Hello all,
This coming month’s journal club will look at the diagnostic accuracy and utility of CT angiogram in the evaluation of patients presenting to the ED with acute GI hemorrhage.
The PGY-1, PGY-2, and PGY-4 articles will be appraised using the Diagnostic Test form; the PGY-3 article will be appraised using the Meta-Analysis form.
Vignette
You’re working a shift at one of your community emergency medicine sites one evening when you encounter Mr. X, a 65-year-old gentleman presenting with bright red bloody bowel movements. He endorses 3 large, bloody bowel movements in the last 2 hours. He has a past history of hypertension, hyperlipidemia, and coronary artery disease with two prior stents. His medications include clopidogrel and baby aspirin. He denies abdominal pain, alcohol use, or recent NSAID or steroid use.
While his initial vital signs are stable, Mr. X becomes hypotensive while you are awaiting his initial labs. You ask the nurse to hang two units of blood, start the patient on a proton pump inhibitor, and place a consult to the gastroenterologist on-call. The patient’s hemoglobin comes back at 6.3 g/dL (baseline 11.4) and his PTT and INR are normal. GI calls back and asks that you continue to resuscitate the patient and admit to the ICU for colonoscopy the following morning.
When speaking with the intensivist, she recommends getting a CT angiogram to evaluate for a source of bleeding and potentially facilitate embolization to achieve hemostasis. You gladly order the CTA but wonder whether there are any studies on the clinical utility of this modality, which specific patients benefit, and whether there is any benefit in presumed upper GI bleeds. You head to the literature and begin your search…
PICO Question
Population: Adult patients presenting with upper or lower GI hemorrhage
Intervention: CT angiography to identify and localized sites of bleeding
Comparison: Standard of care
Outcome: Hemostasis, transfusion needs, hospital length of stay, need for surgical
intervention, mortality
Article 1: Yoon W, Jeong YY, Shin SS, Lim HS, Song SG, Jang NG, Kim JK, Kang HK.
Acute massive gastrointestinal bleeding: detection and localization with arterial
phase multi-detector row helical CT. Radiology. 2006 Apr;239(1):160-7. [Answer Key].
Article 2: Kim J, Kim YH, Lee KH, Lee YJ, Park JH. Diagnostic Performance of CT
Angiography in Patients Visiting Emergency Department with Overt Gastrointestinal
Bleeding. Korean J Radiol. 2015 May-Jun;16(3):541-9. [Answer Key].
Article 3: He B, Yang J, Xiao J, Gu J, Chen F, Wang L, Zhao C, Qian J, Gong S. Diagnosis
of lower gastrointestinal bleeding by multi-slice CT angiography: A meta-analysis.
Eur J Radiol. 2017 Aug;93:40-45. [Answer Key].
Article 4: Erno J, Gregoski MJ, Rockey DC. Diagnostic utility of CT angiography
compared with endoscopy in patients with acute GI hemorrhage. Gastrointest
Endosc. 2024 Feb;99(2):257-261.e5. [Answer Key].
Bottom Line
Use of CT angiography (CTA) in the diagnostic evaluation of GI hemorrhage has been well-studied over the last few decades, and the most recent guidelines from the American College of Gastroenterology note that CTA can accurately locate the bleeding source in hemodynamically significant lower GI bleeding. They further note that embolization within 90 minutes of positive CTA can stop bleeding 98% of the time.
A systematic review and meta-analysis published in 2017 identified 14 studies assessing the diagnostic accuracy of CTA for GI hemorrhage. When combined, the authors found a pooled sensitivity and specificity of 90% and 92%, respectively, correlating to a positive likelihood ratio (LR) of 8.15 and negative LR of 0.16. Unfortunately, the authors of this meta-analysis provide no assessment of the quality of the included studies (QUADAS-2) and combined the results of clinically and methodologically heterogeneous studies, resulting in I2 values ranging from 59% to 90%. Specifically, the authors do not talk about how the different studies established true and false positive and negative CTA (identification of bleeding vs. localization, what was used as the “gold standard”) and did not attempt to understand the clinical impact, if any, or CTA.
A retrospective study out of Seoul, Korea identified 111 ED patients who underwent CTA for “overt” GI bleeding between July and December 2010. A definite bleeding focus was identified in 20.7% of patients. Compared with subsequent testing, CTA had a sensitivity and specificity of 84.8% and 96.9%, respectively, correlating to a LR+ of 27 and negative LR of 0.16. There was no single “gold standard” used in this study; instead, multiple methods of subsequent testing were used to evaluate GI bleeding (partial/differential verification bias). Again, there was no assessment of the impact of findings on CTA on subsequent decision-making.
A similar retrospective study conducted at the Medical University of South Carolina between January 1, 2010, and December 1, 2021 enrolled patients with GI hemorrhage with a drop of at least 4 percentage points in hematocrit from baseline, and who underwent CTA. There were 177 eligible patients who underwent CTA prior to any other diagnostic study. Again, there was no single “gold standard” and any follow-up diagnostic study was used as the criterion standard. CTA had a sensitivity and specificity for detecting GI bleeding of 32% and 87%, respectively, with positive and negative LRs or 2.46 and 0.78. The low accuracy observed in this study when compared to other studies is likely multifactorial, including use of a dichotomous outcome (“evidence of bleeding or not”) for CTA in this study rather than consideration of likely sources of bleeding without active extravasation, use of different confirmatory tests, and the prolonged period of patient inclusion spanning 11 years.
The final paper reviewed was a prospective study from South Korea that included only patients with “acute massive GI bleeding,” defined as either requirement of transfusion of at least 4 units of blood during a 24-hour period or hemodynamic instability (SBP < 90 mmHg). All 26 included patients underwent arterial phase multi-detector row CT followed by formal angiography. The overall location-based test characteristics for the detection of GI bleeding on multi–detector row CT were as follows:
· Sensitivity 90.9%
· Specificity 99%
· Accuracy 97.6%
· Positive predictive value 95%
· Negative predictive value 98%
· LR+ 98
· LR- 0.09
There was one false positive CT scan and two false negative CT scans, all involving bleeds in the duodenum.
These data, which varied primarily based on the manner in which CTA results were confirmed to be positive or negative, do not support the routine use of CTA in GI hemorrhage. The final study suggests a high degree of accuracy for patients with “massive” hemorrhage, which is promising for those patients in whom more urgent evaluation is felt to be indicated. Unfortunately, none of the included studies assessed the impact of CTA on patient management, a key component to understanding the utility of any diagnostic test and potential driver of future research on this subject.