Background/Purpose: Serious infections (SIs) are a major concern in RA patients, significantly contributing to increased mortality. We examined the SI risk associated with bDMARDs compared to csDMARDs in a US-wide observational RA cohort.

Methods: RA patients initiating bDMARDs or csDMARDs from 2001 through 2016 in the National Data Bank for Rheumatic Diseases (NDB) were assessed for SIs (infection + intravenous antibiotics, hospitalization or death). DMARDs were categorized into 3 groups: (1) csDMARDs-reference (bDMARD naive) (2) TNF inhibitors (TNFi) (3) Non-TNFi biologics (abatacept, rituximab, tocilizumab, and anakinra). Each patient contributed to the last DMARD reached. Followup continued until the first SI, DMARD discontinuation, death, or end of study period. SIs were attributed to the corresponding DMARD group when the treatment was ongoing or discontinued ≤3 months before SI. Propensity scores (PS) reflecting the probability of receiving a specific DMARD were calculated using multinomial logistic regression models with characteristics at treatment initiation. PS was added to the Cox models as a continuous variable along with time-varying confounders (age, disease duration, comorbidities, HAQ, pain and patient global scores, weighted cumulative exposure [WCE] of glucocorticoids [GC], prior sDMARDs and bDMARDs counts) of which changes over time might alter the SI risk.

Results: 694 (5.9%) first SIs were identified in 11,623 RA patients during a 27,552 patient-years of follow-up. TNFi and non-TNFi biologics-initiators had significantly higher disease activity, disability, and comorbidity scores than csDMARDs initiators. Crude incidence rate (95% CI) in TNFi group was non-significantly higher than csDMARD and non-TNFi biologic initiators (Figure). The PS-only adjusted model showed a non-significant SI risk increase with TNFi (HR 1.11[0.91-1.34], P= 0.309) and non-TNFi biologics (HR 1.26 [0.95-1.68], P= 0.102) whereas further adjustment for time-varying confounders revealed significantly increased risk of SI with both TNFi (HR 1.33 [1.05-1.68], P= 0.019) and non-TNFi (HR 1.48 [1.02-2.16], P= 0.041) compared to csDMARDs. Other factors associated with SI are presented in Table.

Conclusion: TNFi and non-TNFi biologics were associated with an increased SI risk in RA compared to csDMARDs. In addition to DMARDs, older age, comorbidity burden, pulmonary disease, higher disability, disease activity, and cumulative GC exposure were predictive of SI. Assessment and modification of these factors should be completed before and during bDMARD treatment to minimize SI risk.