Comparison of pooled fresh frozen serum to proficiency testing material in College of American Pathologists surveys: cortisol and immunoglobulin E.

CONTEXT: The College of American Pathologists (CAP) provides proficiency testing (PT) surveys to laboratories around the world.
OBJECTIVES: To compare diagnostic assay methods for serum/plasma cortisol and immunoglobulin (Ig) E in terms of their bias and precision, to determine how well CAP PT specimens simulate human serum, and to reassess proficiency test grading criteria in light of these findings.
DESIGN: A participant-blinded, prospective trial. One vial of pooled fresh frozen serum (FFS) and 4 different admixtures of PT material (PTM) were sent to laboratories participating in PT surveys. PARTICIPANTS: Laboratories providing cortisol (>1000) or IgE (>230) results among the subscribers to the CAP surveys, Ligand (General) 2003, set K/KN-A and Chemistry 2003, set C-C. MAIN OUTCOME MEASURES: The main outcome measures were (1) bias among laboratories using the same method (peer groups), defined relative to the median of method means (MedMM); (2) imprecision as measured by the SD and coefficient of variation (CV) about each method mean; and (3) total error across laboratories for the FFS cortisol results, defined as |Bias Relative to Reference Method| + 2 SD.
RESULTS: Cortisol method biases, relative to MedMM, ranged from -22% to 9% for the FFS challenge and from -24% to 36% for comparable PTM challenges. The method biases, relative to the reference method, ranged from -3% to 19% for the FFS challenge. The cortisol method CVs ranged from 4.2% to 13.6% for the FFS challenge and from 4.7% to 12.7% for comparable PTM challenges. Total error across laboratories ranged from 1.4 to 6.9 microg/dL (39 to 190 nmol/L) for the FFS challenge. Immunoglobulin E method biases, relative to MedMM, ranged from -8% to 9% for the FFS challenge and from -7% to 5% for comparable PTM challenges. The IgE method CVs ranged from 3.6% to 6.7% for the FFS challenge and from 3.4% to 9.8% for comparable PTM challenges.
CONCLUSIONS: The bias for cortisol results was less with FFS than with PTM, but imprecision was comparable. The FFS MedMM was 8.5% higher than the reference value. Fresh frozen serum and PTM bias and imprecision for IgE methods were each less than 10%. Because some of the methods demonstrated greater bias when analyzing PTM than FFS, peer group grading of both these analytes is appropriate.