Medication Use Evaluation (MUE): Criteria, Sampling, the Data Form, and the P&T Report

A medication use evaluation (MUE) is not a chart audit with a pharmacy label on it. ASHP defines it as “a systematic and interdisciplinary performance improvement method with an overarching goal of optimizing patient outcomes via ongoing evaluation and improvement of medication utilization” (ASHP Guidelines on Medication-Use Evaluation, Am J Health-Syst Pharm 2021;78(2):168–175). The distinction matters operationally, because it changes what you have to produce: an audit ends at a compliance rate, an MUE does not end until something about prescribing, dispensing, administration or monitoring has changed and been re-measured.

This page is the end-to-end procedure for a health-system pharmacist or P&T coordinator who has been told to run one next quarter: how to scope it, how to write criteria that survive committee review, how to size and draw the sample, what belongs on the data-collection form, and how to structure the report the Pharmacy and Therapeutics committee actually acts on. It also draws the line between an MUE, a drug use evaluation (DUE), a drug utilization review (DUR) and a straightforward audit — four words that get used interchangeably in practice and are not interchangeable in the source guideline.

MUE vs DUE vs DUR vs audit: the distinction that governs your design

ASHP’s 2021 guideline is explicit that MUE encompasses DUE in its broadest application, rather than being a synonym for it. DUE and DUR are described as criteria-based, drug- or disease-specific assessments that establish appropriate medication use at the individual patient level. MUE is the wider performance-improvement method: multidisciplinary, outcome-oriented, and cyclical.

Practically, the four terms differ along four axes — unit of judgment, what the output is, who owns it, and whether re-measurement is part of the definition:

Dimension	MUE	DUE / DUR	Chart audit
Unit of judgment	The medication-use process across a population	The individual patient’s therapy against drug- or disease-specific criteria	A discrete documented behaviour against a fixed rule
Primary question	Is our use of this medication producing the outcomes it should, and where does the process break?	Was this patient’s use appropriate per criteria?	Did the required thing happen, yes or no?
Team	Interdisciplinary by definition — prescriber, pharmacist, nurse, administrator at minimum	Usually pharmacy-led, criteria reviewed by prescribers	Often a single function (quality, pharmacy, nursing)
Output	Intervention plus re-measurement; a closed improvement cycle	Appropriateness rate; individual-level interventions	A compliance percentage
Ends when	The change is implemented and the follow-up cycle shows it held	The review period closes	The report is filed
Where the term is used	Health-system pharmacy and accreditation	Health-system pharmacy (DUE) and managed care / payer settings (DUR)	Any quality function

One further separation worth holding: DUR is the term you will most often meet in the payer and pharmacy-benefit world, where prospective DUR runs as an automated claims-adjudication edit at the point of dispensing and retrospective DUR runs against claims history. That is a different data substrate (claims, not charts) and a different decision-maker (the plan, not the medical staff). If your brief says “DUR” and your data source is your own EHR, you are almost certainly being asked for an MUE.

Why this matters for design, not just vocabulary: if you scope the work as an audit, you will build a form that captures only the compliance variable and you will have nothing to explain why the process failed when the committee asks. An MUE data set has to carry the process variables — who ordered, at what point in the stay, whether an order set was used, whether a pharmacist intervention was documented — because those are what the intervention is aimed at.

What creates the requirement

Three separate layers create an expectation that a US hospital evaluates its own medication use, and they are worth distinguishing because they impose different obligations.

The CMS Conditions of Participation

42 CFR 482.25, the pharmaceutical services Condition of Participation, is the federal floor. Its text (eCFR § 482.25) states that “the medical staff is responsible for developing policies and procedures that minimize drug errors,” a function that “may be delegated to the hospital’s organized pharmaceutical service.” Two standards under it are load-bearing for MUE work:

• § 482.25(b)(9) — a formulary system must be established by the medical staff. This is the regulatory root of the P&T committee’s authority; the MUE is the evidence base that formulary system runs on.
• § 482.25(b)(6) — drug administration errors, adverse drug reactions and incompatibilities must be reported immediately to the attending physician and, where appropriate, to the hospital’s quality assessment and performance improvement programme. That reporting stream is one of the most reliable MUE triggers you have.

Note what § 482.25 does not say: it does not use the phrase “medication use evaluation,” does not mandate a number of MUEs per year, and does not prescribe a methodology. The CoP creates the formulary system and the QAPI linkage; the method is yours to design.

The accreditation layer

The Joint Commission’s Medication Management (MM) chapter, together with its Performance Improvement requirements, is where the practical survey expectation lives for accredited hospitals — evaluating medication management processes and acting on the findings. We are not citing a specific standard or element-of-performance number here, because we could not read it. jointcommission.org returns HTTP 403 to automated retrieval and has no usable archived copy; asserting an EP number we had not verified against the current manual would be exactly the kind of confident-but-unchecked citation that gets repeated across the web. Check the current edition of your own accreditation manual (or your accreditor’s equivalent — DNV’s NIAHO standards and the CoP crosswalk differ) for the clause your surveyor will actually cite, and note that the January 2026 replacement of the Joint Commission’s National Patient Safety Goals chapter with a National Performance Goals chapter for hospital and critical access hospital programmes changed chapter structure in ways worth re-reading rather than assuming; see our guide to the National Patient Safety Goals for that change.

Your own medical staff bylaws

In most health systems, the binding obligation to run a given MUE comes from the P&T committee charter or medical staff bylaws — frequently phrased as a required number of evaluations per year, or as a condition attached to a restricted formulary addition. Read that document before designing anything; it usually specifies reporting cadence and who has to approve criteria, and those two constraints set your timeline.

Step 1 — Establish the need and write the charter

ASHP’s guideline puts “establishing the need for the MUE” first, and this is where most MUEs are won or lost. An MUE with a weak trigger produces a report nobody acts on, which is the single most common failure mode in this work.

Legitimate triggers, roughly in descending order of how easily they survive committee scrutiny:

• A formulary condition. The drug was added with a stipulation that use be evaluated after a defined period. This is the strongest trigger because the commitment already exists.
• A safety signal. Clustered adverse drug events, a near-miss pattern in your event reporting system, or an external alert. If the signal came through the incident-reporting stream, be deliberate about the patient safety work product privilege boundary — an MUE conducted for accreditation and formulary purposes is generally not PSO work product, and mixing the two data sets can compromise the protection on the material that is.
• High cost with unclear benefit. A high-spend agent whose use has grown faster than the population that should be receiving it.
• Known practice variation. Wide inter-prescriber or inter-unit variation on a medication where a defensible standard exists.
• A new guideline or a new order set. Evaluating adherence after implementation, which doubles as order-set validation.
• A regulatory or measure linkage. Medication elements that feed a reported measure — the antibiotic timing components of SEP-1, for instance — where an internal MUE gives you a diagnosis your measure abstraction cannot.

Write the trigger into a one-page charter before you write a single criterion. The charter should state: the trigger and the sponsor; the specific question in one sentence; the population (inclusion and exclusion); the study period; the data source; who is on the team; the reporting date; and what decision the committee will be asked to make. If you cannot write the last item — the decision on offer — the MUE is not ready to start.

Step 2 — Assemble the team

ASHP specifies a minimum composition for the body responsible for an MUE: prescriber, pharmacist, nurse and administrator. Treat that as a floor rather than a target. In practice you will also want:

• An informatics or data analyst who can build the report, because the difference between a two-week MUE and a two-month one is almost entirely whether the extract is automated.
• A prescriber champion from the specialty being evaluated — not a generic medical staff representative. Criteria written without one will be contested at the committee table, at which point the data becomes irrelevant.
• Whoever owns the intervention you are likely to land on. If the plausible fix is an order-set change, the order-set owner should have seen the criteria before data collection, not after.

ASHP also notes that the responsible body needs formal organisational recognition to support the resulting changes. That is the practical argument for routing the MUE through P&T rather than running it as a pharmacy department project: the committee has the standing to change the formulary and the order sets, and a departmental report does not.

Step 3 — Criteria and thresholds

Criteria are the heart of the design. ASHP is specific on two points that are routinely skipped: criteria should be built from guidelines, treatment protocols and standards of care through an interdisciplinary consensus process, and criteria must be communicated to all affected professionals before the evaluation of care begins. Retrospectively applying criteria clinicians never saw is both methodologically weak and politically fatal.

The four criteria categories

Structure your criteria set into categories rather than a flat list. The categories map onto the medication-use process ASHP describes — prescribing, preparation, dispensing, administration, monitoring — and the mapping tells you where an intervention would have to sit:

1. Indication / appropriateness. Was there a documented indication within the approved set? Was a required diagnostic result available before the order?
2. Process. Was the approved order set used? Was the ordered route consistent with policy? Was a required approval or restriction honoured?
3. Monitoring. Were the required baseline and follow-up parameters obtained within the interval your protocol defines?
4. Outcome. Clinical response, adverse events attributable to the agent, length of therapy, escalation or de-escalation.

Each criterion needs four attributes to be usable: the criterion statement, the threshold (what proportion of cases should meet it), the data element that operationalises it, and the exception set that makes a non-compliant case defensible. A criterion without a written exception set produces false non-compliance and destroys the report’s credibility in the room.

What a criteria table actually looks like

The example below is an illustrative composite, written to show the shape of a criteria table, not to recommend any therapy. It uses an intravenous-to-oral conversion MUE because that is the design pattern most systems run first — the criteria are almost entirely process criteria and the data elements are all discretely captured. The thresholds shown are placeholders; your committee sets them, and the section below explains on what basis.

#	Criterion	Category	Threshold	Data element	Exceptions
1	A documented indication within the approved list is present at the time of the initial order	Indication	Set by committee	Indication field on order; problem list at order time	Order placed pre-diagnosis in a documented emergency
2	The approved order set was used for the initial order	Process	Set by committee	Order-set identifier on the order record	Order placed in a location without the order set deployed
3	Conversion to the oral route was ordered within the interval defined in the conversion policy, once the policy’s conversion criteria were met	Process	Set by committee	Route change timestamp; first documented time all conversion criteria were satisfied	Documented inability to take oral medication; NPO status; malabsorption noted; palliative goals documented
4	Required monitoring parameters obtained at the interval the protocol specifies	Monitoring	Set by committee	Lab result timestamps relative to first dose	Patient discharged before the interval elapsed; patient declined
5	No adverse drug event attributable to the agent documented during the admission	Outcome	Set by committee	ADR report linkage; allergy-list additions during stay	—

Setting thresholds honestly

ASHP advises establishing thresholds from past performance, utilisation reports, or published benchmarks where those are available. The failure mode to avoid is the reflexive 100%: a threshold of 100% on a criterion with a real exception set guarantees an apparent failure and tells you nothing about whether the process is in control.

Three defensible bases, in order of preference:

1. Your own baseline. If you have run this MUE before, or can pull the metric retrospectively, set the threshold as a defined improvement over the measured baseline. This is the strongest basis because it is unarguable and it makes the follow-up cycle interpretable.
2. A published benchmark, cited by name in the report, with the caveat that a benchmark drawn from a different case-mix is a reference point rather than a target.
3. Committee consensus, explicitly minuted as consensus. This is legitimate — it is often the only option for a novel agent — but label it, so the next cycle knows the threshold was a judgment rather than a measurement.

Where a criterion is genuinely absolute — a hard restriction, a required approval — 100% is the right threshold and any exception is a finding by definition. Keep those criteria separate from the graded ones in the table so the two are not read the same way.

Step 4 — The sampling frame and how many charts

ASHP’s guidance on sampling is deliberately non-prescriptive: any set sample size should represent a larger group or population. It gives no magic number, and neither does any other authority — anyone quoting “30 charts” as a standard is quoting convention, not a source. What you owe the committee instead is a sampling frame you can defend.

Define the frame before the sample

The frame is the complete list of encounters eligible for evaluation. Write it as an executable definition:

• Population. Encounters, not patients, unless you have a specific reason — a readmitted patient is two exposures to the process.
• Exposure definition. At least one administered dose, or at least one order? These give materially different denominators, and the ordered-but-not-administered gap is often itself the finding.
• Window. A defined admission or order date range. Avoid straddling a known process change — a new order set going live mid-window makes the result uninterpretable.
• Exclusions. Clinical trial patients (route through the investigational drug service instead), care areas out of scope, encounters shorter than the criteria can apply to.

Choosing the sample

Take the whole frame if the extract is automated and the frame is small enough to review. A census removes the sampling argument entirely and is usually achievable for a restricted agent.

Where manual review forces a sample, three rules do most of the work:

1. Sample randomly from the frame, and record the seed or the method. Convenience samples — the first 30 encounters, or one unit’s charts — will be challenged, and the challenge will be correct.
2. Size the sample against the precision you need at the threshold. The question is whether your sample can distinguish performance above the threshold from performance below it. A sample that yields a confidence interval spanning the threshold cannot support a conclusion in either direction. As a rough orientation: a proportion estimated from 30 observations carries a margin of error in the region of ±18 percentage points, from 100 observations roughly ±10, and from 400 roughly ±5. If your threshold is 80% and you observe 75% on 30 charts, you have not demonstrated anything.
3. Stratify only when you intend to report by stratum. If the committee will want the result split by service line or by unit, stratify and size each stratum — otherwise you will present a subgroup finding resting on four charts.

State the achieved sample, the frame size and the sampling method in the report. Reviewers who see those three numbers stop arguing about method and start arguing about the finding, which is where you want them.

Step 5 — Build the data-collection form

ASHP does not publish a standardised MUE form; the guideline’s position is that the team should agree on data sources and their interpretation before collection begins, and should establish a data dictionary standardising variable definitions and collection sources. It also advises generating reports from the electronic medical record wherever feasible, to minimise manual chart review.

Read those two together and the design rule follows: the form is the physical embodiment of the data dictionary, and every field on it belongs to one of exactly two classes — extracted (pulled by report, never typed) or abstracted (requires human judgment on the chart). Mark the class on the form itself. Fields that could be extracted but are being typed are your main source of both delay and error.

Field inventory

A working form has five blocks:

1. Identifiers and audit trail. Encounter identifier, a study ID that de-identifies the record for analysis, abstractor initials, abstraction date, and a “second reviewer required” flag. Keep the identifier-to-study-ID key separate from the analysis file.
2. Denominator confirmation. The fields that prove this encounter belongs in the frame — exposure date, care area, first dose timestamp. This block is what lets you defend inclusion later.
3. One field per criterion, phrased as the criterion is phrased. Not free text. Use Met / Not met / Exception applies / Not evaluable, with a mandatory exception-code field whenever “exception applies” is selected. Pre-code the exceptions from the criteria table; if abstractors are inventing exception text, your exception set was incomplete and should be revised before collection continues.
4. Process context variables. Ordering service, whether an order set was used, whether a pharmacist intervention was documented, time of day and day of week of the order. None of these are criteria. They are the variables that turn a compliance rate into a diagnosis — they are how you find out that non-compliance is concentrated on weekend admissions, or in one location where the order set was never deployed.
5. Outcome fields. Length of therapy, escalation or de-escalation, documented ADR, and any measure element the MUE feeds.

Calibrate before you collect

Run the form against 5–10 charts with every abstractor before the real collection starts, and compare the results field by field. Disagreement at this stage is cheap and is almost always a criterion-wording problem rather than an abstractor problem. Also build a rule for what happens when an abstractor cannot tell: a “not evaluable” option with a reason code, because forcing a judgment into Met / Not met is how a documentation problem gets silently recorded as a prescribing problem.

Two documentation habits that make the difference at committee: record the exact query or report definition used for every extracted field, and freeze the data set with a date. An MUE whose numbers move between the draft and the presentation loses the room.

Step 6 — Analyse for cause, not just for rate

ASHP maps MUE onto the FOCUS-PDCA improvement framework — Find the process to target, Organize the team, Clarify current knowledge, Understand causes of variation, Select the process improvement, then Plan, Do, Check, Act. The two steps most often skipped are “Understand causes of variation” and “Check.” Skipping the first gives you an intervention aimed at the wrong thing; skipping the second means you never learn whether it worked.

Report each criterion three ways: the raw proportion meeting it, the proportion after exceptions are removed from the denominator, and the count of exceptions by code. Those three numbers separate three genuinely different problems — clinicians doing the wrong thing, clinicians doing the right thing for a documented reason your criteria did not anticipate, and clinicians doing the right thing but not documenting it. Each has a different intervention, and a single blended percentage hides which one you have.

Then cross the criterion results against the process context variables. Concentration is the finding: if non-compliance is uniform across services, times and locations, you are looking at a knowledge or protocol problem; if it concentrates in one location, one shift or one ordering pathway, you are looking at a system problem and the intervention is structural. Where the analysis points at a system failure rather than a knowledge gap, the same discipline applies as in any QAPI performance improvement project — identify the process factor, not the individual.

Step 7 — Select and implement the intervention

Rank candidate interventions by how little they depend on individual recall. In rough descending order of durability:

1. Remove the option. Formulary restriction, removal of a dosage form, or an order that cannot be placed outside the approved pathway.
2. Change the default. Order-set defaults, auto-stop dates, linked monitoring orders that fire with the medication order.
3. Add a targeted decision-support alert. Effective but expensive in alert burden — weigh it against your existing alert load before adding one, for the same reason discussed in early warning score implementation: an alert that fires too often is an alert that is dismissed.
4. Add a pharmacist review step. Reliable while staffed; it is a control that consumes a person.
5. Education and feedback. Necessary but the weakest as a standalone; effects decay, which is precisely why the follow-up cycle exists.

Write the intervention with an owner, a date and a measure. “Pharmacy will educate prescribers” is not an intervention; “the conversion criteria will be added as a default linked order in order set X, owned by the order-set committee, live by [date], re-measured at 90 days” is.

Step 8 — The P&T committee report

The report is the deliverable, and the committee reads it under time pressure with a dozen other agenda items. Structure it so a member who reads only the first page can vote correctly.

1. The ask, first. One paragraph: what was evaluated, what was found, and what decision is requested. If the answer is “no action required,” say that on line one.
2. Background and trigger. Why this MUE, and what commitment or signal prompted it.
3. Methods. Population and frame definition, study period, sample size and sampling method, data sources, criteria approval date and who approved them. This section is short but it is what makes the findings admissible.
4. Criteria table with results. The criteria as approved, each with its threshold, the observed rate, and met-or-not against the threshold. Same table the team worked from, now with two more columns.
5. Findings and analysis. Exception patterns, the process-variable cross-tabs, and the causal reading. Any “not evaluable” volume goes here, not in a footnote — it is a documentation finding.
6. Limitations. Named honestly: retrospective design, documentation dependence, single site, sample precision. A limitations section is a credibility asset, not a concession.
7. Recommendations. Each with an owner, a date and how it will be measured. Distinguish those the committee can enact itself (formulary and restriction changes) from those requiring another body (order sets, informatics, nursing policy).
8. Follow-up plan. The re-measurement date and what would constitute success, agreed at the same meeting. Deciding this later means it does not get decided.

Attach the criteria and the data dictionary as appendices. When the next cycle runs — or a surveyor asks — those two documents are what prove the evaluation was designed rather than assembled.

Step 9 — Close the loop

An MUE that stops at the report is a DUE with extra meetings. ASHP’s framing — ongoing evaluation and improvement, mapped onto PDCA — means the follow-up cycle is part of the method, not an optional extra. ASHP also notes that successful MUE programmes have structures supporting rapid-cycle data collection, analysis and intervention. Rapid-cycle re-measurement on a narrowed criteria set, at 60 to 90 days, beats a full repeat evaluation at twelve months: it is cheaper, it catches a failed intervention while the team still remembers the design, and it is the only way to distinguish a real improvement from a Hawthorne effect.

Re-measure using the identical frame definition and identical criteria wording. If either changes, record the change explicitly and expect to be asked whether the improvement is real or definitional.

Prospective, concurrent and retrospective designs

The three timings are not interchangeable, and choosing the wrong one is a common design error.

• Retrospective — review after discharge. Cheapest, cleanest data, no observation effect, and no ability to help the patients being reviewed. The default for formulary and utilisation questions.
• Concurrent — review during the admission. Allows real-time intervention and captures information that never reaches the discharged chart, at the cost of being labour-intensive and introducing an observation effect that inflates your compliance rate. Appropriate when the safety signal is active.
• Prospective — criteria applied at the point of ordering, usually as a decision-support rule or a required approval. Strictly speaking this is a control rather than an evaluation, and it is what prospective DUR means in the payer setting. It prevents rather than measures; you still need a retrospective evaluation to know whether the control is working and how often it is overridden.

A common and sensible pattern is a retrospective MUE to establish the problem, a prospective control as the intervention, and a rapid-cycle retrospective re-measurement to confirm the control held.

MUE and antimicrobial stewardship: overlapping, not identical

Antimicrobial stewardship programmes run MUEs, and MUE is one of the oldest tools in stewardship’s kit. The two are not the same thing and should not be collapsed.

An antimicrobial stewardship programme is a standing, staffed, continuously operating function with its own CMS Condition of Participation and its own core-elements framework, and it works from a data infrastructure — the cumulative antibiogram, days-of-therapy denominators, resistance trends — that runs whether or not any particular evaluation is underway. An MUE is a bounded project with a start, a criteria set, a sample and a report.

Where they meet: a stewardship programme is frequently the sponsor of an antimicrobial MUE, and the MUE’s report goes to both the stewardship committee and P&T. Where they diverge: an MUE on a non-anti-infective agent has nothing to do with stewardship, and stewardship’s day-to-day prospective audit-and-feedback is a continuous intervention rather than a criteria-based evaluation with a defined denominator. If your question is “how do we run our antimicrobial programme,” that is stewardship. If your question is “did our use of this specific agent over this specific period meet criteria, and what do we change,” that is an MUE.

Two measurement notes for antimicrobial and other utilisation MUEs: the defined daily dose (DDD) is a WHO-maintained statistical unit for cross-institutional comparison and is not a dosing recommendation — using it as a utilisation denominator is legitimate, using it as a clinical reference is not. And where an MUE touches transitions of care, the medication history quality on which it depends is a function of your medication reconciliation process, which means a reconciliation failure will present as a prescribing finding unless you look for it.

Where MUE meets formulary decisions and the AMCP dossier

MUE feeds the formulary system in both directions. Post-addition, it is how a conditional formulary approval is discharged. Pre-addition, the evidence package a manufacturer supplies is structured by the AMCP Format for Formulary Submissions — version 5.0, published April 2024 in the Journal of Managed Care & Specialty Pharmacy, which added guidance on digital therapeutics, health disparities and pre-approval information exchange decks.

The practical link for an MUE designer: a dossier tells you what evidence the manufacturer thinks supports the product, including the clinical endpoints and the economic model. Those endpoints are a reasonable starting list for your outcome criteria — you are, in effect, testing locally whether the dossier’s claims hold in your population. That framing also makes the MUE report far more useful to the committee than a compliance rate, because it answers the question the committee actually asked when it approved the addition. Note the audience difference: the AMCP Format is a managed-care and payer instrument, so treat it as a source of endpoint definitions rather than as a hospital operating procedure. Where an MUE’s finding is a spend question rather than a therapy question, the decision may belong with your value analysis committee or your supply chain function instead of P&T.

Five failure modes worth designing against

1. Criteria written after the data was pulled. Guarantees criteria shaped to the available fields and destroys the report’s standing. Approve criteria, in the minutes, before extraction.
2. No exception set. Produces a non-compliance rate composed largely of appropriate care, which prescribers will identify within about ninety seconds of the report appearing.
3. A denominator nobody can reproduce. If “patients on drug X” is not written as an executable definition, the next cycle’s number is not comparable to this one and the trend is meaningless.
4. An intervention with no owner. The most common reason an MUE produces no measurable change. Owner, date, measure — or it is a suggestion.
5. No follow-up cycle scheduled at the same meeting. Once the meeting ends, the follow-up competes with the next quarter’s agenda and loses.

Frequently asked questions

What is a medication use evaluation?

Per ASHP’s 2021 guideline, a systematic and interdisciplinary performance improvement method aimed at optimising patient outcomes through ongoing evaluation and improvement of medication use. It can examine any part of the medication-use process — prescribing, preparation, dispensing, administration or monitoring — and it is defined by including intervention and re-measurement, not just measurement.

Is an MUE the same as a drug use evaluation?

No. ASHP states that MUE encompasses DUE in its broadest application. DUE and DUR are criteria-based, drug- or disease-specific assessments of appropriateness at the individual patient level; MUE is the wider, multidisciplinary, outcome-oriented improvement method that contains them. DUR specifically is the term used in payer and pharmacy-benefit settings, where it operates on claims data.

How many MUEs does a hospital have to do per year?

No federal regulation sets a number. 42 CFR 482.25 requires a medical-staff-established formulary system and error-minimising policies but does not specify MUE frequency or method. Any number applying to you comes from your medical staff bylaws, your P&T charter, or your accreditor’s requirements — check those documents rather than assuming a standard figure.

How many charts should an MUE review?

ASHP does not specify a number, only that the sample should represent the larger population. Take the full frame where the extract is automated. Where sampling is necessary, sample randomly and size it so the confidence interval around your observed rate does not straddle your threshold — roughly ±18 percentage points at 30 observations, ±10 at 100, ±5 at 400.

Who approves MUE criteria?

The P&T committee or its designated subcommittee, through an interdisciplinary consensus process, before evaluation of care begins. ASHP is explicit that criteria must be communicated to all affected professionals in advance. Approval and communication dates belong in the methods section of your report.

What is the difference between an MUE and a chart audit?

An audit tests whether a documented behaviour matched a fixed rule and ends at a compliance percentage. An MUE evaluates the medication-use process against clinically-derived criteria, seeks the cause of variation, implements an intervention and re-measures. An MUE that stops at a percentage has been run as an audit.

Does an MUE fall under patient safety work product privilege?

Generally not by default. An evaluation conducted to meet accreditation, formulary or regulatory obligations is usually outside the protected space, and mixing PSO-protected event data into an MUE data set risks the protection on that material. Involve your risk or legal function on scope before combining data sources — see PSO reporting and the work product privilege.

What does FOCUS-PDCA mean in an MUE?

ASHP maps many MUE steps onto FOCUS-PDCA: Find the process, Organize the team, Clarify current knowledge, Understand causes of variation, Select the improvement, then Plan, Do, Check, Act. The value of naming the framework is that it makes “Understand causes of variation” and “Check” explicit steps rather than optional ones.

Sources verified

• ASHP Guidelines on Medication-Use Evaluation, Am J Health-Syst Pharm 2021;78(2):168–175, doi:10.1093/ajhp/zxaa393 — the definition, the DUE/DUR distinction, the FOCUS-PDCA mapping, minimum team composition, criteria-communication requirement, sampling and threshold guidance, and the data-dictionary requirement all come from this document.
• 42 CFR § 482.25, Condition of Participation: Pharmaceutical services — eCFR, current text. Read directly; quoted provisions verified verbatim.
• AMCP Format for Formulary Submissions, version 5.0 (April 2024), AMCP, published in JMCP vol. 30, no. 4-b.
• Joint Commission Medication Management standards — not independently verified. jointcommission.org returns HTTP 403 to automated retrieval and no usable archive was available, so no standard or element-of-performance number is asserted on this page. Consult your current accreditation manual directly.

This guide sits in CASRAI’s patient safety cluster, alongside related operational guidance on antimicrobial stewardship, medication reconciliation, QAPI documentation and the tall man lettering conventions a P&T committee applies to look-alike drug names.