The Deadly-Fungus Count Going Around Counts Carriers, Not Patients — and U.S. Candida auris Illnesses Are Below Last Year

Illustration. AI-generated laboratory scene; not a photograph of any specific facility.

The number moving through the news this week is 3,130, described as cases of the fungus Candida auris across 23 states. It comes from a row in CDC (Centers for Disease Control and Prevention)’s weekly notifiable-disease tables, and that row counts something narrower than the word case suggests: people found to be carrying the fungus, usually on their skin, who may have no symptoms at all.

CDC keeps a second row for the cases found the other way: the ones the fungus turned up in when a sample was taken to work out why a patient was ill. On the same date, July 18, 2026, that row stood at 3,302, reported by 27 of the health departments that file these tables rather than 23.

Setting those two numbers against each other tells you nothing, because they count different things: one counts detections on a colonization swab, the other detections in a specimen taken to find out why a patient was ill. CDC counts them separately, and someone already counted on the illness row is not counted again on the carrier row. The comparison that does mean something is each row against itself a year earlier. The coverage does make that comparison — but on the carrier row, which it presents as illnesses.

Neither figure is new this week: both are routine weekly surveillance, current as of July 18, and CDC’s most recent published review of the fungus came out on July 2, 2026.

Key facts

Illnesses (clinical cases), week ending July 18, 2026: 3,302 year to date, vs. 4,102 for the same week of 2025 — about 20% fewer.
Carrier detections (screening cases), week ending July 18, 2026: 3,130 year to date, vs. 3,348 for the same week of 2025.
Latest week in the tables as of August 5, 2026, ending July 25, 2026: illnesses at 3,437 year to date, vs. 4,290 for the same week of 2025 — still below last year.
Both rows are voluntary state reports to CDC’s weekly surveillance tables and can shift after the fact.
Who it reaches: the fungus spreads in hospitals and long-term care facilities and threatens patients who are already seriously ill; risk to healthy people going about ordinary life is minimal.

A swab is not a diagnosis

Most of the confusion in this week’s coverage comes from mixing up the two ways CDC counts this fungus.

A clinical case is the fungus found in a specimen taken to work out why someone is ill — most often a urine sample or a blood draw. A screening case is the fungus found on a colonization swab, usually from the armpit and groin or the rectum, in a patient who may have no symptoms at all. Colonization means the yeast is living on a person’s body. It is not the same as being sick.

The two are tallied on separate lines. The tables record detections at a moment in time rather than following individuals forward: nothing in them shows whether a person found to be carrying the fungus later turns up on the illness row.

With the two lines held apart, the newest week published is easy to read. For the week ending July 25, illnesses were again below the same week of 2025, and so was the carrier row. If you are going to repeat one figure for illness, 3,437 as of July 25 is the one. The 3,130 in this week’s coverage is a carrier figure, and it comes from the week before that one.

Which states lead depends on which row you read

The state figures repeated across the coverage are all from the carrier line, for the year to date through mid-July 2026:

Texas: 706
Michigan: 503
Illinois: 366

Rank the states by illness (clinical cases) in that same period and the order changes:

California: 826
Texas: 391
Ohio: 274
Tennessee: 274
Georgia: 193

California leads the country on the illness row, and it is not among the states the coverage puts at the top.

Texas is worth spelling out, because the widely syndicated version of this story — Nexstar’s, republished by outlets including The Hill — says Texas has already passed its whole 2025 total. Put the two numbers side by side and it has not: 706 carrier detections from January through mid-July of 2026, against 1,314 for all of 2025. Half a year of counting is a little over half of last year’s total, which is roughly what you would expect.

The 660 that version uses is not an invented number. It sits in the same table, in the column that carries the previous year’s count for the same weeks — Texas’s carrier detections through mid-July of 2025. The mistake is calling a part-year figure an annual total. Read correctly, though, it says something this article has to say too: in Texas, carrier detections are running above where they were a year ago, 706 against 660, even while both national rows are below.

The multi-year climb is real — but the numbers behind it come from two different places

Over a span of years, the direction is upward, and that is a different question from this year against last year rather than a contradiction of it. CDC’s surveillance review published July 2, 2026 counted 2,882 clinical cases in 2022, 4,428 in 2023 and 6,197 in 2024, with colonization detections running about twice as high in each of those years and roughly doubling over the same span.

What cannot be done is to extend that line with a number from somewhere else. The weekly tables closed 2025 at 7,308 clinical cases — a bigger number than the review’s 2024 figure, but from a different counting system, and one that points the other way inside its own series. The same weekly tables put 2024 at 7,702, which makes the last full year lower than the one before it, not higher. The two sources disagree even more sharply on colonization: for 2024, the review counted more than twice what the weekly tables hold. A trend has to be read down one column at a time.

Some of the climb the review does show may reflect wider testing rather than more illness. Bloodstream infections — a steadier measure than total case counts, as infectious-disease researcher John Drake has argued — went from 991 in 2022 to 1,586 in 2024, a rise of roughly 60%. That is slower than the growth in total case counts, which suggests wider testing explains part of the difference between the two.

Who this fungus actually reaches

Candida auris is a drug-resistant yeast that CDC classifies as an urgent antimicrobial-resistance threat, its highest tier and the first such designation given to a fungus; the World Health Organization put it in the critical-priority group of its 2022 fungal pathogen list.

It travels in healthcare settings — on contaminated surfaces, on shared equipment, on the hands and clothing of staff — and it survives on surfaces for long stretches. It is not a respiratory infection, and it is not something that moves through the community. The people it threatens are those already seriously ill: patients on ventilators or with central lines and catheters, and residents of long-term acute care and skilled-nursing facilities. Risk to healthy people going about ordinary life is minimal.

For the patients whose infection becomes invasive — the fungus getting off the skin and into the bloodstream or deeper in the body, in people already seriously ill in a hospital or a long-term care facility, not carriers and not healthy people in the community — CDC’s estimate is that 30% to 60% die. The estimate rests on a limited number of patients, and CDC cautions against reading it too literally: most of those patients had other serious illnesses that raised their risk of death on their own, which makes any one death hard to attribute to the fungus. Neither row counts deaths: the weekly tables record detections only, and no national death count for 2026 sits behind the figures in this week’s coverage.

The setting is the pattern. In CDC’s 2022–2024 review, about three-quarters of illnesses were identified in acute care hospitals, and most of the remainder in hospitals that handle long stays. The people counted skew old: the large majority were 45 or older.

Treatment is genuinely harder than for ordinary yeast infections. Some strains resist all three major classes of antifungal drug, including echinocandins, which are held in reserve; others resist the azole drugs and sometimes amphotericin B.

One other fungus has been circulating in the same conversation: Sporothrix brasiliensis, spread by cats in South America. It is a separate story, and it has not been identified in the United States.

Where to check the numbers yourself

The tables everyone is quoting are public. CDC posts them on data.cdc.gov as its NNDSS Weekly Data — the same weekly notifiable-disease tables the news figures come from — under the dataset id x9gk-5huc. Each week carries a Candida auris clinical row and a Candida auris screening row, with the year-to-date total beside the same period a year earlier. Whichever number turns up in a headline, that pair of rows will tell you which one it is — and the column beside it will tell you whether the number is going up or down. The tables label them Candida auris; newer scientific naming writes the same yeast Candidozyma auris, and some state health departments have started using that spelling, so both names point to the same organism.