(212) An Active Safety Surveillance Using Real-World Evidence (ASSURE) Approach to Pharmacovigilance Signal Evaluation: The case of methylphenidate and potential signal for obsessive-compulsive disorders

Background: Routine pharmacovigilance showed a possible association between methylphenidate (MPH) and obsessive-compulsive disorders (OCD) or trichotillomania (TTM). To inform signal evaluation and causality assessment, we conducted a rapid cycle pharmacoepidemiologic analysis using our ASSURE (Active Safety Surveillance Using Real-World Evidence) framework.

Objectives: To evaluate the association between MPH and OCD/TTM among patients with attention-deficit hyperactivity disorder (ADHD)

Methods: We defined four pillars of this analysis (TCIO) - Target: MPH; Comparators: atomoxetine, lisdexamfetamine, and dextroamphetamine; Indication: ADHD (age ≥ 6); Outcomes: OCD and TTM. We conducted a targeted literature review and consulted with subject-matter experts to develop outcome phenotypes. Quantitative feasibility assessment in several multi-national databases, including insurance claims and electronic health records (e.g., Merative MarketScan®, Optum’s de-identified Clinformatics® Data Mart Database (Clinformatics®), IQVIA PharMetrics® Plus and Ambulatory EMR) confirmed nine of them as fit-for-purpose. New-user comparative cohort and self-controlled case series (SCCS) designs were employed; the cohort design estimated hazard ratio in both intention-to-treat analysis (ITT, 60 days follow-up, informed by clinical case reports) and on-treatment analysis (all-time exposed) using large-scale propensity score adjustment with 1:100 variable-ratio matching. SCCS design estimated the risk on-treatment compared to unexposed time. We assessed pre-defined diagnostics to determine whether to unblind results from each analysis and then pooled the estimates across databases using random effects meta-analysis.

Results: Compared with atomoxetine, the pooled hazard ratio for OCD was 0.91 (95%CI: 0.71-1.18) in the on-treatment analysis and 0.91 (0.70-1.19) in the ITT analysis. Similar hazard ratios were observed in contrast with dextroamphetamine (1.04; 0.88-1.23 and 1.05; 0.87-1.26) and lisdexamfetamine (1.00; 0.86-1.17 and 1.05;0.89-1.25) and consistent with on-treatment analyses. Null results with wider confidence intervals were observed in TTM analysis. SCCS failed outcome trend diagnostics in most databases and results were not reported. The risk estimates in age subgroups were similar.

Conclusions: Our analysis did not show an association between methylphenidate and OCD or TTM when contrasted with stimulant and non-stimulant alternative treatments. Comparator selection was complex because spontaneous reports suggested similar associations with all ADHD treatments (non-placebo active comparator). This observation may imply a similar risk of OCD/TTM between available therapies but may not rule out a causal association.