Session Type: Abstract Submissions (ACR)
Background/Purpose: Biologic agents, due to their effect on disease activity, may reduce the risk of premature mortality in rheumatoid arthritis (RA). We evaluated the association between exposure to biologics and risk of mortality in RA, using a population-based RA cohort with administrative health data.
Methods: Using administrative billing data from the Ministry of Health, we assembled a population-based cohort including all RA cases in the province who received care for RA between 01/1996 and 03/2006, using previously published RA criteria, with follow-up until 03/2010. Administrative data was obtained on all medications since 01/1996; MD visits, hospitalizations, and tests since 01/1990. For this study we identified all RA cases who used a biologic agent (anti-TNF, rituximab, anakinra or abatacept) during follow-up. Each biologic user was matched with one RA control who never used a biologic but used at least 3 DMARDs (to mimic coverage requirements) and with a recent (within 6 mos) change in DMARD. Controls were also matched on age, sex, calendar year of inclusion and closest propensity score, using a greedy matching technique. Matched controls were given the date of initiation of first biologic of the user they were matched to. A propensity score (PS) was calculated at time of initiation using markers of RA severity, as well as co-morbidities increasing risk of death. Despite selecting controls with the closest PS, matching was imperfect; therefore PS quintiles were added to the final multivariable model. Cox proportional hazard model (PHM) was used to estimate risk of death associated with biologic exposure, evaluated as a time dependent variable representing current or recent use of anti-TNF, where cases were considered exposed for up to 3 months after discontinuation. Time analyzed was from date of initiation to death or end of follow-up. PHM analysis was also adjusted for age, sex, RA duration, Charlson co-morbidity score, PS quintiles and the variables included in the PS model that were not balanced. Sensitivity analyses were carried out to test the robustness of results.
Results: Our sample includes 2156 biologic users and 2156 matched controls (mean (SD) age: 56.3(14.6), 74.7% females). We observed 573 deaths (326 in controls; 247 in biologic users). Exposure to biologics was associated with a reduced risk of death (aHR (95%CI): 0.25 (0.18; 0.36), p < 0.0001). A sensitivity analysis not requiring matched controls to have used 3 prior DMARDs or a recent change in DMARD yielded almost identical results (aHR (95%CI): 0.26 (0.18; 0.36), p < 0.0001). Another sensitivity analysis, without use of PS, but where the PS variables were allowed to enter the PHM and controls were only required prior use of one DMARD, yielded similar results (aHR (95%CI): 0.31 (0.22; 0.45), p < 0.0001). Limitations of our study are those inherent to observational study, including possible effect of residual or unmeasured confounding, and selection bias from non-random allocation of treatment.
Conclusion: In a population-based cohort, exposure to biologics was associated with a significant reduction in mortality. Given the increased mortality risk of RA, this has important implications for health policy makers, health care providers and people with arthritis.
E. C. Sayre,
« Back to 2012 ACR/ARHP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/biologics-and-mortality-risk-in-rheumatoid-arthritis-results-of-a-population-based-study/