[PLEASE NOTE:  For the most complete & updated material on alcohol withdrawal, please see the Internet Book of Critical Care Chapter on this topic here]

This is part I of an Oktoberfest series on alcohol withdrawal.  Part II on ketamine will be released in two weeks.

At Genius General hospital, we have been using phenobarbital monotherapy for alcohol withdrawal increasingly since 2014.  Our experience has been quite positive.  Per my (admittedly biased) recollections:

• Phenobarbital is uniformly effective (phenobarbital-refractory alcohol withdrawal doesn’t seem to exist, although under-dosing of phenobarbital certainly does exist).
• Patients without other ICU issues can be loaded incrementally with phenobarbital over a couple days and then discharged to the ward.  There is no need to keep patients in the ICU for week-long benzodiazepine infusions.
• The phenomenon of propylene glycol intoxication due to large-volume benzodiazepine administration has quietly disappeared.¹
• When dosed incrementally within an organized treatment scheme, phenobarbital shouldn't precipitate intubation.  Some patients will still get intubated for other reasons (e.g. pneumonia).

I should do a formal retrospective cohort study on this, but I simply don’t have time.  Fortunately, the ICU group at Northshore/Long Island Jewish has done exactly that.

Oks et al.  2018,  The safety and utility of phenobarbital use for the treatment of severe alcohol withdrawal syndrome in the medical intensive care unit.

This is a retrospective cohort study describing 86 admissions to the ICU for alcohol withdrawal between 2011-2015.²  86% were treated with benzodiazepines before ICU admission, usually on the general ward.  The average dose of benzodiazepine before ICU admission was equivalent to 23 mg of lorazepam.  Following ICU admission, all benzodiazepines were discontinued and patients were treated solely with IV phenobarbital.  Doses of 130 mg IV were given every 15 minutes in a symptom-triggered fashion to target a Richmond Agitation and Sedation Scale (RASS) of 0 to -1:

Patient characteristics are below.  An average ICU length-of-stay of five days is slightly higher than I would have expected, but this still compares favorably to many series utilizing benzodiazepines (including those in the next blog post).  Complex patients with additional problems likely increased the mean length-of-stay (e.g. alcohol withdrawal plus GI hemorrhage).

Phenobarbital has a very long half-life (~80 hours), so repeated doses will accumulate.  This is the intended strategy of phenobarbital dosing, which essentially amounts to gradually ramping up the patient’s phenobarbital level until symptoms abate.  Ultimately it is the cumulative phenobarbital dose over a few days that determines the patient’s serum phenobarbital level.  The total phenobarbital doses reported in this study were substantial:

80% of these patients didn’t require intubation.  Furthermore, intubation was never required as a consequence of phenobarbital use.  In fact, the only intubations due to drug-induced obtundation occurred as a result of benzodiazepines administered prior to the delivery of any phenobarbital:

This is probably the most important finding of the study.  A popular misconception is that phenobarbital will cause patients to require intubation.  This study demonstrated quite the opposite:

• Structured use of phenobarbital may reduce the requirement for intubation, compared to a benzodiazepine-based treatment strategy.
• Patients can receive a lot of phenobarbital and still not require intubation.  There were no differences in phenobarbital dose between intubated versus non-intubated patients.  The non-intubated patients received an average phenobarbital dose of 1,954 mg.

Oks et al. protocol strengths

The greatest strength of their protocol was the use of frequent (q15 minutes) and relatively small (130 mg) intravenous doses of phenobarbital.  Most prior studies have used less frequent dosing of phenobarbital (e.g. q30 min or q60 min) to allow for assessment of the effect of one dose before giving the next dose.  However, IV phenobarbital distributes rapidly, so a 15 minute dosing interval is arguably ideal.  Using a short dosing interval allows for relatively small doses to be given successively, facilitating precise titration to effect.

Logistically this strategy requires having a supply of phenobarbital vials available in the unit for immediate use (rather than ordering each dose from pharmacy, with an associated delay time).  Not all hospitals may be able to accommodate this.  Every 15 minute dosing isn’t mandatory to titrate phenobarbital, but it is elegant.

Another strength was their targeting of a RASS value, rather than a formal alcohol-withdrawal score (e.g. CIWA).  CIWA scoring among critically ill patients can be confounded by other medical problems.  I have more trust in the nurse’s clinical judgment than a CIWA score.  As demonstrated in this study, it is perfectly adequate to simply dose phenobarbital to the level of arousal (RASS), while ignoring other components of the CIWA scale (e.g. diaphoresis, nausea, headache).

Oks et al. protocol weaknesses

Their protocol is very similar to a protocol I was using in 2015 as shown here:

This protocol generally worked extremely well.  However, we did encounter one patient who received an excessive dose of phenobarbital (the patient did fine, but upon retrospective review his phenobarbital dose and levels were excessive).  The problem with the above protocol is that if the patient develops another cause of agitation besides alcohol withdrawal, the patient may wind up receiving enormous doses of phenobarbital.  The crux of the matter is as follows:

• Agitated delirium due to alcohol withdrawal should be treated with phenobarbital.
• Agitated delirium due to any other cause should not be treated with phenobarbital (e.g. benzodiazepine-induced delirium, sleep deprivation-induced delirium, sepsis-induced delirium).  Attempting to treat these forms of delirium with phenobarbital may be harmful.

The protocol was consequently modified as shown below, with greater attention to the cumulative phenobarbital dosing.  After the patient has received a relatively large cumulative dose of phenobarbital (30 mg/kg), additional phenobarbital is generally curtailed.  At this point it is assumed that the patient has received sufficient treatment for alcohol withdrawal.  Any residual delirium beyond this point is unlikely to be related to alcohol withdrawal, so it may be more appropriately treated with an antipsychotic.  This issue has been previously discussed on the blog in more detail (see: Pitfall #1 Looping Paradox.)

This newer version has worked well, providing a mechanism that helps to avoid over-use of phenobarbital among patients with multifactorial delirium.  When in doubt, it can be useful to check a serum phenobarbital level (with a target level of roughly ~10-40 ug/mL, extrapolating from the therapeutic levels for epilepsy).³   Data from Oks et al suggest that the “maximal” dose of 30 mg/kg used in this protocol might be too low, but this nevertheless seems like a reasonable approach until more data becomes available (especially in morbid obesity).

Updated phenobarbital guideline

Below is my updated guideline to phenobarbital dosing, based on this new data.  Phenobarbital titration has been changed to 130 mg IV q15 minutes with a RASS target.  Although a variety of dosing strategies are possible, currently this strategy is most strongly supported for phenobarbital monotherapy within a critical care context.

Limitations

As a retrospective cohort study, this cannot prove causality.  Specifically, it cannot prove that a phenobarbital strategy is superior to a benzodiazepine strategy.  The ICU at LIJ/Northshore is one of the best in the country, so it’s conceivable that they could have achieved these same results with a different medication.

Nonetheless, these observational data are impressive (particularly the low intubation rate).  I suspect that this largely reflects the underlying strength of phenobarbital as an agent for alcohol withdrawal.  At the least, it unequivocally demonstrates that phenobarbital monotherapy can be safe and effective.

Exercise caution in severe cirrhosis

One patient in this study was intubated for hepatic encephalopathy following treatment with phenobarbital.  Details regarding this case aren’t provided, but this is consistent with my experience.  To better illustrate this issue, let's consider the following case from Genius General is helpful:

A 38-year-old man was admitted to the ICU following intubation for status epilepticus in the context of alcohol withdrawal.  He received several doses of phenobarbital with the dual intention of seizure control and treatment of alcohol withdrawal.  In total, he received 20 mg/kg phenobarbital, which pushed his serum phenobarbital level to 22 ug/ml (this is incidentally at the lower end of the therapeutic range for epilepsy, 15-40 ug/mL).  While he received these dose increments, he remained agitated.  Extubation was initially delayed due to atelectasis.  When he was finally deemed ready for extubation based on respiratory parameters, he was unexpectedly found to be somnolent.  He had developed hepatic encephalopathy during his ICU course, which may have synergized with phenobarbital to cause somnolence.  He tolerated extubation and was treated for hepatic encephalopathy, with improvement.  However, his somnolence nearly precluded extubation and thereby threatened to prolong his intubation.

As explored previously, the pharmacokinetics of phenobarbital are extremely predictable.  This allows us to give 10-30 mg/kg phenobarbital to anyone, with confidence that this will not cause a toxic phenobarbital level (>40 ug/mL).  However, in patients with a very tenuous mental status for another reason (e.g. hepatic encephalopathy), it is possible that even a therapeutic phenobarbital level could compromise mental status.

Cirrhosis is not a contraindication to phenobarbital, but be careful.  Try to use doses on the lower end (e.g. cumulative doses of <15 mg/kg).  If there is concern for worsening mental status due to hepatic encephalopathy, treat this promptly and aggressively.

P.S. – Be skeptical of guidelines

This is a newer therapy which isn’t necessarily understood by traditional guidelines and reviews.  For example, the guidelines above are utterly, comically wrong on this.⁴  Oks 2018 includes a nice literature review in their recent paper, so for accurate information that’s a better place to start.²

• Oks et al. shows that symptom-triggered IV phenobarbital monotherapy can be safe and effective, even in a cohort of sick ICU patients who failed therapy with benzodiazepine (prior to ICU admission).
• Patients received large cumulative doses of phenobarbital, yet 80% didn’t require intubation.  No intubations were attributed to phenobarbital-induced obtundation, whereas six were attributed to benzodiazepine-induced obtundation.  This suggests that adequate use of phenobarbital for alcohol withdrawal may actually help prevent intubation.
• Oks et al. demonstrates that small doses of phenobarbital may be given frequently (130 mg IV q15 minutes PRN), allowing precise yet aggressive titration of the cumulative phenobarbital dose.
• Phenobarbital may be titrated until the patient is awake/calm or slightly sleepy (target RASS -1 to 0), without using a formal CIWA score.
• In order to treat alcohol withdrawal, you really need just one order. Phenobarbital 130 mg IV q15min PRN RASS>0.  Or, if your shop doesn’t use RASS:  phenobarbital 130 mg IV q15 min PRN anxiety/agitation.  That’s it.  One order to sedate them all.
• To avoid running into trouble: exercise caution in cirrhotics and avoid cumulative phenobarbital doses >30 mg/kg.

current phenobarbital protocol:

More information:  The phenobarbital series

• Phenobarb I (12/2014)
• Phenobarb II (10/2015)
• Phenobarb III (12/2017)

[PLEASE NOTE:  For the most complete & updated material on alcohol withdrawal, please see the Internet Book of Critical Care Chapter on this topic here]

• Author
• Recent Posts

Social Me

Josh Farkas

Josh is the creator of PulmCrit.org. He is an associate professor of Pulmonary and Critical Care Medicine at the University of Vermont.

Social Me

Latest posts by Josh Farkas (see all)

• PulmCrit Wee – Loading infusion auto-titration (LIAT) for infused medications with intermediate half-lives - March 23, 2025
• PulmCrit Wee: Michelin Chest Syndrome - March 15, 2025
• PulmCrit: ADAPT and SCREEN trials are full of sound and fury, signifying little - December 13, 2024