This is a remarkable paper and has big implications for testing and the diagnostic algorithm, especially given who the authors are. The blunt takeaway is do not use salivary cortisol for adrenal Cushing's because it doesn't work. They recommend dex test instead but we know that test has problems too--yes, even in adrenal cases. It's not clear to me if this is generalizable to all mild Cushing's, nor does it appear like cyclical or episodic Cushing's was considered. The other thing is that the technically "better" assay (LCMS) has worse sensitivity than the older, cheaper one (EIA). This is the same thing we saw with the UFC where the older RIAs cross-reacted with cortisol metabolites, so we traded it for the tandem mass spec that produces fewer false positives and more false negatives. I've made all of these points before with my local endo who cited one of the authors of this paper to refute me!