ADWE-07 How many capsule endoscopy cases can be read before accuracy is affected?. (8th June 2018)

• Record Type: Journal Article
• Authors: Beg, Sabina; Card, Tim; Sidhu, Reena; Wronska, Ewa; Ragunath, Krish
• Is Part Of: Gut. Volume 67(2018)Supplement 1
• Page Start: A164
• Page End: A164
• Publication Date: 2018-06-08
• Subjects: Gastroenterology -- Periodicals; 616.33
• DOI: 10.1136/gutjnl-2018-BSGAbstracts.326
• Languages: English
• ISSNs: 0017-5749

Abstract

Abstract : Introduction: The interpretation of Small Bowel Capsule Endoscopy (SBCE) requires a high level of concentration. An abnormality may be present on just a few of the many thousands of images presented for interpretation. It is unknown whether fatigue affects the accuracy of SBCE reporting or how many SBCE can be read in one session. Methods: 32 participants (16 Experienced readers and 16 Novices) were invited to participate in this study. Each was asked to read 6 consecutive pre-selected SBCE cases, these were presented in a random order. All readings took place using the single view mode, with readers able to choose the frames per second viewed from a pre-defined range. Fatigue was measured subjectively using a Likert scale and objectively using a computer based Psychomotor Vigilance Test (PVT). These measures were performed at prior to commencing the study and after every second capsule read. Accuracy in lesion detection was determined by comparison with a gold standard reading, derived from the non-consecutive readings of two experienced readers. Accuracy was plotted against reading order. Results: In keeping with published data, high intra-observer variability amongst the participants was observed. Experienced readers demonstrated a mean correct detection rate of 48.3% (SD:16.1), compared to 21.3% (SD:15.1) amongst Novices. The accuracy of Experienced readers declined after interpreting just a single SBCE case (p=0.01) and plateaued thereafter. Novice readers demonstrated no significant change across time points, with a trend towards improvement, perhaps indicating skill acquisition during the study. The mean reading time to read a single SBCE case was 32 mins. When analysed with respect to reading order a statistically significant reduction in reading time was observed (p=0.05). Reading times were on average 25% faster when reading Case 6 compared to Case 1, representing a mean reduction of 9 mins and 36 secs (range 9–11 mins). Reading at higher frame rates was associated with a reduction in accuracy, which was most pronounced amongst Novice readers. No significant relationship between subjective fatigue or PVT and correct lesion detection was demonstrated. Conclusions: This is the first study to demonstrate that accuracy in SBCE declines after reading a single capsule study. This phenomenon should be considered when reading high risk cases or when a SBCE case has been reported as normal, despite high clinical suspicion.