The Screening Everyone Trusts
Somewhere in your late forties, friends start trading colonoscopy stories the way earlier generations traded wisdom-teeth sagas. The prep is miserable, the nap is glorious, and the procedure itself arrives wrapped in a medical consensus so solid that the US Preventive Services Task Force gives colorectal screening its top grade for adults 45 to 75 while the American Cancer Society says to start at 45. The logic is simple and seductive. A camera finds precancerous polyps, the doctor snips them out, and a cancer that would have killed you never gets born. Screening does not merely detect disease; it prevents it — the promise that makes colonoscopy different from every other cancer screen.
There was just one problem. Until 2022, no randomized trial had ever tested whether inviting healthy people for colonoscopies actually reduces deaths. Mammography, Pap smears, and stool blood tests all earned their reputations in randomized trials. Colonoscopy skipped that step, riding on observational studies in which screened patients reliably died less often. When the first randomized trial finally reported, an editorialist in the New England Journal of Medicine called the results both surprising and disappointing.
The Trial That Rewrote the Question
The Nordic-European Initiative on Colorectal Cancer, NordICC, enrolled 84,583 men and women aged 55 to 64 in Norway, Poland, and Sweden between 2009 and 2014 and randomized them two to one: 28,217 received an invitation to a single screening colonoscopy while 56,366 received nothing. Only 42 percent of the invited ever showed up for the procedure. Ten years later, the first report landed in the New England Journal of Medicine. Invitations cut colorectal cancer incidence by 18 percent, but deaths fell just 10 percent, a gap the statistics could not distinguish from chance.
Thirteen Years In, the Answer Has Barely Moved
In May 2026, The Lancet published the 13-year update, and the picture shifted only slightly. Colorectal cancer cases stood at 1.46 percent in the invited group versus 1.80 percent in the unscreened group, a relative risk of 0.81. Among the 42 percent who actually underwent the procedure, the per-protocol estimate looked stronger at 0.55. But on the endpoint that justifies screening mandates, deaths, the numbers refused to separate: 0.41 percent versus 0.47 percent, a relative risk of 0.88 with a confidence interval of 0.68 to 1.08, straddling the line of no effect. Even the per-protocol mortality estimate of 0.70 carried an interval of 0.26 to 1.25. Three more years of follow-up did not rescue the mortality result.
The benefit concentrated on the left side: distal cancers fell 21 percent while proximal cancers barely budged, with the interaction test returning p less than 0.0001. That split matches the mechanics of the procedure, since the right colon hides flat, hard-to-see growths behind folds.
A Trial Built for a World That No Longer Exists
The designers planned for a world in which 0.82 percent of unscreened participants would die of colorectal cancer during follow-up. What arrived was 0.47 percent, barely half the expected figure, because colorectal cancer mortality has fallen for years as surgery, chemotherapy, and radiation improved. An underpowered trial cannot prove the null; it can only fail to reject it. Nobody seems to have published the obvious follow-up calculation.
Under the trial's original assumptions, roughly 57,000 participants would have delivered 80 percent power, and NordICC's 84,583 looked comfortably adequate. Plug in the world that actually materialized, a 0.47 percent baseline mortality and a true relative risk of 0.88, and the required number explodes to about 420,000 participants, roughly five times the trial's actual size. The trial was neither sloppy nor small. It was built for a 2009 disease burden and graded on a 2026 one. A second way to see the same thinness: at the point estimate, about 1,700 screening invitations prevent one colorectal cancer death over 13 years, and even the optimistic end of the confidence interval implies only around 150 deaths averted per 100,000 invitations.
The Strongest Defense of the Procedure
Gastroenterologists have not been quiet about the trial's limits, and their core objection deserves full weight. The intention-to-screen analysis answers a policy question, what happens when you mail invitations to 28,217 people, rather than a medical one, because only 42 percent of invitees ever lay on the table. The skippers dilute the result by design. David Lieberman of Oregon Health and Science University, speaking for the American Gastroenterological Association, put it bluntly: colonoscopy screening is effective if it is completed.
A 2023 modeling study in Gastroenterology backs him up. Van den Berg and colleagues simulated the NordICC design and found that given the low adherence and the secular decline in colorectal mortality, a null mortality result is exactly what the models predict, even if colonoscopy genuinely prevents deaths among those screened. On this reading, the trial does not overturn the observational evidence; it shows what a strong procedure looks like when partial uptake meets a population whose baseline risk has already fallen.
What We Didn't Prove
NordICC tested one invitation to one colonoscopy in adults aged 55 to 64 in three European countries, which is not the same as a screening program. Nobody received a second colonoscopy, nobody got a fecal immunochemical test as the primary screen, and the participants entered the trial more than a decade ago in health systems different from, say, American adults starting at 45. The trial also cannot say colonoscopy does not save lives. A null result from an underpowered trial is absence of evidence, not evidence of absence. The ongoing US CONFIRM trial, randomizing colonoscopy against FIT with results expected around 2029, may finally answer the mortality question. Until then, the strongest honest statement is narrow: we have no randomized evidence that screening colonoscopy reduces deaths from colorectal cancer, plus a solid explanation for why the first attempt could not show it.
The Bottom Line
The colonoscopy consensus stood for decades on observational evidence, and NordICC has not demolished it. What the 13-year data have demolished is the casual certainty that the procedure's mortality benefit was proven. A screening test can find and remove precancers, cut incidence by a fifth, and still fail to move the death rate, because deaths are rare, adherence is partial, and treatment keeps improving. That is not an argument against the camera. It is an argument for honesty about what the strongest evidence actually says.
What You Can Do
Do not cancel your screening because of this trial. Every major guideline still recommends colorectal screening from 45, and the procedure's ability to find and remove precancerous polyps is not in dispute; the trial itself showed incidence falling. If the prep fills you with dread, ask your doctor about the fecal immunochemical test, a yearly at-home stool test that major guidelines accept as an alternative and that sidesteps the whole adherence problem. If you do get a colonoscopy, ask about your endoscopist's adenoma detection rate, the quality metric that predicts whether the procedure finds what it should. And keep the trial's plainest lesson in mind. The invitation only works if you show up.
Related Reading
- Every Instinct Says to Fight Cancer with Surgery. A 15-Year Trial of 1,643 Men Found That Monitoring Prostate Cancer Achieves the Same Survival Rate.
- Medicare Penalized Hospitals for Readmitting Patients. A Study of 8 Million Cases Found It Increased Deaths.
- Vitamin D Is the World's Most Popular Bone Supplement. The Largest Trial Ever Conducted Found It Does Not Prevent Fractures.