Data mining analysis of psychosis with mefloquine and other antimalarial drugs

Purpose Mefloquine is used for treatment and prophylaxis of chloroquine‐resistant malaria. In 2003, thirteen years after its introduction into the US market, the manufacturer strengthened the warnings about psychiatric reactions. Although this signal was detected early, over the years there was great debate and publicity over the strength of its association with mefloquine use. To add to the body of evidence, we applied data mining methods to evaluate the early, pre‐publicity safety signals collected in the Food and Drug Administration's (FDA) Adverse Events Reporting System (AERS) database. Methods A data mining system, using the Multi‐Item Gamma‐Poisson Shrinker (MGPS) methodology, has been recently developed at the Center for Drug Evaluation and Research/FDA, for improving the early detection of safety signals from the AERS database. We used our data mining system to examine the progression of higher‐than‐expected psychiatric signal scores with mefloquine and other anti‐malarial drugs. Results A strong signal score for psychosis (about 30 times over expected) was seen the first year of mefloquine's marketing in the US. An association with psychosis was also seen with quinacrine, but not with other anti‐malarial drugs. Conclusion MGPS identified that mefloquine's signal of psychosis appeared prior to widespread publicity. This signal was not identified with other quinoline antimalarials. MGPS also identified known class and therapeutic effects with antimalarial drugs. Clinical Pharmacology & Therapeutics (2004) 75 , P77–P77; doi: 10.1016/j.clpt.2003.11.290