The performance of some comput-erized drug-interaction screeningprograms has been found to be

lacking in sensitivity and specificity.1

There are numerous reasons why theseprograms fail to meet the needs of phar-macists, including the following:• They may miss interactions• They may use classification systems

that are based on rules of question-able relevance

• They may rely on literature reportswithout informed review or evaluation

• They may assume that all drugs in aclass will interact in a similar mannerPerhaps the most frustrating problem

with these screening programs is theirinclusion of numerous interactions thatare of dubious clinical significance orrelevance. As a result, the pharmacistmust override an excessive number ofdrug-interaction alerts before he or shecan continue processing the prescrip-tion order. The pharmacist thus be-comes desensitized to drug-interactionalerts, and there is a real possibility thata drug interaction will slip through un-detected and cause patient harm andlegal unpleasantness.

Several recent studies have focusedattention on the problem of computer-ized drug-interaction alerts and howpractitioners perceive them. In a studyof >30 million prescription claims re-viewed by a pharmacy benefit manage-ment company, the computer screen-

ing system identified ~250,000 po-tential interactions.2 By applying filtersto select interactions that would morelikely represent a risk to the patient,this number was reduced to ~65,500cases. Pharmacists reviewed each ofthese cases and determined that~12,800 were clinically relevant, basedon predetermined criteria. Althoughone could debate the criteria used todefine the clinically relevant drug inter-actions, it is important to note that thepharmacist review reduced the numberof clinically relevant alerts generatedby the computer by >94%.

The number of drug-interaction alertsgenerated by a screening program mayprove to be bothersome for pharmacists,but what happens when physicians areexposed to the same process of drug-interaction screening during physicianorder entry? A group of general practi-tioners in the United Kingdom conduct-ed a survey of responses to computer-generated drug-interaction alerts.3

Twenty-two percent of thephysicians admitted to over-riding the alerts withoutobtaining more informationon the potential interaction.When asked why theyignored the alerts, 98% ofthe physicians said that theybelieved the drug interac-tion was not serious, and87% thought that it was notrelevant to their patient.Drug therapy was changedin 13% of the patients,based on the drug-interac-tion alerts.

Another study involved

internists in primary care practice usinga physician order-entry system with adrug-interaction screening program.The physicians overrode almost 95% ofthe drug-interaction alerts, including89% of the level 1 (severe) and 96% ofthe level 2 (moderate) interactions.4

About 25% of the times when an alertwas generated by the computer, thephysician exited the order screen andthen attempted to reorder the drug thathad previously produced the alert.Reasons given for overriding the alertsincluded the following:• The patient was no longer taking the

interacting medication• The interaction was not clinically

significant• The patient was stable on the

combination• The benefit of the treatment out-

weighed the risk of the interactionWhat should be done to improve

computer-based drug-interaction screen-ing? Are we expecting too much of these

Computerized Drug-Interaction Alerts: IsAnybody Paying Attention?John R. Horn, PharmD, FCCP, and Philip D. Hansten, PharmD

Drug Interactions: Insights and Observations

Drs. Horn and Hansten are both professorsof pharmacy at the University of WashingtonSchool of Pharmacy.

Pharmacy Times February 2004 1

2 Pharmacy Times February 2004

programs? It has been said that predict-ing clinically significant drug interac-tions “is not rocket science.” No, it is notrocket science, but it is much more diffi-cult. As we all have experienced, mostpatients exposed to potential drug inter-actions do not develop an adverse out-come. The goal, however, should be that

no patient is ever injured by a drug inter-action. Although few data are availableon the clinical outcomes of potentialdrug interactions, one study of 538 geri-atric patients exposed to potentiallyinteracting drugs found that 130 pa-tients experienced side effects thought tobe a direct result of the interaction.5

The risk that a drug interaction willproduce an adverse outcome is modi-fied by a variety of factors, includingthe dose of the drugs, their route ofadministration, the therapeutic index,the degree of first-pass metabolism, thepatient’s concomitant diseases andintrinsic enzyme activity, and, perhapsmost importantly, the recognition bythe prescriber that an interaction is pos-sible. Anyone who expects a computerinteraction-screening program to takethese and other factors into accountbefore providing an alert is going toremain disappointed for the foreseeablefuture.

Computer-based drug-interactionscreening programs could be improved.Until a program is available with thefeatures pharmacists want, here are afew suggestions:• Use the computer as a drug-interac-

tion screening tool• Rely on other sources of drug-inter-

action information to supplementthe program, particularly when de-tails regarding the interaction ormanagement options are inadequateor missing

• Use your training as a pharmacist toevaluate potential interactions on apatient-specific basis

• When indicated, provide the pre-scriber with noninteracting alterna-tives and other management adviceAs demonstrated by the study by

Peng and associates,2 a pharmacist’s in-put can drastically alter the computer’soutput.

For a list of references, send a stamped, self-addressed envelope to:

References Department, Attn. D. Ryan, Pharmacy Times, 241 Forsgate Drive,

Jamesburg, NJ 08831; or send an e-mailrequest to: dryan@mwc.com.

Drug Interactions: Insights and Observations