Detection of intestinal metaplasia is subject to significant sampling error. It increases with segment length and number of biopsies taken. In the majority of patients, if sufficient biopsies are taken over time, intestinal metaplasia will be demonstrated. The decision to offer surveillance should not be based upon the presence or absence of intestinal metaplasia at index endoscopy as the risk of dysplasia and adenocarcinoma is similar in both groups.

Barrett's columnar-lined oesophagus is the precursor lesion for oesophageal adenocarcinoma. The overall rate of progression to adenocarcinoma is 0.59% per annum. A large prospective multicentre trial is recruiting to assess the role of aspirin as a chemoprotective agent in prevention of development of cancer as well as cardiovascular protection in patients with Barrett's oesophagus. This retrospective analysis of the large UK National Barrett's Oesophagus Registry database seeks to analyse this question from within its large natural history study cohort. Multicentre UK retrospective cohort compared patients known to have been taking aspirin with those who did not take aspirin during the course of surveillance for columnar-lined oesophagus. End point was development of dysplasia or oesophageal adenocarcinoma. Analysis was undertaken using Cox's proportional hazard ratio. Total follow-up was 3683 patient-years. Eighty-six patients were taking aspirin, 650 were not taking aspirin (reference group). Numbers of patients developing all grades of dysplasia and adenocarcinoma were: 13 aspirin (15.1%) and 97 no aspirin (14.9%) (hazard ratio 0.723, 95% confidence interval 0.410-1.310, P = 0.294), high-grade dysplasia and adenocarcinoma: five aspirin (5.8%) and 25 no aspirin (3.8%) (hazard ratio 0.898, 95% confidence interval 0.340-2.368, P = 0.827) and adenocarcinoma: four aspirin (4.7%) and 16 no aspirin (2.5%) (hazard ratio 1.092, 95% confidence interval 0.358-3.335, P = 0.877). No significant difference was observed in hazard of developing dysplasia or adenocarcinoma between patients taking aspirin and those not taking aspirin during the course of follow-up of surveillance for columnar-lined oesophagus. In conclusion, no difference in risk of development of dysplasia or adenocarcinoma was observed between patients taking aspirin and those not taking aspirin in this large cohort.

Oesophageal adenocarcinoma occurs more commonly in older patients and is more frequent in males than females. Once columnar-lined oesophagus had been diagnosed, there were no other demographic or lifestyle factors which were predictive of the development of incident adenocarcinoma in this cohort.