Drug Interaction Classification Systems

Anyone who uses any compendiumof drug?drug interactions—a book,a personal digital assistant (PDA), or acomputer-based screening program—isfamiliar with the various schemes thathave been devised to classify drug interactions.These classification systemswere developed during a time when littlewas known about the mechanismsof drug interactions and even less wasknown about the potential to causepatient harm. As a result, a variety ofparameters have been included in theschemes, such as the "quality" or quantityof the published data concerningthe interaction. Because interactionsgenerally could not be predicted, onehad to wait for a report of an interactionto appear in the literature.

Today, most interactions can be predictedbased on simple pharmacologicproperties. The need for interaction documentationis reduced to defining themagnitude of the effect of the precipitantdrug on the object drug. In addition,although some of the classificationsystems are based on very different criteria,most users assume that the systemsare the same and are surprised that theuse of different criteria result in differentclassifications of the same interaction. Itis important that one read and understandthe criteria employed to determinethe interaction classifications inthe compendium one uses.

There is an innate desire to simplifythe whole issue. No one can possiblymemorize all the potential drug interactionsthat have been identified todate, and new interacting drug pairs areidentified every month. This fact hasled to the common assumption thatmost drug classes have homogeneousinteraction profiles, when such is actuallyquite a rare occurrence. "Just tellme the 10 most significant drug interactions"is a pleading we often hear.Unfortunately, we are never going to beable to list the 10 or even the 100 mostsignificant drug interactions. The largeinterpatient variability in the magnitudeof the effect makes predicting theclinical outcome of an interaction inany one patient nearly impossible.

It is common for drug interaction studiesdone in healthy individuals undercontrolled conditions to demonstrate a 5-to 7-fold difference in effect betweenparticipants. This variability would beexpected to increase with the addition ofdifferent doses, routes of administration,formulations, sequences of drug administration,genetics, and other modifiers ofdrug elimination or response (eg, food,environment, disease).

Knowledge of the mechanisms ofdrug interactions has increased markedlyin the past few years, yet almostnothing is known about their epidemiology.Thus, developing comprehensive,sophisticated, "expert" systems toclassify drug interactions for whichalmost no data exist seems to be, atbest, a futile effort. At worst, it maylead to errors of interpretation.

When a classification system is appliedto a collection of drug interactions, usersimmediately make generalizations regardingthe importance of the interactions.For example, some computerizedscreening systems are adjusted so thatonly interactions with the most severerating are flagged for pharmacist review.There are, of course, no data to supportsuch arbitrary delineation of interactionalerts, and there exists a real medicolegalrisk if a patient is harmed by a "low-risk"interaction that was ignored. In otherinstallations, pharmacists routinely overridedrug interaction alerts, knowing thatmost of the time the patient will not sufferan adverse outcome. The unpredictabilityof drug interaction outcomeshinders one's ability to generalize interactionseverity classifications.

Pharmacists must learn to thinkabout every drug interaction that has apotential to cause patient harm. Howdoes one know which interactions willcause a particular patient harm? Onesimply cannot know. Until data areavailable on the risk of an adverse outcomefrom an interaction, one must erron the side of patient safety. Considerationof the therapeutic range of theobject drug, dosages of the drugs, concurrentdisease state(s), patient demographics,and many other variables isnecessary to estimate the risk to thepatient. No book, PDA, or computer-basedclassification system can replacethe pharmacist's informed evaluation.An interaction that is likely to cause anadverse outcome in one patient mayhave no effect on a different patient.

Here are some tips for pharmacists:

• Learn to recognize the factors thatalter a patient's risk for an adverseevent when he or she is exposed tointeracting drug pairs
• Consider the risk of the potentialinteraction against the benefit ofadministering the drugs
• If the risk to the patient appears tobe greater than the expected benefit,identify a suitable alternativefor one of the drugs

Drug interaction classification systemsshould be used for general guidance.One should find a classificationsystem that uses appropriate criteria toguide one's response to a potentiallyinteracting drug pair. One shouldunderstand the classification system'slimitations and then regard each interaction?patient pair as unique. By evaluatingthe risk and benefit of a potentialdrug interaction to a patient, thepharmacist can decide on a suitablecourse of action.

Drs. Horn and Hansten are both professorsof pharmacy at the University of WashingtonSchool of Pharmacy. For an electronic versionof this article, including references ifany, visit www.hanstenandhorn.com.