Background Recent studies have demonstrated a declining trend in RA mortality relative to the general population (1). This improvement in mortality could be due to improvement in incident risk of cardiovascular events that are the leading cause of excess deaths in RA (2).
Objectives Our objective was to assess secular trends in ten-year incident risk of cerebrovascular accident (CVA) in incident cohorts of RA versus general population controls, using administrative health data.
Methods We conducted a retrospective study of a population-based cohort of incident RA cases who first met previously published RA criteria between 01/01/1997 and 31/12/2004 in British Columbia followed until 31/12/2014, with general population controls matched 2:1 on gender, age, and index year. Individuals were excluded if they had a diagnosis of CVA prior to index date. Incident CVA was defined as first CVA during follow-up using ICD codes 9 code 433, 434/ICD-10 code I64, I63) in Hospital Discharge data or death certificate in Vital Statistics data. RA and general population cohorts were stratified according to year of RA incidence, defined according to first RA visit, using a 7-year wash-out period. Incident rates (IRs) of CVA for RA and general population cohorts, as well as incident rate ratios (IRRs), with 95% confidence intervals (CI) were calculated per calendar years of incidence. Multivariable Cox Proportional Hazard models with left truncation were used to estimate risk of CVA in RA relative to general population while controlling for potential confounders, with contribution of person time of follow-up starting from index date (second RA visit) to avoid immortal time bias and censoring at ten years from incident year, or last health care utilization. To examine whether secular trends differed in RA relative to general population, an interaction term was tested between the RA indicator and year of RA incidence. To account for non-linear effect of cohort year, we compared cox regression models with linear, quadratic, and flexible spline forms of the cohort-year effects and the model with the best AIC was used to interpret the data.
Results 23,545 RA individuals (65.7% female; mean [SD] age 58.11[16.82] years) and 47,090 controls experienced 658 and 1,220 incident CVA respectively. A linear spline Cox model with a knot at year 1999 was selected to fit the CVA events. The change in CVA risk over time differed significantly in RA vs. general population after 1999 [p=0.0488], but not before 1999 [p=0.06]. A significant decline in risk of CVA was observed over the calendar years of incidence after 1999 in RA [0.91 (0.86, 0.96); p=.0003] but not in the general population [0.97(0.93, 1.01); p=0.1019].
Conclusion Our findings suggest that the risk of CVA has significantly declined over time in people with RA onset from 1999 onwards, but not in the general population.
References  Lacaille, D., et al., Improvement in 5-year mortality in incident rheumatoid arthritis compared with the general population—closing the mortality gap. Annals of the Rheumatic Diseases, 2016.
 Myasoedova, E. and S.E. Gabriel, Overview of rheumatoid arthritis and mortality in relation to cardiovascular disease, in Handbook of Cardiovascular Disease Management in Rheumatoid Arthritis, A.G. Semb, Editor. 2017, Springer International Publishing: Cham. p. 1-17.
Disclosure of Interests Kiana Yazdani: None declared, Hui Xie: None declared, Antonio Aviña: None declared, Yufei Zheng: None declared, Michal Abrahamowicz: None declared, Diane Lacaille Grant/research support from: Bristol-Myers Squibb and Eli Lilly Canada
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.