Neuropsychiatric adverse effects of oseltamivir in the FDA Adverse Event Reporting System, 1999-2012

Publications from Roche and a case-control study suggest that there is no evidence, or plausible mechanism of action, to link neuropsychiatric adverse events to Tamiflu (oseltamivir).1 2 Cochrane Collaborators, the BMJ, and others, however, contend that many of Roche’s data remain unavailable.3

We examined neuropsychiatric adverse events associated with oseltamivir in the US FDA’s Adverse Event Reporting System (FAERS) from October 1999 to August 2012 by using a data mining platform (RxFilterTM; www.AdverseEvents.com4) to generate case report counts and reporting ratios for multiple MedDRA (Medical Dictionary for Regulatory Activities) categories, including those queried in the Roche papers. Reporting odds ratios were calculated using standard methods, but numerator and denominator variables were constrained to the period of time beginning with oseltamivir’s FDA approval date and ending with the most recent FAERS data.

The table⇓ shows the number of primary suspect cases found for a selection of MedDRA terms with corresponding reporting odds ratio scores. Scores of note included “abnormal behaviour” (29.35), “psychiatric and behavioural symptoms not otherwise classified” (15.36), “delirium” (13.50), “hallucination” (12.00), and “perception disturbances” (8.41). Finally, “life threatening” and “death” were listed 188 and 81 times, respectively, in case reports in the largest category searched, “psychiatric disorders (system organ class)”. Approximately half the reports originated in Japan and the rest were from the US and elsewhere.

FAERS post-marketing data over the past decade suggest disproportionally increased reporting of certain neuropsychiatric adverse events linked to oseltamivir. While these data support the cautionary warnings added to the drug’s label in several countries, they cannot ascertain causality or incidence. Post-marketing data are subject to many biases, such as marked under-reporting, masking, amplification, and confounding by comorbidities.5 6 Despite such limitations, we hope that these data offer an update for clinicians, while awaiting the release of all trial data.