Clinical Predictors of a Positive Ct Angiogram Study Used for the Evaluation of Acute Gastrointestinal Hemorrhage.

BACKGROUND: Acute gastrointestinal (GI) bleeding is one of the leading causes of emergency department visits and hospital admissions. CT angiography (CTA) has had an expanding role in the evaluation of acute GI bleeding because it is rapidly performed, widely available, reasonably sensitive and provides precise localization when positive. We attempted to identify patient and clinical characteristics that predict CTA results in order to help guide the utilization of this modality in patients with acute GI bleeding. METHODS: In this retrospective study, we analyzed all CTAs performed for the evaluation of GI bleeding in the Duke University healthcare system between October 2019 and March 2020. We captured patient characteristics including age, sex, vital signs, hemoglobin, platelets, PT/INR, and anticoagulation status. Study indications were grouped by suspected source of bleeding: upper GI bleeding (hematemesis or coffee-ground emesis) vs small bowel bleeding (melena or "dark stools") vs lower GI bleeding (hematochezia or bright red blood per rectum (BRBPR)). Chi-square, Wilcoxon, t test, and multivariate logistic regression were used to describe and assess the relationship between patient characteristics and study outcomes (Table 1). Table 1 Univariate analysis of patient characteristics by CT angiography outcome Patient Characteristics by Positive CT for GI Bleed No (N = 274) Yes (N = 43) Total (N = 317) p value Gender 0.451  Female 138 (50.4%) 19 (44.2%) 157 (49.5%)  Male 136 (49.6%) 24 (55.8%) 160 (50.5%) Age, median (Q1,Q3) 65 (51,75) 70 (62,80) 66 (52, 76)  < 0.012 Heart rate, median (Q1,Q3) 86 (74,100) 89 (72,98) 86 (74, 99) 0.782 MAP, mean (SD) 87.32 (15.52) 81.72 (16.53) 86.56 0.033 Shock index, median (Q1,Q3) 0.70 (0.58, 0.85) 0.78 (0.55, 1.00) 0.71 (0.58, 0.85) 0.352 Hemoglobin 0.332  N 273 43 316  Median (Q1, Q3) 8.50 (6.90, 11.00) 7.70 (6.50, 11.30) 8.45 (6.90, 11.00) Baseline hemoglobin 0.202  N 258 39 297  Median (Q1, Q3) 11.20 (9.40, 13.00) 12.00 (9.40, 14.00) 11.20 (9.40, 13.00) Hemoglobin drop from baseline 0.062  N 258 39 297  Median (Q1, Q3) 2.10 (0.60, 3.70) 2.70 (1.20, 4.80) 2.20 (0.70, 3.80) Platelets, median (Q1, Q3) 219.5 (141, 301) 183 (139, 246) 217 (139, 282) 0.102 INR 0.272  N 263 42 305  Median (Q1, Q3) 1.10 (1.00, 1.30) 1.20 (1.00, 1.30) 1.10 (1.00, 1.30) Anticoagulation 0.131  No 155 (56.6%) 19 (44.2%) 174 (54.9%)  Yes 119 (43.4%) 24 (55.8%) 143 (45.1%) Upper GI bleeding 0.401  No 251 (91.6%) 41 (95.3%) 292 (92.1%)  Yes 23 (8.4%) 2 (4.7%) 25 (7.9%) Small Bowel bleeding 0.761  No 216 (78.8%) 33 (76.7%) 249 (78.5%)  Yes 58 (21.2%) 10 (23.3%) 68 (21.5%) Lower GI bleeding 0.091  No 134 (48.9%) 15 (34.9%) 149 (47.0%)  Yes 140 (51.1%) 28 (65.1%) 168 (53.0%) 1Chi-Square 2Wilcoxon 3Equal Variance T-Test RESULTS: A total of 317 patients underwent CTA between October 2019 and March 2020. Forty-three patients (13.6%) had a CTA positive for active bleeding. Multivariable logistic regression showed that after controlling for age, mean arterial pressure (MAP) and indication, only a hemoglobin drop from baseline was significantly associated with a positive CTA. For each 1 g / dL drop in hemoglobin from the patient's baseline, the odds of a positive CT increased by 1.17 (OR 1.17 95% CI 1.00 - 1.36, p = 0.04). Age (OR 1.02 95% CI 0.99 - 1.04, p = 0.06) and hematochezia / BRBPR (OR 2.09 95% CI 0.94-4.64, p = 0.07) approached statistical significance. CONCLUSIONS: In patients who present to the hospital with GI bleeding, CTA can be a helpful triage tool that is most helpful in older patients with suspected lower GI bleeding with a drop in hemoglobin from baseline. Other clinical factors including MAP and the use of anticoagulants were not predictive of a positive CTA.

Duke Scholars

Author Daniel Murray Wild Medicine, Gastroenterology

Author Alice Parish  

Author Donna Niedzwiecki Biostatistics & Bioinformatics, Division of Biostatistics

Author Charles Yoon Kim Radiology, Interventional Radiology