I used to have protocols that insisted that thiamine (Vitamin B1) be given before glucose.
Prolonged thiamine deﬁciency can lead to the development of Korsakoff syndrome, characterized by permanent mental impairment with confabulation and memory deﬁcits.
Maybe that explains how the requirement for thiamine before glucose came about – confabulation. Many of our traditional treatments are better explained by
narrative fallacy confabulation, based on a persuasive hypothesis, than by any evidence of benefit to the patient.
in starvation, when glucose stores are depleted, the brain suffers most acutely as it relies almost entirely on glucose for energy and there is little thiamine available to act as cofactor for the conversion of any small amount of remaining glucose to ATP.
How should that affect patient care?
The prevailing teaching in medical school curricula and in medical textbooks is that if thiamine deﬁciency is suspected, thiamine supplementation should be given before administering glucose (10,11). The theory behind this is that if a thiamine-deﬁcient patient is given a glucose load, meager thiamine stores would rapidly be exhausted, glycolysis further limited, and Wernicke encephalopathy would promptly ensue (9,12–14). 
Reference 10 is the 6th edition (2004) of the Emergency Medicine Manual by Tintinalli. Something that would be accepted as an authoritative reference by many people.
The 19 articles reviewed included expert opinion (n = 2), case reports (n = 13), and animal models (n = 4).
In other words – nothing.
They found an old expired expert opinion, some case reports that should generate some hypotheses for research, and some animal models that also should generate some hypotheses for further research, but nothing that should be the basis of treatment of humans.
The animal studies were rated the best overall, with outstanding quality,
Old McDonald would be thrilled.
Google and Google Scholar search identiﬁed 1 additional non-peer-reviewed report (Gussow, 2007) 
Incidentally, there is evidence that thiamine is taken up and utilized by cells significantly more slowly than is glucose, taking away even a theoretical justification for the thiamine-before-glucose canard.
Dr. Leon Gussow writes The Poison Review and has reviewed this paper. He provides a lot of useful information in his original paper from 5 years ago.
The most often cited source claiming a link between glucose loading and acute onset of Wernicke encephalopathy in thiamine-deﬁcient patients is a four-case series by Watson et al. from 1981 (9). However, as pointed out by Hack and Hoffman in 1998, none of these cases involved the acute administration of glucose (20).
The most often cited source – only four patients – and none of them received the treatment being referenced.
Do any of these people read the paper before citing it?
Do any of these people understand the difference between giving a treatment and not giving a treatment?
Maybe there are valid reasons for citing a paper that is completely irrelevant.
All other cases clearly showed deterioration in mental status after prolonged or massive (>2 L of 5%) glucose infusion, or showed evidence of Wernicke encephalopathy before glucose administration.
>2 L of 5% glucose is more than four 50 ml ampules of D50W (Dextrose 50% in Water). I have never given close to that much to any one patient.
So although the evidence does not support rapid deterioration after glucose administration, it likewise does not seem to support the lack of rapid deterioration.
Because we are in a largely evidence-free zone.
We can make all sorts of claims – as long as we do not expect to use any good evidence to support these claims.
Since we are not using evidence, we should ignore the following suggestion from the authors.
Patients with hypoglycemia should be restored to normoglycemia as quickly as possible (repeated dosing of dextrose 50% in adults until normoglycemia is achieved).
Where is there any evidence that 50% dextrose is any better than an infusion of 10% dextrose?
After demonstrating the lack of evidence supporting one tradition, they blithely encourage adherence to a different tradition. There is no good reason for us to be giving such hyperosmolar fluids peripherally. If we are going to use a peripheral line, we should be using a much more dilute form of dextrose. The total amount of dextrose administered will be significantly less. The risk of extravasation will be less. The cost will be less. The treatment time will not be different.
10% dextrose is a much better idea than 50% dextrose.
Both the median total dose of dextrose administered and post-treatment blood sugar level were significantly higher in the 50% group, and these subjects were more likely to have received the maximum permitted dose of 25 g (table 2).
Equal benefit with 10% dextrose, but less likelihood of harm.
We continue to use 50% dextrose because . . . ?
Two of the subjects contacted by the researchers after treatment reported that before the study they had often had difficulty bringing their blood glucose back to their expected usual level after being treated by paramedics using 50% dextrose.
We need to question more of our traditional treatments.
 Myths of Toxicology: Thiamine Before Dextrose
Gussow, Leon MD
Emergency Medicine News:
April 2007 – Volume 29 – Issue 4 – pp 3,11
Free Full Text from Emergency Medicine News
 Myths and misconceptions of Wernicke’s encephalopathy: what every emergency physician should know.
Donnino MW, Vega J, Miller J, Walsh M.
Ann Emerg Med. 2007 Dec;50(6):715-21. Epub 2007 Aug 3. Review.
PMID: 17681641 [PubMed – indexed for MEDLINE]
 Dextrose 10% or 50% in the treatment of hypoglycaemia out of hospital? A randomised controlled trial.
Moore C, Woollard M.
Emerg Med J. 2005 Jul;22(7):512-5.
PMID: 15983093 [PubMed – indexed for MEDLINE]
Schabelman, E., & Kuo, D. (2012). Glucose before Thiamine for Wernicke Encephalopathy: A Literature Review The Journal of Emergency Medicine, 42 (4), 488-494 DOI: 10.1016/j.jemermed.2011.05.076
Moore, C. (2005). Dextrose 10% or 50% in the treatment of hypoglycaemia out of hospital? A randomised controlled trial Emergency Medicine Journal, 22 (7), 512-515 DOI: 10.1136/emj.2004.020693