Background: At Janssen R&D, a new initiative called ASSURE (Active Safety Surveillance Using Real World Evidence) was established to attempt to systematically incorporate observational data from administrative claims and electronic health records as part of safety signal evaluations which typically rely mainly on clinical trial and spontaneous adverse reporting data. In June 2022, a signal for epistaxis in methylphenidate (MPH) users was opened for evaluation.
Objectives: To evaluate a safety signal for epistaxis in methylphenidate (MPH) users using observational data.
Methods: A retrospective analysis using observational healthcare data from 8 databases (6 United States, 1 Germany, 1 Japan) assessing the association between MPH use and epistaxis was conducted. The study used comparative cohort (CC) and self-controlled cases series (SCCS) designs. Relative risks (RRs; including hazard ratios for CC, incidence rate ratios for SCCS) and 95% confidence intervals (CIs) were reported, negative controls were used for empirical calibration. Results were pooled across databases using meta-analysis. Analyses were stratified by age (6-12, 13-17, 18-64 years) and database. Two at-risk periods were analyzed for the CC design (30 days post-index, all time on treatment). All patients met diagnostic criteria for attention deficit hyperactivity disorder (ADHD) and were users of MPH or 1 of 3 comparators (lisdexamfetamine, dextroamphetamine, atomoxetine). First use was considered index date. Cases of epistaxis were defined by a diagnosis or procedure specific to epistaxis. A new epistaxis event required no evidence of epistaxis in the previous 7 days.
Results: Epistaxis was observed in ~1% of new users of MPH during their treatment, with similar proportions observed for comparators. RRs from meta-analyses of the CC designs ranged from 0.96 to 1.16. During the first 30 days post-index, all meta-analytic 95% CIs contained the null effect of 1.0 with upper bounds reaching no higher than 1.40 across each age strata and for each comparator. Similar RRs were seen for the on-treatment analysis, but meta-analytic CIs were narrower due to longer observation time leading to one CI (vs. dextroamphetamine in 6-12 year olds) above 1.0 (95% CI: 1.01-1.24). All CI upper bounds remained below 1.40. For the SCCS analysis, all meta-analysis CIs covered 1.0 and pooled RRs ranged from 0.92 to 1.12.
Conclusions: No consistent association was found between MPH use and epistaxis events, across age groups, databases, comparators, and study designs. If a causal association does exist, the magnitude of the increased risk is likely < 40% based on upper bounds of the CIs obtained from meta-analyses. Incorporating observational data adds context and depth to the typical signal evaluation process.
