Researchers who searched a decade of national Veterans Affairs emergency department records found no cases of Wernicke's encephalopathy among alcohol-intoxicated patients who received intravenous dextrose before thiamine. The retrospective, cross-sectional study by Jamie Jasti, Katherine Sherman, David Gummin and Andrew Farkas, published online in Academic Emergency Medicine on 28 August 2025 and in print that November, covered visits to all VA emergency departments between 2010 and 2019. Of 120 encounters by 114 individual patients meeting the criteria (an alcohol intoxication code or a serum ethanol above 50 mg/dL, plus dextrose-containing fluids, with anyone given thiamine first excluded), none carried a diagnosis of Wernicke's encephalopathy by billing code or by manual chart review in the emergency department, during hospitalisation, or at follow-up within 90 days.

The finding bears on a rule taught for generations: that thiamine must go in before glucose, on the reasoning that a sugar load in a thiamine-depleted brain can tip a patient into Wernicke's encephalopathy, the acute confusional and eye-movement syndrome that can leave permanent amnesia as Korsakoff's syndrome. The patients in this sample were the ones the rule was written for. Median age was 59, 104 (91 per cent) were male, and among those with a recorded AUDIT-C screen 94 per cent screened positive for alcohol use disorder, while 90 per cent had a documented history of alcohol abuse. The authors conclude that treatment of hypoglycaemia should not be delayed, while describing thiamine in intoxicated patients as low risk and potentially beneficial.

The evidence the rule rested on was never strong. A 2012 literature review in the Journal of Emergency Medicine by Esteban Schabelman and Dick Kuo found the case for it built on case reports, concluded that delaying glucose in a hypoglycaemic patient could not be recommended, and advised thiamine promptly after or alongside restoration of normal blood sugar. The new study is the first national dataset to put a number on the question, and the number is zero.

It is a small zero. Only 120 encounters across ten years of a national health system met the criteria, which sets a wide margin of uncertainty around a rare outcome, and the design was a chart review rather than a trial. Because the outcome depended on a clinician recording the diagnosis or coding it, any case that went unrecognised would not appear in the count. The cohort was overwhelmingly male veterans, and the follow-up window closed at 90 days.