LAUSR

We thank Krupp et al. for their insightful and well-thought comments on our article. We agree with them that there were some methodological flaws in our article. Studies on the association of elevated fibroblast growth factor-23 (FGF-23) levels with cardiovascular events and mortality from the general population were relatively sparse and produced less consistent results. However, considering the facts that associations of elevated FGF-23 levels with risk of cardiovascular disease, heart failure, stroke, and cardiovascular or all-caused mortality were still statically significant (Fig. 1) after we removed the studies [1–4] reporting risk estimates increased from reference category to second quartile, decreased in third, and increased again. A recent publication of the Framingham Heart Study [5] was included in our update meta-analysis. In addition, the predictive value of the FGF-23 was supported by a continuous variable analysis [4]. The included studies calculated the hazard for several categories of FGF-23 levels. We only pooled risk estimate for the top versus the lowest category of FGF-23 levels. A dose-response analysis will more clearly show the association of FGF-23 levels with cardiovascular and mortality outcomes. However, not all the original studies clearly reported the median value of FGF-23 levels or total persontime for incidence data. Lack of such data prevents us to conduct a dose-response relationship analysis. This issue will be considered in