From insight to impact: proving that an improvement changed care

Define the problem before selecting the solution

“Improve communication” is too broad to test. An illustrative service might instead identify that relatives cannot find out who is responding to a concern. The problem is then a missing ownership and acknowledgement process, not simply the absence of another communication policy. Ask the people affected what a useful response would look like before designing it.

Gather enough information to distinguish an isolated event from a recurring weakness. Review the context and any immediate safety implications first. Longer-term investigation must not delay urgent protection or required reporting. Where the facts remain disputed, record the uncertainty and seek appropriate professional support rather than force a convenient explanation into an action plan.

Understand the conditions that produced the event

An employee may have made an error, but that does not explain why the system failed to prevent or detect it. Examine the information available, task design, workload, supervision and equipment. These are suggested lines of enquiry. Their relevance depends on the event, and they should not be used to excuse deliberate misconduct or to predetermine responsibility.

Repeated training is useful when a skill gap has been established. It is a weak default response when workers already understand the task but cannot complete it within the available conditions. An improvement plan should explain why the selected change addresses the evidence found. Otherwise the same problem can return under a newly signed training record.

Choose measures people can recognise

For the illustrative communication problem, a provider might examine acknowledgement time, whether the named contact actually responded and whether relatives understood the next step. Counting the number of messages sent would measure activity rather than whether the communication helped. Include the people receiving care where the concern relates to their preferences or experience.

Establish a baseline before changing the process where it is safe and practicable. Explain the sample and period, including missing information. Our editorial recommendation is to use a small number of measures that can be maintained, alongside qualitative accounts. A complicated scorecard that employees cannot update reliably creates another source of uncertainty rather than stronger evidence.

Test unintended effects alongside the intended benefit

An intervention that speeds one task may displace work elsewhere. A faster electronic form might reduce documentation time but make important detail harder to find. A fixed telephone update schedule may help some families while excluding those unable to answer at that time. Ask who benefits, who does not and whether the change creates new burdens.

Innovation should therefore include a review point and a route to adjust or stop the change. Agree who will decide and what information they need. An initial favourable response is useful, but it does not establish sustained benefit across all users. Avoid presenting a short trial as a proven solution without acknowledging its scope and the uncertainty that remains.

Keep learning after the action is marked complete

NICE’s guide includes continuing the work after implementation. Care Circle suggests a separate outcome review when sufficient experience has accumulated, with the interval chosen for the issue rather than a universal thirty-day rule. Ask whether the original concern recurred and whether the change remained usable across shifts and service conditions.

Share a concise account of the problem, attempted response, results and limits. Another service may adapt the lesson, but it should not copy a process without checking its own circumstances. Documenting a failed intervention can be valuable if it explains what was learnt and prevents the same assumption elsewhere. The purpose is better judgement and care, not an uninterrupted collection of success stories.

Distinguish implementation failure from an ineffective idea. A proposed response may be reasonable but never reach the staff or people expected to use it. Before abandoning it, establish what was actually introduced, how consistently it operated and whether users understood it. Conversely, where a change was delivered as intended but produced no useful result, avoid repeating it simply because time has already been invested. Keeping these possibilities separate helps the next decision address the actual weakness rather than protect the original plan.