Bowel Cancer Screening: some of the facts
A group of critics of breast cancer screening have indicated that they may seek judicial review of the contents of the leaflet sent out to women asking them to attend for mammographic screening.
The signatories of a letter to The Sunday Times (July 31) included Professor Michael Baum of University College London and Peter Gotzsche of the Nordic Cochrane Centre, both long-term sceptics of the value of breast screening. Their letter was prompted by a BMJ paper that casts more doubt on whether screening, or simply better treatment, have been responsible for reducing breast cancer deaths. It concludes (1) that screening has no perceptible effect.
The critics argue that the leaflet fails to reflect these uncertainties or provide women with unbiased advice. Under pressure, the NHS produced a revised version in 2010 but it remains “patronising, coercive in tone and parsimonious with the facts” as well as being in breach of General Medical Council guidelines, the letter writers say. Their criticisms have been met with stonewalling. “In desperation we would like to initiate the process for an independent judicial review”.
While they are about it, they might also take a look at the leaflet Bowel Cancer Screening – The Facts, which is sent out to all those over 60. It has been criticised by consultant gastroenterologist James Penston, who devotes a chapter of his book Stats.con to the issue. (Published by London Press and available from Amazon at £12.99, Stats.con is an interesting read for anybody curious about how statistical and epidemiological methods have gained such a stranglehold in medical research. As his title implies, Dr Penston thinks they are a con.)
The most substantial objection to the bowel cancer leaflet is that it plays the familiar trick of expressing the results of screening in relative, not absolute, terms. “Regular bowel cancer screening has been shown to reduce the risk of dying from bowel cancer by 16 per cent” it says, citing the Cochrane database. Actually, Dr Penston says, the figure is 15 per cent at the two-year screening interval the NHS provides, but we’ll let that pass.
The figure is based on good randomised control trials, but like all relative risk figures it exaggerates the benefits. To know what it actually means, individuals need to know their actual risk of dying from the disease, about which the leaflet is silent. It says that “about one in 20” people in the UK will develop bowel cancer during their lifetime, and 16,000 people a year die of it, both claims that make the risks sound large. Reading the leaflet quickly might easily cause you to conflate the two claims and conclude that one in 20 people dies of colon cancer.
In fact, the figure in the most recent RCT (2) was 0.8 per cent over ten years, which was reduced to 0.7 per cent among those who were invited to screening. That means for every 1,000 people invited for screening over ten years, a single death from bowel cancer is prevented. Among those who actually accept the invitation and are screened – about half of those invited – the odds are better. For this group, one in 500 who participated in screening for ten years was saved from dying from bowel cancer. But if the leaflet said that, relatively few would sign up.
The position is actually slightly worse than this, because the trial showed no reduction in the overall risks of death. When groups invited to be screened were compared with control groups not issued with invitations, all-cause mortality was unchanged.
This was a very big trial, involving over 150,000 people aged between 45 and 74 and followed up for 18 years. (The very fact that it was so large tells us that the outcome expected was small.) Half were offered screening, half were not. Over the 18 years, 20,421 people offered screening died from a variety of causes (26.8 per cent) while 20,336 people who were not offered screening died (26.7 per cent). The effect of colon cancer screening on total deaths was so small that it was swamped by other, commoner, causes of death. This, naturally, isn’t mentioned in the leaflet. “Parsimonious with the facts” would be the polite way to categorise this omission. See table, below:
Bowel cancer screening is soon to be extended by using a different technique, flexible sigmoidoscopy. The UK National Screening Committee, which recommended it, said that in people aged 55 to 64 it could reduce mortality by 43 per cent in those screened. But when they come to write the patient leaflet, it’s to be hoped that they express the benefits in absolute terms. “Using propaganda to promote screening programmes in unacceptable” Dr Penston wrote in a letter to BMJ. “If the data are robust they can be presented honestly.”
- Breast cancer mortality in neighbouring European countries with different levels of screening but similar access to treatment: trend analysis of WHO mortality database, Philippe Autier, Mathieu Boniol, Anna Gavin and Lars J Vatten BMJ BMJ 2011;343:d4411 doi: 10.1136/bmj.d4411
- Effect of faecal occult blood screening on mortality from colorectal cancer: results from a randomised control trial, JH Scholefield, S Moss, CM Mangham and JD Hardcastle, Gut, 2002, June 50(6) 840-844