Letters Section Editor: Stephen J. Lurie,
MD, PhD, Senior Editor.
In Reply: Dr Wierzbicki and colleagues suggest
that inclusion of hsCRP might increase the predictive accuracy of the FRS
to a greater extent than CACS. We have examined the relative predictive power
of hsCRP and CACS1 and found that both hsCRP
and CACS independently contribute to coronary event prediction. In response
to their suggestion to evaluate the predictive ability of hsCRP level in addition
to FRS, we calculated the ROC area for a model that includes FRS plus hsCRP
level (ROC area = 0.64) and found that this model was not significantly better
than FRS alone (ROC area = 0.63, P = .09) and had
poorer discrimination than that for FRS plus CACS (ROC area = 0.68, P<.01). In addition, we note that use of risk models
combining Framingham risk and hsCRP (or other factors) are neither standard
nor widely available at present. However, we agree that they warrant further
development and critical evaluation.
Detrano RC, Greenland P, Doherty TM, Azen SP, LaBree L. Use of Coronary Calcification Scores to Predict Coronary Heart Disease—Reply. JAMA. 2004;291(15):1831. doi:10.1001/jama.291.15.1832-b