Sep 17, 1990 - bias or analytical inaccuracy, 5mea8 is the stable mea- surement standard ..... specific analytical systems, but the present model is suitable for ...
CLIN.CHEM.37/5, 656-661 (1991)
Laboratory Process Specifications for Assuring Quality in the U.S. National Cholesterol Education Program James
0. Westgard,’
Per Hyltoft Petersen,2 and Donald A. Wiebe’
We have assessed the laboratory specifications necessary for ensuring that cholesterol testing processes satisfy the quality required by the U.S. National Cholesterol Education Program (NCEP). A model for setting process specifications has been developed to relate the NCEP guidelines for medical interpretation of a cholesterol test to the pre-analytical and analytical variables that can affect a test result. Using this model, we derived specifications for the imprecision (coefficient of variation, CV, or standard deviation, s) and inaccuracy (bias) that are allowable under stable operation, as well as the qualitycontrol procedures (control rules and number of control measurements) that are necessary to detect unstable operation. The NCEP goals of an allowable CV 3% and an allowable bias no greater than ±3% are inadequate for assuring the quality of an individual or single cholesterol test when monitoring performance with many of the statisticalquality-control procedures currently used in the U.S. With quality-control procedures having two control measurements per run, a CV of 3% is allowable only when bias is zero; a CV 2% is necessary if bias is ±3%. With quality-control procedures having four control measurements per run, a CV of 3% is allowable when bias is ± 1.5%; a CV 2.5% is required if bias is as large as ±3%. For two serial tests, the NCEP 3% goals are adequate for current quality-control procedures having four control measurements per run. Additional Keyphrases:
quality control
-
precision
statistics
control rules
Specifications for inaccuracy, imprecision, and quality control are necessary to establish analytical processes capableof ensuring the quality required by the National Cholesterol Education Program (NCEP) (1). The initial NCEP guidelines (1987) state that imprecision is acceptable if the CV is s5% and that inaccuracy is acceptable if the bias is 2390 mg/L (6.18 mmol/L)1. Likewise, a patient having a true “high” homeostatic set point of 2400 mg/L should have only a 5% chance of
