Evaluation of methods and protocols for hemoglobin screening of prospective whole blood donors.

Fifty-nine volunteers were screened, as if for blood donation, with the use of simultaneous fingerstick (FS), earlobe (EL), and venipuncture (VP) samples tested by copper sulfate density and two instruments for rapid assay, Statcrit and HemoCue. The "true" hemoglobin was determined on the venipuncture sample by cyanmethemoglobin assay. Eligibility (pass/fail) was established for each site/result with the use of standard and investigational cut-offs. Results were used to predict the performance of two-method protocols beginning with copper sulfate followed by FS or EL using Statcrit or HemoCue. Individually, the method sensitivities for hemoglobin below cut-off were low (12.5-62.5%) with the use of standard cut-offs. Assuming 5% prevalence of low hemoglobin, two-method protocols would inappropriately pass more than half of those with low hemoglobin, whereas the proportion of deferred donors with adequate "true" hemoglobin ranged widely (2.8-72.3%). For some methods, investigational cut-offs achieved improved sensitivity with no effect on specificity. Despite standard hemoglobin "requirements" for blood donation, the ability of the investigated methods to correctly classify donors is poor and varies considerably with method and protocol. With cut-offs as specified, the standards for FS and EL samples are not equivalent; the EL cut-offs are too low and the method is inherently less sensitive. Standardization may be better achieved by specifying both required hemoglobin and minimum performance for screening methods.