Colonoscopy as a screening examination

A trial on colonoscopy for colorectal cancer prevention showed that a physician invitation for a screening check-up reduced incidence but not mortality.

Translated from the original Italian version

The NordICC study

The primary analysis of the NordICC study (The Northern-European Initiative on Colorectal Cancer) showed that colonoscopy reduced the risk of colon cancer by only about one-fifth, far below past estimates of the test's effectiveness, and provided no significant reduction in colon cancer mortality. Many gastroenterologists, including Michael Bretthauer himself, the Norwegian physician and researcher who conducted the study, reacted to the results with a mixture of shock, disappointment and even mild disbelief.

This is the first randomised study to show the results of exposing people to colonoscopy screening versus no colonoscopy. Many had expected different results. The study certainly does not invalidate colonoscopy as a useful screening tool, but perhaps leads to a reassessment of its position as the gold standard for colon cancer screening.

Colorectal cancer screening

Colorectal cancer screening is a preventive intervention that can make use of several methods. In addition to colonoscopy, these include, for example, faecal occult blood test (FOBT), rectosigmoidoscopy, faecal immunochemical test (FIT).  In the United States, the USPSTF (United States Preventive Services Task Force) has endorsed the policy that any screening test is better than none. In this situation, although rectosigmoidoscopy had the strongest data and should have been given priority, doctors often preferred colonoscopy to examine the entire colon (and perhaps also because the test is well reimbursed).

Past research has consistently shown that colonoscopy can greatly reduce the incidence and mortality of colon cancer by 70%. But none of these studies were large randomised trials, which are widely regarded as the pinnacle of evidence in clinical research. So Bretthauer, (University of Oslo and Oslo University Hospital), and colleagues initiated one a decade ago, recruiting more than 80,000 people between the ages of 55 and 64 in Poland, Norway, and Sweden to test whether colonoscopy was really as effective as everyone believed.

Approximately 28,000 of the participants were randomly selected to undergo a colonoscopy (the 'invited' group), while the others continued with their usual care, which did not include regular colonoscopy screening (when the study began, there were few or no organised colonoscopy screening programmes at population level in these countries).

The results of the NordICC study

The researchers then tracked colonoscopies, colon cancer diagnoses, deaths from colon cancer and deaths from any cause. Over 10 years of follow-up, the invitation to screening colonoscopy modestly reduced the risk of a colon cancer diagnosis, but did not significantly reduce the risk of death from colon cancer. Cancer survival was almost identical in both groups.  All-cause mortality was also the same.

The specific numbers of the primary outcome were:

Nuances in the results interpretation

The results are inconsistent with some past investigations of other colon cancer screenings. "We know from other screening tests that we can reduce cancer mortality to a greater extent," said Jason Dominitz, who wrote an accompanying editorial in NEJM and did not work on the study. Dominitz pointed out that rectosigmoidoscopy, which examines only a smaller portion of the colon, has been shown to reduce colon cancer mortality in randomised trials. "Colonoscopy is rectosigmoidoscopy with something else added, so one would think that it cannot be less effective than rectosigmoidoscopy," he said.

Cancer treatment has also advanced over the past two decades and the study had only 10 years of follow-up, both factors that make it harder to see a mortality benefit from screening. "They're doing a 15-year follow-up and I would expect to see a significant reduction in long-term cancer mortality," Dominitz added. "Time will tell us more."

But nuances in the interpretation of the data abound, Dominitz said. For example, a minority of participants invited for colonoscopy, only 42%, actually showed up for the exam. This may have diluted the benefits of colonoscopy seen in the study. This observation underlies the study's main criticism: one cannot benefit from a colonoscopy if one has not had it.  

John Mandrola and Vinay Prasad, in their website Sensible Medicine, offer two rebuttals to this argument. The first is that their apparently low post-invitation acceptance rate is not very different from the US experience. Current surveys suggest that 60-70% of Americans undergo screening for colon cancer. This rate has increased in recent years. Although this is a rough estimate, it is not entirely dissimilar to the rates achieved in NordICC and is close to the rates achieved in one enrolment site, Norway, which has not reported different outcomes.

The second objection delves into the meanders of study conduct and statistics. Although the idea of counting only those who undergo the procedure makes theoretical sense, in reality it is complicated. The authors conducted what is called a 'per-protocol' analysis, in which those who actually underwent the procedure are counted against their usual care.

In statistics, however, purists believe that study results should be based on the group that was randomised and not on what happens after randomisation. This principle is called 'intention-to-treat' or, in this case, 'intention-to-screen'.

Why? Because the purpose of randomisation is to balance known and unknown confounders, to balance the distributions of outcomes in the absence of an effect. A per-protocol analysis negates randomisation. People who undergo colonoscopy screening might be different from those who shrug. They might be wealthier, or more likely to engage in other health behaviours, or more docile and compliant. Indeed, if one relies on per-protocol, why randomise?  The authors adjusted for covariates in their per-protocol analysis; they could have done so in an observational dataset.

However, they note, with all these limitations, the per-protocol analysis (the best-case scenario) found that the risk of death from colorectal cancer was 0.15% in the invited group compared to 0.30% in the usual care group. This is a 50% reduction in relative terms. In absolute terms, the difference in risk is 0.15% or 15 per 10,000.

Bretthauer himself reported that the secondary analysis is not as robust as the primary or intention-to-treat analysis. "The intention-to-treat analysis is the best methodology, the analysis in which to put all one's trust," said Bretthauer from Oslo. This led him to consider that he and everyone else in the colon cancer field may have been wrong about the usefulness of colonoscopy.

The value of colonoscopy as a screening test after the NordICC study

"It's not the trump card we thought we had," Bretthauer said. "I think we overestimated colonoscopy. If you look at what the gastroenterology societies say, and I'm a member of them myself so they are my collaborators, they were talking about a 70, 80 or even 90 per cent reduction in colon cancer if everyone underwent colonoscopy. This data does not prove it'.

Rather, he said, the real benefit of colonoscopy screening might lie somewhere between the primary and secondary tests in his study. 'If you undergo a colonoscopy, you can reduce your risk of getting colorectal cancer by 20-30%,' Bretthauer said. This brings it more in line with other major tests for colorectal cancer, which test stools for signs of cancer, abnormal DNA or blood, and can be performed at home.

This raises an important point for politicians, Bretthauer added. Colonoscopies are more expensive, more time-consuming and more unpleasant for patients. Many European countries opposed the idea of allocating public health funds to a large and expensive programme, when faecal testing was cheaper, simpler and had been more widely used in some studies. "Now the European approach makes much more sense. It is not only cheaper, but perhaps just as effective,' Bretthauer said.

Although NordICC's findings mainly concern screening programmes, its results also inform individual decision-making. A person who decides to undergo a colonoscopy takes on a considerable burden. The 'preparation' (which is a euphemism for having to poop more than 12 times) takes up most of the day before the examination and the next day is unproductive because of the procedure and its sedation. The burden would be fine if there was a proportional reduction in risk. A person may think: what are my chances of avoiding a colorectal death with and without the procedure?

Immediate effects on health policies

This is not the end of studies on colorectal screening. We will find out more from a Swedish study comparing colonoscopy with faecal immunochemical testing or usual care. And there are two ongoing US studies that will shed light.

For now, however, these are surprising results that should have an immediate influence on decisions about screening programmes and individual choices.

  1. Bretthauer M, Løberg M, Wieszczy P, Kalager M, Emilsson L, Garborg K, Rupinski M, Dekker E, Spaander M, Bugajski M, Holme Ø, Zauber AG, Pilonis ND, Mroz A, Kuipers EJ, Shi J, Hernán MA, Adami HO, Regula J, Hoff G, Kaminski MF; NordICC Study Group. Effect of Colonoscopy Screening on Risks of Colorectal Cancer and Related Death. N Engl J Med. 2022 Oct 9. doi: 10.1056/NEJMoa2208375. Epub ahead of print. PMID: 36214590.
  2. Dominitz JA, Robertson DJ. Understanding the Results of a Randomized Trial of Screening Colonoscopy. N Engl J Med. 2022 Oct 9. doi: 10.1056/NEJMe2211595. Epub ahead of print. PMID: 36214591.
  3. Mandrola J, Prasad V. Screening Colonoscopy Misses the Mark in its First Real Test. Sensible Medicine. Oct 10, 2022
  4. Chen A. In gold-standard trial, invitation to colonoscopy reduced cancer incidence but not death. STAT. Oct 9, 2022