Background/Purpose: Observational studies have shown an increased
risk of mortality in patients with RA, though none have done so with patients
from all 50 US states nor is it clear how this risk
may vary over time. We examined this risk since 1998 by calendar year, duration
of RA, and treatment.

Methods: RA and non-inflammatory rheumatic disease (NIRD) patients were studied from
1998 through 2011 in the National Data Bank for Rheumatic Diseases (NDB), a longitudinal
observational study following patients from all regions of the US through biannual
questionnaires. Patients were required to have at least 2 observations or 1
observation and died before 2012; those recruited through FDA-mandated safety
registries were excluded. Mortality was confirmed through National Death
Index-matched death records. We calculated standardized mortality rate ratios (SMRRs)
of RA vs. NIRD based on age-, race- and sex-stratified US population data
(CDC.gov). Cox regression models with time varying covariates were used to
investigate the risk of mortality among diagnoses.

Results: Among 15,791 RA patients, 3,531 patients died (22.4%). The mean
(median) time in the study was 5.0 (4.0) years with a total of 79,841
patient-years of follow-up. For the 4,058 NIRD patients, 947 patients died (23.3%)
with a mean (median) time in the study of 5.0 (3.9) years with a total of 20,500
patient-years of follow-up. The overall SMRR for RA vs.
NIRD patients was 1.48 (95% CI 1.38-1.78).  For RA duration ≤5 years, the SMRR was 1.35 (1.20-1.52). For earlier calendar cut-offs,
the SMRR was 1.56 (1.38-1.63) for before 2008 and 1.74 (1.42-2.15) for before
2004.  Adjusted for age, age²
and sex, the hazard ratio (HR) for RA vs. NIRD was 1.40 (1.30-1.50). Adjusting
for additional confounders such as HAQ, pain, patient global, employment, marital
status, BMI, and smoking, resulted in an HR of 1.27 (1.17-1.38). Worse disease
activity markers, higher comorbidity index, and smoking were all
associated with an increased risk of mortality, as opposed to higher education,
being employed, and married. When analyzing RA patients only, hierarchical DMARD/biologic treatments
showed only cytotoxic DMARD use being associated with an increased risk compared to monotherapy MTX (Table).  

Conclusion: In this large US cohort, the risk of mortality for RA patients
remained elevated compared to NIRD patients. We found trends that the SMRRs
decreased over the last 14 years, yet unlike other US studies, we found no
associated mortality benefit of biologics over MTX monotherapy after adjusting
for multiple confounders.

Table. Cox
regression model for mortality in RA patients