Problems and practical considerations in assessing accuracy with NIST SRM 909a: report of defective vials.

During an experimental period of 12 months in 1992-1993, while we were comparing the effectiveness of monthly vs quarterly use of the National Institute for Standards and Technology Standard Reference Material (NIST SRM) 909a as an accuracy material for the projected 30-year Fernald Medical Monitoring Program, we encountered three random defective vials with a glucose recovery of less than 30% of the NIST-assigned value. Analysis with five different multichannel instruments confirmed the original finding. Concomitant glucose recovery from adjacent vials was 97%-104%, as determined by using the same instruments, reagents, calibrators, and quality-control criteria on the same days. Recoveries of uric acid and cholesterol were also low (53-75% and 75-80%, respectively) in the three defective vials. Other analytes were unaffected. Studies to identify the cause of the defective vials were carried out with microbiological, electron microscopic, and biochemical techniques. When used for accuracy studies, each vial of NIST SRM 909a should have a concomitant check for glucose recovery to detect whether the vial is defective.