler, Collins, Halligan et al, 2011). This review identified 6,202 papers written about the topic but after applying rigorous standards of selection (eg RCTs and difference in difference modelling) only 15 studies met the usual Cochrane inclusion criteria. The authors concluded that the evidence was extremely limited(!) They were positive about the potential of some specialist nurse staffing but emphasised that their findings should be treated with “extreme caution” due to the limited evidence currently available. As with the primary care evidence base for nursing the failure of research funders and researchers to commission, design and carry out rigorous studies is indicative of poorly focused policy research.

(c)What is the scope for nurses replacing doctors in hospital care? Simplistic international comparisons show that for instance nurse anaesthetists work in Swedish, Dutch and American hospitals. In the UK, the Royal College of Anaesthetists has guardedly and reluctantly begun to ease restrictions on the scope of practice of nurse anaesthetists. A study of the use of nurse endoscopists found no significant differences in clinical outcomes between doctors and nurses but conclude that doctors were more cost effective (Richardson et al, BMJ, 2009, doi:10.1136/bmjb270). As ever in seeking to answer to the question of nurse-doctor substitution in hospitals the problem is paucity of good evaluative studies of effectiveness and cost effectiveness and the dominance of faith based opinion.

(d)In many areas of hospital care specialist nurse employment has increased. Again the issue is whether these are complements (extending capacity) and/or substitutes providing care and reducing the employment of eg consultants. If the latter where is the evidence of relative cost effectiveness. The reluctance to articulate and address these questions in policy analysis and research ensures reform is largely without an evidence base.

1.1.3 The challenge of “grade inflation”

In a period of austerity critical attention should be paid to advocates of “more” and “better” training. These siren calls posit vague benefits and ignore the opportunity costs of such investments.

Two comments are particularly pertinent. Firstly, before investing in more training for doctors, there is a need for the reform of medical school education whereby instead of each school examining in approximate relation to GMC guidelines, there should be national examinations so that, for example, training in management, finance and health economics for practitioners is mandatory. Note that there are national exams and increased management training in the US medical schools. Many UK medical schools still deny students education about the health care system in which they will work and it is increasingly essential for them to acquire knowledge of funding systems and rationing devices such as technology assessment.

Secondly, the decision to make nursing a graduate profession is costly and benefits are unclear. Many tasks on the ward (eg feeding, cleaning and other aspects of physical care of patients) do not require a degree and can be carried out by nursing assistants and Assistant Practitioners. However there is qualitative evidence that the skills of APs are underused (Spilsbury et al, JHSRP, 2011).

This reinforces the case for questioning the progression to a wholly graduate profession and to evaluate carefully the cost effectiveness of competing nurse skill levels in relation to patient needs. Modest numbers of graduate nurses are needed for tasks requiring complex skills but many (the majority) of patient needs in hospitals may be met cost-effectively by a non-graduate workforce. With RCNs controlling staffing, skill substitution may be inhibited.

1.2 Variations in consultant productivity

Planning the future doctor workforce continues to ignore evidence of considerable variations in consultant clinical activity, for example research by Bloor, Freemantle and Maynard (2004 and 2008). This demonstrates that the activity rates of consultants are highly variable (Bloor et al, JHSRP, 2004). This creates the potential for incentivising improvements which remove poor outliers and shift the mean of the distribution. If such incentives were introduced cost effectively they could increase productivity from the existing workforce. It is to be noted that data about theses variations were commissioned by DH from Bloor and circulated several times by the Department and the NHS Institute of Innovation and Improvement to all trusts in the last decade (see “Delivering quality and value: consultant clinical activity”, at
Its impact has been limited due, inter alia, to lack of incentives to change for managers and clinicians.

It should also be noted that Bloor’s research demonstrated that the activity rates of female consultants across 10 specialties is 10–20% less than their male counterparts (Bloor et al, JRSM,2008). With the “feminisation” of the medical workforce the implications of this work of investment in new consultant posts and/or nurse substitution are obvious but ignored by Whitehall and workforce planners.

The slow permeation of the DH funded research findings such as these into routine workforce planning is unfortunate.

2. Demand Side Issues

The focus on supply side issues (eg skill mix) must be complemented by more sophisticated modelling of the demand for health care. The starting point for demand side estimates of workforce forecasting is demographic projections, in particular the ageing of the population and the resources demands these processes are likely to make. This usually involves the application of fixed coefficients (eg a GP-population ratio).

This approach is a contentious area, where the conventional view is that ageing will add 1% or more to health care demand. This number is a “guesstimate” and has been contested for several decades by radicals such as the American physician James Fries (eg Fries et al, 2011, Journal of Ageing Research, 2011, 261702). He asserts that there is evidence that successive cohorts of Americans are increasingly healthy and likely to be less demanding over the life cycle as their death in old age will be swifter and less resource intensive. This he labels the “compression of morbidity”. While there has been good evidence supporting this hypothesis, it may be that “healthier” cohorts are now in the past, and other researchers emphasise the potential impact of increasing rates of dementia/Alzheimer’s and particularly the increase in obesity and diabetes, which may drive up the demand for health and social care.

The usual assumption in UK workforce forecasts (eg the Centre for Workforce Intelligence, 2010) is a constant relationship between population increases and patient “need”. An alternative has been proposed by Birch et al (2007), Canadian Public Policy, 207, 33 (supplement). This suggests a fourfold approach to demand estimation:

(i)Demography: using data on the size, age and distribution of the population including emigration and immigration over time.

(ii)Epidemiology: develop an assessment of the normatively assessed health care needs of the population.

(iii)Estimate the level of service to be provided.

(iv)Incorporate forecasts of productivity of health care providers, which are products of incentives/workforce contracts and skill mix.

Despite the initial involvement of Manchester University (where Birch works part time) in the Centre for Workforce Intelligence these suggestions do not appear to have been taken trialled in government forecasting.

3. Modelling the NHS Workforce

Despite recent assertions in Whitehall of “world class workforce policies”, practice has remained relatively static and poor. One example of the lack of progress in workforce planning is the failure to model all labour inputs rather than primarily focus on doctors. Such an approach would facilitate more explicit consideration of substitution possibilities. Instead forecasting tends to be insular, dominated by forecasting of medical practitioners and poor in dealing with between and within sector substitution opportunities.


(1)Despite sustained advocacy of improvements in workforce planning practice remains myopic, inadequate and dominated by the market for physicians.

(2)On the supply side, the reluctance to model, innovate and evaluate skill mix options is remarkable and sustained by limited awareness on the NHS of research findings and the failure of R&D to enhance knowledge significantly, particularly in the last decade when funding has been generous.

(3)Demand side modelling remain dominated by un-evidenced assumptions about fixed ratios and a reluctance to complement demographic and epidemiological data with forecasts of trends in productivity and level of service to be provide.

(4)Whitehall could and should do better in NHS workforce planning!

December 2011

Prepared 22nd May 2012