A clinic announces that discharge paperwork is now completed 30 minutes faster. That sounds like improvement, but patients are calling the next day because they did not understand the plan, and nurses are spending extra time correcting mistakes. Healthcare quality improvement asks what actually became better for patients—and whether the faster process harmed another part of care.
The gold standard of healthcare quality improvement connects a meaningful patient or service aim to safe change, reliable measurement, patient participation and ongoing learning. This guide covers eighteen controls, two original fictional cases, twelve FAQs and the existing Gold Standard network of health, evidence and human factors.
This subject belongs to The Gold Standard. It is an educational quality framework, not clinical accreditation, approval of a workflow change or an individual treatment recommendation.
Quick information: the six quality-improvement questions
- What outcome matters to the patient or service?
- Who is eligible and which denominator measures the problem?
- Which process change is supported by evidence and clinical authority?
- How will outcome, process and balancing measures be tracked?
- Who could be missed or harmed by the redesign?
- What observations justify sustaining, adapting or stopping the change?
Chapter index
- 1. Begin with the patient benefit that is missing
- 2. Keep quality improvement and clinical research distinct
- 3. Use WHO’s quality domains as a shared vocabulary
- 4. Define the population and the denominator before data collection
- 5. Listen to patients, families and frontline teams
- 6. Map the actual care process and handoffs
- 7. Write an improvement aim that can be tested
- 8. Choose outcome, process and balancing measures
- 9. Check data definitions, quality and missingness
- 10. Distinguish statistical variation from meaningful change
- 11. Test interventions in small safe cycles
- 12. Keep changes compatible with professional care standards
- 13. Check equity and access instead of averages alone
- 14. Measure unintended burden on staff and services
- 15. Evaluate harm and safety events with honest interpretation
- 16. Distinguish implementation from reliable scale-up
- 17. Preserve audit trails, corrections and accountability
- 18. Transfer learning to a new clinical context carefully
1. Begin with the patient benefit that is missing
Healthcare quality improvement (QI) seeks to make the delivery of appropriate care safer, more effective, people-centred, timely, equitable, integrated and efficient. A completed checklist or a shorter queue is useful only when it supports the care people need. A system can become faster while producing worse handovers or increasing the workload of already stretched staff.
Start with a clearly defined service gap that patients and professionals recognise. Name the eligible population, care setting, timescale and desired improvement. The gold standard is a measurable gain supported by evidence and balanced against possible harm, not a dashboard designed to make a project look successful.
2. Keep quality improvement and clinical research distinct
A local improvement initiative may test how existing evidence-based care is delivered, while a clinical study can ask whether a treatment itself is beneficial or safe. These activities overlap but have different governance, ethical and causal questions. A successful local workflow change does not necessarily establish a new treatment’s general efficacy.
Determine what is being changed and who has authority to approve it. Involve clinical governance, patient safety and appropriate ethical review before work that may require it. A QI project should never bypass the rules for clinical research or alter treatment outside professional authorisation.
3. Use WHO’s quality domains as a shared vocabulary
WHO describes high-quality services through dimensions including safety, effectiveness, people-centredness, timeliness, equity, integration and efficiency. These aims can conflict in practical operations, so an initiative should specify which dimension it targets and which others must not deteriorate.
For example, shorter discharge times are not automatically a gain if patients leave without appropriate information or follow-up. A multidisciplinary team can use the domains to surface competing goals. The final assessment should state the desired improvement and the conditions under which it is acceptable.
4. Define the population and the denominator before data collection
An appointment completion rate may count appointments, unique patients, clinically eligible cases or all referrals. Those denominators are different. Changing the definition between baseline and follow-up can produce an apparent improvement that never happened in actual care.
Write inclusion rules, exclusions, observation period and data source in advance. Confirm whether repeat visits by one person appear once or many times. A denominator that omits hard-to-reach patients can make performance look better precisely by excluding those who need better access.
5. Listen to patients, families and frontline teams
Staff performing a care process often know where it breaks, while patients and carers can identify barriers invisible in operational records. A discharge letter may be generated on time but arrive in language the recipient cannot understand. A reminder system may reach only people with reliable phone access.
Use respectful interviews, feedback and process observation with appropriate consent and privacy safeguards. Invite affected people to help define useful outcomes, not just review a finished project. Practical experience can reveal which redesign is feasible and which creates additional burden.
6. Map the actual care process and handoffs
Care often passes between reception, assessment, laboratory services, clinicians, pharmacy and follow-up. Delays and mistakes may arise at transitions rather than within one department. A process map should include decisions, information requirements, expected responsibility and conditions where work stops.
Observe what happens during busy shifts and incomplete information, not only under ideal conditions. A handover form may be correctly completed but still fail to convey an urgent pending test. Distinguish paperwork compliance from actual transfer of responsibility and clinical understanding.
7. Write an improvement aim that can be tested
The Institute for Healthcare Improvement’s Model for Improvement asks what the team is trying to accomplish, how it will know a change is an improvement, and what changes might result in improvement. A useful aim defines a population, measurable quantity and time frame.
An aim such as ‘improve patient experience’ is too broad for direct measurement. A stronger aim identifies an aspect of care and an appropriate measure, while preserving safety and equity. Goals should be ambitious enough to matter yet grounded in available services and professional standards.
8. Choose outcome, process and balancing measures
IHI distinguishes outcome measures, which indicate the impact on people or the service; process measures, which show whether intended steps are being delivered; and balancing measures, which reveal unwanted effects elsewhere. A team measuring only process compliance cannot reliably infer improvement in health outcomes.
Choose a small set directly tied to the aim and likely failure modes. For faster patient discharge, track relevant follow-up completion and safety concerns alongside process turnaround. Explain which measures are direct outcomes, proxies or operational indicators, and keep their denominators and dates visible.
9. Check data definitions, quality and missingness
Healthcare information systems combine records from different people, workflows and technologies. Duplicate encounters, missing timestamps, coding changes and delayed updates can affect an indicator. Data quality is not improved by excluding inconvenient records without explanation.
WHO’s 2025 guide on measuring quality of maternal, newborn, child and adolescent services emphasises matching indicators to aims, strengthening health information and assessing data quality. Its service emphasis should be respected rather than presented as a complete universal checklist for every specialty. Keep original records and a traceable correction process.
10. Distinguish statistical variation from meaningful change
A measure can fluctuate from week to week even when the underlying process has not changed. An increase after one improvement meeting may reflect ordinary variation, seasonal case mix, changed staffing or a real intervention effect. A single before-and-after percentage is often weak evidence.
Display observations across time with adequate denominators, identify changes to definitions and investigate outliers. Run charts and other suitable methods can help teams notice process patterns, but the interpretation still depends on the data-generating system. Do not label every favourable point a breakthrough.
11. Test interventions in small safe cycles
The IHI Plan–Do–Study–Act cycle involves planning a limited change, trying it, studying what happened and deciding how to adapt. Small tests can reveal problems before an intervention is spread widely. They require a prediction and planned observation, not just a change made quickly.
Choose test conditions that are safe, authorised and appropriate to the care context. Record what was planned, what actually happened and what staff and patients experienced. A failed test can be valuable evidence; suppressing it leads others to repeat the same error at greater scale.
12. Keep changes compatible with professional care standards
An operational team may be tempted to remove checks that appear to slow work. Some are unnecessary duplication; others are critical patient-safety barriers. Determining the difference requires clinical and systems expertise, evidence, and applicable standards of care.
A QI method does not authorise new clinical decisions or medications. Establish what changes are permitted, how oversight works and how unexpected incidents escalate. Protect staff’s ability to stop and raise safety concerns. Speed without safe clinical governance is not success.
13. Check equity and access instead of averages alone
Average improvements can hide widening gaps for people with disability, limited language access, long travel distances or fewer resources. A new digital booking tool may reduce overall queue time while excluding people who cannot use it. A high overall percentage can coexist with persistently poor outcomes in a subgroup.
Break down performance by relevant access factors when lawful, useful and privacy-preserving. Compare participation, not just results among those who reached the service. WHO’s quality dimensions include equity; responsible measurement asks whose experience improved and whose did not.
14. Measure unintended burden on staff and services
Moving work from one team to another can make one department’s performance look better without improving the entire pathway. Nurses may spend additional time fixing discharge errors, or a streamlined scheduling rule may produce excessive follow-up calls. These balancing effects matter for sustainability and patient outcomes.
Track relevant workforce workload, handover failures, duplicate tasks and downstream demand. Staff feedback is meaningful evidence, though it should be combined with quantitative records where available. A change that exhausts the people delivering care can be difficult to sustain even if a short pilot looks favourable.
15. Evaluate harm and safety events with honest interpretation
A count of reported incidents can rise because the system became safer to report into, not necessarily because care became more dangerous. Conversely, fewer recorded adverse events can reflect underreporting. Safety measurement needs definitions, exposure denominators, reporting culture and appropriate clinical review.
Distinguish actual events from near misses and potentially preventable harm. A QI project should create a route for incident review and correction without turning reports into blame statistics. Do not infer that a change is safe solely because a short trial recorded no adverse events.
16. Distinguish implementation from reliable scale-up
A practice that works with one motivated team may fail across different shifts, facilities or patient populations. Scaling involves staffing, training, equipment, leadership, feedback and the ability to adapt without undermining the underlying mechanism. A training attendance count does not prove competent use on a busy ward.
After repeated tests, document which components must remain consistent and which may be adapted. WHO’s clinical quality-improvement resources encourage practical, data-informed improvement at the point of care. Monitor after implementation rather than declaring permanent success on launch day.
17. Preserve audit trails, corrections and accountability
Another team should be able to reconstruct the project aim, baseline, interventions, dates, staff responsibilities, measures, patient feedback, safety reviews and revisions. Statistical analyses should not be rewritten quietly when a definition changes. Corrections and unexpected results are part of the evidence.
A useful quality report is neither a wall of numbers nor a success story without limitations. It explains what improved, what remains uncertain, what was not tested and who has responsibility for continued monitoring. Protect patient privacy while preserving enough traceability for authorised audit.
18. Transfer learning to a new clinical context carefully
An improvement programme in one hospital cannot be copied unchanged into a rural clinic or a different specialty. Case mix, service resources, communication, staff skills and patient needs differ. The transferable method is to define a meaningful aim, map the workflow, test safe changes, evaluate and revise.
The gold-standard outcome is a care system that learns, protects people and improves their experience and health—not simply one that produces more reports or higher checklist-completion percentages.
Worked fictional case A: process improvement with a balancing problem
At a fictional clinic, 210 of 300 eligible discharges include an appropriately documented handover before the initial improvement period ends. The process rate is 210 ÷ 300 = 70%. In a later equal-sized invented period, 255 of 300 meet the handover process definition, a rate of 85%. The observed change is +15 percentage points. The numbers describe a process, not proof of better health outcomes.
Suppose unplanned clarification calls recorded during the same equal-sized periods increase from 20 to 39. Those rates are 6.7% and 13.0% of the period-specific discharge counts, assuming one defined event per discharge. The increased call burden may indicate a problem or an improvement in access to clarification; it requires review of content and appropriateness. A team cannot declare full success from the handover percentage alone.
Worked fictional case B: response bias in a patient survey
An invented post-visit survey is offered to 200 eligible patients, but only 80 return usable answers. Among respondents, 64 say the information they received was clear, yielding 64 ÷ 80 = 80% positive responses. Among all 200 offered patients, only 64 ÷ 200 = 32% are known to have reported clarity. The remaining 120 people’s views are unknown, not automatically negative or positive.
A better evaluation examines response patterns, accessibility, language, and whether those surveyed represent the full eligible service. It separates perceived clarity from measured understanding and later care outcomes. These invented figures do not establish any actual clinic’s performance.
Healthcare QI evidence and safety audit
- Define the care aim, population, dates and governance.
- Preserve baseline and follow-up measures with matching denominators.
- Record interventions and small tests in sufficient detail to repeat.
- Include patient experience and clinical safety review.
- Track process, outcome and balancing measures separately.
- Examine missing data, case mix, variation and equity.
- Keep adverse events and near misses distinguishable.
- Document revisions, accountable decision-makers and follow-up plans.
Frequently asked questions
What is healthcare quality improvement?
A structured effort to make delivery of appropriate care safer, more effective, timely, equitable and people-centred.
Is healthcare QI the same as medical research?
Not always. QI often studies local care delivery, while research may test generalisable interventions or treatments; governance requirements differ.
What is PDSA?
Plan–Do–Study–Act: a cycle for testing and learning from carefully defined changes.
What is an outcome measure?
A measure of the result experienced by patients or the relevant care system.
What is a process measure?
A measure of whether intended steps in the care pathway were completed as designed.
What is a balancing measure?
A measure checking whether an improvement in one part creates problems elsewhere.
Does higher checklist compliance always mean safer care?
No. It is one process indicator and must be related to meaningful outcomes and potential harms.
Why use run charts?
They show performance over time and help distinguish a pattern from a single favourable observation.
Can QI justify removing clinical safety checks?
Not without appropriate evidence, authority and clinical governance.
Why involve patients?
They can reveal obstacles and outcomes that operational records or staff assumptions miss.
What if a pilot produces worse results?
The team should investigate, adapt or stop the change appropriately and report what was learned.
What makes results transferable?
Clear definitions, reproducible interventions, comparison of different contexts and continued safety monitoring.
Connected Gold Standard reading across health and systems
Health informatics supplies data provenance, biostatistics helps evaluate uncertainty, health psychology identifies behaviour and access barriers, human factors examines workflow design, and health economics addresses the resource consequences. No one discipline substitutes for clinical governance.
- The Gold Standard of Health
- The Gold Standard of Health Informatics
- The Gold Standard of Health Communication
- The Gold Standard of Health Psychology
- The Gold Standard of Epidemiology
- The Gold Standard of Biostatistics
- The Gold Standard of Human Factors
- The Gold Standard of Health Economics
Official sources and scope of this guide
- WHO — Quality of Care Overview
- WHO — Quality Health Services Toolkit
- WHO — Measuring and Monitoring Quality of Care (2025)
- WHO — Clinical Quality Improvement Programme (EQuIP)
- IHI — Model for Improvement
- IHI — Establishing Outcome, Process and Balancing Measures
- AHRQ — Plan–Do–Study–Act Cycle
WHO, IHI and AHRQ sources were reviewed on 29 October 2026. The WHO 2025 measurement guide specifically focuses on maternal, newborn, child and adolescent health services and provides transferable measurement principles, not a universal clinical protocol. Every patient count, rate and outcome in the worked cases is fictional. Actual service changes require competent local clinical and safety review.
Return to chapter index · The Gold Standard hub · How X Works