Improvement, resources and public value

Why this matters

NHS teams are repeatedly asked to improve, but activity is often mistaken for effect. A new form, meeting, policy or training session may be completed without showing whether care became safer, timelier or more equitable. The pressure to demonstrate action can produce long lists of tasks that never test the original problem.

Improvement leadership begins with a patient or service need, develops an explanation of the present system, tests a change and examines consequence. It also acknowledges that time, attention, workforce and money are limited. A change that succeeds only by moving burden to another team is not yet a reliable improvement.

Begin with the problem, not the favoured solution

Describe who experiences the problem, what happens, how often, where and with what consequence. Observe the work and speak to the people who use and deliver it. The process in policy may differ from the process in practice. Avoid defining the problem as absence of your preferred answer, such as 'staff have not completed the new template'.

A useful problem statement might say that important escalation information is missing from weekend handovers in a defined proportion of cases, creating delay and repeated clarification. This leaves several possible causes and responses open. It also provides a starting point for measurement.

Audit, assurance, research and improvement

Audit compares practice with an explicit standard. Assurance asks whether controls and evidence give confidence that responsibilities are being met. Research seeks generalisable knowledge through an appropriate design. Improvement uses disciplined local testing to make a process better. One project may use more than one method, but the claim should match the method.

A before-and-after result does not automatically prove that the intervention caused the change. An audit action plan is not the same as iterative testing. State what the work can and cannot establish, and seek research or governance advice when the boundary matters.

Write an aim and give measures different jobs

A useful aim identifies the group or process, the desired change, the amount and the time period. It should be ambitious enough to matter and specific enough to test. Measures then perform different jobs: an outcome measure asks whether the result improved, a process measure checks whether the change was carried out, and a balancing measure looks for unintended harm or displaced burden.

For a handover project, completeness may be a process measure, delay in appropriate escalation may be an outcome, and time taken to prepare the handover may be a balancing measure. Counting staff trained demonstrates reach, not necessarily improvement.

Learn from variation and small tests

A single monthly average can hide important variation between wards, times or patient groups. Review data over time and understand whether change exceeds ordinary fluctuation. Combine numbers with accounts from patients and staff to understand why the pattern occurs.

Small tests are disciplined, not timid. Predict what will happen, test with a manageable group, review the result and adapt before wider implementation. Begin small enough to learn safely but not so small that the test cannot reveal the operational problem. Record what was changed between cycles.

Implementation is part of the intervention

A change does not exist in practice merely because it was approved. Staff need information, capability, time, equipment and a route to resolve exceptions. Digital access, night and weekend work, temporary staff and cross-boundary teams should be considered from the start.

Identify who owns the change after the project team leaves. Build review into routine governance and decide what should trigger correction or withdrawal. Spread should preserve the function that produced benefit rather than copying every visible feature without understanding context.

Measure equity and burden

Average improvement can coexist with widening inequality. Examine access, experience and outcome across relevant groups and places. Ask who must do additional work, travel further, use digital access or rely on unpaid family support. Include these consequences in the design and measures.

Staff capacity is a patient-safety resource. A process that saves organisational money by adding unmeasured clinical administration may reduce time for assessment or supervision. Improvement should name burden and test whether it is proportionate.

Resources, priorities and public value

Resource includes money, beds, equipment, estate, data, time, attention and workforce capability. Cost, saving and affordability are different. A cheaper option may be unaffordable within the current budget, and a promised saving may not release cash if the underlying capacity remains required.

Every choice has an opportunity cost: the benefit that cannot be pursued because resources are used here. Public value asks whether the decision produces worthwhile outcomes fairly and legitimately, not simply whether one department meets its target. Make trade-offs visible and identify who bears them.

Scale, sustain and review

Before scaling, confirm that the change produced benefit, that the important mechanism is understood and that the receiving settings have the conditions needed. A pilot supported by exceptional enthusiasm may not survive routine workload. Test the assumptions most likely to fail at scale.

Sustainability requires ownership, capability, resources, measurement and permission to correct. Review whether the original problem still matters and whether the intervention remains the best response. Improvement is a learning system, not a permanent requirement to defend yesterday's solution.

• Define the problem from patient and staff experience.
• Choose outcome, process and balancing measures.
• Test on a scale that can produce useful learning.
• Name equity, burden and opportunity cost.
• Assign long-term ownership and a review trigger.