Implementation, sustainability and spread

• Distinguish testing from implementation and implementation from spread.
• Identify the infrastructure needed to make a change routine and reliable.
• Design measures for sustainability and detect decay after project attention ends.
• Plan spread while adapting to context and preserving essential functions.
• Recognise when leadership, resources or policy barriers require escalation rather than more local testing.

Many projects show short-term gains while an enthusiastic team is watching. Sustainability asks whether the new way of working persists when attention, personnel and pressures change. The change must become part of the operating system rather than remain a project.

Implementation therefore involves roles, workflows, documentation, technology, training, induction, supplies, escalation and accountability. IHI describes implementation as a permanent change in how work is done, usually after testing under varied conditions.

A sustainability plan should begin before the project closes. If nobody owns the measure or process after the QI team disbands, deterioration is predictable.

Before spread, distinguish the change's essential function from its local form. The core function is what must be preserved for the mechanism to work; the form is how a particular team delivers it.

For example, the core function may be “a senior clinician reviews escalation decisions before transfer”, while one site uses a daily huddle and another uses a digital queue. Insisting on identical form can make spread fail; allowing the core function to disappear makes the intervention meaningless.

Document which components are fixed, adaptable or still uncertain.

Reliable implementation often requires removing the old pathway, changing defaults, updating policies, embedding prompts in digital systems, adding training to induction and ensuring materials are available. Education alone rarely creates sustained reliability where the workflow itself remains unchanged.

Assign process ownership. The owner is responsible for monitoring and response, not necessarily for doing every task. Define what happens when performance deteriorates and who has authority to act.

Where the change crosses organisational boundaries, responsibility must be agreed rather than assumed.

Continue a small set of meaningful measures after implementation. The frequency can reduce as reliability improves, but it should be sufficient to detect drift. Run charts or control charts can reveal whether gains persist.

Include balancing measures if the risk remains relevant. A faster discharge process should still watch readmissions, medication problems or patient experience if these could worsen.

Plan review points at 3, 6 and 12 months or at intervals appropriate to the process. Sustainability is demonstrated by performance over time, not by the existence of a new policy.

Common causes include staff turnover, competing priorities, workarounds, removal of project resources, technology changes, unclear ownership, policy conflict and unrecognised workload. Some interventions depend on a charismatic champion and fail when that person leaves.

Use failure as information about the system. If a change cannot survive ordinary operational conditions, the implementation design is incomplete.

Build resilience by simplifying the process, aligning it with existing work, creating visible feedback and ensuring that staff understand the purpose as well as the steps.

Spread requires readiness assessment in the receiving setting. Map the new context, identify local stakeholders, compare workflows and test the change again at small scale. Do not assume that evidence from one site removes the need for local learning.

Use a spread plan with target sites, sequence, support, measures and feedback. Early adopters can help, but do not mistake enthusiasm for representativeness. Later sites may face different constraints.

Preserve the core function while adapting form. Track whether adaptation changes outcomes.

Some barriers cannot be solved by frontline ingenuity. Staffing levels, procurement, information systems, cross-boundary contracts and policy may require senior decisions. An improvement team should know when to escalate rather than endlessly redesign around a structural constraint.

Leaders support improvement by creating psychological safety, aligning priorities, providing data and resources, removing barriers and avoiding contradictory targets. They should also protect teams from “project overload”.

Implementation fails when governance asks teams to deliver a change without authority, time or infrastructure.

A change that improves the average can widen inequality. Monitor access, uptake, outcomes and burden across relevant groups. Digital pathways, language, transport, disability, socioeconomic position and rurality may alter who benefits.

Involve affected communities in adaptation. The aim is not merely to offer the same process to everyone but to achieve fair opportunity and outcomes.

If a subgroup is disadvantaged, respond during implementation rather than describing the inequity as an incidental limitation.

A hospital tests pharmacist-led medication reconciliation on one ward and reduces discrepancies. Implementation then requires staffing cover, role definitions, an electronic field, weekend contingency, induction training and monitoring. When spread reaches a surgical ward, the timing and staff mix differ, so the pathway is retested rather than copied unchanged.

Three months later, the process measure begins to fall on weekends. The team investigates and finds that locum pharmacists cannot access the electronic template. Fixing access restores performance.

This is sustainability work: detecting and correcting decay, not declaring the original project complete.

List the elements that must remain in place for your change to work. Identify the owner, monitoring measure, trigger for action, training/induction requirement, digital or policy dependency and contingency for staff turnover.

Then choose one potential spread site and identify what is likely to differ. Plan the first local test. Use the Stage 4 Sustainability and Spread worksheet to convert these decisions into an operational plan.

• Assuming education alone will hardwire a new process.
• Stopping measurement as soon as the project reports success.
• Copying a change unchanged into a different context.
• Relying on one champion without process ownership.
• Using further PDSA cycles to compensate for barriers that require executive action.

1. Institute for Healthcare Improvement (IHI): Model for Improvement; Testing Changes; Establishing Measures; Implementing Changes. https://www.ihi.org/library/model-for-improvement
2. NHS England: How to improve care-related processes in general practice; Model for Improvement and change principles. https://www.england.nhs.uk/long-read/how-to-improve-care-related-processes-in-general-practice/

Educational reference. Verify current local governance requirements before starting a project.