The general method is to compare the new instrument with another known accurate method by measuring an identical group of blood samples and then comparing the statistical mean absolute deviation (MAD, or MARD) and the relativity (correlation coefficient) of their results. When MAD is less 15%, it can be used as a primary reference, the output signals of which may be considered a direct reference standard for the patient’s medicine dosage. Throughout the world, the most common standard reference instrument is YSI glucose analyzer (Yellow Spring Instrument, Glucose Analyzer). The deviation and the correlation coefficient are an absolutely accurate evaluation of measuring methods.
In clinical practice, there is another accurate appraisal method which is called Error Grid Analysis. Introduced by Dr. W.L. Clarke and his colleagues, it evaluates the accuracy of a single glucose self-test. According to the clinical significance and the possibility of error, the dispersed X-Y related chart is divided into the areas of A, B, C, D, E, and etc. The data in area A is clinically accurate; in area B, acceptable; in areas C, D, etc. the data is unreliable. While there may be minimal inaccuracies in area B, these are not significant enough to warrant rejection of the data; thus, areas A and B are the clinically significant results.