Accuracy in clinical laboratories participating in regional quality control programs.

Few studies have assessed analytic bias (accuracy) in clinical chemistry laboratories. Using the data base of a large regional quality control program, we studied, during a six-year period, analytic bias of participant "all-methods" results, as well as selected commonly used individual methods, for calcium, cholesterol, glucose, potassium, and sodium. Analytic bias was assessed by comparison with "benchmark" measurements by United States Bureau of Standards definitive methods and/or United States Centers for Disease Control clinical reference methods. The all-methods bias at selected clinically relevant analyte levels was +0.1% at 10 mg/dL for calcium; -0.7% for cholesterol at 200 mg/dL; +4.5% for glucose at 120 mg/dL; +0.5% for potassium at 4 mmole/L; and 0% for sodium at 145 mmole/L. There was sufficient agreement between benchmark and participant all-methods results for these analytes to suggest that the all-methods results may be useful analytic bias benchmarks in future regional quality control programs that use similar lyophilized human serum pools.