The risk of uveitis in patients with JIA receiving etanercept: the challenges of analysing real-world data

Objectives: To describe and compare the occurrence of newly diagnosed uveitis in children with juvenile idiopathic arthritis (JIA) receiving methotrexate, etanercept, adalimumab, and infliximab. 
Methods: This on-drug analysis included patients within UK JIA registries (BSPAR-ETN & BCRD) with non-systemic disease, registered at methotrexate or biologic start with no history of uveitis. Follow-up began from date of first treatment, continuing until first uveitis, discontinuation of registered drug, most recent follow-up up or death, whichever came first. Hazard ratios (HR) comparing risk of uveitis between drugs were calculated using propensity adjusted Cox regression. 
Results: 2294 patients were included (943 methotrexate, 304 dalimumab/infliximab, 1047 etanercept). There were 44 reported cases of uveitis (27 methotrexate, 16 etanercept, 1 adalimumab). Unadjusted HR showed a reduced risk of uveitis in biologic cohorts compared with methotrexate. After adjusting for propensity deciles, there was no significant difference in the risk of uveitis between patients receiving etanercept or methotrexate (HR 0.5 (0.2-1.1)). Fully adjusted comparisons were not possible for adalimumab/infliximab as there were too few events.
Conclusions: In this first paper to compare the rate of new onset uveitis across the three main anti-TNF therapies used in JIA, a new diagnosis of uveitis is less common among patients starting biologics compared with methotrexate, although this did not reach statistical significance. The suggested protective effect of etanercept is likely explained by confounding, whereby patients in the methotrexate cohort are younger and earlier in disease, therefore more ‘at risk’ of developing uveitis compared with etanercept patients.