Internal quality control, proficiency testing, and the clinical relevance of laboratory testing.

Estimates of instrument-specific analytical performance for measurement of four chemistry analytes (calcium, cholesterol, creatinine, and glucose) were derived from 21 quality control programs that utilized the College of American Pathologists Quality Assurance Service and from 12 College of American Pathologists Comprehensive Chemistry Survey sets. The quality control-derived estimates of analytical bias were larger and level-dependent when compared with the survey results. Possible explanations for these differences include "matrix effects" due to glycol stabilization of the control materials studied and analyst bias when testing survey specimens. A number of instruments from the Quality Assurance Service pools failed to meet the allowable analytical error goals for calcium and creatinine that serve as the basis for the "fixed limits" evaluation criteria of the survey program. The methods for glucose analysis met the prescribed evaluation limits. Cholesterol analyses need significant improvement to meet the analytical requirements of long-term testing of individuals.