Patients in the intensive care unit (ICU) are particularly vulnerable to adverse drug reactions (ADRs), many of which are preventable. Various algorithms can be used to help clinicians determine the likelihood that a medication is responsible for an ADR. A new study compared three pharmacovigilance algorithms to test their accuracy. All three tools were found to produce similar results for retrospectively evaluating ADRs in the ICU. However, some variability was found when they were used to analyze ADRs prospectively while the patient was in the hospital.
The three instruments studied were the modified Kramer, Naranjo, and Jones. Each evaluates similar or different criteria and provides its own scoring system and categorization of ADRs. During the first phase of the study, researchers used the algorithms to evaluate a random sample of 261 administrations of medication antidotes after patients were discharged from the ICU (i.e., retrospectively). During the second phase, they used an active medication monitoring system to prospectively evaluate five abnormal laboratory values while patients were in the ICU to determine the likelihood of an ADR.
All three algorithms displayed moderate to excellent agreement when it came to analyzing ADRs in the ICU following discharge. As a result, any one of them can be used for retrospective analysis of ADRs with confidence, note the researchers. They did find more variability among the three tools when they were used to evaluate ADRs on the day they occurred during hospitalization. In these cases, the researchers recommend that more than one algorithm be used to achieve a more definitive ADR assessment.
The study was supported in part by AHRQ (HS17695). See "Comparison of three pharmacovigilance algorithms in the ICU setting," by Sandra L. Kane-Gill, Pharm.D., M.S., Elizabeth A. Forsberg, Pharm.D., Margaret M. Verrico, B.S., and Steven M. Handler, M.D., Ph.D., in Drug Safety 35(8), pp. 645-653, 2012.