Introduction: Drug persistence may be a surrogate marker that reflects both long-term efficacy and safety in clinical settings, and tuberculosis (TB) is considered as one of the most important opportunistic infections after the biological treatment in rheumatoid arthritis (RA). We aimed to compare drug persistence and incidence of TB between tumor necrosis factor alpha (TNFα) inhibitors and tocilizumab in patients with RA using data from the Korean Health Insurance Review and Assessment Service database.

Methods: In this analysis, 5449 patients with RA who started TNFα inhibitors, such as adalimumab, etanercept, infliximab, and golimumab or tocilizumab, as the first-line biological therapy between January 2014 and December 2017 were analyzed and followed up until December 2019. Drug persistence was defined as the duration from initiation to first discontinuation, and TB was defined as the prescription of > 2 anti-TB medications after the initiation of biologics.

Results: TNFα inhibitors and tocilizumab were prescribed in 4202 (adalimumab, 1413; etanercept, 1100; infliximab, 769; golimumab 920) and 1247 patients with RA, respectively. During the analysis period, 2090 (49.7%) and 477 (38.3%) patients with RA discontinued TNFα inhibitors and tocilizumab, respectively, and 42 patients with RA developed TB (TNFα inhibitors, 33; tocilizumab, 9). After adjustment for confounding factors, TNFα inhibitors were significantly associated with a higher risk of discontinuation compared with tocilizumab (hazard ratio (HR) 1.63, p < 0.001). In subgroup analysis, all types of TNFα inhibitors, except for infliximab, demonstrated a significantly lower persistence rate compared with tocilizumab. There was no significant difference in TB incidence between tocilizumab and TNFα inhibitors. In subgroup analysis, infliximab has a significantly higher risk of TB compared with tocilizumab (HR 2.84, p = 0.02).

Conclusion: In this analysis, tocilizumab had longer persistence than TNFα inhibitors with a similar incidence of TB. Our analysis has limitations: (1) The HIRA database lacks clinical details like disease activity and joint damage extent, potentially influencing the analysis results. (2) Reasons for discontinuing biological agents were not available. (3) TB diagnoses may be inaccurate because of missing microbiological results. (4) We did not analyze the impact of treating latent TB infection on TB development post-biological treatment, despite mandatory screening in Korea.