Medical specialty training and transition to the GP workforce: technical document

9 December 2025

1 Overview

Data source names	TPM, NPCCD, LRMP
Update frequency	Annual
Census dates	30 September (GP headcount)

2 Data sources

2.1 Turas Programme Management (TPM)

TPM is managed by NHS Education for Scotland (NES) and holds information on specialty registrars, their trainers and programmes. TPM receives data on specialty registrars and their programmes from the Oriel recruitment portal, which holds data on the outcome of specialty training recruitment. Records in TPM are then managed and updated locally by administrative teams in the Scotland Deanery.

2.2 National Primary Care Contractor Database (NPCCD)

The NPCCD is a centralised database of primary care clinicians and general practice details, held at Public Health Scotland but with data maintained by NHS Boards.

NPCCD is the authoritative source of information on all general practitioner (GP) and GP practice data in Scotland. Legislation states that all practitioners must be on the performer list (NPCCD) and validated prior to commencing work.

The trained workforce includes those working in substantive roles (performer, performer salaried and performer retainee) in a general practice. Those working only in the Out of Hours service or only in locum roles are not included.

2.3 List of Registered Medical Practitioners (LRMP)

NES receives downloads of the General Medical Council’s (GMC) list of registered medical practitioners. The LRMP contains data on doctors’ sex and the country of their primary medical qualification. These data are more complete and of a higher quality than what is recorded for the same doctors in TPM, and they are the preferred data source for these variables. Record linkage is carried out using doctors’ professional registration number.

3 Methods

3.1 Number starting training

The intake to GP specialty training (GPST) in a given year is the headcount of doctors whose programme start date fell in that year and who started in the first year of the programme, i.e., who started at Specialty Training Grade 1 (ST1).

3.2 Number completing training

A doctor completes specialty training having achieved Outcome 6 in their final Annual Review of Competency Progression (ARCP). Outcome 6 is the Deanery’s recommendation for completion of training. The doctor then applies through their college/faculty for the Certificate of Completion of Training (CCT). Following a successful application, the General Medical Council (GMC) awards the CCT and the doctor is added to the Specialist Register.

The number completing GP specialty training in a given year is the headcount of doctors who had a CCT date in that year and were recorded in TPM as having reached the end of their training.

3.3 Probability of completing training and length of training

3.3.1 Overview

Survival analysis methods are used to examine whether and when doctors successfully complete training.

These methods produce completion probabilities, which can be interpreted as the probability of completing within a certain duration after starting.

For example, a three year completion probability of 0.4 can be interpreted as meaning that the probability of a doctor successfully completing training within three years is 0.4.

These probabilities are estimated using the Kaplan-Meier statistic.

3.3.2 Detail

The Kaplan-Meier statistic comes from a field of statistics known as Survival Analysis. It typically estimates, for each time point, the number of people who have not yet experienced some event (that is, who have “survived” it), as a proportion of the number of people who were at risk of experiencing the event immediately prior to that time.

Calculating this proportion at each time point and multiplying it by the same proportions from all previous times builds the survival function. The survival function estimates the probability of being “event-free” by a given time, conditional upon having been event-free up until that time point.

The complement of this function (1 minus the survival function) estimates instead the cumulative probability of having experienced the event by a certain time.

Applied to our GP specialty registrars, the event of interest is completing the programme and attaining CCT, and the at-risk group are those doctors who have not yet completed. Since we want to estimate the probability of doctors completing by a given time, we use the complement of the survival function. The at-risk group naturally gets smaller over time as doctors complete and are no longer at risk of doing so in the future. Doctors who have not completed training by the time of our analysis will always be counted in the at-risk group for each time point.

The data are prepared for analysis as follows:

Given that a doctor has started GPST1, they are coded as Complete if they attain their CCT, and the duration in days between the start of their training and their CCT is recorded. If a doctor has started but not completed GPST training by the time of our analysis, they are coded as Not Complete, and the duration in days between the start of their training and the time of our analysis is recorded. For doctors who are coded Not Complete, their data are said to be censored because their outcome is not visible to us at the time of the analysis: they may go on to complete their training later or they may leave the programme without completing it.

Survival probabilities are estimated for each event-time in the data, i.e., for each observed combination of number of days and number of CCTs. Survival probabilities are subtracted from 1 to produce completion probabilities, which estimate the probability of completing GPST training by a given time.

3.3.3 Interpretation

The completion probability values in our charts and tables can be interpreted directly as the proportion of doctors in our sample who attained their CCT by a certain time after starting training. For example, a completion probability of 0.4 at around 1,095 days (or three years) would indicate that 0.4, or 40%, of doctors in our sample completed within three years.

In probability terms, we would say there is a probability of 0.4 that a doctor starting GPST training will complete within three years. More intuitively, we could say there is a 40% chance that a doctor starting GPST training will complete within three years.

A further use of completion probabilities relates to planning. Given that a certain number of doctors has started training, these probabilities can be used to forecast the numbers of newly qualified GPs at certain points in the future. They can also be used to calculate the required input to training in order to reach some future target of newly qualified GPs. If we require, say, ten newly qualified GPs in three years’ time, we will want to see 10/0.4 doctors beginning training now. That is, we estimate that we will need a cohort of 25 doctors starting training now to have 10 ten newly qualified GPs in three years’ time. (This is a simplified example that does not take account of the outputs of previous cohorts.)

Completion probabilities can be used to compare completion and the duration of training between groups, for example, between training cohorts or between male and female doctors.

We refer to completion probabilities as estimates. While they are based on straightforward sample proportions, they are estimates in the sense that some observations in the samlple are censored, that is, we are unable to determine the outcome of some doctors’ training at the time of the analysis. They are also estimates in the sense that they are statistics based on a sample of data, and that we use them to predict (or estimate) outcomes of future cohorts whose training has not yet been observed.

3.4 Transition to employment

The probability of transitioning to the GP workforce follows the methodology for the probability of completing training. We are interested in whether a doctor who has been awarded a CCT subsequently appears in the GP workforce data, and the duration between CCT and first appearance in the data as a practising GP. This gives an estimate of the size and timing of the inflow from training into the GP workforce.

In this publication, the GP workforce includes those working in substantive roles (performer, performer salaried and performer retainee) in a general practice. Those working only in the Out of Hours service or only in locum roles are not included.

The event of interest is a doctor’s first appearance in the NPCCD GP workforce data, and the duration measured is the number of days between the CCT date and the first contract start date in NPCCD. Doctors who CCT but who have not appeared in NPCCD by the time of our analysis are treated as censored.

3.4.1 Interpretation

Transition probabilities can be interpreted directly as the proportion of doctors in our sample who joined the GP workforce by a certain time after completing training. For example, a transition probability of 0.4 at 30 days would indicate that 0.4, or 40%, of doctors in our sample joined the workforce by 30 days post-CCT.

In probability terms, we would say there is a probability of 0.4 that a doctor completing GPST training will join the workforce within 30 days. More intuitively, we could say there is a 40% chance that a doctor completing GPST training will join the workforce within 30 days.

Like the training completion probabilities, these employment transition probabilities are based on sample proportions, but are estimates in the sense that they account for censored observations, i.e., doctors who may join the workforce later, and in the sense that we can use them to predict transition for future cohorts of GP registrars.

3.5 Stocks and flows

3.5.1 Overview

Stocks and flows are terms used to understand the size, and changes in the size, of the workforce. Stocks are measured at a particular point in time; flows are measured over a period of time. The size of the stock changes according to the relative size of inflows and outflows.

3.5.2 Detail

In the context of the GP workforce, the stock of GPs is measured as the headcount of practising GPs on 30 September each year. Inflow is measured as the headcount of GPs who joined or returned to the workforce in the previous 12 months, and outflow as the headcount of GPs who left the workforce in the previous 12 months.

For example, a GP who appears in the workforce data for the first time on 30 September 2024 is counted as an inflow in the 12 months up to 30 September 2024. A GP who returned to practice after several years and was observed in the 30 September 2024 data is also counted as an inflow. A GP who was observed in the data on 30 September 2023 but not on 30 September 2024 is counted as an outflow in the 12 months up to 30 September 2024.

3.5.3 Interpretation

When plotted as time series, outflows are reported at the beginning of the year and inflows are reported at the end of the end of the year. For example, the data points for inflow and outflow at 30 September 2023 would indicate that x many GPs joined the workforce since September 2022, and that y many GPs left the workforce between September 2023 and September 2024.

Inflows can be decomposed into a number of sources such as inflows from specialty training and inflows from returners. We can also analyse inflows and outflows by personal characteristics such as age and sex.

4 What data are published?

Number starting training: a time series of the number of doctors commencing GP specialty training.
Duration of training: survival-based analysis of the probability and timing of GP registrars completing training.
Number completing training: a time series of the number of doctors completing GP specialty training.
Transition to the workforce: survival-based analysis of the probability and timing of joining the GP workforce post-CCT.
Inflow from training: a time series of the number of doctors joining the workforce after completing training.
Stock and flow: time series of the stock of GPs and workforce inflows and outflows, with source of inflow.
Outflow by age group: time series of outflows, with outflows per age band expressed as a percentage of total outflow.

Statistics can be examined by doctors’ sex and source of primary medical qualification for numbers of GP registrars starting and completing training, duration of training, transition to the workforce, and workforce stocks and flows.

5 Data quality

5.1 TPM

For the release of these official statistics in development, the TPM data have been initially assessed using the Quality Assurance of Administrative Data (QAAD) framework, against the same basic-to-enhanced assurance levels required for our Official Statistics releases. Feedback from users will form part of our ongoing assessment during the development period.

5.2 NPCCD

NPCCD is maintained by NHS Boards to manage their performer lists. Contract details may be updated to indicate changes for a time after the contract has ended. Currently, all data is subject to revision, meaning historical changes can affect the time series - although these changes tend to be small.

5.3 LRMP

The GMC ensures that every doctor has the right knowledge, skills, qualifications and experience to work across the UK. It does this by maintaining official lists of registered doctors. To retain their licence to practice, doctors must keep their knowledge and skills up to date through the GMC’s process of revalidation. The Register is updated daily.

The GMC licences certain organisations to download daily updates to the complete register. As part of this licence agreement, NES is permitted to use register data for non-profit research including the production of official statistics.