(BSPOT24) Data Mining for Adverse Events of Montelukast in Children and Adolescents: A Tree-Based Scan Statistics Analysis of Swedish Nationwide Data

Background: Montelukast, a leukotriene receptor antagonist, is efficacious and widely used in children and adolescents with asthma and allergic rhinitis. It is primarily an add-on drug in patients with insufficient response to inhaled corticosteroids, but it is also used as a first-line treatment. Concerns about potential adverse effects of montelukast, including neuropsychiatric events, have been raised previously. However, the main source of safety data has been pharmacovigilance reports.

Objectives: To screen for signals of previously undocumented adverse events of montelukast in children and adolescents with asthma or allergic rhinitis.

Methods: This was a data mining study based on Swedish nationwide data from routine clinical practice (2007-2021). An age-stratified propensity score-matched analysis was performed where inhaled corticosteroids and long-acting beta-agonists served as active comparators in age groups 6 months-5 years and 6-17 years, respectively. The propensity score matched cohort was balanced on general potential confounders, including sex, age, socioeconomic factors, place of residency, and general health care use. Patients were followed as long as they were on treatment (maximum 1 yr). We screened for signals of adverse events based on incident physician-assigned diagnosis codes (ICD-10) from inpatient and outpatient contacts in specialist care. Tree-based scan statistics using an unconditional Bernoulli model was applied to identify events with unexpectedly high incidence (significance level 5%). Diagnoses at all classification levels from disease chapters to four-position codes were considered with adjustment for multiple testing. We used Monte Carlo simulation (n=9999) to estimate p-values.

Results: We identified 25,017 and 30,898 matched pairs of montelukast initiators and comparators, in the younger and older age groups, respectively (mean age 2.4 and 11.8 years; 59% and 57% were male; mean follow-up 4.0 and 5.8 months, respectively). We observed five signals of events with unexpectedly high rates: R05 Cough (risk difference 0.57%; p-value < 0.001), J35 Chronic diseases of tonsils and adenoids (0.45%; 0.009), L50 Urticaria (0.22%; 0.001), J22 Unspecified acute lower respiratory infection (0.16%; < 0.001), and J40 Bronchitis, not specified as acute or chronic (0.15%; 0.05; only three-position codes presented). All identified adverse events were considered to be likely related to the underlying diseases or comorbidities rather than the treatment.

Conclusions: Based on a matched sample of more than 55,000 children and adolescents initiating montelukast, we identified five signals but none of them were likely related to the drug.