Ultrasound G.E.L. – GI Bleed Risk

Originally published on Ultrasound G.E.L. on 3/30/20 – Visit HERE to listen to accompanying PODCAST! Reposted with permission.

Follow Dr. Michael Prats, MD (@PratsEM) from Ultrasound G.E.L. team!

Results of a Prospective Study to Evaluate the Impact of Point-of-Care Ultrasound in the Enhancement of Gastrointestinal Bleeding Risk Scores

J Ultrasound Med Feb 2020 – Pubmed Link

Take Home Points

1. In patients with GI bleeding, POCUS evidence of hypovolemia is associated with varied composite outcomes of adverse events.

2. Adding POCUS to GI bleeding risk scores may be helpful, but requires more research for validation and patient centered outcomes.


For once we are not here to talk about the diagnostic accuracy of POCUS. Can POCUS identify gastrointestinal (GI) bleeding?! It has been reported, but it’s probably not that accurate. HOWEVER perhaps ultrasound can help these patients in another way. The problem with GI bleeding is that, like many other disease processes, there is a wide spectrum. It can be challenging to determine who is safe to go home and who has a high risk of an untoward outcome. Of course, several prediction rules based on known risk factors have been developed – but these are not perfect. What if we could not only use risk factors but assess their current physiology? The authors of this study wonder if point of care ultrasound to assess the preload of a patient could be helpful in determining their risk of needing an intervention in the near future.

Check out The Evidence Atlas – Echocardiography for a review of prior evidence on point of care cardiac ultrasound.


Are POCUS parameters associated with adverse events in patients with GI bleeding?

Does integrating POCUS into existing GI bleed risk scores improve their predictive performance?


Single center, Emergency department patients presenting with GI bleeding


  • Adults ≥18 years old
  • GI bleeding in past 24 hours


  • Previous ischemic arterial disease in last 3 months
  • Previous trauma or surgery with blood loss
  • Pregnancy
  • Hemodynamic instability requiring immediate ICU admission


Patients evaluated, get history and physical, labs ordered

For suspected Upper GI Bleeding → Rockall score and Glasgow Blatchford score were calculated

For suspected Lower GI Bleeding → Velayos score was calculated

Prior to or <30 minutes after the start of IV Fluids, point-of-care ultrasound (POCUS) exam was performed by someone not treating the patients.

  • He performed IVC ultrasound, measured collapsibility index
  • He performed focused echo, looked for evidence of hypovolemia, and calculated left ventricular outflow tract VTI
  • He performed these before and after a passive leg raise maneuver

Primary outcome was association of POCUS parameters with adverse events during follow up.

Early Adverse Events were:

  • Rebleeding
  • Additional therapy including blood transfusion
  • Additional intervention to control bleeding
  • Hypotension (mean arterial pressure <60 mm Hg)
  • Hypoxia (<92%)
  • Drop in hemoglobin, pH or increase in lactate, urea, creatinine after 24 hours of stability

Late Adverse Events were:

  • Rebleeding
  • Additional blood transfusion
  • Interventions to control bleeding
  • Re-admission
  • Death within 30 days

Secondary outcome was to see if integrating ultrasound parameters into the existing risk scores would enhance their performance in detecting adverse events.

Follow up during hospitalization and 1 month after recruitment

Calculated they need 75 patients to detect a 24% difference in complications

Who did the ultrasounds?

One emergency physician with good experience with POCUS

The Scan

Phased Array


Their protocol was this:

  1. Inferior vena cava (IVC) in long axis → calculate collapsibility index (max IVC diameter – min IVC diameter)/max IVC diameter). Measured 2 cm from IVC right atrial junction on B-mode (not on M-mode!)
  2. Parasternal long axis view → measured LV outflow tract diameter and recorded the presence or absence of “kissing” LV walls (systolic obliteration of the left ventricle)
  3. Apical 5-chamber view → measured LVOT velocity time integral (VTI). 5 measures at 2 different times and took the highest one
  4. Perform passive leg raise and repeat steps 1-3

Learn how to do Ultrasound for LVOT VTI from 5 Minute Sono!

Learn how to do Ultrasound for IVC from 5 Minute Sono!


Check out Echo Pathology on the POCUS Atlas!



n = 203 patients enrolled over 2 years

  • Mean age late 60s
  • 86.8% had comorbidities
  • 62.6% were lower GI bleed
  • LGI causes were hemorrhoidal (22%), diverticular (19.7%), vascular (18.1%), inflammatory (14.2%), neoplastic (13.4%)
  • UGI causes were ulcer (42.1%), neoplastic (15.8%), varices (15.8%), Mallory-Weiss (13.2%)
  • Risk in LGI was 66.1% intermediate, 27.6% high, 6.3% low
  • Risk in UGI was high in 73.7-81.5% depending on score used
  • 72.4% had early AEs, most commonly decreased hemoglobin
  • 27.6% had late AEs, most commonly intervention to control bleeding

Primary Outcome – POCUS features associated with AEs

UGI + IVC <1 cm had more AEs with OR 1.48 (CI 1.15-1.90, p = 0.036)

LV kissing walls had more AEs had OR 3.8 (CI 1.32-10.96, p = 0.01)

LGI + IVC CI >50% after PLR had more early AEs with OR 3.6 (CI 1.46-9, p = 0.004)

The other parameters including 10% change in VTI, IVC CI before PLR were not significantly different between groups with and without AEs.

Secondary Outcomes– Adding POCUS to Risk Scores

Since these rules are designed mostly to be sensitive and try to identify a population safe to go home, the decrease in false negatives is most important to look at.

Rockall Score

  • Adding LV kissing walls to Rockall score increased performance in predicting late AEs
  • Increase sensitivity 9.6%
  • Decreased false-negatives from 17.1% to 13.1%
  • Increased true positives from 73.7% to 78.7%
  • AUC improved from 77.6% to 80.3%


  • Adding LV kissing walls to GBS increased sensitivity but decreased specificity
  • Increase sensitivity 9.3%
  • Decreased false negatives from 21.1% to 17.1%
  • Increased true positives 72.4% to 74.7%
  • AUC improved from 72.5% to 73.2%


  • Adding IVC >50% after PLR, IVC <1cm, and LV kissing walls increased performance for both early and late AEs
  • For early AEs, Increase sensitivity 4.9%
  • For Early AEs – AUC increased from 55.7% to 72.2%
  • For Late AEs – Increased sensitivity 29.8%
  • For Late AEs – AUC increased from 57.5% to 66.9%


Convenience sample, single center, single well-trained operator

Excluding critically ill patients may blunt the improvement seen by ultrasound.

Composite outcome of adverse events. Whenever outcomes are combined, you have to be careful. Not all of these outcomes are of equivalent importance. I would care much more if a patient required an additional surgical or endoscopic procedure as opposed to seeing a small increase in their creatinine. For some of these “adverse events”, hospitalization will not prevent or help them in any way. In this study, the majority of the early events were hemoglobin and creatinine changes (instead of bleeding or hypotension which are arguable more serious). The late events were mostly requiring some intervention or coming back to the ED, much less likely being admitted to the ICU or death. Furthermore, some of these outcomes could be unrelated to a worsening GI bleed – hypotension, hypoxia, lab changes are all non-specific occurrences that could confound this outcome. On the other hand, since the authors were trying to be as sensitive as possible, grouping all these together could theoretically make sure nothing remotely bad was missed.

Do the findings make sense? Why would some signs be predictive of early AEs and some signs be predictive of late? Why would this change between upper and lower bleeding? There are a lot of questions here, and I don’t think the findings are entirely explainable based on the hypothesis of the authors. There is always the chance that there are other confounders making certain findings statistically significant.


Does using POCUS to risk stratify GI bleeds make sense? The idea here is that you are attempting to find the patient who looks good, but may be at risk for adverse events. What the ultrasound adds is an understanding of the patient’s likelihood of being hypovolemic. The problem is that you are still only getting a snapshot in time. They may be hemorrhaging briskly, but are not hypovolemic when you scan them. An ultrasound alone does not obtain all the information you need to determine their risk for bad outcomes.

There are actually many other Lower GI bleeding scores, although it does not seem any one is that good. It is not recommended to rely on these scores for the most part.

Take Home Points

1. In patients with GI bleeding, POCUS evidence of hypovolemia is associated with varied composite outcomes of adverse events.

2. Adding POCUS to GI bleeding risk scores may be helpful, but requires more research for validation and patient centered outcomes.

Our score

2 Probes

Expert Reviewer for this Post


Meghan Herbst, MD @EUSmkh

Meghan directs the clinical curriculum for UConn School of Medicine and the ultrasound curriculum for the UConn Emergency Medicine Residency

Reviewer’s Comments

I don’t think the high mortality of GI bleeds means necessarily that it is useful to have effective decision rules and predictive models. I think we need to be vigilant and thoughtful about the etiology of the bleed and the patient risk. I have never used the Rockall score, Glasgow-Blatchford score or Velayos scores. It would be helpful to mention in the introduction (if true) how often patients with GI bleeds are poorly risk stratified clinically and therefore the need for PoCUS in this population. Based on this study, I don’t agree with the authors’ statement, “it can reasonably be assumed that with a more accurate predictive ability of risk stratification, this could help better allocate the limited resources in high-risk patients, safely discharge low-risk patients, and then lead to a better clinical outcome in patients with GI bleeding”, nor do I agree with the final sentence based on the study. The marginal improvement in risk stratification scores unfortunately were not very clinically meaningful.


Cite this post as

Michael Prats. POCUS in GI Bleed Risk. Ultrasound G.E.L. Podcast Blog. Published on March 30, 2020. Accessed on February 26, 2021. Available at https://www.ultrasoundgel.org/89.

Leave a Reply

Your email address will not be published. Required fields are marked *