Abstract No. 315 - Differentiating variables that predict positive angiograms in gastrointestinal hemorrhage

• Published in: Journal of vascular and interventional radiology Vol. 24; no. 4; p. S137
• Main Authors: Lee, J.H., Rashid, S., Hargreaves, E.L., Yudd, A., Censullo, M.L.
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.04.2013
• ISSN: 1051-0443; 1535-7732
• DOI: 10.1016/j.jvir.2013.01.340

Nuclear medicine tagged red blood cell (NM RBC) studies are frequently performed prior to catheter angiography for detection and treatment of gastrointestinal (GI) tract bleeding. Given the relatively higher sensitivity of nuclear scintigraphy, prescreening prior to catheter angiography would be useful. Here, we retrospectively determined useful predictors for catheter angiography GI bleed detection. A retrospective chart review was performed with patients whom had both a positive NM RBC study and a catheter angiogram for the evaluation of GI bleeding. Each study was evaluated for time elapsed before positive NM RBC study (NM RBC Time) and before performance of catheter angiogram (CA Time). In addition, clinical variables including severity of bleed, Intensive Care Unit (ICU) status, cardiovascular vital signs, changes in hemoglobin/hematocrit (H/H), and transfusion status were evaluated. Simple t-tests were used to identify differentiating variables, which were then incorporated into predictive discriminant functions. 95 studies were derived from 85 patients. 30/95 were positive on CA for GI bleed (80% colon; 20% small bowel). There was no bias on the outcome of the catheter angiography for gender (p=0.66) or age (p=0.19). NM RBC time correlated negatively with a positive catheter angiogram (p=0.03) with 93% becoming positive within 60 minutes. Similarly CA time was also negatively correlated (p=0.02) with 90% of the positive CA performed by 8 hours. Clinical factors at the time of referral including ICU admission (p=0.1), number of transfusions (p=0.15), laboratory values such as change in H/H (p=0.25), and vital signs including hypotension (p=0.48) and tachycardia (p=0.18) did not differentiate CA outcome. Discriminant functions involving 5 differentiating variables were able to successfully predict CA outcomes 79% of the time: 1) severity of hemorrhage 2) NM RBC Time, 3) hemoglobin, 4) heart rate at time of referral, and 5) difference in heart rate from the patient’s baseline. Statistically significant predictors for positive CA are routinely available at time of consultation. They include prompt NM RBC Time and prompt performance of CA.