Alcohol Withdrawal Syndrome

On a fairly regular basis, many doctors fail when it comes to indentifying people with 'alcohol use disorders'. This failure is often due to simply not asking the right questions. The consequences of such failure may be substantial, especially if the person has major surgery or develops a severe intercurrent illness, and then manifests acute alcohol withdrawal.

How do we identify alcohol use disorders?

Alcohol is the number one abused drug worldwide - this abuse results in many deaths, much illness, and staggering financial costs. Alcohol abuse has an extremely strong association with violent crime and also motor vehicle crashes. Yet all too often we as physicians gloss over the problem and miss the diagnosis. Over ten percent of 'clients' seen by general practitioners in many Westernised populations may actually have an alcohol use disorder. Much effort has been devoted to developing screening tools that help detect alcohol use disorders. The simplest tool is perhaps the CAGE questionnaire, composed of just four questions:
  1. Have you ever felt the need to CUT DOWN on your use of alcohol?
  2. Has anyone ANNOYED you by criticizing your use of alcohol?
  3. Have you ever felt GUILTY because of something you've done while drinking?
  4. Have you ever taken a drink (EYE OPENER ) to steady your nerves or get over a hangover?

Some of the questions may need to be 'culturally adjusted'. If two or more of the CAGE questions are answered 'Yes', (conventionally regarded as a 'positive' test), then the specificity is 89+% - this implies a high likelihood that the patient is alcoholic - there are very few false positives. Unfortunately, the sensitivity is far less, so many persons with an alcohol use disorder will be missed. Moving the cutoff to three questions, one achieves nearly 100% specificity, but at the further expense of sensitivity, which probably becomes a lot less than 50%.

There are several more complex tests than CAGE - 'TWEAK' with five questions (adding amnesia and tolerance, and removing 'annoyed' - designed to identify at-risk drinking in pregnant women), the World Health Organization's ten-question AUDIT, and the 25 (or 24) question Michigan Alcoholism Screening Test (MAST),  to mention just a few. The bottom line is that it's important to use some test to look for an alcohol use disorder - for example, we should probably screen every person undergoing major surgery.

An important component of history taking in such patients is to write down the raw information - not just a CAGE score, or whatever. How many units of alcohol does the patient consume per day? (one unit = 12g of alcohol ~ 375 ml beer ~ a glass of wine ~ a tot of spirits). Document past alcohol use, binge drinking, and the effects of the alcohol on the patient's behaviour, all in a non-confrontational fashion designed to extract maximal information! (A patient is probably more likely to admit to large amounts of alcohol consumption in conversation with a fellow drinker, than in discussion with a nun)! Does the patient use alcohol in risky situations (driving a vehicle, operating a chain saw)? Does drinking result in failure to meet social obligations? Have there been unsuccessful attempts to cut down? Have there been symptoms of withdrawal in the past (a biiig warning sign)?

Alcohol Withdrawal

The alcohol withdrawal syndrome has been variously called "delirium tremens" or "DTs" in the past. Here we explore manifestations, differential diagnosis, assessment of severity, consequences of withdrawal, and comorbidity. We then move on to management.

Manifestations of alcohol withdrawal

Florid alcohol withdrawal syndrome (AWS) presents with:

Note that the seizures commonly seen after alcohol withdrawal ("rum fits") occur soon - often about 6-8 hours after stopping drinking. This is long before AWS itself becomes severe, something which frequently takes three or four days! After a witnessed seizure related to alcohol withdrawal, 2mg of lorazepam IV is very effective in preventing a repeated seizure [N Engl J Med 1999 Mar 25;340(12):915-9], with a 3% recurrence of seizures as compared to 24% in a placebo group! Regarding further workup of seizures, it has been reported [BMJ 1994 Oct 15;309(6960):986-9] that the pick-up level of lesions on CT brain is very low in persons presenting with a first seizure, no focal neurological sign, and a 'Yes' to two or more CAGE questions. (Of interest is the recent report that serum homocysteine levels are substantially elevated in patients with alcohol withdrawal seizures [Neuroreport 2000 Aug 21;11(12):2749-52] - this may be due to actions at the NMDA receptor).

Differential diagnosis of alcohol withdrawal

There is a vast number of disorders that can result in confusion and systemic illness that mimics alcohol withdrawal. Particularly in the post-operative period, one should be aware of the commoner causes of 'ICU psychosis':

How severe is the withdrawal?

As with screening for alcohol use disorders, several tools have been used. The most-often used is probably the "Addiction Research Foundation C linical I nstitute for W ithdrawal A ssessment" scale, often mysteriously abbreviated to CIWA-Ar. This is complex and we won't repeat it in full here. (The scale is readily available on the Web. Modifications have been proposed). The basic idea is one of points rating of each of ten features of AWS, with a score of 8 points or more indicating withdrawal; once up to about 20, medication should be increased, or started, if not already prescribed. The maximum score is 67. The factors considered are nausea & vomiting, tremor, paroxysmal sweats, anxiety, agitation, tactile disturbance, auditory disturbances, visual disturbances, headache/fullness in the head, and clouding of the sensorium. Note that heart rate and blood pressure are NOT components of the system.

Consequences of Withdrawal

Severe withdrawal is associated with major morbidity and mortality, especially in a post-operative context {See BJS 1999 86 869}, where wound complications, bleeding, infection and cardiopulmonary compromise are prominent.

It may be important to tease out the relative contributions of alcohol withdrawal and associated co-morbidity (such as cirrhosis and, notably, malnutrition), to poor outcomes. As far as we're aware, nobody has considered this topic in detail, but a few authors have looked at some aspects:

Studies of alcohol withdrawal are confounded by their heterogeneity. Estimates of incidence of withdrawal range from 0 to 100%, presumably depending on the severity of alcohol use, and the reason for withdrawal! A reasonable guess as to the likelihood of severe withdrawal is in the region of 10%.


One should be aware of co-morbidities in patients undergoing AWS, notably the complications of chronic alcohol use including:

Managing Alcohol Withdrawal

Florid alcohol withdrawal can have a devastating effect on the patient, with an uncooperative and restless patient who may ultimately require heavy sedation. Large quantities of sedative exact their own toll, and one may end up needing to intubate the patient to protect the airway - aspiration pneumonitis/pneumonia may result from failing to do so!

One possible option for prevention of alcohol withdrawal is simply to allow the patient to continue drinking. No physician has been bold enough to publish a study that explores this option, although the author would consider it quite reasonable to permit say one unit of alcohol TDS in patients with an acute severe medical illness (such as a recent myocardial infarction) rather than have to cope with the additional problem of severe AWS. Most physicians go for withdrawal + benzodiazepines, although I think such a decision should be carefully negotiated with the patient - it seems singularly pointless to impose withdrawal on someone who will subsequently immediately resume drinking!

Several moderate-sized studies indicate that careful titration of benzodiazepines (according to CIWA-Ar score, for example) can prevent the more severe manifestations of AWS. There has been a general tendency to manage more and more AWS on an out-patient basis , but such management is relatively contra-indicated where there is a past history of florid withdrawal, recent heavy alcohol consumption, or seizures. In addition, substantial co-morbid disease may suggest that out-patient management is inappropriate.

Key elements of management include:

  1. An appropriate 'subdued' environment;
  2. appropriate nutrition, with B vitamin supplementation notably 100 mg thiamine/day;
  3. Pharmacological managment, thoroughly reviewed by Mayo-Smith.
The mainstay of therapy is benzodiazepines - a vast variety has been used. Shorter-acting benzodiazepines (such as lorazepam and oxazepam) are favoured by many because of the presumed lower risk of heavy sedation. Others favour longer acting agents such as diazepam or chlordiazepoxide because of a 'lower risk of breakthrough seizures'.


These are generally given PO or IV; some have used IM lorazepam in cases where IV access is difficult, although this seems sub-optimal. When one converts from PO to IV dosing, it is probably appropriate to halve, or more than halve the dose. Here's a table that gives approximately equivalent doses, and details of half-life.

halazepam 40 mg Long (14)§
oxazepam* 30 mg Med (5-20)
chlordiazepoxide 25+ mg Long (5-30)§ 25-100mg Q6hr PO
chlorazepate 15-20 mg Long (30-100)§
flurazepam 15 mg Long §
temazepam 15 mg Med (~6-24)
diazepam 10 mg Long (20-80+)§ 2-10mg Q2hr IV
prazepam 10 mg Long (30-100)§
midazolam 2+ mg short 2-20mg per hour IV infusion
(costly with no apparent benefit)
lorazepam * 1 mg Med (10-20) 1-2+mg Q2-4hr
clonazepam 0.5+ mg Long (18-50)§
alprazolam 0.25+ mg Med (6-12+)
triazolam 0.25+ mg Short (1.5-5)
* : no significant active metabolites; all the others have!
§ : action prolonged markedly due to active metabolites
(modified from Jaeger, and other sources)

As mentioned, some authorities use a fixed-dose regimen, but many have moved towards titrating benzodiazepine according to response. With long-acting agents such as chlordiazepoxide, it may be more reasonable to load the patient until sedated and then administer very little more benzodiazepine; conversely a "symptom-triggered" approach should perhaps use shorter-acting agents. The risk of use of long-acting agents is mainly over-sedation. In a recent rather select group of 216 patients (of 536 patients reviewed), Jaeger et al found on retrospective analysis that symptom-triggered therapy was associated with a lower rate of 'delirium tremens' when compared with usual (fixed-dose) therapy; the rates were 6.9% and 20.5% respectively.

Another study of symptom-triggered therapy (STT) is that of Saitz et al, who in a prospective randomised trial (n=101) found a remarkable difference in duration of treatment (9 hr vs 68 hr) and chlordiazepoxide use when comparing STT and fixed-schedule therapy. The message I get from this study is that conventional tapered therapy is 'overkill', although one must be careful in extrapolating from the study population to one's own! Note that the Saitz study was quite carefully designed, and that severity of alcoholism was assessed (using yet another obscure tool called QIAD, the Quantitative Inventory of Alcohol Disorders). There were also 128 exclusions from the study - for any seizures, opiate or benzodiazepine withdrawal, beta-blocker use, and more.

A very recent, similar study by Daeppen et al supports the conclusions of Saitz. In 117 patients withdrawing from alcohol dependence (mostly white males), only 39% of the symptom-triggered group required oxazepam, and less benzodiazepine was used for a shorter period of time, when compared with controls given fixed-schedule oxazepam. The fixed schedule regimen was 30mg 6 hourly for 4 doses, then 15mg 6 hourly for 8 doses, omitting doses when the patient was somnolent or sleeping; the symptom-triggered approach used titration of therapy according to CIWA-Ar score (score 8-15: 15mg oxazepam; score> 15: 30mg), with additional doses every 1/2 hour according to score. The study was placebo-controlled. Group differences persisted even after excluding those who did not require oxazepam.

Other agents

The following have been used with varying success:


  1. Chang PH, Steinberg MB. Med Clin North Am. 2001 Sep;85(5):1191-212 {A good review of alcohol withdrawal, well worth reading}

  2. Am Fam Physician 1999 Sep 15;60(4):1175-83 {An AFP review of outpatient detoxification of the addicted or alcoholic patient, worth a cautious read}.

  3. Mayo-Smith F. JAMA 1997 278(2) 144-151. {A meta-analysis of thirty years of pharmacological management of alcohol withdrawal. The full text is available here}.

  4. Jaeger TM et al. Mayo Clin Proc 2001 Jul;76(7):695-701 {A less-than-perfect study comparing "symptom-triggered" and conventional therapy for alcohol withdrawal}. Also look at Saitz:

  5. Saitz R, et al. JAMA 1994 272 519-523 {A prospective study (randomised, double-blind) comparing QID chlordiazepoxide and symptom-triggered therapy}.

  6. Daeppen et al: Arch Intern Med 2002 May 27;162(10):1117-21 {A recent trial of symptom-triggered therapy}.

  7. Here's a good web review on DT's in ICU by David Crippen.

  8. A superb, detailed review of benzodiazepines in PDF format.

  9. Tanaka E. Journal of Clinical Pharmacy and Therapeutics (1999) 24, 347-355 Benzodiazepine-drug interactions available as a PDF from UCSD.

  10. Dawe S, & Mattick RP: "Review of diagnostic screening instruments for alcohol and other drug use and other psychiatric disorders" Australian Government Publishing Service. ISBN 0 644 39747 0. {a comprehensive review of screening tools}.