Introduction In order to detect all cases of Barrett's oesophagus (BE) in the population, with the aim of reducing adenocarcinoma associated mortality, screening needs to be considered. Unsedated transnasal endoscopy (TNE) can potentially offer a safer and less expensive alternative to conventional sedated endoscopy (CE) for detecting BE.

Aims (1) To evaluate the sensitivity and specificity of TNE in diagnosing BE compared with CE. (2) To assess patient acceptability of TNE for a future multicentre screening trial.

Methods Patients with known BE under surveillance and those referred for routine endoscopy for other reasons underwent both TNE and CE in a randomised cross-over design. TNE findings of endoscopic BE, biopsy size and presence of intestinal metaplasia (IM) were compared against CE, which was used as gold standard. Pre and post endoscopy short-form spielberger state-trait questionnaires (SF6), a visual analogue scale and a single question addressing preference for endoscopy type were used to measure patient acceptability of the procedures.

Results Of 57 patients randomised, 50 completed the study. All 22 BE cases with columnar lined epithelium (CLE) on CE were correctly diagnosed by TNE. Although the biopsy size of TNE was significantly smaller compared to CE (mean size 2.59 vs 4.74 p<0.0001), it detected an equal number of IM cases and there was one case of low grade dysplasia which interestingly was not picked up by CE. Abstract 019 shows sensitivity, specificity and diagnostic accuracy (including Pearson-Clopper CI) for diagnosing BE and IM using TNE compared with CE as the gold standard. The mean difference between the two procedures in determining the length of BE was 0.02. The 95% Bland and Altman limits of agreement was between −2.08 and 2.13 with 49 cases (98%) falling within the pre-determined level of agreement of ±1.5 cm. The mean pre and post procedure anxiety score for TNE (13.3 vs 14.0) and CE (13.2 vs 13.5) were not significantly different (p=0.15). The post endoscopy experience of TNE was comparable to CE (mean 5.0 vs 4.9, p=0.77).

Conclusion TNE is accurate in diagnosing BE and detecting IM. Despite the smaller biopsy size, it detected an equal number of IM cases and showed potential to detect dysplasia. It was well tolerated and preferred by patients and may be an efficient screening method for BE.