NUR 5310 Module 4 Decision Support Review Example

The Duplicate Therapy Alert at Admission: Override Patterns, Alert Fatigue and a Governance Plan for Tuning One Rule

Student Name

Master of Science in Nursing (MSN), Thomas Edison State University

NUR-5310: Nursing Informatics: Concepts and Issues

Mentor Name

Month Day, Year

The Duplicate Therapy Alert at Admission: Override Patterns, Alert Fatigue and a Governance Plan for Tuning One Rule

Clinical decision support (CDS) is intended to bring the right knowledge to the clinician at the moment of a decision. On our medical units, one rule has become the clearest example of support that no longer supports: the duplicate therapy alert that fires during admission medication reconciliation. This paper describes the rule and its behavior, reviews evidence on alert overrides and alert fatigue, analyzes why this rule misfires after external medication histories were imported, and proposes a governance process for deciding who may change a rule, on what evidence, and how changes are monitored.

The Rule and What It Does

The duplicate therapy rule compares each medication on the admission list with every other active or pending medication and fires when two belong to the same therapeutic class. It was designed to catch true duplicates, such as a patient taking a brand-name and a generic version of the same antihypertensive, or a home benzodiazepine continued alongside a new inpatient order for another. Kuperman et al. (2007) classed duplicate therapy checking as one of the basic forms of medication-related decision support, alongside drug-allergy, drug-interaction, basic dosing and formulary checks, and noted that the usefulness of any check depends on how well the underlying knowledge base and the patient's data are represented.

That last point explains our problem. Since external pharmacy fill histories began appearing on the admission screen, the list being checked often contains every fill from the past year, including discontinued drugs, dose changes recorded as separate fills, and the same drug dispensed by two pharmacies. Each of these looks like a duplicate to the rule.

The rule was configured years ago, when admission lists were typed by hand from the patient's report and rarely held more than a dozen entries. Nobody revisited its logic when the data feeding it changed, which is a common pattern: rules are built, switched on and then left alone while the record around them evolves.

What the Override Data Show

Our informatics nurse pulled three months of alert data for the medical units. The duplicate therapy alert fired at a rate of several per admission, and nurses and pharmacists overrode the great majority of them, usually selecting "already addressed" or "patient no longer taking." A sample review of fifty overridden alerts found that most were triggered by fills of the same medication at different strengths or from different pharmacies; only a handful identified a genuine duplication that the reconciler then corrected.

These findings match published experience. Nanji et al. (2014) reviewed 157,483 medication-related alerts on more than two million outpatient orders and found that 52.6% were overridden; duplicate drug alerts were the most common type, and reviewers judged only about half of a sample of overrides to be justified, a share that ran from 12% for kidney dosing advice to 92% for allergy warnings. The variation matters for us: the issue is not that clinicians ignore alerts in general but that some rules generate far more noise than signal.

Alert Fatigue and Its Mechanisms

Alert fatigue is often blamed for overrides, but the term covers more than one mechanism. Ancker et al. (2017) tested two explanations in data from 112 primary care clinicians: cognitive overload from more and more complex work, and desensitization to the same alert over time. Acceptance fell as clinicians received more alerts per encounter and more repeated alerts for the same patient; every extra reminder in a visit cut the odds of acceptance by about 30%. Workload by itself was not associated with lower acceptance, and newly deployed alerts did not lose effect over time. The authors concluded that reducing repeat alerts within the same patient was a promising target.

Our duplicate therapy alert is a textbook repeat generator. One patient with twelve months of fills for a single drug at two strengths can trigger several alerts in one reconciliation, each requiring a click. The danger is that a reconciler who has clicked past ten false duplicates will click past the eleventh, real one.

Nurses are not passive recipients of these alerts. On our units, nurses complete about half of admission reconciliations at night, when no pharmacist is present, so the effect of noisy alerts falls disproportionately on them, and their override behavior is a direct measure of whether the rule is working.

Proposed Changes to the Rule

The team proposed three technical changes. First, compare only medications the reconciler has marked as currently taken, rather than every external fill, so the rule fires after reconciliation decisions rather than before them. Second, suppress the alert when the two entries share the same ingredient and differ only in strength or dispensing pharmacy, and instead display them grouped on the screen. Third, show a one-line explanation of why the alert fired, naming both entries, so the reconciler can judge it quickly. Interruptive alerts would be kept for true class duplications involving high-risk drugs such as opioids, anticoagulants and sedatives.

Each change was weighed against the harm the rule exists to prevent. Grouping same-ingredient fills could, in theory, hide a real duplication if a patient were taking two strengths of the same drug at once on purpose and a third prescriber added another. To guard against that, the grouped display will show the total daily dose across the entries and flag any total above the usual maximum. The pharmacist reviewing the reconciled list the next morning remains a second check, and any change will be reversed if monitoring shows a missed duplicate reaching the patient.

Governance: Who Tunes the Rule

No single department should change a safety rule on its own. Our hospital's CDS committee includes pharmacy, nursing, medicine, informatics and patient safety. We proposed that rule changes require a written request stating the problem, the data behind it and the expected effect; review by a pharmacist and an informatics nurse; testing in a non-production environment with real cases; and approval by the committee. After go-live, the committee should review the rule's firing rate, override rate and a sample of overrides at thirty and ninety days, and the safety team should watch event reports for any missed duplication. Nurses at the bedside need a simple route to report a misfiring alert, such as a feedback button on the alert itself, because they are often the first to notice a rule that no longer fits the workflow.

The goal of tuning is not fewer alerts for their own sake. It is a rule that fires when it should, and is trusted when it does.

References

Ancker, J. S., Edwards, A., Nosal, S., Hauser, D., Mauer, E., & Kaushal, R. (2017). Effects of workload, work complexity, and repeated alerts on alert fatigue in a clinical decision support system. BMC Medical Informatics and Decision Making, 17(1), 36. https://doi.org/10.1186/s12911-017-0430-8

Kuperman, G. J., Bobb, A., Payne, T. H., Avery, A. J., Gandhi, T. K., Burns, G., Classen, D. C., & Bates, D. W. (2007). Medication-related clinical decision support in computerized provider order entry systems: A review. Journal of the American Medical Informatics Association, 14(1), 29-40. https://doi.org/10.1197/jamia.M2170

Nanji, K. C., Slight, S. P., Seger, D. L., Cho, I., Fiskio, J. M., Redden, L. M., Volk, L. A., & Bates, D. W. (2014). Overrides of medication-related clinical decision support alerts in outpatients. Journal of the American Medical Informatics Association, 21(3), 487-491. https://doi.org/10.1136/amiajnl-2013-001813