Pantoprazole or Placebo for Stress Ulcer Prophylaxis (POP-UP): Randomized Double-Blind Exploratory Study, by Shane Selvanderan et al .
How many things we do daily in auto-mode? I mean, like the ACLS, of course when we’re running a code we’re thinking what the fu#k happened to the patient, reversible causes and things like that, but the algorithm, that’s in auto-mode. Epi, compressions, hey, you, push harder, check rhythm…. When you tie your shoes while talking to someone, or even when you pretend you’re listening that fellow colleague telling you about the awesome weekend he had on the beach while you were on call. Sometimes auto-mode is good, so you can save your “real brain” for what matters. But sometimes, auto-mode sucks! Like when the nurse comes at you and say: Hey Doc, Mr Mustard has no prescription for gastrointestinal (GI) bleeding prophylaxis. BOOM! Auto-mode kicks in, and there are you, prescribing Mr Mustard a good life-saving dose of omeprazole! It gets worse when your hospital has checklists, with a blank box and the words: GI bleeding prophylaxis. Today, we’re gonna discuss the POP-UP trial (Pantoprazole or Placebo for Stress Ulcer Prophylaxis: Randomized Double-Blind Exploratory Study) , published two months ago in Critical Care Medicine.
But first things first. Ok, clinically important GI bleeding might increase mortality and ICU-LOS. It makes sense, if the guy bleed enough to need transfusions or get hypotensive, his chances of a bad outcome increases. I think we all agree to that. We also know what are the risk factors for clinically important GI bleeding. In 1994, Deborah Cook published a study in NEJM  evaluating potential risk factors for clinically important GI bleeding, and I think we should spend few seconds here. It was far from a perfect study, and unfortunately, some people cannot see its limitations. After gathering all the data, they did a simple regression analysis, the variables significantly associated with GI bleeding (P<0.05) were entered into a multiple regression analysis and tested for interaction. Here it is:
If you understand the difference between the two analysis, good, but if don’t, you’ll probably do what is easier: you’ll go for the amazing simple regression and its low p values. In simple regression analysis, you use a single variable (independent variable) to predict the value of another variable (dependent variable), for instance, you’ll use sepsis (independent variable) to predict bleeding (dependent variable). The use of a simple regression to build a multiple regression model is very common. Multiple regression is used when you want to predict a variable (dependent variable) based on two or more variables (independent variables), it allows you to determine the relative contribution of each variable to the overall model. And when we talk about interaction, it means that the effect of one independent variable might depend on other independent variable. Some guys just looked at the simple regression, that’s why you probably heard that all the factors listed in this the table are risk factors for GI bleeding. But no! You have to look to the multiple regression model, and there, we can see that the only two statistically significant risk factors were: Respiratory failure (mechanical ventilation >48h) and coagulopathy (platelets <50K, INR>1.5 or PTT>2), note that anticoagulant therapy is different from coagulopathy.
Now, we know that clinically important GI bleeding is related to worse outcomes and what are its risk factors(we do have other risk factors, but let’s stick to these ones). The big question is: is there any drug that can prevent clinically important GI bleeding in patients under risk? We have tons of studies evaluating this subject, from the use of sucralfate, ranitidine, and omeprazole, among others. A meta-analysis  showed no mortality benefit, but it showed a decrease in GI bleeding (note that GI bleeding is different from clinically important GI bleeding) with stress ulcer prophylaxis (SUP), however, the quality of evidence was low. So, I said all this bullshit to finally talk about our study. Our Australian friends designed a randomized controlled trial comparing pantoprazole vs placebo for prophylaxis against GI bleeding. This is important because we don’t have any good studies comparing proton pump inhibitors (PPI) against placebo for SUP. Also, the use of PPI might be associated with ventilator-associated pneumonia and Clostridium difficile infection.
How the did it
They designed a single center, prospective, RCT. Inclusion criteria were: patients expected do be mechanically ventilated >24h, and expected to receive enteral nutrition within 48h of admission. Exclusion criteria: admission with GI bleeding, history of peptic ulcer, use of steroids (100mg of prednisolone or more), pregnancy, upper GI or cardiac surgery, and a few more. Since both treatments were considered standard of care at the institution they used a delayed opt-out consent. Intervention was pantoprazole 40mg IV once a day or placebo, until the patient was no longer mechanically ventilated or a maximum of 14 days. Primary outcome was a composite of clinically significant GI bleeding, VAP, and C. diff infection. Since it was an exploratory study, they expected that data collected during 1 year period would be sufficient.
What they found
From all 654 eligible patients, 436 were excluded (most of them were receiving acid suppressive drugs prior the ICU admission), and 216 were randomized (107 to intervention and 109 to placebo). Patients were around 52 y/o and about half of them were receiving vasoactive drugs. The table below shows the processes of care of each group while receiving the study drug:
More than 80% of patients received enteral nutrition, however, both groups had high rates of intolerance to enteral feeding (defined as gastric residual volume >250mL at least once on any day). I think it’s a waste of time measure gastric residual volume, and use a gastric residual volume >250ml as a marker of intolerance to enteral feeding is a joke. It’s interesting that mote than 30% of patients had hemostatic dysfunction, good.
The first interesting finding was that there were no episodes of clinically significant GI bleeding, 1 patient in placebo group and two in pantoprazole group had VAP, and although a total of 70 patients were tested for C. diff infection, only 1 in intervention group had a positive sample. A total of 9 patients had overt bleeding (6 in placebo and 3 in pantoprazole, p = 0.50).
Well, this was an exploratory study, therefore, don’t expect to change your practice based on its results. But I do expect that these results make you think about how we’ve been doing things. It seems that clinically significant GI bleeding in critically ill patients is an uncommon phenomenon. The use of pantoprazole did not increase the rate of adverse events (VAP and C. diff infection), but do not use this as an excuse next time you prescribe PPI to your patients. I got bit scared when I saw that 56% of excluded patients were already receiving acid suppressive drugs, and this reinforces my beliefs that the indiscriminate use of acid suppressive drugs in ICUs is just the tip of the iceberg. Furthermore, we are always suspicious about C. diff every time our patients let out a juicy fart, 32% of patients were investigated for C. diff, with only one positive result. Why don’t we take a deep breath before ordering a C. diff test, and think thoroughly about other possible and more common causes of diarrhea? Ok, it’s time for a huge multicenter RCT comparing placebo vs acid suppressive drugs, also including fasting patients. My message is, although we don’t wanna see our patients bleeding (here I’m referring to clinically important GI bleeding), we’re still not sure if any treatment can prevent it from happening. I don’t think it’s a sin to prescribe acid suppressive drugs to high risk patients, but sometimes, prescribing those drugs become a bad habit, hard to break. “Oh, but is just pantoprazole”. C’mon, be smart!
1. Selvanderan SP, Summers MJ, Finnis ME, et al. Pantoprazole or Placebo for Stress Ulcer Prophylaxis (POP-UP): Randomized Double-Blind Exploratory Study. Crit Care Med. 2016;44(10):1842-1850.
2. Cook DJ, Fuller HD, Guyatt GH, et al. Risk factors for gastrointestinal bleeding in critically ill patients. Canadian Critical Care Trials Group. N Engl J Med. 1994;330(6):377-381.
3. Krag M, Perner A, Wetterslev J, Wise MP, Hylander Møller M. Stress ulcer prophylaxis versus placebo or no prophylaxis in critically ill patients. A systematic review of randomised clinical trials with meta-analysis and trial sequential analysis. Intensive Care Med. 2014;40(1):11-22.