Diagnosing Barrett's oesophagus: factors related to agreement between endoscopy and histology

Few previous studies have addressed the agreement between endoscopy and histology regarding Barrett's oesophagus in unselected endoscopy patients. Our aim was to quantify this agreement, and to study its relation to clinical and endoscopic characteristics in consecutive patients coming for first-time gastroscopy.

We invited consecutive patients aged 18-79 years and endoscoped for the first time at endoscopy units exclusively serving defined catchment areas in southeast Sweden. Endoscopic and clinical data were recorded according to a predetermined protocol, and biopsies were taken from the distal oesophagus in all patients.

Among 705 patients included, 17% [95% confidence interval (CI): 14-20] had endoscopically visible columnar mucosa above the oesophagogastric junction and 38% (95% CI: 34-42) had columnar mucosa in at least one biopsy irrespective of the endoscopic finding. The overall concordance between endoscopy and histology regarding presence (or absence) of columnar mucosa above the oesophagogastric junction was 74% (95% CI: 71-77) and the agreement beyond chance, as measured by Kappa (kappa) statistics, was fair, kappa=0.38 (95% CI: 0.32-0.45). The agreement between the endoscopic assessment and intestinal metaplasia at biopsy was 86% (95% CI: 83-88), but kappa was only 0.31 (95% CI: 0.21-0.41). Our data were consistent with a lower threshold for macroscopic detection of columnar epithelium above the oesophagogastric junction, when risk factors for Barrett's oesophagus were present.

The agreement between macroscopic and microscopic assessments of Barrett's oesophagus is no more than fair, and partly dependent on the presence of patient characteristics suggestive of pathology in this region.