Results of a nationwide proficiency test for carcinoembryonic antigen.

A proficiency testing survey for carcinoembryonic antigen (CEA) was conducted by the Center for Disease Control. The results from 125 laboratories throughout the United States indicate that many laboratories perform satisfactorily, but some laboratories need substantial improvement. Failure to follow manufacturer's recommendations and failure to heed the indications of "out of control" control results were two of the reasons for poor performance. Results reported for samples with CEA levels of larger than or equal to 20 ng/ml showed that the direct method produced significantly higher values than the indirect method on either whole of diluted plasma. Almost one-fourth of the results reported in this survey were placed in the wrong nominal group. It was determined that the results were log normally distributed and, consequently, that statistical methods that are appropriate for this distribtuion should be used for the analysis of CEA results. Most of the variation observed was the result of poor comparability between laboratories rather than lack of precision within the laboratory. This indicates that better performance could be achieved by better standardization and closer adherence to established procedures.