In Reply: We agree with Dr Sierra-Johnson and colleagues that the modest number of CHD events in women in our investigation limits our ability to detect small differences in predictive performance of different lipids, and additional studies of larger samples are warranted. Regarding the criticism of the use of a composite CHD end point for risk prediction, this approach is consistent with our prior publications1
and other risk prediction algorithms in the literature.2 Furthermore,
such an approach reflects clinical practice, in which the objective is to predict and prevent risk of all CHD events and not just one of its individual components. We did not analyze “hard”
CHD events for this reason and because of a further reduction in statistical power to detect associations (due to a reduction in numbers of events),
a concern also voiced by the writers.
Ingelsson E, Pencina MJ, D’Agostino RB, Vasan RS. Utility of Different Lipid Measures to Predict Coronary Heart Disease—Reply. JAMA. 2008;299(1):35-36. doi:10.1001/jama.2007.5