Outcome, process and balancing measures in nursing QI help you decide whether a local change improves care, whether the intended work happens and whether another part of the service is affected. A useful measurement plan connects these questions with practical definitions, reliable data and a clear aim.
This guide focuses on designing that plan. It includes a worked nursing example, a measure specification table and checks for interpreting results over time. For the complete project structure, use our nursing quality improvement project guide.
Outcome, process and balancing measures in nursing QI: quick answer
An outcome measure describes the result you want to improve. A process measure tracks whether an important action is performed as intended. A balancing measure checks for consequences elsewhere in the system, such as extra workload or delayed access.
Choose a small, purposeful set linked to the project aim. Define exactly what each measure counts, who is eligible, how often data will be collected and how the result will be reviewed. A label such as “better documentation” is too vague to guide consistent collection.
The Institute for Healthcare Improvement describes these three types as a family of measures. The combination gives a more useful account of change than a single favourable percentage (IHI, n.d.-a).
What makes a measure an outcome?
An outcome connects with the result named in the aim. It might concern patient experience, access, a safety event or the timeliness of a service response. The choice should reflect what matters in the particular setting.
Suppose a community team wants patients to understand their follow-up arrangements. Counting how many leaflets were distributed describes an activity. Assessing whether patients can correctly identify their next contact arrangement comes closer to the intended result.
However, no single measure captures every aspect of care. A brief understanding check may depend on its wording, timing and administration. Explain what the measure can establish and what remains outside its scope.
Do not choose an outcome simply because it is already easy to extract. First establish whether it answers the aim. Then decide whether the information can be collected accurately and proportionately during the project.
What makes a measure a process?
A process measure describes an action believed to contribute to the result. It helps explain whether the change was delivered. For a discharge communication project, this might be the proportion of eligible consultations using a planned teach-back prompt.
The measure must specify what counts as completion. A tick box saying “education given” may not show that the intended communication steps occurred. A clearer definition identifies the core action and the evidence accepted for recording it.
Process data can help explain an unchanged outcome. If the new approach was rarely used, the team has limited evidence about what happens when it is delivered consistently. If delivery was high but the outcome did not improve, the proposed mechanism needs another look.
High compliance is useful information, but it is not automatically proof of better care. A process target can be achieved while the patient outcome remains unchanged. Keep both questions visible in the report.
What makes a measure a balancing measure?
A balancing measure examines a plausible consequence of the change outside the main target. For example, a longer communication step might improve understanding while extending consultation time or delaying the next patient.
Choose a consequence that the project could reasonably influence. “Staff satisfaction” without a collection method is too broad. A defined measure of time, interruptions or reported burden may give the team a clearer signal.
Balancing measures should not be added only after a problem appears. Consider likely trade-offs during planning, then monitor them alongside the main result. A favourable outcome chart can otherwise conceal a cost elsewhere in the pathway.
A balancing measure is also not necessarily an adverse event. It may reveal a neutral or positive consequence, or show that a feared disruption was not observed. Report the actual pattern without treating every unchanged value as proof of safety.
Worked example: follow-up understanding in a community clinic
This is an invented project for teaching. It does not describe a service’s actual results. A community clinic identifies that some patients leave without a clear understanding of when to contact the nursing team again.
The proposed aim is to increase the proportion of eligible patients who correctly describe their next contact arrangement from a baseline level established by the team to a locally agreed target within ten weeks. The change is a short teach-back prompt at the end of the consultation.
| Type | Proposed measure | Reason for inclusion |
|---|---|---|
| Outcome | Percentage correctly describing the next contact arrangement | Addresses the patient understanding aim |
| Process | Percentage of eligible consultations using the agreed prompt | Tracks delivery of the intended action |
| Balancing | Median consultation duration | Checks a plausible time burden |
| Additional context | Number of eligible consultations each week | Helps interpret changing workload and denominators |
The team should agree what a correct response means before collecting the baseline. For example, the assessment could require the patient to identify both the planned date or trigger and the contact route. The definition must be appropriate to the patient group and communication needs.
Patients who need interpretation, accessible information or supported communication should be considered during design. Excluding them for convenience could make the project appear successful while overlooking an important part of the service population.
What the example would not establish
A ten-week local project may show a pattern of improved understanding. It would not automatically show fewer admissions, better long-term health or the same result in every clinic. Those claims require other measures and a suitable evaluation.
The outcome also depends on when and how the patient is asked. A response immediately after teaching may differ from recall several days later. Describe the timing so readers understand the result.
The example illustrates alignment rather than prescribing a universal target. Set targets through local evidence, stakeholder judgement and feasibility. Do not copy a percentage because it appears ambitious in another project.
Write an operational definition for every measure
A measure specification turns an idea into a repeatable collection procedure. Two people should reach the same recording decision when they examine the same eligible case. If they cannot, the definition needs refinement.
| Specification field | What to record |
|---|---|
| Purpose | The question the measure answers |
| Eligible population | Who or which encounters enter the denominator |
| Numerator | What event or criterion counts as success |
| Exclusions | Justified cases left out and the reason |
| Data source | Where the information comes from |
| Timing | When the observation is made |
| Collection schedule | How often and by whom data are recorded |
| Missing data | How unavailable information is distinguished and reported |
| Review action | Who reviews the result and what prompts a response |
The IHI measures worksheet provides a resource for organising measures and collection plans (IHI, n.d.-b). Adapt the plan to your setting rather than treating a blank worksheet as a completed method.
For the clinic example, the denominator might be all eligible consultations during defined collection sessions. The numerator for the outcome would be consultations where the patient meets the agreed understanding criterion. State how refusals or unavailable assessments are handled.
Check numerators and denominators before percentages
Percentages can look impressive while hiding a small denominator. Nine successful cases out of ten produce 90%, but so do 90 out of 100. The two results contain different amounts of information.
Display the counts alongside the percentage, particularly when weekly activity varies. A reader should know whether an apparent change reflects more successful cases, a different denominator or both.
Check whether the numerator is genuinely a subset of the denominator. Counting completed actions across all encounters while dividing by a narrower group can produce an invalid percentage. The eligibility rules must align.
Do not treat missing documentation as confirmed non-delivery without justification. “Not recorded” and “not performed” can differ. If the measure is documentation itself, absence may be a relevant failure; if it is actual care delivery, the inference requires care.
Choose a collection interval that supports learning
A single before-and-after percentage can conceal variation. Repeated observations over time let the team see how the process behaves during testing and implementation. The collection schedule should suit the project’s pace and activity level.
Weekly collection may be useful for a busy clinic. A rare event may require another interval or a different approach. Very small weekly denominators can create unstable percentages that are difficult to interpret.
Choose the schedule before implementation where possible. Record any later change and its reason. If baseline data were collected differently from follow-up data, discuss whether the apparent difference could reflect the collection method.
Use our nursing PDSA guide for planning small tests of change. The measurement plan should support those tests by answering specific questions, rather than collecting large quantities of unrelated information.
Plot the data and annotate changes
A run chart can help display a measure across time when used appropriately. Label the time points and units clearly. Annotate when the prompt was introduced, revised or extended so readers can compare the pattern with the implementation record.
Do not interpret every upward movement as evidence of improvement. Routine variation, case mix and workload can change the result. Use a recognised interpretation method appropriate to the chart and available observations.
A statistical process control chart requires the right chart type and assumptions. It is not simply a line graph with invented limits. Obtain advice if formal signal rules or control limits form part of the evaluation.
Present the process and balancing results alongside the outcome. A chart showing improved understanding is more informative when readers can see whether prompt use increased and consultation duration changed during the same period.
Test collection before the project depends on it
Try the recording procedure on a small number of suitable cases before routine collection. Check whether the information exists, whether the form is understandable and whether the work fits the service’s schedule.
Ask two collectors to apply the definition to the same sample where feasible. Disagreement may reveal ambiguous wording or an unreliable data source. Resolve the issue before interpreting a long series of measurements.
Record changes to the definition. If “prompt used” initially meant any verbal reminder but later required all planned components, the two periods are not directly equivalent. A project log should make the change visible.
Keep the source data secure and proportionate to the project. A dissertation rarely needs identifiable patient information in its results table. Follow the setting’s agreed governance and information-handling requirements.
Consider whose experience the average hides
An overall improvement can coexist with uneven experience across groups. A communication change may work well for some patients and poorly for those facing a language or accessibility barrier.
Consider whether relevant subgroup information can be collected lawfully and interpreted meaningfully. Small groups create both privacy and precision concerns. Avoid publishing identifiable combinations or making strong conclusions from a handful of cases.
Use stakeholder input to decide which differences matter. A patient representative may identify a barrier that routine documentation misses. Their contribution can help refine the measure and the change itself.
Do not add demographic variables merely because they are available. Every collected field should have a justified purpose and fit the project’s approvals and data plan.
Interpret the family of measures together
Suppose the clinic observes better understanding, higher prompt use and longer consultations. That pattern suggests a possible benefit accompanied by a time cost. The team should discuss whether the trade-off is acceptable and whether the workflow can be adapted.
If understanding improves while prompt use remains low, investigate other influences. Staff may have changed another communication practice, the assessed population may differ or the process recording may be inaccurate.
If prompt use rises but understanding does not change, the action may be insufficient or poorly targeted. Examine whether the prompt was delivered meaningfully, rather than assuming a completed tick box proves effective communication.
Use the nursing logic-model guide to describe the proposed pathway between activities and results. The measures then test parts of that pathway without making the diagram itself proof of causation.
Write the measures section of your QI report
Explain why each measure was selected and how it was collected. Provide operational definitions, eligibility, schedule and missing-data handling. State who collected the observations and how consistency was checked.
The results should show what happened to each planned measure, including unfavourable or unclear findings. Avoid reporting only the outcome that improved. If a balancing measure was unavailable, explain the gap and its implications.
SQUIRE 2.0 provides reporting guidance for healthcare improvement and emphasises the intervention, its context and its evaluation (Ogrinc et al., 2016).
The discussion should consider context and competing explanations. Staffing changes, appointment patterns or another service initiative may influence the trend. Describe these alongside the intervention record.
Report the limitations of short follow-up. A positive pattern during supervised testing may not persist after support is withdrawn. Sustainability needs an appropriate monitoring plan rather than a confident prediction.
Common measurement mistakes
- Using an activity count as proof of patient benefit.
- Choosing a target without baseline or local rationale.
- Changing eligibility rules without recording the change.
- Presenting percentages without denominators.
- Ignoring unavailable or incomplete observations.
- Collecting so many measures that the process becomes unreliable.
- Leaving balancing measures until the final discussion.
- Claiming causation from a simple time trend.
These problems often begin during planning. Review the proposed data table with someone who works in the setting. They can identify whether the definitions reflect the real pathway and whether the collection burden is manageable.
A final measurement-plan check
Before implementation, confirm that the aim and outcome align. Each process measure should track a relevant action, and each balancing measure should address a plausible consequence. Verify that collectors can apply every definition consistently.
Then check the planned presentation. Can you show the data over time, supply the denominators and explain when changes occurred? If the project can only produce one final percentage, revisit whether the collection plan supports the learning you intend.
Finally, identify who will act on an adverse pattern. Monitoring is useful only if the team knows when to review the change, adjust it or seek advice. Agree that responsibility through the local project process.
Support for designing a practical measurement plan
Our nursing capstone project support can help you review measure definitions, alignment and report structure. Bring your local aim, available data sources and university brief. Support should strengthen your own planning and interpretation rather than create invented project findings.
Frequently asked questions
How many measures should a nursing QI project include?
Use a manageable family linked to the aim and likely consequences. There is no benefit in collecting many indicators that the team cannot define, review or act upon.
Is documentation an outcome or process measure?
Its role depends on the aim. Documentation may be the direct target of one project and a process supporting a patient outcome in another.
Can staff time be a balancing measure?
Yes, when extra time is a plausible consequence. Define the unit, start and end points, collection procedure and review plan clearly.
Do all measures need a numerator and denominator?
No. Percentages require them, while time or score measures need other specifications. Every measure still requires a clear operational definition.
Does reaching a process target prove improvement?
No. It shows that the recorded action reached its target. Outcome and balancing data are needed to understand the wider pattern.
Measure the result, delivery and consequences
A useful QI measurement plan connects the intended benefit with the actions and trade-offs involved in achieving it. Define each measure before collection, retain denominators and review patterns over time. Together, these choices give your nursing project a more credible account of what changed and what remains uncertain.
References
- Institute for Healthcare Improvement (n.d.-a) How to Improve: Model for Improvement: Establishing Measures. Available at: IHI guidance (Accessed: 30 September 2026).
- Institute for Healthcare Improvement (n.d.-b) Quality Improvement Project Measures Worksheet. Available at: IHI worksheet (Accessed: 30 September 2026).
- Ogrinc, G. et al. (2016) ‘SQUIRE 2.0: revised publication guidelines from a detailed consensus process’, BMJ Quality & Safety, 25(12), pp. 986–992. Available at: Full article.
