The clinical history often suggests alcohol problems, which are seen in 15 to 20 percent of patients in primary care and hospitalized patients. Since alcohol withdrawal can be complicated by seizures and delirium and is more severe in persons with more previous episodes of withdrawal or other illnesses, careful evaluation is essential. Such evaluation should include assessment for anemia, thrombocytopenia, and elevated liver-enzyme levels. Specific symptoms during sedative or alcohol withdrawal that may dictate pharmacotherapy include auditory and tactile disturbances and seizures.
Significant components of the withdrawal syndrome after chronic alcohol use reflect reduced neurotransmission in type A γ-aminobutyric acid pathways and enhanced neurotransmission in glutamate (N-methyl-D-aspartate) pathways. Seizures during withdrawal probably result from this altered neurobiology, but the anticonvulsant medication phenytoin is not an effective treatment for alcohol withdrawal. Benzodiazepines are effective, probably owing to cross-tolerance with ethanol at the type A γ-aminobutyric acid receptor, where carbamazepine and divalproate are also believed to have their main actions.
On the basis of this neurobiology, antagonism at the N-methyl-D-aspartate receptor is another potential mechanism for relieving symptoms of alcohol and sedative withdrawal and may be important for reducing the toxic effects from repeated episodes of alcohol withdrawal.
Withdrawal symptoms can be quantified to allow symptom-triggered therapy, in which the patient receives medication only when symptoms exceed a threshold of severity, rather than on a fixed schedule. This approach is as effective as fixed-dose therapy but requires significantly less medication and leads to a more rapid detoxification. It does, however, require careful and frequent monitoring with a validated withdrawal-symptom scale such as the Clinical Institute Withdrawal Assessment for Alcohol. The categories on this scale are sweating, anxiety, tremor, auditory or visual disturbances, agitation, nausea and vomiting, tactile disturbances, headache, and orientation. Total symptom scores of more than 15 on this scale or a history of withdrawal seizures indicates that medications should be started at presentation.
Without medication, alcohol-withdrawal symptoms might be expected to peak about 72 hours after the last use of alcohol, but medications can reduce symptoms within hours. In patients with delirium tremens, management with medication requires high doses of benzodiazepines (e.g., 5 to 10 mg of diazepam by intravenous injection, repeated in two to four hours if seizures occur). Unless delirium is present, medication is typically needed for no more than seven days after the last use of alcohol, although some patients will report withdrawal symptoms, including sleep problems, for several more weeks. Protracted symptoms may precipitate relapse.
Withdrawal from benzodiazepines and other sedatives produces more psychomotor and autonomic nervous system signs than does withdrawal from alcohol, and these signs start between 2 and 10 days after abrupt discontinuation. If medications are used, treatment with anticonvulsant drugs such as carbamazepine will need to be continued for about two weeks, and the dose of benzodiazepines gradually tapered.
Benzodiazepines and Barbiturates
Two major reviews of pharmacotherapy for alcohol withdrawal concluded that benzodiazepines are the treatment of choice on the basis of several outcomes, including the severity of the alcohol-withdrawal syndrome, occurrence of delirium and seizures, adverse effects of the medication, and completion of withdrawal, as well as subsequent entry into rehabilitation. A meta-analysis comparing benzodiazepines with placebo or with an active control drug included 11 trials, representing a total of 1286 patients. There was more often a clinically significant reduction of symptoms within two days with benzodiazepines than with placebo (common odds ratio, 3.28; 95 percent confidence interval, 1.30 to 8.28).
In addition, in six prospective trials, benzodiazepines, particularly longer-acting ones, were more effective than placebo in reducing the incidence of seizures (risk reduction, 7.7 seizures per 100 patients treated; P=0.003) and delirium (risk reduction, 4.9 cases of delirium per 100 patients treated; P=0.04). However, the potential for abuse is greater with agents that have a rapid onset of action, including diazepam, alprazolam, and lorazepam, than for those with slower onset of action, such as chlordiazepoxide, oxazepam, and halazepam. Although phenobarbital has a low potential for abuse as compared with other barbiturates and is used by about 10 percent of substance-abuse programs in the United States, its use is supported by only a few controlled studies. Furthermore, phenobarbital has a poorer safety profile than benzodiazepines: it can cause respiratory depression when used in high doses or when combined with alcohol, as may happen with outpatients.
Other drugs, particularly carbamazepine, do not differ significantly from the benzodiazepines in terms of adverse events (common odds ratio for adverse events associated with benzodiazepines as compared with carbamazepine, 0.67; 95 percent confidence interval, 0.34 to 1.32). Dropout rates in the first seven days are slightly lower with benzodiazepines (common odds ratio, 0.68; 95 percent confidence interval, 0.47 to 0.97). Thus, a critical issue may be the potential effectiveness of anticonvulsant agents for outpatient treatment in more severe as well as mild-to-moderate alcohol withdrawal.
The treatment of benzodiazepine or barbiturate dependence has involved either tapering dosages of the agent of dependence or substituting a longer-acting benzodiazepine or phenobarbital, with a gradual reduction in dose over a period of two weeks. Tables in Smith and Wesson provide dosage equivalents for use in substituting longer-acting for shorter-acting agents.
Although they may be useful as adjuvant treatments, most other agents are unsuitable for use alone during alcohol withdrawal. The phenothiazines and haloperidol reduce signs and symptoms of withdrawal but are significantly less effective than benzodiazepines in preventing delirium (difference in risk, 6.6 cases per 100 patients) and seizures (difference in risk, 12.4 seizures per 100 patients) (P<0.01 for both comparisons). Beta-adrenergic antagonists and clonidine reduce autonomic manifestations of withdrawal but have no known anticonvulsant activity. Symptoms of early withdrawal or impeding delirium may be masked by propranolol. Centrally acting alpha-adrenergic agonists such as clonidine ameliorate symptoms in patients with mild-to-moderate withdrawal but probably do not reduce delirium or seizures. They may be used in conjunction with benzodiazepines in patients with coexisting conditions such as coronary artery disease. For benzodiazepine withdrawal, however, they have not found much use even when low therapeutic doses are discontinued.
Although phenytoin has no primary role in the treatment of alcohol-withdrawal symptoms, other anticonvulsant agents, such as carbamazepine, have been in clinical use for this purpose for three decades. Carbamazepine is superior to placebo and equal in efficacy to phenobarbital and oxazepam for patients with mild-to-moderate withdrawal symptoms. Carbamazepine has no significant toxic effects on the blood or the liver when used in seven-day protocols for alcohol withdrawal; it reduces emotional distress better and permits a faster return to work than does oxazepam. Carbamazepine has well-documented anticonvulsant activity and prevents alcohol-withdrawal seizures in studies in animals, but data from trials in humans are limited. It does not potentiate the central nervous system and respiratory depression caused by alcohol, does not inhibit memory (which occurs with even small doses of benzodiazepines), and has no potential for abuse. However, dizziness, vomiting, and nausea are common side effects, particularly at the initial dose of 800 mg per day. Carbamazepine has not been evaluated for treating delirium tremens.
In a recent study of 136 patients with alcohol-withdrawal symptoms, patients treated with carbamazepine had fewer protracted symptoms than did those receiving lorazepam, five days after the medications were stopped. This difference in symptom levels persisted for a week. Furthermore, relapse to alcohol use during a follow-up period of three months was less common in the carbamazepine group. Adverse effects such as dizziness and incoordination were also less common in the carbamazepine group (7 percent vs. 20 percent). Some early studies have suggested that carbamazepine is efficacious in patients undergoing benzodiazepine withdrawal, but the data are less extensive than are those for alcohol.
Valproate may also reduce symptoms of alcohol withdrawal. Data have been reported for a total of approximately 2500 patients. The trials have generally been open-label, and not all have reported seizure rates. Two double-blind, randomized studies have been published, however. In these trials, patients treated with 1000 to 1200 mg of valproate for four to seven days had fewer seizures, dropped out less frequently, had less severe withdrawal symptoms, and used less oxazepam than did controls treated with placebo or carbamazepine.
No controlled trials have been published on the use of gabapentin for alcohol-withdrawal symptoms.
From the Departments of Psychiatry (T.R.K.) and Medicine (P.G.O.), Yale University School of Medicine, New Haven, Conn.; and Veterans Affairs Connecticut Healthcare System, West Haven, Conn. (T.R.K.).
Thomas R. Kosten, M.D., and Patrick G. O’Connor, M.D., M.P.H.