NHS England’s modelling for the Long Term Workforce Plan

Recommendation 1: NHSE should develop a modelling pipeline whose different parts are fully integrated to avoid manual processing, and which creates the capability to more easily test and produce outputs for a wider range of policy options. The pipeline, or a simplified version of it, should also be more easily shareable to allow for greater scrutiny outside NHSE.

Ref Page 26, 2.20 · Implemented Q4 2025/26

Recommendation 10: Ahead of producing the next version of the models, NHSE should review measures of productivity, and other workforce capacity models, to ensure that it is applying the most appropriate measure for its modelling.

Ref Page 36, 3.15 · Implemented Q4 2025/26

Recommendation 11: NHSE's modellers should consider whether it is reasonable to expect the same rate of increasing productivity from a workforce increasingly composed of newly qualified staff.

Ref Page 36, 3.15 · Implemented Q4 2025/25

Recommendation 12: Future versions of the modelling would be improved by greater consultation and by taking better account of the factors that could limit the expansion of education and training, and of any reductions in service that might result from rapid increases in the amount of on-the-job training.

Ref Page 40, 3.27 · Implemented Q4 2025/26

Recommendation 13: NHSE's modelling should incorporate more realistic assumptions on the minimum level of international recruitment of doctors in future.

Ref Page 41, 3.31 · Implemented Q4 2025/26

Recommendation 14: Given the long time frames required for medical training, NHSE should stress test their assumptions on training places for medical students by running their modelling over a longer time period than that covered by the LTWP, to ensure that assumptions about training places do not create long-term risks of excess supply of doctors.

Ref Page 41, 3.31 · Implemented Q4 2025/25

Recommendation 15: Ahead of the next version of the LTWP, NHSE modellers should consult further with a wide range of stakeholders, including Royal Colleges, healthcare providers and other health bodies about the appropriate substitution rates to use when meeting demand for one type of healthcare worker by employing another and should attempt to understand the real-world workload effects on general practice of changes in roles in primary care. NHSE should also be clear about how any proposed changes in skill mix fit with its wider agenda of progressively moving care out of hospital settings and into the community.

Ref Page 44, 3.36 · Implemented Q4 2025/25

Recommendation 16: NHSE analysts should document and share their long-term assumptions, for each iteration of the LTWP, within NHSE and with HM Treasury and DHSC. NHSE analysts across the organisation should use common assumptions where possible.

Ref Page 44, 3.37 · Implemented Q4 2025/26

Recommendation 2: NHSE should improve documentation across the whole of the modelling pipeline to make it easier for future analysts to understand the modelling and any limitations, reproduce model outputs, and update the model when producing future versions.

Ref Page 28, 2.28 · Implemented Q2 2025/26

Recommendation 3: Given NHSE's commitment to update the LTWP at regular intervals, it should assign responsibility for understanding the entire modelling pipeline to a small team and ensure that there is sufficient knowledge of the Python and triangulation models available in the team. NHSE should complete a succession plan for the modelling.

Ref Page 28, 2.28 · Implemented Q2 2025/26

Recommendation 4: NHSE should ensure quality assurance practices take place in a timely manner, so analysts have sufficient time to respond accordingly.

Ref Page 29, 2.31 · Implemented Q4 2025/26

Recommendation 5: NHSE should revisit internal quality assurance arrangements for existing models when they are used for a new purpose, such as the models used to provide input data for the workforce modelling, and ensure independent scrutiny is evidenced accordingly.

Ref Page 29, 2.31 · Implemented Q4 2025/26

Recommendation 6: In the next version of the model, NHSE should communicate outputs for a wider range of plausible future scenarios. NHSE should consider different methods and processes which would allow it to better capture uncertainty in its modelling.

Ref Page 31, 3.3 · Implemented Q4 2025/26

Recommendation 7: Modelling of this kind requires gathering assumptions about the future of the NHS in one place. This presents an opportunity to expose those assumptions widely to scrutiny and challenge, both internally and externally. Assumptions should be generated in transparent and systematic consultation with external stakeholders. NHSE should make more of this opportunity the next time it repeats its modelling.

Ref Page 3, 3.3 · Implemented Q4 2025/26

Recommendation 8: In future versions of the modelling, NHSE should model and communicate a wider range of different scenarios for the demand NHS services may face. In our view, this could include scenarios in which different amounts of total funding are available, and upside and downside scenarios for demand for particular services.

Ref Page 32, 3.7 · Implemented Q4 2025/26

Recommendation 9: The modelling (at 0.8% per annum productivity increase) and the LTWP (which describes 1.5% to 2% per annum productivity increase) present productivity differently. In future, NHSE should more clearly explain both internally and in public how it has calculated different measures of productivity and how these relate to the National Statistics produced by the ONS.

Ref Page 35, 3.15 · Implemented Q4 2025/26