Quality improvement
Test a clear change against meaningful measures and look for unintended effects.
Quick revision summary
Define the aim and a useful set of measures
Define a specific aim for a population and time frame. An outcome measure describes the result for patients or the service; a process measure shows whether the intended action happened; a balancing measure identifies a possible adverse consequence elsewhere.
Use operational definitions and data over time
Specify the population, intended improvement and time frame. Define each measure so another person would collect it the same way.
Test, adapt and make ownership explicit
A plan–do–study–act cycle starts with a prediction and a manageable test. Compare what happened with what was expected, then adopt, adapt or abandon the change.
Common mistake
An apparent improvement in one measure may conceal a harmful change elsewhere.
Improvement that demonstrates a better service
Test a clear change against meaningful measures and look for unintended effects.
Define the aim and a useful set of measures
Define a specific aim for a population and time frame. An outcome measure describes the result for patients or the service; a process measure shows whether the intended action happened; a balancing measure identifies a possible adverse consequence elsewhere. For example, faster referral handling may improve time to triage, but could increase incomplete referrals or workload if implemented badly. Collect a baseline and record operational definitions so measurements remain comparable.
Plan–do–study–act cycles test a change on a manageable scale before wider adoption. Study what happened against the prediction and modify the next test. A before-and-after percentage alone may be affected by case mix or ordinary variation. Plot observations over time and annotate changes where useful. Agree who owns the new process, training and ongoing monitoring so a successful test survives rotation or staff change. Share limitations as well as improvements.
Key distinctions in practice
| Finding or option | What it means | Clinical consequence |
|---|---|---|
| Outcome measure | The result that matters | Examples include time to care or symptom improvement. |
| Process measure | Whether the change was performed | Check reliable delivery, not just policy existence. |
| Balancing measure | A potential trade-off | Look for displaced workload or missed care. |
| PDSA cycle | A small planned test and learning step | Compare prediction with evidence and adapt. |
Use operational definitions and data over time
Specify the population, intended improvement and time frame. Define each measure so another person would collect it the same way. For a referral project, “complete referral” should name the required information and denominator rather than depend on the reviewer’s impression. Record a baseline and consider whether case mix or changing workload affects comparison.
Outcome measures describe what matters to patients or the service; process measures show whether the change was delivered; balancing measures look for adverse consequences elsewhere. Plot data over time where appropriate and annotate interventions. A single favourable week may be ordinary variation, and a before-and-after comparison cannot automatically establish causation.
Test, adapt and make ownership explicit
A plan–do–study–act cycle starts with a prediction and a manageable test. Compare what happened with what was expected, then adopt, adapt or abandon the change. Involve the people who use the process, including patients when relevant, because an apparently simple improvement can create work or barriers elsewhere.
For a successful change, decide who maintains training, supplies and measurement. Include a handover plan for staff turnover and review whether the benefit persists. Distinguish improvement from data collection: an audit identifies a gap against a standard, while completing its cycle requires action and remeasurement. Report limitations and balancing effects as clearly as the favourable result.
A distinction to remember
An apparent improvement in one measure may conceal a harmful change elsewhere.
A new referral form improves completion rates but delays urgent calls. How should the project respond?
Reveal the reasoning
The balancing measure reveals harm. Adapt the process to preserve prompt urgent communication and reassess overall benefit before wider adoption.
Guidance to read alongside this chapter
Test your recall
What does a balancing measure detect?
Show answer
Unintended consequences of a change.
