Quality Improvement, Clinical Audit and Research in Healthcare
Building the project: aims, teams, charters and theories of change
How to convert a recognised problem into an executable project with a measurable aim, appropriate team, clear scope, charter and explicit theory of change.
Learning objectives
• Write a specific improvement aim that describes who, what, where, how much and by when. • Define scope and exclusions without making the project artificially easy. • Build an interprofessional team around the process and assign meaningful roles. • Use a project charter and driver diagram/theory of change to connect interventions to outcomes. • Plan sponsorship, governance, resources and ownership before launch.
Learning objectives
- Write a specific improvement aim that describes who, what, where, how much and by when.
- Define scope and exclusions without making the project artificially easy.
- Build an interprofessional team around the process and assign meaningful roles.
- Use a project charter and driver diagram/theory of change to connect interventions to outcomes.
- Plan sponsorship, governance, resources and ownership before launch.
Aims that guide decisions
A useful aim gives the team a destination. “Improve discharge communication” is too vague. “Increase the proportion of eligible medical discharges from Ward A with a reconciled medicines list reaching the GP by 18:00 on the day of discharge from 62% to 90% by 31 March, without increasing pharmacy overtime” is more useful. It identifies the population, process, target, time and an important balancing concern.
Targets need a rationale
Targets can come from evidence-based standards, statutory requirements, organisational priorities, benchmark performance or a locally agreed improvement ambition. Avoid choosing 100% reflexively. Some measures have legitimate exclusions or require professional judgement. Record why the target is appropriate.
Scope and boundaries
Define included settings, patient groups, shifts, process start/end points and exclusions. Scope should make the project manageable, not manipulate the denominator. A pilot may begin on one ward, but the team should state whether the long-term intention is local improvement or wider implementation.
Build the team around the work
Include people who perform and receive each key step, not only project enthusiasts or senior leaders. Identify a project lead, sponsor, front-line representatives, data support, patient/service-user input and relevant subject/governance expertise. Clarify who can authorise tests, who collects data, who can remove barriers and who will own the process after the project.
Project charter
A one-page charter should include problem, aim, scope, team, measures, change theory, governance route, risks, milestones and post-project ownership. The charter is a live agreement, not a bureaucratic form. Revisit it when the problem, scope or evidence changes.
Driver diagrams and theories of change
A driver diagram links the aim to the conditions that must change (primary drivers), more specific contributing processes (secondary drivers) and testable change ideas. It forces teams to explain why an intervention should influence the outcome. A change idea with no credible link to the aim is a weak candidate for testing.
Worked example: reducing omitted medicines
Aim: reduce clinically significant omitted time-critical medicines on a community hospital ward from 11 per 1,000 administration opportunities to fewer than 4 by 30 June. Primary drivers: reliable identification of time-critical medicines; timely supply; clear escalation when doses are unavailable. Change ideas include an electronic flag, pharmacy stock review and an escalation prompt. Measures include omission rate (outcome), percentage flagged correctly (process), and urgent pharmacy calls/out-of-hours workload (balancing).
Project start-up meeting
Use the first meeting to agree the problem and aim, not to allocate data collection immediately. Walk through the current process, identify missing perspectives, agree governance, choose a minimum useful measure set, identify one small testable change, and schedule the first learning review. End with named owners and dates.
