Article Text

SAT0044 The Role of C-Reactive Protein or Erythrocyte Sedimentation Rate in Predicting Cardiovascular Outcomes in Rheumatoid Arthritis: Analysis of Data from US Managed Care Organization
  1. J. An1,
  2. E. Alemao2,
  3. T. Cheetham1,
  4. K. Reynolds1,
  5. H. Kawabata2,
  6. D. Solomon3
  1. 1Kaiser Permanente Southern California, Downey
  2. 2Bristol-Myers Squibb, Princeton
  3. 3Brigham and Women's Hospital, Boston, United States


Background Patients with RA are at a 2- to 3-fold higher risk for cardiovascular (CV) events. Disease activity in RA is often assessed by C-reactive protein (CRP) or erythrocyte sedimentation rate (ESR), and there is growing evidence that the disease activity in RA is associated with increased CV risks.

Objectives To test whether CRP or ESR improves CV risk prediction in RA patients.

Methods Patients aged ≥18 years and with ≥2 RA diagnoses with a DMARD were identified between 01/01/2007 and 12/31/2011 within Kaiser Permanente Southern California. Individuals were followed from the first RA diagnosis or prescription date (index) until death, disenrollment, study end period (12/31/2011) or first CV outcome (myocardial infarction, angina, stroke, transient ischaemic attack, intermittent claudication, heart failure or coronary heart disease death). The base-case prediction model included traditional predictors of CV events similar to the Framingham Heart Study model (FHS).1 Patients with a previous history of CV events or patients with any missing values were excluded from the analysis. Event-free survival distributions by baseline CRP tertiles (≤7.4, 7.4–17.8, ≥17.8 mg/L) or ESR tertiles (male: ≤13, 13–36, ≥36 mm/h; female: ≤20, 20–40, ≥40 mm/h) were compared using a Kaplan–Meier log rank test. Multivariable Cox proportional hazards regression models were used to test the relation between risk predictors and CV outcomes. Harrell's C-index and net reclassification improvement (NRI) were used to evaluate the discriminatory ability of the prediction model, including CRP/ESR compared to the FHS prediction model.

Results A total of 7277 RA patients were included in the cohort with median follow-up of 3.8 years. The mean (SD) age was 55.7 (13.9) years, with 79% being female. At baseline, 54% had hypertension or were taking antihypertensive medications, 51% had dyslipidaemia, 14% had diabetes and 12% were current smokers. Only 39% of the study population had baseline CRP values, with a mean (SD) of 20.8 (28.5) mg/L, whereas 99% had baseline ESR values, with a mean (SD) of 31.3 (28.0) mm/h for males and 35.0 (25.3) mm/h for females. CV event-free survival was significantly lower in patients with a high CRP or ESR range (log rank test: p<0.001). Using a base-case FHS prediction model, all of the factors except for dyslipidaemia were significant and positively correlated with the risk of CV outcomes (C-index: 0.7762). Including CRP/ESR as an additional factor, the new model slightly improved the model discrimination (C-index: 0.7784, NRI: 1.112%). As well, there was a trend towards a higher CRP/ESR being correlated with an increased risk of CV outcomes (hazard ratio [HR] of high vs low CRP/ESR=1.25; 95% CI 0.98, 1.61, HR of medium vs low CRP/ESR=1.28; 95% CI 0.99, 1.65).

Conclusions Baseline CRP or ESR were marginally associated with an increased risk of CV outcomes and the inclusion of CRP/ESR in the FHS model resulted in a small improvement in the reclassification of CV risk.


  1. Wilson PWF. Circ Cardiovasc Qual Outcomes 2008 November

Disclosure of Interest J. An Grant/research support: BMS, Merck, Genentech, E. Alemao Shareholder of: BMS, Employee of: BMS, T. Cheetham Grant/research support: BMS, Gilead, K. Reynolds: None declared, H. Kawabata: None declared, D. Solomon: None declared

DOI 10.1136/annrheumdis-2014-eular.1742

Statistics from

Request permissions

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.