Case example
A middle-aged man was admitted to the ICU for refractory alcohol withdrawal. Prior to arriving in the ICU he had been treated aggressively with an escalating regimen of IV diazepam, without any improvement. Upon arrival in the ICU he had impressive tremors but was not delirious.
He was given an initial dose of 260 mg IV phenobarbital followed by 130 mg IV Q30 minutes as needed. With each dose, his symptoms improved incrementally. After receiving about 1200 mg of phenobarbital his symptoms resolved, leaving him awake and calm. He was observed for a day prior to transfer out of the ICU, but required no additional treatment.
Introduction
Benzodiazepines have traditionally been the mainstay of treatment for alcohol withdrawal. This dates back to a RCT in 1969 which compared a benzodiazepine (chlordiazepoxide), antipsychotics, antihistamines, and thiamine. Unfortunately, there has never been an adequate RCT comparing a benzodiazepine versus barbiturate.
Recently there has been increasing recognition that phenobarbital has advantages compared to benzodiazepines. Phenobarbital has been shown to be beneficial both for initial up-front loading, and also for patients with symptoms refractory to benzodiazepines. As explored in a prior post, this led to a treatment strategy which started and ended with phenobarbital:
Over time, emerging evidence and clinical experience has led us to doubt whether benzodiazepines offered an advantage compared to phenobarbital monotherapy:
Advantages of phenobarbital monotherapy
Neuroscience: Phenobarbital is theoretically superior to benzodiazepines
Alcohol suppresses the brain via multiple mechanisms, including enhancement of inhibitory GABA receptors and suppression of excitatory glutaminergic receptors. The brain adapts to chronic alcoholism by down-regulatinginhibitory GABA receptors and up-regulatingexcitatory glutaminergic receptors (Rao 2015). Such adaptation allows alcoholics to survive with blood alcohol levels which would kill most people. Unfortunately, this also causes withdrawal: both down-regulation of inhibitory GABA receptors and up-regulation of excitatory glutamate receptors excites the brain.
Benzodiazepines stimulate inhibitory GABA receptors, which may improve alcohol withdrawal. However, low GABA activity is only part of the problem. In contrast, barbiturates have dual activity, simultaneously enhancing GABA activity and suppressing glutaminergic activity. This dual mechanism of action is well matched to the pathophysiology of alcohol withdrawal, making barbiturates theoretically superior to benzodiazepines.
Clinical experience: Barbiturates are more powerful than benzodiazepines
Synergistic activity on two of the most important neurotransmitter systems makes barbiturates more powerful than benzodiazepines. This power is observed in other neurologic disorders as well (e.g. seizures are often refractory to benzodiazepines, but rarely refractory to barbiturates). Barbiturates are the hammer of neurotherapeutics: there is no positive symptom which will not respond to a barbiturate. Regarding alcohol withdrawal, it is widely recognized that a subset of patients with severe withdrawal will fail to respond to benzodiazepines, yet will subsequently respond to phenobarbital (as in the case above; Hack 2006).
Less delirium & paradoxical reactions?
Benzodiazepines are notorious for causing delirium. For example, one study of intubated ICU patients found that nearly all patients who received >20 mg of lorazepam developed delirium (Pandharipande 2006). Less commonly, lower doses of benzodiazepine elicit agitated delirium, known as a paradoxical reaction. As discussed last week, paradoxical reactions are more common in alcoholism, with a rate of ~2% in this population (Tae 2014). Is it possible that benzodiazepine-induced delirium and paradoxical reactions complicate the treatment of alcohol withdrawal without our being aware of these reactions?
Moore 2014 reported a retrospective case series describing the use of flumazenil to evaluate for benzodiazepine-induced delirium in patients undergoing alcohol withdrawal. Their general practice was to suspect benzodiazepine-induced delirium and trial flumazenil among patients with persistent confusion whose withdrawal seemed to have resolved (e.g. normal vital signs, no hyperreflexia). Among 74 patients in whom a response to flumazenil was recorded, 84% improved while only two patients experienced increased anxiety. This study suggests that much of the prolonged delirium observed in patients undergoing alcohol withdrawal may actually be benzodiazepine-induced delirium rather than alcohol withdrawal itself:
Unlike benzodiazepines, phenobarbital doesn't cause paradoxical reactions (Ives 1991). This may reflect phenobarbital's more balanced inhibitory effect on the brain via two neurotransmitters, which protects against disinhibition.
Simplified pharmacology: Choose one GABAergic medication
Benzodiazepines and barbiturates act synergistically on the GABA receptors (benzodiazepines increase the duration of channel opening, whereas barbiturates increase the frequency of channel opening). In some situations, this may cause patients to be unexpectedly sensitive to these medications. For example, a patient loaded with 2,000 mg of phenobarbital might be at risk from over-sedation if treated with a usual dose of benzodiazepine.
The clinical effect of phenobarbital alone is more predictable. As discussed previously, phenobarbital administration elicits very predictable serum level of drug (figure below). In the absence of confounding factors (e.g., benzodiazepines, other neurologic problems), the safe level of phenobarbital is well established. This dose-response relationship may be helpful when considering safe drug doses and monitoring patient response to therapy. Removing benzodiazepines from the picture simplifies this pharmacology and allows these relationships to be used more reliably.
 |
| Relationship between cumulative phenobarbital dose and plasma phenobarbital concentration among patients treated for alcohol withdrawal (Tangmose 2010). We have added green lines indicating the plasma therapeutic range for phenobarbital (64-172 micromol/L = 15-40 ug/ml), an orange line indicating the level at which mild signs of toxicity are usually noted such as ataxia and nystagmus (225 micromol/L = 50 ug/ml), and a red line indicating the lowest level which has been associated with stupor or coma (>280 micromol/L = 65 ug/ml)(Lee 2013). |
Seizure prophylaxis
Seizure is one of the most dangerous complications of alcohol withdrawal. Among the agents used to treat alcohol withdrawal, barbiturates probably have the greatest anti-epileptic activity. Combining this efficacy with the long duration of phenobarbital may provide patients with excellent seizure prophylaxis.
There is little evidence regarding the relative efficacy of benzodiazepines versus barbiturates for alcohol withdrawal seizure. One report found that the addition of primidone (a phenobarbital precursor) to chlordiazepoxide reduced the seizure rate (table below; Smith, 1976). This suggests that in alcohol withdrawal, as with other causes of seizure, barbiturates offer anti-epileptic activity above and beyond that of a long-acting benzodiazepine.
Improved pharmacokinetics with intravenous phenobarbital?
Historically, phenobarbital has often been given orally for the management of alcohol withdrawal. When administered orally, the time of onset is less reliable compared to intravenous administration. Thus, oral administration could increase the risk of administering several doses before the first doses are fully absorbed ("dose stacking"), with eventual overdose.
Theoretically, intravenous phenobarbital should allow greater control over pharmacokinetics, with avoidance of dose stacking and improved safety. It is unclear how much added benefit intravenous administration provides compared to oral phenobarbital. For critically ill patients, the intravenous route is typically used to err on the side of safety. However, for less ill patients without gastrointestinal problems, slower administration of oral phenobarbital has been demonstrated to be safe (Tangmose 2010).
Evidence regarding phenobarbital monotherapy
“In Denmark barbital, a long-acting barbiturate, has been the drug of choice in the treatment of DT for many years. It was introduced in the beginning of this century (Moller 1909). In following discussions the importance of repeated, often large doses, (that is, 0.5-1 gram), in the initial stage of the disorder was stressed. The aim of the treatment was to sedate the patient to such a degree that he fell asleep and then slept for several hours. After this “critical sleep” the symptoms often disappeared completely. The treatment as outlined in the first reports was found so favorable that barbital has been preferred by Danish psychiatrists for several decades” - Kramp 1978
Although barbiturate monotherapy is currently perceived as a new idea, it is actually a very old idea. In Denmark, barbiturate monotherapy was used for over a century with considerable success. In the United States, a survey of inpatient alcohol treatment programs performed in 1992 estimated that 11% of all patients received barbiturates (Saitz 1995). Unfortunately this practice largely pre-dated evidence-based medicine. Available studies are as follows:
Kramp 1978 This is the only available prospective RCT that compares barbiturate to benzodiazepine for patients with severe alcohol withdrawal. 91 patients were randomized to receive intramuscular diazepam vs. oral barbital (an early long-acting barbiturate). To preserve blinding of physicians and patients, all patients were treated simultaneously with intramuscular injections and oral medication, one of which was a placebo. Although this study has substantial methodological flaws, barbiturate was found to be superior among the patients with the most severe withdrawal symptoms (table below). Three patients in the diazepam group were initially refractory to therapy, leading clinicians to un-blind themselves. Overall these results are consistent with our current recognition that a subset of patients with severe withdrawal will be refractory to benzodiazepines.
Ives 1991 These authors describe a protocol involving a loading dose of 15 mg/kg phenobarbital followed by a fixed taper over a week. For breakthrough agitation, patients were allowed to receive low doses of lorazepam. Although little data is provided, they reported that this protocol was used successfully in over seventy patients at the University of North Carolina at Chapel Hill from 1982-1990. We are aware of success utilizing a similar protocol more recently at some hospitals in the Northeast USA.
Michaelsen 2010 This is a retrospective cohort study comparing outcomes from the treatment of delirium tremens at two hospitals in Denmark between 1998-2006. During this period one hospital (Rigshospitalet) used oral phenobarbital, whereas the other hospital (Bispebjerg) transitioned from oral phenobarbital to intravenous diazepam in 2002. There was no difference between diazepam and phenobarbital in terms of length of delirium tremens, mortality, or pneumonia (table below). A sub-population (9%) in the diazepam group failed treatment and required phenobarbital.
Transition to diazepam correlated with an increase in the rate of delirium associated with alcohol withdrawal at Bispebjerg (53 cases from 1998-2002 vs. 88 cases from 2002-2006). Perhaps this reflects some patients who developed benzodiazepine-induced delirium.
Hendey 2011 This is a RCT comparing phenobarbital vs. lorazepam for patients presenting to the emergency department with mild to moderate alcohol withdrawal. 44 patients were randomized to treatment with lorazepam or phenobarbital (260 mg IV followed by 130 mg IV PRN), with the majority subsequently discharged home. The two treatments were equivalent, although the study was underpowered.
Phenobarbital monotherapy: Nuts and bolts
Phenobarbital monotherapy is extremely simple, as it amounts to a dose-titration using a single medication. Phenobarbital has a half-life of about three days, so successive doses will accumulate in an additive fashion. The main requirement for using phenobarbital is patience, because it may take some time to reach an effective level. However, this time investment is worth it, because once a therapeutic level is reached, little additional therapy may be needed (the phenobarbital will gradually auto-titrate off, providing therapy for days).
The main decision which needs to be made is whether the patient qualifies for an initial loading dose of 10 mg/kg ideal body weight (1). For most patients who are initially presenting with alcohol withdrawal, 10 mg/kg of phenobarbital may be given safely (discussed previously here). This dose will produce a serum level of phenobarbital around 15 ug/ml, which by itself isn't nearly enough to cause respiratory suppression. Indeed, Ives 1991 used a 15 mg/kg loading dose divided over three doses. However, if the patient has other active neurologic issues or has received a significant amount of sedating medication (especially benzodiazepine), there is a possibility that this dose could cause excessive sedation. When in doubt, it is safer (albeit slower) to omit the loading dose and simply use an incremental IV phenobarbital titration:
As discussed above, we prefer to use intravenous phenobarbital initially until there is resolution of symptoms (which will typically occur in the emergency department or intensive care unit). Following improvement and transfer out of these locations, patients may be located in a psychiatric or medical ward, which is unfamiliar with the use of intravenous phenobarbital. In this situation, oral or intramuscular phenobarbital may be used for mild or moderate withdrawal symptoms (Tangmose 2010).
Phenobarbital and benzodiazepines act synergistically on the GABA receptors. Therefore, after receiving a large cumulative dose of phenobarbital (>>10 mg/kg), patients may be at increased risk of over-sedation if they receive benzodiazepines. This should be communicated to providers who will be caring for these patients after leaving the ICU or ED. Ideally, patients who have been stabilized using a large dose of phenobarbital should continue with a phenobarbital monotherapy strategy.
- Although benzodiazepines are regarded as the mainstay of treatment for alcohol withdrawal, there has never been an adequately powered RCT comparing benzodiazepines vs. phenobarbital.
- Benzodiazepines occasionally fail to control alcohol withdrawal, and may promote agitated delirium. In contrast, phenobarbital is more effective and doesn't cause paradoxical agitation.
- Some countries have extensive experience treating alcohol withdrawal with phenobarbital monotherapy. Available evidence supports the safety and efficacy of this approach.
- Phenobarbital monotherapy consists of a gradual dose titration as shown below. Once a therapeutic phenobarbital level is reached, this will gradually auto-taper and provide ongoing protection from seizures or recurrent withdrawal.
Coauthored with Dr. Ryan Clouser, neurointensivist colleague and drinking buddy.
Related posts: Delerium Tremens part I - Posted last year, this covers some background about alcohol withdrawal. Our current approach has changed from the algorithm described in that post, but the general rationale and pharmacology is the same.
Notes
(1) We use ideal body weight based as previously described in alcohol withdrawal (e.g. Ives 1991). Since phenobarbital is water-soluble, dosing based on actual weight in the setting of morbid obesity could lead to excessive doses.
Image credits: https://en.wikipedia.org/wiki/Shaolin_Kung_Fu#/media/File:Shi_DeRu_and_Shi_DeYang.jpg