P270 Ultrasonography interobserver agreement in Crohn Disease between experienced and non expert operators
Marra, A.(1);Dell'Aquila, P.(2);Di Leo, A.(3);Principi, M.B.(3);
(1)IRCCS "Casa Sollievo della Sofferenza", Gastroenterology and Endoscopy Units, San Giovanni Rotondo, Italy;(2)University of Bari " Aldo Moro", Internal Medicine, Bari, Italy;(3)University of Bari " Aldo Moro", Section of Gastroenterology- Department of Emergency and Organ Transplantation-Internal Medicine, Bari, Italy; Ultrasound Study Group
Background
Bowel ultrasound (US) is a safe, noninvasive, accessible, reproducible, and cost-effective resource technique for assessing bowel inflammation in CD. Despite the performance and advantages of bowel US, its incorporation into routine gastroenterology practice is internationally inconsistent. To date, this technique is often branded as ‘‘operator -dependent’’ and it has value only in expert hands. Nevertheless, some evidence showed that bowel US in CD could be standardized with a good reproducibility among expert operators. This result has been confirmed more recently. Interobserver agreement between operation with several degree of experience in bowel US and its learning curve need to be investigated further. The aim of this study is to assess the interobserver agreement between experienced and non-experts sonographers in the evaluation of established bowel US parameters used in CD.
Methods
Twenty-eight patients with an established diagnosis of CD underwent bowel US examination by an experienced sonographer. After baseline examination, a small intestine contrast-enhanced ultrasonography (SICUS) by use of polyethylene glycol (PEG) was also be performed. Every exam was recorded and for each segment of the bowel a videotape was registered at baseline and after PEG intake. Every video was showed to both 10 blinded expert and 10 blinded non- expert operators. Interobserver agreement was scored by kappa statistics and AC1.
Results
Agreement among 16 sonographers for the evaluation of bowel wall thickening was high (K = 0.70; AC1 = 0.82), comparable between expert operators (K = 0.72; AC1 = 0.81) and non-experts (K = 0.69; AC1 = 0.83). Agreement was moderate for stratification and vascularization according to Cohen's analysis (k = 0.60), and excellent at AC1 with values of 0.90 and 0.84, respectively. The presence of lymph nodes was reproducible in both groups with k values of 0.41 and 0.51, whereas that on abdominal free fluid was poor (0.09- 0.16). For the evaluation of complications (stenosis, fistulas, abscesses), the agreement differs significantly, resulting poor with Cohen's K (0.14-0.37) and excellent with AC1 analysis (0.98-0.94), in both groups.
Conclusion
In this study, most of the US parameters used in CD showed moderate/substantial agreement without any significant difference between experienced and non-expert operators. A different agreement based on the statistical analysis for some parameters was observed. Standardized training between operators would be useful to reduce the variability of the learning curve.