• An appropriate type of regression analysis should be performed and biases at the medical decision levels and at the lower and upper limits of the measuring interval should be calculated along with the confidence interval of each bias estimate.
• Total error (central 95% region of observed differences between the candidate test and CM) should be estimated for the entire measuring interval of the candidate test, and for 3 subintervals (low, medium and high), as described in CLSI EP21.
• The measuring interval of the CM should be at least as wide as the measuring interval of the candidate test. If there are samples with either candidate test or CM values outside of the corresponding measuring intervals, these samples should be analyzed separately.
• If one of the medically important points of the candidate test includes the Limit of Blank (LoB)/Limit of Detection (LoD)/Limit of Quantitation (LoQ), then some additional calculations for samples with very low levels of analyte may be needed for appropriate evaluation and comparison of the LoB/LoD/LoQ of the candidate test in the hands of untrained operators (see CLSI EP17).

‍