Evaluation of blood lead proficiency testing: comparison of open and blind paradigms

Most proficiency testing (PT) programs operate with an open design in which clearly identified performance samples are distributed directly to participating laboratories on a shipping schedule announced in advance. In this study, we examine the effectiveness of assessing clinical laboratory performance for blood lead with an open PT by comparing its results with a double-blinded testing protocol.

Aliquots from up to 72 blood lead performance pools from the New York State Department of Health and the Wisconsin State Laboratory of Hygiene were disguised as routine patient specimens and submitted in two phases to up to 42 certified clinical laboratories for blood lead analysis. These 42 laboratories also received aliquots of the same performance samples for blood lead analysis directly from the "open" PT program provider.

Data reported under blind and open strategies were scored against acceptable target ranges using the Clinical Laboratory Improvement Amendments of 1988 (CLIA '88) criteria established for blood lead, i.e., +/- 0.19 micromol/L (+/- 4 microg/dL) or +/- 10%, whichever is greater. Performance differences between the strategies were also assessed. We found that 17.7% of all blind PT results were classified as unacceptable compared with only 4.5% of open PT results (P <0.001). In phase 1, 13 of 22 laboratories (60%) exhibited a statistically significant difference (P <0.05) between their blind and open PT performances, although for 6 laboratories the poorer blind performance may not necessarily have led to unsuccessful PT participation under CLIA '88 criteria. Seven (32%) laboratories had unsuccessful aggregate performance (<80%) under blind testing while maintaining successful performance in open testing. Of these seven, two had gross discrepancies motivating further investigation.

The data suggest that although approximately 60% of clinical laboratories make special efforts to improve analytical performance on open PT samples relative to performance achieved for routine patient specimens, in most cases the differences are clinically insignificant and would not likely affect cumulative PT performance. Occasional use of blind PT may deter the inclination to treat performance samples more carefully.