Abstract 5906: Red Cell Distribution Width (RDW) is a Strong Predictor of Outcome and is Weakly Associated With Clinical Variables in Patients at Risk for Cardiovascular Events
Background Red cell distribution width (RDW) is one of the strongest predictors of outcome for patients with coronary disease or heart failure, yet little is known of the basis of this association.
Methods Within the MURDOCK CV study of cardiovascular risk classification in 6447 patients undergoing coronary angiography at Duke University Medical Center for suspected ischemic heart disease from 2001–2007, we examined the univariable associations of clinical variables with RDW and developed a multivariable linear regression model to identify factors independently associated with RDW using the R square selection method.
Results The median RDW for the study population was 13.5% (IQR 12.9%–14.3%) and the reference range for the clinical laboratory was (11.5% to 14.5%). In Cox proportional hazards models of clinical events in our study population, RDW was the 4th strongest predictor of both death (X2 44) and death/MI (X2 46). In univariable linear regression models, 14 variables were significantly associated with RDW (hemoglobin, heart failure severity, BUN, white race, ejection fraction, female, Charlson index, heart rate, valvular disease, age, outpatient status, WBC, smoking, and history of CABG). Hemoglobin explained 14% of the variance in RDW, but no other single variable explained more than 5%. In multivariable modeling considering all available covariates, a model containing female sex, white race, weight, systolic BP, diastolic BP, smoking, valve disease, heart failure severity, history of bypass surgery, history of MI, ejection fraction, outpatient status, Charlson index, BUN, hemoglobin, sodium and WBC provided the best balance R square and number of model variables, explaining 22% of the variance in RDW.
Conclusions RDW is a powerful independent predictor of future death and death/MI in patients referred for coronary angiography. However, only a small proportion of the variance in RDW is explained by readily available clinical characteristics. Thus, further study of the underpinnings of the association of RDW with clinical events may provide novel mechanistic insights into cardiovascular risk.