Abstract 6201: Ankle-brachial Index And Fibrinogen Improve Cardiovascular Risk Prediction And Classification
BACKGROUND: An abnormal ankle-brachial index (ABI) and fibrinogen have been shown to be associated with a higher risk of cardiovascular disease (CVD) but the incremental usefulness of adding them into risk prediction models has not been established.
DESIGN: A Retrospective Longitudinal “Follow-up”.
SETTING: The Atherosclerosis Risk in Communities (ARIC) study, 1987–2001.
PARTICIPANTS: Data from 11,888 White and African American (26%) men and women, aged 45–64 years, without known history of CVD, diabetes, or atherosclerotic vascular disease at baseline, with available standard CVD risk factors, ABI and fibrinogen.
MEASUREMENTS: The main outcome was CVD defined as MI, coronary death or stroke. Multivariable proportional hazards models were constructed with ABI and fibrinogen added to the pool of standard risk factors (age, gender, smoking, LDL and HDL cholesterol, systolic blood pressure, antihypertensive treatment and race). Additionally risk was assessed in different strata defined by abnormal ABI (ABI<0.9 in either leg), abnormal fibrinogen (>=400 mg/dl) and high Framingham risk score (risk >=20% based on a model with only standard risk factors).
RESULTS: During 10-year follow-up, 691 CVD events occurred. All established risk factors were significantly (all p-values <0.0001) associated with the outcome leading to a “standard model” with a C statistic of 0.75. When added to this model, both ABI and fibrinogen were highly significant with respective hazard ratios of 0.85 (95% CI 0.8–0.91) and 1.19 (95% CI 1.12–1.28) (both p-values <0.0001). The C statistic of new model was 0.76. Risk reclassification, as measured by the net percentage of people correctly reclassified to a more accurate risk category (with 6% and 20% used as thresholds) was 3.3% (p-value =0.019). The presence of both abnormal ABI and fibrinogen increased the age, gender and race-adjusted 10-year CVD risk from 5.4% to 25.3% (95% CI 11.6–36.2%) in general population and to 15.0 % (95% CI 1.9–25.9%) in people with Framingham risk <20%.
CONCLUSION: The results suggest that the addition of inexpensive and readily available markers like ABI and fibrinogen to traditional CVD prediction algorithms can improve the risk stratification beyond that offered by a model based on standard risk factors.