The value of multidetector-row computed tomography for localization of obscure acute gastrointestinal bleeding.

PURPOSE There are no simple guidelines on when to perform multidetector-row computed tomography (MDCT) for diagnosis of obscure acute gastrointestinal bleeding (AGIB). We used a risk scoring system to evaluate the diagnostic power of MDCT for patients with obscure AGIB. MATERIALS AND METHODS Ninety-two patients with obscure AGIB who were referred for an MDCT scan after unsuccessful endoscopic treatment at presentation were studied. We recorded clinical data and calculated Blatchford score for each patient. Patients who required transfusion more than 500mL of blood to maintain the vital signs were classified as high-risk patients. Two radiologists independently reviewed and categorized MDCT signs of obscure AGIB. Discordant findings were resolved by consensus. One-way ANOVA was used to compare clinical data between two groups; kappa statistics were used to estimate agreement on MDCT findings between radiologists. RESULTS Of the 92 patients, 62 (67.4%) were classified as high-risk patients. Blatchford scores of high-risk patients were significantly greater than those of low-risk patients. Sensitivity for MDCT diagnosing obscure AGIB was 81% in high-risk patients, as compared with 50% in the low-risk. When used in conjunction with selection of the cut-off value of 13 in Blatchford scoring system, the sensitivity and specificity of MDCT were 70.9% and 73.7%, respectively. Contrast extravasation was the most specific sign of AGIB (k=.87), recognition of which would have improved diagnostic accuracy. CONCLUSIONS With the aid of Blatchford scoring system for evaluating the disease severity, MDCT can localize the bleeders of obscure AGIB more efficiently.