Prescribing Benzodiazepines in Psychiatry, With Carl Salzman, MD

Joseph F. Goldberg, MD, in this installment of "Brain Trust: Conversations in Psychopharmacology," sits down with Carl Salzman, MD, to discuss the enduring safety and appropriate use of benzodiazepines in clinical practice. Despite 6 decades of accumulated pharmacologic and clinical data, benzodiazepines remain widely misunderstood, with prescribers often caught between undertreating anxious patients and fear of dependence or abuse. National survey data indicate that roughly 1 in 10 US adults report benzodiazepine use in a given year, with a smaller subset using these agents regularly and as prescribed.¹ Contemporary guidance continues to emphasize short-term use when possible, alongside individualized, risk-based decision-making for patients who require longer-term therapy.² Salzman draws on decades of research and clinical experience to unpack when benzodiazepines are the right choice, how to assess candidacy for long-term use, and where these agents fit relative to antidepressants and antipsychotics.

See the full video podcast episode of this conversation here.

Joseph F. Goldberg, MD: Our topic today is benzodiazepines and their misunderstood status—perceptions of them, the Beers criteria, and the fear among so many clinicians that they are to be avoided. I want to start by asking: when you are inevitably asked, are benzodiazepines all good or all bad, how do you typically respond?

Carl Salzman, MD: We have had benzodiazepines since the 1960s, with Librium being the first of the drugs. They revolutionized the treatment of anxiety disorders. Prior to them, we had only barbiturates and Miltown (meprobamate)-type drugs, which were much more dangerous and had many side effects. These drugs produced a reliable sense of relaxation, calmness, occasionally sedation, and very few side effects, depending on dose. If you took a lot of them, they were sedating and could affect memory and balance, but those were doses way above what was normally used—Librium was usually something like 10 milligrams 2 or 3 times a day. Because they were so novel, research began, and research has continued with the benzodiazepines right up to the present. So we really understand these drugs very well, from a pharmacologic point of view as well as from a clinical point of view, and it is very clear from 60 years of experience that when appropriately used, these drugs are safe and effective. I want to emphasize appropriately used—the right drug, for the right person, at the right dose, for the right period of time. That is good clinical practice, basically, because that is what you would want to do for a patient anyway. But these drugs have not always been used that way, and they are sometimes overused. They are certainly misused, and they can be abused, and I think we need to talk about all of those things.

The question of misuse is undefined, but I think most people think that means taking the drug without a doctor's prescription. So how many people take these drugs without a prescription? I can answer that to some degree of accuracy, because there have been a number of surveys. The one I think is most interesting was done before the turn of the century: a door-to-door survey in a suburb of the San Francisco Bay Area. People who opened the door were asked: have you taken one of these drugs even once in the past 12 months? About 11% said yes, at least once. The follow-up question—have you taken it more than once—dropped to about 6%, or 5.5%.Those numbers were compared with European countries, and it turned out Americans were using benzodiazepines less than most other countries in Europe, except for Britain. Subsequent studies have found the numbers are still more or less the same—about 10% have taken a benzodiazepine on occasion, and about 5% to 6% take them on a regular, prescribed basis.

Is that good or bad? That depends on how you see the world. If prescribed use is effective and helpful, that is very good, because at therapeutically prescribed doses, these drugs are remarkably safe. What about the additional people taking them without a prescription? As clinicians, we know that many people, psychiatrists included, keep benzodiazepines in the medicine chest at home, or take them when they travel. For example, many years ago at the American Psychiatric Association meeting, I was giving a talk on benzodiazepines. There were about 500 people in the room, and I asked how many had brought a benzodiazepine with them to the meeting, even if they did not plan to take it. About a third said yes. Why? Mostly to help sleep, since they were in a different location, and some for flight anxiety. The point is that we psychiatrists, who should know about these drugs, might take them unprescribed on occasion. Is that misuse? Is it bad? You would have to make that judgment yourself. I do not think it is bad, because there is no evidence that this leads to further use or drug addiction.

The second question that comes up is that these drugs are sedative hypnotics, and like all sedative hypnotics, they work at the GABA-chloride ionophore complex, and they can cause a physiologic dependence if taken regularly over time. If you take the drug at a prescribed dose—say, Xanax 0.5 milligrams 2 or 3 times a day—when will you really be dependent, and what is that dependence like? If you take it regularly, within a few weeks your receptor system will probably be modified such that if you suddenly stopped the pills, you might have withdrawal symptoms. It depends on the dose and the duration—the more you have taken, the worse the withdrawal, but it is rarely very severe unless very large amounts were taken over a long period, which is not standard prescribing. At therapeutic doses, withdrawal symptoms are relatively mild and short-lived, and easily treated by tapering over a week or 2, rather than stopping abruptly. Some people have trouble discontinuing a benzodiazepine—we can come back to that. By and large, over 60 years, even in chronic use, these are safe and effective drugs, and there is no evidence of dose escalation—patients at a therapeutic dose tend to stay there, and may even decrease it slightly over time. Dose escalation is rare.

Goldberg: I would imagine that is one of the sticking points for most prescribers—how well do I know this person who is only going to use this before their colonoscopy, or other prescribed indication?

Salzman: Very good question. How do we do this in the office, in real life? If we know the patient—if this is somebody we have known for a while—we have a sense of whether they are trustworthy, whether they will take it as prescribed. That means knowing the patient and building a therapeutic alliance. But what if you do not know the patient? This is a matter of clinical sensitivity and experience. I do not think there are any hard rules, but I think everybody would agree: the patient comes in and says, “Doc, I just moved to the city, and the dog ate my prescription,” or “I lost my prescription—could I have a Valium?” A red flag goes up. You say, “Wait a minute, maybe we ought to talk a little more—maybe you should come back next week.” If the patient does not show up, you have probably identified a potential abuser.

There is no official way to figure this out, except good clinical practice—doing a good interview, finding out why they might need a benzodiazepine, whether they have taken it before, whether there is a family history, what the symptoms are, and whether this is an acute crisis or something chronic. Some people take benzodiazepines regularly, and it is reasonable; others take them regularly, and it is not. Because these drugs can have negative effects at high doses, we want to be careful, as with any sedative hypnotic.

Goldberg: May I ask about 2 particular psychiatric diagnoses? We have patients coming to us in mental health who presumably have a psychiatric condition that may be anxiety, or may coexist with anxiety, but certain conditions arouse particular concern—the VA practice guidelines frown on long-term benzodiazepine use in posttraumatic stress disorder, and the literature on borderline personality disorder speaks of a kind of insatiable way of allaying anxiety, which is where the problems come in. Could you speak to those?

Salzman: Sure, that is an excellent question, because it is true that there are populations of patients likely to be long-term users, and that might or might not be appropriate, depending on the individual patient. In my own opinion—and people can very much disagree with it—making a generalization about a class of people, and whether they should take benzodiazepines, is not warranted, because even within a class of people (veterans, people with personality disorders, even people with substance use disorders) may actually benefit from an occasional benzodiazepine. To restrict them and say, “No, you cannot have them,” I think, is not good practice. Again, this is my opinion.

But what I have noticed in recent years is that because everybody is so worried about benzodiazepines, patients who should legitimately get one are not given it, and doctors are made to feel guilty, even worse, if they prescribe them when the general opinion is that they should not. Now, I will admit there are plenty of people who prescribe benzodiazepines who should not, or who prescribe them ill-advisedly—there is a lot of data about that too. There are doctors who prescribe benzodiazepines to many people, ill-advisedly, and those patients do become dependent on them. I certainly do not approve of that.

Goldberg: Maybe this is about sizing up candidacy—the same way a surgeon decides whether you are a candidate for an aortic valve replacement, or for CPAP. I guess the other question is: what else would you use for long-term management of a chronic condition?

Salzman: Right—so we can talk about the options for long-term use of benzodiazepines, and how they compare with other substances used for long-term use. If we are thinking about anxiety, or anxiety-spectrum disorders—DSM conditions—the 2 classes of drugs used for long-term use are benzodiazepines and antidepressants. Fortunately, we have enough data to guide us as to which would be preferred in general, though for any individual patient, the choice is still a clinical one. We know that serotonergic antidepressants, the selective serotonin reuptake inhibitors (SSRIs), can be very useful for treating long-term, chronic anxiety disorders, so this is a reasonable use of SSRIs, and the data suggest they are effective in most cases. What the data do not show is the side effects of long-term SSRI use in people functioning in midlife who have an active sexual life—SSRIs have sexual side effects for a reasonable number of people, so the benefit of the SSRI might be tempered by that potential problem, among other SSRI side effects. I am not saying this should eliminate the SSRI—I have prescribed them for long-term anxiety, like everybody else—but there is that potential problem.

On the other hand, benzodiazepines for long-term use have been very well studied—there is no dose escalation, and if the patient is not a substance abuser or alcohol abuser, long-term benzodiazepine use might be safe as well as effective. There are very few head-to-head studies comparing long-term use for anxiety with SSRIs versus benzodiazepines. There is one study in social anxiety, by Mark Pollack, MD, comparing benzodiazepines with SSRIs, which showed the benzodiazepines were more effective, and patients liked them more; other studies showed the SSRIs were effective.³ But most people taking an SSRI for a long-term anxiety disorder are also taking a benzodiazepine.

Goldberg: Or they are also taking a plethora of things—many of our listeners have patients on an SSRI, an atypical antipsychotic, hydroxyzine. For really difficult cases, we often have a boatload of things in the picture.

Salzman: Right. So one of the things I practice myself, and try to teach our residents, is to ask very carefully about alcohol, which is the primary issue with benzodiazepines, because they both work at the GABA-chloride ionophore. There is a severe drug interaction, and for somebody who is a heavy drinker, giving a benzodiazepine might not be the best idea. I would certainly consider trying something else first, rather than going straight to a benzodiazepine.

Goldberg: That is a pretty lethal combo, is not it, to mix and match your benzodiazepine with your vodka?

Salzman: That is correct—benzodiazepines are not lethal in overdose, except when combined with other sedative hypnotics, like alcohol; then they are dangerous, and that should be clearly understood by the prescriber. If you think the patient is also using other substances of abuse, benzodiazepines might not be a reasonable choice.

This is a good time to talk about benzodiazepines being abused—they are, by substance abusers, alcohol abusers, and opiate, cocaine, and other street drug users, and we want to be very careful not to provide these people with benzodiazepines. But benzodiazepines by themselves are not abused for pleasure—people do not go to Xanax parties on Saturday night. They may go to parties where Xanax is used, but it is always in conjunction with something else. So the idea of abuse is real and serious—I do not mean to minimize it—but by themselves, benzodiazepines are only abused by people who might be taking them for personal reasons, feeling they are not getting better, hoping the benzodiazepine will fix their disordered life, personality, or interpersonal problems.

Goldberg: I would bet a fair number of our listeners are thinking about a patient right now who fits that description—where it is not simply assuaging psychic or somatic anxiety related to something like generalized anxiety disorder, panic, or social phobia, but something a little more existentially pervasive—somebody who just cannot seem to self-soothe, who has never acquired alternative means to keep themselves intact.

Salzman: If there is one message I think ought to get across, it is that the people who should not be taking benzodiazepines may be taking them in the hope that it will help straighten out a disordered life they are otherwise having trouble with, and that there will be some magic in these pills. It can be understandable—some people's lives are in terrible shape, and they are not good candidates for psychotherapy, and they come in for a pill.

For example, I was talking with a young psychiatrist who lived in an area of Massachusetts where there were not a lot of psychiatrists, so medical doctors were sending him all kinds of patients they felt needed psychiatric care. Many of these patients came in complaining that they were so unhappy and so anxious, and could they please have their clonazepam. These patients had heard, or had experience, that somehow clonazepam was going to be helpful, and was certainly going to help them sleep. This poor psychiatrist had patients coming in one after another, asking for clonazepam, or Xanax, or something. He asked me what he should do; he did not think all these people needed it, but he did not know how to say no, and when he tried, these patients would get very angry, and he was sometimes threatened physically. I suggested he put up a sign in the waiting room saying, “This doctor does not prescribe clonazepam for any reason,” and have the person answering the phone say the same thing. The next week, 50% of the people who had been showing up disappeared, and the remaining people had legitimate reasons to take benzodiazepines. But that other 50% is suffering, and needs help of some kind. They may think help looks like a Xanax pill, but the psychiatrist or other mental health evaluator might say, “I am going to identify the nature of the problem you are having, in a way that is a little different than how you are identifying it. What you think is going to be helpful may not necessarily be what I think could be helpful.” You almost have to sneak up on somebody to say, “I want to help you. I think I have something that can help, but I do not know that what you envision is the answer.” That is good psychiatric practice—that is what I would hope people have learned in their training, and do themselves.

Goldberg: I think you are spot on. What I would add is that you might say to the patient, “I hear that you are really suffering, but maybe a second interview would be really helpful, so we could really understand what might be best. Would you be willing to come back for a second interview?” In other words, do not prescribe the drug right off the bat.

Salzman: My guess is that people who are coming for the pill might not come back—they are shopping, they know what they want, and you are trying to treat them, but they just want the refill. That may filter out that population.

Goldberg: Can I ask—when I was in training, there was this term, tell me if you have heard it—“ego glue.” When I was in training, antipsychotics were thought to maybe address, to use some old-fashioned language, the looseness in the minds of some people who just cannot seem to bind their fundamental ontological anxiety together, and that maybe a dopamine blocker, for all its own problems, could help. I am curious if you have an opinion about that.

Salzman: I actually agree with that. I think that for some people, a low dose of an antipsychotic is the correct choice, not a benzodiazepine—that is not somebody I would immediately be thinking of for a benzodiazepine. I would want to rule out bipolarity, substance abuse, and maybe even personality disorder of the borderline type, because patients with borderline personality disorder can be substance abusers and benzodiazepine abusers. But for people who really cannot get it together, who cannot perform normally, whose thoughts are erratic, sometimes a low dose of an antipsychotic can be very helpful. Aripiprazole is a very nice drug in low doses for that, because it does not produce a lot of side effects.

Goldberg: Do you think a dopamine blocker actually goes more directly to, let's say, the fight-or-flight limbic processes, as opposed to the GABA system altogether—that fear-based hyperarousal might respond better to an anti-dopamine drug?

Salzman: That is a very good question—I would have to speculate. I would look for other kinds of problems—sleep, and family history. We just ran our big master class over the weekend, and this question came up a number of times, and all of our speakers agreed that one of the key pieces of information to guide you in picking a class of medications is family history.

I will give an actual example: I was interviewing somebody, and I was not sure what was going on, so I asked whether anybody in the family had any troubles. They said no. I said, is there anybody who does not sleep very much, who is up a lot at night? He said, “Oh, you mean like grandpa? Yeah, grandpa is always up at night, chopping wood at 2 in the morning.” I said, well, what about your sleep? He said, “Yeah, I do not sleep so much—I am lucky, I do not need to sleep.” And suddenly there was the bipolarity, which otherwise would have been missed.

Goldberg: If it were simple insomnia, or primary insomnia, would you favor nowadays a benzodiazepine-receptor agonist, like a Z-drug, over a benzodiazepine, to spare benzodiazepine availability for anxiety, and not use it for sleep?

Salzman: I am not the best person to ask about sleep and treating sleep, but I can tell you what the sleep people say. They say the Z-drugs are preferred over benzodiazepines, because the GABA system has many subtypes, and the Z-drugs tend to work on different subunits than the benzodiazepine agonists. So they would recommend that, usually with a recommendation of no more than a couple of weeks of nightly use, because the drugs lose their efficacy. But patients who take them every night say they do not lose efficacy at all—they are very helpful. Do not quote me on that, though, and do not use me as a guideline for Z-drugs, because I rarely prescribe them—I have never taken one, so I do not really know the answer. I do think short half-life benzodiazepines, like lorazepam, are sometimes very helpful for sleep when people are in a state of arousal—namely, crisis or illness—and we discussed this in the master class as well. The idea about a short half-life is that it is reasonably rapidly absorbed. Lorazepam is highly lipid soluble, and there is no hangover when you wake up in the morning. It is not the most rapidly absorbed, however—diazepam, Valium, is more rapidly absorbed, but Valium has a longer half-life. That longer half-life does not matter if you are only taking 1 or 2 nightly pills, but if you are taking it regularly, the drug tends to accumulate, and pretty soon you have a hangover and are sedated during the day. So long half-life benzodiazepines are not recommended for that.

Let me make a little sales pitch. I said we know the pharmacology of these drugs—that is one of the reasons benzodiazepines are so useful. That pharmacology suggests long half-life drugs have a problem if someone is taking them acutely, because they stay in the body longer, so side effects are produced longer, and short half-life drugs do not have that problem. On the other hand, if you are taking a benzodiazepine for chronic suppression of anxiety, long half-life drugs, taken once a day, might actually be safer and as effective. In either case, if you are taking the drug daily or nightly for more than 3 to 4 weeks, the drug should probably be tapered off slowly when the patient wants to stop.

Dr Goldberg is a clinical professor of psychiatry at The Icahn School of Medicine at Mount Sinai in New York, NY and the immediate-past president of the American Society of Clinical Psychopharmacology.

Dr Salzman is a professor of psychiatry at Harvard Medical School in Boston.

References

1. Olfson M, King M, Schoenbaum M. Benzodiazepine use in the United States. JAMA Psychiatry. 2015;72(2):136-142.

2. Brunner E, Chen CA, Klein T, et al; Clinical Guideline Committee (CGC) Members; ASAM Staff and Contractors. Joint clinical practice guideline on benzodiazepine tapering: considerations when risks outweigh benefits. J Gen Intern Med. 2025;40(12):2814-2859.

3. Gomez AF, Barthel AL, Hofmann SG. Comparing the efficacy of benzodiazepines and serotonergic anti-depressants for adults with generalized anxiety disorder: a meta-analytic review. Expert Opin Pharmacother. 2018;19(8):883-894.