It was an uncomfortable moment for people who perform colonoscopies. In October, a massive randomized clinical trial in Europe presented its initial results¹, which suggested that, as a screening tool, colonoscopies don’t save as many lives as expected.

Researchers were perplexed because the procedure had long been considered a true success story in cancer screening. But after the study results were compared with data from other trials, colonoscopy to be seemed less effective than simpler screening methods that assess only part of the colon. Jason Dominitz, a physician at the Veterans Health Administration in Seattle, Washington, says it was like suggesting that mammography on only one breast was better than scanning both. It didn’t make sense.

A media frenzy followed, and headlines were blunt, declaring that colonoscopies might not be effective or prevent deaths at all. But when Dominitz dug deeper, the trial results reflected where and how the study was conducted and the complexity of the questions it was trying to answer. “It is really important to not just read the headline,” says Dominitz, who is also director of the colorectal-cancer screening programme run by the US Department of Veteran Affairs.

The controversial embryo tests that promise a better baby

A closer look at the European study, on its own and in the context of other studies, shows that colonoscopies do in fact substantially reduce the risk of developing colorectal cancer and dying from it. They are still considered by many experts to be one of the best ways to screen for the disease. But for any screening procedure, there are trade-offs both for individuals and at the public-health level. As scientists are working out the details of which tests to recommend, the reaction to the study illustrates how difficult it is to interpret and communicate research on cancer screening.

“It’s really important to look at all the evidence in totality,” says Jennifer Croswell, a public-health researcher at the National Cancer Institute (NCI) in Bethesda, Maryland, who specializes in cancer screening. “This was a complicated trial to sort through.”

Benefits of catching cancer early

Colorectal cancer is the third most common form of cancer and the second leading cause of cancer deaths globally, behind lung cancer. In 2020, 1.9 million people were diagnosed with it and 900,000 died from it around the world. Rates are highest in high-income countries, but are growing in low- and middle-income countries.

There is plenty of evidence to show that detecting the disease early saves lives. If colorectal cancer is found before it can spread, the 5-year survival rate is about 91%, according to the NCI, compared with 15% if the cancer has metastasized. There are signs that screening programmes have made a difference, particularly in the United States. In 2000, 38% of US adults over 50 were getting screened; that figure rose to 66% by 2018. Colorectal cancer rates dropped in that period and deaths fell from about 20 to 13 per 100,000 people.

In 1995, the US Preventive Services Task Force (USPSTF) began to recommend that colorectal-cancer screening start at age 50. But the vast increase in screening is often credited in part to the television news presenter Katie Couric, who broadcast her own colonoscopy in 2000 after her husband died from the disease. Other celebrities have since promoted the procedure, and the USPSTF lowered the recommended starting age for screening to 45 in 2021.

Colonoscopy has long been the most popular form of colorectal-cancer screening in the United States. It is highly valued because it allows doctors to examine the entire colon for signs of cancer and to remove polyps — abnormal growths that can become cancerous — during the process. But scientists have struggled to compare colonoscopy’s effectiveness with that of less invasive methods, such as external imaging, stool-sampling techniques or flexible sigmoidoscopy, which looks at only half the colon.

Colonoscopy is much less common in Europe, in part owing to questions about whether the test is too invasive and expensive to be worth recommending, says Michael Bretthauer, a gastroenterologist at the University of Oslo. To address these questions, he and his colleagues planned a randomized trial of colonoscopies. Starting in 2009, they recruited more than 84,000 people aged 55 to 64 from Norway, Poland and Sweden. Some were invited to get screened. Others received their usual health care but no such invitation.

With about ten years of follow-up data, Bretthauer and colleagues released their attention-grabbing results in October 2022, seemingly suggesting that colonoscopies had a smaller benefit than expected. There was just an 18% reduction in the risk of developing cancer among those who had been invited to get colonoscopies, and no significant reduction in the risk of death.

But the study itself offered layers of interpretation that cast colonoscopies in a more favourable light. Overall, only 42% of people in the group that had been invited to get colonoscopies actually got one. If the compliance rate had been 100%, the researchers’ analysis showed, the test would have reduced cancer risk by 31% — from 1.22% to 0.84% — and it would have reduced the risk of death from colorectal cancer by 50% — from 0.3% to 0.15%.

Those benefits are significant, says Chyke Doubeni, a family doctor and colonoscopy researcher at the Ohio State University in Columbus, and there are reasons to think that they could be larger in other circumstances, especially in populations that experience disproportionately high rates of the disease. And despite the huge scale of the European study, ten years of follow-up is a relatively short period of time for colorectal-cancer development, says Amy Knudsen, who studies disease simulation models to inform cancer-care policy at Massachusetts General Hospital and Harvard Medical School, both in Boston. “I think we’re only going to see the impact of colonoscopy increase the longer we follow up,” she says. The European study is continuing to track participants.

Dominitz also notes that roughly one-third of endoscopists in the study found polyps in less than 25% of colonoscopies performed. Typically, more than half of people develop polyps in their lifetimes, he says. The low rate of detection, primarily in Sweden, reflects relatively low risks of colorectal cancer in that country for reasons that have yet to be clarified, Bretthauer says.

Could CAR-T-cell therapy offer hope to children with cancer?

Some of the confusion about the study’s results might be related to the kinds of question that it was trying to answer, which came from both a public-health and an individual-health perspective, says Bretthauer. “If I talk to you as a patient, I would say, ‘You could probably reduce your risk of getting colon cancer by 31%,’” he says. “If I talk to a politician who says, ‘I’m wondering if I should start to introduce this,’ I would say, ‘Well, you can expect an 18% effect in your city.’” But that distinction was not often picked up in coverage of the study, says Knudsen.

“In the US media, I feel like the headlines were all talking about the trial results showing that colonoscopy was not good, and I don’t think that’s what the trial shows at all,” she says. “I was very disappointed.”

Other evidence

No single study can be used in isolation to evaluate something as complicated as cancer screening. For years, evidence has accumulated in support of colonoscopies, including several large studies that have suggested that the procedure has major benefits in some cases. In a 2021 review², the USPSTF found that, across observational and modelling studies, the risk of developing colorectal cancer is lower for people who get screening colonoscopies, with estimates ranging from a 40% to 69% reduction, and that their risk of death is also lower, with estimates ranging from a 29% to 88% reduction (see ‘Catching cancer’). For every 1,000 people screened with colonoscopy, modelling suggests that as many as 28 lives are saved by early detection.

Although the European study has been portrayed as disappointing, Doubeni says the results are consistent with previous data. Randomized trials are important, he adds, but observational studies have the benefit of reflecting real-world conditions. “I think there’s fairly good evidence at this point, even outside of clinical trials, that colonoscopy is effective,” he says. “I have no doubt about that.”

There are other options for screening: CT colonography (which uses computed tomography imaging instead of a scope), flexible sigmoidoscopy and two main types of stool test. One, called a fecal immunochemical test (FIT), uses antibodies to detect signs of haemoglobin (a protein in red blood cells) in stool, which can be indicative of cancer. The other, FIT–DNA, detects haemoglobin and DNA that gets shed from the lining of the colon and rectum.

Each type of test comes with trade-offs between potential benefits, risks, discomfort, cost and logistics. Getting a colonoscopy, for example, requires going 24 hours without solid food and undergoing sedation while a skilled doctor inserts a flexible tube equipped with a camera into the rectum. Complications can happen — studies show that, for every 10,000 procedures, there are an average of 3 bowel perforations², which sometimes require another surgery to repair. And even without complications, recovery is not instantaneous.

Stool tests don’t involve bowel prep or sedation, but they need to be done as often as once a year; by contrast, colonoscopies are needed once every 10 years as long as the results are negative. Any positive result from another type of test requires a colonoscopy to confirm it, Doubeni says. And false positives are common, with rates as high as 13% for FIT-DNA and up to 5% for FIT.

A study published last year³ found that people who didn’t get a follow-up colonoscopy after a positive FIT stool test were twice as likely to die as those who did. “There is no test to do for colon-cancer screening that will not require colonoscopy to be done for you to get a benefit,” he says. “It is important to underscore this.”

How to improve the diagnosis of prostate cancer

Modelling suggests that the benefits are similar across screening types. In an analysis for the latest USPSTF report by the NCI-funded Cancer Intervention and Surveillance Modelling Network⁴, the number of lives saved ranged from 24 per 1,000 people screened with flexible sigmoidoscopy to 28 per 1,000 for colonoscopy. “They’re all in the same ballpark,” Dominitz says. “The maximum spread here is four lives, and for those four people and their families, of course this is huge. But they’re all really close. It’s not like one is really dominant over the other.”

Trials that have tested screening methods head-to-head suggest that compliance has a lot to do with these comparable success rates. Colonoscopies might be the best at seeing everything in the colon, but fewer people are motivated to get them. In a Swedish trial called SCREESCO⁵, 35% of people who were advised to get a colonoscopy got one, compared with 55% of people in the FIT group. And in Spain, a study⁶ called COLONPREV randomized more than 53,000 people into two groups — one received advice to get a colonoscopy, and the other was told to undergo FIT stool screening every two years. After 10 years, 30 cancers had been detected in the colonoscopy group, compared with 33 in the FIT group — largely because of people’s behaviour. Only 25% of people in the colonoscopy group got screened, compared with 34% in the FIT group.

Dominitz is co-leading a study that is directly comparing cancer outcomes in 50,000 veterans who will be screened with either an annual FIT test or colonoscopy. So far, adherence in the study has been high, and screening colonoscopies have found precancerous polyps in 46% of participants. It’s the first study of its kind in the United States, and, Dominitz says, it should more clearly help people to make decisions about how to get screened. “It gets more to the question of the patient in front of you who says, ‘I want to be screened. What should I do?’” Results will be available in 2028.

People looking for guidance now will get different advice depending on where they live. Canada and the United Kingdom recommend stool-based screening (starting at age 50 in Canada and 60 in the United Kingdom), whereas the push for colonoscopies is mostly in the United States.

Bretthauer thinks that people should make their own decisions with all of the information available; he chose to get a colonoscopy, but he has colleagues in other countries who did not, on the basis of the same data.

Offering choices seems to be key to increasing screening rates in people of all ages. In one study⁷, 69% of people who were advised to get a colonoscopy or stool testing got screened, compared with 38% who were advised to get just a colonoscopy. “For a screening test to be successful, it has to be acceptable to patients,” Croswell says, and colonoscopies can be hard to stomach. “There’s a yuck factor to having a scope stuck up your rectum. People are not necessarily excited about that.” Compliance could also suffer if the test is too costly or difficult to schedule.

Researchers are continuing to monitor evidence to refine guidelines — but specialists say that, for colorectal cancer, getting any kind of screening is better than nothing. It’s still not clear which test is best, Dominitz says. But stool tests, sigmoidoscopy and colonoscopy are all effective, as long as people do them. “Screening only works,” he says, “if it gets done.”