Ferritin

What the research found.

• In one study the average ferritin was 4,752 micrograms per liter in adult-onset Still disease, against 1,571 in the controls. The spread around each average was enormous, at 9,599 and 3,807. Both of those are bigger than the averages they belong to.

  Fautrel and colleagues, Journal of Rheumatology, 2001

• One review pooled eight studies covering 556 patients. The ferritin cut-offs those studies used ranged from 400 to 5,000 micrograms per liter. Some used no fixed number at all, asking instead for five times the top of normal, which is more than a tenfold spread in what researchers called diagnostic.

  Lee and Song, Zeitschrift für Rheumatologie, 2025

• A set of rules from 2016 helps spot a complication called macrophage activation syndrome, and one of those rules is a ferritin level. One team used it on adults with Still disease rather than on the children it was built for. It cleared just 15.0 percent of the people who didn't have the complication.

  Tada and colleagues, Rheumatology International, 2018

• One study of 206 Still disease patients tested a ferritin cut-off of 3,500 micrograms per liter. It caught 85 percent of the people who had macrophage activation syndrome. When it read below the cut-off it was right 97 percent of the time, which is the useful half of that finding.

  Javaux and colleagues, Journal of Clinical Medicine, 2021

What the number means.

What the test measures, twice over

Ferritin is the protein that stores iron inside your cells, and the amount in your blood normally reflects how much iron you have in reserve. That makes a low ferritin one of the more reliable tests in medicine, because it means your iron is low and it shows that before anemia appears. Most people know the test for that job, and it's the simple one.

Ferritin does a second job too, rising whenever the body is inflamed, whatever your iron is doing. It also rises in infection, in liver disease, and in some cancers, and those two jobs cause almost every misunderstanding about this test. A raised ferritin can mean stored iron or it can mean inflammation, and the number alone can't tell you which.

Why it's ordered in Still disease

Adult-onset Still disease pushes ferritin to some of the highest values seen anywhere in medicine, which is why the test turns up in its assessment. One study compared 49 patients with the illness against 120 controls, where the average ferritin was 4,752 micrograms per liter in the patients and 1,571 in the controls. That difference reached a P value of 0.029.

Look at the spread around those averages before you conclude anything. It was 9,599 in the patients and 3,807 in the controls, and both are bigger than the averages they belong to, which is what heavily overlapping groups look like. Plenty of controls had high values and plenty of patients had lower ones.

So the difference is real and the averages aren't cut-offs. Don't use them as cut-offs, because an average tells you where the middle of a group sits and says nothing about where one person falls inside it. That distinction recurs through this whole page.

Why there's no agreed cut-off

A 2025 review pooled eight studies covering 556 patients with adult-onset Still disease and 763 controls. The most useful thing the review reports is how much those studies disagreed. Their ferritin cut-offs ranged from 400 to 5,000 micrograms per liter, and some used no fixed number at all, asking for at least five times the top of normal instead. That's more than a tenfold spread in what different research groups called diagnostic.

The same review gives the pooled figures. Ferritin caught 60.4 percent of cases and cleared 85.7 percent of the people who didn't have the disease, a positive result multiplied the odds by 4.179, and the whole test scored 11.32 on one combined measure. Read together, that describes a moderately useful test with no settled operating point.

So this page publishes no cut-off for the diagnosis, because there isn't one to publish. Offer any single figure as the level that diagnoses Still disease and you manufacture an agreement the research doesn't contain. That is worth refusing to do.

The macrophage activation syndrome problem

Macrophage activation syndrome is a serious complication of Still disease, and catching it early counts for a great deal. A set of classification rules came out in 2016, worked out in children with a related illness, so the obvious question was whether they hold up in adults. One study set out to answer that directly.

It tested the rules against lab results from 76 adults with Still disease. Taken as a whole, the rules caught 100 percent of the cases and cleared 70 percent of the people who didn't have the complication. When they said yes they were right 47.1 percent of the time. When they said no they were right 100 percent of the time, and the individual rules performed unevenly, because the triglyceride rule caught only 46.7 percent of cases and the ferritin rule cleared only 15 percent.

Clearing only 15 percent means the ferritin rule flagged nearly everybody who didn't have the syndrome. The reason is built into the illness, because ferritin is already very high when Still disease is active, so a cut-off designed to catch a further rise can't sort one from the other. That's a limitation of the rule rather than a failing of anybody using it.

The authors then dropped the triglyceride rule and re-tuned the rest, which got the rules to catching 100 percent and clearing 93 percent in those 76 adults. Note what that means. It's a set of rules tuned on the very patients it was tested on, and it has never been checked against a fresh group.

Using ferritin the other way round

A separate study of 206 Still disease patients found a more defensible use for the number. It tested a ferritin cut-off of 3,500 micrograms per liter. That caught 85 percent of the people who had macrophage activation syndrome, and when it read below the cut-off it was right 97 percent of the time.

That second figure is a specific kind of usefulness. A ferritin under that level makes the complication unlikely, which is genuinely reassuring to somebody waiting for an answer. It doesn't mean a value above it settles the diagnosis, and that lopsidedness is the practical content of the finding.

That study found something else worth knowing. Very high values at the time of the Still disease diagnosis predicted getting the complication later, which says something about the future rather than about now. It's a reason for closer watching rather than a test result to act on today.

Glycosylated ferritin, which moves the other way

Ferritin often gets ordered alongside glycosylated ferritin, and the pairing exists because the two move apart in Still disease. Total ferritin rises while the glycosylated share falls. In the 2001 study it was 15.9 percent in patients against 31.5 percent in controls, and the two tests together do better than either one alone.

Now pair them properly, because that's where the numbers get interesting. A glycosylated ferritin at or below 20 percent, plus a total ferritin above the top of normal, caught 70.5 percent of cases and cleared 83.2 percent of healthy people. Raise the ferritin bar to five times normal and the figures move to 43.2 percent and 92.9 percent.

So that change spent 27 points of catching to buy about 10 points of clearing. That is the kind of trade every cut-off in medicine makes, and it is hardly ever stated out loud to the person whose result it is. Glycosylated ferritin has its own page here.

What to do with a high result

Suppose you've been given a very high ferritin. The useful question is what else is going on, because fevers, a rash, a sore throat, and sore joints point one way, while a recent infection, liver disease, or heavy drinking point somewhere else entirely. Too much iron is a separate possibility, which other iron tests help sort out, and a ferritin pushed up by inflammation doesn't mean excess iron.

This isn't something to work out alone, and the range of possibilities is why. Take the result to a doctor who can set it beside your exam and your history, because that is where it acquires a meaning. A number this variable earns everything from its context, and the context is what a consultation supplies.

Questions patients ask.