To the Editor: Bolin et al,1 in their editorial accompanying articles by Yusoff et al2 and Bampton et al,3 took the opportunity to make their case for endoscopic screening for colorectal cancer. We believe that their editorial is seriously misleading.

1: It is misleading to suggest that the 27 case–control and cohort studies in the meta-analysis by Johns and Houston4 stratified the index case by age at diagnosis. The 2.25 risk quoted by Bolin et al refers to the overall risk of first-degree relatives in families with one affected relative. In those studies in which age was stratified in the meta-analysis, there is a spectrum of risk, with families with onset of bowel cancer at an older age having lifetime relative risks much less than the average. A subanalysis of seven studies with age stratification showed the risk to be 1.82 (95% CI, 1.47–2.25), where the index case was over 59 years at diagnosis.

Whether a 1.8-fold risk elevation warrants colonoscopic surveillance could be debated. "First do no harm" is an important axiom in well-patient screening, so one should aim for an order of magnitude of benefit over risk, which in this situation is not secured until the patient being screened is older than the suggested 40 years of age.

2: The recommendation that colonoscopic follow-up of patients with only small, tubular, distal adenomas can be at less frequent intervals is not based on the US National Polyp Study,5 as suggested in the editorial. It is based on the large cohort study of Atkin et al,6 who reported that patients with this finding were actually at below-average risk (relative risk, 0.5) for subsequent colorectal cancer after prolonged follow-up. This occurred despite the inevitable "miss rates". The editorial by Bolin et al handles this issue unconvincingly. The main message of the US National Polyp Study5 was that follow-up (except in exceptional circumstances of numerous polyps, or incomplete removal of malignant polyps) is not needed at 12 months — after 3 years is adequate. The National Health and Medical Research Council guidelines extend this to 4–6 years in the low risk groups, as defined by Atkin et al.6 The Atkin et al data, however, are only Level 3 evidence.

3: Bolin et al1 completely miss the point about pilot programs of screening with faecal occult blood testing (FOBT). There is no intention to confirm evidence of mortality reduction. The pilot studies are neither designed to, nor capable of, doing this. Mortality reduction from FOBT is well established on Level 1 evidence. The pilot studies are in place to answer the very practical questions of

• how to implement large-scale screening programs in Australia;

• what logistic and resource issues are involved;

• how compliance and acceptance will best be secured; and

• how to approach difficult-to-access populations (perhaps with low health insurance rates as distinct from populations well supplied by colonoscopy services).

The central issue in advocating a menu of options to individuals versus more prescriptive screening (based on FOBT) is whether the height of the scientific bar should be at Level 1 evidence or modestly robust Level 3 evidence (flexible sigmoidoscopy) or the less robust Level 3 evidence (colonoscopy), complemented by certain appeals to logic (carefully crafted in the editorial). Medical initiatives based on less than Level 1 evidence have a history of being shown to be wrong, and the concept of colonoscopic surveillance is not immune from this outcome. Where a significant outlay from the public purse is involved, the Federal Government is being appropriately prudent in acting on Level 1 evidence.