Comparative risk of hospitalized infection between biological agents in rheumatoid arthritis patients: A multicenter retrospective cohort study in Japan

OBJECTIVE Knowing the risk of hospitalized infection associated with individual biological agents is an important factor in selecting the best treatment option for patients with rheumatoid arthritis (RA). This study examined the comparative risk of hospitalized infection between biological agents in a routine care setting. METHODS We used data for all RA patients who had first begun biological therapy at rheumatology divisions of participating community hospitals in Japan between January 2009 and December 2014. New treatment episodes with etanercept, infliximab, adalimumab, abatacept, or tocilizumab were included. Patients were allowed to contribute multiple treatment episodes with different biological agents. Incidence rates (IRs) of hospitalized infection during the first year of follow-up were examined. Cox regression analysis was used to calculate hazard ratios (HRs) for overall hospitalized infection and for pulmonary hospitalized infection, adjusting for possible confounders. RESULTS A total of 1596 new treatment episodes were identified. The incidence of overall hospitalized infection during the first year was 86 with 1239 person-years (PYs), yielding a crude IR of 6.9 per 100 PYs (95% confidence interval [CI], 5.6-8.6). After correction for confounders, no significant difference in risk of hospitalized infection was observed between treatment groups: adjusted HRs (95% CI) were 1.54 (0.78-3.04) for infliximab, 1.72 (0.88-3.34) for adalimumab, 1.11 (0.55-2.21) for abatacept, and 1.02 (0.55-1.87) for tocilizumab compared with etanercept. Patient-specific factors such as age, RA functional class, body mass index (BMI), prednisolone use, and chronic lung disease contributed more to the risk of hospitalized infection than specific biological agents. The incidence of pulmonary hospitalized infection was 50 and a crude IR of 4.0 per 100 PYs (95% CI, 3.1-5.3). After adjustment for confounders, adalimumab had a significantly higher HR for pulmonary hospitalized infection compared with tocilizumab: an adjusted HR (95% CI) was 4.43 (1.72-11.37) for adalimumab. BMI, prednisolone use, diabetes mellitus, and chronic lung disease were also significant factors associated with the risk of pulmonary hospitalized infection. CONCLUSIONS The magnitude of the risk of overall hospitalized infection was not determined by the type of biological agents, and patient-specific risk factors had more impact on the risk of hospitalized infection. For pulmonary hospitalized infections, the use of adalimumab was significantly associated with a greater risk of this complication than tocilizumab use.