Risk of tuberculosis comparison in new users of antitumour necrosis factor-α and with existing disease-modifying antirheumatic drug therapy.

WHAT IS KNOWN AND OBJECTIVE: Patients with rheumatic disease are at risk for infections. Evaluating antitumour necrosis factor (anti-TNF) drug-associated risk of infections requires justification of baseline risk in the population at high risk of infection. We examined the incidence of active tuberculosis (TB) and its risk factors in patients with rheumatic disease started with anti-TNF-α therapy or with existing disease-modifying antirheumatic drug (DMARD) therapy.

METHODS: A retrospective cohort study of anti-TNF-α therapy new users (anti-TNF-α group) and those starting with a DMARD after the failure of at least one other DMARD or who had added to existing DMARD treatment (DMARD group) for rheumatic disease in the largest medical setting in Taiwan from 1 January 2005 through 31 November 2013 was conducted to determine relative risk of TB between patient groups. Patients in the DMARD group were stratified into "mild" and "severe" disease severity as proxies for low and high background risk of infection.

RESULTS AND DISCUSSION: A total of 3640 patients were enrolled (anti-TNF: 955; DMARD: 2685). The incidence of TB was 903.9/100 000 patient-years for anti-TNF-α new users and 391.7/100 000 patient-years for DMARD switchers. In Cox regression model, adjusted HR for TB in the anti-TNF-α group was higher than for the entire DMARD group (aHR, 2.41; 95% confidence interval [CI], 1.2-4.85), subgroup with mild disease (2.91; 1.31-6.47) and subgroup with severe disease (1.65; 0.68-4.03). Significant independent risk factors for TB were being male, age ≥60 years, history of respiratory disease, glucocorticoids dose >7.5 mg/d and living in a TB-prevalent region.

WHAT IS NEW AND CONCLUSION: Anti-TNF-α therapy was independently associated with increased risk of TB in patients with mild disease, but it was not significantly correlated in patients with severe disease after adjusting for confounders.