TCPR: How should we taper psychiatric meds?

Dr. Horowitz: I have three major principles. 1) Do it slowly, often months and sometimes more than a year for long-term users. 2) Do it at a rate that the patient can tolerate. Everyone’s a bit different, so there’s some trial and error here. 3) Go much slower at the end because small doses of psychiatric drugs have much larger effects than people would expect them to have. 

TCPR: Slower toward the end. That’s how we’re taught to taper benzodiazepines. 

Dr. Horowitz: Yes. It’s called a hyperbolic or proportional taper, and I think it applies to most psychiatric medications. So, you might start by lowering escitalopram by 10%–20% of the most recent dose, like a 5 mg reduction from 20 mg. But once you get to 5 mg/day, lower by 1 mg reductions, and even smaller right at the very end. Some people will need to go as slowly as 5% of their last dose a month.

TCPR: What is the basis for the hyperbolic taper?

Dr. Horowitz: On a neurobiological level, small doses of psychiatric drugs have much larger effects than you’d think, which produces a hyperbolic curve. At low doses, the receptors for that drug are mostly unsaturated—“open for business.” So every milligram of the drug has a large effect. At higher doses, the receptors are more saturated, and every milligram has less and less effect. For example, 2 mg of citalopram sounds like a homeopathic dose, but it has about half the effect of 20 mg on the serotonin transporter receptor (Horowitz MA and Taylor D, Lancet Psychiatry 2019;6(6):538–546).

TCPR: But 2 mg citalopram doesn’t have much clinical effect.

Dr. Horowitz: I’m not sure that dose has been specifically looked at in studies. But if we look at dose-response curves, where the dose is plotted against clinical effects, it follows the same hyperbola as the neurobiology. There’s a steep increase in the lower dose range, then it plateaus. That is true for most antidepressants, antipsychotics, and benzodiazepines (Leucht S et al, Am J Psychiatry 2020;177(4):342–353; Furukawa TA et al, Lancet Psychiatry 2019;6(7):601–609).

TCPR: Are there clinical studies of the hyperbolic taper?

Dr. Horowitz: A randomized trial is underway, but right now the taper is largely based on pharmacological principles and prospective cohort studies. The observational data are built on thousands of people who couldn’t come off their antidepressant using a traditional linear approach over a few weeks. In one of those studies, 71% of respondents were unable to come off their antidepressant with a linear taper. When they switched to a much more gradual hyperbolic taper over months, 72% were able to stop successfully. When patients were asked how severe their withdrawal was, they rated it very severe (7/7) with the traditional method, but much milder (2/7) with the hyperbolic taper (Groot PC and van Os J, Ther Adv Psychopharmacol 2021;11:20451253211039327).

TCPR: How do patients get doses like 1 mg of escitalopram?

Dr. Horowitz: For some doses, you can use pill splitters, which are available at most pharmacies. For smaller doses, use liquid formulations. When those aren’t available, compounding pharmacies can make specialized doses (eg, tapering strips). As a last resort—if professional compounding is not around—patients can liquefy their pills with a commercial product like Ora-Plus. Patients crush up their pills and emulsify them in this sugary liquid. A little math is required to make sure you dilute it in the right proportion, and the suspension should be carefully shaken before use. The NHS in the UK advises people who have trouble swallowing tablets on how to do this.

TCPR: Are there medications that can’t be crushed?

Dr. Horowitz: If you crush extended-release meds, you’ll often lose the extended-release effect but will still have the drug. Duloxetine won’t work if it is crushed. It comes in beads that are covered with a gastro-resistant coating, and if that coat breaks the drug is neutralized in the stomach acid. Unfortunately, for people who have trouble coming off duloxetine, their only option is to open up the capsules and count or weigh the beads. 

TCPR: Is there any logic to switching to another antidepressant for the taper? 

Dr. Horowitz: People have looked at switching to a similar medication with a longer half-life for the taper, like diazepam or fluoxetine. For benzodiazepines, the studies are equivocal (Denis C et al, Cochrane Database Syst Rev 2006;3:CD005194). I judge it by the individual. In some ways, it’s easier to stop a drug that they are accustomed to rather than introducing a new one. And fluoxetine is not “self-tapering” as advertised. It has a longer half-life than other drugs but not long enough for the months-or-longer taper that many longer-term users require, and about half of people experience withdrawal effects from it, which can be delayed in onset because of its half-life. When it comes to antidepressants, I generally don’t recommend switching. I think the selective serotonin reuptake inhibitors (SSRIs) as a class are more dissimilar than benzodiazepines are as a class, although they are all SSRIs. The serotonin/norepinephrine reuptake inhibitors (SNRIs) are even more dissimilar to fluoxetine because of the noradrenergic effects. 

TCPR: For some, these withdrawal protocols require a lot of effort.

Dr. Horowitz: Yes, and I think it is worthwhile. Textbooks describe withdrawal effects as mild and self-limiting, and they are, for people who stopped a medication after taking it for eight to 12 weeks (which is what the industry-sponsored trials focused on). But the longer patients are on a medication, the more their brain adapts to it. In those cases, withdrawal effects are often severe and can be disabling: intense headaches, suicidality, panic attacks, and sometimes akathisia. 

TCPR: Which meds cause akathisia during withdrawal?

Dr. Horowitz: It is well known during long-term antipsychotic use, but it can also happen when withdrawing from antidepressants, benzodiazepines, antipsychotics, and gabapentinoids. Patients are often pacing. They feel terrorized and restless. We did a survey of people who went to support groups for antidepressant withdrawal—so a very select group—and two in three experienced akathisia. It is often misdiagnosed as agitated depression or a manic state (Moncrieff J et al, J Aff Dis Reports 2024;16:100765). 

TCPR: You’ve talked about antidepressants. How does the hyperbolic taper relate to other meds? 

Dr. Horowitz: The basic principle is the same, where low doses have a proportionately larger effect than high doses. It’s called the law of mass action. It’s true for antipsychotics and their effect on D2 and 5HT2 receptors (Horowitz MA et al, Schizophr Bul 2021;47(4):1116–1129), and it’s true with benzodiazepines. Prof. Heather Ashton noticed that people could lower diazepam from 20 mg to 19 mg, but not from 5 mg to 4 mg (Ashton CH. Benzodiazepines: How They Work and How to Withdraw [aka The Ashton Manual]. Newcastle-upon-Tyne, England: Newcastle University, 2002). There’s a nonlinear relationship going on. Because it’s a general pharmacological principle, it actually applies to all psychiatric drugs and all their targets.

TCPR: How long do withdrawal effects tend to last? 

Dr. Horowitz: Long-lasting withdrawal effects are possible. It’s not the time it takes for the drug to leave the body, but rather the time it takes for the body to become used to the presence of less drug. That process takes longer: weeks, months, or even years. One effect of SSRI antidepressants is reduced sensitivity of serotonin receptors, and we still see that change for up to four years after stopping the antidepressant in neuroimaging studies of people who’ve been exposed to long-term antidepressants (Horowitz MA et al, CNS Drugs 2023;37(2):143–157). 

TCPR: Some have argued that these syndromes are not withdrawal but relapse, a return of the original condition.

Dr. Horowitz: Relapse is certainly possible after stopping treatment, but we know that withdrawal is also involved because we see these problems in people who were prescribed antidepressants for reasons other than mental health, like pain, menopause, and healthy volunteers. Among patients with psychiatric disorders, the withdrawal symptoms are often different from the symptoms they originally presented with (Moncrieff et al, 2024). Also, controlled studies find that the faster the med is stopped, the more likely a patient will experience problems (Baldessarini RJ et al, Am J Psychiatry 2010;167(8):934–941). Relapse, by contrast, should not depend on the rate that the med is stopped. Fast or slow, the underlying condition is going to reveal itself when the treatment is taken away. When we see a difference between abrupt and gradual withdrawal, it is due to withdrawal effects, not relapse.

TCPR: How do you tell the difference between withdrawal effects and return of the condition?

Dr. Horowitz: The first thing is to ask if the symptoms are different from the patient’s ­original symptoms. If they started an antidepressant for fatigue and depression and now—after stopping it—they’re having shooting pains, panic, insomnia, then it’s more likely a withdrawal effect. So I think that is a very big misunderstanding about how long these effects can last for and how severe they can be. 

TCPR: If it is withdrawal, do you restart the medication?

Dr. Horowitz: For acute withdrawal, like if the patient forgot their meds at the beach, restarting will quickly take care of the problem. But for people with protracted withdrawal (months of withdrawal), things are more difficult to predict. Restarting may work, but it may take a couple months for things to settle down. Then there are some who actually get worse on reinstatement. 

TCPR: So the longer the withdrawal symptoms have gone on, the harder they are to treat. 

Dr. Horowitz: Yes. We don’t understand protracted withdrawal well, but it’s as if there is some injury from coming off too quickly. The best approach is to prevent the problem, which is why I advocate for a long, hyperbolic taper. Otherwise, it’s very difficult to put the toothpaste back in the tube. 

TCPR: Withdrawal from benzos peaks, on average, one to two weeks after stopping them. What about ­antidepressants?

Dr. Horowitz: Often the symptoms are worse after a week or two, but there is a lot of variability. I have also seen people whose symptoms peak months after stopping. It may be that these kinds of protracted or delayed withdrawals are more common in people who have been on the drug for many years. In neuroimaging studies, antidepressants linger longer in the central nervous system than in the serum. It can take several weeks for the med to dissociate from the serotonin transporter (Sørensen A et al, Mol Psychiatry 2022;27(1):192–201). I wonder if that slower dissociation may explain delayed withdrawal effects. Or perhaps it is simply downstream effects that take time to accumulate (like the game Mouse Trap).

TCPR: With antidepressants, we worry most about withdrawal from SSRIs and SNRIs, but what about the other classes?

Dr. Horowitz: Withdrawal effects have been described for every antidepressant, from mirtazapine to monoamine oxidase inhibitors. Any drug that causes changes in the brain, adaptations that take longer to resolve than the drug takes to be removed from the body, will cause withdrawal effects (Reidenberg MM, J Pharmacol Exp Ther 2011;339:324–348). It’s driven by the law of homeostasis and persisting changes. When you are attending a loud concert, your eardrums become less sensitive to sound. When you walk out in the quiet street, your friends’ voices sound muffled for a few minutes—the time taken for your eardrums to relax. Likewise, if you increase levels of a neurotransmitter, the brain will become less sensitive to it. Then, when you remove the drug, the brain can take months or longer to go back to its “predrug” sensitivity (“factory settings”)—and the person experiences withdrawal for that period.

TCPR: What do we worry about with antipsychotic withdrawal?

Dr. Horowitz: There’s insomnia, headache, dizziness, but the big thing is withdrawal psychosis. Stopping antipsychotics can lead to a relapse of psychotic symptoms. This can sometimes be more than just a return of an underlying condition. There are cases of people with no psychiatric conditions who were prescribed dopamine antagonists like domperidone for nausea or for trouble lactating after giving birth (because antipsychotics can raise prolactin), and frank psychosis developed in some patients when they stopped abruptly: Capgras delusions, paranoid delusions, command hallucinations. The most common explanation is dopamine supersensitivity, which has been described since the 1970s (Moncrieff J, Acta Psychiatr Scand 2006;114(1):3–13). But antipsychotic withdrawal also causes insomnia, and insomnia can trigger psychosis. 

TCPR: I understand tardive dyskinesia can also worsen when an antipsychotic is stopped.

Dr. Horowitz: Yes, and that is another example of dopamine supersensitivity. The dopamine receptor becomes more sensitive when it is blocked by an antipsychotic. When that medication is taken away, the receptor gets exposed to more dopamine, which can cause withdrawal dyskinesias. 

TCPR: What do we worry about with anticonvulsive withdrawal?

Dr. Horowitz: Here the research is thinner. In my experience, people have less trouble coming off anticonvulsants than ­antidepressants, benzodiazepines and antipsychotics, but I have still seen people with trouble. 

TCPR: And with lithium?

Dr. Horowitz: With lithium, we have very good evidence of strong withdrawal effects. People who come off lithium have a seven-fold increased risk of mania or depression compared to their predrug baseline (Suppes T et al, Arch Gen Psychiatry 1991;48:1082–1088). In other words, before the drug was stopped, they had an episode every year, and now they have an episode within seven weeks. People need to come off very slowly. In controlled trials, patients did much better stopping it over four weeks instead of abruptly, but I think four weeks is still too fast. For long-term users, I would taper off over at least a year, with a hyperbolic pattern. 

TCPR: When should we consider deprescribing?

Dr. Horowitz: There are a lot of situations where you need to know how to stop a medication. 1) The harms, like sexual side effects, emotional numbing, and weight gain, may outweigh the benefits. 2) Patients may be on the medication longer than guidelines recommend. Most antidepressant guidelines recommend six to 12 months of treatment for a single episode of anxiety or depression. A lot of people start a medication in the context of an acute stressor and may be able to come off after the stressor has resolved. 3) Polypharmacy is another reason to consider stopping, especially as people get older where the additive side effects can become problematic.

TCPR: Thank you for your time, Dr. Horowitz.