Performance of gastrointestinal pathologists within a national digital review panel for Barrett's oesophagus in the Netherlands: results of 80 prospective biopsy reviews

Abstract

AIMS: The histopathological diagnosis of low-grade dysplasia (LGD) in Barrett's oesophagus (BO) is associated with poor interobserver agreement and guidelines dictate expert review. To facilitate nationwide expert review in the Netherlands, a web-based digital review panel has been set up, which currently consists of eight 'core' pathologists. The aim of this study was to evaluate if other pathologists from the Dutch BO expert centres qualify for the expert panel by assessing their performance in 80 consecutive LGD reviews submitted to the panel. METHODS: Pathologists independently assessed digital slides in two phases. Both phases consisted of 40 cases, with a group discussion after phase I. For all cases, a previous consensus diagnosis made by five core pathologists was available, which was used as reference. The following criteria were used: (1) percentage of 'indefinite for dysplasia' diagnoses, (2) percentage agreement with consensus diagnosis and (3) proportion of cases with a consensus diagnosis of dysplasia underdiagnosed as non-dysplastic. Benchmarks were based on scores of the core pathologists. RESULTS: After phase I, 1/7 pathologists met the benchmark score for all quality criteria, yet three pathologists only marginally failed the agreement with consensus diagnosis (score 68.3%, benchmark 69%). After a group discussion and phase II, 5/6 remaining aspirant panel members qualified with all scores within the benchmark range. CONCLUSIONS: The Dutch BO review panel now consists of 14 pathologists, who-after structured assessments and group discussions-can be considered homogeneous in their review of biopsies with LGD.