It has been appropriately pointed out by Bock that a 5% deviation from "true" value may be too restrictive, especially when only one or two samples are being evaluated at a reference and at a clinical laboratory. The value of 9.5% recommended by the National Cholesterol Education Program is perhaps more practical and allows for each laboratory to be within 5% of the target value. We used the 5% figure in the editorial1 because we believed that if a series of cross-checked samples is consistently more than 5% off of the true value (Centers for Disease Control laboratory value), then something is wrong in the laboratory being evaluated. We appreciate Bock's comments on this point.
Dujovne CA, Harris WS. Accuracy of Cholesterol Measurements-Reply. Arch Intern Med. 1991;151(8):1677. doi:10.1001/archinte.1991.00400080155043