Reference Standards and Comparators
The choice and application of a reference standard or comparator shape what an evaluation can establish.
#A reference standard defines the target condition
FDA uses the STARD definition of a reference standard as the best available method for establishing whether the target condition is present. It may be one method or a combination that includes clinical follow-up. STARD distinguishes this from an error-free gold standard.
Read the target condition and how the reference classification was obtained. Best available does not mean infallible. A familiar test name alone does not explain the algorithm, interpretation or follow-up used to classify participants.
Evidence: FDA (March 2007): reporting diagnostic-test evaluation results / STARD 2015: explanation of diagnostic-accuracy reporting items
#Agreement is not automatically accuracy
FDA distinguishes comparison with a reference standard from comparison with another method that is not a reference standard. For the latter, its guidance recommends agreement terminology rather than directly interpreting the comparison as sensitivity and specificity. It warns that two tests can agree while both are wrong.
Read what the comparator establishes before accepting an accuracy claim. Disagreement does not by itself identify which test is mistaken. These distinctions come from FDA's March 2007 guidance for qualitative diagnostic-test reporting, not a complete rulebook for every measurement study.
Evidence: FDA (March 2007): reporting diagnostic-test evaluation results
#Execution and available information matter
STARD asks for enough detail about the index test and reference standard to allow replication, and the rationale for the chosen standard. It also asks which clinical information and test results were available to the people performing or interpreting each assessment. Such information can influence interpretation.
Read who assessed the result, how it was assessed and what they knew. Naming a reference standard does not settle how independently it was applied. This reporting guidance helps readers inspect methods; it does not certify that a reported design is sound.
Evidence: STARD 2015: explanation of diagnostic-accuracy reporting items / STARD 2015: diagnostic-accuracy reporting statement
#Incomplete reference assessment needs explanation
FDA describes verification bias when the reference standard is not applied fully and the analysis fails to account for that design. STARD asks how missing and indeterminate test results were handled, and for the interval and interventions between the index test and reference standard.
Read who received which assessment and whether missing results or changing clinical circumstances could affect the comparison. Specialized designs need suitable analysis. Mynd is not selecting a clinical reference method, prescribing that analysis or validating a particular test here.
Evidence: FDA (March 2007): reporting diagnostic-test evaluation results / STARD 2015: explanation of diagnostic-accuracy reporting items
Source note
The sections above were checked against the linked sources. No clinical review has been performed. This is general research education, not a clinical guideline.