Antibiotics Started for Sepsis, in a Patient Who May Not Have One
Severe pancreatitis meets every criterion a sepsis alert is built to catch — fever, tachycardia, leukocytosis — without there necessarily being any infection behind it, and the multiple negative trials testing exactly this scenario say something the order set doesn't know.
Desmond A., a 52-year-old warehouse manager, was admitted three days ago with severe acute pancreatitis, initially attributed to gallstones, complicated by roughly forty percent necrosis of the pancreatic body on CT obtained on hospital day two. He met SIRS criteria on presentation — fever to 38.9°C, heart rate 118, respiratory rate 24, white count 19,400/µL — and was started on broad-spectrum piperacillin-tazobactam in the emergency department under a sepsis order set triggered automatically by those same vital signs, before the pancreatitis diagnosis had even been fully established.
Three days later, the clinical picture is genuinely mixed rather than clearly resolving in either direction. He remains critically ill, still requiring ICU-level fluid and pain management, and his necrosis on repeat imaging has not changed meaningfully. Before this admission he had no significant past medical history beyond well-controlled hypertension, no prior pancreatitis episodes, and no alcohol use his family could corroborate — gallstones, confirmed on his admission ultrasound, remain the clear and sufficient explanation for why this happened, which matters because it removes any competing infectious source (an ascending cholangitis, for instance) that might otherwise complicate today's antibiotic question. Two sets of blood cultures and one set drawn from a percutaneous aspirate of the necrotic collection, obtained specifically to look for infected necrosis, have all returned negative to date, and repeat CT shows no gas within the necrotic collection — the specific radiographic sign most associated with infection. Three negative sources and an absent radiographic sign are, between them, the entire case for stopping — several randomized trials, including Isenmann et al.'s ciprofloxacin/metronidazole prophylaxis study and Dellinger et al.'s meropenem prophylaxis trial, tested prophylactic antibiotics specifically in severe and necrotizing pancreatitis without documented infection, and found no mortality benefit from continuing them — evidence now reflected directly in IAP/APA joint guidance against routine prophylactic antibiotics in sterile pancreatic necrosis, a guideline written for exactly the population Desmond's own imaging and cultures now place him in.
ICU round, hospital day 4, cultures negative to date
I'd rather keep antibiotics running a few more days than stop them in a patient this sick with forty percent necrosis. If infected necrosis develops and we've already de-escalated, that's a genuinely dangerous miss in exactly the population where it matters most.
I understand the instinct, but I'd point to what the trial evidence in this exact population actually shows: Isenmann's and Dellinger's randomized trials both tested prophylactic antibiotics in severe, even necrotizing pancreatitis without documented infection, and neither found a mortality benefit from continuing them. That's not a general antimicrobial-stewardship preference — it's a direct answer to the specific question in front of us.
I don't think 'he's very sick' and 'he has an infection' are the same finding here — his cultures and the aspirate from the necrotic collection itself are both negative, and continuing antibiotics without evidence behind them isn't free; it carries real C. difficile and resistance-selection risk in a patient who's going to be in this ICU for a while.
I'd frame this even more directly: the trials you're both referencing weren't testing pancreatitis severity in the abstract, they were testing exactly this scenario — severe or necrotizing pancreatitis, with antibiotics started empirically and no documented infection found. His absent gas on imaging and three negative culture sources, including the aspirate specifically obtained to look for infected necrosis, place him squarely in the population that evidence describes, not in some different, higher-risk category severity alone might otherwise suggest.
So: de-escalate today. Not because he isn't sick — he clearly is — but because the specific thing antibiotics would be treating hasn't been found despite genuinely looking for it, and the randomized evidence in this exact population says continuing them doesn't change his outcome.
Agreed: piperacillin-tazobactam discontinued, with explicit criteria set for restarting empiric coverage — any new fever, hemodynamic decline, or gas seen on a future scan — so the surgical resident's real concern has a concrete, monitored answer rather than being resolved by simply overriding it.
Desmond remained afebrile and hemodynamically stable over the following 72 hours off antibiotics, with a repeat CT at day 7 showing stable, still-sterile-appearing necrosis — consistent with the de-escalation decision, though the team noted the surveillance plan would continue for the full expected necrosis-maturation window.