General Practice Workforce, 31 May 2022

The figures presented in this publication are a snapshot of GPs, Nurses, Direct Patient Care (DPC) and administrative staff working in general practice at the end of the reporting period, which includes any staff on fixed term or zero hours contracts.

Annexes B and C of the Excel Bulletin provide information about GPs working in an ad-hoc or short-term capacity as it is important to understand the extent of their contribution to general practice. However, we do not include counts of these ad-hoc locums (also sometimes known as sessional GPs) in the main publication totals for reasons described in the Using this publication section which contains more details about the ad-hoc locum workforce.

We collect information about job role, contracted and working hours, gender, age, and ethnicity, but no information relating to earnings or expenses.

We always recommend that you use the most recent publication for up-to-date and time-series figures. Furthermore, for the release presenting statistics for 31 December 2021, we implemented a range of methodological changes (please refer to Methodological Review and Changes of the December 2021 publication for details) and revised the entire time series back to September 2015. Although the earlier publications are available from the publication archive, we encourage all users to refer to current figures.

We collect record-level data directly from general practices using the National Workforce Reporting Service (NWRS) which is an online collection system. Please see https://digital.nhs.uk/data-and-information/areas-of-interest/workforce/national-workforce-reporting-system-nwrs-workforce-census-module/the-new-national-workforce-reporting-service for more details about this tool.

Although it is a contractual requirement for practices to submit their workforce data, not all practices do so. However, the completeness and coverage of the data set has increased since the first collection in September 2015 when only 88.1% of practices submitted data.

Until December 2018, we received data about some practices from four Health Education England (HEE) regions which collected workforce data directly from the practices in their regions although only three submitted data for the September 2015 collection.

Table 4 shows the count and percentage of practices submitting data and the overall coverage and it is notable that the completeness of the collection increased dramatically for December 2016.

Table 4: Data submission and coverage

We ask practices to ensure that information about their workforce and details of any ad-hoc locums that worked at the practice during the reporting period are up to date at the end of each month and extract the data automatically.

The NWRS has inbuilt data validation functionality to reduce data entry errors such as limiting job roles to those within the Workforce Data set (NWD), checking the format of national insurance numbers or ensuring only numbers are entered for numeric fields.

In some cases, practices provide staff details but do not include information about working hours and in these cases, we calculate full-time equivalent (FTE) estimates (referred to as partial estimates) as described in the Using this publication section.

Where practices have not provided any workforce data for entire staff groups, we calculate headcount and full-time equivalent estimates, aggregated to CCG level (full estimates). For Direct Patient Care staff, we only calculate full estimates for practices if they have not provided any Direct Patient Care staff data and are missing at least one other entire staff group. See Estimates for a full explanation of the estimation methodology.

Impact of COVID-19

There is some evidence to suggest that the completeness and coverage of the data extracted from the NWRS for March and June 2020 were adversely affected by a range of issues, including the COVID-19 pandemic, and it is likely that the exceptional pressures on the general practice workforce meant not all practices were able to update their NWRS data in time for the extracts on 31 March and 30 June 2020. However, from September 2020 onwards, we are confident practices had returned to normal levels of NWRS data completion in time for each quarterly extract.

Nonetheless, the possible impact of the data quality issues should be taken into consideration when considering figures for these two quarters.

We have been collecting individual-level data from general practices and publishing this series of Official Statistics since September 2015 when a new data source was introduced, and we have been working closely with practices to improve the quality, completeness and coverage of the data submitted.

All general practices are contractually required to provide data on their workforce. However, a small number of general practices do not do so or submit incomplete data. In addition, some records fail data validation and are excluded. The completeness and coverage of the collection has improved since September 2015 when only 88.1% of practices submitted data; this percentage had reached 99.7% by May 2022.

We collect information about the general practice workforce directly from practices using a live, online collection tool. The completeness and coverage of the data collection has improved since we first collected wMDS in September 2015, but nonetheless there are still some data quality considerations.

Full estimates

There are some practices where all the provided data for a staff group is of poor quality and has to be removed. We calculate estimates for headcount and full-time equivalence for those practices which did not provide valid and/or complete data for one or more staff groups; this could be due to poor data quality or no submitted data.

To produce estimates for incomplete or missing data, we use the valid data submitted by the other practices during the reporting period, along with information about their registered patient population. Our estimation methodology takes the practice patient populations into account to address potential issues that could arise if the practices providing no data or poor-quality data were not of a typical or average size. We know the registered patient counts for 98-99% of practices and use the average registered patient count of a practice’s Sustainability and Transformation Plan (STP) area where this information is not available.

We produce our estimates as follows:

1. For every job role and STP, we calculate a ratio of FTE per registered patient. This uses the total FTE and registered patient count for all the practices that supplied valid data.
2. We calculate FTE estimates at CCG-level by taking the STP job role ratio calculated in step (i) and multiplying it by the total registered patient count for all practices in that CCG that did not supply valid data for the applicable staff group.
3. We use the same principles to calculate headcount estimates.
4. We aggregate the estimates from CCG level to provide higher level figures at national and sub-national levels.

We produce estimates independently for each of the four staff groups (GPs, Nurses, Direct Patient Care and Admin/Non-clinical staff). This means that if a practice submits no data, or invalid data for a single staff group, their submission for the other three staff groups is still treated as valid.

For the Direct Patient Care staff group, using the same estimation methodology is not appropriate for the following reasons.

• Practices are more likely to be operating without Direct Patient Care staff. This is borne out by the fact the percentage of practices who do not record Direct Patient Care staff in NWRS is much higher than the other staff groups
• The introduction of Primary Care Networks means staff may be employed by the PCN instead. Estimating for them in general practice would therefore over-estimate this part of the primary care workforce
• This is the largest staff group in terms of number of job roles with 18, but no practices who submit information about DPC staff have records for every Direct Patient Care job role. The methodology can only estimate for all job roles within a staff group and so it is highly likely that this leads to overall over-estimation of DPC staff
• Estimation testing (see the Methodological Change Notice) showed erratic results for the DPC staff group, with large overestimates

To limit the degree of over-estimation of Direct Patient Care staff, where a practice has provided information about GPs, nurses and admin/non-clinical staff and only the Direct Patient Care staff group is missing, we do not calculate estimates for that practice. This alternative approach is based on an assumption that the majority of practices who provide information about three staff groups but not Direct Patient Care are likely to not employ Direct Patient Care staff. 

We are aware that seasonal variation affects General Practice workforce figures, but as our estimation process calculates ratios using valid data for the applicable reporting period, any seasonal variation should be mitigated.

It is important to note that we do not produce wholly estimated records at individual practice level. At general practice level, we only estimate missing FTE where the staff member’s record is otherwise valid and complete and is simply missing contracted or working hours (see Partial Estimates below).

We do not allocate estimated records any age, gender, or country of qualification information; for these data items, any estimated records are reported against ‘Unknown’.

We no longer produce estimates for GPs in training grades as the timely and complete TIS data means this is not required. However, some historical figures do include some fully estimated records for GPs in training and locums:

• Following the introduction of Health Education England’s (HEE) Trainee Information System (TIS) as the data source for GPs in training, we calculated some estimates for missing data for September 2015 to March 2018. This is discussed in further detail in the GPs in Training (formerly referred to as GP registrars) section. These estimates are available only at a national level and regional figures from June 2018 onwards are not comparable with earlier figures.
• Early in 2017 we introduced new guidance relating to recording data about the locum GP workforce and in March 2017 we noticed a large increase in their numbers. Following a consultation with stakeholders, we calculated FTE estimates for the locum GP workforce to account for records that should have been submitted between September 2015 and December 2016. These figures are published at CCG level but are available only at a high level and do not include personal characteristic such as gender or age so caution should be used when considering such breakdowns in earlier reporting periods. There remains a small ‘step change’ between December 2016 and March 2017, but the overall counts of GP FTEs are believed to be consistent and comparable.
  • It was not possible to calculate estimated headcount figures for these locum GPs so there is an unavoidable break in the headcount time series between December 2016 and March 2017. Please refer to the Locums section in the Using this publication section.
• Due to data quality issues with the ad-hoc locum data collection in the last quarter of 2020, the ad-hoc locum figures in Annex B of the Excel Bulletin tables include some estimated FTE figures for December 2020. However, it is not possible to calculate headcount estimates for these ad-hoc locum GPs.

When we create estimates in this way, we do not allocate the records any personal characteristics such as age, gender, or country of qualification information; but report these against the ‘Unknown’ category. This means that figures by these personal characteristics are not comparable across time periods.

Partial estimates (estimated FTE)

We calculate estimated FTE figures for otherwise valid staff records where no information about contracted or average working hours has been provided. The estimates are calculated using the national average FTE for the applicable job role. We refer to these figures are ‘partial estimates’ and the scale of these estimates varies by staff group. Annex A in the Excel Bulletin tables shows the percentage of partially estimated figures for each staff group.

We are aware that some practices make retrospective changes to their workforce data after we have already extracted their information and published the statistics, for example, by changing a job role or start date. We reflect any such changes from the release of subsequent statistics, but we cannot retrospectively amend the time series as this would result in all figures being continually subject to revision. However, the scale of records affected by such retrospective changes has been found to be generally stable across quarters.

The collection includes details of staff on long-term absence, for example due to sickness or maternity/paternity leave. It is possible that FTE and headcount figures for all staff groups are inflated by the inclusion of temporary staff – such as locum GPs – that work at practices to cover for these staff absences. Although we used to collect information about staff absence, the data completeness was very low and it was not possible to adjust FTE and headcount numbers to mitigate for this potential double-counting. We are not currently collecting details of staff absence.

From June 2021 onwards, we have removed the FTE and headcount figures of ad-hoc locums from the overall totals, and are publishing them separately in Annex B the Excel Bulletin for England, and in Annex C by CCG. Although these GPs provide an immensely valuable contribution to general practice, including them in the figures is likely to result in double-counting, which distorts our understanding of capacity within the general practice workforce and inflates the rate of GPs per 100,000 registered patients. This is because many of the sessions worked by these ad-hoc locums are to cover for the short-term absence of other GPs who usually work in the practice and are therefore also included in the practice’s total workforce.

Moreover, there is a delay in the availability of ad-hoc locum-related data. The negative impact of this delay on the reporting of ad-hoc locums was exacerbated by the move to monthly collections from June 2021. Figures presented for the two most recent months in Annexes B and C are marked as provisional and are updated in subsequent months. Separating these figures from the overall GP workforce totals means that the figures in the main tables are not required to be provisional when first published.

We have been collecting individual-level data and publishing the General Practice Workforce series of Official Statistics since September 2015 when a new data source was introduced and figures in this series are not comparable with any general practice workforce figures based upon previous data sources.

As described in the Using this publication section, it is not always possible or appropriate to compare figures over time.

We urge anyone considering GP workforce statistics, to make any comparisons only across full years, for example September to September, rather than from one quarter or month to the next. This is because data is affected by seasonal variation. 

In addition, the inclusion of some estimated figures for GPs in training and locum GPs mean that not all figures in the times series are comparable back to the beginning of the data collection.

Table 5 shows the earliest point at which year-on-year may be made:

• As outlined in the Using this publication section, where GP counts include GPs in training, estimates for September 2015 to March 2018 mean that regional comparisons can be made only from June 2018.
• Similarly, where locum counts are included, there is an unavoidable break in the headcount time series and year-on-year figures are comparable only from December 2017.

Table 5: Earliest date in this publication series where year-on-year (whole year) figures may be compared

1. Adjustment applied to September 2015 to December 2016 data to account for an improvement in GP locum recording.
2. Adjustment applied to September 2015 to March 2018 data to account for a changing data source for GPs in training.
3. Change in data source of GPs in training led to a notable reduction in GPs in training of 'unknown' location and region.
4. Some records within the Admin/non-Clinical staff group could not allocated to a specific job role in September 2016 and were therefore categorised as Not Stated for that reporting period. As a result, job role figures are comparable only from December 2016.

Time series figures

All published data are comparable with the same point in earlier years unless otherwise stated and shown notably by a vertical line in the Excel Bulletin tables.

There is a break in the headcount time series for locum GPs at March 2017 that affects the figures for locums, All GPs and All Fully Qualified GPs.

There are also two breaks in the headcount and full-time equivalent (FTE) time series for Admin/Non-Clinical staff role-level figures, due to the presence in September 2016 of multiple Admin/Non-Clinical records with a ‘Not Stated’ staff role.

Impact of ‘unknown’ data

In some cases, staff records may not include details of age, gender, ethnicity, or other characteristics and in such cases, we report the FTE and headcount figures as ‘Unknown’. Therefore, figures by these personal characteristics are not comparable over time.

The relevance of NHS workforce data is maintained by reference to working groups who oversee both data and reporting standards. Major changes to both are subject to approval by the Data Alliance Partnership Board (DAPB) which replaced the Data Coordination Board in 2021.

Significant changes to workforce publications (e.g., frequency or methodology) are subject to consultation, in line with the Code of Practice for Statistics.

We publish figures as quickly as possible after extract and announce our publication schedule on our website at https://digital.nhs.uk/services/organisation-data-service/data-downloads/production-schedule.

From July 2021,  main workforce figures are published monthly. However, we release some regional and joiner / leaver figures less frequently - see the Using this publication section and will include details of the planned content in the Changes in this series section.

FTE figures are rounded to the nearest whole number.

Totals may not add to the sum of their components as a result of rounding.

We use the following symbols in tables:

..          not applicable

-           zero

0          greater than zero but less than 0.5

ND      No data

|           A time-series break, i.e., figures either side of the break are not comparable