Alcohol Withdrawal Seizures
Alcohol withdrawal seizures are perhaps the most rigorously studied complication of AWS. Although alcohol withdrawal seizures are generally self-limited, benzodiazepines are preferred for patients with persistent or recurrent alcohol withdrawal seizures. In a randomized, placebo-controlled trial of 229 subjects with alcohol withdrawal seizures, 2 mg of intramuscular lorazepam reduced the risk of recurrent seizure from 24% to 3% (P < 0.001) at 6 hours, and the need for hospital admission from 42% to 29% (P = 0.0222).30 However, whether this interrupts the natural history of progression to DTs is not known. There is no role for phenytoin in either treatment or prevention of alcohol withdrawal seizures. In multiple trials, phenytoin was ineffective in preventing recurrence of alcohol withdrawal seizure.1,18,82 The most likely explanation for the failure of phenytoin is its inability to regulate GABA or NMDA receptors, the principle mediators of the development of seizures in alcohol withdrawal. One exception to this lack of usefulness occurs in the alcoholic patient with a nonalcohol withdrawal-mediated seizure, or a history of underlying seizure disorders.
In the early stages of alcohol withdrawal, many patients are able to self-medicate with additional alcohol consumption. Among those who seek medical attention, many patients with AWS can be safely managed as outpatients. Outpatient management has significant cost savings with little effect on treatment outcome when the patient is sufficiently stable to tolerate this option.43 In one study, patients who exhibited a current lack of intoxication, no history of either DTs or alcohol withdrawal seizures, no comorbid psychiatric or medical disorders, and a CIWA-Ar score of less than 8 were safely managed as outpatients.2 Patients not meeting these criteria were referred to inpatient detoxification centers or medical units, depending on the severity of withdrawal and other comorbid conditions.
For all patients with AWS, the initial stages of therapy remain the same, and should include a thorough assessment to identify coexisting medical, psychiatric, or toxicologic disorders. In particular, an assessment for central nervous system trauma and infection should include the appropriate use of computed tomography and lumbar puncture. Patients with altered cognition or an elevated body temperature should receive antibiotics as appropriate, pending the results of a lumbar puncture. In concert with this approach, adequate supportive care should be instituted, with attention to identifying treatable causes of abnormal vital signs and clinical findings.
Chronic alcohol consumption leads to severe vitamin and nutritional deficiencies and electrolytes disturbances that should be corrected.25,49 Specifically, parenteral thiamine should be given to all patients to prevent the development of Wernicke encephalopathy. It is generally suggested that thiamine should be given prior to the administration of dextrose to prevent precipitation of Wernicke encephalopathy.73,100 Although there is little evidence to support this approach, the administration of thiamine (see Antidote in Depth A25: Thiamine Hydrochloride) and dextrose together is a reasonable practice. Because there is typically a high incidence of intravascular volume depletion among alcoholics, all patients should receive adequate volume resuscitation. Of 39 deaths between 1915 and 1936 attributed to DTs in which volume status was recorded, all subjects were volume depleted.75 Finally, for patients with AWS, particularly if severe, prevention of nosocomial complications is paramount for reducing hospital stay. Currently, in addition to adequate volume replacement, we recommend that all patients be kept with the head of the bed elevated to prevent pulmonary aspiration, and that deep vein thrombosis prophylaxis be given if the patient is bed bound for an extended period.
The association of severe alcohol withdrawal with severe psychomotor agitation led to early use of sedative-hypnotics. One of the first randomized trials compared chlorpromazine to paraldehyde. In both study arms there was a 0% mortality, suggesting equivalency of the two treatments.34 Over the ensuing years, numerous trials documented similar efficacy between paraldehyde, benzodiazepines, and antipsychotics.16,34,92 However, in a landmark study, 547 patients were randomized to 1 of 4 drugs (chlordiazepoxide, chlorpromazine, hydroxyzine, and thiamine) or to placebo for the treatment of alcohol withdrawal.55 Patients receiving chlordiazepoxide had the lowest incidence of both DTs and alcohol withdrawal seizures, establishing benzodiazepines as a first-line agent for treatment of AWS. Of note, use of chlorpromazine, an antipsychotic, was associated with a significant increase in the incidence of seizures in both humans55 and animal models.10
Since this study, numerous trials have compared different routes of administration among various sedative-hypnotics, both to each other and to placebo. Because of the historical use noted above, chlordiazepoxide remains widely used in outpatient and inpatient detoxification clinics. Oral benzodiazepine administration is generally effective in patients with early or mild AWS, although initial rapid titration with an intravenous regimen may be more efficient. Benzodiazepines administered intravenously have a rapid onset of action, and have long displaced paraldehyde as the sedative-hypnotic of choice for acute control. Among the benzodiazepines, diazepam offers the most rapid time to peak clinical effects, which limits oversedation that may occur following the administration of drugs with slower onset to the peak drug effect, such as lorazepam. Because of the delayed peak clinical effect of lorazepam of approximately 10–20 minutes, several doses may be administered in rapid succession with little clinical effect, followed by the appearance of oversedation from the cumulative doses. Midazolam may be administered intramuscularly if intravenous access is not available, but intramuscular injection significantly delays the time to both onset and peak clinical effect. Although no significant differences are observed between benzodiazepines and barbiturates in terms of mortality or the duration of delirium, the improved pharmacokinetic profile and ease of administration favor benzodiazepines as the preferred initial agent (see A24: Benzodiazepines).2,63
Other pharmacokinetic factors and experience confirm that diazepam is preferred for initial intravenous use in patients with moderate to severe AWS. Diazepam has a long half-life [43± 13 hours] and has active metabolites (desmethyldiazepam and oxazepam). The prolonged half-life (48–72 hours) of desmethyldiazepam further extends the effective duration of action of the initial dose of diazepam.107 A retrospective review reported that the use of a single benzodiazepine rather than multiple benzodiazepines was a marker for treatment success in surgical patients experiencing alcohol withdrawal during surgical admission.77 These data suggest that it is more important to rapidly sedate the patient with adequate doses of a single benzodiazepine than to use multiple benzodiazepines in hopes of finding an effective regimen. Finally, it should be noted that in patients with advanced liver disease, the use of diazepam may result in a very prolonged period of sedation because of impaired clearance of the parent compound and its metabolites. Consequently, in these patients, a benzodiazepine without active metabolites, such as lorazepam, may be a better drug.
The initial management of patients with AWS/DTs should include rapid titration with intravenous benzodiazepine to achieve sedation. The goal of therapy is to have the patient sedated but breathing spontaneously, with normal vital signs. Although normalization of vital signs is not a mandatory therapeutic endpoint, abnormal vital signs despite adequate sedation should prompt a search for comorbidities. In many patients, complete sedation with diazepam may allow for autotitration; that is, as the AWS resolves, the blood concentrations of diazepam and desmethyldiazepam decrease, allowing gradual recovery. In practicality, most patients need periodic redosing with diazepam to maintain adequate sedation. This is particularly important in patients with AWS with an elevated blood alcohol concentration.
Multiple studies now suggest that if additional doses are required, they should be administered based on symptoms ("symptom triggered"), as opposed to a fixed dosing schedule. In two randomized controlled trials, administration of benzodiazepine in a symptom-triggered fashion reduced both the total amount of benzodiazepine and the duration of treatment.26,86 In these trials, benzodiazepines were administered every hour as long as the CIWA-Ar score remained greater than 8–10. In both trials, symptom-triggered therapy resulted in a 4- to 6-fold reduction in the duration of therapy and a 4- to 5-fold reduction in the total amount of benzodiazepine administered, with no increase in withdrawal seizures or adverse events.26,86 Symptom-triggered doses in patients with moderate or severe AWS should be diazepam 10–20 mg IV or lorazepam 2–4 mg.26,86 For less-symptomatic patients, oral chlordiazepoxide 50–100 mg should be administered. However, it is important to note that the decision to treat in the symptom-triggered group was made based on CIWA-Ar score (usually >8), which demonstrates the usefulness of standardized scoring and evaluation tools. It should also be noted that in both of these trials, patients had very mild withdrawal symptoms, with mean CIWA-Ar scores of 9–11. While experience suggests that this same regimen is also effective in patients with serious withdrawal and/or DTs it has not been validated in this population. Furthermore, it must be emphasized that protocolized use of the CIWA-Ar score is dependent on a significant history of recent heavy drinking and a communicative subject. A prospective analysis of complex medical/surgical patients enrolled in a CIWA-Ar protocol suggested more than 50% patients failed to meet these criteria, leading to unnecessary treatment.44
Resistant Alcohol Withdrawal and Delirium Tremens
There is a subgroup of patients with AWS who require very large doses of diazepam, or another comparable drug to achieve initial sedation.40,68,78,105 This same group often has exceedingly high benzodiazepine requirements to maintain this level of sedation. Subjects with resistant AWS and DTs may have benzodiazepine requirements that exceed 2600 mg of diazepam within the first 24 hours, and generally require admission to an intensive care or stepdown unit.105 Patients admitted to the Bellevue hospital medical ICU for resistant alcohol withdrawal had very high diazepam requirements, with a mean of 234 mg (range: 10–1490 mg) required in the first 24 hours, and individual doses of diazepam that often exceeded 100 mg, to control their agitation. At Bellevue, these patients comprise approximately 5% of all ICU admissions, with nearly 40% of patients requiring mechanical ventilation and a mean ICU length of stay of 5.7 days.36
The approach to the management of resistant AWS depends on several factors, including the availability of an intensive care unit bed. In the ICU, despite the perception of failure of high benzodiazepine requirements, we favor administration of benzodiazepines in a symptom-triggered fashion. This approach was confirmed in a study of patients who developed AWS postoperatively in the ICU.88 In this study, a symptom-triggered strategy resulted in a shorter length of stay and a lower incidence of mechanical ventilation than did continuous infusion of midazolam.88 Patients who receive this therapy generally respond to bolus doses of diazepam, which results in a brief period of sedation followed by recrudescence of their AWS. However, the dose range and drug required to achieve this may be dramatically different than what is observed in subjects with nonresistant AWS. In a recent study, use of escalating bolus doses of diazepam, up to 200 mg as an individual dose, combined with phenobarbital in subjects with continued benzodiazepine resistance (defined as the requirement for bolus doses more frequently than every hour), reduced the need for mechanical ventilation by nearly 50%.36 In non-ICU settings, the ability to administer frequent intravenous doses of diazepam is limited, and the use of intravenous infusions with secondary sedative agents may be more practical.
In instances of extreme benzodiazepine resistance, patients often receive a second GABAergic drug because of "failure" of benzodiazepine therapy. Phenobarbital, given in combination with a benzodiazepine, in intravenous doses of up to 260 mg, is a reasonable choice.36 Caution is required to avoid stacking doses of phenobarbital, as the onset of clinical effect takes approximately 20–40 minutes.45,53 Alternatively, propofol in standard doses may be administered. Although propofol has a rapid onset, it is difficult to titrate, and high dose or long-term use is associated with profound metabolic consequences.22,23,74 However, in a recent observational study, propofol was safely administered to 21 subjects intubated for severe benzodiazepine resistant DTs.36 The main drawback to the use of these drugs is their narrow therapeutic—toxic index, with the potential for profound respiratory depression. This is especially true for propofol, which should generally only be used in the setting of mechanical ventilation. Both of these agents can act synergistically with benzodiazepines to enhance GABA-induced chloride channel opening. In addition, propofol uniquely antagonizes NMDA receptors, thus reducing the excitatory component of AWS.27,41,51
Ethanol consumption is a common and effective means by which alcoholics can self-medicate to treat and/or prevent mild alcohol withdrawal. Consequently, some hospitals still administer ethanol for either prophylaxis and/or treatment of AWS. In one survey, 72% of 122 hospitals surveyed had administered either IV or oral ethanol for these indications.9 Despite its widespread use, little randomized controlled data support its use. In one trial, 39 trauma patients without liver or CNS disease were successfully treated with 10% ethanol infusion for treatment of presumed AWS.24 Conversely, IV ethanol was no more effective than flunitrazepam in one trial and inferior to diazepam in another in the prevention of AWS in postoperative surgical patients.31,101 Although the authors did not report any adverse effects in these trials, the necessity for frequent blood alcohol monitoring, unpredictable elimination kinetics, potential for significant hepatic complications, the postulated adverse effects of ethanol on wound healing, and the difficulty in safely administering this therapy make it inappropriate to recommend this regimen.29,48,73
Numerous studies have investigated the use of sympatholytics to control the autonomic symptoms of alcohol withdrawal. Both β-adrenergic antagonists and clonidine reduced blood pressure and heart rate in randomized, placebo-controlled trials.4,64,106 However, the inability of these xenobiotics to address the underlying pathophysiologic mechanism of AWS, and subsequently control the neurologic manifestations, makes them suboptimal as sole therapeutic xenobiotics. There are additional concerns that by altering the physiologic parameters that serve as classic markers for AWS severity, there is a risk of underadministering necessary amounts of benzodiazepines.106 This was observed in a randomized controlled trial of the central α-agonist lofexidine.56 Finally, while animal evidence suggests a potential role for dexmedetomidine for AWS, there are only scattered case reports documenting its use in humans.84,85 Consequently, we do not recommend using these drugs for the treatment of AWS until it becomes clear that other standard therapies have failed.
The theoretical benefits of magnesium supplementation are based both on the high prevalence of magnesium deficiency in alcoholics and its usefulness in preventing seizures in other disorders, including eclampsia.5,49 Furthermore, magnesium deficiency has many clinical similarities to AWS, clouding the differential diagnosis. Numerous studies have evaluated the efficacy of magnesium supplementation. However, in a randomized, placebo-controlled trial, intravenous magnesium sulfate had no effect on either severity of alcohol withdrawal or incidence of withdrawal seizures.103 Consequently, aside from repletion of electrolyte abnormalities, there is no indication for routine administration of magnesium for the treatment of AWS.
Carbamazepine has been used in multiple trials for treatment of mild AWS, more commonly in Europe where an intravenous preparation is available. In animal studies, carbamazepine increases both the central nervous system GABA concentrations and the seizure threshold in alcohol withdrawal models.20 In humans, carbamazepine is superior to placebo and equally efficacious as benzodiazepines for treatment of mild to moderate AWS in both inpatients and outpatients.7,70,71 Similar data have been obtained with valproic acid, which appears to have a benzodiazepine-sparing effect in patients with mild withdrawal.83 In contrast, a recent randomized, placebo-controlled study of the newer anticonvulsant oxcarbazepine showed no difference between this xenobiotic and placebo in inpatient detoxification.60 Consequently, while this class of drugs may be reasonably recommended as adjuncts, they should not be used as monotherapy for treatment of established AWS and DT.