The authors of the article cited above respond:

摘要

To the Editor: Krouwer (1) makes a valid point that total error for a measurement procedure must include analytical nonspeci-ficity influences (referred to as "random patient interferences" by Krouwer) as well as mean calibration bias and imprecision components. Fig. 3 in the report from the Laboratory Working Group (LWG) of the National Kidney Disease Education Program showed boundaries for combinations of bias and imprecision that would contribute no more than a 10% change in root mean square error for estimates of glomerular filtration rate calculated from a serum creatinine measurement (2). The emphasis in that report was on reducing the calibration bias and imprecision of creatinine measurement procedures to reduce the variability in estimates of glomerular filtration rate.