CT colonography for detecting colorectal cancer and polyps
Doug Altman, Susan Mallett, Jon Deeks with Steve Halligan (CR-UK, St. Mark’s Hospital)
CT colonography (‘virtual colonoscopy’) is a novel CT-based technique for detection of colorectal cancer and precursor polyps. The technique has disseminated rapidly into the healthcare system both in the UK (where it tends to be used in patients with symptomatic cancer) and abroad (where the focus is more on screening). There are now a number of reports detailing within-patient comparisons with conventional video-endoscopy. The European Association of Radiology and European Society of Gastrointestinal and Abdominal radiology (ESGAR) commissioned Steve Halligan to investigate the new technology with the aim of producing pan-European guidelines for research and accreditation.
The first stage in the research process was a systematic review of the new health technology, the preliminary results of which were orally presented by Dr Halligan at the annual meeting of ESGAR in June 2003. A paper describing the findings has been submitted for publication.
During the review process it became clear that generally used methods of data collection were illogical and their presentation often incomplete. This hinged around a complex interaction between the size of the target lesion (colorectal polyps), the category assigned (small, medium, large), and the risk of developing subsequent cancer based on that category. Further, the reference standard (video-colonoscopy) is often inaccurate and also unblinded to the significance of category. The net result is that method comparison studies have been relatively poor to date. We have developed guidelines for sensible data collection for method comparison studies of CT colonography and to recommendations for meaningful analysis. ESGAR has committed to endorse and disseminate these recommendations.
It is clear from our analysis of accumulated data that measurement error during CT colonography has been relatively neglected. This is surprising since the risk assigned to a polyp is directly dependent upon size. We plan to carry out a study to determine the inter- and intra-observer measurement error for polyps identified by CT colonography. We will also investigate whether there is a difference in performance of readers of differing experience.
Our collaboration with ESGAR has spawned ESGAR1, a multi-centre study of observer performance for CT colonography, also led by Dr Halligan. This study aims to determine the observer performance of recognised CT colonography ‘experts’ from 10 to 15 European centres, with the aim of developing evidence-based guidelines for accreditation in the technique. ESGAR1 will also determine the performance of ‘novices’ who have been subject to directed training in the technique in an attempt to develop evidenced-based guidelines for training. We will provide both methodological advice and statistical support for the study.
