First, because of relatively limited statistical power, we restricted the number of covariates included in our statistical models. Covariates in our fully adjusted models were selected a priori on the basis of previously published work that identified these covariates as potential confounders of the association of biomarkers with mortality. For example, previous studies demonstrate that the ankle–brachial index and black race are each associated with increased mortality among persons with PAD (1, 2), even though these characteristics were not statistically significantly associated with mortality in univariate analyses in our study. Nonetheless, we agree with Dr. Fujita that statistical adjustment for HDL cholesterol level is warranted in our analyses, because lower HDL cholesterol levels were statistically significantly associated with increased mortality in our PAD cohort. Therefore, we repeated our fully adjusted statistical analyses of biomarker levels with cardiovascular disease and all-cause mortality, adding adjustment for HDL cholesterol level. Our results were unchanged in the analyses. In all of these fully adjusted models, HDL cholesterol level was statistically significantly or nearly statistically significantly associated with cardiovascular disease mortality.

Second, at baseline, correlation coefficients of CRP level with body mass index, number of cardiovascular diseases, and HDL cholesterol level were 0.112 (P = 0.033), 0.030 (P = 0.56), and –0.076 (P = 0.141), respectively.

Finally, we agree that some past studies suggest that CRP level may not add incrementally to traditional atherosclerotic disease risk factors in predicting cardiovascular events (3). However, the purpose of our study was to determine whether elevated biomarker levels were more strongly associated with near-term than later-term all-cause and cardiovascular disease mortality in persons with PAD. We did not attempt to address the question of whether biomarkers add to the predictive ability of traditional atherosclerotic risk factors for cardiovascular events. Similarly, data are insufficient to provide recommendations on a threshold CRP or serum amyloid A value that predicts near-term risk for cardiovascular events in persons with PAD. Further study involving a larger sample is needed to confirm our findings and address whether a threshold biomarker level is appropriate for identifying patients with PAD who are at high risk for near-term events.