Author Affiliations: University of California, San Francisco (Dr Peralta) (firstname.lastname@example.org); and San Francisco VA Medical Center, San Francisco (Dr Shlipak).
In Reply: Dr Fabri points out that the robustness of a predictive model would be enhanced if different cutoff values were used to define CKD, rather than the clinical cut points that we used. However, our study was intended to provide the clinician with a framework that could easily be used to risk-stratify persons classified as having CKD by creatinine level. Moreover, our data suggest that multiple markers may identify persons at high risk for death and end-stage renal disease who are missed by assessing only creatinine level. If our intention had been to evaluate the best predictive model, then we likely would have found that different cut points for each marker might have been more appropriate; we previously found that even mildly reduced estimated GFR (eGFR) identified by cystatin C was associated with adverse events.1 We believe having different cut points for each marker would be less readily applicable in clinical practice.
Peralta CA, Shlipak MG. Biomarkers for Detecting and Risk-Stratifying Chronic Kidney Disease—Reply. JAMA. 2011;306(6):611-612. doi:10.1001/jama.2011.1112