Proton pump inhibitors did not improve outcomes in ICUs (SUP-ICU trial)
Proton pump inhibitors (PPIs) are widely prescribed to mechanically ventilated patients to prevent gastrointestinal (GI) bleeding. The evidence for their use is not strong; individual randomized trials have not shown a clear benefit over histamine blockers, but when many trials were pooled into meta-analyses, PPIs do appear to be superior. The Surviving Sepsis Guidelines have advised preferential use of PPIs over other agents in mechanically ventilated patients with sepsis.
Concerning risk signals associated with PPIs emerged in the mid-2010s, such as increased observed rates of Clostridium difficile and ventilator-associated pneumonia among patients treated with PPIs. Many patients started on PPIs in-hospital continue them after discharge; long-term PPI use has been associated with risk for chronic kidney disease, dementia, and vitamin deficiencies. (All these risks are observed associations and none have ever been proven to be due to PPIs.)
The expert whose work pushed clinical practice toward wide use of PPIs published a 2018 review concluding that the drugs' benefits are probably smaller and their risks greater than originally believed. She cautioned that it's not possible today to confidently identify which patients will benefit from PPIs as prophylaxis against bleeding ulcers.
A new randomized trial (the largest yet), published in the New England Journal of Medicine, did not show obvious net benefit from proton pump inhibitors as GI prophylaxis in the ICU. Fortunately, nor were any short-term risks identified.
Authors randomized 3,298 ICU patients in five European countries and the U.K. to receive pantoprazole or placebo during their ICU stay. Enrolled patients were all at high risk for bleeding, with shock, anticoagulant use, renal-replacement therapy, mechanical ventilation, liver disease, or coagulopathy. Many patients had multiple risk factors and were quite ill as a group, with organ failure present in most patients in both arms (median SOFA scores = 9).
At 90 days, there were no differences in mortality or in the composite outcome of "clinically important events" (including GI bleeding, C. difficile infection, or myocardial ischemia) -- about 22% in each group.
Patients receiving pantoprazole did have numerically lower rates of clinically important GI bleeding (2.5% vs. 4.2%), but the trial was not designed to say whether this was statistically conclusive. Drilling down:
Forty-seven more placebo patients required red cell transfusion compared to pantoprazole-treated patients (535 vs. 488).
10 more patients receiving placebo were transfused ≥2 units of packed red cells, compared to those receiving pantoprazole (39 vs. 29).
13 more placebo patients required new or increased vasopressors (35 vs. 22).
12 more placebo patients required endoscopy (28 vs. 12), and two more each required surgery (5 vs. 3) or embolization (4 vs. 2).
The percentage of patients contracting C. difficile or pneumonia was virtually identical in both groups.
Just for fun, if we pretended these event rates were "truth" and could be replicated and extrapolated to large populations, the numbers needed to treat with PPIs for GI bleed prevention in the highest risk patients would be:
Any red cell transfusion: 35
≥2 units of packed red cells: 164
Vasopressor use/dosage increase, or avoidance of endoscopy: ~126
What It Means
It would be inappropriate to conclude there are no benefits from PPIs as gastrointestinal prophylaxis based on this trial. It suggests just the opposite: pantoprazole likely did prevent some significant GI bleeding, but the composite outcome blurred all the other relevant events together, limiting conclusions about each individually.
That said, such a large study in high-risk patients suggests PPIs aren't very helpful as prophylaxis. It also (inferentially) weakens the supposed advantages of PPIs over histamine antagonists. If PPIs have such limited benefits over placebo in the highest-risk patients, their relative benefits over H2Bs are likely even smaller, and their benefit in less-ill ICU patients minimal or absent.
A meta-analysis of 96 eligible studies enrolling 7,293 patients concluded that PPIs are superior to H2Bs for prevention of GI bleeding in high-risk patients, and even more superior to placebo, but "probably" increase the risk for ventilator-associated pneumonia. Increased VAP rates weren't seen here, leaving this an unresolved question, as well.
Editorialists concluded,
"Prophylaxis with a PPI, if initiated, should be reserved for seriously ill patients who are at high risk for this complication. ... the [reduced] rate of important upper gastrointestinal bleeding may still support the recommendation of using a prophylactic PPI, particularly given the absence of [excess adverse events in the] pantoprazole group."
Source: NEJM