Ethanol produces its toxic effects (relaxation, euphoria, disinhibition, slurred speech, ataxia, sedation, stupor, coma, and respiratory depression; see Chapter 119) through modulation of a variety of neuroreceptors and ion channels [4]. It acts, in part, by interacting with the γ-aminobutyric acid (GABA_A) receptor complex, potentiating inhibitory GABAergic receptor function by inducing chloride flux through the chloride channels of the receptor complex [5]. Ethanol also inhibits excitatory N-methyl-D-aspartate (NMDA) glutamate receptor function, contributing to impaired cognition and blackouts associated with chronic ethanol use [6]. Inhibition of NMDA receptor function changes intracellular calcium levels and, as a result, affects cell-signaling cascades, including phosphorylation [7]. Other neurotransmitter systems affected by ethanol include dopamine and serotonin [8]. Ethanol has been found to affect 5-hydroxytryptamine receptor function by increasing the potency with which agonists bind this receptor [4]. Ethanol consumption may also result in an increase in endogenous opiates, contributing to its euphoric effect [9]. In addition, ethanol may exert its effect by altering the lipid matrix of cell membranes [10]. Although it was not recognized until the 1950s that delirium was a manifestation of ethanol withdrawal rather than toxicity, it is now clear that the hallmarks of ethanol and other sedative–hypnotic intoxication are distinctly different from the manifestations of withdrawal from these agents [11,12].

Ethanol withdrawal produces a hyperadrenergic state characterized by intense sympathetic nervous system activation. This may be due in part to compensatory central nervous system (CNS) mechanisms that counteract the depressant effects of ethanol intoxication. During withdrawal, these compensatory mechanisms are unopposed, resulting in increased neural stimulation [13]. In support of this theory, elevated levels of plasma and urinary catecholamines have been associated with tachycardia, elevated blood pressure, and tremors observed in withdrawing patients [14]. A decrease in the inhibitory activity of presynaptic α₂-receptors has been demonstrated and may explain, in part, the increase in norepinephrine levels [15]. In addition, an increase in β-adrenergic receptors during withdrawal has been demonstrated [16]. One study showed an increase in plasma levels of the dopamine metabolite homovanillic acid in patients presenting with delirium tremens [17].

Compensatory changes in number and function of inhibitory GABA_A receptors and excitatory NMDA glutamate receptors during chronic ethanol use may contribute to the CNS stimulation brought on by the cessation of ethanol. The abrupt withdrawal of the GABA-potentiating effects of ethanol leads to a disinhibition of neural pathways in the CNS [18]. During withdrawal, ethanol’s enhancing effect on chloride flux is lost, resulting in a decrease in GABAergic functioning. Tachycardia, diaphoresis, tremors, anxiety, and seizures have been associated with this reduction in GABA-induced chloride flux [19]. Upregulation in NMDA glutamate receptors and changes in their receptor subunit composition increases calcium flux through these receptors [20]. This likely contributes to the excitotoxic neuronal cell death associated with ethanol withdrawal [21]. Repeated episodes of withdrawal increase the propensity for ethanol withdrawal seizures through altered GABA_A and NMDA receptor function [22,23]. Because NMDA receptors mediate dopaminergic transmission, the increased NMDA receptor function that occurs during withdrawal may also lead to decreased dopaminergic and serotonergic transmission, contributing to alcohol craving [7].

Ethanol withdrawal occurs when a dependent patient suddenly stops drinking or drinks at a slower rate than previously. In either case, a significant drop in the serum ethanol level occurs. In chronic alcoholics, signs of withdrawal are commonly present even when their serum ethanol concentrations are higher than 100 mg per dL [24]. Patients admitted to the ICU with ethanol withdrawal often have a significant underlying disease that has led to an inability to maintain an ethanol intake adequate to prevent withdrawal. Alcoholic gastritis, hepatitis, pancreatitis, and pneumonia commonly precipitate decreased ethanol use and withdrawal. These patients typically present to the hospital after 24 to 48 hours of abdominal pain or fever and may be tremulous or have had a withdrawal seizure. Another type of ICU patient prone to withdrawal is one who has continued to imbibe ethanol nearly to the moment of arrival at the hospital. Intoxicated patients are prone to experience traumatic events and arrive in the operating room, recovery room, or ICU still intoxicated. A history of ethanol abuse or previous withdrawal may not be available in the postoperative or intubated patient when initial signs of withdrawal occur. Failure to recognize ethanol withdrawal in the seriously ill or injured patient may lead to prolonged complications [13].

Ethanol withdrawal results in a variety of signs and symptoms that vary in severity and duration. In their landmark article, Victor and Adams [12] described withdrawal as a tremulous–hallucinating–epileptic–delirious state. Although this description is often used to divide ethanol withdrawal syndrome into four stages, it is important to remember that the various manifestations of ethanol withdrawal form a progressive continuum of severity. A patient in ethanol withdrawal may exhibit one or more of these manifestations. The sequence of clinical events may be inconsistent. The severity of the withdrawal is often dose-dependent, with more severe reactions associated with heavier and longer periods of drinking [24]. It has been suggested that repeated withdrawal episodes produce a kindling effect, such that each subsequent withdrawal elicits increasingly more severe reactions [15,23,24].

Tremulousness and seizures are the most common clinical manifestations of ethanol withdrawal. They tend to occur early and are generally considered mild-to-moderate ethanol withdrawal symptoms. Delirium tremens is a late manifestation of ethanol withdrawal and constitutes the most serious clinical presentation. Although dramatic and life threatening, delirium tremens is but one aspect of ethanol withdrawal and affects 5% of withdrawal patients [25].

Mild ethanol withdrawal is usually characterized by a period of acute tremulousness (the “shakes”). It begins 6 to 8 hours after a reduction in ethanol intake [24,26]. Patients usually complain of tremulousness, nausea, vomiting, anorexia, anxiety, and insomnia. Physical examination reveals evidence of mild CNS and autonomic hyperactivity, which includes tachycardia, mild hypertension, hyperreflexia, irritability, and a resting tremor. Occasionally, significant tremor may not be appreciated despite the patient’s complaint of feeling “shaky inside.” Despite the fact that patients in delirium tremens have evidence of significant disorientation, this milder form of withdrawal is characterized by a clear sensorium,
although the patient may have a minor disorientation to time. Symptoms of mild ethanol withdrawal usually peak between 24 and 36 hours, and 75% to 80% of these patients recover uneventfully in a few days. Approximately 20% to 25% of patients presenting with mild ethanol withdrawal progress to serious withdrawal manifestations, which include seizures, hallucinations, or delirium tremens. However, it is impossible to reliably predict which patients will deteriorate [24].

Seizures that occur in alcoholics may or may not be due to ethanol withdrawal. Although ethanol withdrawal accounts for many of these seizures, other common causes include preexisting idiopathic and post-traumatic epilepsy [11,12]. Other complications of ethanol abuse not necessarily associated with withdrawal, such as hypoglycemia, hypomagnesemia, and hyponatremia, may also precipitate seizure activity [27]. Ethanol intoxication itself is not thought to be proconvulsant [28]. Alcoholic patients with a history of epilepsy appear to have a greater incidence of seizures than those without a preexisting seizure disorder. Failure to comply with anticonvulsant regimens may, in part, account for this. Brief abstinence (even overnight) may also lower the seizure threshold sufficiently to provoke seizures in susceptible patients. Because management strategies differ depending on whether the patient has a history of previous seizure disorder unrelated to ethanol withdrawal, differentiating between them becomes important [29].

Early studies showed that as many as 25% to 33% of patients in ethanol withdrawal demonstrate seizure activity [11,12]. Most ethanol withdrawal seizures (“rum fits”) occur between 7 and 48 hours after cessation or relative abstinence from drinking [30]. Mild-to-moderate signs of withdrawal may precede the seizures, or the seizure may herald the onset of ethanol withdrawal. They are short, generalized, tonic–clonic seizures, 40% of which are limited to a single isolated event. Often a short burst of two to six seizures with normal sensorium between seizures occurs over a few hours. Patients with ethanol withdrawal seizures usually have normal baseline electroencephalograms, in contrast to those with underlying seizure disorders. Status epilepticus or recurrent seizure activity lasting longer than 6 hours is distinctly uncommon in ethanol withdrawal and suggests another diagnosis [31].

Ethanol-related seizures may foreshadow the development of delirium tremens. In one series of patients with ethanol withdrawal seizures, delirium tremens developed in 33% [32]. In some patients, postictal confusion blended imperceptibly into delirium tremens. Approximately 40% of patients in whom delirium tremens subsequently developed exhibited an initial clearing followed by the onset of delirium tremens 12 hours to 5 days later.

Disordered perceptions characterized by hallucinations and nightmares were noted in 25% of tremulous patients in early withdrawal by Victor and Adams [12]. The hallucinations were predominantly visual in nature, auditory only in 20% of cases, and rarely tactile or olfactory. Commonly described visual phenomena in this setting may include the graphic depiction of bugs crawling on the walls or bed [32].

A subset of hallucinating patients does not demonstrate tremulousness or other signs of sympathetic hyperactivity. Known as acute alcoholic hallucinosis, this uncommon clinical presentation (occurring in 2% of the patients of Victor and Adams) is a distinct manifestation of ethanol withdrawal that usually begins within 8 to 48 hours of cessation of drinking [12]. It is characterized by disabling auditory hallucinations, often of a persecutory nature. These patients display no evidence of formal thought disorder, have no personal or family history of schizophrenia, and are usually oriented to person and place. In most cases, symptoms last for 1 to 6 days, although they may persist for months and come to resemble chronic paranoid schizophrenia. These symptoms usually respond to therapy with cross-tolerant agents such as benzodiazepines [33].

Delirium tremens is characterized by a significant alteration of sensorium associated with dramatic autonomic and CNS hyperactivity. Only 5% of patients who exhibit any of the previously discussed manifestations of ethanol withdrawal progress to delirium tremens. Delirium tremens appears to be more common in patients with a history of significant withdrawal and a long history of ethanol use. Patients in whom delirium tremens develops may not have demonstrated earlier signs of withdrawal. Other patients who have had ethanol withdrawal seizures or hallucinations may deceptively improve before the onset of delirium tremens, which is rarely seen before 48 to 72 hours after cessation or reduction in drinking and may be delayed for as long as 5 to 14 days [12,26]. These patients are truly delirious, exhibiting disorientation, global confusion, hallucinations, and delusions. Speech is unintelligible. Psychomotor disturbances, such as picking at bedclothes, significant restlessness, and agitation, are common and often require the use of physical restraints. Autonomic disturbances, such as tachycardia, hypertension, tachypnea, hyperpyrexia, diaphoresis, and mydriasis, are present. Cardiac dysrhythmias may also occur [34]. Seizures rarely occur during delirium tremens [26]. Concomitant illness, trauma, seizures, or therapeutic drugs may mask or modify the typical presentation.

Mortality for delirium tremens varies with the presence of underlying disease. Higher mortality is associated with superimposed pneumonia, meningitis, pancreatitis, gastrointestinal bleeding, and major trauma. In the untreated patient without serious coexisting medical disease, mortality usually is a consequence of severe dehydration or hyperthermia, or both, precipitating cardiovascular collapse [35]. Before adequate therapeutic agents were available, a mortality rate of 24% to 35% was cited in the literature [36]. This had decreased to 5% to 10% with the use of barbiturates and paraldehyde [37]. The use of benzodiazepines and intensive supportive care and earlier recognition of withdrawal should further reduce mortality in the absence of significant underlying disease [18].

The differential diagnosis of ethanol withdrawal includes other causes of a hyperadrenergic state. Most importantly, ethanol-related hypoglycemia needs to be differentiated from withdrawal. Clinically, these two conditions may appear remarkably similar, although only hypoglycemia rapidly improves after intravenous (IV) glucose administration [38].

Intoxication with sympathomimetic agents such as cocaine or amphetamine shares many features with ethanol withdrawal, including signs and symptoms of adrenergic excess. Overdose of monamine oxidase inhibitors, phencyclidine, anticholingergic agents, and lithium, as well as neuroleptic malignant syndrome and serotonin syndrome, may all demonstrate marked agitation and confusion [39]. In the elderly patient, almost any therapeutic drug may be associated with delirium [40]. Withdrawal from other sedative–hypnotics, such as benzodiazepines, barbiturates, GHB, and baclofen, may precipitate a delirium-tremens-like state (see following discussion).

Significant underlying metabolic, traumatic, and infectious disorders should be excluded in the patient with altered mental status associated with ethanol withdrawal. Differentiation may require lumbar puncture, laboratory tests, and computed tomographic scan. These include CNS emergencies, such as intracranial bleeds, meningitis, and encephalitis; metabolic causes, including hypoxia, hypercarbia, sepsis, thiamine deficiency, and sodium and calcium abnormalities; and endocrine disturbances, such as thyroid storm and pheochromocytoma. Distinguishing between delirium tremens and hepatic encephalopathy may be difficult, especially because these conditions often coexist [41].

A successful strategy in treating ethanol withdrawal must address several key goals: alleviation of symptoms, prevention of progression of withdrawal to a more serious stage, avoidance of complications, treatment of coexisting medical problems, and planning for long-term rehabilitation and drug independence [26]. Initial management involves securing the airway, breathing, and circulation. Patients with an altered level of consciousness require oxygen and IV administration of at least 100 mg thiamine and 50 g glucose. The latter two substrates are particularly important, as Wernicke’s encephalopathy and hypoglycemia may be confused or coexist with ethanol withdrawal. Severely agitated patients may initially require physical restraints to prevent injury and facilitate sedation. Prolonged use of physical restraints without adequate sedation, however, may be detrimental because agitated patients quite often continue to struggle against their restraints. Such activity perpetuates the risk for hyperthermia, muscle destruction, and resultant myoglobinuric renal failure. Volume resuscitation, correction of electrolyte abnormalities, and vigilance in the diagnosis and treatment of coexisting medical and surgical disorders are vital in reducing morbidity and mortality in the patient with delirium tremens [37,42].

Achievement of adequate sedation is the cornerstone of successful treatment of ethanol withdrawal [43]. Sedation alleviates the excitatory manifestations of withdrawal, prevents progression to delirium tremens, and prevents common complications of agitation, including trauma, rhabdomyolysis, and hyperthermia. Although many agents have been used over the years, benzodiazepines have proved the most effective [43,44,45,46,47]. Benzodiazepines, unlike the neuroleptics, are cross-tolerant with ethanol and function as a replacement drug for the short-acting ethanol, increasing the affinity of GABA for the GABA_A receptor [48].

Diazepam (Valium), chlordiazepoxide (Librium), and lorazepam (Ativan) are the most commonly used parenteral agents. All three drugs can easily be given intravenously to facilitate rapid sedation and titration of effect. Of these agents, only lorazepam has reliable intramuscular (IM) absorption [24,49]. Diazepam and chlordiazepoxide are long-acting agents with active metabolites that prolong their therapeutic effect, avoiding the need for frequent dosing that is associated with shorter-acting agents. Lorazepam, a shorter-acting agent, has no active metabolites and is better tolerated in the elderly and in patients with hepatic dysfunction, producing less sedation. Prolonged therapy (e.g., > 1 month) with high-dose IV lorazepam, however, has also been associated with acute tubular necrosis secondary to the polyethylene glycol used as the lorazepam diluent [50]. Continuous IV infusion of midazolam, a short-acting agent, has also been recommended in the treatment of delirium tremens [51]. However, this approach requires more vigilant monitoring and does not provide the advantages of a long-acting benzodiazepine that is gradually eliminated over several days. Midazolam infusion is also considerably more expensive than therapy with longer-acting agents [52].

The benzodiazepine of choice in the treatment of ethanol withdrawal remains controversial [53,54]. Although many investigators have suggested that lorazepam may be the preferred agent [13,37,55], long-acting benzodiazepines such as diazepam may be more effective in preventing ethanol withdrawal seizures and contributing to smoother withdrawal with less breakthrough or rebound symptoms [56,57].

Symptom-triggered benzodiazepine treatment for alcohol withdrawal is strongly encouraged [58]. The Clinical Institute Withdrawal Assessment for Alcohol (CIWA-A) scale is a reliable, validated scale to assess severity of alcohol withdrawal so treatment can be appropriately titrated and individualized. It includes subjective parameters such as anxiety, auditory and visual disturbances, headache, and nausea as well as objective parameters such as tremor, sweating, agitation, and clouding of sensorium. [59] The dose of benzodiazepines needed to achieve adequate sedation varies considerably depending on the patient’s tolerance. Although oral therapy may be appropriate in patients with mild withdrawal, those with significant signs of withdrawal require IV treatment. Therapy with an IV benzodiazepine is titrated to the patient’s needs by the use of frequent boluses until withdrawal symptoms subside. Using such a front-loading technique helps avoid undertreatment or excessive sedation [60,61]. For example, 5 to 20 mg of diazepam can be administered to the patient every 5 minutes until he or she is quietly asleep but can be easily awakened. Initial safe titration of benzodiazepines requires continual reevaluation by an observer at the bedside. In patients with moderate withdrawal symptoms, a study showed that using a symptom-triggered approach, instead of a fixed-schedule approach, resulted in the administration of less total medication and fewer hours of medication (9 hours vs. 68 hours) [62,63]. A recent study in a surgical ICU demonstrated that this symptom-orientated bolus-titrated approach decreases the severity and duration of alcohol withdrawal symptoms, resulting in reduced medication requirements, fewer days of ventilation, lower incidence of pneumonia, and shorter ICU stay [64].

Failure to obtain adequate sedation with standard doses of the chosen agent should not prompt a switch to an alternative benzodiazepine. Some patients require very high doses to achieve sedation; cases of patients receiving more than 1,000 mg diazepam during 24 hours have been reported [62]. Recent research into GABA receptor physiology suggests that resistance to large doses of benzodiazepines in some patients with alcohol withdrawal may be due to alterations in GABA_A receptor subunits [65]. Chronic ethanol exposure produces upregulation of GABA_A receptor α₄ subunits that are insensitive to benzodiazepines, and downregulation of benzodiazepine-sensitive α₁ subunits. If a patient with severe alcohol withdrawal does not respond to large doses of a benzodiazepine, administration of an alternative agent may be warranted. A drug such as a barbiturate, which acts on the GABA_A receptor regardless of its specific α subunit composition, would be appropriate.

Recent research also suggests that changes in NMDA glutamate receptor physiology may be important in both clinical signs and symptoms of ethanol withdrawal and the excitotoxic neuronal cell death that may occur. In animal studies, NMDA receptor antagonists may attenuate the development of ethanol dependence if administered concomitantly, and may prevent withdrawal seizures and neuronal excitotoxicity if given during periods of withdrawal [20]. Patients who are refractory to high dose GABA_A agonists may potentially benefit from addressing the glutaminergic as well as the GABergic manifestations of ethanol withdrawal. Options here are limited, but drugs such as propofol, which possess both GABA agonist and NMDA antagonist properties, may be particularly helpful.

Adequate early treatment with benzodiazepines usually suppresses significant manifestations of withdrawal and prevents progression to delirium tremens. If delirium tremens is already manifest, sedation with a benzodiazepine does not completely reverse mental status abnormalities. This may be a consequence of the incomplete cross-tolerance of benzodiazepine with ethanol or perhaps the lack of immediate reversibility of some of the CNS effects of withdrawal [66].

Barbiturates, particularly intermediate and long-acting agents such as pentobarbital and phenobarbital, are an alternative class of cross-tolerant sedative–hypnotic agents that can be used in the treatment of ethanol withdrawal [67]. Although excess sedation and a greater tendency to produce respiratory
depression may be more of a concern with barbiturates as compared with benzodiazepines, the drugs are still titrated until the patient is quietly asleep but easily awakened [68]. Phenobarbital dosages more than 20 mg per kg may be required. Withdrawal patients with idiopathic or post-traumatic epilepsy who require maintenance anticonvulsant levels may particularly benefit from this alternative strategy. Phenobarbital may also be useful for those patients who are resistant to benzodiazepine therapy.

Propofol, a sedative–hypnotic agent used for induction and maintenance of anesthesia, has been used successfully for treatment of severe ethanol withdrawal that is resistant to large doses of benzodiazepines (> 1,000 mg per day) [69,70,71]. Like ethanol, it acts as an agonist at the GABA_A receptor and also inhibits the NMDA receptor. Its onset of action is rapid, it is easily titratable, and sedative effects wear off quickly after short-term use (< 72 hours). The fact that it addresses the glutaminergic as well as the GABAergic aspects of ethanol withdrawal may be one reason for its increased apparent effectiveness in patients resistant to standard therapy with benzodiazepines. Disadvantages of its use include high cost and prolonged sedation when it is used for extended periods [72]. No controlled trials have compared propofol and benzodiazepines for treatment of ethanol withdrawal.

Intravenous and oral ethanol have been used to suppress withdrawal and continue to be used by some medical practitioners, especially surgeons [73,74]. However, IV ethanol intensifies the biochemical abnormalities associated with ethanol metabolism, shifting energy production toward lactate and ketogenesis [75]. The use of ethanol in the treatment of ethanol withdrawal is not recommended [76].

The use of phenothiazines and butyrophenones to treat ethanol withdrawal has been associated with excessive fatalities [42,75,77]. These agents have been shown to lower the seizure threshold, induce hypotension, impair thermoregulation, and precipitate dystonic reactions [78,79,80]. These drugs have no role in the management of sedative–hypnotic withdrawal [81].