Goldacre, Michael J. FFPH; Lambert, Trevor W. MSc
Medical training takes many years and is expensive. The implicit expectation is that most doctors will use their training by going into medical practice; however, at least in the United Kingdom (England, Scotland, Wales, and Northern Ireland), no legal or financial mechanism compels them to do so. Tuition fees for medical students in the United Kingdom are £9,000 per annum, and graduates repay these via percentage deductions from their salary after graduation whether they are working in medicine or not. This cost to the student, however, is modest compared with the total cost, including cost to the public purse, which was estimated in 2011 to be £261,000 per student over the five- or six-year course of his or her education.1 Knowledge both of the number of doctors who qualify each year and of the percentage who will subsequently practice is fundamental for assessing the outcomes of medical education and for future medical workforce planning.
However, accurate information on attrition has been hard to find. The authors of a 2004 U.S. study2 concluded that both the American Medical Association’s Masterfile data and the data from studies of physicians’ self-reported intentions to leave had shortcomings as measures of actual physician behavior. Two studies (published in 20033 and 20104) of the leaving intentions of general practitioners in England showed a rise in intentions to leave related to declining job satisfaction, but did not quantify actual attrition. A 2010 U.S. study of attrition from emergency medicine clinical practice5 was unusual in providing attrition estimates. It reported that, overall, 87% of trained and certified emergency medicine physicians remained in practice. It also provided attrition rates by number of years post graduation: a range of 5% to 9% at 2 to 15 years post graduation, 18% at 20 years post graduation, and 25% at 30 years post graduation.5
In the United Kingdom, as in many other places, there is a ceiling on the number of students admitted to medical school. The UK Health Departments decide the number of students to admit each year on the basis of expert estimations of both the future number of doctors needed to practice and the number of medical graduates who will remain in medical practice in the medium and long term. In addition to financial and other costs, ethical considerations are important for estimating and planning the number of doctors who will practice medicine. For example, patients allow medical students to study them, their bodies, and their illnesses on the basis of the understanding that they are contributing to the training of the next generation of doctors. Ethical issues with respect to patients’ interactions with medical trainees may arise if a large percentage of physicians-in-training do not, in fact, eventually practice medicine.
We authors are staff in the Medical Careers Research Group (MCRG). The MCRG has surveyed graduates from medical schools in the United Kingdom for over 30 years, in cohorts defined by year of graduation. This surveying has enabled us to estimate with a high degree of precision the level of participation in medicine in the National Health Service (NHS) by the graduates from each cohort at various stages in their postgraduate careers.6–8 The NHS is the publicly funded health care system of the United Kingdom. It was established in 1948, is funded through general taxation, and provides services that are generally free to residents of the United Kingdom at the point of use. The great majority of UK doctors work primarily in the NHS, though they are also able to undertake “private” work outside the NHS that is funded either directly by patients or by the patients’ private insurance companies. In this report, we extend the MCRG work on estimating participation in the NHS6–8 to provide findings on the number and percentage of graduates from UK medical schools who are in medical practice not only within the NHS but also elsewhere.
We calculated these numbers and percentages by combining our estimates of NHS participation with (1) estimates of the number of UK-trained physicians practicing in the United Kingdom, but not in the NHS, and (2) estimates of the number of UK-trained physicians practicing medicine outside of the United Kingdom.
The MCRG has tracked the careers of graduates from UK medical schools by conducting postal surveys of the career intentions and progression of graduates from all medical schools in the United Kingdom in selected year-of-graduation cohorts. The MCRG surveys the doctors toward the end of their first and third years after graduation and at longer time intervals thereafter. The studies commenced in 1975 and continue still. The MCRG has studied each cohort contemporaneously; for example, the first survey of the 1974 graduates took place in 1975, and the first survey of the 2002 graduates took place in 2003. This ensures that the results are not subject to recall bias. The intention has been to initiate study of new cohorts at three-year intervals; however, the availability of funding and the wishes of the funding body govern the ability to do so.
In this report we draw on data covering the careers, including the latest known or most recent career destination, of UK graduates from the years 1974, 1977, 1983, 1988, 1993, 1996, 1999, 2000, and 2002. Funding was not available in the late 1980s and early 1990s, which means that only one new inception cohort was initiated between 1983 and 1993 (i.e., the one in 1988). Lack of funding at this time also affected the means to maintain exactly the preferred schedule of follow-up surveys. The MCRG surveyed both the cohorts of 1999 and 2000 because the group wished to investigate year-on-year changes at a time of particular changes to postgraduate medical education in the United Kingdom.
The timing of the MCRG surveys relates to training milestones in the United Kingdom as follows (although the details have changed over the decades covered by the surveys). Graduation from medical school is followed by a two-year foundation program entailing several specialty attachments (or rotations), after which doctors enter a specialty training program (residency) which typically lasts, in the case of hospital specialties, six years, or, in the case of general practice (family medicine), three years. Year 2 results, therefore, cover doctors who are completing their foundation rotations and are about to start their specialty training; year 5 results cover doctors in specialty training (or coming to the end of it, in the case of family practitioners); and finally, results from year 10 onward cover doctors who have, in the majority of cases, completed their training and are in “career” posts. These are posts—notably, “hospital consultants” and “principals in general practice,” to use UK terminology—in which the doctor is in independent professional practice rather than working under the supervision of a more senior doctor.
Each MCRG survey covers the participants’ career choices and intentions, training, and actual career posts—as well as their views and attitudes on issues of career relevance, such as job satisfaction, the quality of the training they have received, and their views about future career opportunities. Graduates also have the opportunity to comment in an open-prose format on any aspect of their career that they wish.
The doctors who received the first survey of each cohort make up the whole cohort as it was at the time of graduation. Later surveys, whenever conducted, also start from the original roll of all graduates. For example, the survey of the graduates of 1993, undertaken in 2010, attempted to contact all doctors who graduated in 1993 (using updated addresses), excluding only those doctors who had previously declined to participate, who were untraceable after exhaustive searching, or who were known to have died. To practice medicine in the United Kingdom, a doctor must register with the national General Medical Council (GMC). For the initial survey, the MCRG obtains addresses from the doctors’ registration with the GMC. For subsequent surveys, the addresses are those from the then-current GMC Medical Register, the Medical Directory (a commercially produced directory of UK doctors in which entry is optional), and the addresses supplied by the doctors themselves in response to successive surveys. Typically, the MCRG sends up to four reminders, each including a copy of the questionnaire, to doctors who have not yet responded to the survey.
Governance and ethical considerations
Before 2005 in the United Kingdom, full GMC address details for all registered doctors were publicly available, and the GMC provided data to us on that basis. Since then, addresses have been confidential and the GMC provides them to research groups on an exceptional basis only. Further, doctors now maintain the right to have their details withheld from researchers (thus far, few doctors have asked for their contact details to be withheld). The GMC has deemed that the MCRG’s research objectives, and the availability of results from the MCRG’s surveys, support its functions of regulation and ensuring standards of medical practice. The GMC takes an interest, for example, in responses to the statement “My experience at medical school prepared me well for the jobs I have undertaken so far.” The MCRG’s Information Sharing Agreement includes the option for the GMC to suggest questions for the group to include in the questionnaires. For example, at the request of the GMC, the MCRG recently added a question about whether doctors who achieve consultant status (i.e., the level at which they practice medicine without requiring supervision) felt that their training had been “long enough, and good enough, to enable [them] to practice adequately.” The respondents are told that the GMC is a source of contact information. The individual responses from doctors (as distinct from aggregated statistics) are never released by the MCRG to anybody, including the GMC or the Department of Health (DH; the funding body); doctors are informed of this.
In addition to the agreement with the GMC, this research has full national NHS research ethical committee approval (REC reference 04/Q1907/48). The MCRG ensures confidentiality by restricting access to respondents’ identities to the survey administration staff who send the mailings and receive the replies. The researchers in the team have access only to anonymized response data.
Each MCRG survey asks doctors to provide details of all posts they have had since the previous survey, including start and end dates, location(s), specialty, grade (i.e., level of seniority, similar to intern, resident, attending physician), and working time (whether on a full-time or less-than-full-time contract). Using these data, we have built a continuous job history for each respondent, and we have used this history, in turn, to determine the main appointment of each doctor on September 30 of each year after his or her graduation (September 30 is the annual workforce census date in the NHS).
We classified each doctor as either a “home” student (i.e., living in the United Kingdom) or an “overseas” student (i.e., not living in the United Kingdom), at the time of his or her entry into medical school, according to the location the doctor reported.
Doctors with academic posts who are involved in teaching and/or research, doctors with clinical commitments, and doctors in NHS clinical administrative posts have honorary NHS contracts (in addition to their academic or management contracts). All our calculations include all doctors with NHS contracts, whatever the detail of their posts and whether they are full-time or part-time in the NHS.
Data from national workforce records on doctors working in the NHS
Simply knowing that a doctor is on the GMC register (as we do) is no indication of whether the doctor is working in the NHS as a doctor. Many registered doctors are not in medical practice in the NHS, and many are not in medical practice at all. The latter include, in particular, doctors who have retired and doctors who trained outside the United Kingdom and who have been unable to obtain a medical post within it. We therefore augment our survey data with employment records. The English DH in earlier years, and the NHS Information Centre (IC) more recently, have constructed a census of all doctors working in the NHS on September 30 of each year. Using the doctor’s GMC registration number, the DH and IC can determine whether each doctor in the MCRG cohorts was in the NHS on the census date in any given year. The DH and IC have regularly provided this annual census information to the MCRG since 1989.
Capture–recapture calculations of NHS participation
The MCRG has used capture–recapture methods,9,10 which in our case entails combining our survey data and NHS employment data, to provide a highly accurate estimate of the number and percentage of doctors working in the NHS at a specified number of years after graduation.
To estimate total participation in medicine for each cohort, we added two other groups of doctors (i.e., in addition to the physicians working in the NHS) to the capture–recapture estimates (see below). These groups were (1) the physicians who responded to our surveys who were working in medicine in the United Kingdom, but outside the NHS, and (2) the doctors who responded to our surveys who were working in medicine outside the United Kingdom. We also included for each cohort our estimates of the minimum and maximum percentages (see below) of doctors likely to be working in medicine anywhere in the world.
For the capture–recapture analysis, we classified NHS employment status at the NHS census point for each graduate of each cohort as one of the following:
(a) known to both MCRG and DH/IC to be working in the NHS,
(b) known to MCRG, but not to DH/IC, to be working in the NHS (these physicians are most likely working through an intermediary employment agency, which markets services to the NHS; thus, their names do not appear in the DH/IC records), or
(c) known to DH/IC, but not to MCRG, to be working in the NHS (these are physicians who have not responded to our survey).
We used these three categories to calculate the size of a fourth “unobserved” group (d), those also working in the NHS, but not known as such to either the MCRG or the DH/IC, using the formula d = bc / (a + 1). We were then able to calculate an estimated total number (e) in the NHS using the formula a + b + c + d.6–8
Next, we were able to compare the total number of physicians in the NHS (e) with the total of all doctors in each cohort (f), excluding the few known to be deceased or who declined to participate, in order to ascertain the percentage of the cohort estimated to be in the NHS at each census time ([e/f] × 100).
Minimum and maximum estimate calculations
To calculate the likely minimum percentage and the likely maximum percentage of UK medical school graduates working in medicine anywhere in the world, we proceeded as described in the following example of the 2002 graduates, two years after graduation. Through capture–recapture analysis, we estimated that in 2004, of the 4,224 UK medical school graduates from the class of 2002, 3,590 (85.0%) were working in the NHS of the United Kingdom, and the remaining 634 were not working in the NHS. Using data from the MCRG surveys, we were able to account for 231 of the 634: We knew, because they told us in their replies, that 38 graduates were working in medicine in the United Kingdom outside the NHS and that 150 graduates were in medicine outside the United Kingdom (therefore, 188 graduates were working in medicine outside the NHS). Forty-three responders were not working in medicine at all.
We determined our estimate of the minimum number/percentage of those in medicine by assuming that the 188 responders working in medicine outside the United Kingdom were the only ones of the 634 to be working in medicine. Hence, we estimated that the minimum number in medicine was 3,590 + 188 = 3,778, or 89.4% of the cohort (of 4,224 graduates).
Subtracting the 3,590 graduates working for the NHS and the other 231 for whom we could account from the 4,224 total graduates, there are 403 doctors unaccounted for. To determine our estimate of the maximum number/percentage in medicine, we assumed that, just as 81.4% (188 of 231) of physician responders not in the NHS were working in medicine, so 81.4% (the same proportion) of the unaccounted-for doctors were also working in medicine. Hence, the estimated maximum number of unaccounted-for doctors who were, in fact, in medicine but outside the NHS, would be 328 (81.4% × 403). Thus, the maximum number of doctors in medicine would be the 3,590 in the NHS (as determined by the capture–recapture analysis), plus the 188 non-NHS responders, plus the 328 extrapolated from responder data to the unaccounted-for doctors. This total is 4,106, or 97.2%, of 4,224 doctors.
Thus, the range run is 89.4% to 97.2%, and the likely true value lies somewhere in between.
Inclusion and exclusion criteria
We excluded graduates from Northern Ireland, as well as physicians who were working in medicine in Northern Ireland at the time of the last MCRG survey, because DH/IC employment data were not available for that part of the United Kingdom. We also excluded those whose medical school we did not know and (as mentioned) those we knew to be deceased, and those who declined to participate during the most recent survey. Finally, we excluded those whom we knew were working in medicine, but at a location unknown to us.
Years of graduation included
Our results include data for the most recent survey year for which both our survey data and the DH/IC employment data were available. DH/IC data were available only for the years after 1987; hence, we did not include the early years of the older cohorts in our analysis.
Year 2 results cover the graduates of 1988, 1993, 1996, 1999, 2000, and 2002. Year 5 results cover the graduates from 1983, 1988, 1993, 1996, 1999, and 2000. Year 10 results cover graduates from 1977, 1983, 1988, 1993, and 1996. Year 15 results cover graduates from 1974, 1977, 1983, and 1988. Years 20 and 25 results cover graduates from 1974 and 1977.
We conducted our analyses using IBM SPSS Statistics Software version 20 (IBM UK, Portsmouth) and Microsoft Excel 2010 (Microsoft UK, Reading and London).
Tables 1–3 show the estimated maximum percentage and estimated minimum percentage of doctors in the combined cohorts who were participating in medicine 2, 5, 10, 15, 20, and 25 years after graduation. The numbers on which we based these percentages appear in Appendixes 1–3. Table 1 and Appendix 1 show the results for all doctors, and separately for women and for men. Table 2 and Appendix 2 convey the results for home doctors (i.e., those who came to their UK-based medical school from family homes in the United Kingdom), and Table 3 and Appendix 3 provide the results for overseas doctors (i.e., those whose family home immediately prior to medical school was outside the United Kingdom).
All UK graduates
Overall, we estimate (Table 1) that, 2 years after graduation, between 91.3% and 96.1% of UK graduates across all of the cohorts we studied would still be working in medicine, that after 5 years between 90.1% and 94.8% would be doing so, and that after 10 years between 90.6% and 95.5% would still be in medicine. Given our lowest estimates for participation in medicine, our highest estimates for attrition from medicine were 8.7% at 2 years, 9.9% at 5 years, and 9.4% at 10 years.
Two years after graduation, women were working in medicine at similar rates to men (Table 1); the rates were between 92.6% and 95.6% for women compared with between 91.3% and 97.0% for men. In later years, the participation rates for women tended to be a little lower than those for men, but differences remained small (Table 1).
UK graduates whose family home prior to medical school was in the United Kingdom
Restricting the analysis to doctors known to be from family homes in the United Kingdom (Table 2) produced slightly higher rates of participation. We estimate that between 94.7% and 96.6% of UK graduates from UK homes, over all the cohorts studied, were working in medicine after 2 years, between 92.9% and 95.4% were still in medicine after 5 years, and between 94.0% and 96.1% were still in medicine after 10 years. Women had similar or slightly lower participation rates than men (Table 2).
UK graduates whose family home prior to medical school was outside the United Kingdom
Compared with graduates from UK homes, much higher proportions of graduates from family homes outside the United Kingdom were working in medicine abroad, and correspondingly smaller proportions were in the NHS; however, the total of those not working in medicine at all were similar (Table 3). We estimate that between 88.5% and 94.6% of the graduates from non-UK homes, over all the cohorts studied, were working in medicine after 2 years, between 85.6% and 93.9% after 5 years, and between 82.4% and 95.1% after 10 years. The wider ranges at longer time intervals represent greater uncertainty about the employment characteristics of nonresponders as, increasingly over time, the MCRG lost touch with some of the UK-trained doctors from non-UK family homes. Again, women who were from outside of the United Kingdom when they began medical school had only slightly higher attrition rates than did their male peers (Table 3).
Doctors from UK homes working in medicine but not in the NHS
In our cohorts, we estimated that the percentages of UK-trained doctors, from UK homes, who were working in UK medicine but not in the NHS, were between 1.5% and 1.9% after 2 years, between 2.3% and 3.1% after 5 years, and between 2.6% and 3.3% after 10 years (Table 2). These doctors include those employed in, for example, the civil service (e.g., the DH), the Health Protection Agency (which covers communicable disease control and environmental hazards), the Armed Forces, and the Prison Service (all of which are outside the NHS). Doctors practicing in the United Kingdom, but not in the NHS, may also be in clinical practice without an NHS contract (i.e., “private practice”) or in occupational health (also in the private sector).
We estimated that the percentages of home UK medical school graduates who were working in medicine but outside the United Kingdom were between 4.5% and 5.9% after 2 years, between 4.6% and 6.2% after 5 years, and between 4.9% and 6.3% after 10 years (Table 2).
Percentages in both groups (those working in medicine in the United Kingdom but not for the NHS and those working in medicine outside of the United Kingdom) tended to be a little higher for men than for women (Table 2).
Results for individual cohorts
Supplemental Digital Tables 1–9, http://links.lww.com/ACADMED/A123, show the estimates of participation for individual year-of-graduation cohorts for all UK medical school graduates, for UK medical school graduates from the United Kingdom, and for UK medical school graduates from outside the United Kingdom, from which the aggregate data in Tables 1 to 3 have been derived.
Comparing graduates (in different cohorts) at the same stage in their careers, we found no evidence of any appreciable temporal trend by year of graduation in the rates of participation in medicine, or indeed of any substantial variation among cohorts, comparing the cohorts at the same career stage. For example, for all UK graduates the minimum estimate for participation in medicine anywhere varies among cohorts (Supplemental Digital Table 1) from 89.4% to 93.0% after 2 years, from 88.9% to 91.0% after 5 years, and from 88.5% to 91.8% after 10 years—all ranges of variation which are modest.
We found that the great majority of UK-trained medical graduates work in medicine for many years after graduation. This is no doubt in part because they have made a substantial commitment of time, personal resources, and intellectual energy in the pursuit of a medical career. It probably also reflects good recruitment decisions in the selection of students for entry into medical school. This trend has been stable across graduation cohorts; in particular, recent graduates are no more likely to leave UK medicine than older generations of doctors. We also found that the great majority of graduates from UK medical schools remain in the United Kingdom to practice medicine. This means that, on current evidence, medical educators and medical workforce planners in the United Kingdom can gauge with confidence the likely number of doctors from each graduating cohort who will be in the UK medical workforce for years to come.
According to conventional wisdom, the increased number of women entering medical school around the world may be wasteful if women do not stay in the medical workforce. Our findings show that gender differences in participation are modest; women are only very slightly less likely than men to continue to follow a medical career for many years after graduation. We observed just a slight dip in the percentage of women working in medicine 5 and 10 years after graduation—years that likely coincide with taking time off work for child rearing. However, many women may work less than full-time during the child-rearing years. This part-time work by women (and to a lesser but increasing extent by men) is an important factor to take into account in planning service requirements in terms of whole (i.e., full)-time equivalent employees. The calculation of whole-time equivalent work is not straightforward using capture–recapture methodology and is beyond the scope of this report. The MCRG has reported elsewhere on the number and percentage of respondents who work part-time.6,11
The percentage, in UK medical schools, of students whose homes prior to medical school were outside the United Kingdom is fairly small (about 7%). Our data show that a majority stay in the United Kingdom for postgraduate training and practice. The percentage who eventually leave the United Kingdom increases gradually with time, but even 20 years after graduation, about half are still in practice in the UK NHS (Table 3). Those involved in workforce planning and/or in creating policies regarding permission for continued residence in the United Kingdom need to take account of our evidence showing that many doctors from overseas have eventually left the United Kingdom, at least in the past.
Strengths and weaknesses of the study
The major strengths of the study are its large national scale, long time span, and continuity with the same cohorts for, in some cases, over 30 years. As in any survey-based study, our analysis is open to the possibility of responder bias. In this study, such a bias may lead to overestimation or underestimation of participation in medicine; however, our combination of survey data and annual DH employment data, along with our use of capture–recapture analysis, means that our estimation of NHS participation, which forms the great majority of employment undertaken by the cohorts, is likely to be very accurate.
We used two methods to estimate participation in medicine overall, including work undertaken outside the NHS but in the United Kingdom, and work undertaken in medicine outside the United Kingdom. The minimum estimate assumed that only the doctors who reported on their questionnaires that they were working in medicine but not in the NHS were the only ones doing so. This is certainly an underestimate; some nonresponders were likely working in medicine, particularly those who had left the United Kingdom without leaving a forwarding address. Likewise, the maximum estimate assumed that the percentage of doctors in the whole cohort who work in non-NHS UK medicine, who work in medicine abroad, and who are not in medicine (whether or not abroad) have the same characteristics of practicing in medicine as those of our survey respondents. This is almost certainly an overestimate. Survey respondents may underrepresent those not in medicine or in medicine abroad, because graduates no longer working in medicine or working in medicine abroad may be harder to trace than those working as doctors in the United Kingdom. Even if traced, those outside medicine may be more likely to decide that the survey does not apply to them and may, therefore, be less likely to respond than those working in medicine. This means that the relative number of responders in these three groups (in non-NHS UK medicine, in medicine abroad, not in medicine) would overestimate those in medicine. Overestimation is probably fairly small, given both the high percentage of graduates known to be in medicine in the UK NHS (from our capture–recapture analysis) and the appreciable number of non-NHS doctors who reported on their surveys that they are in medicine.
We did not know the family home of all the doctors in the cohorts. For example, of the 3,482 doctors in the 1993 cohort, we knew that 2,921 were from UK homes when they started medical school and that 170 were from overseas when they began medical school, leaving 391 whose origins were unknown. We included the last group in Table 1, which gives results for all doctors in the cohorts, but not in Tables 2 and 3, which are defined by place of origin.
Finally, the profile of retention of UK-trained medical graduates staying in medical practice may apply to other high-income countries, but this cannot be assumed. We recommend that others, outside of the United Kingdom, consider applying methods like ours to study the careers of their medical graduates.
Comparisons with other studies, main findings, interpretation, policy implications
Other studies on attrition in medicine cover attrition from specific specialties only,5,12,13 do not cover a national sample,14 or cover attrition over fairly short periods of time.15 Further, these studies cover intentions and motivations to leave medicine rather than measures of actual outcomes.16 To the best of our knowledge, ours is the only study that is national in scope and that spans such a long period of time. We provide best estimates for the minimum and maximum percentage of physicians participating in medicine. The minimums are certainly underestimates, and the maximums are overestimates; the true percentages lie somewhere in between.
In summary, the evidence suggests that the vast majority of UK medical school graduates continue to work in medicine for many years after they graduate. As an approximation for policy assumptions, we suggest that at least 90% is a highly reliable estimate. The actual percentages may be appreciably higher. For planning purposes, we note that the midpoint of our range of estimation (Table 1) for the percentage in medicine at 2, 5, 10, 15, 20, and 25 years after graduation is, respectively, 94%, 93%, 93%, 93%, 90%, and 90%. Hence, attrition from medicine can be expected to be no more than 10%, and likely less, in the early postgraduate years.
Acknowledgments: The authors are very grateful to all the doctors who participated in the surveys. The authors wish to thank Jean Davidson for analytical assistance and Emma Ayers, Janet Justice, and Alison Stockford for data preparation and administration.
Funding/Support: The UK Medical Careers Research Group is funded by the Policy Research Programme of the English Department of Health. The Unit of Health Care Epidemiology is funded by the English National Institute for Health Research.
Other disclosures: None.
Ethical approval: This study was approved by the Central Office for Research Ethics Committees, following referral to the Brighton Mid Sussex and East Sussex local research ethics committees.
Disclaimer: This is an independent report commissioned and funded by the Policy Research Programme in the Department of Health. The views expressed are not necessarily those of the Department of Health.