Thiamine IV Therapy: Mechanism, Evidence and the Dose That Came From Nowhere

In short: thiamine is the one water-soluble vitamin in this category with an unarguable intravenous emergency, and the figure most infusion menus use for it has no provenance. The licensed labels are specific and narrow. The FDA-approved label for Thiamine Hydrochloride Injection, revised September 2025, gives 100 mg intravenously then 50–100 mg intramuscularly daily for Wernicke-Korsakoff syndrome; the UK licensed combination delivers 250 mg of thiamine per ampoule and its July 2025 label specifies two to three pairs every eight hours for suspected Wernicke's encephalopathy — 500 to 750 mg per administration. The 100 mg of the traditional banana bag was never chosen from a trial. A 2016 review in Critical Care Medicine concluded exactly that and recommended 200–500 mg every eight hours instead. The sepsis hypothesis has been tested and failed — VICTAS reported 25 ventilator- and vasopressor-free days against 26 for placebo in 501 patients. And a new deficiency population has appeared in the safety literature: 15 reported cases of Wernicke encephalopathy during GLP-1 receptor agonist treatment, 13 of 15 with gastrointestinal symptoms. The right question for a clinic is not what to add to a drip, but whether the person in front of it is depleted.

The one B vitamin where the route argument is real

Most intravenous vitamin claims rest on a bioavailability argument that does not survive inspection. Thiamine is the exception, and understanding why also shows where the exception stops.

Human thiamine stores are small — on the order of 25 to 30 mg in an adult, turning over fast enough that clinical deficiency can develop within weeks of inadequate intake rather than months or years. That is the first half of the route argument. The second half is intestinal: at low luminal concentrations absorption runs through an active, carrier-mediated and therefore saturable transporter, with greatest activity in the duodenum. Published estimates put the ceiling of that mechanism at roughly 5 to 8 mg from a single oral dose.

Stop there, as many supplier sheets do, and the conclusion looks obvious: oral thiamine is capped at a few milligrams, so a patient needing hundreds must be given it intravenously. The literature does not support the conclusion in that form. Above about 2.5 mg, passive diffusion contributes, and a pharmacokinetic study of high-dose oral thiamine hydrochloride in healthy subjects found that absorption was not saturable up to 1,500 mg, with higher oral doses producing higher overall exposure. The clinical review published in the Journal of Nutrition and Metabolism in February 2026 reads both findings together and concludes that both routes are clinically acceptable, with the choice driven by gastrointestinal access and urgency rather than by an absolute absorption ceiling.

So the honest version is narrower and more useful. The intravenous route earns its place when the gut is unreliable, when the patient is vomiting, when absorption is compromised, or when the clinical picture will not wait for an oral dose to be taken, absorbed and distributed. Those are real situations. "Oral cannot work" is not the reason — "this patient cannot use oral, and the time course matters" is.

Three enzymes, and what stops when the cofactor goes

Thiamine itself does almost nothing. Phosphorylated to thiamine diphosphate, it becomes the cofactor for a small set of enzymes that sit at the centre of carbohydrate metabolism, and the deficiency syndrome is the sum of their failure.

Pyruvate dehydrogenase converts pyruvate to acetyl-CoA, the step that admits glucose-derived carbon to the citric acid cycle. Alpha-ketoglutarate dehydrogenase carries the cycle past one of its own control points. Both reactions reduce NAD⁺ to NADH, which is what the respiratory chain and coenzyme Q10 downstream of it actually consume. Without thiamine diphosphate, pyruvate accumulates, is shunted to lactate, and aerobic ATP production falls in the tissues least able to tolerate it — brain and myocardium first.

The third enzyme explains a quieter consequence. Transketolase moves two-carbon units in the non-oxidative branch of the pentose phosphate pathway, the system that supplies NADPH and ribose-5-phosphate. NADPH is what regenerates reduced glutathione. A thiamine-depleted cell is therefore not only energy-limited but less able to maintain its own antioxidant capacity — which is the mechanistic reason transketolase activity was used for decades as a functional index of thiamine status.

Two clinical facts follow directly from this biochemistry, and both matter at the bedside. First, giving glucose to a thiamine-depleted patient increases demand for exactly the cofactor that is missing, which is why the EFNS guideline instructs that thiamine be given before any carbohydrate, and why the FDA label lists administration of intravenous dextrose to a person with marginal thiamine status as one of its indications. Second, magnesium is a cofactor for the enzyme that phosphorylates thiamine to its active form, so hypomagnesaemia can blunt the response to a correctly sized thiamine dose. Neither point is a dosing instruction; both are reasons the sequence is not arbitrary.

What the licensed labels actually say

Thiamine is one of the few actives in this field where a buyer can read a licensed indication rather than infer one, and the two major labels are worth reading side by side because they do not match.

The FDA-approved label for Thiamine Hydrochloride Injection USP, 100 mg/mL, in the version revised September 2025, indicates the product for thiamine deficiency or beriberi in its dry and wet forms, and for use where rapid restoration of thiamine is necessary — Wernicke's encephalopathy, infantile beriberi with acute collapse, cardiovascular disease due to thiamine deficiency, and neuritis of pregnancy where vomiting is severe. Its stated doses are modest: 10 to 20 mg intramuscularly three times daily for up to two weeks in beriberi, followed by an oral multivitamin; 100 mg intravenously as an initial dose in Wernicke-Korsakoff syndrome, then 50 to 100 mg intramuscularly daily; 100 mg in the first few litres of intravenous fluid where dextrose is being given to someone with marginal thiamine status.

The UK licensed product is a different object. Vitamin B&C Intravenous High Potency — long known by its brand name — is supplied as paired 5 mL ampoules. Ampoule 1 contains thiamine hydrochloride 250 mg with riboflavin 4 mg and pyridoxine hydrochloride 50 mg; ampoule 2 contains ascorbic acid 500 mg, nicotinamide 160 mg and glucose 1,000 mg. The Summary of Product Characteristics revised in July 2025 directs that for Wernicke's encephalopathy two to three pairs be diluted in 50 to 100 mL of physiological saline or 5 per cent glucose and infused over 30 minutes every eight hours.

Run the arithmetic and the divergence is stark. The UK regimen delivers 500 to 750 mg of thiamine per administration and 1,500 to 2,250 mg per day. The US label's figure for the same syndrome is 100 mg once, then 50 to 100 mg daily. Both documents are current, both are regulator-approved, and they differ by more than an order of magnitude for the same indication. That is not a reason to disregard either. It is a reason to notice that the number on a label is a regulatory artefact of a specific product and dossier, not a physiological constant — and to be sceptical of any supplier who quotes "the" licensed thiamine dose without naming which one.

The dose that came from nowhere

Between those labels sits the most-copied thiamine dose in practice, and it has no evidentiary parent at all.

The traditional banana bag — 100 mg of thiamine, 1 mg of folic acid and magnesium in a litre of dextrose, named for the colour riboflavin gives it — became a fixed order set in emergency departments and then migrated, more or less intact, into commercial infusion menus. The review published in Critical Care Medicine in August 2016 under the title Unpeeling the Evidence for the Banana Bag examined its components one at a time. Its finding on thiamine was blunt: the 100 mg figure was not chosen on the basis of clinical trial evidence, and on pharmacokinetic grounds it is unlikely to optimise delivery of thiamine to the central nervous system in patients with alcohol use disorder. The authors recommended 200 to 500 mg intravenously every eight hours in its place.

Guidelines independently converge somewhere above the inherited figure. The EFNS guideline of 2010 recommends 200 mg three times daily, preferably intravenous, diluted in 100 mL and given over 30 minutes, and explicitly before any carbohydrate. The 2026 clinical review reports general agreement across guidelines on at least 500 mg of parenteral thiamine per day for three to five days in treatment of Wernicke's encephalopathy, with 200 to 300 mg daily for prophylaxis in at-risk patients, and notes that symptoms typically begin to resolve within 24 to 48 hours when the dose is adequate.

The commercial reading of this is uncomfortable in both directions. A wellness drip carrying 100 mg of thiamine is not delivering a guideline treatment dose for the one condition where thiamine is genuinely urgent; it is delivering a figure that a 2016 review criticised precisely because it had no derivation. Equally, the guideline doses are treatment regimens for a neurological emergency in a depleted patient, and transplanting them into an elective menu for people who eat normally substitutes one unexamined number for another. What a clinic can defensibly do is state which figure its protocol uses and where that figure came from. A protocol sheet that cannot answer that question is carrying a number by inheritance.

What the randomised evidence found

Sorted by claim, with the trial that tested it and the result as published.

Read down the table and the shape is consistent with the rest of this series. Where thiamine corrects a measured or strongly suspected deficit, it works and the clinical literature treats the matter as settled without a placebo arm. Where it was promoted as a metabolic intervention in people who are not deficient, the properly powered test was run and returned nothing.

VICTAS is the cleanest version of that second result. Published in JAMA in February 2021, it randomised 501 patients with sepsis and respiratory or cardiovascular dysfunction to intravenous vitamin C, thiamine and hydrocortisone or to placebo for up to 96 hours. The median was 25 ventilator- and vasopressor-free days in the treatment group against 26 in placebo, with no mortality difference. The metabolic-resuscitation cocktail that dominated critical care discussion for four years did not survive its own trial — the same arc the high-dose vitamin C literature followed, and worth remembering whenever a combination is marketed before it is tested.

A new deficiency population

The clinically interesting development in thiamine is not a new indication. It is a new way for patients to become depleted, and it sits directly in the path of the aesthetic and longevity sector.

A pharmacovigilance study and literature review published in Clinical Nutrition identified 15 cases of Wernicke encephalopathy associated with GLP-1 receptor agonist treatment: 13 from the FDA Adverse Event Reporting System, one from the published literature and one from the authors' own centre. Most involved semaglutide, 8 of 15, or tirzepatide, 6 of 15, and 14 of the 15 were reported during 2023 and 2024. Thirteen of the 15 patients reported gastrointestinal manifestations. Disproportionality analysis gave a reporting odds ratio of 2.35, with a confidence interval of 1.38 to 4.01. Long-term neurological sequelae were recorded in 7 of the 11 patients for whom follow-up data existed.

The caveats belong in the same paragraph as the finding. Spontaneous reporting databases establish a signal, not causation; reporting is influenced by publicity; and 15 cases set against tens of millions of prescriptions describe something rare. The authors frame it as a potentially rare but severe adverse event warranting increased clinical awareness, particularly in patients with severe gastrointestinal symptoms — which is a different statement from a claim of drug toxicity.

For a clinic the practical content is still substantial, because the mechanism requires no novel pharmacology. Thiamine stores are small, prolonged vomiting or markedly reduced intake depletes them within weeks, and rapid weight loss raises metabolic demand while intake falls. A service that prescribes weight-loss pharmacotherapy is, by construction, managing a cohort in which reduced intake is the intended effect. The response to that is clinical attention to intake, symptoms and neurological signs — not an infusion added to a package as insurance. A drip given to a patient who is eating adequately treats nothing; a missed Wernicke's in a patient who is not eating is a preventable neurological injury, and the evidence above records that it often does not fully reverse.

Anaphylaxis: the label and the data disagree

Thiamine carries one of the clearest documented divergences between a current regulatory warning and current prospective evidence, and buyers should know both sides of it rather than one.

The FDA-approved label warns that serious hypersensitivity and anaphylactic reactions can occur, particularly after repeated administration, states that deaths have resulted from intravenous or intramuscular use, and advises a skin test where sensitivity is suspected — one-hundredth of the dose intradermally with 30 minutes of observation. That warning language is long-standing and remains in the September 2025 revision.

The largest prospective safety series points the other way. Published in Annals of Emergency Medicine in 1989, it followed 989 consecutive patients given 100 mg of thiamine hydrochloride as an intravenous bolus, across 1,070 doses. Twelve adverse reactions were recorded in total, 1.1 per cent: eleven were transient local irritation, and one — 0.093 per cent — was generalised pruritus. There was no anaphylaxis. The authors concluded that thiamine hydrochloride may be given intravenously without undue concern and that intradermal test doses are not warranted in patients without a previous allergic reaction.

The UK Summary of Product Characteristics resolves the tension in the way a clinic can actually operationalise: serious allergic reactions such as anaphylactic shock may occur rarely, warning signs may include sneezing or mild asthma, and facilities for treating anaphylaxis should be available whenever the product is administered. Preparedness rather than pre-testing. Any supplier conversation about an intravenous B-vitamin presentation should establish that the product's own label has been read, not a generic claim that the molecule is well tolerated.

Measuring status without delaying treatment

Thiamine testing is better than it was and still cannot carry the weight that is often placed on it.

The contemporary assay is whole blood or erythrocyte thiamine diphosphate by high-performance liquid chromatography with fluorescence detection. It has largely displaced the erythrocyte transketolase activation coefficient, which measures function indirectly and whose methodological problems were revisited in an analysis published in Annals of the New York Academy of Sciences in 2026. Whole blood is generally preferred over washed erythrocytes for population work, because washing is laborious and introduces its own analytical variability.

The limitation is the one that matters commercially. There is no international consensus on the cut-off that defines deficiency, which means a result sitting low in a reference interval is an invitation to assess the patient, not a diagnosis and not a purchase order. Guideline bodies are unusually direct about the sequence: ESPEN advises that red cell or whole blood thiamine diphosphate may be collected but that treatment should not be based on the result, and ASPEN notes that levels should not be obtained for emergency management, though they may inform subsequent care. Wernicke's encephalopathy is diagnosed clinically and treated before the laboratory answers.

Which leaves a clean operational rule, and one that runs against the commercial grain: a test that cannot be acted on faster than the clinical picture is not a gate on treatment, and a normal-looking number is not a reason to infuse something else instead.

Before the order goes out

Four questions specific to this molecule, each of which a supplier should be able to answer in writing.

One — which salt, and how much thiamine. Thiamine hydrochloride and thiamine mononitrate are not interchangeable by mass, and the derivatives marketed for better absorption — fursultiamine, benfotiamine, sulbutiamine, developed largely in Japan from the 1950s — are different compounds with different pharmacology, not stronger versions of the same one. A specification reading "vitamin B1 250 mg" without naming the salt is incomplete.

Two — what else is in the ampoule, and specifically whether there is sulfite. This is the thiamine-specific stability question and it is frequently missed. Bisulfite and metabisulfite, used as antioxidants in some injectable solutions, cleave thiamine irreversibly into its thiazole and pyrimidine halves. The effect was documented in American Journal of Hospital Pharmacy in 1981 and has been confirmed repeatedly since in parenteral nutrition work; thiamine has been shown to remain stable in complete admixtures in multi-layered bags for at least 28 days provided metabisulfite-containing amino acid solutions are avoided. The published studies do not agree on the precise pH dependence of the reaction — the 1981 work reported stability inversely related to pH, while a 1988 study in JPEN found losses at pH 5.5 but not at higher pH — and that disagreement is a reason to ask for product-specific data rather than to rely on a general rule. Degradation also tracks sulfite concentration, temperature, light and oxygen exposure, which is why this parameter belongs next to the cold chain and storage questions rather than apart from them.

Three — the dilution and the infusion time the presentation was specified for. Both licensed labels say this explicitly: slow intravenous administration in the US label, dilution in 50 to 100 mL and infusion over 30 minutes in the UK one. A concentrate quoted without its intended dilution and rate describes a solution nobody infuses.

Four — who made it. For a parenteral B-vitamin the specification, the stability data and the batch record sit with whoever ran the process. EFBA manufactures the IVIXIR series in its own GMP-certified facilities, and the manufacturer and wholesale page sets out how that supply side is organised. Thiamine appears as a declared component of IVIXIR MyerSence MD, alongside riboflavin, nicotinamide, dexpanthenol, pyridoxine, vitamin B12, magnesium chloride and sodium ascorbate, at stated concentrations — the B-complex group as a declared set rather than a phrase. Where the methylation vitamins are the subject, IVIXIR Metilplus is the separate concept, and the B12 and folate article deals with the route question there, which is not the same question as this one.

None of this argues against thiamine. It argues for knowing which thiamine, at which figure, for which patient. The molecule has something very few actives in this category have — a condition where an infusion changes the outcome within 48 hours, and a body of guidance that says so. That makes the unexamined 100 mg harder to defend, not easier. IVIXIR concepts are defined sets of molecules at stated concentrations for professional use — not medicinal products, and not treatments for any condition — and decisions about indication and patient belong to the practitioner. What can be specified can be assessed, and a dose nobody chose cannot be.