(rewritten 3/2019)

Summary: Antidepressant withdrawal can last for many months in some people. Maybe a lot of people, that’s not clear. Tapering one’s dose very slowly, more slowly than commonly recommended, can make a huge difference.  While some people can go fast, stopping their antidepressant over a few weeks, many people will be better off going down over months, by very small steps.  Gently warn others about this antidepressant withdrawal risk.

Avoid withdrawal: taper slowly

Some people can get off antidepressants over a few weeks. Interestingly, we don’t really know how many can do this. We don’t hear from them! I worry that one of the reasons that over 10% of the U.S. adult population is taking an antidepressant (can you believe that?) is because so many have tried to stop and ran into antidepressant withdrawal problems.  In one study, withdrawal effects were reported by 55%, and addiction by 27%.

You’ve likely read about, and may already have experienced, the many weird symptoms that can arise when your antidepressant dose goes down. Here’s one way of looking at it, from a great recent review even if I don’t agree with their recommendations:

From Horowitz and Taylor, Lancet 2019
Antidepressant withdrawal symptoms

Some people even get these effects in between doses, or if they miss a few doses. Those folks are almost certain to have big problems with withdrawal when they try to taper off. Fortunately there’s a solution: taper slowly. But how slowly?

Slowly means 4 months; or longer

The paper I wrote back in 2011 suggested that it might take about 4 months, on average, for the brain chemistry to change from on-antidepressant to off-antidepressant, after the last dose of a rapid taper. This suggests that to avoid experiencing physical changes, one would need to make the taper around 4 months.

Two obvious goals of antidepressant withdrawal are:

  1. Get off the medication
  2. Do so without making things worse

These two goals determine the taper rate (in my opinion; there are others…). Go as fast as you can without getting too much worse. That varies for every patient.

Small steps

What doesn’t vary much is the benefit of using small steps. Maybe bigger at first and smaller later. Maybe small at first but going faster, then slower later.  I don’t agree with the authors of the “better paper” I’ve cited who suggest a “hyperbolic taper”: bigger steps at first, like cutting a big dose in half; then slower, and very slowly at the end.

If anything I’ve recommended almost the opposite: small steps at first to make sure that they’re almost certain not to turn you into a wreck.  For people who are really afraid of going down, those first steps must be tiny.  We brainstorm on how tiny my patient wants to make it: are they willing to chop pills? Shall we switch to a liquid version now, or later?

The skill in fussing over tiny doses can become really useful at the end of the process, because the reductions at the end need to be small to avoid big percentage drops.  Obvious, right? The drop from 20 mg of citalopram to 15 mg is one quarter of the dose. But the drop from 5 mg to 2.5 mg is one half of the dose. So at the end, the steps need to be about as small as you can make them.

But not everybody needs to fuss with a liquid version and an eyedropper. Many people can manage just by chopping the pills they have, especially if they can manage quartering them. Some antidepressants work this way, but some don’t.

Watch closely

You’ll want to make sure to identify withdrawal symptoms before you take another step down. If in doubt, usually better to wait.  Some people report withdrawal sensations coming on long after their last step down, but usually you’d seen them within 3-4 days if not sooner. The trick is not to take another step down lest you compound the withdrawal

Specific antidepressants

Without going through them all, here are a few that stick out:

  • fluoxetine/Prozac: has a very long lifetime in the bloodstream, like weeks for each pill. So when you go down, the dip in blood level is smoothed out over several weeks. Faster at first, then slowly approaching your new dose (“exponential” decrease). That’s why fluoxetine is one of the easiest to get off. It’ “autotapers”. Sometimes we’ll switch people off their antidepressant to fluoxetine, then taper the fluoxetine — the “Prozac bridge“, we used to call it.
  • paroxetine/Paxil: sort of the opposite of fluoxetine, with a very short lifetime in the bloodstream for many people. Thus tends to cause more severe antidepressant withdrawal.  Which means lots of pill chopping on the way down, to avoid problems. Fortunately there’s a liquid for the last few steps.
  • venlafaxine/Effexor:  is one of the hardest to manage, because it’s a capsule full of beads (over 100 even in the smallest capsule); and no liquid routinely available. If you’re on venlafaxine, I’d refer you to the user-specialists for venlafaxine at survivingantidepressants.org. Start at their front door, it’s a very supportive organization. Thousands of users, so you’ll have lots of company.

Three specific approaches

1. You’re Bold

You’re not afraid of withdrawal and you’re willing to have a few symptoms, maybe even bad ones, for a few days. Okay, you can go faster and learn whether you need to slow down. Still worth chopping your pills in half or maybe even quarters if you can manage it. And if you get even a hint of withdrawal symptoms, wait until they’re gone before taking another step. Make the next step smaller if you can.

2. You’re not so bold

You’re not so bold but you’re ready. Here’s my favored approach. Ask your prescribing provider for the smallest dose made, to use in addition to the ones you have now, as you go down.  (Look up the smallest dose for her as she may not know them all. )  You may be taking a big pill now but you’re gonna need those small pills for later steps anyway. If you don’t think so, you’re ready for approach #1 above!

Now, lower your dose by the smallest step you can. For example, say you were taking citalopram 30 mg. You get some 10 mg, the smallest made, and you chop them in half to make a 5 mg. Now go down to 25 mg, using whatever combination of your old pills and your new ones is needed.

If no withdrawal symptoms, great. Wait a week and try another step. If you prefer to be cautious, wait two weeks or even longer. If you do get withdrawal symptoms, it should be fairly mild because you made that first step so small.  Key idea: wait until those symptoms are gone for at least 4 days before you take another step.

If your withdrawal symptoms are severe, really intolerable, at any step down, you can go back up to where you just were one step before, and wait a week.  If you can make a smaller reduction than you’ve been making (e.g. now you’re ready to try quartering pills; or ready to get a liquid version), that would really help. If not, you can just wait longer in between steps down. How long? That depends on how bad it was and how fast you want to go.  If you don’t want to go through that again, wait as long as a month. If it happens again (should be milder this time), maybe next time wait two months. You get the idea…

3. You need some support

Fortunately there’s a great resource for you. At the same outfit I mentioned for the venlafaxine folks, survivingantidepressants.org, the brilliant moderator strongly advocates reducing your dose “10% per month“.  This will thus take you 10 months or so.  For some people, that’s exactly what they need to do. Start at the front door.

How long will antidepressant withdrawal last?

That varies. For some, the symptoms are minor and go away quickly, within days. For others, they last months. For a few, they can last a year or more, according to a 2018 survey.  Unfortunately, understanding those long withdrawals gets really tangled up with “return of symptoms” . Are these people going through a relapse into previous — or new — symptoms? Or is there a withdrawal component in there too?

Antidepressant withdrawal effects overlap with the symptoms of depression and anxiety (mood, sleep, cognitive, even physical and gastrointestinal) so much that sorting this out can be really tricky. This has led some physicians to write essays that appear — at least to those who’ve been through this — to minimize the problem, even though their analysis of the little data we have may be logical.

I’ll avoid adding to the public alarms that have been raised about antidepressant safety, in light of the large numbers of people having withdrawal problems. As will be obvious elsewhere on my website, I’m not a big fan of antidepressants. It would be easy to fan flames here. Not much use in that. As you can see, I’m sticking here to simply talking about getting off of them.


STOP HERE unless you want to dig much deeper.

The neurologic data

Some data from neurology research may be relevant to understanding antidepressant withdrawal in mood disorders.  I published a small paper on this in 2011.Phelps  Below you’ll find the paper’s ideas in plain English.  It’s still kind of technical. If you’re interested, give it a try.

Here’s the scenario. These data come from a study of patients with narcolepsy.Ristanovic  In that illness, people can have sudden episodes of falling asleep, during the daytime. They literally fall, sometimes, collapsing to the ground. Such episodes are called cataplexy.

Antidepressant medications can decrease the frequency of cataplexy in patients with narcolepsy. Thus when a new medication for cataplexy was being studied, it was necessary to taper patients off of any antidepressant they were already taking prior to starting the study medication. After the study was over, the authors recognized that something striking had happened even before the study medication was begun. When antidepressants were stopped, it took weeks, many weeks, for the impact of that medication withdrawal to finally subside.

Since antidepressants are thought to prevent cataplexy, it makes sense that the frequency of episodes would increase when the antidepressant was stopped. But as you can see in the graph below, the increase lasted a long time, suggesting that the physiologic changes associated with removing an antidepressant take months to stabilize, despite a 3-week tapering (which begins at the point labeled “Start”):



(The time scale represents 4 points in the study, not days, so the times shown are rough but accurate )

You may be asking “hey, wait a minute, they ended up back where they started. Doesn’t that mean the antidepressant wasn’t doing anything, before it was stopped?” The authors of this study raise that question too.

But that’s for the narcolepsy researchers to figure out. The point here is to recognize how long, physiologically, the effects of stopping an antidepressant persisted.

Bottom line: these data suggest that when an antidepressant is stopped, the brain response can take four months. Whether these data apply in bipolar disorder is an open question. But until now, I had never seen any data consistent with the idea that the process could take this long, even though I was recommending to patients that we should take four months to stop an antidepressant (based on one expert’s recommendation, and my subsequent clinical experience suggesting that he was right).

Another colleague helped list 31 guidelines for more slowly stopping antidepressants based on his experience, which matches mine.