Structured terminology for capsule endoscopy : Results of retrospective testing and validation in 766 small-bowel investigations

Background and study aims: Capsule endoscopy (CE) is an effective means of investigating the small bowel in patients with gastrointestinal diseases. Computerized reports are frequently used in endoscopy, and the Minimal Standard Terminology (MST) has been promoted by endoscopy societies as the official vocabulary for endoscopy. The aims of this study were to design a lexicon for CE reports based on the principles of the MST and to validate lists of terms for describing findings and reasons for performing a CE by cross-matching them with the results of CE procedures collected during ongoing clinical studies. Materials and methods: A consensus-based Capsule Endoscopy Structured Terminology (CEST) was developed by experts involved in CE studies. Lists of terms suitable for CE were designed for the various sections of an endoscopic report. They were then correlated with the corresponding MST lists for duodenal and intestinal endoscopy. The results of 766 CE procedures, collected in an electronic case record form (eCRF), were analyzed to provide lists of reasons for performing the procedures and of the findings. The eCRF provided only a limited number of items for each data field, along with free-text facilities. Only descriptions pertaining to the small bowel were analyzed. Lists of terms were then reviewed by two experts to group obvious synonyms. The accuracy of the CEST was defined beforehand as the capability to describe 90% of entries. Results: A total of 766 CE procedures were analyzed. The eCRF included 824 entries as reasons for the examination in 655 CEs (1.3 per procedure). These represented 122 different expressions. After grouping of synonyms, 28 expressions remained. Among them, 10 were matched with terms from the list of reasons for performing CE offered in the CEST. These were the most frequently used, accounting for 768 entries in this field (93.2%). All eCRFs contained at least one description of findings. A total of 109 CE procedures were classified as normal (14.3%). A total of 2624 entries for abnormal findings were recorded for 657 procedures (4.0 per procedure). In all, 213 different expressions were used to describe abnormal findings. After grouping of synonyms, 52 expressions remained. Among these, 27 were matched with terms from the list of findings in the CEST, covering 2403 entries (91.6%). Conclusions: In this study, CEST terms were capable of describing more than 90% of the reasons for performance and of the findings in an unselected set of CE procedures. CEST is therefore suitable for use as the standard lexicon for CE reports. Adopted as a standard, it could significantly improve the quality of the data collected and reported in CE studies.