Home/Cortisol 101/Why cortisol disorders are hard to diagnose

Medical information

Why cortisol disorders are so hard to diagnose.

Cushing's syndrome takes an average of 34 months to identify. Adrenal insufficiency is given a wrong diagnosis first in roughly two thirds of cases. Neither of those numbers is a story about inattentive doctors. They are the predictable output of five specific problems in how cortisol is measured, every one of them documented in the endocrine literature. Here they are, and here is what they mean for the result sitting in your file.

Written by the Auromone team · Sourced to the Endocrine Society clinical practice guidelines, Endocrine Reviews, and peer-reviewed literature, each claim linked below · Last updated 13 July 2026

The short answer

Five structural reasons. The symptoms are ordinary and belong to a dozen commoner conditions. Cortisol runs on a steep daily curve, so the hour of the sample is part of the result. Some cortisol excess is cyclic and vanishes between episodes. Different laboratory assays return different numbers from the same sample. And a standard blood test measures total cortisol, which oral estrogen inflates and low blood protein deflates.

The counterweight matters as much as the list. Most people who read a page like this do not have a cortisol disorder, and treating "hard to diagnose" as evidence that you have one is an efficient way to delay the diagnosis you actually need. Read the warnings first.

This page explains why a class of medical conditions is difficult to diagnose. It is not medical advice, it is not a diagnosis, and no wearable or at-home kit can diagnose a cortisol disorder. Read the emergency guidance below.

Read this before anything else

An article about the limits of testing is exactly the kind of article that can send a sick person home to wait, or send a well person chasing the wrong hormone for a year. So the warnings come first, before any of the explaining.

1. If you take a steroid, do not stop it because of anything on this page, and when you are ill the dose usually has to go up rather than stay flat. Abruptly stopping corticosteroids after prolonged use is a recognized cause of adrenal crisis, which is life-threatening. The dose belongs to the doctor who prescribed it. Published guidance on preventing adrenal crisis sets out two sick-day rules: double the usual oral glucocorticoid dose during fever, during any illness that puts you to bed, and whenever you need antibiotics; and inject rather than swallow during severe illness or persistent vomiting. If you are vomiting, you cannot absorb an oral steroid tablet, which makes being unable to keep medication down an emergency in itself, not a reason to lie down and rest. If you have diagnosed adrenal insufficiency you should already have an injectable hydrocortisone kit, training to use it, and a steroid emergency card. If you do not have all three, ask for them by name at your next appointment, and ask your prescriber to write your sick-day plan down. The full picture is in what coming off prednisone does to your own cortisol production.

2. These need emergency care now, not an appointment, and not tomorrow:

  • Severe weakness, drowsiness or confusion, severe vomiting and diarrhoea, severe pain in the abdomen, lower back or legs, or collapse. This is an adrenal crisis, the emergency form of low cortisol. It looks exactly like a bad stomach bug, because a stomach bug is its most common trigger: gastroenteritis precipitates 35 to 45% of episodes and fever a further 17 to 24%. If you carry an emergency hydrocortisone injection, inject 100 mg into the muscle now, then call emergency services. If you do not carry one, call emergency services and say: "I may have adrenal insufficiency and may need hydrocortisone." The Endocrine Society recommends immediate parenteral (injected) hydrocortisone, 100 mg, before diagnostic results are available. Treatment must not wait for an assay, which is the one place in this entire story where the testing problem has already been solved.
  • Weight loss with severe muscle weakness, blood pressure that will not come down, and low potassium. This is the pattern of ectopic ACTH syndrome, in which a tumour outside the pituitary drives the adrenal glands, and small cell lung cancer is one of its recognized sources (7 of 39 ectopic cases in the series below, 18%; bronchial carcinoids are the other leading cause of ectopic ACTH). In that series of 7 small-cell patients, potassium was severely low (2.12 mmol/L, against 2.7 in the other causes), all 7 had high blood pressure, all 7 had profound muscle weakness, and all 7 lost weight (median 5 kg). Only 1 of the 7 had stretch marks. 5 of the 7 developed infections, 2 developed sepsis, and mean survival was 2.3 months. That is not a testing puzzle to work through over months. Ask for electrolytes and a chest image now, not another cortisol result.
  • Sudden breathlessness, chest pain, or a swollen, painful calf. Cortisol excess makes blood clot more readily. People with Cushing's syndrome carry roughly 10 times the general-population risk of venous thromboembolism, highest in the first year after diagnosis. Call emergency services.
  • A sudden, severe headache with blurred or double vision, a drooping eyelid, or loss of side vision. Most cortisol excess begins in a pituitary tumour, and a sudden bleed into one, called pituitary apoplexy, is a neurosurgical emergency that also shuts down cortisol production within hours. Call emergency services. Slowly worsening loss of side vision is not tonight's emergency, but it is a this-week appointment, not a this-year one.
  • Feeling suddenly and seriously unwell, with or without a fever. The Endocrine Society states that severe hypercortisolism "impairs immunity and predisposes to severe, systemic infection and/or sepsis due to bacterial, fungal, and opportunistic pathogens." Sepsis kills by the hour. Go to an emergency department or call emergency services now, tell them you may be immunosuppressed by cortisol or by steroid medication, and do not wait to see whether a fever appears.
  • Thoughts of harming yourself. Depression is reported in 55 to 81% of people with Cushing's syndrome and psychosis in about 8%, and a psychiatric label is one of the two commonest wrong first answers given to people who turn out to have adrenal insufficiency. This is not something to sit on while tests are arranged. In Canada and the United States, call or text 988. Elsewhere, call your local emergency number.

Every sign above is useful as a positive. Not one of them is safe as a negative

On a page about diagnostic failure this is the part that matters most, and it is the part that gets skipped. A sign worth acting on when it is present is almost never worth trusting when it is absent.

  • No fever does not rule out an adrenal crisis or a serious infection. Steroid medication and high cortisol both suppress the inflammatory response that produces a fever. Never use a normal temperature to talk yourself out of getting seen.
  • No darkening of the skin does not rule out adrenal insufficiency. The Endocrine Society lists hyperpigmentation as a feature of primary disease only. If the problem sits at the pituitary, or is caused by steroid medication, the skin looks completely normal.
  • Normal potassium does not rule out a lung tumour, and normal sodium does not rule out adrenal insufficiency. Each removes a clue, not a possibility.
  • Not looking like the photographs does not rule out cortisol excess. In the lung cancer series above, the round face and the purple stretch marks were mostly absent. Their absence was the norm, not the reassurance.
  • A normal cortisol result does not rule out a cortisol disorder, which is the whole subject of this page. It also does not rule out anything else, because it did not look. One hormone at one moment is not a clean bill of health.

And the trap this page sets for its own reader: "hard to diagnose" is not "you have it." The move that follows from everything below is never to reject a result and go quiet. It is to go back with a more specific question, and to keep the rest of the differential open while you do.

No wearable and no at-home kit can diagnose any of this, ours included. We make a cortisol sensor, so we should be the ones to say it. What follows is a description of clinical medicine, not an argument for a device.

Five groups quietly decide this page is not about them

  • Men. Cushing's syndrome is diagnosed more often in women, and the illustrations follow the epidemiology, so men read the material and put it down. In the small cell lung cancer series above, 5 of the 7 patients were men.
  • People whose steroid does not come in a tablet. Inhalers, nasal sprays, creams, eye drops and joint injections all reach the bloodstream, and all belong on the drug history that precedes a cortisol test. If you take a steroid by any route, the drug-history sections of this page are about you.
  • People who already had a normal test. That result may well be a genuine answer, and for most readers it will be. It may also be a sample taken at the wrong hour, run on a different assay, or drawn during a quiet phase of a cyclic disease. Which of those it is cannot be settled by reading. It is settled by going back with the specifics below.
  • People whose drug is not a steroid at all. Long-term opioids, megestrol, and the immune checkpoint inhibitors used in cancer treatment (ipilimumab, nivolumab, pembrolizumab) can all suppress the pituitary–adrenal axis and cause adrenal insufficiency. None of them feels like a steroid to the person taking it. All of them belong on the drug history.
  • People told they have a harmless adrenal nodule. An adrenal incidentaloma is one of the groups the Endocrine Society explicitly says to test for cortisol excess. "Nothing to worry about" on a scan report is not the same sentence as "your cortisol has been checked."

Why are cortisol disorders so hard to diagnose?

Because the symptoms are ordinary and the hormone moves. Those two facts drive everything else. Cortisol disorders present with fatigue, weight change, low mood, high blood pressure, weakness and broken sleep, which is also a description of a hard year for a large share of the population. And unlike a bone, which is either broken or not, cortisol is a quantity that changes by the hour, is measured differently by different machines, and travels through the blood attached to a protein whose concentration is itself variable.

The result is a diagnosis reached late, or reached only after a wrong one. A meta-analysis of 5,367 patients across 44 studies put the mean time to a Cushing's syndrome diagnosis at 34 months, and found it had not improved between the cohorts diagnosed before and after 2000. It varied by cause: 14 months for ectopic disease, 30 for adrenal, 38 for pituitary. On the low-cortisol side, a cross-sectional study of 216 patients with adrenal insufficiency found that 68% were given a false diagnosis first, most commonly psychiatric or gastrointestinal, that more than 67% consulted at least 3 physicians, and that 20% waited more than 5 years for the right answer.

Five mechanisms produce that pattern, and they stack.

The reason What it actually is What it does to a test result
1. The symptoms belong to everything Fatigue, weight gain or loss, low mood, hypertension, poor sleep, weakness The test is never ordered, or is ordered years late. 68% of adrenal insufficiency patients were misdiagnosed first
2. Cortisol is a curve, not a level It rises from 3 to 4am, peaks at 7 to 9am, and falls to very low levels by midnight The hour of the sample is part of the result. A number without a time attached is not interpretable
3. Some cortisol excess is episodic Cyclic hypercortisolism: peaks of real excess separated by normal or even low troughs A scheduled test can land in a trough and come back clean. Across 212 reported cases, 6% had unnecessary surgery from misclassification
4. The assay is not neutral Antibody-based immunoassays cross-react with cortisol metabolites. Mass spectrometry does not The same sample returns different numbers on different machines. The Endocrine Society prints its own cut-off with the words "assay dependent"
5. The blood test measures the wrong fraction Roughly 90% of serum cortisol is bound to protein. Only the free ~10% is biologically active Oral estrogen inflates the total. Low blood protein deflates it. Neither changes the hormone your body is actually responding to

Why does the time of day change a cortisol test result?

Because there is no such thing as "your cortisol level." There is only your cortisol at a stated time. The Endocrine Society guideline sets out the shape plainly: "the level of serum cortisol begins to rise at 0300–0400 h, reaches a peak at 0700–0900 h, and then falls for the rest of the day to very low levels when the person is unstressed and asleep at midnight."

The steepest stretch of that curve is the one nearest most people's appointments. A 2025 review in Endocrine Reviews reports that free salivary cortisol increases between 50% and 156% during the first 30 to 45 minutes after waking. A sample taken 20 minutes after you got out of bed and a sample taken 90 minutes after you got out of bed are reading two different parts of a fast-moving quantity. The same review states that "temporal accuracy of sampling is critical", and that failing to record it produces inaccurate estimates.

Clinical medicine knows this and has built its tests around it. The late-night salivary cortisol test exists for precisely this reason. The guideline notes that a single sleeping serum cortisol above 1.8 µg/dL (50 nmol/L) had 100% sensitivity for Cushing's syndrome, at only about 20% specificity, meaning it flags four out of five people who do not have the disease. It is a research measurement drawn through a line in a sleeping inpatient, not the tube you spit into at home, and a value above it is a reason to test further, never a diagnosis. The sensitivity figure works at all only because a healthy midnight value is meant to sit near the floor. Loss of that late-night nadir is, in the guideline's words, "a consistent biochemical abnormality in patients with Cushing's syndrome."

Two consequences follow, and they cut against each other. First, a random cortisol drawn at whatever hour you happened to attend is not a screening test, and the guideline recommends against using one as though it were. Second, a timing-based test is only as good as the assumption underneath it. The guideline warns that the late-night test "assumes a nadir of cortisol in the late evening" and "may not be appropriate for shift workers or those with variable bedtimes." If you work nights, say so before you spit in the tube, because otherwise your sample is being read against a curve that is not yours.

And some cortisol excess arrives in episodes

In cyclic hypercortisolism, cortisol production comes in peaks separated by troughs in which secretion is normal or even low. A scheduled test lands where it lands. The Endocrine Society guideline states outright that 24-hour urinary free cortisol "can be normal if a patient has cyclic disease." A systematic review in The Lancet Diabetes & Endocrinology pooled 212 cases from 118 articles, found the source was a pituitary tumour in 67%, an ectopic tumour in 17% and an adrenal tumour in 11%, and reported that 6% of patients, 12 of the 212, had unnecessary surgery due to misclassification.

Notice that the error runs both ways. A trough sends a sick person home. A peak, in the wrong patient, sends a well person to an operating theatre. Which is why the answer to cyclic disease is not one better test. It is more laboratory tests (late-night salivary cortisol, 24-hour urinary free cortisol, dexamethasone suppression), in the right form, spread across time, ordered and interpreted by a clinician who is following you. A consumer device measuring a wellness signal does not substitute for any of those, and no number of readings turns it into one. The long version of that argument, and what to ask for, is in what a normal cortisol result does and does not settle.

Can two laboratories get different cortisol results from the same blood?

Yes, and by more than most people would guess. The Endocrine Society guideline puts it in one line: "Results for a single sample measured in various assays may be quite different." The reason is chemical. Many steroids are structurally similar to cortisol, so "antibody-based immunoassays such as unextracted RIA and ELISA can be affected by cross-reactivity with cortisol metabolites and synthetic glucocorticoids," while "structurally based assays such as HPLC and tandem mass spectrometry (LC-MS/MS) do not pose this problem." Those last three words matter to anyone taking a steroid, because the drug you were prescribed can be counted by the machine as cortisol you made yourself, letting a suppressed adrenal gland read as a healthy one. A 2017 review in the Annals of Clinical Biochemistry reaches the same verdict about the machines most hospitals actually run: automated immunoassays "lack specificity and show significant inter-assay differences."

The clearest demonstration sits inside the stimulation test used to diagnose adrenal insufficiency. In a study of 100 healthy volunteers, the same blood was measured on four different immunoassays. The 5th percentile of the 30-minute cortisol response, which is effectively the line between a normal result and an abnormal one, ranged from 510 to 626 nmol/L depending on which assay was used. Women showed significantly higher stimulated responses than men. The authors' conclusion is the whole point of this section:

"The definition of the 'normal' response to Synacthen should be both gender and method related at all time points."
Clark PM, et al. Clinical Endocrinology, 1998.

The guidelines carry the caveat openly, in plain sight. The Endocrine Society's 2016 adrenal insufficiency guideline gives its diagnostic threshold as "peak cortisol levels below 500 nmol/L (18 µg/dL) (assay dependent) at 30 or 60 minutes." That parenthesis is doing an enormous amount of work. It means the number your result was compared against is a property of the machine as much as of medicine, and that a threshold inherited from an older, less specific assay can misclassify a real patient in either direction.

This is not a scandal and it is not hidden. It is a known limitation that laboratory medicine discusses in the open. But it is a reason that "my cortisol was normal" is an incomplete sentence until you know which test, at what hour, on which assay.

Can the contraceptive pill or HRT affect a cortisol test?

Oral estrogen can, and it is the confounder most likely to be sitting in a reader's own cupboard. Roughly 90% of the cortisol in your blood is not free. It is bound to cortisol-binding globulin and to albumin, and only the remaining ~10% is biologically active. A standard serum cortisol test measures the total: bound plus free. So anything that changes the amount of binding protein changes the number without changing the hormone your tissues are responding to.

Estrogen raises cortisol-binding globulin. The consequence, in the Endocrine Society's own words, is stark: "False-positive rates for the overnight DST are seen in 50% of women taking the oral contraceptive pill." Half. The guideline's instruction follows: "wherever possible, estrogen-containing drugs should be withdrawn for 6 wk before testing or retesting." The 2016 adrenal insufficiency guideline says the same thing from the other direction, noting that estrogen-containing oral contraceptives "will result in higher CBG with a corresponding rise in cortisol."

The mirror image is the more dangerous one, because it points toward a diagnosis rather than away from one. When binding protein is low, total cortisol reads low while free cortisol is untouched. In a study published in the New England Journal of Medicine, 66 critically ill patients were tested alongside 33 healthy volunteers. In 14 of the patients the stimulated total cortisol came back subnormal, and every one of those 14 had low blood protein. Their free cortisol was not low at all. It was several times higher than the volunteers'. On the total number they looked adrenally insufficient. On the fraction that actually acts on the body, they were not.

Two other things move a cortisol result in the same quiet way, and are worth naming:

None of this is exotic, and none of it is anybody's fault. It is simply why the drug and alcohol history comes before the blood test, and why leaving something off it, including the pill, the patch, the inhaler and the two glasses of wine, can cost months.

Does "hard to diagnose" mean you have one?

Almost certainly not. But read the next paragraph before you act on that. Everything above is true, every one of those failure modes is real, and the base rate still dominates. Cortisol disorders are uncommon. The symptom cluster that brought you here is not. For cortisol excess, the Endocrine Society is specific about who should be tested: people with unusual features for their age, people with multiple and progressive features, children whose height percentile is falling while their weight climbs, and anyone found to have an adrenal incidentaloma. Its position on everybody else is a single sentence: "We recommend against widespread testing for Cushing's syndrome in any other patient group." If you are in one of those four groups, that sentence is not about you.

And that sentence is about Cushing's syndrome only. For low cortisol the guidance runs the other way, and this is the paragraph that actually keeps people alive. The Endocrine Society's adrenal insufficiency guideline recommends testing every patient with indicative symptoms or signs, and asks explicitly for a low diagnostic and therapeutic threshold in acutely ill patients and in anyone with predisposing factors: unexplained volume depletion, low blood pressure, low sodium, high potassium, fever, abdominal pain, darkening skin, or, in children, low blood sugar. Adrenal insufficiency is the disorder on this page that kills people while they wait, and the study cited above is a study of precisely that: 68% of those patients were sent away with a psychiatric or gastrointestinal label first. If your symptoms point down rather than up (weight loss, nausea, dizziness on standing, salt craving, deepening exhaustion), do not let the paragraph above talk you out of asking.

That matters not as reassurance but as a warning, because the wrong hormone is an extremely effective place to hide a real diagnosis. The same fatigue, weight gain, low mood, high blood pressure and broken sleep are produced by conditions that are far more common, that are treatable, and that a cortisol test does not investigate at all: obstructive sleep apnea, thyroid disease, iron deficiency, polycystic ovary syndrome, poorly controlled diabetes, alcohol use, depression, and the medications already in your cupboard. Someone who spends 18 months insisting on cortisol testing has not been thorough. They have been late for something else.

So the useful thing to take from this page is narrow. Take the specifics, not the suspicion:

An endocrinologist is the specialist for this. If your features are unusual for your age, or multiplying, that is precisely the group the guideline says to test, so ask for the referral by name. For the honest account of what the research does and does not support at the vaguer end of this territory, see what the evidence actually shows about HPA axis dysfunction. The low-cortisol side is covered in how adrenal insufficiency is actually diagnosed.

And plainly, on ourselves: nothing above is an argument for a wearable, ours or anyone else's. A general wellness reading is not a laboratory assay. It has no clinical threshold, it is not on the same scale as a clinical result, and it cannot be compared to one. There is no consumer shortcut around a physician here, and anyone offering you one is selling you something. If what you want is to understand the hormone itself, what it does and how it moves through an ordinary day, that is Cortisol 101.

This page is general medical information, not medical advice, and not a diagnosis. It summarizes published clinical guidance from the sources below. Cortisol disorders are diagnosed and treated by qualified healthcare professionals, never by a wearable or an at-home test. Never start, stop or change a steroid medication without speaking to the doctor who prescribed it. If you suspect an adrenal crisis, a blood clot, or a serious infection, seek emergency care immediately.

References

Take the specifics to a doctor, not the suspicion.

Ask which test you had and at what hour, whether it was repeated, and what else is being considered. Hand over the complete drug list, including the pill, the HRT, the inhaler and the alcohol. If your features are unusual for your age or are getting worse, ask for a referral to an endocrinologist. And if any of the emergency signs above apply to you, call emergency services now.

Keep reading

Related guides

Normal test, same symptomsWhat a normal cortisol result does and does not settle, and what cyclic disease is. Low cortisolAdrenal insufficiency, how it is diagnosed, and the crisis warning signs. Cortisol 101What cortisol is, the daily curve, and how it's measured.
Straight answers

Diagnosis FAQ

Why is Cushing's syndrome so hard to diagnose?

Because its symptoms are ordinary and its hormone moves. Weight gain, fatigue, low mood, high blood pressure and disturbed sleep are common in the general population, so the syndrome is rarely the first thing anyone considers. On top of that, cortisol swings across the day, some cases are cyclic and produce normal results between episodes, and different laboratory assays return different numbers from the same sample. A meta-analysis of 5,367 patients found a mean time to diagnosis of 34 months, and that figure had not improved between the cohorts diagnosed before and after 2000.

Why does the time of day change a cortisol test result?

Because cortisol is a curve, not a fixed level. The Endocrine Society guideline describes serum cortisol beginning to rise at 3 to 4am, reaching a peak between 7 and 9am, and then falling for the rest of the day to very low levels when a person is unstressed and asleep at midnight. Salivary cortisol rises between 50% and 156% in the first 30 to 45 minutes after waking. The hour you were sampled is part of your result, which is why a late-night salivary cortisol test exists at all.

Can two laboratories get different cortisol results from the same blood?

Yes. The Endocrine Society guideline states that results for a single sample measured in various assays may be quite different, because antibody-based immunoassays cross-react with cortisol metabolites while mass-spectrometry methods do not. In a study of 100 healthy volunteers, the 5th percentile of the 30-minute cortisol response to a stimulation test ranged from 510 to 626 nmol/L depending on which of four immunoassays was used. The Endocrine Society's own cut-off for adrenal insufficiency carries the words assay dependent.

Can the contraceptive pill or HRT affect a cortisol test?

Oral estrogen can. Estrogen raises cortisol-binding globulin, and a standard blood test measures total cortisol, roughly 90% of which is bound to protein. The Endocrine Society guideline reports false-positive rates on the overnight dexamethasone suppression test in 50% of women taking the oral contraceptive pill, and advises that estrogen-containing drugs be withdrawn for 6 weeks before testing or retesting wherever possible. Tell whoever orders the test everything you take, including the things you do not think of as drugs.

Can a wearable or an at-home kit diagnose a cortisol disorder?

No. We make a cortisol sensor, so we should be the ones to say it plainly. Diagnosis is made by a physician, using laboratory assays with defined clinical thresholds, repeated over time, and followed by imaging and further testing to find a cause. A general wellness reading is not on that scale and cannot be compared to a clinical result. Nothing on this page is a reason to buy a device. It is a reason to ask your doctor a more specific question.