Workforce & Appointments by OC level

NHS England · GP Workforce Census, Appointments in General Practice · February 2026

Tertile design: 6,018 practices with OC data, grouped by OC submission rate (Feb 2026, per 1,000 patients per working day). This uses the full set of practices with OC submission data. The rush analysis uses a smaller subset (4,306 practices) which additionally required telephony (CBT) data.
Low OC (<1.58/1k/day, n=2,006) · Mid OC (1.58–8.48/1k/day, n=2,006) · High OC (>8.48/1k/day, n=2,006).
Sources: GP Workforce Statistics (Feb 2026, practice-level detailed). Appointments in General Practice (Feb 2026, practice-level crosstab). OC Submissions via OC Systems in General Practice (Feb 2026).

1. Workforce

Staff FTE per 1,000 patients

Monthly snapshot · Feb 2026 GP Workforce Census

Reading: Almost identical staffing across all three groups. GP FTE per 1k ranges from 0.58 (Low OC) to 0.61 (High OC). Nurses, DPC, and admin are similarly flat. OC adoption has not led to measurably different staffing levels.

Practice size & staff mix

Median list size and total clinical FTE per 1k

Reading: High-OC practices are substantially larger (median 10,880 vs 6,842 patients). Larger practices may find it easier to adopt OC due to infrastructure and admin capacity. Total clinical FTE per 1k is marginally higher in high-OC practices (1.18 vs 1.09).

Workforce summary

	Patients	Median list	GP FTE/1k	GP excl T&L	Nurse FTE/1k	DPC FTE/1k	Admin FTE/1k
Low OC	15.9m	6,842	0.58	0.43	0.25	0.26	1.16
Mid OC	21.6m	9,098	0.60	0.44	0.25	0.27	1.19
High OC	25.2m	10,880	0.61	0.45	0.27	0.30	1.23

GP excl T&L = fully qualified GPs excluding trainees and locums. DPC = direct patient care (pharmacists, physiotherapists, physician associates, etc.). Workforce is essentially flat across tertiles. The main difference is practice size: high-OC practices serve 59% more patients.

2. Total appointments

Appointments per 1,000 patients · by mode

Monthly totals · Feb 2026 · all appointment types, all staff

Reading: High-OC practices record 507 appointments per 1k vs 453 for low-OC — 55 more per 1k (12% higher). Face-to-face is lower in high OC (285 vs 316/1k). Telephone is higher (127 vs 102/1k). Video/Online is nearly four times higher (76 vs 20/1k) — though in practice most of this is text-based OC processing rather than video consultations. But 55 extra recorded appointments is a fraction of the input: high-OC practices process ~240 more OC submissions per 1k/month than low-OC. Most of the effort of reviewing and responding to those submissions does not show up here.

3. What kinds of appointments — and who delivers them?

Caveat on appointment categories: The AIGP data classifies each appointment into a “National Category” — General Consultation Routine, General Consultation Acute, Clinical Triage, Planned Clinics, etc. We do not fully understand how practices assign these categories, whether it is done consistently, or how clinical system defaults and OC workflows influence the classification. A rise in “Clinical Triage” may reflect genuine new triage activity, automatic coding by the system when an OC submission is processed, or simply a change in how existing work is labelled. The numbers below should be read with this uncertainty in mind.

Appointment category per 1,000 patients · GP vs Other Practice Staff

Monthly totals · each category has three bars (Low / Mid / High OC). Darker shade = GP, lighter shade = Other Practice Staff. · Feb 2026

Reading: Clinical Triage nearly triples from 33/1k (Low) to 80/1k (High). GP triage work more than doubles (21→46/1k) and other staff triage nearly triples (12→34/1k). Meanwhile, GP routine consultations fall from 79 to 63/1k — so part of what was previously a GP consultation is now a GP triage contact. Acute consultations, planned clinics, and planned procedures are stable across tertiles and have similar staff splits. OC doesn’t shift work from GPs to other staff — it adds a triage layer that loads both.

4. What does a GP session look like?

Assumptions & caveats: This metric is derived, not directly published. We take GP appointments per 1,000 patients (from AIGP) and divide by sessions per 1,000 patients, where 1 FTE = 9 sessions/week and Feb = 4 weeks = 36 sessions per FTE. We use GP FTE excluding trainees & locums from the workforce census. This assumes all sessions are equal (they are not — some are shorter, some are admin-heavy) and that FTE translates neatly into patient-facing sessions (it does not — GPs also do paperwork, training, and management). The 2018 baseline uses national totals rather than practice-level data and cannot be split by tertile — OC barely existed in 2018, so a tertile comparison would not be meaningful. You could retrospectively identify the same practices, but many have merged, changed, or started OC at very different times, making the comparison unreliable.

Crucially, in mid- and high-OC practices, GPs also spend time processing online consultations. This work — reading forms, triaging, messaging patients back — may not always generate a recorded appointment but still consumes session time. The “appointments per session” metric therefore understates total GP workload in high-OC practices.

GP appointments per session · by mode

Monthly data (Feb) · 1 FTE = 9 sessions/week = 36 sessions in Feb · GP excl trainees & locums · Feb 2018 national baseline + Feb 2026 by OC tertile

Reading: In 2018, a GP session was 9.4 face-to-face and 2.2 telephone — almost entirely in the room or on the phone (11.8 total). By 2026, even low-OC practices have shifted: fewer face-to-face (8.8), more telephone (3.6), and a small video/online slice (0.6). High-OC practices have moved further: 7.0 face-to-face, 4.3 telephone, 2.3 online. The session has also got busier — 13.9 recorded appointments vs 11.8 in 2018. But in high-OC practices, GPs are also processing online consultations outside these recorded appointments.

GP appointments per session · summary

	Face-to-face	Telephone	Video/Online	Home visit	Total
Feb 2018 (national)	9.4	2.2	0.1	0.1	11.8
Low OC (2026)	8.8	3.6	0.6	0.1	13.2
Mid OC (2026)	7.8	4.4	1.4	0.2	13.8
High OC (2026)	7.0	4.3	2.3	0.2	13.9

Key point: The GP’s session has got busier (+2 recorded appointments) and the channel mix has shifted dramatically. In 2018, 80% of a session was face-to-face. By 2026, even without OC it’s down to 67%; with high OC it’s 50%. The extra appointments per session are phone and online contacts — the face-to-face count has actually fallen. These figures count only recorded appointments; time spent reading and responding to OC submissions is additional unrecorded workload.
Sources: Feb 2018 baseline from AIGP national totals + GP Workforce Census Mar 2018 (27,892 FTE excl registrars & locums). Feb 2026 from practice-level data.

5. Exploring appointment mode

Appointment mode (face-to-face, telephone, video/online) shows the largest variation between OC tertiles. Here we break it down by National Category to see where the shift is happening. The same caveats about category classification apply.

Triage appointments by mode · per 1,000 patients

Monthly totals · Feb 2026

Reading: In Low-OC practices, triage is mainly by telephone (19/1k, 57%). In High-OC practices, triage is dominated by Video/Online (41/1k, 51%). Total Video/Online appointments across all categories are 76/1k for High OC — so triage alone accounts for over half of all online appointments. The “digital shift” in the appointment data is largely a triage phenomenon.

Routine consultations by mode · per 1,000 patients

Monthly totals · Feb 2026

Reading: Routine consultations barely shift mode. Face-to-face dominates at 66–74% regardless of OC level. Video/Online is 2–3/1k — trivial. Telephone is modestly higher in high OC practices (37 vs 35/1k). The actual consultations still happen face-to-face; the digital channel is used for the triage step that precedes them.

6. How quickly are patients seen?

Time from booking to appointment (%)

Monthly data · Feb 2026

Reading: Same-day appointments: 42% (Low) to 46% (High). Within one week: 69% to 71%. The differences are modest. High-OC practices are marginally quicker to see patients, but the gap is small.

7. DNA rate

Did Not Attend rate (%)

Monthly data · Feb 2026 · as proportion of attended + DNA

Reading: DNA rate is virtually identical: 4.3% (Low), 4.1% (Mid), 3.9% (High). The marginal improvement in high-OC practices may reflect higher same-day appointment rates (patients less likely to DNA a same-day slot) rather than OC itself.

Appendix: The triage count is driven by the clinical system

Triage appointments per 1,000 patients · EMIS vs TPP × OC tertile

Clinical system from GPAD supplier field · Feb 2026

Reading: TPP (SystmOne) practices record more triage at every OC level: 37/1k (Low) vs 31/1k on EMIS, and 108/1k (High) vs 61/1k on EMIS. The clinical system roughly doubles the triage count at high OC levels. The gap between systems is wider at higher OC levels — suggesting the way SystmOne records OC processing amplifies the triage count more than EMIS does.

Total appointments per 1,000 patients · EMIS vs TPP × OC tertile

Feb 2026

Reading: TPP practices record substantially more appointments overall at every OC level (490–562/1k vs 425–465/1k on EMIS). The gap grows with OC adoption, driven by triage inflation. This means the earlier finding that “high-OC practices record 12% more appointments” is partly a clinical system artefact: TPP practices are over-represented in the high-OC tertile.

Appendix (cont.): Does the OC supplier matter within the same clinical system?

Triage/1k · TPP (SystmOne) practices

By OC supplier · single-OC-supplier only · n≥15

Reading: On TPP, most OC suppliers produce similar triage (51–57/1k). The outlier is TPP’s own integrated OC at 165/1k — three times higher. This is the workflow where SystmOne’s OC and appointment book are fully integrated, auto-generating a triage entry for each submission. Continuum is also higher (99/1k).

Triage/1k · EMIS practices

By OC supplier · single-OC-supplier only · n≥15

Reading: On EMIS, OC supplier variation is more modest. Engage is highest (77/1k), Blinx next (70/1k), Continuum (69/1k), with Accurx, eConsult, Klinik, and PATCHS clustering at 46–49/1k. Evergreen and Silicon Practice are lowest (28–39/1k).

Summary table · clinical system × OC tertile

System	Tertile	Practices	Total/1k	Triage/1k	Routine/1k	Acute/1k
EMIS	Low	1,153	425.1	30.9	143.7	80.0
	Mid	1,217	448.6	50.6	126.6	84.2
	High	1,149	465.4	60.8	120.8	87.6
TPP	Low	850	490.3	36.6	134.0	91.6
	Mid	789	528.6	66.8	116.7	91.7
	High	856	562.3	107.5	104.7	88.5

Key point: On both systems, triage is higher and routine is lower in high OC practices. But TPP records 50–80% more triage than EMIS at the same OC level. Total appointments are 65–97/1k higher on TPP. This is a recording difference, not a workload difference. Acute consultations are stable across all cells. EMIS covers 3,519 of 6,014 practices (59%); TPP covers 2,495 (41%).

8. Summary

Workforce: Essentially identical across OC tertiles. GP, nurse, DPC, and admin staffing per 1,000 patients shows no meaningful variation. The main structural difference is practice size: high-OC practices are 59% larger (median 10.9k vs 6.8k).

Appointments: High-OC practices record 507 appointments per 1k vs 453 for low-OC. Face-to-face appointments are lower (285 vs 316/1k) and telephone higher (127 vs 102/1k). The “video/online” category nearly quadruples (20→76/1k) — but this is largely text-based OC processing rather than video consultations. Routine consultations fall; acute, planned clinics, and planned procedures are stable. Wait times are marginally shorter (46% same-day vs 42%). DNA rates are similar.

OC processing effort: High-OC practices process ~12.9 OC submissions per 1k patients per working day, compared with 0.8 in low-OC. The effort of reviewing and responding to those submissions — reading forms, triaging, messaging patients back — is extremely unlikely to be fully captured in the appointment data. The recorded appointments show the output; the input effort is largely invisible.

The clinical system effect (see appendix): TPP (SystmOne) practices record 50–80% more triage than EMIS practices at the same OC level. We don’t fully understand why. If you use SystmOne and have a view on what drives this, we’d welcome your feedback.

Implication: The argument for OC-driven triage is not that it reduces total clinical workload — it is that it ensures the right clinician sees each patient, and that GPs do only the work that requires a GP. The appointment data cannot tell us whether that is happening. What we can see is that in medium- and high-OC practices, there is significant work associated with managing OC that is not recorded in the appointment data. Whether that unmeasured effort is offset by more efficient use of GP time remains an open question.