We would like to thank McCulloch et al for their comments and try to respond to the issues they raised regarding our recent study.

First, our aim in the study was to examine the association between 25-hydroxyvitamin D (25[OH]D) and C-reactive protein (CRP) in a cross-sectional setting. Subjects taking 25(OH)D supplements are likely to have higher serum levels of 25(OH)D at the time of CRP measurement. Thus, 25(OH)D supplementation should be reflected in their serum levels. The only situation in which this may not be true is when a subject begins taking 25(OH)D only a few days before the sample blood draw for a study. However, this is likely to happen so infrequently that it is unlikely to have had any effects on our results.

Second, we chose to present the y axis using a logarithmic scale to emphasize that our model assumed a linear relation. By not using log-transformed CRP values, the linear assumption would have been obscured.

Third, the participants with 25(OH)D levels <21 ng/ml had a mean CRP level of 0.22 mg/dL, whereas those with 25(OH)D levels >21 ng/ml had a mean CRP level of 0.17 mg/dL. It is important to realize that these means included subjects with a wide range of 25(OH)D levels. Because the slope of the relation between CRP and 25(OH)D was steeper for subjects with 25(OH)D levels <21 ng/ml, overall CRP levels were higher at the lower levels of 25(OH)D. Beta coefficients represent change in CRP per unit change in 25(OH)D and not means after adjusting for other variables.

Fourth, one can indeed follow a statistically significant approach for model selection, as suggested by McCulloch et al. However, our model selection approach was based on biologic significance between variables selected a priori with the outcome. We believe that a biologic approach provides more realistic and biologically meaningful adjustments in assessing independent effects of 25(OH)D on CRP.

Fifth, our study was limited in examining the relation between 25(OH)D status and CRP levels, which was reiterated in the “Results” and “Discussion” sections as well. We do plan to examine all-cause and cardiovascular mortality with 25(OH)D status in asymptomatic subjects as well.

Last, as mentioned in our report, other biomarkers of inflammation were available for a relatively small number of subjects, which would have limited the number of participants in the analysis, so we chose to stay with serum CRP for now.