"MURAL" model to predict bleeding from mural-based lesions in potential small bowel bleeding may improve diagnostic capability and decrease cost
Issued Date
2022-12-02
Resource Type
ISSN
00257974
eISSN
15365964
Scopus ID
2-s2.0-85143567972
Pubmed ID
36482571
Journal Title
Medicine (United States)
Volume
101
Issue
48
Rights Holder(s)
SCOPUS
Bibliographic Citation
Medicine (United States) Vol.101 No.48 (2022) , E31989
Suggested Citation
Limsrivilai J., Chaemsupaphan T., Khamplod S., Srisajjakul S., Kositamongkol C., Phisalprapa P., Maipang K., Kaosombatwattana U., Pausawasdi N., Charatcharoenwitthaya P., Leelakusolvong S., Pongprasobchai S. "MURAL" model to predict bleeding from mural-based lesions in potential small bowel bleeding may improve diagnostic capability and decrease cost. Medicine (United States) Vol.101 No.48 (2022) , E31989. doi:10.1097/MD.0000000000031989 Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/87152
Title
"MURAL" model to predict bleeding from mural-based lesions in potential small bowel bleeding may improve diagnostic capability and decrease cost
Author's Affiliation
Other Contributor(s)
Abstract
In potential small bowel bleeding, video capsule endoscopy (VCE) is excellent to detect mucosal lesions, while mural-based lesions are better detected by computed tomography enterography (CTE). A predictive tool to identify mural-based lesions should guide selecting investigations. In this retrospective study, we developed and validated the "MURAL"model based on logistic regression to predicts bleeding from mural-based lesions. Cost-effectiveness analysis comparing diagnostic strategy among VCE, CTE, and MURAL model was performed. Of 296 patients, 196 and 100 patients were randomly included in the derivative and validation cohorts, respectively. The MURAL model comprises 5 parameters: age, presence of atherosclerosis, chronic kidney disease, antiplatelet use, and serum albumin level. The area under the receiver operating characteristic curve was 0.778 and 0.821 for the derivative and validation cohorts, respectively. At a cutoff value of 24.2%, the model identified mural-based lesions with 70% sensitivity and 83% specificity in the validation cohort. Cost-effectiveness analysis revealed that application of the MURAL model demonstrated a comparable missed lesion rate but had a lower missed tumor rate, and lower cost compared to VCE strategy. The model for predicting mural-based lesions provide some guidance in investigative decision-making, which may improve diagnostic efficiency and reduce costs.