The short version: SSRIs are a class of medications that affect emotions, used for depression, anxiety, anger, etc. My usual regimen is to start Lexapro 5 mg, go up to 10 mg after one week, and potentially go up to 20 mg later if that isn’t enough. The most common side effects are loss of sex drive and decreased emotional range; other side effects include tiredness, wiredness, weight gain, etc. You can go into withdrawal if you stop SSRIs, especially if you stop them suddenly. A good doctor can help you avoid or manage this withdrawal, and you shouldn’t let it scare you out of taking SSRIs if you need them.

The long version:

1. What do SSRIs do?

SSRIs affect emotional processing.

The usual narrative is that SSRIs are “antidepressants”. I think that’s an oversimplification. SSRIs have no particular relationship to depression – they’re approved for anxiety, social anxiety, OCD, panic disorder, eating disorders, mood issues related to menstruation, etc, etc. Whatever’s going on with your emotions, there’s a pretty good chance that SSRIs will have some kind of effect.

Too vague? I’m tempted to describe SSRIs as blunting emotions, especially strong emotions and negative emotions. This explains a lot of their use – they blunt anger, blunt anxiety, blunt panic, blunt depression. It also explains a lot of their side effects – they blunt libido and sexual desire, and if you take too high a dose they blunt everything and leave you feeling emotionless. But this isn’t exactly right. Some people whose depression feels like an emotionless gray cloud say that SSRIs bring them back to feeling normal. Other people say SSRIs affect depression symptoms which are completely unrelated to emotion.

My guess is that if we genuinely understood SSRIs, we would think of them as doing something underneath the surface which is almost, but not quite, the same thing as blunting emotions. Some people, when confronted with too many negative emotions to bear, reactively change their emotional processing style so that they can’t feel anything at all. This is a good strategy for avoiding negative emotion, but leaves them unmotivated and depressed. SSRIs blunt the negative emotions, and after the brain realizes it’s not overwhelmed by negative emotions anymore it’s more willing to “come out of its shell” and start trying to feel normally. Probably a better understanding of the dynamical systems perspective on depression would help us understand this better.

For now, a simple model might be that SSRIs blunt negative emotions, may sometimes blunt positive emotions at high doses and in susceptible individuals, and do complicated things downstream of that which sometimes help people recover from depression, even if the depression doesn’t “feel” especially emotion-laden.

I use this nonstandard explanation of SSRI effects because it helps cut through some of the questions that bog people down. The most common question is “How can I be sure I really have real depression or a real anxiety disorder?” This question is meaningless; people don’t neatly separate into two groups, “has an anxiety disorder” and “doesn’t have an anxiety disorder”. There’s just a spectrum of people with more or fewer or different anxiety symptoms. If you have enough anxiety that it’s significantly interfering with your life, and you don’t have any good way to get rid of whatever’s causing your anxiety, then you might want to try an SSRI to blunt your negative emotions.

A lot of other people get bogged down in the question of whether their depression/anxiety is “chemical” or because of “life events” (in technical terms, “endogenous” vs. “exogenous”). There is currently no strong evidence for or against the claim that SSRIs treat endogenous depression better than they treat exogenous (see 1, 2, 3, 4, 5, 6 for various conflicting claims), and current guidelines don’t distinguish between them. Some people are confused to hear that medications can help with problems caused by life events. But this shouldn’t be so surprising – alcohol, marijuana, and other recreational drugs also (temporarily) relieve anxiety, regardless of cause. There are many specific differences between these drugs and SSRIs, but the basic principle is the same. If you have some negative emotions, for whatever reason, SSRIs will blunt them.

In case I’m sounding like a pharma company shill, keep in mind that you don’t always want your emotions blunted. Some negative emotions are useful – normal physiological anxiety is your brain’s way of telling you not to poke lions with sticks. And at very high doses, SSRIs can blunt positive emotions along with the negative. People at these doses tend to describe themselves as feeling like zombies or robots, and as not being able to feel or enjoy anything.

But if you’re having extreme, intolerable negative emotions, and you want something to reduce their intensity, SSRIs may be a good choice for you.

2. How do SSRIs work?

The short answer is: nobody knows!

SSRIs block the serotonin transporter, a protein that takes serotonin out of synapses (the gaps between neurons). That means serotonin stays in the synapses longer. That means they raise the effective amount of serotonin in your brain.

And that blunts negative emotions? Unclear. For one thing, SSRIs block the serotonin transporter instantly, but SSRIs take about a month to work. One theory is that it takes a month to desensitize serotonin autoreceptors, another kind of protein that would otherwise adjust for this effect and cancel it out. Another theory is that this works through a much more complicated cascade of actions where serotonin, over the course of weeks, upregulates a chemical called BDNF, which causes the creation of new synapses throughout the brain. Probably this is good in some way, though we don’t yet understand exactly how.

All of this is still very preliminary. We know that serotonin must be involved in mood somehow, because there are dozens of different antidepressants whose only commonality is that they all affect serotonin (or closely related monoamines like norepinephrine). And we know that deliberately depleting lab animals and humans of serotonin makes them depressed. But right now we can’t say much more than that.

3. How effective are SSRIs?

In 1998, Irving Kirsch published a meta-analysis showing that most benefits of antidepressants were due to placebo. In 2008, he published a better, more complete meta-analysis showing the same thing. In 2018, a team led by Andrea Cipriani published its own meta-analysis, and although the statistics are confusing and nobody really remarked on this at the time, it mostly confirmed Kirsch’s estimate of antidepressant efficacy.

Let’s go over exactly what was found. Kirsch reports his statistics in effect size. Antidepressants, on average, had an effect size of 1.2. But on closer inspection, about 0.9 of that was placebo, and only 0.3 of that was the drug itself. A lot of people had a lot of debate over whether those numbers were exactly accurate, but after Cipriani’s contribution I think it’s fair to say that they are close.

Kirsch argued that this should turn us against antidepressants. Conventionally, effect sizes of 0.2 are considered “small”, 0.5 “medium”, and 0.8 “large”. 0.3 is small, and in fact so small it’s probably clinically insignificant – the average patient (or doctor) might not even notice it. So although antidepressants themselves have a large effect size (1.2), most of that is placebo, and the drug itself is so small we shouldn’t care about it very much.

I have three qualms with Kirsch’s analysis.

First, antidepressants tend to have smaller effect sizes when measured on formal instruments (as Kirsch’s studies did) than when measured by patient preference. That is, there are a bunch of tests that ask you a bunch of questions about your feelings and symptoms, and you can add them up and call that a “depression score”, and if you do that, antidepressants have an effect size of 0.3. Or you can ask patients “how depressed do you feel on a scale of 1-10”, and if you do that, antidepressants have an effect size of 0.5. I think the latter is better, because it’s what we actually care about (how patients are doing), and the tests are kind of dumb and ask about a lot of symptoms most people realistically aren’t experiencing. So plausibly we should think of antidepressants as having an effect size of 0.5, which is at least “medium”.

But second, antidepressants aren’t a one-size-fits-all solution. Suppose 50% of patients are “responders” and 50% are “nonresponders” (source: personal experience; this study gives similar numbers but this sort of thing is very hard to operationalize and I will just go with personal experience). If the total population has an average effect size of 0.5, then responders must have an average effect size of 1.0. Nobody ever claimed SSRIs work for everyone – just that in the people who they work for, they do a good job. As for the other people, they stop their medication trial after a month and try something else, which will hopefully work for them. The situation is actually even more complex than this, because a small minority of patients do worse on SSRIs. These patients bring down everyone’s average, and then studies find that “on average” “participants” only get an effect size of 0.5. Fine, but in real life the people who felt worse on SSRIs would have stopped them immediately.

And third, if you count the placebo effect, Kirsch’s studies showed SSRIs work super great. Of course, it’s kind of embarrassing to be relying on placebo effects, especially when you’re giving an active drug with real side effects. But do you have any better ideas? I can’t actually give out sugar pills; they’d take away my license. And you can’t voluntarily take sugar pills and pretend they’re antidepressants, because you know what you’re doing.

So my reading of Kirsch is something like: antidepressants will work for about 50% of people. For those people, they will have a large real effect size of 1.0, plus a large placebo effect size of 0.9, for a very large total effect size of 1.9.

3.1: Are antidepressants just “active placebos”?

Kirsch also added that maybe the apparent non-placebo effect of SSRIs was just because the drugs were “active placebos” as opposed to the “passive placebo” of a sugar pill. That is, the drugs caused real side effects, which made people realize they were taking an active medication, making it feel “more real” than a simple placebo and ruining blinding.

But this time he’s just wrong. See eg Quitkin et all (2000), which finds that giving placebos with side effects doesn’t increase placebo response rate, and Hieronymous et al (2017), which finds the same real antidepressant efficacy even in the subgroup of patients who have no side effects.

4. Which SSRI is the best?

Different SSRIs have different advantages and disadvantages, and some people will do better on one than another for mysterious reasons – but unless you have some specific reason otherwise, I would start with escitalopram (Lexapro).

Here’s a table of various rankings of five common SSRIs (a sixth SSRI, fluvoxamine, isn’t used often enough to show up in most rankings).

Medication AskAPatient WebMD SSC Cipriani
Fluoxetine 7.1 3.3 3.32 5.5 1.52
Escitalopram 7.2 3.3 3.38 5.7 1.68
Citalopram 7.2 3.3 3.39 5.4 1.52
Sertraline 6.9 3.3 3.35 5.6 1.67
Paroxetine 6.4 2.9 3.00 4.6 1.75, AskAPatient, and WebMD are all sites that ask patients to rate their drugs (the first 1-10, the second two 1-5). SSC is the SSC survey, where 8000 blog readers rated the medications they’d taken 1-10. Cipriani is the odds ratio found in Cipriani’s meta-analysis of antidepressant effectiveness.

All five rankings find that most SSRIs are very similar to each other, except that the first four find that paroxetine is much worse than the others. Cipriani doesn’t find this, probably because he is only ranking efficacy, and the problem with paroxetine isn’t that it works less well, it’s that it has many more side effects.

If you look at all the rankings together, escitalopram looks like it probably wins by a hair. This matches my impression of psychiatric consensus and what patients usually think.

When might you start with an SSRI other than escitalopram? Fluoxetine has a reputation of being more stimulating than other SSRIs, so if you’re really tired all the time, it might be a good choice (and if you already can’t sleep, it might be a bad one). It’s also less likely to cause withdrawal on discontinuation than other SSRIs, so it might be a good choice if you’re really worried about that. Finally, medication response is very genetically linked, so if your mother or brother or cousin did well on one of the SSRIs, you probably will do well on that one too and it should be the first one you try.

There are some genetic tests that purport to tell you which SSRI will work best for you. I’m very skeptical of these, but if you’ve already had them for another reason, you might as well consider what they have to say.

5. What side effects might I get on an SSRI?

Your first week taking SSRIs, as your body adjusts, you’ll probably get various annoying side effects including nausea, stomachache, diarrhea, or headache. These usually go away after three days to a week, as your body adjusts.

After a few weeks, when the medication is starting to work, about 50% to 70% of people will notice some longer-term side effects. Different surveys say different things about exactly how common these are – I’m going to go with my own, which found (note that “has severe” numbers are included in “has at all”, so you don’t need to add the columns up):

Side effect % has at all % has severe
Sexual difficulties 52% 11%
Emotional blunting 39% 8%
Fatigue 24% 6%
Cognitive problems 19% 3%
Makes depression worse 7% 2%
Weight gain ? ?

Other surveys found lower rates, probably because they waited for patients to name them instead of asking specifically, or because they only counted side effects patients rated as bothersome.

Sexual difficulties are by far the most common side effect, but the specific difficulty various from person to person. It can be any or all of decreased libido, difficulty orgasming, difficulty getting an erection, difficulty enjoying sex, or decreased sensation in the genitals. These usually go away a few weeks to months after stopping the medication, but in rare cases they might linger for months or years, and there are a few people who say their sexual side effects never went away. These cases are very unusual and still not well understood.

When everything goes right, SSRIs blunt negative but not positive emotions. But many people even at reasonable doses will notice that their most extreme positive emotions become a little less extreme (this may be part of the problem with sex). At higher doses, or in unlucky people, all emotions including positive emotions may feel washed-out and gray. This isn’t part of the expected medication reponse, and usually means you need to be on a lower dose. Some people will not be able to find a dose that works for them without emotional blunting, and these people may not want to stay on SSRIs.

Fatigue and cognitive problems are less common. I tend to associate cognitive problems mostly with paroxetine, which is famous for them, and you can usually avoid them by not choosing that medication.

I don’t have good data on weight gain, but I agree with the unnamed experts who guess 25%. Some optimists argue that since one symptom of depression can be under-eating, this is just the medication working as intended, but a lot of patients don’t feel that way.

That’s the bad news. The good news is – you can avoid a lot of these side effects by not taking paroxetine, the worst offender. And if you have side effects on one SSRI, you can try switching to another; not all SSRIs will cause the same side effects in the same people. If you have the same side effect on two SSRIs, it might not be worth trying a third.

6. Are there any long-term dangers of SSRIs?

This is a hard question and we still don’t know the answer.

We can’t do randomized experiments where we make people take SSRIs for decades. So the people who take SSRIs for decades tend to be people with a lot of depression and anxiety. These people predictably do worse in a lot of ways. Are they doing worse because they’re taking SSRIs for decades, or because of their depression and anxiety? Unclear, and the various statistical tricks people use to try to get around this problem don’t inspire confidence.

There are various conflicting studies showing that SSRIs increase or decrease risk of various conditions. How does it all balance out? Maslej et al investigate the effects of antidepressants on all-cause mortality. The results pose an interpretational challenge. One reasonable summary would be that overall, there was no effect on mortality. But the authors argue this hides a deeper pattern; the medications had no effect on (or decreased) mortality in cardiovascular patients, and increased it in everyone else. This makes a sort of sense; SSRIs probably help in cardiovascular disease because they decrease blood clotting, but they might make other conditions worse. See here and here for debate on these issues.

If we take Maslej’s results at face value, non-cardiovascular patients could expect to lose up to three years of life on average if they were on SSRIs forever. Should you be concerned? My position is probably not. For one thing, you don’t know whether or not you will have cardiovascular problems when you’re older, and the sample of everyone (cardiovascular patients plus noncardiovascular patients) neither gained nor lost mortality due to SSRIs. For another thing, probably this is less about some effect of SSRIs that builds up over your entire life, and more about whether or not you’re on SSRIs at the specific moment where you have some other condition – so healthy young people don’t have to worry except insofar as their decisions now affect whether they’ll still be taking SSRIs when they’re older and sicker. Most important, either you don’t need to be on SSRIs forever, or you do need to be on SSRIs forever. That is – you’ll try an SSRI. If it doesn’t work, you’ll come off it, no harm done. If it does work, you’ll stay on it for a few months while you work on your depression, then try coming off it. The only people who will be on SSRIs forever are people with severe mental health issues, who feel much better on an SSRI, and nothing else works for them. If that’s you, probably an SSRI is worth whatever mortality risk it gives you.

7. How hard are SSRIs to come off of?

SSRI withdrawal is definitely real, and can sometimes be severe, but I would warn people against listening to the horror stories you’ve probably seen online or in the news.

(Problems coming off SSRIs are usually called “discontinuation syndrome”, because some people decided “withdrawal” sounds drug-related and stigmatizing. But biologically it’s a withdrawal process, and to avoid confusion I’ll be referring to it as such.)

In my survey, about 60% of people who came off SSRIs reported no withdrawal, 35% reported mild-to-moderate withdrawal, and 5% reported severe withdrawal. Other studies find somewhat different numbers – see here for one particularly large and controversial one – but my experience supervising maybe a hundred SSRI tapers backs up my survey. Withdrawal symptoms are diverse and hard-to-categorize, with different people noticing different things, but among the most common are itching, dizziness, nausea, fatigue, muscle aches, and chills. Some people also get “brain zaps”, a weird feeling of occasional non-painful (but extremely distracting) electric-shock-like sensations in their head. On average these symptoms last a few days, but some people will have them for weeks or even months.

Risk factors for SSRI withdrawal include: being on high-dose SSRIs, being on SSRIs for a long time, being on paroxetine, and coming off an SSRI too quickly. A person who stops paroxetine cold turkey after ten years will almost definitely have severe withdrawal for weeks or months. A person who stops escitalopram in a gradual taper after three months will almost definitely have little or no problem.

Given that this is real and potentially very unpleasant, why do I argue against listening to the horror stories?

First, because it’s very hard to experience SSRI withdrawal if you have an adequate taper. One one level this is a circular argument, since a taper that causes withdrawal wasn’t adequate. On another level, I think it’s true and should guide practice. I talk to patients about how they want to balance the risks and benefits of a shorter vs. longer taper, and we usually agree on a medium-length one. If they try the medium-length one and start experiencing withdrawal, then we back up to whatever the last dose they didn’t have withdrawal on was, and restart from there with a slower taper. If the slower taper still causes symptoms, we try an even slower one, until we get it right. So far this process has never failed me, and it’s hard for me to understand what it would even mean for it to fail. This paper has some great discussion of the principles behind very long tapers, and serves as my taper of last resort if everything else goes badly. I have never had to go all the way to the taper described in that paper; so far everyone has done okay with something less complicated.

(my taper of first resort is something like the schedule shown here with two weeks between steps, a little slower on paroxetine)

Second, because a short trial of SSRIs is probably worth it. It takes a long time – usually many months or years – before your body becomes so dependent on SSRIs that it’s likely to suffer severe withdrawal from them. So go on an SSRI and see if it treats your symptoms. If it doesn’t work, you can go off it after a month or two and not worry about any of this. If it does work, then the relevant question becomes whether a chance of SSRI withdrawal is better or worse than depression (for most people, the answer is better).

8. What’s the right dose for an SSRI?

Different for different people. Using this paper and normal clinical practice as a source, here are some rules of thumb (all in mg):

Medication Low dose Standard dose High dose
Fluoxetine 20 40 80
Paroxetine 20 40 60
Sertraline 50 100 200
Fluvoxamine 50 100 300
Citalopram 10 20 40
Escitalopram 5 10 20

In most cases, the dose marked “high” here is the maximum dose for which the medication has FDA approval.

I usually tell patients to start on half the low dose for one week to let their body adjust, then go up to the low dose and stay there for a month. At the end of the month, if they’re feeling even slightly better, I ask them to try going to the medium dose for a month. If they confirm that going up to the medium dose made them feel even better, and they’re still not where they want to be, I ask them to try the high dose. If at any point a dose escalation doesn’t improve things, they go back to the previous dose.

This paper speculates on SSRI doses, saying that:

In some cases this matches my table above; in other cases it doesn’t – taken seriously, it would recommend prescribing escitalopram at over three times the FDA-recommended maximum dose. Needless to say I don’t do this, though I sometimes ignore the official 20 mg cap of escitalopram in favor of an unofficial 30 mg one. I think if this paper were right, we would see escitalopram being much less effective than other antidepressants (since we’re currently prescribing it at only a third the right dose), whereas in fact we see the opposite (see above, where escitalopram usually gets the highest patient ratings – although someone who was against SSRIs could draw the opposite conclusion here and argue that maybe people like escitalopram best because we consistently underdose it!).

There is some evidence that obsessive compulsive disorders respond well to higher SSRI doses than other conditions, and some authorities suggest going above the dose marked “high” above for treatment-refractory OCD patients. This is rarely dangerous (high-doses of some antidepressants can raise the risk of a heart condition called “QT prolongation”, but this is generally overblown except in a few very specific people especially at risk of cardiac issues) but going above the maximum dose significantly raises the risk of usual side effects like sexual issues or emotional blunting.

9. How long will I need to be on an SSRI?

SSRIs do not permanently cure any condition. At best, they suppress the condition while you are on them. So you will need to be on an SSRI until your condition goes away, or you feel ready to deal with it without medication.

The average length of a major depressive episode is between three months and a year, with most sources saying six months. Current guidelines state that if SSRIs treat your depressive episode, you should stay on SSRIs for about six months, then try stopping the SSRIs to see if your depression has gone away on its own. This suggests some picture of depression episodes “continuing to go on” “below the surface” even when SSRIs have rendered them asymptomatic; I don’t know if anyone actually believes this or whether this is a productive way to think about things. But I think the six-month suggestion is a good rule of thumb.

If you stop SSRIs after six months and immediately feel depressed again, then probably you should try again after another six months to a year. If you keep trying this, and every time you get off the SSRI you get depressed again and again, you may need to stay on the SSRI until something changes, or even permanently.

If you’re taking SSRIs to deal with depression, anxiety, or other negative emotions caused by life circumstances, you should probably match the course of the SSRIs to the course of your life circumstances. For example, if you’re anxious because of your terrible job, and SSRIs help relieve that anxiety, you should probably stay on them until you can get a less terrible job. After that, you can try stopping them and seeing if your anxiety is better.

If you’re taking SSRIs for a trait condition like trait anxiety, trait dysthymia, or OCD – a condition you’ve had most of your life and which just seems to be a part of how you are – then you’ll need to stay on SSRIs unless you find some other way to deal with the trait condition. IE if you’re just naturally an extremely anxious person, and you need SSRIs to be able to function without your anxiety getting in the way, you will need to stay on SSRIs until you find some other way of dealing with your anxiety (lifestyle changes? therapy?)

Some people stay on SSRIs forever. There’s no shame in this, and there may not be any health risks; see the section on side effects for more details.

10. Will an SSRI change me into a different person?

Sometimes people ask this question from a philosophical perspective, so I’ll start with a philosophy-style answer: does caffeine change you into a different person? Does alcohol? Does stress? Does depression? Does pregnancy? Does the birth control pill? Does falling in love? All of these change your brain chemistry, some of them in profound ways. Some of them change how you act, pretty noticeably. Does that mean you’re a different person? Read Derek Parfit for a good answer; my mediocre answer is something like “these things make you act differently but don’t make you feel a rupture in your sense of continuity with yourself.”

SSRIs increase the amount of serotonin in your brain. But so does going out in the sunlight. I’m not trying to trivialize going out in the sunlight – I think, feel, and act very differently after a long walk in the sun compared to being shut up in my room with the windows down all day. But I don’t philosophically agonize over this difference and I don’t think you should philosophically agonize over whatever differences SSRIs cause.

Other times people ask this question from a more practical perspective: will I think and act so differently that it will cause me problems, or at least feel creepy? I think the answer is a guarded “no”. Guarded because there is one very real example of this happening – the emotional blunting some people get on high doses of SSRIs. I think this is mostly avoidable. I think most people won’t notice any creepy changes in their thought processes or sense of self besides that one.

I was on SSRIs for about five years as a child. In retrospect, I had moderate emotional blunting. At the time, I didn’t realize this, because I was young, I had limited experience with normal emotions, and it just seemed like I was an unusually calm person. I think I made the right choice in going on them when I did and the right choice in going off them when I did. If there’s one thing I want people to take away from this essay, it’s that trying SSRIs is an experiment, not a commitment. SSRIs won’t change your personality in a way you can’t get back from. If you’re on the fence, then try them, see if the benefits outweigh the risks, and make an informed decision.