Pharmacoeconomics and Safe Medication Use

for ADE detection. The pharmacy may design daily reports that contain all the abnormal drug

serum concentrations such as the phenytoin, digoxin, lithium, vancomycin, and aminoglycoside

concentrations, or other laboratory monitoring values (e.g., lisinopril and potassium). The clinical

pharmacist can then review all cases of supratherapeutic concentrations and laboratory abnormalities

by performing a targeted medical record review for ADEs. Although these are basic CDS functions

that include medications and laboratory values as the knowledge for the alert, more advanced,

“smarter” alerts are being used that include risk scoring–based approaches that help prioritize the

pharmacist’s reviews.

Identified events must be confirmed for causality. Many instruments can help link the drug with

the event. Structured instruments create a more reliable and valid assessment. A sample ADE form

with the Naranjo criteria for causality determination is provided in Appendix 1. This is the most

commonly used ADR causality instrument, though its reliability and validity in the critically ill

population need improvement. Another instrument specific for drug-induced liver injury is the

RUCAM. Causality assessment includes temporal sequence, dechallenge (removal of the suspect

drug), rechallenge (reintroduction of the suspect drug), evaluation of other causes, objective evidence

that is available or obtained, and history of a reaction to a similar drug.

Prospective

Direct observation

One method of medication error detection is through direct observation; this can be accomplished

using the medication pass method, in which the pharmacist observes nurses in the medication

administration process and notes any errors that occur.

ii.

Direct observation provides the unique advantage of capturing medication administration

errors that are not typically identified with other detection methods. Other examples of

direct observation include nurses observing nurses or physicians observing physicians during

medication administration. This is considered prospective because observation occurs in real

time but errors are evaluated later.

iii.

A potential limitation of direct observation is that if an individual is aware of being observed,

the individual may change behavior to be “optimal” (i.e., the “Hawthorne effect”).

CDS with alert generation can be used for prospective or real-time surveillance.

Prescribing – Providers can be notified of high-risk scenarios during the ordering process in

order to prevent medication errors, such as prescribing enoxaparin for a patient with a CrCl less

than 30 mL/minute. Alerts during medication ordering should be used judiciously and created

with greater precision to avoid alert fatigue

ii.

Administering—Dosing calculators integrated into the EHR for nursing-driven protocols (eg,

insulin infusions or sliding scales, anticoagulant infusions).

iii.

Order verification – Pharmacists receive preventive alerts during order verification to avoid

medication errors and the potential for ADEs (e.g., drug-drug interactions, duplicate therapy).

iv.

High-risk scenarios in real time – Pharmacists may receive advanced alerts outside order

verification, indicating when a patient is initiated on a drug, but the patient’s clinical scenario

presents a risk (e.g., patient receiving an epidural and initiated on a systemic anticoagulant).

Role of artificial intelligence in medication use process (J Am Coll Clin Pharm. 2025;8(4):302-310)

Prescribing

Incorporation of patient-specific clinical data to develop personalized treatment plans, including

pharmacogenomics, to reduce alert fatigue

ii.

Integration of diagnostics and patient data to reduce time to diagnosis