9.2. Medical Workforce Dynamics and Patient Care

On 1 August 2004, the European Working Time Directive (EUWTD), a measure intended to limit the working week to 48 hours for all workers within the European Union (EU), became mandatory for junior doctors working for the National Health Service (NHS) in Britain. The intuition behind this piece of health and safety legislation was to reduce the fatigue experienced by junior doctors by limiting doctor's working hours and so improve the quality of patient care. This rationale, though compelling, does not address the non-clinical effects of the EUWTD, in particular, the fundamental change that the directive has on doctors' working patterns, in-service training and work–life balance.

These concerns were of personal and professional interest to Dr Mark Ratnarajah, a paediatric specialist registrar based in London, who, at the time, was also enrolled on the Executive MBA programme at London Business School. He decided to conduct a project to consider the effects of the directive on junior doctors' career decisions and the consequences of these decisions on the medical workforce and quality of patient care. Based on his experience and knowledge of the UK National Health System (NHS), he developed a system dynamics model to consider the broad implications of the directive and to explore alternative courses of action. His work is summarised below.^[]

^[] I am grateful to Dr Ratnarajah for permission to use this material, which is based on the management report he wrote as part of the Executive MBA degree at London Business School. The report (Ratnarajah, 2004) received the Student Prize of the UK Chapter of the System Dynamics Society in 2005.

9.2.1. Background

Since the inception of the NHS over 50 years ago, junior doctors have formed the backbone of the service. Medical cover at night is provided by the junior ranking doctor, who remains resident in the hospital overnight with senior support from the consultant specialist in charge, (who usually consults from home and is available to be called into hospital if the need arises). Traditionally, a junior doctor would be expected to work a full day, followed by an over night period on-call and then work the following day (up to 32 hours resident in the hospital). Although this work schedule accounts for only one and a half days of doctor cover, it constitutes over half the 48 weekly working hours permissible under the EUWTD. The directive also stipulates that there must be a minimum of 11 hours rest between continuous periods of work and no doctor working a night duty will be allowed to work at any time during the previous or following daytime. As a consequence of the restrictions, junior doctors will be expected to adopt a full-shift pattern of work involving shorter, more frequent, serial periods of work, usually covering the day/evening and night separately. This transition presents a major challenge to NHS hospitals. Traditionally, junior doctors have worked a 72-hour week. When the 48-hour week and full-shift system is fully implemented, the number of hours of doctor cover lost will be between 208 296 and 476 638 hours, which is the equivalent of 4 300–9 900 junior doctors. These numbers are based on estimates published by the British Medical Association.

A system dynamics model was developed to examine the consequences of the working time directive on the UK medical workforce and junior hospital doctors in particular. The project was conducted in two stages. In stage one, a workforce planning model was built to explore how hospitals will cope with the expected loss of junior doctor cover and the transition to full-shift work patterns. This model focuses on the tangible effects of the directive on the total hours available from junior doctors. In stage two, the model was extended to include intangible effects of the working time directive on the work-life balance and morale of junior doctors and potential knock-on consequences to doctors quitting the medical profession. The models' structure was derived from the modeller's own decade of personal experience as a physician trained in the NHS. Parameters were gleaned from government health care policy documents and from journal articles about the medical profession. These publications are listed in the footnote below as a practical example of the use of formal information sources for model building.^[]

^[] Information sources for parameters in the medical workforce model:

Source 1: 'Hospital Doctors – The European Working Time Directive', May 2004, British Medical Association.

Source 2: 'Medical schools delivering doctors of the future', Department of Health Report, March 2004.

Source 3: Review Body on Doctors and Dentists Remuneration, 33^rd Report 2004.

Source 4: Trevor W Lambert, Jean M Davidson, Julie Evans & Michael J Goldacre. Doctors' reasons for rejecting initial choices of specialties as long-term careers. Medical Education,37(4), 312–318.

Source 5: Chesser, S., Bowman, K. & Phillips, H. (2002). The European Working Time Directive and the training of surgeons. British Medical Journal Careers Focus, 31/8, 69–70.

Source 6: 'The future healthcare workforce', Discussion paper 9. February 2002. BMA Health Policy and Economic Research Unit.

Source 7: 'An organisation with a memory', Report of an expert group on learning from adverse events in the NHS. 2000, Department of Health.

Source 8: Mann, C. & Guly, H. (1998). Is the emergency (999) service being misused? A retrospective analysis. British Medical Journal, 316, 437–438.

Source 9: Patients' Charter, Department of Health, revised 1996.

Source 10: White, C. (1998). The loss of doctors can be stemmed. British Medical Journal, 316, 1851; White, C. (2004). Why do doctors leave the profession? BMA Health and Economic Research Unit, April.

Source 11: Survey of SAS doctors and dentists report, Health policy and Economic Research unit, July 2004.

Source 12: P Burn, 'EUWTD, the European dimension'. Royal college of Physicians, Nov 2002.

9.2.2. Medical Workforce Planning Model

Hospitals are professional service organisations that rely on teams of highly trained doctors and nurses to deliver patient care. To achieve high standards of care it is important for hospitals to maintain the right balance of personnel. The medical workforce planning model represents the number of medical students, junior doctors and specialist doctors working in the National Health Service and examines how this balance will change over time under the provisions of the European Working Time Directive. As mentioned above, the directive will reduce medical cover by the equivalent of 4 300–9 900 junior doctors (see p. 316, footnote 2, source 1). The model examines how hospitals will adapt to this shortage through a combination of additional training and overseas recruitment.

Figure 9.6 shows the composition of the medical workforce in terms of specialist doctors, junior doctors, non-UK resident doctors and medical students. The dynamic hypothesis is that the Working Time Directive will alter the balance of medical professionals, which in turn will affect the service capacity of hospitals and patient care. The hypothesis was conceptualised directly in terms of stocks and flows of doctors (rather than causal loops) as this seemed a natural way to think about service capacity. In the top half of the figure is a commonly used stock and flow network known as an experience chain (or aging chain). Medical students are recruited into medical schools where they spend time in training to become junior doctors. They then undergo further specialist training to take lifetime appointments as specialist doctors (or consultants) before eventually retiring. Along the way, some junior doctors take non-hospital appointments as general practitioners and others leave the medical profession altogether. In equilibrium, the rate of medical student recruitment is exactly equal to the rate at which doctors of all types are leaving hospitals (the sum of the rates of non-hospital appointments, junior doctor attrition and specialist doctor retirement).

A characteristic of experience and aging chains is that they are slow to achieve an equilibrium or appropriate balance among the different categories represented in the chain. If one were starting a hospital from scratch, with only a handful of newly-graduated medical students, then it would take decades to build a cadre of specialist doctors and achieve the kind of consultant-led care typical of hospitals in developed nations. In the meantime, the hospital would run with a relatively inexperienced medical workforce. Similarly, if, through policy change, one were to alter the proportion of trained doctors needed at a point in the chain then it would take hospitals years to re-establish an appropriate balance of professionals. Such re-balancing is one important consequence of the EUWTD as it will require hospitals to employ a larger proportion of junior doctors in order to achieve the same medical cover as in the past. Incidentally, a similar slow process of category balance is observed in the urban dynamics model, which contains two linked aging chains: one for housing and another for business. City growth is characterised by a high proportion of new buildings and enterprises. The transition from growth to equilibrium results in a larger proportion of old houses and declining businesses, with a consequent decline in the quality of urban life.

Consider now the factors determining the flow rates of doctors in Figure 9.6 and the composition of medical staff. We will start with medical students and trace the influences throughout the experience chain all the way to specialist doctors. We will also consider the balancing effect of non-UK resident doctors shown in the lower half of Figure 9.6.

The number of medical students recruited each year from secondary schools or as mature students is determined by the Department of Health. The inflow is treated as an exogenous variable, beyond the control of individual hospitals and independent of the EUWTD. Historically, medical school places in the UK have increased by more than 50 per cent in the seven year period 1997–2004. This growth was achieved through expansion of existing medical departments and the building of entirely new medical schools. By 2004, the intake reached approximately 6000 medical students per year according to a report from the Department of Health (see p. 316, footnote 2, source 2). At the time of the study, no further capacity expansion was planned and therefore the model was parameterised with a static recruitment rate of 6 000 medical students per year, a number which applies across the model's 15-year time horizon.

It requires, on average, five years for a medical student to qualify as a doctor in the UK and take a junior post in a hospital. This delay is captured in the duration of medical school training that controls the outflow from the level of medical students. A typical depletion formulation is used in which the rate of medical school training is the ratio of medical students to the duration of training. The flow of trained students then accumulates in the pool of junior doctors. From this pool, a significant annual percentage will choose a career path in general practice or primary health care, away from hospital medicine. Others will leave the medical profession. The proportion lost to General Practice (shown in Figure 9.6 as the annual fractional loss to non-hospital appointments) is currently 2 per cent per year (see p. 316, footnote 2, source 3). Those who quit medicine are represented by the junior doctor attrition rate, which is estimated in the medical literature to be 1.2 per cent annually. The outflows are simple depletion formulations, proportional to the number of junior doctors, where the proportionality depends on the annual percentages quoted above (see p. 316, footnote 2, source 4). Note that we assume the attrition rate is unaffected by working environment changes, an assumption broadly consistent with the thinking behind the Working Time Directive. This assumption is relaxed in the stage two model, where the attrition rate depends also on junior doctors' morale.

Doctors' postgraduate training in recent years has undergone a process of formalisation. Prior to the introduction of the EUWTD, it had taken an average of nine years for a junior doctor to achieve consultant status. The Directive is set to increase this period for reasons that are now explained and are represented by the influences on the duration of specialist training in Figure 9.6. Although most postgraduate education takes place through autonomous learning at work, formal clinical in-service training (for instance in surgical specialities) and seminars can only occur under direct senior supervision during normal working hours. Hence, any reduction in the number of working hours or increase in the number of shifts worked outside of normal working hours will have the effect of extending the time required to train. The duration of specialist training is formulated as a weighted average of the traditional duration of training (nine years) and the duration of training under the EUWTD (11.45 years), where the weighting is defined by the percentage of workforce compliant (which gradually increases from an initial value of 57 per cent to 100 per cent over two years). The extra 2.45 years of training is a result of an estimated 79 per cent reduction in the number of daytime hours worked under a full shift system assuming a 48-hour working week (see p. 316, footnote 2, source 5).

The final influence on the composition of the medical workforce comes from non-UK resident doctors. The high quality of training and service offering within the National Health Service has consistently attracted overseas doctors wishing to undertake sabbatical periods of work and for some to train towards attaining permanent residency in the United Kingdom. More recently, with the relaxation of workforce transfer among European Union member states and an inherent delay in satisfying any required expansion of the health care workforce with domestically trained doctors, increasing numbers of overseas doctors are being recruited to make up the shortfall. Effectively, they are a source of spare medical 'capacity'. The rate of non-UK resident doctor recruitment is influenced by the workforce deficit and the time to recruit (an example of a standard stock adjustment formulation). The workforce deficit is the difference between the target workforce and current aggregate workforce consisting of UK trained junior doctors and non-UK doctors. The target workforce grows as more hospitals become compliant with the working time directive. The time to recruit is assumed to be six months (0.5 years) from the time of advertising the post to successful candidates passing the prerequisite admittance exams for starting work in a UK hospital. Once recruited, overseas doctors are assumed to take a temporary post of four years (limited by the length of a UK working visa) and do not undertake specialist training during their period of employment. Hence, the attrition rate is formulated as the ratio of non-UK resident doctors to the duration of a working visa.

Specialist doctors (consultants) are replenished by the rate of junior doctors training and diminished by senior consultants retiring. The rate at which senior consultants retire depends on the average tenure of consultants, which is assumed to be 32 years. Although concerns have been raised about consultants choosing to retire early, the number of them that have decided to do so is currently insignificant and so is not included in the model.^[]

^[] The role of consultants in this particular hospital model is rather passive. They simply accumulate at the end of the medical personnel chain and spend their time in the stock of specialist doctors until they retire, without any particular effect on the delivery of patient care. Clearly, in real life, consultants play a vital role in service delivery and junior doctor training that the model overlooks. This simplification might be justified on the grounds that the model's dynamic hypothesis focuses on the changing work-patterns and number of junior hospital doctors as a consequence of EUWTD and that the number of consultants and their activities are relatively stable under EUWTD and therefore not dynamically significant. However, this hidden assumption could (and perhaps should) be tested, particularly in the way it affects junior doctor training. The present formulation assumes that the duration of specialist training will increase because junior doctors' working hours decrease and so there are fewer opportunities in a typical working week for them to interact with consultants. However, it is common in models of service organisations to also represent the feedback effects from the number of senior/experienced personnel on the rate at which on-the-job training can be delivered. The lower the proportion of senior/experienced personnel in an organisation the more thinly their time is spread and the less on-the-job training takes place. An example, as it applies to hospitals, is presented in Winch and Derrick (2006) and readers are invited to modify the hospital model in the CD folder for Chapter 9 to include the supervision of junior doctors by consultants and the feedback effects on flow-through of doctors (and possibly patient care). Sample formulations of on-the-job training in a service organisation can be found in the staff sector of the airline model in the CD folder for Chapter 6 – but these formulations will need to be modified for a hospital.

9.2.3. Quality of Patient Care

There are many factors that may influence the quality of patient care. The two most pertinent to current medical practice, and presumed to be under the influence of the EUWTD, are clinical error rate and patient to doctor ratio or work schedule pressure. Both are shown in Figure 9.7.

Clinical errors often stem from confusion about a patient's condition. This causal effect is represented in the top half of Figure 9.7. Dissemination of patient information among clinical professionals is vital to the continuation of the highest level of medical care and assurance that an individual patient's clinical progress is followed. The medical handover is the keystone of this process and an important part of continuous professional learning. Doctors who have finished a shift provide detailed verbal and written communication to the next shift of doctors about their patients' clinical status and progress. Under the EUWTD, the number of shifts per week will need to increase, as will the number of handovers between shifts. As the pressure of time keeping increases, in order to stay compliant with the maximum duration of a shift, handovers are likely to become hurried and less detailed in content (see p. 316, footnote 2, source 6). Therefore, we hypothesise that the change in the handover process under the EUWTD will cause the actual error rate (benchmarked against the expected error rate prior to the introduction of the EUWTD) to increase in proportion to the relative increase in the number of handovers (see p. 316, footnote 2, source 7).

The other determinant of the quality of patient care is the patient to doctor ratio shown in the bottom half of Figure 9.7. The patient:doctor ratio is used to calculate a work schedule pressure, which acts as a crude measure of organisational stress. The work schedule pressure represents the amount of work expected by each doctor and is calculated by the ratio of the number of acutely admitted patients per year (growing by 8.9 per cent annually; see p. 316, footnote 2, source 8) to the sum of junior and non-UK trainee doctors per year, and multiplied by the expected maximum patient waiting time, which, according to the Department of Health's Patients' Charter, is three hours (see p. 316, footnote 2, source 9).

The effect of the error rate and patient load on the quality of patient care is calculated by deriving equally weighted, normalised ratios for current to expected work schedule pressure and the relative error rate.

9.2.4. Base Run – Changing Composition of the Medical Workforce

Figure 9.8 is the base run of the medical workforce planning model and shows the changing composition of the medical workforce during and after the implementation of the European Working Time Directive. The simulation runs for 15 years, which represents the average time for a doctor to be trained from medical student to clinical specialist. The gradual ramping up of medical students (line 2) reflects the throughput effect of planned capacity expansion of medical schools by the Department of Health over the previous seven years. By year 7.5, the total number of medical students levels off at 30000 (a maximum of 6000 students per year in each of the five years at medical school).

The effect of the increase in medical student recruitment is reflected in the gradual rise in numbers of junior doctors (line 1), as the inflow of newly graduated junior doctors is greater than the combined outflow of junior doctors leaving the profession, becoming general practioners or specialists. In addition, the prolongation of junior doctor's postgraduate training under the EUWTD contributes to the rise in the number of junior doctors.

Specialist doctors remain in post longer than other medical personnel in training (medical students and junior doctors), on average 32 years before retirement. The number of specialists swells over time as the rate of junior doctors completing their training is greater than the specialist retirement rate.

During the simulation the perceived workforce deficit is smoothed over the five years running up to the EUWTD amendment, due to take effect in 2009, in order to emulate the prospective behaviour of workforce planners in preparing for the expected shortfall in the medical workforce. As a consequence, the number of non-UK resident doctors (Figure 9.8, line 4) increases over the first five years and then gradually declines as UK trained junior doctors begin to make up an increasing proportion of the workforce. Therefore, the non-UK resident doctors act as a sort of 'swing' workforce provider, because of their short time to recruitment and limited duration of service. Note that the EUWTD amendment referred to above reduces working hours from an interim target of 58 hours per week to the final target of 48 hours per week.

In summary, the simulation shows a gradual adjustment in the composition of the UK medical workforce in response to changes in junior doctors' work patterns. There is a temporary rise in overseas doctors that gradually subsides as more UK-trained junior doctors become available through the training pipeline of medical schools. The number of specialist doctors rises steadily over the entire 15 year horizon, reflecting both the increased supply of junior doctors and an initial shortage of specialists in the National Health Service.

9.2.5. Base Run – Quality of Patient Care

The two determinants of the quality of patient care affected by the introduction of the EUWTD are the relative clinical error rate and relative work schedule pressure. In Figure 9.9, the quality of patient care (line 3) steadily declines to approximately half its initial value over the 15 years of the simulation under the influence of the EUWTD. Much of the early decline in quality of patient care is due to a rise in the relative error rate (line 2). The trajectory matches the shape of EUWTD compliance, reflecting a gradual transfer of junior doctors to full-shift working and more frequent handovers that cause errors.

The continuing decline in quality of patient care is due to growth in relative work schedule pressure (line 1). The total number of non-specialist doctors does not keep pace with the assumed 8.9 per cent annual growth in patient admissions. An imbalance arises because recruitment of non-UK resident doctors is aimed at maintaining 24-hour doctor cover in hospital, rather than maintaining a fixed and effective patient:doctor ratio. This lack of coordination between medical staff and patients may seem surprising when viewed from the reader's detached perspective outside the National Health Service. Ambiguity in the patient:doctor ratio, however, is entirely plausible at the sharp-end of hospital practice due to the inertia of the experience chain and the added complexity of staff administration caused by the Working Time Directive itself. Bear in mind that the total number of hospital doctors grows substantially during the simulation, from almost 75 000 to 104 000 – a rise of nearly 40 per cent. Against this backdrop of staff growth, a decline in quality of patient care will, in practice, appear anomalous.

9.2.6. Intangible Effects of the European Working Time Directive

The base case model assumes that the effects of the working time directive on hospitals are fully captured in the time budget of junior doctors. If each junior doctor works fewer hours a week then there will need to be more of them to provide the same level of cover. This balancing act may be dynamically complex but essentially if the time budget is restored then quality of patient care will not be diminished, except for the increase in clinical errors due to more frequent medical handovers. What if new work patterns also have indirect effects on the morale of junior doctors due to changes in their work-life balance, duration of training and other working conditions? Dr Ratnarajah was in no doubt that these intangible effects existed and was curious to investigate their potential impact. The quest to represent and test intangibles led to a second stage of modelling that retained all the structure and relationships of the base case model and added the new concept of doctors' morale and the various influences on it.

9.2.7. Modelling Junior Doctor Morale

In the extended model, morale is an integral part of junior doctors' decision-making process to either stay or leave medical training, and is influenced by changes in junior doctors' working conditions (in the base case junior doctors were considered resilient to any working environment change). The level of morale has a direct impact on the rate of junior doctor attrition: the lower the morale, the higher the rate of junior doctor attrition. The equation formulation is simply a variant of standard stock depletion in which the attrition rate depends on the product of junior doctors and the annual attrition fraction, multiplied by the inverse of morale.

Here morale is conceived as a dimensionless index that runs on a scale from zero to one (though strictly speaking the minimum value is slightly greater than zero). Morale is initialised at a value of one which is taken to be the normal state of morale prevailing in the pre-EUWTD era. A change in morale is represented as an asset stock adjustment process brought about through a difference in indicated morale and the previous level of morale, as shown in Figure 9.10. Indicated morale is an equally weighted multiple of the key factors, as reported within the profession itself, that contribute to junior doctors morale. The factors include the quality of patient care, relative duration of junior doctor training, relative work life balance, flexibility of the rota and aggregate pay discrepancy (see p. 316, footnote 2, source 10).

Each of these morale factors is defined as a dimensionless index with a normal value of one. So, for example, the relative duration of training is equal to the ratio of the duration of specialist training (under EUWTD) to the traditional duration of specialist training (pre-EUWTD). The factors are combined in a multiplicative formulation as shown in the equations at the bottom of Figure 9.10. The reason for multiplying the determinants of indicated morale, rather than applying an additive formulation, is that each determinant has a spill-over effect on the other determinants of morale and, therefore, should not be considered as independent effects. The change in morale is represented as a bi-flow because morale can go up as well as down, depending on the prevailing conditions. The timescale for junior doctors' morale to change is estimated to be approximately one year.

Such multiplicative formulations, involving dimensionless indices normalised at one, are common whenever there is a need to combine the effects on a flow rate of two or more co-dependent variables. A similar formulation is used in the urban dynamics model to combine the five components of city attractiveness that control the arrival of underemployed population into an urban area. Multiplication of normalised indices can result in a surprisingly low resultant value if several determinants in the expression fall below their normal value of one. For example, if each determinant of morale takes a value of 0.8 (20 per cent below normal) then indicated morale, which in this case would be equal to the fifth power of 0.8, falls to only 0.328 (two-thirds or almost 70 per cent below normal).

9.2.8. Overview of the Complete Model

The complete model represents the workplace pressures behind junior doctor morale and adds several new feedback loops in the representation of medical workforce dynamics. The resulting sector map is shown in Figure 9.11. On the left is medical workforce planning and on the right is quality of patient care management – the two sectors already described. In addition, there are new sectors for work–life balance, doctors' remuneration and factors influencing morale. The light shaded region shows how the boundary of the model has been expanded. The new module neatly overlays the base case model and allows us to examine the incremental impact of morale and junior doctor attrition on the implementation of the working time directive.

9.2.9. The Formulation of Work–life Balance and Flexibility

Recent results from a longitudinal cohort study of doctors who graduated from UK Medical schools, carried out by the British Medical Association, suggest that the current generation of junior doctors view their commitment to the NHS very differently to that of their predecessors. While prepared to fulfil a reasonable contract, the current crop of young doctors also demands time for self-fulfilment and family life. For those doctors who choose to leave the NHS (15–20% of domestically trained doctors will not be working for the NHS a few years after graduation), the majority cite issues with work–life balance, risk of litigation, flexibility, remuneration, quality of training and workload as their reasons for resigning. Therefore, despite the fact that the implementation of the EUWTD has reduced absolute numbers of hours worked per week, there may be unforeseen consequences that are unacceptable to the current generation of junior doctors and that may yet manifest as changes in the junior doctors' attrition rate and quality of health service provision.

Work–life balance in the model is a measure of the number of social hours available after each shift. Under EUWTD, shift work eats into social activity. The main assumption is that post-shift work recovery time is often incompatible with social activity. These ideas are represented in the concepts linking average shifts per week to relative work–life balance in Figure 9.12. A typical pre-EUWTD 1:6 rota (one shift worked every six days) allows 8 social hours per shift whereas an EU compliant rota of three or more shifts per week allows 6 or less social hours per shift. Relative work–life balance is formulated as the ratio of social hours available under EUWTD to social hours available pre-EUWTD.

Flexibility of shift work is also of concern to junior doctors. More flexibility is better. The pre-EUWTD rota allowed for 24-hour shifts to be swapped among co-workers on the same rota with relative ease, because of the discrete nature of the working pattern and the relatively long working week. By introducing a full-shift system, under the Working Time Directive, the process of swapping shifts becomes more complex. There are additional restrictions from adhering to the minimum rest requirement between shifts and through having to work a pattern of serial night/day shifts (for many doctors the introduction of a EUWTD compliant rota will result in having designated periods of annual leave). In the model, the measure of flexibility (shown as flexibility of rota in the lower-right of Figure 9.12) is formulated as the ratio of the average number of shifts worked per week under EUWTD to the number of shifts worked pre-EUWTD.

9.2.10. Simulations of the Complete Model

As in the base case, simulations were run over a time horizon of 15 years. It should be remembered that the simulator is not intended to be an accurate forecast of the future, but merely to provide a recognisable and compelling story for the possible outcomes of various policy decisions and assumptions within an internally consistent framework. This caveat is particularly relevant to the interpretation of junior doctor morale with which we begin.

Figure 9.13 shows an exponential decline in morale from a value of one to only 0.1 over the first five years of the simulation. Contributing to this precipitous fall are increases in both the relative duration of training and relative error rate, together with declines in the quality of patient care, work–life balance and rota flexibility. Junior doctors come under increasing pressure as the five determinants of indicated morale take their toll. Intuitively the exponential decline in morale fits with the notion that its determinants have a combined effect much stronger than each individual effect in isolation. Of course, we are only seeing the implications of our assumptions, which may be overstated or incomplete. Nevertheless, the simulation gives pause for thought and challenges us to explain why apparently reasonable descriptions of junior doctor working conditions, supported by evidence from the medical profession's own journals, produce such a detrimental effect. The extreme outcome is not implausible. When one considers that historically individual doctors have had little direct influence in effecting change in their working conditions, one can surmise that an inability to respond to onerous working conditions would contribute to a pattern of sustained decline in morale.

Anecdotally, a recent medical workforce questionnaire has shown a sustained decrease in the levels of job satisfaction after graduation. Over half of the 2 500 doctors surveyed, in a recent British Medical Association study reported that morale had declined over the past five years (see p. 316, footnote 2, source 11).

If we accept the simulated behaviour of junior doctor morale then what are the knock-on consequences for hospitals? Simulations of the complete model allow us to explore this scenario. First, though, consider what you think would happen to the number and composition of hospital doctors and the quality of patient care if morale were to plummet. The base runs in Figures 9.8 and 9.9 are useful benchmarks for this preliminary thought experiment. Relative to the base case we now make just one additional assumption – that low morale accelerates junior doctor attrition. The ramifications are far reaching.

Figure 9.14 shows the number of junior doctors (line 1) is significantly eroded by the increasing attrition rate from low morale. At the start of the simulation there are about 4000 junior doctors and by the end there are only 2000 – a 50 per cent reduction over 15 years. Some reduction is to be expected under the circumstances. What is more surprising is the resilience of hospitals in the face of this mass exodus of talent – at least in the short run. The number of non-UK resident doctors (line 4) increases over the first five years to make up for the shortfall in the numbers of junior doctors and the target workforce. By year 4 of the simulation, the number of non-UK resident doctors exceeds the number of UK trained junior doctors – despite the fact that the number of UK medical students (line 2) follows exactly the same trajectory as in the base run.

The shifting composition of junior doctors has serious long-run consequences for the supply of specialist doctors. Initially, the stock of specialist doctors (line 3) rises because the inflow of trained junior doctors is greater than the outflow of specialist doctors retiring. However, as the number of junior doctors undertaking specialist training declines, due to the increasing attrition rate, the rate at which specialist doctors retire eventually exceeds the rate of junior doctor training and so by year 8 the population of specialist doctors also begins to decline. The rate of junior doctor attrition is gradually stemmed by year 5 through a more gradual decline in morale. The change is due to the slight increases in flexibility and work–life balance brought about by the reduction in the number of hours and reduced frequency of shifts worked per week under the amended EUWTD.

The quality of patient care is shown in the top half of Figure 9.15. For comparison, the equivalent chart from the base run is shown in the bottom half of the figure. In both charts, quality of care (line 1) declines throughout the simulation due to a rise in both the clinical error rate and work schedule pressure. Again the total number of non-specialist doctors does not keep pace with the assumed 8.9 per cent annual growth in patient admissions, as shown by the rising patient:doctor ratio (line 2), which ends up slightly higher in the complete model than in the base run.

9.2.11. Conclusions from the Medical Workforce Study

The purpose of the modelling project was to examine the effects of the European Working Time Directive on junior doctors' career decisions and the consequences of these decisions on the medical workforce and quality of patient care. The use of system dynamics provided a framework with which to consider these possibilities and a useful tool to explore alternative courses of action.

The results of the base run demonstrate that the target medical workforce (required to fulfil 24 hours continuous hospital cover, compliant with the EUWTD) can be achieved under current UK recruitment practices and that the non-UK resident doctors serve to temporarily fill any workforce deficit. This outcome implicitly assumes that UK trained junior doctors are resilient to changes in their work environment in the sense that the fraction who eventually quit the medical profession remains constant. Hence, any increase in medical student intake or in the target workforce numbers serves to increase the long-term numbers of UK trained junior doctors.

Given that the planned capacity expansion of UK medical school intake has already been achieved, the patient:doctor ratio will continue to rise as the numbers of acute patient admissions grow year on year while the size of the medical workforce remains static. The consequences are likely to be higher demands placed on the existing medical workforce and a continual erosion in the quality of patient care as the service pressure mounts on already over-stretched hospital staff.

This issue of work schedule pressure can only really be addressed by further workforce expansion in line with increases in patient admissions or improvements in the efficiency of in-patient management. For instance, focusing resources on primary care, to manage non-urgent cases in the community, would reduce the burden on accident and emergency departments. In addition, re-assigning routine clinical and administrative tasks to suitably trained health care professionals would free junior doctors to treat more patients and undertake more appropriate tasks relative to their level of training.

When the level of junior doctors' morale is factored into the complete model, the simulation demonstrates a precipitous decline in the numbers of UK trained junior doctors brought about by an increasing attrition rate. The fall in morale is caused by the decline in the quality of patient care, work–life balance, rota flexibility and an increase in the duration of specialist training. Interestingly, the perception of remuneration in the form of the aggregate pay discrepancy has little effect on junior doctors' attrition rate, as relative pay remains almost constant over time. This outcome reinforces the notion that pay plays a less important role in junior doctors' career decision-making processes than other work and lifestyle-related factors.

As a consequence of low morale, there is an increase in junior doctor attrition and a progressive loss of junior doctors in exchange for non-UK resident doctors, who not only make up the shortfall in the target workforce numbers but also constitute an increasing majority of the overall junior doctor workforce. The fall in the number of UK trained junior doctors is contrary to the base run and to policymakers' optimistic expectations about the effects of the EUWTD.

There remains a question mark over what the service impact will be of having less UK trained doctors? In the model, no value judgement has been made about the relative competence of non-UK trained versus UK trained junior doctors. Nevertheless, unless more non-UK trained doctors are recruited as specialist trainees (rather than only for service provision) or the loss of UK trained junior doctors is stemmed, there will be a long-term reduction in consultant specialists and the ability of the NHS to deliver a consultant-led service.

Sensitivity analyses on the complete model suggest an inevitable decline in junior doctor numbers irrespective of policy changes in the level of recruitment of medical student or delays in implementation of the EUWTD amendment. This result suggests that quantitative policy changes alone are insufficient to ensure a sustainable workforce. Without consideration of the qualitative effects of the EUWTD on junior doctors' morale, it would appear that the introduction of the EUWTD is unlikely to result in a successful outcome.

Ultimately, time will tell as to what the effects of the EUWTD will be on junior doctor attrition and the medical workforce. However, there are a number of areas highlighted by the simulator that are worth pursuing further. First, more accurate data about the relative weight or importance of the determinants of morale would build more confidence in the model's intangible sectors and provide a cross-check to attitudes held by junior doctors in the NHS. Second, a postponement in the implementation of the EUWTD amendment due in 2009 would be a prudent and viable option for the UK government to consider. However, as sensitivity tests of the complete model demonstrate, any delay in the implementation of the EUWTD amendment on its own is unlikely to improve the long-term outcome. In addition, there needs to be active collaboration between policymakers and the medical workforce to address important lifestyle and training issues affected by the EUWTD. Third, other European member states that have already been operating under the EUWTD would be a valuable source of information and experience. The implementation of EUWTD policy in the UK may benefit from sharing best practices and areas of concern with other EU countries. In particular, Denmark and the Netherlands have fully implemented the EUWTD and, for the most part, within their current workforce capacity (see p. 316, footnote 2, source 12). Despite the demographics and population distribution of these countries being different to the UK, it may be useful to identify the critical success factors in managing the medical workforce to effectively implement policy.

Finally, in an attempt to minimise the impact of the EUWTD on extended specialist training, alternatives need to be actively pursued to allow for the most effective training while working under a full-shift system. One possible solution, currently under consideration by the Royal Colleges, is the introduction of hospital resident consultants, who could streamline in-patient management and be available for teaching throughout the shift. In addition, the patient:doctor ratio for acute admissions would be reduced if consultants are considered to be directly involved in managing acute patients rather than consulting from home.