The correct use of terminology to describe performance results is important to ensure safe and efficient use of a diagnostic device. Where possible, this guide uses internationally recognized terminologies and definitions, collected in the Harmonized Terminology Database of the Clinical and Laboratory Standards Institute (CLSI). 1 This guide also uses terms defined in the STARD (STAndards for Reporting of Diagnostic Accuracy) initiative. 2 The STARD initiative refers to diagnostic accuracy studies. Although the STARD initiative does not specifically target studies to demonstrate the equivalence of diagnostic devices, many of the reporting concepts are nevertheless applicable. Total percentage compliance alone does not sufficiently characterize the conformity of a test with the non-reference standard. Two different 2×2 tables can have the same global highlight with the same value for (b+c), but have very different values for b and c individually. It is therefore more appropriate to point out a pair of agree measures, a positive percentage agreement (PFA) and a negative percentage agreement (PFA): the choice of the comparative benchmark and the methods of comparison and reporting are influenced by the existence and/or practical applicability of a reference standard. Depending on the availability of a reference standard, the FDA makes the following recommendations on the choice of comparison benchmark: You should not use results modified or updated by different dissolution to assess the sensitivity and specificity of a new test or an agreement between a new test and a non-reference standard. Comparing a new test with a non-reference standard does not achieve real performance.

If the new test is better than the reference standard, the agreement will be bad. Alternatively, the agreement could be bad, given that the non-reference standard is quite accurate and the new test is imprecise. There is no statistical solution to determine what the scenario of the real situation is. Table 6C shows that the percentage of concordance between the new test and the non-reference standard remains 88.2% (39+6)/51) and 98.2% ((4+660) /676) for unconditional subjects (standard reference column−). However, the total over-compliance in percentage associating subjects with or without conditions is 97.5% ((39+6+4+660) /727), higher than the initial 95.9%. What shows a more dramatic difference is that the positive overcommit rate is 76.8% (43/(43+13)) versus 90.9%, and the negative percentage rate is slightly higher (99.2% (666/(666+5)) versus 97.2%).