Antidepressant Safety Discussions: 
The 2004-5 FDA Warnings
(revised 9/2007)

Update 9/2007: the most balanced summary on this issue that I have seen since the FDA's initial warning has just been published in a prominent psychiatry journal (Leckman).  Allow me to offer you a version of this essay in plain English, if you are not a physician or someone with similar medical training.

Update 5/2007: the FDA has just revised the suicide warning for antidepressants.On May 2 day they suggested to antidepressant manufacturers that the black box warning be extended to include those between the ages of 18 and 24. In their announcement, they noted that the risk is small; and that antidepressants are effective treatments for many young people, indeed probably lowering their suicide risk. At issue is a small subgroup whose thinking about suicide can become much more dramatic, and may even include homicidal thinking. Here is that FDA advisory.


On June 30th, 2005, the FDA formally announced their inquiry into suicidal thought/behavior in adults taking antidepressant medications. This followed their 2004 inquiry about this same risk in children and adolescents. The following essay provides links to their Warnings, and offers my interpretation of them.  Some bipolar researchers believe bipolar disorders represent a possible explanation for suicidality connected to antidepressants; this hypothesis is explained and their work linked.  Here is a brief summary of my conclusions from all this, before all the detail begins. 

With this announcement in 2005, I think the FDA is signaling their continued concern about the safety of antidepressant medications when given without adequate screening or education of patients. With such screening and education, antidepressants remain a very important part of the treatment of depression and should not be avoided on the basis of these safety concerns alone. 

There will be continued controversy, but in my view it's just a matter of balancing risks and potential benefits, while aware that one of the risks of antidepressants is a possible increase in agitation and anxiety, sometimes even including suicidal and homicidal thinking. If one accepts that summary, then the rest of this essay is largely unnecessary. 

In the meantime, prescribers should be aware that antidepressants now include a recommendation on their label stating that patients who are to receive them should be screened for bipolar disorder, including a family history of psychiatric problems.  All this, in addition to explaining the association between antidepressants and suicide; obviously a challenge in a primary care setting. I've constructed a one-page version of a now-standard screening tool (the MDQ) which accomplishes all this while you see another patient, but you must be careful when interpreting its results, as explained here.   

In my view, the simplest way to explain all that the FDA has recommended patients understand about this risk issue is to combine the warning with an education about "soft" or subtle bipolarity. Several authors have speculated that such underlying bipolarity might account for much if not all of the adverse events the FDA is warning about (e.g. Berk). Children may be at greater risk because those depressed enough to warrant consideration of antidepressants are much more likely to have bipolar disorder than a similar sample of adults; and children may also be more likely to experience this particular effect of antidepressants.Faedda 

Therefore the solution, in my view, to this problem of antidepressant-associated suicidality, is to screen all depressed patients for bipolar disorder, as the FDA recommends prior to prescription of an antidepressant; and then educate all who manifest even subtle signs (even if well less than DSM diagnostic thresholds, or MDQ cut-off scores) about soft bipolarity.  Only after that education should one then discuss with the patient the diagnostic likelihood of bipolar disorder and her/his risk-tolerance regarding antidepressants. 



  • Introduction
  • The FDA reports themselves
  • Brief summary of the high points
      - Causation possible, not yet established
      - Warn patients and monitor closely after starting
      - Anxiety, agitation, irritability, insomnia 
         may indicate increased risk
      - Must screen for bipolar disorder, 
         get a family history of mood problems, before AD use
  • A brief history of the warning
  • Some of the data the FDA reviewed
  • A case example: no depression before medication, 
    suicide in 2 days
  • Identifying risk: A working hypothesis 
  • Summary and guidelines


FDA reports
From their overview page on this issue, the FDA offers you: 

Brief summary of the high points: 
  - Causation possible, not yet established
  - Warn patients and monitor closely after starting
  - Anxiety, agitation, irritability, insomnia may indicate increased risk
  - Must screen for bipolar disorder before antidepressant use
For more on each of these, read the bullet points in the FDA advisory, the exact "Warning" information

History -- These medications have been around for years; why is the FDA ruling now? 
"Pediatric Exclusivity" (PE) is the answer.  Several companies were pursuing this for their antidepressants.  PE means a company can continue its patent rights 6 months beyond the expiration date (that can be worth a lot of money).  But it also led the FDA to review their and the companies' data on effectiveness and safety more closely.  These same data prompted a similar review in Britain, where the FDA equivalent, the Committee on Safety of Medicines (CSM), ended up ruling on this issue first with a near ban on antidepressant use in children.  In April 2004 the European Union's Agency for the Evaluation of Medicinal Products issued similar recommendations

These reviews focus on whether antidepressants can actually cause suicidal thinking and behavior.  This issue had first been raised in 1990 when Dr. Martin Teicher and colleagues at Harvard published 6 cases which seemed to implicate antidepressants in the suicidal thinking of several patients.  The FDA convened a hearing in 1991 to look at this issue -- basically the same inquiry as is now taking place, but focused on adults rather than children.  They concluded that antidepressants do not cause suicidality. 

Here is an FDA summary of that 1991 hearing.  Comparing that process to the 2/3/2004 hearing, it is clear that a different approach was taken this time.  One of the most striking differences is that families and patients were allowed to present their experiences.  It seems from the 2004 transcript (5th bullet, 5th sub-bullet) that these presentations had a substantial effect on the committee members (e.g. closing remarks, page 400).  

There have been allegations (and explanations for same) about the process the FDA used for handling this issue.  In any case, the FDA has now commented on the issue of antidepressants and suicidality, this time initially focusing on children, though ultimately ruling on all use of antidepressants.   

Data from the FDA hearing  Feb. 3,2004; and the conclusions offered: 

There were three primary data sources considered:  

  1. Manufacturer's data from previous studies (Dr. Mosholder's presentation)

  2. Adverse events reported to the FDA or to manufacturers (Dr. Iyasu's presentation); 

  3. Families' and patients accounts of suicidality.

Here are some select slides from these sources: 

A. Data from manufacturers
 As shown below, in 15 studies of antidepressants in children the rate of suicidality ("possibly suicide-related events") was nearly unchanged by being on medications, compared to being on placebo.  However, a major trigger for the new FDA review was the markedly increased suicidality in a few studies, shown below in red  (from Dr. Laughren's summary presentation, slide #13).  The venlafaxine (Effexor) data led that drug's manufacturer to send its own "Dear Doctor" letter last fall warning of a risk in children. 

Drug/Study Number
“Possibly Suicide-Related”
Risk Ratio

B.  Adverse events data

including by medication type: 

C. Case examples from families 

As you can imagine, the crucial history would be a man or woman who had never had suicidal thinking, and ideally, would have been given the medication for some reason other than depression.  There was just such a case presented at this hearing.   (Here is a summary of 14 more such cases from among the 53 testimonies). 

Here is the crucial slide from the son's testimony, describing his father. 

As  you can see,  this man did not have depression before the medication was started.  Thus in this case it is easier to presume that the medication, rather than a sudden worsening of depression,  actually "caused" his suicide. 

But the FDA is wisely trying to have more to go on than cases like this.  After all, they are ruling on a class of medications now being given to millions of people.  They feel, understandably, that they simply must get this right and not make a mistake.  So they have to have something more certain to go on than testimony like this. 

They got a little more such evidence at their February 2004 hearing, from an interesting new data source:  the Internet.  An organization called the Child and Adolescent Bipolar Foundation, whose entire testimony is available on their excellent website,; presented this crucial report (all italics theirs):

Over the four years since CABF first launched its Web site,, numerous parents have reported on our Message Boards the results of their children using of antidepressants. Some report that their children first became suicidal immediately or shortly after being prescribed an antidepressant, and that these alarming symptoms stopped only when the medication was stopped. Others report that their child’s prior existing depression with suicidal ideation quickly escalated into rapid cycling/mixed states (the state with highest risk of suicide for any patient with bipolar disorder) soon after their child began treatment with an antidepressant, and that these increased symptoms subsided when the antidepressant was stopped. Conversely, other parents report that their children had suicidal thoughts or actions before taking any antidepressant, or which developed during treatment for bipolar disorder, and that the suicidal ideation and behavior improved when an antidepressant was added to the child’s treatment plan.

Here is CABF's conclusion: 

There is no reason to doubt the veracity of any of these reports. The reports indicate that in some subgroup of children suicidal ideation and behavior may emerge for the first time (or worsen) when a child is given an antidepressant.

Making sense of all this:  a working hypothesis

How do we proceed while the FDA continues to seek further data?  We need a working hypothesis that might suggest who is at risk, and who can use these medications safely (though all should now receive a warning, the FDA says).  

First, a good hypothesis should explain why, if the case reports implicate antidepressants as causing suicidality and homicidality, the overall data to date don't clearly show this.  For more on this, see my speculations on why we might not see suicidality go up -- including more case report details from the FDA hearing.   

Secondly, who can take an antidepressant safely? Is there a group of patients who are likely to get this suicidal reaction?  Can they be identified, as a group?  And finally, the crucial issue:  can a doctor tell that her/his patient is a member of that group? Can that be done quickly in a busy primary care practice?

The Bipolar Hypothesis

[Update 5/2005: Two recent articles proffer exactly the same point of view.Berk, Akiskal] Amongst the explanations offered for the apparent increase in suicidality with antidepressants, only one connects very directly with extensive previous experience: bipolar disorder.  By comparison, the only other coherent clinical identity associated with antidepressants and suicidality is "akathisia", but akathisia has not generally been thought to be an acute risk for suicidality when caused by antipsychotics (i.e. there is not an extensive literature on akathisia and suicide, as there is for bipolar disorder and suicide). 

There is no doubt that antidepressants can induce manic symptoms.  There is also great suspicion that antidepressants can induce mixed statesMontgomery, which are associated with increased suicide risk.Simpson  Therefore we can posit that at least one mechanism by which antidepressants could induce suicidality is by inducing bipolar manic or mixed states.  The FDA warning implies this model.  They warned of the following symptoms, most of which are directly associated with bipolar disorder (italics mine):  

Anxiety, agitation, ... insomnia, irritability, hostility, impulsivity, ... severe restlessness, hypomania, and mania  have been reported in adult and pediatric patients being treated with antidepressants for major depressive disorder as well as for other indications, both psychiatric and nonpsychiatric.

Proceeding on the basis of the bipolar hypothesis
Perhaps only patients with clearly severe depressive symptoms would have potential benefit great enough to justify the risk of an antidepressant, if everyone is at equal risk.  We could recommend a lot more psychotherapy, for those less severe cases; and if good psychotherapy is available in your area, that's actually a pretty good idea on how to approach this dilemma.

But if we are going to use antidepressants in patients who do not have severe depression (e.g. obsessive compulsive disorder; and those patients with anxiety disorders who cannot access the psychotherapies developed specifically for panic disorder, social phobia; and PTSD; and most especially, milder cases of depression), then it certainly would be helpful to be able to tell patients whether this Warning really applies to them.  

In my practice I have been using, even prior to the FDA warning, the "bipolar hypothesis":  I tell patients that if we screen for bipolar disorder and find little to support it in their history and family, then they can probably take antidepressants with little risk of somehow triggering something unexpected.  Note the word "probably"; remember, this is a working hypothesis.  As the FDA has indicated, we are not in a position to offer conclusions or, in this case, firm reassurances.  

Whether you use this "working hypothesis" or not, the FDA has indicated that you must screen for bipolar disorder. Once that screening has been shown to be negative, then you may give the antidepressant if the patient prefers, or won't go to or can't access psychotherapy, or if the depression is severe.   Here is that section of the FDA warning:

Because antidepressants are believed to have the potential for inducing manic episodes in patients with bipolar disorder, there is a concern about using antidepressants alone in this population. Therefore, patients should be adequately screened to determine if they are at risk for bipolar disorder before initiating antidepressant treatment so that they can be appropriately monitored during treatment. Such screening should include a detailed psychiatric history, including a family history of suicide, bipolar disorder, and depression.

How is a busy primary care doctor going to get that detailed history?  One way to save some time on this step is to use the "MDQ" -- the Mood Disorders Questionnaire, a validated bipolar screening instrument.Hirschfeld   You can download a Word version from the Primary Care Providers' Resource Center.   

The family history recommended by the FDA can be very telling in some patients.  Here are the red flags:

Finally, when you screen for bipolar disorder, what are you going to do when you get one (e.g. a "positive" MDQ)?  Hopefully you could refer all such patients to a psychiatrist.  But even if you have that luxury, may I recommend that you also send the patient to the Bipolar II material on this website (I'm not making any money when you do, by the way; see Funding).  Because many patients will reject a "bipolar" label, associating it with Bipolar I (including connotations of psychosis), most patients will need an immediate primer on bipolar disorder variations lest they say "I'm not bipolar; I've never had a manic episode!". 


  1. The FDA warning calls for screening for bipolar disorder including a family history check.  Here's a modified MDQ screening tool for this step. 
  2. Refer positives to a psychiatrist, if you can; and to the Bipolar II material on this website.
  3. Follow the FDA warning instructions.  Here's one way: 

    Include in your usual PAR the TIES to risk:  say to your patient

    "Antidepressants can occasionally cause some really strange reactions.   Contact me if you get 

    • T: unusual thoughts -- including violent ones, or suicidal thinking; 

    • I:  marked irritability,

    • E: too much energy

    • S: severe sleep problems.

    Don't just stop the medication -- call me. " 

    Note that a similar warning has been included at the bottom of the modified MDQ, which can serve as a one-step approach to screening and explaining these risk to patients, in a busy primary care setting. 

  4. See the patient back within two weeks unless you have a very tight phone follow-up system; and again by one month (most of the adverse reactions were in this early phase of treatment, though you have to stay vigilant after this time as well).
  5. Learn more about Bipolar II (see the clinician's version; it's trimmed down from the patients' version) as you go; a working hypothesis (above) predicts we'll find these are the patients who have the reaction the FDA is warning about.