Cortisol testing goes wrong when the sample does not match the question. There are only two clinical questions worth testing for in most people: is there too much cortisol, which is Cushing syndrome, or too little, which is adrenal insufficiency. Each has its own validated test, its own timing and its own cut-off. The Endocrine Society names both sets and, unusually for a guideline, also names the tests that should not be used.

Testing for too much cortisol

The Endocrine Society guideline on the diagnosis of Cushing's syndrome recommends one of four initial tests: urine free cortisol on at least two collections, late-night salivary cortisol on two occasions, a 1 mg overnight dexamethasone suppression test, or a longer low dose suppression test at 2 mg daily for 48 hours. On the overnight test, a morning cortisol above 1.8 mcg/dL, which is 50 nmol/L, is a positive screen. These tests work because they capture either total daily output or the loss of the normal night-time low point.

Testing for too little cortisol

For suspected adrenal insufficiency, the Endocrine Society guideline uses a morning cortisol below 5 mcg/dL, which is 140 nmol/L, together with ACTH, as a preliminary result suggesting adrenal insufficiency. Diagnosis is confirmed with a cosyntropin stimulation test: 250 mcg given intravenously, with cortisol measured at 30 or 60 minutes. A peak below 18 mcg/dL, which is 500 nmol/L, indicates adrenal insufficiency, and the exact threshold depends on the assay used. This is a real and treatable disease, and untreated it can be fatal.

Tests that do not answer either question

The Cushing guideline states directly: "We recommend against the use of the following to test for Cushing's syndrome: Random serum cortisol or plasma ACTH levels, Urinary 17-ketosteroids, Insulin tolerance test, Loperamide test." That list matters because random cortisol is the test most often ordered by consumer panels. The four-point salivary cortisol curve sold as an adrenal stress index is a separate matter: late-night salivary cortisol is validated for Cushing screening, but the daily curve interpreted as adrenal fatigue is not a recognised diagnostic test.

Question	Correct test	Threshold
Too much cortisol	1 mg overnight dexamethasone suppression	Morning cortisol above 1.8 mcg/dL is positive
Too much cortisol	Late-night salivary cortisol, two samples	Loss of the night-time trough
Too little cortisol	Morning cortisol with ACTH	Below 5 mcg/dL is suggestive
Too little cortisol	Cosyntropin 250 mcg stimulation test	Peak below 18 mcg/dL indicates deficiency
Neither	Random daytime serum cortisol	Recommended against by the Endocrine Society

If cortisol testing has been suggested, ask which of the two questions is being asked and which validated test answers it. If the plan is a random daytime cortisol or a saliva curve read as a stress pattern, ask what treatment would change based on the result.

Interfering factors are common and are the reason screening tests are repeated before they are believed. Oral oestrogen raises cortisol binding globulin and therefore total serum cortisol, and should generally be stopped for around six weeks before dexamethasone suppression testing. Shift work, an irregular sleep schedule and untreated depression can all disturb late-night salivary cortisol. Dexamethasone metabolism is altered by CYP3A4 inducers such as phenytoin, carbamazepine and rifampicin, which produce false positives.

Urine free cortisol requires a complete 24-hour collection and is unreliable in significant renal impairment. In pregnancy, urine free cortisol is preferred over dexamethasone suppression. Where cyclical Cushing syndrome is suspected, repeated collections over time are required, since a single normal result does not exclude it.

Sources

Testing for too much cortisol

Testing for too little cortisol

Tests that do not answer either question

Sources

Related