Background/Purpose: Higher disease activity might be associated with a higher risk of developing infections among patients with rheumatoid arthritis (RA). Since no randomized trial has been conducted using disease activity as the main exposure, this association may be confounded due to the effects of other RA treatments or comorbidities. To examine the association between RA disease activity and hospitalized infection using data from the U.S. Corrona RA registry linked to Medicare claims data for outcome ascertainment. We hypothesized that patients who started in moderate disease activity and subsequently attained low disease activity or remission had lower rates of infection.

Methods: Using CORRONA RA patients with Medicare coverage in 2006-2014, we identified eligible patients who had at least one visit with moderate disease activity based on the clinical disease activity index (CDAI between 10 and 22). Hospitalized infection was assessed in Medicare data. Follow-up started at the second Corrona visit and ended at the earliest date of: hospitalized infection, high disease activity (CDAI >22), loss to follow-up, or 12/31/2014. We calculated the incidence rate of hospitalized infection for patients in remission, low and moderate disease activity and estimated the effect of CDAI on hospitalized infection by controlling for time-dependent confounders using marginal structural models (MSM). Each observation was weighted using stabilized inverse-probability-of-treatment weights (IPTW), truncated at the 1^st and 99^th percentile.

Results: A total of 1,618 RA patients were eligible for analysis, for which 212 hospitalized infections were identified over mean follow-up of 2.2 years. These patients had mean (SD) age of 69.0 (10.0) years and 77.2% were female.  The crude incidence of hospitalized infection was 4.0 per 100 person years for patients in remission, 7.1 for low disease activity and 7.for in moderate disease activity. Using MSMs, and referent to being in remission, the hazard ratio of hospitalized infection associated with moderate disease activity was 1.19 (95% CI: 0.42-3.42) and for low disease activity was 0.78 (95% CI: 0.37-1.67). Baseline factors that were significantly associated with an increased risk for infection included older age, duration of RA, glucocorticoid use, and history of hospitalized infection.

Conclusion: In rheumatoid arthritis patients, the crude incidence of hospitalized infection was higher for patients in low or moderate disease activity compared to those in remission. However, these differences were greatly attenuated after controlling for confounding using marginal structural models to account for the interplay of disease activity, RA treatments, treatment switching, and other potential confounders.