Real‐World Outcomes and Safety of JAK Inhibitors in Paediatric Alopecia Areata: A Retrospective Multicentre Study

INTRODUCTION: Alopecia areata (AA) is a non-scarring, immune-mediated hair loss disorder with substantial psychosocial impact in children. Robust paediatric real-world evidence regarding Janus kinase inhibitors (JAKi) for this disease is limited. In this study, we describe real-world clinical outcomes and safety of JAKi for paediatric AA. METHODS: We conducted a retrospective, observational multicentre study across 11 Spanish hospitals including patients aged 0-18 years who initiated a first JAKi (baricitinib, tofacitinib, or ritlecitinib) in routine care. The primary outcome was Severity of Alopecia Tool (SALT)-50. Secondary outcomes included the absolute SALT trajectory, eyebrow/eyelash responses, adverse events, and switching to a second JAKi. RESULTS: Forty-six patients were included (baricitinib 22/46 [47.8%], tofacitinib 14/46 [30.4%], ritlecitinib 10/46 [21.7%]; mean age 11.3 ± 4.0 years). Eight patients (21.6%) achieved SALT-50 at weeks 4-8, 21 patients (58.3%) at week 16, 20 patients (80.0%) at week 48, and 14 patients (82.4%) at week 72. Eyebrow and eyelash responses improved concordantly. At week 16, higher baseline SALT reduced the odds of response (β = -0.061, SE 0.029; p = 0.034). Unadjusted response rates varied across drug groups; however, treatment allocation was non-random, and groups were markedly imbalanced at baseline, so drug-level comparisons are exploratory and likely confounded by indication. Response was lower in universalis (unadjusted p = 0.008). Adverse events occurred in 16 of 46 patients (34.8%), mostly mild, with two patients (4.3%) discontinuing treatment. A second JAKi was used in seven patients (15.2%): Three complete, two partial, and two absent responses occurred post-switch. DISCUSSION: In real-world paediatric practice, we observed improvements in scalp, eyebrow, and eyelash involvement with JAKi use, with a generally favourable tolerability profile. Given the observational design, baseline imbalances, and missing data, causal inference and between-drug comparisons are not supported.