Critical Care Mini Fellowship: GI

The Critical Care Mini fellowship crew had an excellent discussion (and dinner) on October 21st led by Dr. Wendy Chan (moi!) and facilitated by Dr. Ashika Jain.  We share our review of the literature with you below.  Thank you for all who participated.  I have included links to the orginal articles embedded within the title of the article.  Also, we welcome senior residents and PGYIIs to join our monthly meeting.  Contact Dr. Bains or Dr. Jain for more information.


Transfusion Strategies for Acute Upper Gastrointestinal Bleeding

Background: In critically ill patients, the goal of transfusion is to prevent tissue hypoxia however there is some controversy regarding the safest and most effective target hemoglobin. In 1999, the TRICC trial demonstrated improved survival in critically ill patients with restrictive strategy, however this study excluded GI Bleeding.

Design: This was a large single center RCT enrolling 921 patients with acute severe upper GI bleeding. They excluded patients with massive exsanguinating bleeds, vasculopaths, recent transfusion, trauma, or surgery, those with low risk of further bleeding, and LGIB. The average hemoglobin was 9.5 on admission. The most common etiology of the bleeding was a bleeding peptic ulcer. 31% of patients had liver cirrhosis. Patients were randomized either to a restrictive strategy (Transfuse if Hgb ≤ 7 with a post-transfusion target of 7-9 g/dl) or a liberal strategy (Transfuse if Hgb ≤ 9 with a post-transfusion target of 9-11. In both groups if transfusion threshold was met, 1 unit of pRBCs was transfused, hemoglobin was reassessed, and an additional unit was transfused if the patient was still below the threshold.   This protocol continued until the patient was discharged. The patient was also transfused at any time there was symptomatic anemia, massive bleeding, or surgical intervention. All patients underwent emergent gastroscopy within the first 6 hours.

Primary Outcome: all-cause mortality at 45 days. Mortality was lower in the restrictive group compared to the liberal group 5% vs 9%. Subgroup analysis also showed in patients with cirrhosis, risk of death was also lower in the restrictive group if Child-Pugh A or B, equivocal in group C. This was likely because the prognostication for a Child-Pugh C score is so poor overall (<50%), the impact of transfusion strategy was likely negligible.

More people in the restrictive group (51%) vs the liberal group (14%) did not receive a transfusion and the restrictive group was transfused a small quantity (1.5 units vs 3.7 units). This is hardly surprising because from a purely numbers perspective, a higher overall target hemoglobin is going to mean more patients need to be transfused a greater number of units.

Secondary Outcomes: rate of further bleeding, rate of in-hospital complications. Length of stay was significantly shorter in the restrictive transfusion strategy group 9.6 days vs 11.5 days. The rate of further bleeding was significantly lower in the restrictive strategy group (10% vs 16%). Overall complications were significantly lower in the restrictive group (40 vs 48%), with significantly fewer transfusion reactions (3% vs 9%) and cardiac complications (11% vs 16%). Other complications such as TIA, AKI, bacterial infections, and pulmonary complications were not statistically different between groups.

Limitations: This excluded low risk and high risk GI bleeding, not blinded, 6 hour gastroscopy/intervention unattainable at most institutions, protocol did include a balanced transfusion of RBC + platelets + FFP, so those receiving more units of RBC may have developed a coagulopathy, a greater percentage of patients in restrictive group received transfusion for symptomatic anemia or massive bleed, which likely skewed results.

BOTTOM LINE: A restrictive transfusion strategy significantly reduced mortality, risk of further bleeding, length of stay, and overall complications in patients with acute upper GI bleed.

Moving on to paper two, which was discussed over homemade roasted figs with a balsamic reduction poured over vanilla ice cream.

Proton pump inhibitors versus histamine 2 receptor antagonists for stress ulcer prophylaxis in critically ill patients: a systematic review and meta-analysis.

Background: A stress ulcer with clinically important GI bleeding occurs in 1-4% of critically ill patients who are not on GI prophylaxis. Mortality from ICU patients with bleeding complications approaches 50%, which is why intensivists began administrating drugs for stress ulcer prophylaxis. The general principle is that maintaining a gastric pH above 3.5-5.0 prevents gastric injury. Both histamine 2 receptor agonists and proton pump inhibitors have been shown in multiple RCTs to reduce the risk of bleeds.  At Kings County, the majority of those practicing reach for a PPI as first-line prophylaxis, however nationally and world-wide, at the time of this study (March 2013), the majority of our colleagues (67%) initiated prophylaxis with an H2 blocker. PPIs have a stronger effect on raising the gastric pH, however whether this translated into improved patient outcomes as compared to H2 blockers was a big unknown.

Design: Systematic review and meta-analysis comparing PPIs to H2 blockers for the prevention of UGI bleed in 1,720 critically ill patients in fourteen different trials.  Sub-group analysis found no difference when comparing route of drug administration, frequency of dosing, ICU type, or Asian vs. non-Asian trial setting.

Primary Outcome: PPIs were associated with lower risk of overt bleeding compared to H2 blockers. (RR 0.35; 95% CI 0.21-0.59) The number needed to prophylaxis was estimated to be 30.

Secondary Outcomes: There was no significant difference in incidence of nosocomial pneumonia, all-cause ICU mortality, or ICU length of stay between groups.  There was no data on C.diff infections

Limitations: Potential for publication bias and small number of events.

BOTTOM LINE: PPIs are more effective than H2 blockers in reducing clinically important GI bleeds however do not appear to affect mortality, ICU length of stay, or risk of nosocomial pneumonia.  It is also important to consider both H2 blockers and PPIs have important drug interactions and given no difference is seen in mortality, one should select a medication with fewest potential for medication interactions and adverse reactions in the individual patient.

The following two tabs change content below.


Senior EM Resident at SUNY Downstate / Kings County Hospital, EM/Critical Care Blogger, Medical Student Education Curriculum Co-Chair, has a blackbelt in "keepin' it real"

Leave a Reply

Your email address will not be published.