Use of atomoxetine and suicidal ideation in children and adolescents: Results of an observational cohort study within general practice in England

Aim: To investigate the association between atomoxetine, a drug used in the treatment of Attention Deficit Hyperactivity Disorder (ADHD), and suicidal ideation, within a cohort of 2-18-year-old patients in England. Methods: The study was conducted using the observational cohort technique of Modified prescription event monitoring (M-PEM). Patients prescribed atomoxetine were identified from dispensed prescriptions issued by primary care physicians. A customised postal GP questionnaire was used to capture outcome data for suicidal ideation. A matched pair cohort analysis was performed within patients to compare the risk of suicidal ideation in the period after starting atomoxetine with the risk prior to starting atomoxetine; this was stratified by age and concomitant use of methylphenidate. Additional information on patient characteristics, and events of interest was also collected; individual cases of suicidal ideation were qualitatively assessed for drug relatedness. Results: Of the final cohort (n = 4509); 85.5% male (n = 3857), median age 11 years (IQR: 9,14). Primary prescribing indication for atomoxetine was ADHD (n = 4261, 94.6%). Almost a quarter of the cohort had been co-prescribed methylphenidate. Results of the matched pair cohort analysis indicated that the period after starting atomoxetine was not associated with an increase in the incidence of suicidal ideation compared to the period prior to starting treatment (RR: 0.71; CI: 0.48-1.07; P-value: 0.104). Individual case assessment of suicidal ideation suggested a causal association within a number of cases. Conclusions: This study found no evidence of an increased risk of suicidal ideation during treatment with atomoxetine, compared to the period prior to starting treatment. Amongst age specific subgroups, this risk may change. Nonetheless, individual case assessment suggested a causal relationship in some patients, hence physicians need to be aware of the possibility of developing this event, and furthermore consider how best to detect it in this paediatric population. This study demonstrates the importance of combining quantitative statistical analyses with a qualitative case series assessment. (C) 2016 Elsevier Masson SAS. All rights reserved.