Expressions of accuracy in the evaluation of biomedical instrumentation.

This review discusses and compares the most common and appropriate expressions and representations of accuracy in the evaluation of biomedical instrumentation. Instruments used to measure blood pressure illustrate the various expressions of accuracy in common use. Accuracy of biomedical instruments can be defined as a measure of system error. In system error there is a consistent, systematic bias inherent in the method or instrument being tested. Two cases or conditions in which accuracy may be determined were identified and analyzed: 1) static or single-target conditions, and 2) dynamic or continuum-of-target conditions. Expressions of accuracy tested included the mean difference/standard deviation (MD/SD) method and the regression analysis (RA) method. In the static, single-target cases examined, the MD/SD method proved acceptable. In the dynamic, continuum-of-targets cases examined, the RA method proved the most accurate and useful expression of instrumental accuracy. The MD/SD method could be significantly misleading in these dynamic target cases. Since the majority of biomedical instruments are used in dynamic or continuum-of-targets situations, RA is, generally, the method of choice.