Currently there is a lorazepam shortage in the United States. This has caused a surge of interest into using phenobarbital to manage alcohol withdrawal. I've received several e-mails over the past few weeks about this. It's been two years since my last post about phenobarbital, so here's an update focusing on lessons learned in the interim.
Background: Why use phenobarbital monotherapy?
#critcarepearls During lecture today: "no such thing as alcohol withdrawal, only phenobarbital deficiency"
— Anthony Provenzano (@docpro89) December 1, 2017
This has been explored in more detail previously. To summarize:
- Superior neurochemistry: Alcohol withdrawal involves deficient inhibitory GABA-receptor activity as well as excessive stimulatory NMDA-receptor activity. Benzodiazepines only affect GABA receptors, not NMDA receptors. In comparison, phenobarbital stimulates GABA receptors and inhibits
NMDA receptors(1). [Erratum- phenobarbital affects AMPA-type glutamate receptors, not NMDA-type glutamate receptors – Thanks to Meghan Spyres for pointing this out. Clinical implications are the same.].
- Reliable efficacy: A small subset of patients fail to respond to benzodiazepines (Hack 2006). Phenobarbital is effective in such patients, possibly because of superior neurochemistry.
- Predictable efficacy: Phenobarbital is effective at a cumulative dose of roughly 10-20 mg/kg. Compare this to benzodiazepines, where the effective dose varies enormously (e.g. some patients respond to 40 mg diazepam, whereas others may require 400 mg). Predictable efficacy helps you know when to stop giving more medication (more on this later on).
- Predictable pharmacokinetics: If you keep track of the cumulative dose of phenobarbital administered, the blood level can be easily predicted (and vice versa). This makes it easy to understand how much drug the patient has on board.
- Wide therapeutic index: Phenobarbital is effective for alcohol withdrawal at a dose of ~10-20 mg/kg (corresponding to a blood level of ~12-25 ug/mL). Severe toxicity (stupor/coma requiring intubation) shouldn't occur below a blood level of ~65 ug/mL if other sedatives aren't on board. This should provide a good margin of safety.
- Avoidance of intubation: There is often fear about respiratory suppression when using phenobarbital. However, when dosed appropriately in a patient who genuinely has alcohol withdrawal, the risk of respiratory suppression is minimal (because of #4 & #5). As explored further below, phenobarbital may reduce intubation risk.
- No paradoxical reactions: Some patients exposed to benzodiazepine or propofol will experience paradoxical agitation. This is more common in alcoholics, possibly explaining why some patients with alcohol withdrawal respond poorly to benzodiazepines. Paradoxical reactions don't seem to occur with phenobarbital.
- Sustained efficacy: Once you achieve a therapeutic phenobarbital level, patients generally do well. Phenobarbital will gradually taper itself off over several days (its half-life is ~80 hours). Patients might need a tad more phenobarbital, but they shouldn't suddenly deteriorate.
- Excellent anti-seizure efficacy: Barbiturates are the most effective anti-epileptic agents available (e.g., they can reliably break super-refractory status epilepticus). A phenobarbital-based strategy provides good protection against alcohol withdrawal seizures.
- Simplicity: Benzodiazepines don't always work, so benzodiazepine-based protocols must involve several medications (e.g. benzodiazepine, phenobarbital, propofol). In contrast, monotherapy with phenobarbital makes things simple. All you need to do is titrate up the patient's phenobarbital level to a therapeutic level. There is no polypharmacy, drug interaction, or confusion about which agent to choose.
- Single drug in IV, IM, and PO formulation: My preference is to give phenobarbital intravenously, to see the clinical effect rapidly and facilitate prompt dose-titration. However, phenobarbital can also be given intramuscularly and orally with excellent absorption. Multiple formulations provides flexibility (e.g. you can load with IV phenobarbital in the ED/ICU, then use PO phenobarbital for additional therapy on the ward).
- Benzodiazepines add nothing to phenobarbital. Adding benzodiazepine to phenobarbital probably doesn't improve efficacy. However, benzodiazepines may reduce the toxic level of phenobarbital – making it harder to dose phenobarbital safely (2).
Overall experience with phenobarbital monotherapy
Although phenobarbital monotherapy is perceived as a new intervention, it's actually an extremely old intervention. Barbiturate monotherapy was used as front-line therapy for alcohol withdrawal for the majority of the twentieth century in some European countries. Old studies comparing the two interventions suggest that barbiturates were equivalent or superior compared to benzodiazepines (Hjermo 2010, Kramp 1978). What's old is new again: currently, some centers are moving (back) to phenobarbital monotherapy.
Beth-Israel Deaconess (Harvard)
This study is available only as an abstract from the 2016 Society of Critical Care Medicine conference:
Further discussion of this study must await complete publication. However, the results are exciting. A preliminary data slide from the 2017 ASHP conference a few weeks ago suggests a reduced rate of intubation while using phenobarbital (3):
— Kyle DeWitt (@EmergPharm) December 6, 2017
Genius General Hospital
For about two years, phenobarbital has been a front-line therapy for severe alcohol withdrawal within two ICUs in the Genius General Intergalactic Hospital Network. One of these ICUs is staffed by intensivists, while the other is staffed by hospitalists (who communicate regularly with the mothership via telemedicine).
Although we haven't formally studied this, it seems to work very well (similar to the Beth Israel data above). Symptoms can generally be controlled within several hours, with a low rate of recurrence. Patients with pure alcohol withdrawal don't require intubation, although intubation may be needed in patients with multiple active problems (e.g. pneumonia complicated by withdrawal). The phenomenon of the patients with alcohol withdrawal who linger in the ICU for days on a potpourri of sedative infusions has disappeared.
The remainder of this post will focus on some pitfalls that have been encountered using phenobarbital, and how they may be resolved. However, bear in mind that our overall experience with phenobarbital has been quite positive.
Pitfall #1 with phenobarbital: Looping paradox
The looping paradox is most common when using benzodiazepines to treat alcohol withdrawal. Benzodiazepines can treat alcohol withdrawal, but they also can cause delirium. This often creates a very confusing picture on hospital day #2-3 if the patient remains delirious:
- If the patient is continuing to suffer from alcohol withdrawal, then additional benzodiazepine is needed. In this case, giving an antipsychotic (e.g. haloperidol) is relatively contraindicated, because it would fail to treat the underlying withdrawal.
- If the patient is suffering from benzodiazepine-induced delirium, then additional benzodiazepine is harmful. In this case, small doses of haloperidol would be beneficial.
In practice, numerous patients wind up receiving excessive benzodiazepine, leading to a state of benzodiazepine-induced delirium. This pitfall was demonstrated by a case series of patients with prolonged delirium who improved following administration of flumazenil (thereby diagnosing iatrogenic benzodiazepine poisoning)(Moore 2014). Flumazenil isn't recommended due to the risk of seizure, but this study proves the concept that patients often wind up in a loop of excessive benzodiazepine administration.
The looping paradox seems to be much less problematic when using phenobarbital, because phenobarbital causes less delirium and paradoxical agitation compared to benzodiazepines. Nonetheless, occasional patients can recieve excess sedation even when using phenobarbital:
- More common cause of excessive dosing: If patients have multi-factorial delirium (e.g. delirium due to alcohol withdrawal plus sepsis), then they may remain delirious even after receiving a therapeutic dose of phenobarbital. If residual delirium is misinterpreted as an indication to give more phenobarbital, they will receive excessive doses of phenobarbital.
- Rarely: Some patients enjoy the effect of IV phenobarbital and feign symptoms of withdrawal in order to continue receiving more phenobarbital.
No patient at Genius General has suffered from harm due to excess phenobarbital (the use of small PRN doses at a fixed rate makes it nearly impossible to push a patient into a dangerously stuporous/comatose state). However, in retrospect it does seem like occasional patients probably received a bit more phenobarbital than they may have absolutely required.
Solution to the looping paradox
When using phenobarbital, patients who are receiving unnecessarily high doses may be identified on the following basis:
- Total cumulative dose of phenobarbital >20 mg/kg should raise concern. The literature is contradictory regarding exactly how much phenobarbital is required to treat alcohol withdrawal. Ives 1991 reported success using a fixed dose of 15 mg/kg. However, reports from Denmark document the use of much greater doses (e.g. Hjermo 2010 described using an average cumulative dose of 3,000 mg). Overall, it seems that most patients will respond to a dose of ~15 mg/kg, but occasional patients might require more.
- Aberrant pattern of phenobarbital use. Patients generally require the greatest doses on the first day, with rapidly decreasing requirements thereafter. If a patient is receiving substantial doses of phenobarbital day after day, this suggests a problem other than simple alcohol withdrawal.
- Failure to respond to PRN doses. 130-260 mg IV phenobarbital alone doesn't always have a huge effect, but it usually causes some improvement. If the patient is obtaining no clinical improvement despite ongoing PRN dosing, this should cause reconsideration of the diagnosis.
The presence of these features should trigger re-evaluation for another concurrent cause of delirium (e.g. fluoroquinolone-induced antibiomania, personality/psychiatric disorder, head trauma, CNS infection). After treatment with >20 mg/kg phenobarbital, alcohol withdrawal has probably been treated adequately. Whatever symptoms remain are likely due to another problem. Thus, phenobarbital should probably be stopped and remaining agitation may be treated with PRN antipsychotics (e.g. haloperidol or olanzapine).
Occasionally an extremely agitated patient is encountered. This seems to occur especially following the administration of benzodiazepines, which may cause paradoxical agitation and exacerbate delirium. Severe agitation cannot be immediately treated with phenobarbital, because phenobarbital must to be given at a controlled rate.
A reasonable strategy here is to use a dexmedetomidine infusion along with phenobarbital. Dexmedetomidine helps achieve rapid immediate sedation. As phenobarbital is administered, the dexmedetomidine is weaned off. The goal is to discontinue dexmedetomidine within 8-12 hours, with transition to phenobarbital monotherapy.
Overall, dexmedetomidine isn't a great drug for alcohol withdrawal for several reasons:
- Doesn't reduce seizure risk
- Doesn't treat underlying problem (GABA hypoactivity & NMDA hyperactivity)
- Not a durable solution to the problem (even if patient responds beautifully to dexmedetomidine, this doesn't get you anywhere – the patient cannot leave the ICU or go home on a dexmedetomidine infusion)
Nonetheless, this is one situation where dexmedetomidine is useful as an adjunctive agent to temporize matters. The goal of this scheme is to achieve for behavioral control and phenobarbital titration without requiring intubation (which would otherwise frequently be required in these highly agitated patients).
Pitfall #3 with phenobarbital: Logistics of drug delivery
A common problem when using phenobarbital is delay in delivering the drug from the pharmacy to the bedside. Phenobarbital titration may involve PRN doses given as frequently as every 30 minutes. Some pharmacies will formulate this within individually mixed mini-bags (e.g. 130-260 mg phenobarbital within 100 ml normal saline). Using individual mini-bags makes it sluggish to achieve phenobarbital titration, because the pharmacy often takes 45-60 minutes to deliver each mini-bag. This problem can be overcome using a variety of solutions:
- Avoid giving patients benzodiazepines, but instead initiate treatment using a 10 mg/kg phenobarbital load up-front (previously discussed here). Starting with front-loaded 10 mg/kg phenobarbital will typically cause a substantial improvement in symptoms within an hour, allowing further dose-titration to occur at a more leisurely rate.
- Stock individual vials of phenobarbital at the point of care. These can be drawn up at the bedside and pushed over 5 minutes as needed.
- Order more phenobarbital from pharmacy than is immediately needed (e.g. 500 mg), then only administer half or a quarter of it at a time.
Parting shot & future directions
It has been said that alcoholics exist to teach physicians humility. These patients are quite complicated, often suffering from several problems simultaneously (e.g., withdrawal, nutritional deficiencies, trauma, infection, psychiatric problems). Arguably the most important task when treating alcohol withdrawal is exclusion of alternative or superimposed problems. Phenobarbital obviously cannot be expected to solve every problem that may exist in an alcoholic patient.
Phenobarbital does, however, appear to be a uniquely effective and safe treatment for alcohol withdrawal (including withdrawal seizures and delirium tremens). Several reasons for the superiority of phenobarbital are listed above. Preliminary evidence suggests that these do indeed translate into meaningful clinical benefit (e.g. faster recovery, fewer complications).
The true power and beauty of phenobarbital emerges when it is used as monotherapy. In my opinion, adding a benzodiazepine to phenobarbital brings nothing to the table other than confusion and toxicity. When used in isolation, phenobarbital has predictable pharmacokinetics and pharmacodynamics, which allow for rapid achievement of a therapeutic level and confident avoidance of toxicity (2). Unfortunately, addition of a benzodiazepine muddies up the relationship between the phenobarbital level and respiratory suppression. The conventional practice of starting with a benzodiazepine and adding phenobarbital for refractory symptoms is thus misguided, preventing the optimal use of phenobarbital.
- Phenobarbital monotherapy has numerous advantages compared to benzodiazepines. Emerging evidence suggests that translates into meaningful clinical benefit.
- If 15 mg/kg phenobarbital fails to achieve clinical resolution, this suggests that the patient may have another problem in addition to alcohol withdrawal. Alternative causes of delirium should be explored.
- For severely agitated patients, dexmedetomidine can be useful initially to gain behavioral control. However, phenobarbital should be rapidly up-titrated to treat the underlying disorder and avoid seizure.
- An updated approach to phenobarbital use is suggested (figure below).
- If you work at a center that is using lorazepam for alcohol withdrawal, the current shortage of lorazepam isn't a threat – it's an opportunity to move to a superior therapy.
- PulmCrit phenobarbital series:
- Phenobarbital in Alcohol Withdrawal (CoreEM)
- There is another reason that phenobarbital may be more powerful than benzodiazepines. Benzodiazepines increase the duration of GABA channel opening, but barbiturates can open the GABA channel by themselves. In a brain with very little endogenous GABA, this could make barbiturates more effective (because benzodiazepines need some endogenous GABA to work, whereas barbiturates don't).
- For example, in the absence of any other medications 10 mg/kg of phenobarbital will achieve a serum level of ~13 ug/ml. This is below the therapeutic level of phenobarbital in epilepsy (which is 15-40 ug/mL), which should be quite safe – especially in an alcoholic with hyperarousal. This allows you to get a lot of phenobarbital (10 mg/kg) on board rapidly, in a safe fashion. However, in the presence of benzodiazepines it is possible that a phenobarbital level of 13 ug/mL might cause synergistic toxicity, so you lose the ability to rapidly load the patient with phenobarbital. Phenobarbital has a very attractive therapeutic index when used as monotherapy (therapeutic level of roughly 15-25 ug/mL, severe toxicity starts to occur >65 ug/mL) – but this index may not be reliable in the face of benzodiazepine use.
- This is a before/after study design (i.e. comparing historical controls to patients treated after design of a phenobarbital-based treatment protocol). As such, it cannot prove causality (due to, for example, confounding from the Hawthorne effect). Nonetheless, it demonstrates that introduction of a phenobarbital-based protocol in a major academic center can improve outcomes.