VIEWPOINT

Drug-interaction monitoring:
Good, bad, and ugly

Erythromycin and theophylline were dispensed to a three-year old boy. The pharmacist on duty was unaware of the potential risk of the interaction between the two medications. The boy suffered severe brain seizures that resulted in permanent brain damage.

Based on cases like this, the court system is establishing that pharmacists have a duty to detect and warn of potentially harmful drug interactions. Additionally, the Omnibus Budget Reconciliation Act (OBRA) '90's prospective drug use review requirement calls for pharmacists to screen all prescriptions, before dispensing, to detect potential problems. These requirements put additional pressure on pharmacists already burdened with too many prescriptions.

The good thing is that pharmacists have available pharmacy software systems that help screen for harmful drug interactions. The bad thing is that, according to a study by Georgetown University and published in U.S. News & World Report, an astonishing proportion of potentially harmful drug interactions go undetected by community pharmacists. The ugly thing is that the software systems pharmacists rely on may be deficient. According to a study by the University of Washington, nine drug-interaction detection software systems in 516 community pharmacies in the state failed to detect significant drug interactions one-third of the time.

Software products fail to detect major drug interactions for a number of reasons. One study suggests that too much data, lack of specific data, variation among drug-interaction classification systems, and differences among data banks have led to software deficiencies. Another study suggests that there are discrepancies in the listing and rating of drug interactions among leading drug information sources. Also, there is variability in the interpretation of data among these software systems.

The root of the problem lies in the drug information database. This produces a domino effect leading to software design problems, resulting in too many drug-interaction warnings. For example, an imbalance in the sensitivity and specificity of the software can produce too many drug-interaction warnings. Too many warnings lead to annoyance and to efforts to bypass the system without proper review and necessary calls to physicians. And the use of outdated software allows warnings to be bypassed, hindering rather than aiding the pharmacist in detecting and preventing drug interactions.

Also, some pharmacists are deficient in their knowledge of and training in these systems. Some of our recent data revealed that many community pharmacists did not know who determines which drug interactions will set off an alarm in their computer system and how often their drug-interaction software system is updated. Pharmacists are thus deterred from using such systems to their fullest potential.

Our data also revealed that the majority of the community pharmacists surveyed rely heavily on software systems to detect drug interactions. They believe that these systems is the best way to detect drug interactions and think their system does a great job of detecting such interactions. Based on the University of Washington study, however, it is clear that many community pharmacists are not aware of the limitations associated with drug-interaction detection software.

Although pharmacists' failure to detect many drug interactions is associated mainly with their software system, there are other factors as well. Patients may fail to provide pharmacists with pertinent information that could help them detect drug interactions, such as neglecting to mention receiving medication samples from a physician.

The scarce amount of time community pharmacists have for detecting and preventing drug interactions is also a problem. Pharmacists have a lot of prescriptions to fill and many other duties, such as counseling, to perform—and often have inadequate staff support.

How can these problems be fixed? The databases must be fixed first (Table 1). Most of the information about drug interactions is based on case reports, leading to an overabundance of drug interactions in databases. To date, there are no population-based outcome studies using a control group to assess the true incidence and severity of interactions from various drug combinations. Thus these spontaneous reports are missing considerable information and adequate causality assessment.

There is also no standard drug-interaction classification system in existence. A system that is oriented toward the management of drug-drug interactions rather than the assessment of them would be helpful. An example is the OpeRational ClassificAtion of drug interactions (ORCA). Some researchers have suggested that a nongovernmental organization should establish and maintain standards for the content of drug-interaction databases. The United States Pharmacopeia (USP) is working on this effort.

Once improvements have been made to the databases, the drug interaction software can be redesigned (Table 2) to be more specific and less sensitive in detecting drug interactions. The drug interactions that trigger an alert should be user-defined, thus reducing the number of insignificant alerts and minimizing the misuse of the software. The software should be changed to allow only a pharmacist to override alerts—and an explanation required must be given for overriding them. Additionally, the software should identify the source of the drug-interaction information.

Community pharmacists should be fully knowledgeable of those drug interactions known to cause harm and not have to over-rely on computers to prevent harm (Table 3). Pharmacists should receive proper training about their software system (including its limitations in detecting drug interactions).

Ideally, community pharmacists should have a pleasant work environment with minimal stress and sufficient staffing. This will allow them to do their job in such a way as to ensure the best health outcomes for all their patients. There should also be more emphasis on the importance of obtaining crucial information from patients—including any Rx drugs, drug samples, OTCs, and herbal supplements they're taking.

Finally, our information reveals that pharmacists are still entering the majority of the prescriptions. Pharmacy technicians should be doing this; typing prescriptions is extremely time-consuming. The less time spent counseling and obtaining pertinent information from patients increases pharmacists' risk for missing potentially harmful drug interactions.

By William N. Kelly, Pharm.D. and Sheila C. Housley

William N. Kelly is professor of clinical and administrative sciences at Mercer University, Southern School of Pharmacy, Atlanta.

Sheila C. Housley is a fourth-year Pharm.D. candidate at Mercer University, Southern School of Pharmacy, Atlanta.

 

William Kelly. Drug interaction monitoring: Good, bad, and ugly. Drug Topics Sep. 1, 2003;147:41.