Background/Purpose: The comparative risk of infection with newer biologic agents, such as abatacept (ABA), in the treatment of juvenile idiopathic arthritis (JIA) has not been reported. We used claims data to compare rates of hospitalized infection among JIA patients initiating ABA or tumor necrosis factor inhibitors (TNFi).

Methods: We combined data from national U.S. Medicaid claims from 2000-2010 and MarketScan claims from 2010-2014. Patients with a physician diagnosis code for JIA before age 16 years and prior to new use of ABA or a TNFi were included. New use of ABA and of each of the 5 TNFi agents was defined by a 6 month baseline period of non-use prior to first observed use. Follow-up began on the day of the new prescription fill or infusion claim and was extended for 90 days beyond the days supplied for each subsequent fill. Study outcome was hospital discharge with any infection as the primary diagnosis. We calculated crude infection rates per 100 person-years and stratified results according to clinical factors during the baseline period: use of oral glucocorticoids (GC), use of a different biologic agent, and inpatient or outpatient infection. We also stratified by systemic JIA (SJIA) as defined by any diagnosis for macrophage activation syndrome or any use of relatively SJIA-specific medications (e.g., anakinra). Multivariable regression analyses were precluded by few observed outcomes.

Results: We identified 5,933 and 257 initiators of TNFi and ABA, respectively. The baseline patient characteristics are in Table 1, and the overall and stratified crude infection rates are in Table 2. ABA was strongly associated with SJIA and with use of a different biologic agent during baseline compared to TNFi. The overall crude infection rates were higher for ABA (4.14 [2.07-8.28]) compared to TNFi (1.50 [1.22-1.85]). SJIA was associated with much higher infection rates among ABA users, and, to a lesser extent, among TNFi users. All 8 ABA patients with infection outcomes had infections during the baseline period, but the overall proportions of patients with baseline infections were similar with ABA and TNFi.

Conclusion: Crude hospitalized infection rates following initiation of ABA were higher compared to TNFi. This may be partially explained by higher proportions of patients with SJIA and recent use of other biologics among ABA initiators compared to TNFi. Infection during baseline was a risk factor for subsequent infection and may importantly be the reason for initiation of ABA in some patients. Evaluation of the comparative safety of non-TNFi biologics in JIA is challenging due to small sample sizes and requires careful consideration of prescriber channeling.

Table 1.

Table 2.