Risk of infections in rheumatoid arthritis patients switching from anti-TNF agents to rituximab, abatacept, or another anti-TNF agent, a retrospective administrative claims analysis.

Objective: This study compared the incidence and hazard of ICD-9-CM-coded infections and severe infections in rheumatoid arthritis (RA) patients treated with subsequent-line (SL) BIOs (BIO) after switching from first-line (FL) anti-TNF therapy (anti-TNF).

Methods: Retrospective analysis of a large U.S. claims database. RA patients initiating an FL anti-TNF between 1/1/2004 and 3/31/2010 were identified and followed forward in time to capture all SL BIO episodes through 3/31/2010. SL BIO episodes were classified into: abatacept, adalimumab, etanercept, infliximab, or rituximab. Multivariate mixed-effects survival models compared the hazard of infections and severe infections across the SL BIO episodes with adjustment for demographic and clinical confounders.

Results: In total, 4332 SL BIO episodes were identified: mean age 55 years; 80% female. In adjusted analyses: when compared to rituximab, the hazard of all infections was significantly higher for adalimumab (hazard ratio [HR] = 1.31, 95% confidence interval [CI] = 1.09-1.55), etanercept (HR = 1.44, 95% CI = 1.20-1.72), and infliximab (HR = 1.30, 95% CI = 1.07-1.57), and insignificantly different for abatacept (HR = 1.18, 95% CI = 0.98-1.41); when compared to rituximab, the hazard of severe infection was significantly higher for infliximab (HR = 1.62, 95% CI = 1.03-2.55), and insignificantly different for abatacept (HR = 1.21, 95% CI = 0.78-1.88), adalimumab (HR = 1.10, 95% CI = 0.72-1.68), and etanercept (HR = 1.27, 95% CI = 0.83-1.95).

Conclusions: In RA patients treated with SL BIO, a 30-44% higher hazard of all infection was observed in anti-TNFs versus rituximab with a 62% higher hazard of severe infection observed in infliximab versus rituximab. This study used a non-randomized, observational design and is therefore subject to confounding from unmeasured factors that influence both treatment choice and infection risk.