Alcohol Withdrawal: The Evidence Behind the Pathway

It is the second morning after admission. A 46-year-old painter fell from a ladder and broke his ankle, and he was admitted to the orthopaedic ward for fixation. On the first day he was cheerful and asked when he could go home. This morning the nurse calls you: his hands are shaking, he is sweating, his pulse is 118, and he keeps asking whether the operation is over. His wife tells you, when you ask, that he drinks a quarter bottle of whisky every evening, more on Sundays, and that his last drink was the night before the fall. Two years ago, when he stopped for a family function, he “had a fit”. This is an illustrative case, not a real patient.

The Alcohol Withdrawal Pathway will help you with him: his PAWSS score, his CIWA-Ar score, which benzodiazepine, at what dose, on which schedule, how much thiamine, and what to start before he goes home. What it no longer carries is most of the why. The mechanism, the trial results and the full agent and relapse prevention tables that used to sit in its Protocols and Evidence tabs have moved here, so that the tool stays quick at the bedside. By the end of this article you will know what stands behind each of those instructions, and where the evidence is firm.

1. Why the brain shakes when the alcohol stops

Why begin with receptors, when the patient in front of you needs a dose? Because the mechanism tells you which drugs can work and which cannot, and that is a bedside decision. This used to be section 1 of the tool’s Evidence tab.

The brain runs on a balance between inhibition and excitation. The main inhibitory signal works through the GABA-A receptor; the main excitatory one through the NMDA glutamate receptor. Alcohol is a positive allosteric modulator at the GABA-A receptor and an antagonist at the NMDA glutamate receptor. A positive allosteric modulator is a drug that makes a receptor respond more strongly to its own natural signal. So alcohol pushes the balance towards inhibition, in two places at once.

The brain adapts to anything that is there every day. Sustained exposure downregulates GABA-A receptors and shifts their subunit composition while upregulating NMDA receptors. Think of a car driven every day with the handbrake half on: the driver learns to press the accelerator harder. Now release the handbrake suddenly. Taking the alcohol away leaves the adapted brain with too little inhibition and too much excitation at the same time. The tremor, the sweating, the racing pulse, the fits and, later, the delirium all come from that one imbalance.

“Then can any sedative settle him?” No, and this is the practical point. Benzodiazepines work because they act at the receptor the alcohol was acting on, and nothing that does not act there can substitute for them. That is why the tool insists that dexmedetomidine is an adjunct only, and must never replace the benzodiazepine: it calms the pulse and the agitation without touching the receptor that is out of balance.

The same adaptation explains kindling. Each successive untreated episode of withdrawal is more severe than the last. That is why a history of previous withdrawal is the single most useful predictive question you can ask. Our painter has already answered it: he had a fit when he stopped two years ago.

2. Symptom-triggered dosing works, where the ward can score it

There are two ways to give the benzodiazepine. In a fixed-dose taper, he gets scheduled doses that come down day by day. In symptom-triggered dosing, the nurse scores him on the CIWA-Ar scale every hour and gives a dose only when the score is 8 or above. Which is better?

Randomised comparisons have consistently shown that symptom-triggered dosing reduces cumulative benzodiazepine by roughly threefold to fourfold and shortens treatment duration, without an increase in seizures or delirium (Saitz 1994). The tool’s regimen table used to put it as roughly a fourfold reduction in cumulative benzodiazepine and a shorter treatment duration compared with fixed dosing. Less drug, a shorter course, and no more fits: that is a good result.

“So why does the tool not simply choose symptom-triggered dosing every time?” Because that result depends entirely on the scoring being done properly and hourly by a nurse trained in the instrument. This is the common mistake, and it is an honest one: the protocol is written on the chart, so it feels as if it is happening. On a ward where the score is copied forward from the last shift, symptom-triggered dosing becomes no dosing. The tool asks you what your ward can actually monitor, and offers a fixed-dose taper where hourly scoring is not realistic. A fixed taper given properly is safer than a symptom-triggered protocol scored badly.

Remember too that six of the ten CIWA-Ar items need the patient to tell you a symptom. Our painter, asking whether his operation is over, may already be beyond reliable self-report. Once delirium has set in, the tool tells you to stop scoring CIWA-Ar and to titrate against his sedation level and his autonomic signs.

3. Reading the benzodiazepine table

Here is the agent table, whole, as the tool carried it on 4 November 2026. On a phone the tool now shows it as two narrow tables: the doses, then the preparations with the notes.

Read the last column first. It tells you why you would choose each drug, and the choice is about the liver. Chlordiazepoxide and diazepam are broken down in the liver into long-lived active metabolites, so they taper themselves smoothly in a patient with a healthy liver, and pile up dangerously in one with cirrhosis or in the elderly. Lorazepam has no active metabolites and does not accumulate. That is why it is the choice in significant liver disease, at 65 and above, and where breathing is a worry. Our painter has no sign of liver disease, so chlordiazepoxide is reasonable. With a previous fit, the tool may also offer front-loading with diazepam.

“Can I switch from one to another using the equivalence line?” Use it as a guide only. The tool keeps the approximate equivalence beside the table: diazepam 10 mg is about chlordiazepoxide 25 mg, about lorazepam 1 mg. Those figures say nothing about how long each one acts, which is exactly what differs between them.

4. Why phenytoin does not prevent the fits

Our painter had a fit last time. Should he get phenytoin this time? It is prescribed for this often enough that the question deserves a straight answer.

Withdrawal seizures arise from a transient receptor imbalance rather than from an epileptic focus, and randomised trials of phenytoin for their prevention and for the prevention of recurrence have been negative (Rathlev 1994). An epileptic focus is a fixed patch of irritable cortex, the thing phenytoin is good at quietening. A withdrawal seizure has no such patch. It is the whole brain, briefly out of balance, and the drug that restores the balance is the one acting at the GABA-A receptor. Intravenous lorazepam has been shown to reduce recurrence after a first withdrawal seizure (D’Onofrio 1999). The benzodiazepine is both the treatment and the prophylaxis.

One caution the tool keeps: a seizure that is focal, follows a head injury, comes with fever, leaves a long postictal state, or falls outside the 6 to 48 hour window is not a simple withdrawal seizure, and needs its own work-up. A man who fell from a ladder has had a head injury until you have looked.

5. Thiamine before glucose, and magnesium with it

Suppose his blood glucose is a little low, and the house officer reaches for a dextrose drip. Why does the tool tell you to give thiamine first?

Thiamine pyrophosphate is the cofactor for transketolase, pyruvate dehydrogenase and alpha-ketoglutarate dehydrogenase. A cofactor is a helper molecule the enzyme cannot work without. These are the enzymes that burn glucose for energy. A glucose load in a thiamine-deplete patient drives these enzymes and can exhaust the remaining thiamine, which is how a well-intentioned dextrose infusion precipitates Wernicke encephalopathy.

All three enzymes also require magnesium, so thiamine repletion can fail entirely while the serum magnesium is low, and hypomagnesaemia is close to universal in chronic alcohol use because of renal magnesium wasting. Hence the tool’s sequence: thiamine, then magnesium, then glucose. The exception is real hypoglycaemia, where glucose goes in at once and thiamine immediately after, because low sugar harms the brain within minutes.

“How much thiamine?” The tool gives the doses, and is honest that the dose for established Wernicke encephalopathy is not settled between the UK and European guidance (Galvin 2010). Thiamine is cheap and close to harmless. An untreated Wernicke encephalopathy is not reversible. Our painter, confused and drinking daily, gets parenteral thiamine now, while the question is being settled.

6. The admission is not over until relapse prevention is decided

Most patients who are detoxified and sent home without a drug to prevent relapse are drinking again soon, and kindling makes the next withdrawal worse. Here is the tool’s relapse prevention table, whole. On a phone it is now two narrow tables: the dose with when to avoid each drug, then the notes.

“Disulfiram only frightens people. Does it really work?” It works through deterrence, so it works only when somebody else gives the tablet. Supervised disulfiram given by a family member has performed well in Indian pragmatic trials, for structural rather than pharmacological reasons (De Sousa 2004). In many Indian homes a spouse or a parent is willing to hand over the tablet every morning, and that supervision is exactly what the drug depends on. Use it deliberately, with the counselling the tool insists on: the patient and the family must know what a reaction feels like, and where alcohol hides, including mouthwashes, tonics and some asava and arishta preparations.

Back to the ward

Our painter is withdrawing on day two, with a previous withdrawal fit. His PAWSS is high. He is drifting towards confusion, so his CIWA-Ar score is no longer trustworthy, and the orthopaedic ward cannot score him every hour. He gets a benzodiazepine on a schedule the ward can deliver, titrated to light sedation, not to a number. He gets parenteral thiamine before any glucose, and his magnesium is checked and corrected. Nobody starts phenytoin. His head is examined properly, because he fell. Before discharge, his wife is asked whether she will supervise a tablet each morning. Each of those steps is a line in the tool. Now you know what stands behind each line.

Try this

A 58-year-old woman with alcohol-related cirrhosis and ascites is admitted with withdrawal. The intern writes chlordiazepoxide 50 mg every 6 hours, and asks whether naltrexone can be started at discharge. What do you change? Illustrative case.

Answer. Two things. Chlordiazepoxide makes long-lived active metabolites that her liver cannot clear, so it will pile up and can tip her into encephalopathy. Choose lorazepam, which does not accumulate, and accept that she will need doses more often. And naltrexone is to be avoided in hepatic failure. Acamprosate, which is renally excreted, is the agent of choice in cirrhosis, provided her creatinine clearance is not below 30 mL/min.

Take-home points

• Withdrawal is too little inhibition and too much excitation. Only a drug acting at the GABA-A receptor corrects it; nothing replaces the benzodiazepine.
• Ask every admitted drinker about previous withdrawal and previous fits. Kindling makes each episode worse.
• Symptom-triggered dosing saves drug only where the ward can score hourly. Otherwise give a fixed taper properly.
• Phenytoin does not prevent withdrawal seizures. Benzodiazepines do. Thiamine goes in before glucose, with magnesium.
• Decide relapse prevention before discharge. In cirrhosis think lorazepam for withdrawal and acamprosate afterwards.

Declaration: the author built and maintains the tool. It is free to use. The figures in this article are those the tool carried in its Protocols and Evidence tabs on 4 November 2026, each from the source named beside it.

References

1. Sullivan JT, Sykora K, Schneiderman J, Naranjo CA, Sellers EM. Assessment of alcohol withdrawal: the revised Clinical Institute Withdrawal Assessment for Alcohol scale (CIWA-Ar). Br J Addict. 1989;84(11):1353-1357.
2. Maldonado JR, Sher Y, Das S, et al. Prospective validation study of the Prediction of Alcohol Withdrawal Severity Scale (PAWSS) in medically ill inpatients. Alcohol Alcohol. 2015;50(5):509-518.
3. The ASAM Clinical Practice Guideline on Alcohol Withdrawal Management. J Addict Med. 2020;14(3S Suppl 1):1-72.
4. Saitz R, Mayo-Smith MF, Roberts MS, Redmond HA, Bernard DR, Calkins DR. Individualized treatment for alcohol withdrawal: a randomized double-blind controlled trial. JAMA. 1994;272(7):519-523.
5. D’Onofrio G, Rathlev NK, Ulrich AS, Fish SS, Freedland ES. Lorazepam for the prevention of recurrent seizures related to alcohol. N Engl J Med. 1999;340(12):915-919.
6. Rathlev NK, D’Onofrio G, Fish SS, et al. The lack of efficacy of phenytoin in the prevention of recurrent alcohol-related seizures. Ann Emerg Med. 1994;23(3):513-518.
7. Galvin R, Brathen G, Ivashynka A, Hillbom M, Tanasescu R, Leone MA. EFNS guidelines for diagnosis, therapy and prevention of Wernicke encephalopathy. Eur J Neurol. 2010;17(12):1408-1418.
8. Caine D, Halliday GM, Kril JJ, Harper CG. Operational criteria for the classification of chronic alcoholics: identification of Wernicke’s encephalopathy. J Neurol Neurosurg Psychiatry. 1997;62(1):51-60.
9. Harper CG, Giles M, Finlay-Jones R. Clinical signs in the Wernicke-Korsakoff complex: a retrospective analysis of 131 cases diagnosed at necropsy. J Neurol Neurosurg Psychiatry. 1986;49(4):341-345.
10. National Institute for Health and Care Excellence. Nutrition support for adults: oral nutrition support, enteral tube feeding and parenteral nutrition. Clinical guideline CG32. London: NICE.
11. Basu D, Ghosh A, Subodh BN, Mattoo SK. Clinical Practice Guidelines for the Management of Alcohol Use Disorders. Indian Psychiatric Society. Indian J Psychiatry.
12. De Sousa A, De Sousa A. A one-year pragmatic trial of naltrexone vs disulfiram in the treatment of alcohol dependence. Alcohol Alcohol. 2004;39(6):528-531.
13. Schuckit MA. Recognition and management of withdrawal delirium (delirium tremens). N Engl J Med. 2014;371(22):2109-2113.