CATR: What is the natural history of alcohol withdrawal and where does delirium tremens (DTs) fit on that timeline?
Dr. Gunderson: The general consensus is that nonspecific symptoms—nausea, anxiety, headache, agitation—emerge within hours of cessation, certainly within the first day. Seizures occur early, as soon as six hours after the last drink. Delirium can occur later, up to seven days, with a classic peak around days three to four. Interestingly, this conventional knowledge comes from a very small set of patients that dates back to the 1950s (Victor M and Adams RD, Res Publ Assoc Res Nerv Ment Dis 1953;32:526–573). My perspective is that this hasn’t been re-examined carefully enough in modern cohorts. In a study I participated in at the VA that examined 600 patients, most seizures occurred early, but some were much later, as far out as day three or four. For delirium, the peak was earlier than expected, between days one and two, and about a quarter of the patients were delirious upon presentation (Ronan M et al, J Addict Med 2024;18(4):389–396). So, I suspect seizures may occur later than the classic teaching and delirium may occur earlier.

CATR: Does treatment with benzodiazepines (BZDs) shift that time course? For example, could it push delirium later?
Dr. Gunderson: There isn’t a good way to know. What I do see, and what I teach, is that “delirium in the hospital” always has a differential. If a patient becomes delirious during management for alcohol withdrawal, DTs are one possibility, but BZD delirium is another. People assume more is better with BZDs, but that’s not the case. BZD toxicity can happen, and I’ve seen delirium that’s driven by the BZD rather than the alcohol withdrawal (Moore PW et al, J Med Toxicol 2014;10(2):126–132).

CATR: Where do hallucinations fit in?
Dr. Gunderson: Interestingly, I don’t encounter classic alcoholic hallucinosis, which is characterized by hallucinations in someone who is otherwise lucid, as often as you might expect. In our study, it showed up in about ten percent of patients. The key is not to assume hallucinations equal DTs. Do a mental status exam. If the patient is otherwise clear, the hallucinosis may be a benign feature of an otherwise mild withdrawal that does not necessarily require specific treatment beyond treating the underlying withdrawal. But patients with DTs, who require intensive care, also have hallucinations. You have to look at the patient in the context of the full clinical picture.

CATR: How do you recognize DTs at the bedside? What distinguishes it from other causes of delirium?
Dr. Gunderson: Most inpatient delirium is hypoactive—think hepatic encephalopathy, or delirium from infection. That kind of delirium is easy to miss. DTs is different. It’s hyperactive, patients are loud and can be quite agitated. The name “tremens” is apt; patients are tremulous, diaphoretic, their vitals run high. 

CATR: You mentioned differential diagnosis a few times. Beyond BZD delirium, what else is on your short list when someone with alcohol use disorder (AUD) becomes delirious?
Dr. Gunderson: Apart from DTs, BZD delirium is number one by frequency in my experience. Hepatic encephalopathy is another one to look out for. For some people with cirrhosis, liver dysfunction becomes clinically evident only when the person gets sick. Those patients tend to have the stigmata of cirrhosis on exam and, again, usually present with hypoactive rather than agitated delirium. Infection is also common. Pneumonia is the classic culprit and can lead to delirium either from a systemic inflammatory response or from respiratory failure. There are other possibilities, but those are the main three outside of DTs: BZD delirium, hepatic encephalopathy, and infection.

CATR: We all learn that DTs can be fatal. What is so dangerous about them?
Dr. Gunderson: I’m not sure DTs have a higher fatality than “run-of-the-mill” delirium when looking at inpatients broadly. For a hospitalist like me, any delirium is a big setback. The danger usually comes through the airway. Delirium can cause aspiration which can lead to pneumonia, chemical pneumonitis, airway obstruction, respiratory failure, and ultimately cardiopulmonary arrest. It’s less common, but cardiac mechanisms exist as well. While not due to DTs per se, electrolyte issues are rife in this population, hypomagnesemia being particularly common. Low magnesium can prolong the QT interval and precipitate Torsades, which can be fatal. 

CATR: Many antipsychotics prolong QT as well, and these are often used in the treatment of hyperactive delirium. Do you worry that they may further compound the risk of arrhythmia?
Dr. Gunderson: Yes, any QT prolonging medication can increase risk of arrhythmia, especially when the patient has electrolyte abnormalities. In addition, antipsychotics lower the seizure threshold. So, I tend not to recommend antipsychotics as a matter of course. That said, they can have a limited role, but whenever antipsychotics are used, I recommend carefully monitoring electrolytes and EKGs. 

CATR: What are risk factors for developing DTs and what are complications to look out for if delirium sets in?
Dr. Gunderson: The literature lists a lot of potential risk factors though findings are inconsistent across studies. In our own study, only a few variables were significantly associated with developing DTs: prior complicated withdrawal or prior delirium, being admitted for a medical reason, and a low platelet count (Ronan et al 2024). Many things you’d expect to matter like peak CIWA, abnormal vitals, even blood alcohol levels, didn’t predict progressing to DTs. Days since the last drink is often cited as a risk factor. The idea here is that people who present “late” are at higher risk; we saw a weak signal, but it wasn’t significant. In short, prior history and a few basic clinical features mattered; a lot of the rest did not.

CATR: Given that prediction is imperfect, what should we do to minimize DTs once the patient is in our care?
Dr. Gunderson: My approach lines up with the American Society of Addiction Medicine’s inpatient guidance (Lindsay DL et al, J Addict Med 2020;14(5):376–392). All hospitalized patients should be treated with symptom-triggered therapy, using a long-acting BZD, typically diazepam. Higher-risk cases receive additional medication in the form of a few loading doses early in treatment. Our medical center shifted from lorazepam to diazepam about five years ago, and that change has made management much easier and, in our experience, safer. 

CATR: So, the first step is risk stratification, and those at high risk receive front-loading with diazepam?
Dr. Gunderson: Exactly. The front-load ensures that the patient receives sufficient treatment as early as possible. And we provide additional medication as needed for elevated CIWA scores. One of the advantages of diazepam is its long-acting metabolites; once you’ve front-loaded, you’re covered. You don’t typically need a taper. I’m always mindful of delayed seizures and having a long-acting BZD on board is reassuring.

CATR: Can you give us some more specifics about your approach?
Dr. Gunderson: All patients receive symptom-triggered diazepam. For patients with very high CIWA scores or who are otherwise concerning, I use our front-load order set: 10 mg diazepam every six hours for 24 hours. That’s a total of 40 mg; roughly equivalent to 8 mg of lorazepam. But not all front-loads are equal. Some patients need only two doses; you can always stop if they’re sedated. Others need more. If I’m worried, I’ll give an extra 10 mg diazepam. If you administer it intravenously, onset is within minutes so you can repeat doses fairly quickly as long as the patient is monitored. Once light sedation is reached, you’re essentially done because of diazepam’s long duration of action. This ensures quick treatment and avoids chasing vital signs with more BZDs. Tachycardia and hypertension make clinicians nervous, but you often can’t normalize those by piling on BZDs alone. Low-dose clonidine can be effective for patients who already have BZDs on board—0.1 mg twice or three times daily as needed is often sufficient. I also like having gabapentin on board; 300 mg three times daily, in my experience, smooths things out and may have additional benefits in AUD (Anton RF et al, JAMA Int Med 2020;180(5):728–736).

CATR: What are some other reasons that you prefer to front load diazepam in high-risk patients?
Dr. Gunderson: Symptomatic treatment is inherently subjective. Some patients answer the CIWA questions in such a way as to receive additional BZDs. It’s not inherently malicious, and patients should report their symptoms, but when PRNs continue for days or even weeks, you can generate sedative-hypnotic dependence in someone who initially “only” had AUD. Our fix was to pair symptom-triggered diazepam with early front-loading, stop BZDs once light sedation is achieved, and auto-expire orders at 48 hours. Some patients do need more than two days; in that case you can reorder intentionally. That small checkpoint at 48 hours has proven to be quite helpful.

CATR: Any other tips?
Dr. Gunderson: Just one caveat: IV diazepam “hits” very quickly in a way that oral medication does not. I’ve had some patients refuse oral, which can become a separate management challenge. But overall, since adopting this diazepam-forward, front-load approach, our ICU transfers and rates of DTs are down in our facility. 

CATR: Clinicians sometimes hesitate to use diazepam in patients with severe liver disease.
Dr. Gunderson: Most prescribers choose lorazepam in that case because it’s less dependent on hepatic metabolism. Personally, in patients with significant liver disease, and also in older adults, I like to start with gabapentin if the withdrawal is not too severe and they have good renal function (Anton RF et al, JAMA Intern Med 2020;180(5):728–736). It isn’t hepatically metabolized, and I’ve found it to be less encephalopathic than BZDs. 900 or 1200 mg divided over 24 hours is a good starting point. If a BZD is still needed, I’ll use lorazepam. I’m cautious about front-loading lorazepam though, and in most cases, I’ll use symptom-triggered dosing. 

CATR: Many of our readers work in psychiatric settings. When should we refer to a medical service or higher level of care?
Dr. Gunderson: Fluid and electrolyte management is crucial, so any need for IV fluids should warrant an emergency department evaluation or referral to medicine. Those are patients who cannot keep down fluids because of nausea or vomiting. Substantially abnormal vital signs or very high CIWA scores should also prompt a referral. Some of these patients, particularly those with a long history of heavy drinking, can have significant electrolyte abnormalities; low magnesium, potassium, and phosphate are the common ones. Metabolic acidosis and acute kidney injury also warrant referral to acute medical treatment. 

CATR: You mentioned antipsychotics can play a role. 
Dr. Gunderson: Antipsychotics should not be used to treat withdrawal itself. They can be useful for behavioral control when BZDs are already on board and the patient still has unsafe behavior. During DTs, patients can display severe agitation, confusion, and they can become violent. You can see this in BZD-induced delirium as well. In these cases, when safety is really jeopardized, an antipsychotic can be useful. A mistake that I’ve seen is assuming that any agitation during alcohol withdrawal is DTs, and providers try to “benzo” their way to behavioral control; in my opinion, that’s how you can end up with oversedation and airway problems. In those circumstances, I usually use an antipsychotic—ideally with electrolytes corrected and QT monitoring. If repeated antipsychotic dosing is required, I monitor the QT with daily EKGs.

CATR: Finally, is there role of phenobarbital in DTs?
Dr. Gunderson: I believe there is. Not for everyone up front, but for the patient who isn’t going the way you want with diazepam, or the one who is starting to drift into delirium with other signs of ongoing withdrawal (tremor, diaphoresis, high vitals) despite BZDs. For those patients, phenobarbital can be your ace in the hole; switching them fully to phenobarbital treatment and discontinuing BZDs can be a good intervention for many of them. (Editor’s note: See our interview with Dr. Nisavic on pg 1 for more).

CATR: Thank you for your time, Dr. Gunderson.