Contents
On this page
- Definition of a performance gap assessment
- Limitations of published figures as a substitute for local measurement
- Specification of a reproducible indicator
- Four distinct explanations for an apparent gap
- Assessment of overuse as well as underuse
- Translation of the gap into feedback and the effects of feedback
- Pitfalls that invalidate an assessment
- Common causes of performance gaps
- Formulation of the gap statement
- Key points
- Application to your study
- Frequently asked questions
Academy · Research Publication · Performance gap assessment
Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting
A performance gap is the measured difference between what a defined population of patients received and what the evidence supports they should have received. It must be measured locally, using a reproducible indicator, because a published national estimate does not constitute evidence about a particular service and clinicians will not accept it as such. Most assessments fail not on the arithmetic but on the denominator.
Every improvement project rests on an implicit claim that a gap exists, yet this claim is seldom made explicit. The gap is typically asserted from the literature, for example a figure from a review or a national audit percentage, and the project proceeds without establishing that the problem exists locally. This omission is costly in two respects. Without a measured baseline, there is nothing against which improvement can be demonstrated, and without local data, the clinicians whose behaviour the project seeks to change have a legitimate and unanswerable objection: the figures do not describe their service.
Learning objectives
- Define a performance indicator with a numerator, denominator, exclusions and data source that another analyst could reproduce
- Distinguish a true performance gap from a documentation gap, a coding artefact and appropriate clinical variation
- Measure a baseline in a way that will withstand challenge from the clinicians it describes
- Assess gaps in both directions, since overuse and underuse coexist
- Turn a measured gap into feedback with a realistic expectation of what feedback achieves
- Write a gap statement that an executive, a reviewer or a sceptical colleague can interrogate
Common misconceptions
- “The literature already establishes the gap.” It establishes that gaps exist in some settings. Buchan, Sewell and Sweet report in the Medical Journal of Australia that 30-40% of patients do not receive treatments of proven effectiveness, an estimate derived from particular settings and sets of conditions that provides no information about an individual unit.
- “A performance gap means underuse.” Gaps occur in both directions. Grol and Wensing, in the same journal, report that 20% or more of care provided is not needed or is potentially harmful, and Buchan and colleagues estimate unnecessary or potentially harmful treatment at 20-25%. An assessment that examines only omissions will therefore overlook a substantial part of the problem.
- “If it is not documented, it did not happen.” This is sometimes true and sometimes a measurement artefact. A documentation gap and a performance gap require entirely different interventions, and conflating them produces a project that improves recording without changing clinical practice.
- “Showing clinicians the gap will close it.” Ivers and colleagues, in the Cochrane Database of Systematic Reviews, found audit and feedback produces small average improvements that vary substantially with design. Wright-Hughes, Willis and colleagues tested six feedback modifications across five UK national clinical audit programmes and found no single modification independently improved intention to act.
- “Clinicians resist the evidence.” In most cases they do not. Melnyk and colleagues surveyed 160 nurses in Worldviews on Evidence-Based Nursing and found that beliefs about the benefit of evidence-based practice (EBP) were high while knowledge was relatively low. The constraint is usually capability, time or authority rather than conviction.
Definition of a performance gap assessment
A performance gap assessment answers one question: for a defined population over a defined period, what proportion received the recommended action, and what proportion should have? It comprises four essential components. A recommended action with a source and a certainty rating 14. A denominator — the patients for whom the action was indicated, with exclusions stated. A numerator — those who received it, with the evidence of receipt specified. Finally, a data source, with the extraction method described.
If any component is missing, the result is an impression rather than an assessment. The component most often missing is the denominator, and it is the component that determines the answer. “Forty per cent of our patients were screened” is meaningless until it is known whether the denominator is all attendances, all eligible attendances, or all attendances where screening was clinically appropriate — three numbers that can differ by a factor of several.
Mainz, in the International Journal for Quality in Health Care, sets out how clinical indicators should be defined and classified for quality improvement, and it is the technical reference for this work 24. The classification matters as well as the definition: a structure indicator, a process indicator and an outcome indicator answer different questions and present different attribution problems, and most performance gap work should address process, because process is where a service has agency and where change is detectable within a project period.
Limitations of published figures as a substitute for local measurement
The published estimates of care gaps are informative but unsuitable as a baseline. Buchan, Sewell and Sweet, in the Medical Journal of Australia, report that 30-40% of patients do not receive treatments of proven effectiveness and that 20-25% receive treatments that are unnecessary or potentially harmful 22,23. Grol and Wensing, in the same journal, cite studies from the United States and the Netherlands suggesting at least 30-40% of patients do not receive care according to current scientific evidence, and that 20% or more of care provided is not needed or potentially harmful 23.
These figures establish that the problem is general and two-directional. They cannot establish that it exists in a particular service, specialty or indicator, and the reason is not merely one of generalisability. The clinicians whose practice the project seeks to change have a valid objection to external data, and presenting data that invites that objection is a strategic error from which a project cannot recover. The most common sequence in unsuccessful improvement work is to cite a national figure, encounter the response that the local population is different, concede the point and lose the support of the clinical team.
The variation literature explains why local measurement is also intellectually necessary. Wennberg, in Science in 1973, documented small-area variation in procedure rates that disease prevalence could not explain 30. Chassin, Kosecoff, Solomon and Brook, in JAMA, examined how coronary angiography was used in practice and found that clinical determinants of appropriateness only partly accounted for it 28. Eddy, in Health Affairs, attributed much of the variation to uncertainty rather than ignorance 29. If variation between services is that large, a national mean is a poor predictor of any particular service, in either direction.
Specification of a reproducible indicator
The test of an indicator definition is operational: it is given to a second analyst with access to the same data to determine whether the same count is produced. Most definitions fail this test, and the failures occur in predictable places.
The denominator’s eligibility criteria. “Eligible patients” must be decomposed into the specific data conditions that identify them — which codes, which fields, which value ranges, which date windows 24. If eligibility requires a clinical judgement that is not recorded in a structured field, the indicator cannot be computed from the record system, and a different indicator, or a notes review with explicit criteria, is required.
Exclusions. Clinically appropriate non-receipt is not a performance gap. Contraindication, patient declination and competing priority are legitimate reasons the recommended action did not occur, and an indicator that does not exclude them overstates the gap and will be rejected on exactly that basis. The appropriate approach is to specify exclusions in advance, state how each is identified in the data, and report how many records fell into each.
Evidence of receipt. “Received the action” requires a data definition: a structured field, a coded procedure, a dispensed prescription or a free-text mention. Each has a different sensitivity, and a numerator dependent on free text measures documentation rather than performance.
The period and the extraction date. Records accrue retrospectively; the same query run two months apart gives different answers. State both dates.
Our pages on electronic health records and challenges in medical data collection cover why the record system resists this kind of specification, and biostatistics support covers the sampling and precision questions.
Indicator specification: the fields that have to be filled for the count to be reproducible
| Element | What must be stated | The usual failure |
|---|---|---|
| Recommended action | The action, its source, and the certainty of the underlying evidence | A guideline named without a certainty rating, so the strength of the gap claim is unknown |
| Denominator | The exact data conditions identifying eligible patients — codes, fields, value ranges, date windows | “Eligible patients” left undecomposed, so no one else can reproduce the count |
| Exclusions | Each exclusion, how it is identified in the data, and how many records it removed | Exclusions applied silently, or not at all, which overstates the gap |
| Numerator | What counts as evidence of receipt, and in which field | A free-text-dependent numerator, which measures documentation rather than performance |
| Period and extraction date | Both dates, separately | Only the period is given, so a re-run produces a different number with no explanation |
| Counts as well as proportions | Numerator and denominator as integers | A percentage on a small denominator, presented as though the movement were signal |
Specification requirements drawn from Mainz J. Defining and classifying clinical indicators for quality improvement. International Journal for Quality in Health Care, with the failure modes compiled from the improvement literature cited here.
Four distinct explanations for an apparent gap
A low indicator value has at least four candidate explanations, and they require entirely different responses. Determining which applies is the analytical core of a performance gap assessment, and it is usually performed poorly or omitted entirely.
A true performance gap. The action was indicated, not contraindicated, not declined, and did not happen. This is the only one that justifies a behaviour-change intervention.
A documentation gap. The action happened and was not recorded in the field the indicator reads. The fix is a change to the recording mechanism, and a behaviour-change programme aimed at this will produce better documentation and no clinical change — a result that looks like success on the indicator and is not.
A coding or data-model artefact. The action was recorded somewhere the query does not look, or under a code the denominator excludes. The fix is in the query or the data model.
Appropriate clinical variation. The action was correctly not taken for reasons the indicator cannot detect. The remedy lies in the indicator’s exclusions, and dismissing this possibility undermines credibility with clinicians.
The practical method for separating them is a small structured notes review on a sample of apparent failures — typically a few dozen records, read against a pre-specified form that asks, for each case, which of the four explanations applies. The review typically requires a day or two and determines whether the entire subsequent project is aimed at the right target. Hulscher, Laurant and Grol, in Quality and Safety in Health Care, make the parallel argument for process evaluation of the intervention itself 26.
Four explanations for a low indicator value, and what each requires
| Explanation | How to recognise it | What it requires | Consequence of misdiagnosis |
|---|---|---|---|
| True performance gap | Notes review confirms the action was indicated, not contraindicated, not declined, and not done | A behaviour-change or system intervention | Nothing — this is the case the usual toolkit is built for |
| Documentation gap | Notes review finds the action recorded in free text or elsewhere but not in the field the indicator reads | A change to the recording mechanism, not to clinical behaviour | The indicator improves, clinical practice does not, and the project reports a success it did not achieve |
| Coding or data-model artefact | The action is recorded under a code the query does not capture, or eligible patients are excluded by the denominator logic | A fix to the query or the data model | A large apparent gap disappears on re-query, after the project has already expended its credibility |
| Appropriate clinical variation | Notes review finds a documented clinical reason the action was correctly not taken | A revision to the indicator’s exclusions | Clinicians are told they underperformed when they did not, and the project permanently loses clinical support |
Framework compiled from the quality measurement and process evaluation literature cited in this article, principally Mainz J in the International Journal for Quality in Health Care on indicator definition and Hulscher MEJL, Laurant MGH, Grol RPTM in Quality and Safety in Health Care on process evaluation.
Assessment of overuse as well as underuse
Most gap assessments are built to detect omission, because omission is easier to specify and more comfortable to discuss. The evidence does not support that asymmetry. Grol and Wensing report that 20% or more of the care provided is not needed or is potentially harmful; Buchan, Sewell and Sweet put unnecessary or potentially harmful treatment at 20-25% of patients 22,23. On those estimates, an assessment that looks only for underuse is examining somewhat over half the problem.
Overuse indicators are harder to construct for a specific reason: the denominator is patients for whom the action was not indicated, which requires identifying the absence of an indication in data designed to record presence. In practice this usually means a narrower, more defensible indicator — a specific test ordered in the presence of a specific exclusion, rather than a general over-testing rate. A narrow, clean overuse indicator is preferable to a broad one that will be contested indefinitely.
The Closing the Quality Gap series from Shojania, McDonald, Wachter and Owens, with the condition-specific volumes by Shojania, Ranji and colleagues, is the systematic treatment of quality improvement strategies across both directions and merits review before an indicator set is designed 17,18. Our page on monitoring clinical risk with biostatistics covers the statistical side of detecting change in rates that are already low.
Gaps run in both directions. The first two figures are from Buchan H, Sewell JR, Sweet M. Translating evidence into practice. Medical Journal of Australia. The third is from Grol R, Wensing M. What drives change? Barriers to and incentives for achieving evidence-based practice, in the same journal, which also reports studies from the United States and the Netherlands suggesting at least 30-40% of patients do not receive care according to current scientific evidence. All are estimates from particular settings rather than universal rates, and none substitutes for a local measurement.
Translation of the gap into feedback and the effects of feedback
A measured gap is not an intervention. The standard next step is audit and feedback, and its evidence base, although the strongest in this field, remains modest. Ivers, Jamtvedt, Flottorp and colleagues, and subsequently Ivers, Yogasingam, Lacroix and colleagues, in the Cochrane Database of Systematic Reviews, found audit and feedback produces small average improvements in professional practice, with effect size depending substantially on how the feedback is constructed and delivered 1,2.
The field has moved from asking whether audit and feedback works to asking how it should be designed. Foy, Skrypak, Alderson, Ivers, McInerney, Stoddart, Ingham and Keenan set out the agenda for revitalising audit and feedback to improve patient care in the BMJ, published in volume 368 as article m213 6. Wagstaff, Vindrola, Fulop and Moonesinghe, in the same journal, argue that responding to local context is the route to improvement 5.
One result should temper any proposal that promises improvement from a redesigned feedback report. Wright-Hughes, Willis, Wilson, Weller and colleagues, in Implementation Science, ran a randomised fractional factorial screening experiment testing six modifications to audit feedback, with clinicians, managers and audit staff drawn from five UK national clinical audit programmes, and found that no single modification independently improved intention to act 3. Willis, Wright-Hughes and colleagues report the wider programme. The reasonable interpretation is that feedback effects do not reside in any single design feature, so a project whose entire theory of change is an improved report has a weak theory of change 3,4.
The implication for sequencing is that the gap assessment should be paired with at least one effort-removing component — a template change, a default, a pre-populated order set — rather than with information alone. Our implementation steps page covers strategy selection, and TRIP interventions covers the component evidence.
Pitfalls that invalidate an assessment
Small numbers presented as rates. A percentage computed on a denominator of 20 moves by five points when one patient changes category 22,23. Report counts alongside proportions, and do not plot a monthly rate on a small denominator as though the movement were signal.
Denominator drift. If the eligibility criteria are refined partway through — usually in response to a clinical objection — the before and after figures are not comparable. The definition should be frozen and versioned, and if it must be changed, the baseline should be recomputed.
Gaming and the measured-not-done problem. Any indicator tied to consequence will be optimised, and the easiest optimisation is to the recording rather than the act. This is a specific reason to prefer indicators where the recording is a byproduct of the action — a dispensed prescription, a returned laboratory result — over indicators where the recording is a separate step.
Attribution to the indicator rather than the practice. An improvement in the measured proportion is an improvement in the measured proportion. Whether it reflects clinical change depends on the documentation question above, which is why the notes review is not optional.
Using a national comparator as a target. A national mean is a description, not a standard. Where the evidence supports a specific threshold, that threshold should be used; where it does not, the target should be stated as a local improvement trajectory rather than a benchmark, with the reasons given.
Common causes of performance gaps
Before the intervention is designed, it is useful to establish where the constraint typically lies, because the default assumption is usually incorrect. Melnyk, Fineout-Overholt, Fischbeck Feinstein and colleagues surveyed 160 nurses and found beliefs about the benefit of evidence-based practice were high while knowledge of EBP was relatively low 21. Melnyk, Gallagher-Ford, Zellefrow, Tucker, Thomas, Sinnott and Tan’s later study of nurses’ EBP competencies, and Melnyk, Gallagher-Ford, Thomas, Troseth, Wyngarden and Szalacha’s study of chief nurse executives, extend the picture upward: executive prioritisation of EBP against competing demands is itself frequently low 7,10.
Grol and Wensing’s catalogue of barriers in the Medical Journal of Australia distributes the causes across the individual, the social context, the organisation and the wider system, with most weight above the individual 23. Grol and Grimshaw, in the Lancet, and Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou, in the Milbank Quarterly, supply the organisational and diffusion accounts 19,25. The practical consequence for a gap assessment is that the measurement should be accompanied by a short diagnostic of cause — capability, opportunity, motivation, or system constraint — because the same gap value arising from different causes needs different interventions, and the assessment is the natural place to find out which.
Formulation of the gap statement
A robust gap statement contains seven elements in one paragraph: the recommended action and its source; the indicator definition or a pointer to it; the measured value with counts as well as proportions; the period and the extraction date; the exclusions and how many records each removed; the result of the notes review distinguishing performance from documentation; and the cause diagnosis. Written in this way, it is three or four sentences longer than the version most projects produce, and it distinguishes a claim that clinicians can challenge productively from one that they can dismiss.
For publication, improvement work is held to reporting standards that most first drafts do not anticipate — see our SQUIRE checklist and EBP and quality improvement checklist pages. Our page on EBP and informed decisions covers the clinical reasoning the indicator is a proxy for, and medical writing support covers the manuscript.
The gap statement at the start of an improvement project
Dismissible in one sentence
The literature indicates that a substantial proportion of eligible patients do not receive evidence-based care, with studies suggesting that 30 to 40 per cent miss out on treatments of proven effectiveness. A preliminary review of our own records suggests that compliance in our service is suboptimal and that there is considerable scope for improvement. A quality improvement project is therefore proposed.
The gap is external. A national or published figure invites the correct objection that the local population differs, and once that objection is made the project has no answer.
“Preliminary review suggests suboptimal compliance” has no indicator. No numerator, no denominator, no exclusions, no data source — so nothing can be reproduced and nothing can later be shown to have improved.
No exclusions. Contraindication, declination and competing priority are all counted as failures, which overstates the gap and guarantees a clinical challenge on exactly that point.
Performance and documentation are not separated. If a large share of the apparent gap is recording, the project will improve the indicator without changing care.
No cause diagnosis. Without knowing whether the constraint is capability, opportunity or motivation, the intervention is chosen by habit rather than by diagnosis.
Challengeable only on specifics
The recommended action is the one specified in the named guideline, where the underlying evidence is rated moderate certainty. The indicator is the proportion of eligible attendances with the action recorded in the structured field, defined in Appendix 1 to the level of codes, fields and date windows, so that an independent analyst can reproduce the count. Over the three months stated, the indicator was [numerator] of [denominator] eligible attendances, extracted on [date] from the record system rather than from recollection or a notes audit. Three exclusions were applied in advance — documented contraindication, documented patient declination and death or transfer within the window — and the number of records removed by each is reported in Appendix 1. We then read a sample of apparent failures against a pre-specified form: the majority were true performance gaps, a minority were documentation gaps where the action appeared in free text but not in the structured field, and a small number were appropriate clinical variation, which has been added to the exclusions and the baseline recomputed accordingly. On the cause diagnosis, staff interviews indicate capability and opportunity rather than motivation, consistent with survey evidence that belief in evidence-based practice is generally high while knowledge and time are the limiting factors. The project therefore pairs feedback with a change to the record template rather than relying on information alone.
The gap is local, measured, dated and sourced, so the obvious objection is pre-empted rather than invited.
The indicator is specified to the point of reproducibility, which is what allows a later value to count as improvement.
Exclusions are pre-specified with counts reported, so the gap is not inflated by clinically appropriate non-receipt.
A notes review separates true performance from documentation and from appropriate variation, and the baseline is recomputed where the review justified it.
The cause diagnosis points to capability and opportunity, which is what selects an effort-removing component over information alone.
Every number in the statement can be interrogated, which is the condition for a productive conversation with the clinicians it describes.
The first version can be dismissed without any data at all. The second can only be challenged on a specific exclusion, a specific code or a specific interpretation — and every one of those challenges improves the assessment rather than ending it.
Key points
Key points
A performance gap is a measured local quantity, not a figure borrowed from the literature. The indicator should be specified so that a second analyst can reproduce the count: the recommended action with its certainty, the denominator decomposed into data conditions, pre-specified exclusions with the records each removed, a numerator that does not depend on free text, and both the period and the extraction date. A sample of apparent failures should then be reviewed to separate a true performance gap from a documentation gap, a coding artefact and appropriate clinical variation; these four explanations require four different responses, and this is the assessment step most often omitted. Gaps should be examined in both directions: Buchan, Sewell and Sweet report 30-40% not receiving treatments of proven effectiveness and 20-25% receiving unnecessary or potentially harmful treatment, and Grol and Wensing put unneeded or potentially harmful care at 20% or more. Feedback alone should be expected to achieve little: the Cochrane reviews find small average effects, and a fractional factorial experiment across five UK national audit programmes found that no single one of six feedback modifications independently improved intention to act. The measured gap should be paired with a measure that reduces the effort required, and the cause should be diagnosed before the intervention is chosen, because belief is usually not the constraint.
Application to your study
Application to your study
- The indicator definition should be written before any query is run. The definition should specify codes, fields, value ranges, date windows, exclusions and extraction method, and a colleague should produce the count independently from it.
- The local baseline should be measured, and both the measurement period and the extraction date should be recorded. A project should not be initiated on the basis of a published national figure, however relevant that figure appears.
- Exclusions should be pre-specified, and the number of records each removed should be reported. Typical exclusions include contraindication, declination, death or transfer within the window, and competing priority. Unreported exclusions invite the challenge that ends projects.
- A sample of apparent failures should be reviewed against a standard form. Each case should be classified as a true performance gap, documentation gap, coding artefact or appropriate variation, and the baseline should be recomputed if the review justifies it.
- An overuse indicator should be developed alongside the underuse indicator. The overuse indicator should be narrow and defensible, such as a specific action in the presence of a specific exclusion, rather than a broad over-testing rate.
- Counts should be reported alongside every proportion. With a small denominator, a percentage changes by several points per patient, and a monthly rate plotted without counts invites over-interpretation.
- The cause should be diagnosed before the intervention is selected. Capability, opportunity, motivation or system constraints may each underlie the same gap value, and each cause requires a different response.
Frequently asked questions
It is a measurement answering one question: for a defined patient population over a defined period, what proportion received a recommended action and what proportion should have. It needs four parts to be an assessment rather than an impression — the recommended action with its source and the certainty of the evidence behind it, a denominator of patients for whom the action was indicated with exclusions stated, a numerator with the evidence of receipt specified, and a data source with the extraction method described. The part most often missing is the denominator, which is also the part that determines the answer: all attendances, all eligible attendances and all clinically appropriate attendances can differ by a large factor.
No, for two reasons. The practical reason is tactical: the clinicians whose practice the project seeks to change have a valid and unanswerable objection to external data, and the usual failure sequence in improvement work is to cite a national figure, encounter the response that the local population differs, and lose the argument permanently. The intellectual reason is that between-service variation is large enough that a national mean poorly predicts any individual service — Wennberg documented this in Science in 1973, and Chassin, Kosecoff, Solomon and Brook found in JAMA that clinical determinants of appropriateness only partly explained how coronary angiography was used. Published estimates establish that gaps exist; they cannot establish that a gap exists in a particular service.
By reviewing a sample of the apparent failures. A few dozen records in which the indicator shows that the action did not happen are read against a pre-specified form asking which of four explanations applies: a true performance gap, a documentation gap where the action occurred but was recorded somewhere the indicator does not look, a coding or data-model artefact where the query misses the record or the denominator wrongly includes the patient, or appropriate clinical variation where the action was correctly not taken. This review takes a day or two and determines whether the whole subsequent project is aimed at the right target. Omitting it risks a project that improves recording and changes no care.
Yes, although most do not. Grol and Wensing report in the Medical Journal of Australia that 20% or more of care provided is not needed or is potentially harmful, and Buchan, Sewell and Sweet put unnecessary or potentially harmful treatment at 20-25% of patients while 30-40% do not receive treatments of proven effectiveness. On those estimates, an assessment examining only omission covers somewhat over half the problem. Overuse indicators are more difficult to construct because the denominator is patients for whom the action was not indicated, which requires identifying the absence of an indication in data designed to record presence. A narrow, clean overuse indicator is preferable to a broad one that is readily disputed.
Less than most project plans assume. Ivers, Jamtvedt, Flottorp and colleagues, and later Ivers, Yogasingam, Lacroix and colleagues, in the Cochrane Database of Systematic Reviews, found audit and feedback produces small average improvements in professional practice, with effect depending substantially on how the feedback is constructed and delivered. More specifically, Wright-Hughes, Willis, Wilson, Weller and colleagues ran a randomised fractional factorial screening experiment on six modifications to audit feedback with clinicians, managers and audit staff drawn from five UK national clinical audit programmes and found no single modification independently improved intention to act. The practical conclusion is that feedback should be paired with a component that removes effort rather than relying on better information.
Five pitfalls recur. The first is percentages reported on small denominators, where one patient moves the rate several points and monthly variation is read as signal; counts should be reported alongside proportions. The second is denominator drift, where eligibility criteria are refined mid-project in response to an objection, making before and after figures incomparable; the definition should be frozen and versioned. The third is numerators that depend on free text, which measure documentation rather than performance. The fourth is indicators where recording is a separate act from the clinical action, which are the ones most easily optimised without clinical change. The fifth is the use of a national mean as a target, when a mean is a description rather than a standard; if the evidence supports no specific threshold, the target should be stated as a local improvement trajectory, with the reasons given.
Capability, opportunity and system constraint rather than belief. Melnyk, Fineout-Overholt, Fischbeck Feinstein and colleagues surveyed 160 nurses in Worldviews on Evidence-Based Nursing and found beliefs about the benefit of evidence-based practice were high while knowledge was relatively low. The same group’s study of chief nurse executives found organisational prioritisation of EBP against competing demands to be low as well. Grol and Wensing’s catalogue of barriers distributes causes across the individual, the social context, the organisation and the wider system, with most weight above the individual. This is why a cause diagnosis belongs in the assessment: the same gap value from different causes needs a different intervention.
In one paragraph containing seven things: the recommended action and its source with the certainty of the underlying evidence; the indicator definition or a pointer to the appendix holding it; the measured value as counts and proportion; the period and the extraction date separately; the pre-specified exclusions with how many records each removed; the result of the notes review separating true performance from documentation and from appropriate variation; and the cause diagnosis. That is three or four sentences longer than most projects write, and it converts a claim that can be dismissed without data into one that can only be challenged on a specific code, exclusion or interpretation — each of which improves the assessment.
References
- Ivers N, Yogasingam S, Lacroix M, et al. Audit and feedback: effects on professional practice. Cochrane Database of Systematic Reviews. 2025:CD000259. doi:10.1002/14651858.CD000259.pub4 read 9 October 2026. doi.org/10.1002/14651858.CD000259.pub4
- Ivers N, Jamtvedt G, Flottorp S, et al. Audit and feedback: effects on professional practice and healthcare outcomes. Cochrane Database of Systematic Reviews. 2012;6:CD000259. doi:10.1002/14651858.CD000259.pub3 read 9 October 2026. doi.org/10.1002/14651858.CD000259.pub3
- Wright-Hughes A, Willis TA, Wilson S, Weller A, Lorencatto F, Althaf M, Seymour V, Farrin AJ, Francis J, Brehaut J, Ivers N, Alderson SL, Brown BC, Feltbower RG, Gale CP, Stanworth SJ, Hartley S, Colquhoun H, Presseau J, Walwyn R, Foy R. A randomised fractional factorial screening experiment to predict effective features of audit and feedback. Implementation Science. 2022;17(1). read 9 October 2026. doaj.org
- Willis TA, Wright-Hughes A, Weller A, Alderson SL, Wilson S, Walwyn R, Wood S, Lorencatto F, Farrin A, Hartley S, Francis J, Seymour V, Brehaut J, Colquhoun H, Grimshaw J, Ivers N, Feltbower R, Keen J, Brown BC, Presseau J, Gale CP, Stanworth SJ, Foy R. Interventions to optimise the outputs of national clinical audits to improve the quality of health care: a multi-method study including RCT. Health and Social Care Delivery Research. 2022;10(15). doi:10.3310/QBBZ1124 read 9 October 2026. doi.org/10.3310/QBBZ1124
- Wagstaff D, Vindrola C, Fulop N, Moonesinghe SR. Revitalising audit and feedback by understanding and responding to local contexts. BMJ. 2020;369:m1467. doi:10.1136/bmj.m1467 read 9 October 2026. doi.org/10.1136/bmj.m1467
- Foy R, Skrypak M, Alderson S, Ivers N, McInerney B, Stoddart J, Ingham J, Keenan D. Revitalising audit and feedback to improve patient care. BMJ. 2020;368:m213. doi:10.1136/bmj.m213 read 9 October 2026. doi.org/10.1136/bmj.m213
- Melnyk BM, Gallagher-Ford L, Zellefrow C, Tucker S, Thomas B, Sinnott LT, Tan A. The first U.S. study on nurses' evidence-based practice competencies indicates major deficits that threaten healthcare quality, safety, and patient outcomes. Worldviews on Evidence-Based Nursing. 2018;15(1):16-25. doi:10.1111/wvn.12269 read 9 October 2026. doi.org/10.1111/wvn.12269
- Buckwalter KC, Cullen L, Hanrahan K, Kleiber C, McCarthy AM, Rakel B, Tucker S, et al. Iowa Model of Evidence-Based Practice: Revisions and validation. Worldviews on Evidence-Based Nursing. 2017;14(3):175-182. doi:10.1111/wvn.12223 read 9 October 2026. doi.org/10.1111/wvn.12223
- Hill JN, Guihan M, Hogan TP, Smith BM, LaVela SL, Weaver FM, Evans CT, et al. Use of the PARIHS Framework for retrospective and prospective implementation evaluations. Worldviews on Evidence-Based Nursing. 2017;14(2):99-107. doi:10.1111/wvn.12211 read 9 October 2026. doi.org/10.1111/wvn.12211
- Melnyk BM, Gallagher-Ford L, Thomas BK, Troseth M, Wyngarden K, Szalacha L. A study of chief nurse executives indicates low prioritization of evidence-based practice and shortcomings in hospital performance metrics across the United States. Worldviews on Evidence-Based Nursing. 2016;13(1):6-14. doi:10.1111/wvn.12133 read 9 October 2026. doi.org/10.1111/wvn.12133
- Schaffer MA, Sandau KE, Diedrick L. Evidence-based practice models for organizational change: overview and practical applications. Journal of Advanced Nursing. 2013;69(5):1197-1209. doi:10.1111/j.1365-2648.2012.06122.x read 9 October 2026. doi.org/10.1111/j.1365-2648.2012.06122.x
- Grimshaw JM, Eccles MP, Lavis JN, Hill SJ, Squires JE. Knowledge translation of research findings. Implementation Science. 2012;7:50. doi:10.1186/1748-5908-7-50 read 9 October 2026. doi.org/10.1186/1748-5908-7-50
- Trochim W, Kane C, Graham M, Pincus HA. Evaluating translational research: a process marker model. Clinical and Translational Science. 2011;4:153-162. doi:10.1111/j.1752-8062.2011.00291.x read 9 October 2026. doi.org/10.1111/j.1752-8062.2011.00291.x
- Balshem H, Helfand M, Schunemann HJ, et al. GRADE guidelines: 3. Rating the quality of evidence. Journal of Clinical Epidemiology. 2011;64(4):401-406. doi:10.1016/j.jclinepi.2010.07.015 read 9 October 2026. doi.org/10.1016/j.jclinepi.2010.07.015
- Morris ZS, Wooding S, Grant J. The answer is 17 years, what is the question: understanding time lags in translational research. Journal of the Royal Society of Medicine. 2011;104(12):510-520. doi:10.1258/jrsm.2011.110180 read 9 October 2026. doi.org/10.1258/jrsm.2011.110180
- Dawes M, Summerskill W, Glasziou P, et al. Sicily statement on evidence-based practice. BMC Medical Education. 2005;5(1):1. doi:10.1186/1472-6920-5-1 read 9 October 2026. doi.org/10.1186/1472-6920-5-1
- Shojania KG, McDonald KM, Wachter RM, Owens DK. Closing the quality gap: a critical analysis of quality improvement strategies, volume 1 - series overview and methodology. Agency for Healthcare Research and Quality (AHRQ Publication No. 04-0051-1), Stanford-UCSF Evidence-based Practice Center. 2004;1. read 9 October 2026. healthpolicy.fsi.stanford.edu
- Shojania KG, Ranji SR, Shaw LK, Charo LN, Lai JC, Rushakoff RJ, McDonald KM, Owens DK. Closing the quality gap: a critical analysis of quality improvement strategies, volume 2 - diabetes mellitus care. Agency for Healthcare Research and Quality. 2004;2. read 9 October 2026. database.inahta.org
- Greenhalgh T, Robert G, Macfarlane F, Bate P, Kyriakidou O. Diffusion of innovations in service organizations: systematic review and recommendations. Milbank Quarterly. 2004;82:581-629. doi:10.1111/j.0887-378X.2004.00325.x read 9 October 2026. doi.org/10.1111/j.0887-378X.2004.00325.x
- Grimshaw JM, Eccles MP. Is evidence-based implementation of evidence-based care possible? Medical Journal of Australia. 2004;180(6):S50. doi:10.5694/j.1326-5377.2004.tb05945.x read 9 October 2026. doi.org/10.5694/j.1326-5377.2004.tb05945.x
- Melnyk BM, Fineout-Overholt E, Fischbeck Feinstein N, Li H, Small L, Wilcox L, Kraus R. Nurses' perceived knowledge, beliefs, skills, and needs regarding evidence-based practice: implications for accelerating the paradigm shift. Worldviews on Evidence-Based Nursing. 2004;1(3):185-193. read 9 October 2026. ovid.com
- Buchan H, Sewell JR, Sweet M. Translating evidence into practice. Medical Journal of Australia. 2004;180(6):S43. doi:10.5694/j.1326-5377.2004.tb05942.x read 9 October 2026. doi.org/10.5694/j.1326-5377.2004.tb05942.x
- Grol R, Wensing M. What drives change? Barriers to and incentives for achieving evidence-based practice. Medical Journal of Australia. 2004;180(6):S57. doi:10.5694/j.1326-5377.2004.tb05948.x read 9 October 2026. doi.org/10.5694/j.1326-5377.2004.tb05948.x
- Mainz J. Defining and classifying clinical indicators for quality improvement. International Journal for Quality in Health Care. 2003;15(6):523-530. doi:10.1093/intqhc/mzg081 read 9 October 2026. doi.org/10.1093/intqhc/mzg081
- Grol R, Grimshaw J. From best evidence to best practice: effective implementation of change in patients' care. Lancet. 2003;362:1225-1230. read 9 October 2026. epistemonikos.org
- Hulscher MEJL, Laurant MGH, Grol RPTM. Process evaluation on quality improvement interventions. Quality and Safety in Health Care. 2003;12(1):40-46. doi:10.1136/qhc.12.1.40 read 9 October 2026. doi.org/10.1136/qhc.12.1.40
- Sackett DL, Rosenberg WMC, Gray JAM, Haynes RB, Richardson WS. Evidence based medicine: what it is and what it isn't. BMJ. 1996;312(7023):71-72. doi:10.1136/bmj.312.7023.71 read 9 October 2026. doi.org/10.1136/bmj.312.7023.71
- Chassin MR, Kosecoff J, Solomon DH, Brook RH. How coronary angiography is used. Clinical determinants of appropriateness. JAMA. 1987;258(18):2543-2547. doi:10.1001/jama.258.18.2543 read 9 October 2026. doi.org/10.1001/jama.258.18.2543
- Eddy DM. Variations in physician practice: the role of uncertainty. Health Affairs. 1984;3(2):74-89. doi:10.1377/hlthaff.3.2.74 read 9 October 2026. doi.org/10.1377/hlthaff.3.2.74
- Wennberg J. Small area variations in health care delivery. Science. 1973;182(4117):1102-1108. doi:10.1126/science.182.4117.1102 read 9 October 2026. doi.org/10.1126/science.182.4117.1102
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 9). Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting. Pubrica. https://pubrica.com/academy/concepts-definitions/performance-gap-assessment/
Pubrica Academy. “Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting.” Pubrica, 9 Oct. 2026, pubrica.com/academy/concepts-definitions/performance-gap-assessment/.
Pubrica Academy (2026) Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting. Pubrica. Available at: https://pubrica.com/academy/concepts-definitions/performance-gap-assessment/ (Accessed: 15 October 2026).
Pubrica Academy. “Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting.” Pubrica. October 9, 2026. https://pubrica.com/academy/concepts-definitions/performance-gap-assessment/
Pubrica Academy, ‘Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting’ (Pubrica, 9 October 2026) <https://pubrica.com/academy/concepts-definitions/performance-gap-assessment/> accessed 15 October 2026.
Pubrica Academy. Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting. [Internet]. Pubrica; 2026 Oct 9 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/concepts-definitions/performance-gap-assessment/
Pubrica Academy, “Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting,” Pubrica, Oct. 9, 2026. [Online]. Available: https://pubrica.com/academy/concepts-definitions/performance-gap-assessment/
@misc{performance_gap_assessment_2026,
author = {{Pubrica Academy}},
title = {Performance Gap Assessment in Clinical Practice: Indicator Specification, Measurement and Reporting},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/concepts-definitions/performance-gap-assessment/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Performance gap assessment worksheet
Complete sections 1 to 4 before querying any data, and section 5 before writing the gap statement. Section 5 is the step most projects skip and the one that determines whether the intervention is aimed at the right thing.
PERFORMANCE GAP ASSESSMENT
1. THE RECOMMENDED ACTION
Action ...............................: ___________________
Source (guideline, review, standard) .: ___________________
Certainty of the underlying evidence .: ___________________
Direction assessed: [ ] underuse [ ] overuse [ ] both
2. DENOMINATOR (decompose into data conditions)
Population in words ..................: ___________________
Identified in data by:
codes / terminologies ..............: ___________________
fields .............................: ___________________
value ranges .......................: ___________________
date window ........................: ___________________
[ ] Eligibility requires a judgement NOT in a structured
field -> this indicator cannot be computed from the
record system; use a notes review with explicit criteria
3. EXCLUSIONS (pre-specify; report counts)
[ ] documented contraindication ...... n removed: ______
[ ] documented patient declination ... n removed: ______
[ ] death or transfer within window .. n removed: ______
[ ] competing clinical priority ...... n removed: ______
[ ] other: _________________________ n removed: ______
4. NUMERATOR AND EXTRACTION
Evidence of receipt ..................: ___________________
Field it is read from ................: ___________________
[ ] Recording is a BYPRODUCT of the action (preferred)
[ ] Recording is a SEPARATE act (gameable - note this)
Period measured ......................: ___________________
Date extracted .......................: ___________________
Result: ______ / ______ ( ______ % )
[ ] A second analyst reproduced this count from section 2-4
5. CLASSIFY A SAMPLE OF APPARENT FAILURES
Sample size reviewed .................: ______
Reviewed against a pre-specified form? [ ] yes
true performance gap ...............: ______
documentation gap ..................: ______
coding / data-model artefact .......: ______
appropriate clinical variation .....: ______
[ ] Exclusions revised as a result; baseline recomputed
Revised baseline .....................: ______ / ______
6. CAUSE DIAGNOSIS (what is the binding constraint?)
[ ] capability (skill, knowledge)
[ ] opportunity (time, access, authority, equipment)
[ ] motivation (belief, priority)
[ ] system (policy, staffing, data, physical constraint)
Evidence for that judgement ..........: ___________________
7. TARGET
[ ] Evidence supports a specific threshold: ______
[ ] No evidence-based threshold exists; target is a local
improvement trajectory, stated as: ______________
[ ] NOT a national mean used as a target
8. GAP STATEMENT (write it in one paragraph, including:)
[ ] action and source, with certainty
[ ] indicator definition or pointer to it
[ ] value as counts AND proportion
[ ] period and extraction date, separately
[ ] exclusions with counts
[ ] notes review result
[ ] cause diagnosis
PHRASES TO DELETE IF PRESENT
[ ] "the literature indicates a substantial proportion"
used in place of a local measurement
[ ] "compliance is suboptimal"
[ ] "considerable scope for improvement"
[ ] a proportion with no denominator shown
[ ] a national average described as the target Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Pubrica reviews the measurement before a project commits to it, examining whether the indicator definition is reproducible from the text alone, whether exclusions are pre-specified with counts, whether the numerator measures the action itself or only its documentation, and whether the gap statement can withstand the objection that the local population is different. Where a draft uses a published national figure as a baseline, Pubrica replaces it with the local measurement plan for which that figure cannot substitute.