Abstract
Background: Little is known about the sensitivity, specificity, and predictive values of transabdominal ultrasonographic (US) findings in a teaching hospital setting. Methods: We carried out a prospective study including 227 patients with symptoms suggestive of inflammatory bowel disorder. The Picker 9200 CS equipment (5-mHz curved-array probe) was used to obtain bowel images. Gastrointestinal endoscopy, enteroclysis, bowel enema, computed tomography scan, or bowel surgery was used as reference. Results: Of 227 patients, 168 had pathologic findings of the bowel as final diagnosis. The overall sensitivity of US was 76%, whereas the positive predictive value was 98%. Overall specificity was 95%. The negative predictive value for bowel disorders was only 58%, since US missed pathologic findings in 48 patients. Subgroup analysis showed a sensitivity of 84% for Crohn's disease, 66% for ulcerative colitis, 46% for bowel tumors, and 60% for diverticulitis. Topographic comparisons showed that US detected inflammatory bowel-wall alterations preferentially in the terminal ileum and colon, whereas abnormalities in the duodenum, jejunum, and rectum were frequently missed (sensitivity, 10%-20%). Conclusions: Positive US findings are useful for the diagnosis of bowel processes. US is highly predictive albeit not disease-specific. Negative US examinations, however, do not exclude pathologic bowel processes. A topographic location of pathologic US findings is mostly confined to the colon.