Antidepressants in Bipolar II: What the Experts Do

Table of Contents

We update the research on antidepressants in bipolar disorder, and 18 experts weigh in on how they use these medications in bipolar II disorder in this podcast.

Listen online or search for “Carlat” in your podcast store

In this month’s Carlat Report we featured a review of a new trial by Nassir Ghaemi on antidepressants in bipolar disorder, and in this companion podcast we’re going to look at what actually goes on behind closed doors in the practices of bipolar experts when they turn to the antidepressant page of their pharmacopeia. First, a little more background.

What the Research Shows

Most of what we know about antidepressants in bipolar disorder comes from studies in bipolar I disorder, and there the answer is clear: Antidepressants should be avoided because they can cause mania. That risk is lower if they are given with an anti-manic mood stabilizer – in other words any mood stabilizer except lamotrigine. This idea was debated for a long time, as some argued that we didn’t have good enough data to say with certainty that antidepressants caused mania. In my mind the question was settled around 2014 when a large study by Alexander Viktorin and colleagues showed the risk of mania was 3-times higher when antidepressants were used on their own – as monotherapy – without a mood stabilizer.

I don’t have any hard evidence that that article put an end to the debate on whether antidepressants cause mania, but here’s what I’ve seen. Lecturers on the subject often ask the audience for a show of hands on whether they think antidepressants are safe in bipolar. Before 2014, a lot of hands were raised in defense of their safety. After 2014, it’s a small minority – fewer than 10% – who raise their hands in support. The editorial that accompanied Victorin’s article put the nail in the coffin – it was titled “Antidepressants in Bipolar I: Never as Monotherapy

KELLIE NEWSOME: So that says never in Bipolar One. What about bipolar II?

CHRIS AIKEN: Viktorin wasn’t able to parse out bipolar I from II in his dataset, so that leaves a big question mark. We do know from other research that antidepressants are safer in bipolar II than bipolar I, but outside of that it’s not clear just how helpful or hurtful antidepressants are in bipolar II disorder. Getting back to pharmacologic dissection, let’s remember that that was used to separate bipolar I from unipolar disorder in 1980, and even there the split was not completely successful. Bipolar II was carved out of the unipolar side in 1994, and there are no FDA approved treatments specifically for bipolar II. Quetiapine – Seroquel – is the medication with the most data in bipolar II – and it worked there for bipolar II depression.  In contrast cariprazine – Vraylar – did not work in the bipolar II studies, so its FDA approval in bipolar depression should be specified as bipolar I depression.

A Few Surprises With Lithium

KELLIE NEWSOME: And what about lithium – isn’t that more for bipolar I because it’s better at treating mania than depression?

CHRIS AIKEN: What I love about bipolar research is that many ideas that seem intuitive – like what you just said about lithium – are contradicted by the research. I can think of one large but non-randomized study where lithium worked better in bipolar II depression than it did in bipolar I depression, and the differences were not trivial. Compared to bipolar I, those with bipolar II were twice as likely to fully remit with lithium and those with bipolar II stayed well 6 times longer on lithium. My take on all the data is that lithium works great in about 30% of people with bipolar – and whether they have bipolar I or bipolar II does not tell you who is going to respond. In our September interview with Dr. Rybakowski we listed the signs of a good lithium responder, and the main one is having a tendency toward classic, euphoric manias or hypomanias – and that can be true in bipolar I or bipolar II.

KELLIE NEWSOME: This is all new to me. I was taught that lithium is much more effective against mania than depression, and depression is much more common in bipolar II than bipolar I.

CHRIS AIKEN: What you said is true in the short term – lithium’s short-term effects in depression don’t always hit the big effect sizes. But when it comes to long-term prevention, lithium had one of the highest effect sizes against both the manic and depressive pole – it tied with quetiapine for that prize. Here’s a tip – never judge lithium by its short-term effects. If it works, it’s main benefits are in the long-term prevention of new episodes. And this 1998 study I’m quoting from Leonardo Tondo and Ross Baldessarini was long-term, with an average follow up of 8 years.  Those benefits are so delayed that you’re likely to miss them in practice, and your patients may only recall their initial response to lithium and tell you it did nothing for them. So you need to use some sort of long-term mood measure to pick up on this. It could be a simple as “rate your depression from 1 to 10.”  What I often see is that patients will say that lithium did nothing, but their mood scores are consistently improved when they are on it – and worse off it.

3 Ways That Antidepressants Worsen Mood

KELLIE NEWSOME: That’s just as important when using antidepressants in bipolar. If your patient gets worse soon after starting an antidepressant, you’re going to notice. They may call you in a full-blown mania, travelling across the country and spending outrageously at every stop, but more likely they will just say they feel worse – more depressed, more anxious, and when you get into the details what emerges is a mixed state. They are tired and wired, driven to do something but don’t know what to do; they are agitated they can’t sit still; their thoughts are racing but they are racing with depressive and anxious themes. Their sleep is all over the place and they are irritable and demanding with their family. In practice, antidepressants are more likely to cause mixed states than overt mania because they just aren’t powerful enough to flip someone from full depression to full mania. Instead, they simply don’t help the depression, and they sprinkle manic symptoms on top of it.

Anyway, you’re probably going to pick up on those changes, but unless you are doing long-term mood charting you may miss rapid cycling on antidepressants, which is the main risk that Dr. Ghaemi’s new study identified. Here’s what it looks like in practice. A patient with bipolar disorder gets depressed. At first, they wait, hoping it will go away on its own. 3 weeks go by, then they call to see you and they get in a week later. You start an antidepressant – venlafaxine Effexor – and tell them it’s going to take 3-6 weeks to work. And like clockwork, 4 weeks later it works. Their depression lifts. But take note of the timeline: The episode lifted 2 months after it began, and that’s about how long the average bipolar depression episode lasts, so this could have been the natural course of things and not the antidepressant. But how are you going to know that? So you continue the antidepressant.

A few months later the same patient calls you. He is depressed again, and says, “Venlafaxine worked so well last time, can we just raise it?” So you raise it from 75mg to 150mg, and low and behold he gets better. Then depression strikes back, this time 6 weeks later. This time he says he knows the cause – there’s been a lot of conflict between him and his wife the past month. Still, he asks if he can raise the dose to help him get through the conflict, and heck it’s only at 150 so you go up to 300mg a day.

The next time you get a call it’s not from the patient. It’s from his wife. “I don’t know what he’s telling you, but he has not been the same since he started that medication. Everything upsets him, and we’re all walking on eggshells around him. His temper is just unbearable, and I had to move out with the kids because he started smashing all the dinner plates when they were making too much noise in their bedroom.”

What has happened is that the medication never worked to begin with. Like I suggested, the patient got better from the natural course of illness, or maybe the antidepressant induced a bit of mild hypomania that he didn’t think to inform you of. But what it did do was start a process of rapid cycling that unfolded over the next year. Every time the dose was raised, he got better, but that’s just because the episodes were getting shorter – he calls you at the peak of the episode, and then cycles out of it. And meanwhile the antidepressant is making the episodes come on faster and faster. First every 3 months, then every 6 weeks. Finally they culminate in a mixed state – the cycling is so fast that the mania and depression are overlapping together, causing him to do impulsive but destructive things that frighten his family.

CHRIS AIKEN: But unless you have a razor-sharp memory you’re not going to pick up on those subtleties of frequency as these events play out over a year. And patients almost never come in complaining about frequency – they complain about their current symptoms; they aren’t as attuned to whether the time interval between their episodes is shortening or widening.

KELLIE NEWSOME: The best way to spot rapid cycling is to use a mood chart. I don’t recommend the daily mood charts, because it’s too much of a burden for patients to fill those out every day, and you don’t need that kind of micro level of detail. Instead, I have them do a mood rating scale once a week and chart that – this way their rating is more accurate because they are rating specific mood symptoms instead of their emotions, and the weekly figures will paint a clear picture after about 6 months – and in mood chart terms anything less than 6 months is just a snapshot and not very useful. Dr. Aiken keeps a weekly version on his website.

CHRIS AIKEN: I’ll tell patients that this is more accurate than a $20,000 brain scan in predicting their medication response, which is true, but if they don’t do it you can track mood at each visit to get a sense of the cycling. And if that fails, here’s a tip. Suspect rapid cycling in bipolar disorder when the patient comes in and it seems like every other visit they are better and every other visit they are worse – it’s the roulette wheel sign – often what’s going on is that they are cycling up and down on their own regardless of what treatments you are using.

Back to the Research

KELLIE NEWSOME: Before we get into the expert opinions, let’s summarize what we know from the research a little further. Most of the studies on antidepressants in bipolar disorder are small, and even when they are controlled the control group is usually another medication instead of a placebo. That’s the case with most of Jay Amersdam’s studies, and most of the studies showing good effects with antidepressants in bipolar II disorder come from his group. Like this one, which showed that venlafaxine worked just as well as lithium in bipolar depression. But it’s not uncommon for lithium to work just as well as a placebo in a small, short-term controlled study like that, so this comparison really doesn’t tell us much. So far the most definitive study we have comes from Gary Sach’s group at Mass General. In 2007 they did a large randomized placebo controlled trial of antidepressant vs. placebo in patients with bipolar I and II depression. For the antidepressant, they used bupropion (Wellbutrin) or paroxetine (Paxil), because at the time these two were thought to be the most favorable in bipolar disorder. The bottom line of this study was that the antidepressant did not work – it did no better than placebo – but then again it didn’t cause mood problems either, but keep in mind it was short term and everyone was taking a solid mood stabilizer.

This study was also able to zoom in on the bipolar II patients, and the antidepressants didn’t look any better in that group, if anything they did worse – although the difference was not statistically significant. So Dr. Aiken, if antidepressants didn’t work in bipolar I or II depression, why do people keep using them in bipolar disorder?

CHRIS AIKEN:  Mixed in the group of bipolar patients, the group that is not supposed to respond to antidepressants, are a small handful of bipolar patients who do respond to antidepressants. Just how small is that handful? Let’s look at the meta-analyses. There are 4 in the past decade – from 2008, 2011, 2016, and 2017 – each involved about 6-7 trials in bipolar depression. One concluded that the antidepressants did not work. The other 3 found some benefit, but in each case that benefit was so small that it probably wouldn’t have passed the FDA – only about one in 12 responded to them, and the overall effect size of 0.17. But the studies that found this small benefit also found harm – mania, mixed states, and rapid cycling – and the chance of causing harm was greater than the chance of doing good.

What the Experts Do

KELLIE NEWSOME: Now that we’ve clarified the data, let’s look at what the experts do in practice. Gordon Parker asked 18 experts whether and how they used antidepressants in bipolar II disorder. Keep in mind this is just about bipolar II – most experts recommend against antidepressants in bipolar I. The results were published in a 2019 textbook on Bipolar II. Dr. Aiken contributed to the textbook and he analyzed the 18 responses. Here’s what he found:

CHRIS AIKEN: The responses fell into 4 categories, based on whether the expert viewed antidepressants as helpful, harmful, or some mix of the two.

At the extremes were two experts who thought that antidepressants were mostly helpful or mostly harmful in bipolar II, so these were both minorities. One felt that they helped bipolar II and did not cause hypomania. That was Jay Amsterdam from the University of Pennsylvania in the US, who published many of the favorable trials of antidepressants in bipolar disorder. But not all of his work is so sunny. Recently he published three uncontrolled trials suggesting that patients with bipolar disorder develop something like tolerance with antidepressants. The more they take them, the more likely they are to stop working.

At the other extreme Tadafumi Kato at the RIKEN Brain Science Institute in Japan, who wrote “I do not use antidepressants for patients with bipolar II, as they can worsen the illness course by inducing manic switches or rapid cycling,” but while Dr. Kato avoids them for bipolar II depression, he did admit to using SSRI antidepressants to treat comorbid anxiety disorders sometimes.

In the middle were two broad types. There were experts who felt antidepressants are helpful in bipolar II but should only be used with a mood stabilizer. These experts are probably swayed by the large studies showing a clear risk of mania with antidepressant monotherapy, but no detectable risk when taken with a mood stabilizer. 10 of the 18 experts fit in this cautiously optimistic group.

The more cautious in the middle group were willing to use antidepressants in bipolar II, but were not so convinced of their safety and efficacy and would only use them as a last resort. They cited the risk of cycling and hypomania on them, and perhaps were influenced by the meta-analyses I mentioned – the ones showing a low chance of benefit and a greater risk of harm with antidepressants. 6 of the 18 experts fell in this cautious pessimistic group.

Experts Opinions on Antidepressants In Bipolar II DepressionPercent endorsing (total=18)
Antidepressants are helpful in bipolar II and do not cause hypomania (endorsed by 1/18 experts).6%
Antidepressants are helpful in bipolar II but are best used with a mood stabilizer to avoid hypomania (endorsed by 10/18 experts).55%
Antidepressants are best avoided or used with a mood stabilizer as a last resort in bipolar II (endorsed by 6/18 experts).33%
Antidepressants should almost always be avoided in bipolar II because of the risk of hypomania and cycling (endorsed by 1/18 experts).6%

Which Antidepressants They Favor

KELLIE NEWSOME: When these experts reached for an antidepressant, they tended to favor bupropion or an SSRI. They thought tricyclics and MAOIs were the most dangerous, but some admitted that they use MAOIs as a last resort. The MAOIs are a double edged sword, and both sides are sharp – they have studies showing a big effect in bipolar depression, as well as studies showing a high risk of mania.

When They Use Antidepressants

Most experts agreed that only a minority of bipolar patients did well on antidepressants, but they only had vague clues as to who those patients were. Past response to antidepressants was of course a guiding light. They also tended to use antidepressants in patients with long-standing depression that did not respond to other options, or if there were comorbidities like anxiety disorders or OCD that might benefit from them. They shied away from antidepressants if the bipolar II patient had recent hypomanic symptoms in the past 6 months, rapid cycling, or a history of problematic hypomanias. So where do you fall in all this Dr. Aiken?

CHRIS AIKEN: I’m cautiously pessimistic.  I’ve seen antidepressants destroy lives in bipolar disorder, and I’ve also seen people who got worse when the antidepressant was taken away. I’ve learned to taper them very slowly when coming off – unless it’s obvious that they are causing harm or the patient is actively manic I would taper them off over 6-18 months.

KELLIE NEWSOME: All these experts are reading the same studies. So how do they come to such different conclusions?

Why Opinions Differ

CHRIS AIKEN: I’ve thought about that a lot, and I’ve come up with 3 differences.

First – Do they believe there are major gaps in the research, and how willing are they to allow their own experience to fill in those gaps? For example, some believed the harms we’ve found with antidepressants in bipolar mainly apply to bipolar I, while others were willing to extrapolate the bipolar I data to bipolar II. For me, it’s a spectrum of harm between the two, and I’ll cite here a study by Jules Angst. Buried in that study is some telling data. They found the risk of mood worsening on an antidepressant rose steadily with the duration of the patient’s hypomania, from 1% with no hypomania to 38% if they had ever had hypomanias lasting more than a week:

One gap that concerns me in the research is – I’m not convinced that antidepressants are safe when used with a mood stabilizer just because the studies say they are. Suppose a small handful of studies showed construction workers did not develop dementia after multiple head injuries on the job as long as they wore a hard hat. I still wouldn’t want to do that job – hard hat or not – maybe they don’t get dementia but the blows to the head could be harming them in other ways – cognitive problems, headaches, irritability. My concern is that antidepressants will worsen the course of the illness, and most of these studies are only measuring whether they cause overt, observable mania. Most are not even looking at subthreshold mania. When something is harmful, and it’s not known to do much good, my approach is to avoid it as much as possible, not simply lower the dose or add a protective mood stabilizer.

Second – and this is a bigger one – experts seem to differ in how dangerous they think hypomania is. Hypomania if the only disorder in the DSM that – by definition – does not cause significant impairment. But significant impairment in this context means stuff that would make you lose your job, your relationships, suffer legal consequences, or get hospitalized. That’s what mania does. But most patients I see with hypomania say that it does cause problems. It may not attract the attention of the police, but it eats away at their relationships. Here’s how one patient described it in a YouTube video.

She’s talking about the positive, euphoric hypomania there, but even then she’s impatient and irritating to others.

KN: One of the experts, Dr. Ayal Schaffer from Toronto, put it this way: “There is sometimes a misconception that hypomania is a benign experience, without significant distress or consequences. I do not find this to be the case. Hypomania, while not as severe as mania, can destabilize the illness, destabilize lives, and leave they patient with an unsettling sense that their illness is not under full control.”

CA: Yes, the loss of self-control is frightening. These patients can no longer control their own thoughts, which race with doom and gloom. They long to shut their mind off, which, when you think about it, is devastating wish that can result in intoxication, overdose, and suicide.

Should We Treat the Happy Hypomania?

KELLIE NEWSOME: But some patients do have euphoric hypomanias that don’t cause much harm. They are confident, energized, they get a lot of work done, and they do all this in a euphoric, giddy feeling. What’s wrong with that?

CHRIS AIKEN: You know, a lot of people who don’t have bipolar experience hypomania after a night of sleep deprivation. It’s brief, but if they make it past the low point in their circadian rhythm – around 3-4 am – they are greeted in the morning with a perky, upbeat mood. Alcohol and cocaine do the same thing. All these things feel good, but they harm the brain, and in bipolar disorder that harm comes down the road – as depression. So when I see a patient who is loving their hypomania and doesn’t want it to end, I’m worried about them.

KELLIE NEWSOME: And how do you get them to accept treatment, I mean if they don’t think anything is wrong?

CHRIS AIKEN: I start by validating their experience. “You are in a really good place right now, so you clearly don’t need much from me. The only thing I can offer is to help you stay in a good place so you don’t crash into depression. Do you have any concern that depression could come back?” And usually they remember that what comes up does come down, and we work out a plan to prevent that, which may involve a mood stabilizer, lowering an antidepressant, or adding in a behavioral intervention like dark therapy.

KELLIE NEWSOME: Do you ever use Despondex?

CHRIS AIKEN: Despondex?

KELLIE NEWSOME: Yea, it was FDA approved in 2009 but I’ve never seen anyone prescribe it. I read about it in in an American newspaper called The Onion.

Sleep Onset Latency

And now for the word of the day… Sleep onset latency

KELLIE NEWSOME: Have you ever heard a patient complain that they can’t fall asleep? What they are talking about is sleep onset latency – the time it takes to fall asleep. This is how it’s measured objectively in a sleep lab. The clock starts when the lights are turned off – or sometimes when the patient gets into bed – and it stops when then patient actually falls asleep – they look for EEG changes, actigraphy, or other behavioral parameters to figure out if the patient is actually asleep. There’s also subjective sleep onset latency, which is usually measured by a straight-forward question on the Pittsburgh Sleep Quality Index: “During the past month, how long (in minutes) does it usually take you to fall asleep at night?”

Most sleep meds improve sleep onset latency by about 10 minutes over placebo, but the placebo effect adds another 15-20 minutes on top of that, giving many patients the illusion that their hypnotic packs a bigger punch than it does. Next week, we’ll look at whether adding a hypnotic can make antidepressants work better. See you next Monday.

Marijuana in Middle Age and Older Adults

CHRIS AIKEN: Or follow us daily for practical updates on our Carlat twitter feed. Today, we cover a new JAMA metaanalysis on how marijuana affects middle aged and older adults. It’s all about the THC. The more THC in the product, the more likely it is to end up causing dizziness or cognitive and perceptual problems. THC is the problematic ingredient, but cannabidiol (CBD) seems beneficial; it usually counteracted any negative effects of THC. CBD is the same product that is used experimentally to treat psychosis and is available as CBD oil. And it didn’t take much THC to cause problems – 10 mg/day was the cut off, about 1/10th of what you’d find in a single joint.

Get Smarter About Mental Health

Our Brain Bulletin decodes mental health updates for you.

It’s free.