Oseltamivir use and severe abnormal behavior in Japanese children and adolescents with influenza: Is a self-controlled case series study applicable?

OBJECTIVE:Since the 1990s, self-controlled designs including self-controlled case series (SCCS) studies have been occasionally used in post-marketing evaluation of drug or vaccine safety. An SCCS study was tentatively applied to evaluate the relationship between oseltamivir use and abnormal behavior Type A (serious abnormal behavior potentially leading to an accident or harm to another person) in influenza patients. METHODS:From the original prospective cohort study with approximately 10,000 Japanese children and adolescents with influenza (aged <18years), 28 subjects (mean age: 7.3years) who developed abnormal behavior Type A after the first visit to the collaborating hospitals/clinics were analyzed. We hypothesized four combination patterns of the effect period (i.e., the period that effect of oseltamivir on occurrence of abnormal behavior Type A is likely) and the control period. Mantel-Haenszel rate ratio (M-H RR) and its 95% confidence interval (CI) were calculated as the relative risk estimate. RESULTS:Among 28 subjects in the SCCS study, 24 subjects (86%) were administered oseltamivir and 4 subjects (14%) were not. Abnormal behavior Type A was more likely to occur in the effect period than the control period in every pattern (M-H RR: 1.90-29.1). We observed the highest estimate when the effect period was set between the initial intake of oseltamivir and Tmax (M-H RR: 29.1, 95% CI: 4.21-201). CONCLUSION:Abnormal behavior Type A was more likely to develop up to approximately 30 times during the period between the initial intake of oseltamivir and Tmax. However, this period overlapped with the early period of influenza where high fever was observed. Since useful approaches to control the influence of the natural disease course of influenza were not available in this study, we could not deny the possibility that abnormal behavior was induced by influenza itself. The SCCS study was not an optimal method to evaluate the relationship between oseltamivir use and abnormal behavior.