A 278,000-person screening trial found more early cancers — and more early harms

Guidelines recommend colorectal cancer screening in people aged 50 to 75 using either colonoscopy or a faecal immunochemical test (FIT), and many national programmes use one-sample biennial FIT [s1]. What is thinner than the recommendations suggest is randomised evidence comparing either method against no screening at all — the comparison that would establish how much benefit, and how much harm, screening actually produces. The authors of a new trial state that need directly [s1].

Nature Medicine published on February 20 the diagnostic-phase results of SCREESCO, a Swedish trial that ran that comparison at national scale [s1].

The design

SCREESCO randomised 278,280 individuals aged 60 by a randomised block method, without masking, into three groups [s1]:

• Once-only primary colonoscopy, at a 1:6 ratio against controls
• Two rounds of two-stool FIT with a low cutoff of 10 μg per gram of faeces, at a 1:2 ratio against controls
• Usual care — no screening

In the analysis, 31,113 individuals were in the primary colonoscopy arm and 60,267 in the FIT arm, against 186,671 colonoscopy controls, of whom 120,521 also served as controls for the FIT comparison [s1].

Two design choices are worth flagging. The FIT protocol used two stool samples per round at a low cutoff, which is more sensitive than the one-sample biennial FIT that most national programmes use [s1] — so the FIT arm here is a more aggressive test than the one many readers will be offered. And the trial enrolled a single age, 60, rather than a range.

What happened to cancer incidence

After a median follow-up of 4.8 years, the incidence rate of colorectal cancer was [s1]:

• Colonoscopy arm: 107.9 per 100,000 person-years, versus 99.9 in controls — incidence rate ratio 1.08 (95% CI, 0.91–1.28)
• FIT arm: 96.0 per 100,000 person-years, versus 103.9 in controls — IRR 0.92 (95% CI, 0.81–1.05)

Neither is a statistically distinguishable difference. Both confidence intervals cross 1.

That is not, by itself, a failure. Over a short follow-up, screening is expected to raise measured incidence by finding cancers earlier than they would otherwise have declared themselves — the lead time effect — before any reduction in incidence from polyp removal has had time to appear. Which is roughly the pattern the two arms show, in opposite directions.

Where the two arms did separate

Rates of stage I–II colorectal cancer — early-stage, and the stage at which the disease is most treatable — were higher in both screening arms [s1]:

• Colonoscopy: IRR 1.38 (95% CI, 1.09–1.74)
• FIT: IRR 1.19 (95% CI, 0.99–1.43)

The colonoscopy result excludes 1; the FIT result narrowly does not.

This is the mechanism by which screening is supposed to work: not fewer cancers in the short term, but the same cancers found sooner. Whether that translates into fewer deaths is the question the diagnostic phase cannot answer, and the authors do not claim it does.

The harms

Rates of cardiovascular and gastrointestinal events were slightly higher in the intervention arms during the first year, and subsequently became more similar to controls [s1].

That temporal pattern is what one would expect if the excess events are procedure-related — concentrated around the time colonoscopies and follow-up colonoscopies are performed, then fading. The authors' own summary is that the increase in cancer detection implies a benefit of screening while the increase in adverse events suggests some initial harm [s1].

How to read a trial like this

Three things constrain what SCREESCO establishes so far.

It is the diagnostic phase. The outcome that matters for a screening programme is colorectal-cancer mortality, and a median 4.8 years of follow-up is short for that. Nothing here tells you whether screening at 60 saves lives.

Randomisation was to assignment, not to procedure. The trial assigned individuals to once-only colonoscopy, to two rounds of two-stool FIT, or to usual care [s1]; it did not and could not compel anyone to attend. Participation rates are not given in the published abstract, and any effect measured across everyone assigned to a screening arm is diluted by those who did not take it up.

The comparators are specific. Once-only colonoscopy at age 60 is not the same as colonoscopy every ten years from 45. Two-sample FIT at a 10 μg/g cutoff, twice, is not one-sample biennial FIT. Results do not transfer cleanly to programmes built differently.

Why it matters anyway

Very few screening modalities in wide use have ever been tested against no screening in a randomised trial at this scale. The trial's value is that it quantifies both sides of the ledger in the same population, with the same follow-up — an early-stage detection gain of roughly 38% in the colonoscopy arm, set against a first-year excess of cardiovascular and gastrointestinal events [s1].

What to watch

The mortality endpoint, which is what the trial was ultimately built to measure, and any published analysis of participation rates by arm.

The trial is registered on ClinicalTrials.gov as NCT02078804 [s1]. This article describes a research finding and is not medical advice; screening decisions belong with a clinician.

Sources

• [s1] Westerberg M, Ludvigsson JF, Metcalfe C, et al. Colonoscopy and fecal immunochemical testing versus usual care in diagnostic colorectal cancer screening: the SCREESCO randomized controlled trial. Nature Medicine, published online 2026-02-20.