Prednisone Use and Risk of Mortality in Patients With Rheumatoid Arthritis: Moderation by Use of Disease-Modifying Antirheumatic Drugs

Abstract

Objective: Medications for rheumatoid arthritis (RA) may affect survival. However, studies often include limited followup and do not account for selection bias in treatment allocation. Using a large longitudinal database, we examined the association between prednisone use and mortality in RA, and whether this risk was modified with concomitant disease-modifying antirheumatic drug (DMARD) use, after controlling for propensity for treatment with prednisone and individual DMARDs.

Methods: In a prospective study of 5,626 patients with RA followed for up to 25 years, we determined the risk of death associated with prednisone use alone and combined treatment of prednisone with methotrexate (MTX) or sulfasalazine. We used the random forests method to generate propensity scores for prednisone use and each DMARD at study entry and during followup. Mortality risks were estimated using multivariate Cox models that included propensity scores.

Results: During followup (median 4.97 years), 666 patients (11.8%) died. In a multivariate, propensity-adjusted model, prednisone use was associated with an increased risk of death (hazard ratio [HR] 2.83 [95% confidence interval (95% CI) 1.03-7.76]). However, there was a significant interaction between prednisone use and MTX use (P = 0.03), so that risk was attenuated when patients were treated with both medications (HR 0.99 [95% CI 0.18-5.36]). However, combination treatment also weakened the protective association of MTX with mortality. Results were similar for sulfasalazine.

Conclusion: Prednisone use was associated with a significantly increased risk of mortality in patients with RA. This association was mitigated by concomitant DMARD use, but combined treatment also negated the previously reported beneficial association of MTX with survival in RA.

Figure 1

Prevalence of prednisone, methotrexate and sulfasalazine use in ARAMIS over time (1981-2005)

Figure 2

Hazard for death associated with prednisone monotherapy and in combination with methotrexate and sulfasalazine. Cox time-varying multivariable regression model, adjusted for age, sex, ethnicity, education level, duration of RA at study entry, BMI, HAQ Disability Index, pain, rheumatology visit (y/n), NSAIDs, DMARDs (excluding the one under study), TNF inhibitors, COX-2 inhibitors, comorbidities and both prednisone and DMARD baseline and time-dependent propensities, stratified by time of study entry and location.