If this is not met, PAWSS does not apply and the remaining items are not scored.
| Strategy | How it is given | When it is the right choice |
|---|---|---|
| Symptom-triggered | Score CIWA-Ar hourly. Give a dose whenever the score is 8 or above, then rescore in one hour. Once three consecutive hourly scores are below 8, move to two hourly, then four hourly, then eight hourly scoring. | Trained nursing with hourly scoring genuinely feasible, and the patient can be scored. Roughly a fourfold reduction in cumulative benzodiazepine and a shorter treatment duration compared with fixed dosing. |
| Fixed-dose taper | Scheduled doses over 5 to 7 days with additional doses available as required, tapering by about 25 per cent per day. | Hourly CIWA-Ar scoring is not realistic, the patient cannot be scored, or ambulatory detoxification is being supervised by family. Given properly it is safer than a symptom-triggered protocol scored badly. |
| Front-loading | Diazepam 20 mg orally every 1 to 2 hours until CIWA-Ar falls below 8 to 10 or the patient is lightly sedated, usually after 3 to 4 doses. The long half-life of diazepam and its active metabolites then provides a self-taper, and no scheduled taper is needed. | Predicted severe withdrawal, a previous seizure or delirium tremens, or a patient already scoring in the severe range. Requires a setting able to observe for over-sedation. |
| Agent | Symptom-triggered dose | Fixed taper | Preparation in India | Notes |
|---|---|---|---|---|
| Chlordiazepoxide | 50 to 100 mg orally per dose, maximum 300 mg in 24 hours | 50 mg every 6 hours for 4 doses, then 25 mg every 6 hours for 8 doses, then stop | Capsules and tablets 10 mg and 25 mg. Oral only: no parenteral preparation in routine use | First line where hepatic function is preserved. Long half-life with active metabolites gives a smooth self-taper |
| Diazepam | 10 to 20 mg orally per dose. In delirium tremens 5 to 10 mg intravenously every 5 to 10 minutes | 10 mg every 6 hours for 4 doses, then 5 mg every 6 hours for 8 doses, then stop | Tablets 5 mg and 10 mg. Injection 10 mg in 2 mL, so 5 mg is 1 mL | The front-loading agent. Half-life 20 to 100 hours with an active metabolite, so accumulation is the intended effect and also the risk |
| Lorazepam | 2 to 4 mg orally or intravenously per dose. In delirium tremens 2 to 4 mg intravenously every 10 to 15 minutes | 2 mg every 6 hours for 4 doses, then 1 mg every 6 hours for 8 doses, then stop | Tablets 1 mg and 2 mg. Injection 4 mg in 2 mL, so 2 mg is 1 mL | Choose in significant liver disease, age 65 and above, and respiratory compromise. Conjugated directly, no active metabolites, so it does not accumulate |
| Oxazepam | 15 to 30 mg orally per dose | 30 mg every 6 hours for 4 doses, then 15 mg every 6 hours for 8 doses | Availability in India is limited and inconsistent | Shares the metabolic advantage of lorazepam. Listed for completeness; lorazepam is the practical choice here |
| Situation | Regimen | What to draw up |
|---|---|---|
| Low risk. Well nourished, eating, no features of Wernicke encephalopathy, no decompensated liver disease | Thiamine 100 mg orally three times daily, continued on discharge | Tablets 100 mg |
| High risk prophylaxis. Malnourished, BMI below 18.5, decompensated liver disease, vomiting, admitted for detoxification, or previous Wernicke encephalopathy | Thiamine 200 to 300 mg intravenously or intramuscularly once daily for 3 to 5 days, then 100 mg orally three times daily | Injection 100 mg/mL in 2 mL ampoules, so 200 mg is one ampoule. Dilute in 100 mL normal saline and infuse over 30 minutes |
| Suspected or established Wernicke encephalopathy. Two or more Caine criteria | Thiamine 500 mg intravenously over 30 minutes, three times daily, for 2 to 3 days. If there is a response, continue 250 mg once daily for a further 3 to 5 days, then 100 mg orally three times daily long term | 500 mg is 5 mL of the 100 mg/mL preparation, that is two and a half 2 mL ampoules, diluted in 100 mL normal saline |
| Step | What to do |
|---|---|
| 1. Escalating intravenous benzodiazepine | Diazepam 5 to 10 mg intravenously every 5 to 10 minutes, or lorazepam 2 to 4 mg intravenously every 10 to 15 minutes, doubling the dose if the previous one had no effect. The endpoint is light sedation with the patient rousable, not a number on a scale. |
| 2. Recognise resistance | Benzodiazepine-resistant delirium tremens is conventionally defined as a requirement above about 200 mg of diazepam equivalent within 3 to 4 hours without adequate control. At that point add a second agent rather than continuing to escalate alone. |
| 3. Phenobarbitone | 10 mg/kg intravenously in 100 mL normal saline over 30 minutes, or 130 to 260 mg intravenously every 15 to 30 minutes. Indian preparation is 200 mg/mL, so a 700 mg dose is 3.5 mL. Cross-tolerant with alcohol at the GABA-A receptor and effective where benzodiazepines are not. Respiratory depression is additive with high-dose benzodiazepine: airway skills and monitoring must be present before the first dose. |
| 4. Dexmedetomidine | 0.2 to 1.4 microgram/kg/hour by infusion, no bolus. Adjunct only. It controls autonomic signs and agitation but has no action at the GABA-A receptor, does not prevent withdrawal seizures and does not treat the underlying receptor adaptation. It must never replace the benzodiazepine. See ICU Infusion Dosing. |
| 5. Antipsychotic | Haloperidol 2.5 to 5 mg intramuscularly or intravenously, only for hallucinations or agitation not controlled by adequate benzodiazepine, and always alongside it. Lowers the seizure threshold and prolongs the QT interval, so an ECG is needed and it is never used as monotherapy. |
| 6. Supportive care | Correct magnesium, potassium and phosphate. Thiamine before glucose. Quiet, well-lit room, orientation cues, familiar attendant. Look actively for the infection, head injury or metabolic cause that is mimicking or driving the delirium. |
| Agent | Dose | Avoid in | Notes |
|---|---|---|---|
| Naltrexone | 25 mg daily for 3 to 4 days to reduce nausea, then 50 mg orally once daily | Current opioid use or opioid dependence, anticipated need for opioid analgesia, acute hepatitis, hepatic failure | Strongest evidence for reducing heavy drinking days. Can be started while the patient is still drinking. Check liver enzymes before starting and periodically |
| Acamprosate | 666 mg orally three times daily above 60 kg; 333 mg three times daily below 60 kg. Halve the dose if creatinine clearance is 30 to 50 mL/min | Creatinine clearance below 30 mL/min | Renally excreted and hepatically safe, which makes it the agent of choice in cirrhosis. Supports maintenance of abstinence. Start once detoxification is complete |
| Disulfiram | 250 mg orally once daily, range 125 to 500 mg. At least 24 hours of abstinence before the first dose | Ischaemic heart disease, significant arrhythmia, severe hepatic disease, psychosis, pregnancy, peripheral neuropathy | Works through deterrence, so supervised administration is what makes it effective. Indian trials with family supervision have found it to perform well. Requires explicit informed consent |
| Baclofen | 10 mg orally three times daily, titrated upwards | Significant renal impairment | Renally cleared and hepatically safe. An option in advanced cirrhosis where the others are unsuitable. Evidence base is smaller and less consistent than for the first three |
Alcohol is a positive allosteric modulator at the GABA-A receptor and an antagonist at the NMDA glutamate receptor. Sustained exposure produces adaptation in both directions: GABA-A receptors are downregulated and their subunit composition changes, while NMDA receptors are upregulated. Remove the alcohol and the adapted brain is left with too little inhibition and too much excitation at the same time, which is the whole clinical picture. The corollary is the reason benzodiazepines work: they act at the same receptor the alcohol was acting on, and nothing that does not act there can substitute for them. It also explains kindling, in which each successive untreated withdrawal is more severe than the last, and why a history of previous withdrawal is the single most useful predictive question you can ask.
Minor withdrawal, meaning tremor, anxiety, sweating, palpitations and gastrointestinal upset, begins 6 to 12 hours after the last drink. Alcoholic hallucinosis appears at 12 to 24 hours and is defined by hallucinations with a clear sensorium, which is what separates it from delirium tremens. Withdrawal seizures occur between 6 and 48 hours with a peak at 12 to 24 hours. Delirium tremens appears later, typically at 48 to 96 hours and occasionally up to day 5. Bedside pearl: a patient who becomes confused 12 hours after the last drink does not have delirium tremens, and the diagnosis to pursue is hepatic encephalopathy, sepsis, hypoglycaemia or a subdural haematoma. A patient who is entirely well at 24 hours is not yet out of the window for delirium.
These are routinely conflated and they answer different questions. CIWA-Ar is a severity instrument for a patient who is already withdrawing, and it drives dosing. PAWSS is a prediction instrument scored on admission, before withdrawal has declared itself, and it drives placement and prophylaxis. The patient who most needs PAWSS is the one admitted for something else entirely, a fractured femur or a chest infection, who looks well on day one and convulses on day two. Bedside pearl: ask every admitted patient the four questions that carry most of the predictive weight: have you had withdrawal before, have you ever had a fit when stopping, have you ever had the DTs, and when was your last drink.
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. That result depends entirely on the scoring being done properly and hourly by a nurse trained in the instrument. On a ward where the score is copied forward from the previous shift, or performed once a day, symptom-triggered dosing becomes no dosing, and the patient who deteriorates does so unobserved. Indian context: nursing ratios on general medical wards in most Indian teaching hospitals do not permit hourly scoring of every patient in withdrawal. This is not a reason to avoid the strategy, but it is a reason to be honest about which wards can support it, and to use a fixed-dose taper elsewhere rather than a symptom-triggered protocol in name only.
Six of the ten CIWA-Ar items require the patient to report a subjective symptom: nausea, anxiety, tactile, auditory and visual disturbances, and headache. A delirious patient cannot report these reliably, an intubated patient cannot report them at all, and a patient who does not share a language with the nurse scoring them will be scored low regardless of how unwell they are. The score in these patients is not merely imprecise, it is falsely reassuring, because the items that cannot be obtained default to zero. Bedside pearl: once delirium has set in, stop scoring CIWA-Ar and switch to titrating against sedation level and autonomic signs. The instrument has done its job by then; continuing to use it substitutes a number for the clinical assessment that is actually needed.
This is worth stating plainly because it is prescribed for this indication regularly. Alcohol 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. Benzodiazepines are both the treatment and the prophylaxis, and intravenous lorazepam has been shown to reduce recurrence after a first withdrawal seizure. Point to note: a patient who is already on phenytoin for established epilepsy continues it, and a first-ever seizure still needs proper evaluation. Focal onset, a prolonged postictal state, head injury, fever or a seizure occurring outside the 6 to 48 hour window all argue against a simple withdrawal seizure and warrant imaging, glucose, sodium and consideration of meningitis. See the Seizure Pathway.
The classic triad of confusion, ataxia and ophthalmoplegia is present together in only about 10 to 16 per cent of cases at presentation, and post-mortem series have consistently found Wernicke encephalopathy that was never diagnosed in life. The Caine criteria, requiring two of dietary deficiency, oculomotor abnormality, cerebellar signs, and either altered mental state or mild memory impairment, are far more sensitive and are the practical bedside standard. Why it matters: thiamine is inexpensive, essentially non-toxic, and the damage it prevents is irreversible once established. The asymmetry of that trade-off is such that any patient with alcohol dependence and any degree of confusion should receive parenteral thiamine while the question is being settled, not after.
Thiamine pyrophosphate is the cofactor for transketolase, pyruvate dehydrogenase and alpha-ketoglutarate dehydrogenase. A glucose load given to a thiamine-deplete patient drives these enzymes and can exhaust the remaining thiamine, which is the mechanism by which a well-intentioned dextrose infusion precipitates Wernicke encephalopathy. All three of these 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. Bedside pearl: the sequence is thiamine, then magnesium, then glucose. The exception is the hypoglycaemic patient, in whom glucose is given at once and thiamine immediately afterwards: hypoglycaemia will cause harm within minutes, and the theoretical risk does not outrank it.
Delirium tremens is a diagnosis of exclusion in a population with an unusually high burden of the things it is confused with. Hepatic encephalopathy shares the confusion and is worsened by the benzodiazepines that treat withdrawal, so asterixis and a search for precipitants are worth the two minutes they take. Subdural haematoma is common in patients who fall and who may be coagulopathic, and it can present days later with fluctuating confusion. Sepsis produces fever, tachycardia and confusion together, and fever is not a feature of uncomplicated withdrawal, so a temperature above 100.4 F should send you looking for pneumonia or spontaneous bacterial peritonitis rather than being attributed to the withdrawal. Hypoglycaemia, hyponatraemia and alcoholic ketoacidosis complete the list. Bedside pearl: the patient whose confusion does not improve with adequate benzodiazepine does not have delirium tremens, or does not have only delirium tremens.
A patient who has been drinking heavily, then stopped because of vomiting and abdominal pain, presenting with a high anion gap metabolic acidosis, ketonuria and a glucose that is normal or low, has alcoholic ketoacidosis. It responds to thiamine followed by glucose-containing saline and does not require insulin. Separately, and more urgently in the Indian setting: a cluster of patients who drank country liquor or hooch, presenting with visual disturbance, a severe high anion gap acidosis and an elevated osmolar gap, have methanol poisoning. That is a toxicological emergency requiring fomepizole or ethanol and haemodialysis, and it is not withdrawal at all. The ABG Diagnostic Pathway will characterise the acidosis, but note that no module in this catalogue currently calculates the osmolar gap or holds antidote dosing, so that part of the decision has to be made from another source.
Chlordiazepoxide and diazepam undergo oxidative metabolism in the liver and generate long-lived active metabolites, principally desmethyldiazepam. That accumulation is exactly what gives them their smooth self-taper in a patient with normal hepatic function, and exactly what makes them dangerous in decompensated cirrhosis and in the elderly, where the same accumulation produces prolonged sedation and can precipitate encephalopathy. Lorazepam, oxazepam and temazepam are conjugated directly by glucuronidation, a pathway that is comparatively preserved in liver disease, and they have no active metabolites. Bedside pearl: the memory aid is the three agents whose names supply the letters L, O and T. In significant liver disease, in the elderly, and where respiratory reserve is limited, choose from those three and accept that you will need to dose more often.
Most patients who are detoxified and discharged without pharmacotherapy relapse quickly, and the readmission is often more severe than the admission because of kindling. Naltrexone has the best evidence for reducing heavy drinking and can be started while the patient is still drinking; acamprosate supports abstinence once achieved and is the choice in cirrhosis because it is renally rather than hepatically cleared; disulfiram works by deterrence and therefore only works when someone else gives it. Indian context: supervised disulfiram given by a family member has performed well in Indian pragmatic trials, and the reason is structural rather than pharmacological. Where a spouse or parent is willing to give the tablet daily and the household is cooperative, that supervision is available in India in a way it often is not elsewhere, and it should be used deliberately rather than treated as an afterthought. The decision belongs before discharge, not at the first follow-up visit that many patients never attend.
The module stops at the point where specialist judgement takes over. It does not manage withdrawal in pregnancy, where untreated withdrawal is more dangerous to the pregnancy than benzodiazepine exposure but the decision belongs with an obstetrician. It does not cover patients below 18 years. It does not manage established Korsakoff syndrome or the rehabilitation that follows it, and it does not substitute for psychiatric assessment of the comorbid depression, anxiety and suicide risk that accompany alcohol dependence at high frequency. It also assumes the diagnosis: a patient who is confused and has been drinking is not necessarily in withdrawal, and pearl 9 lists what else to rule out before committing to that label.
AMA Style:
Umakanth S. Alcohol Withdrawal & CIWA-Ar Management Pathway. MEDiscuss. Published 2026. Accessed .
Vancouver Style:
Umakanth S. Alcohol Withdrawal & CIWA-Ar Management Pathway [Internet]. MEDiscuss.org; 2026 [cited ]. Available from:
Whatever you type here stays on this device and is not sent anywhere. The server receives only a scrambled code made from it, so nobody with access to the server can tell which patient a saved calculation belongs to. Enter the same nickname the next time to see that patient's earlier values. What is stored
