Risk of Significant Infection in Rheumatoid Arthritis Patients Switching Anti-Tumor Necrosis Factor-α Drugs

Objectives To describe rates of first significant infection of rheumatoid arthritis patients who switch between anti-tumor necrosis factor (aTNF) drugs. Methods Subjects with rheumatoid arthritis who received only aTNF drugs were observed in an insurance claims database from January 2001 to December 2007. Nonswitchers (NS) remained on one aTNF throughout the study period (date of the first aTNF claim was the index date); switchers (S) received at least one other aTNF (claim date for the 2nd agent was the index date). Significant infections included those that required intravenous antibiotics or hospitalization. Two attributable risk periods were used: (1) an infection occurring ≤90 days following a claim for an aTNF (90-day) and (2) an infection occurring after the index date (ever-treated). Follow-up was censored at the first occurrence of a significant infection event, end of eligibility, or end of study period. Data were analyzed using Cox regression. Results In 13,752 NS and 2293 S patients, time-stratified rates declined 2- to 3-fold between the first year versus ≥2 years. Risk of significant infection was not different for either attribution model [90-day hazard ratio (HR) = 0.93, 95CI: 0.74 to 1.17, P = 0.55; ever treated HR = 0.94, 95CI: 0.78 to 1.15, P = 0.57]. First and second year rates were similar. Predictors included age ≥50 years; history of significant or opportunistic infection, diabetes, respiratory disease; Charlson score ≥2; or prior hospitalizations. Conclusions The risk of a significant infection was not different between NS and S patients. Regardless of switching status, the rate of infection was greater in the first year. This study was limited by the lack of clinical data to determine the reason for switching.