[Analytical uncertainty--how wrong can a laboratory result be?].

Abstract

Some uncertainty encumbers the outcome of all laboratory tests. Patient conditions and handling of specimens, as well as analytical variation and systematic error will affect the results. In order not to confuse the interpretation of test results, the analytical standard deviation should not exceed one half of the intraindividual biological standard deviation, and systematic error should not exceed 1/16 of the reference interval. However, these goals cannot always be achieved. Moreover, analytical control procedures have limited ability to detect errors in the analytical process. After one analysis of control material, using +/- 2 standard deviations (analytical) from the mean as acceptance limits, the magnitude of a systematic error must be 3.3 times the analytical standard deviation in order to be detected with 90% probability. As a result of such error, patients' results can be released with a total error up to five times the analytical standard deviation. More complicated control procedures may give smaller total error. Practising physicians should be familiar with the variations in lab results, and interpret the results accordingly. Lab results that conflict with results of other investigations should be used with caution.