Stress Ulcer Prophylaxis by Default: Reconsidering a Standing Order in a Low-Risk Patient
A single patient, three days ventilated, on a standing PPI order nobody actively decided to write. The disagreement isn't whether stress ulcer bleeding is worth preventing — it's whether ventilation alone still justifies prophylaxis in a patient who doesn't otherwise look at risk.
N.P. manages a regional marketing team and was in the middle of a product launch when a bad asthma flare, the worst she says she's had since college, sent her to the emergency department three days ago struggling to speak in full sentences. She has had asthma since childhood, generally well-controlled on a daily inhaled corticosteroid-LABA combination, with this flare apparently triggered by a viral upper respiratory infection the week before rather than anything she can point to as a medication lapse. She was intubated within hours for status asthmaticus and has improved steadily since — wheezing resolved, sedation lightening, extubation likely within the next day. She has no history of GI bleeding, no coagulopathy, has never required vasopressors during this admission, and her only steroid exposure has been a standard methylprednisolone burst for the asthma itself, not the kind of prolonged high-dose course that independently raises bleeding risk.
She has been on intravenous pantoprazole since her first ICU day — not because anyone specifically decided she needed it, but because it's part of the unit's standard ventilated-patient admission order-set, built around the classic 1994 finding that mechanical ventilation beyond 48 hours alone raises stress ulcer risk enough to warrant prophylaxis. SUP-ICU (Krag, 2018), the largest and most rigorous trial to test that practice directly in a real, modern ICU population, found no mortality difference with pantoprazole — but its bleeding result is more often flattened than it deserves. Clinically important GI bleeding fell from 4.2% to 2.5%, a relative risk of 0.58 with a confidence interval of 0.40 to 0.86 that does not cross one. It was a secondary outcome reported without adjustment for multiple comparisons, which is a real caveat and a different one from calling the effect absent. She has also tolerated enteral tube feeds without difficulty since ICU day 2, another marker generally associated with lower stress-ulcer risk in the observational literature, though not one of the two classic Cook criteria themselves. So the honest position is narrower than either slogan on offer. The drug does prevent bleeding; what SUP-ICU could not show is that preventing it changes anything that matters at the population level, and what it explicitly did not find was any subgroup in which the benefit concentrated. Deciding about her means reasoning from Cook's original risk factors, which she almost entirely lacks, rather than from a trial result that never separated patients like her out.
A standing order nobody re-examined
I'd continue the pantoprazole. She meets the classic 48-hour ventilation threshold from the original Cook criteria, regardless of her other risk factors, and SUP-ICU's own numbers showed fewer clinically important GI bleeds on pantoprazole, 2.5% versus 4.2% — a relative risk of 0.58, confidence interval 0.40 to 0.86. The trial was null on its primary mortality endpoint, but that bleeding reduction is a real effect, not a trend. Stopping prophylaxis in a ventilated patient is a step back from established practice.
SUP-ICU is the largest, most rigorous trial specifically built to test this practice, in a real modern ICU population, and it found no mortality difference. Her actual risk profile — hemodynamically stable throughout, no coagulopathy, never on vasopressors, only a brief course of ventilation, no history of GI pathology — sits well outside the true high-risk phenotype the original 1994 criteria were built around. Continuing exposes her to the drug's own infection risks, C. difficile and hospital-acquired pneumonia, for a benefit that in someone at her risk level is probably close to zero.
I'll grant you it's an effect and not a trend — I'd only add that it was a secondary outcome, unadjusted for multiple comparisons, and that it's an average across SUP-ICU's whole enrolled population, which included patients in shock and with coagulopathy she doesn't have. An average benefit doesn't mean the benefit is evenly spread across every risk profile inside that average.
I don't think ventilation status alone, on its own, is still the right trigger — but I want to correct something before it hardens into a rule. SUP-ICU did not find that the bleeding risk concentrates in the coagulopathic or shocked patients; it found no significant subgroup interactions at all, including for shock, coagulopathy and mechanical ventilation. The risk-factor list we actually use comes from Cook's 1994 cohort, not from SUP-ICU, and it is the weaker kind of evidence. She has essentially none of those factors. That argues for stopping prophylaxis in her specific case without abandoning the practice as a blanket rule for a patient who actually does carry those other risk factors — this is an individualized-risk call, not a referendum on stress ulcer prophylaxis as a whole.
Agreed: discontinue pantoprazole given her low composite bleeding-risk profile, with an explicit plan to restart if any new risk factor emerges — coagulopathy, shock, renal replacement therapy, or ventilation extending well beyond the next day or two.
Not agreed:
Whether this same reasoning should change the unit's standing order-set default for every ventilated patient, not just hers. The critical care physician wants to keep ventilation-status-alone as the default trigger for the rest of the unit's patients; the clinical pharmacologist wants the order-set itself changed to require a documented individualized risk factor rather than ventilator status alone, a broader practice change neither side treated as settled by tonight's single case.