Thiamine Before Glucose: How Strictly Does the Sequence Matter?
A pregnant patient with severe hyperemesis arrives critically hypoglycemic and already showing an early sign of thiamine deficiency. The debate isn't whether she needs thiamine — it's whether a brief delay to give it first actually protects her, or just feels like it does.
Carol R., a 28-year-old choir teacher, was fifteen minutes into a substitute's shift covering her afternoon class when her downstairs neighbor found her slumped over the kitchen table, unable to answer more than her own name. She is fourteen weeks into her first pregnancy, and for the last three weeks — after two prior emergency-department visits for intractable vomiting, each ending in a bag of IV fluids and a discharge before anyone was confident the nausea was actually controlled — she has kept down perhaps a third of what she used to eat in a day. She has lost twelve pounds since conception, roughly nine percent of her prepregnancy weight, and stopped trying to force solid food a week ago in favor of ginger candy and sips of an electrolyte drink she often loses within the hour.
Her point-of-care glucose in triage reads 31 mg/dL, low enough on its own to account for her flattened responsiveness — cerebral glucose delivery falls off sharply below roughly 50 mg/dL, and neuronal injury from hypoglycemia this severe accrues within minutes, not hours. That number alone would ordinarily settle the sequencing question: treat the hypoglycemia first, worry about anything else second. What complicates it is a finding the resident nearly wrote off as incidental — a fine, sustained nystagmus on lateral gaze, present bilaterally, that her partner says wasn't there two days ago. Nystagmus is one of the four Caine criteria for Wernicke encephalopathy, alongside dietary deficiency, cerebellar dysfunction, and altered mental state; clinical diagnosis requires only two of four, and she already has three. Thiamine is also the cofactor her body needs to actually metabolize the dextrose about to be pushed — pyruvate dehydrogenase and transketolase, both thiamine-dependent, sit directly in the pathway that converts a glucose load into usable energy rather than into a further draw on whatever thiamine reserve three weeks of vomiting has left her. The question dividing the team is not whether she needs thiamine — nobody disputes that — but whether the minute it takes to push it before the dextrose is a real safeguard against tipping her the rest of the way into Wernicke's, or a well-intentioned ritual built on a case-report base thinner than the dogma stacked on top of it.
In triage, before the first line is open
Her glucose is 31. That number, on its own, tells me what to do in the next sixty seconds, and it isn't debate the origin of a case-report literature — it's open the line and push dextrose. Neuroglycopenic injury from hypoglycemia this severe accrues on a timescale of minutes, and every minute we spend deciding whether to draw up thiamine first is a minute her brain is running on nothing. I've looked at the actual basis for "thiamine before glucose, always" more than once, because I give dextrose to at-risk patients constantly. Schabelman and Kuo's 2012 review in the Journal of Emergency Medicine went looking for what that teaching actually rests on. They found nineteen papers, and not one of them rose above the level of a case report or small series — no cohort study, no controlled trial, nothing. More to the point, the human cases they did find describe patients given glucose without thiamine for days on end, not a single emergency-department bolus. The teaching rests on animal models and prolonged-glucose case reports in patients whose deficiency was already severe, not on a demonstrated harm from one push of dextrose.
I'm not arguing she doesn't need thiamine, or that her risk profile is low — she's clearly at real risk. I'm arguing the sequencing itself isn't where the actual danger sits tonight.
I'll take the aside as the real disagreement, then, because on the sequencing question specifically I don't think the absence of case reports carries the weight you're giving it. Three weeks of near-total intake failure, twelve pounds down, and bilateral nystagmus that wasn't there two days ago — that's not a theoretical risk profile, that's two of the four Caine criteria already met before we've given her anything. The EFNS guideline recommends thiamine before or with glucose specifically because the harm being guarded against, if it happens, is largely irreversible, and the safeguard costs a healthy adult essentially nothing to provide. Wernicke's in pregnancy is specifically flagged in the obstetric literature as under-recognized and frequently diagnosed only at autopsy — the absence of case reports may say as much about how rarely anyone looks for it as it does about how rarely it happens.
"A minute spent deciding is a minute her brain runs on nothing" assumes deciding and drawing up thiamine can't happen in parallel with opening the dextrose line — in a functioning bay, they can.
They can, and that's really the whole disagreement dissolved, not adjudicated. Get a second line in — most patients this unwell already have two — and have thiamine physically in hand before the dextrose bag goes up, so the actual delay between them is seconds, not the minutes either of you keeps describing as the stakes. What I'd rather the room actually argue about is the dose someone's about to draw up. Standard hypoglycemia protocols default to a reflexive 100mg IV thiamine, which is what most nursing order sets have preloaded. For a patient with three weeks of established deficiency risk and three of the four Caine criteria already met, the guidance for treating suspected acute Wernicke's isn't a bigger single vial, it's a different schedule: the Royal College of Physicians regimen is 500mg IV three times daily for two to three days before stepping down, and the EFNS guideline's is 200mg IV three times daily until improvement plateaus. Both are six to fifteen times the daily thiamine in that order set, and both give it in divided doses rather than once. If this room settles the sequencing question and then someone pushes the wrong dose because it's what the order set defaults to, we'll have won an argument that didn't matter and lost the one that did.
Agreed within minutes of the glucose result coming back: a second IV line, thiamine 500mg pushed essentially alongside the dextrose bolus rather than meaningfully before or after it, then continued at 500mg three times daily for the next two to three days while she's monitored, with a step-down to once-daily dosing once she is clearly improving. The reflexive 100mg single dose built into the nursing hypoglycemia order set was overridden explicitly, in writing, so it isn't pulled from muscle memory on her next low reading.
Not agreed: whether an MRI to look for the mammillary-body and periaqueductal changes classically described in Wernicke's is worth ordering tonight. The neurologist wants it — a positive scan would settle, for the record, whether she already has structural injury rather than only an isolated exam finding. The emergency medicine physician and pharmacologist both made the same practical point from different angles: nothing about tonight's treatment plan changes based on what the scan shows, since she's already getting full-dose thiamine regardless. The scan was deferred to the inpatient team rather than ordered from the emergency department.