Bowel cancer screening and noise to signal ratio

What a lot of media noise about a study just out in the Lancet on one-off bowel cancer screening using sigmoidoscopy, a test to examine the lower part of the bowel – usually in outpatients, with no sedation. It’s a well designed, large study, true.  But no wonder about the fuss, given the press release from the Lancet, which starts:

The value of cancer screening programmes has been a source of much debate over the years. Colorectal cancer continues to be a major cause of death-it is the third most frequently diagnosed cancer worldwide, accounting for more than 1 million cases and 600,000 deaths every year. In new research published in The Lancet, researchers announce findings of a UK-based trial, which started 16 years ago, to assess the merit (in terms of reduction in mortality and incidence) of a single sigmoidoscopy examination in patients aged 55-64 years to screen for colorectal cancer. Cancer Research UK Chief Executive Harpal Kumar will join the papers authors on the press conference panel to discuss the implications of the study.


A single examination of the lower colon and rectum using sigmoidoscopy, between the ages of 55 and 64 years, reduced colorectal cancer mortality by 43% in those screened and incidence by one third. These are findings of a long-term UK study reported in an Article Online First and in an upcoming edition of The Lancet. The Article is written by Professor Wendy Atkin, Imperial College London, UK, and Professor Jane Wardle, University College London, UK, and colleagues from the UK Flexible Sigmoidoscopy Trial Investigators. The trial was funded by the UK Medical Research Council, Cancer Research UK, The UK National Institute for Health Research, and KeyMed.

Colorectal cancer is the third most frequently diagnosed cancer worldwide, accounting for more than 1 million cases and 600 000 deaths every year. Survival is strongly related to stage at diagnosis, with survival rates of 90% for localised cases. Current screening methods using the faecal occult blood test, which detects early cases, reduce mortality by around 15%, and many countries have introduced screening programmes based on this test.


113,195 people were assigned to the control group and 57,237 to the intervention group, of whom 112,939 and 57,099, respectively, were included in the final analyses. 40,674 (71%) people underwent flexible sigmoidoscopy. During screening and median follow-up of 11 years, 2,524 participants were diagnosed with colorectal cancer (1818 in control group vs 706 in intervention group) and 20,543 died (13 768 vs 6775; 727 certified from colorectal cancer [538 vs 189]). In intention-to-treat analyses (which included people assigned to screening but who did not attend), colorectal cancer incidence in the intervention group was reduced by 23% and mortality by 31%. Incidence of distal colorectal cancer (rectum and lower colon) was reduced by 50%. Incidence of colorectal cancer in people attending screening (excluding non-attendees) was reduced by 33% and mortality by 43%. The numbers needed to be screened to prevent one colorectal cancer diagnosis or death, by the end of the study period, were 191 and 489, respectively.

The authors say: “After 11 years of follow-up, colorectal cancer incidence was reduced by a third and colorectal cancer mortality by more than 40% in those who underwent screening. Confining results to the rectum and sigmoid (lower) colon, incidence was reduced by half in those who were screened.” Furthermore, they point out that 59% (126) of the 215 colorectal cancer cases that developed were detected at screening and very few cases were detected post screening—suggesting that screening has a lasting protective effect.”

There is more, but there is very important detail in the paragraph given above. Mainly, most people who have the test do not benefit from it.

The media coverage seems accepting of the presentation that this is a near-miracle intervention. Harpal Kumar, CEO of cancer research UK is seen here saying that sigmoidscopy is “completely safe” and will prevent anxiety. He wants this test introduced immediately. The press release from Cancer Research UK says that

“…Today the remarkable results of a study… provide hard evidence that a one-off screening procedure could prevent a third of bowel cancers and reduce death rates by nearly half. Given that bowel cancer is the third most common cancer and the second biggest cancer killer in the UK, these are extremely important results. Furthermore, we expect the results to get even better as we monitor the people who took part in this study for a few more years. ”

This press release seems to be the one most media outlets are using. But there is no point telling people that you can reduce a risk by half unless you tell them what their risk was to start off with. If you want to read the paper, you will see that almost half the patients contacted to take part did not because they either said no, didn’t respond, weren’t sure, or had their invitations undelivered. You will also see that 5% of people who had sigmoidoscopy screening were referred on for colonoscopy, which examines the whole of the large bowel and requires bowel preparation and sedation. The complication rate from colonoscopy is usually given as 1 in 1000 leading to bowel perforation, and some estimates of mortality in therapeutic colonoscopy (e.g. where abnormalities are biopsied or removed) are about 1 in 3500. I don’t think it’s fair to say that a bowel screening programme would be without hazards. I also don’t think we know enough about how this would be worked in the real world. Previous estimates of uptake of bowel cancer screening have been based on breast cancer screening uptake – which I’d argue is hardly comparable.

I’ve nothing against bowel screening, but what I am concerned about is overly enthusiastic selling of a test as part of a far more complex program which can deliver both benefit and harm. It’s paternalistic but also disingenuous to think that all screening should be presented to the public as a fait accompli, and something which requires us only to join the queue. The bottom line from this study is that there is just over a 1 in 500 chance of the initial sigmoidoscopy leading to further tests, investigations and treatments which will go on preventing your death due to bowel cancer. There is also a 25 in 500 chance of you having further tests after the initial screening, which may come with futher tests and treatment, as well as risks: and in 24 out of those 25 cases it will not change your cause of death: you’ll die of whatever you were going to anyway.

So yes, it’s a good study. But I don’t think this round of press releases has improved our knowledge about screening or ability to make decisions about it one jot.

One Response to “Bowel cancer screening and noise to signal ratio”

  1. Bill MacGregor April 27, 2015 at 7:35 pm #

    Thanks for this useful review from one who has just received an invite for bowel scope screening! What bothered me about the invite was that apart from age, there was no indication given about the risk factors that affect the growth of the pre-cancerous polyps that the scope aims to find. I guess that if I knew that I was at high risk, I’d be more inclined to accept the invite. Would be interested to hear what others think.

Leave a Reply