Many of you will remember that in October of 2004, the FDA announced to drug companies that they should add black-box warnings about the risk of suicidal behavior in children and adolescents started on antidepressants. This was based on an analysis of placebo-controlled, randomized trials (a total of 24 trials involving over 4,400 patients) showing that patients assigned to antidepressants had roughly double the risk of suicidal behavior [mostly suicidal ideation (SI)]—from 2% for the placebo group to 4% for the drug group.

A later meta-analysis found that while SI was more likely among those taking medications, this risk difference was only significant for the pooled sample of all clinical trials. The risk was not significantly greater for sub-groups of trials, including trials of meds for major depression, OCD, and the non-OCD anxiety disorders (Bridge JA et al, JAMA 2007;297(15):1683–1696).

A few years later, in May of 2007, the FDA determined that there was a similar risk for young adults (18–24 years old) and required black-box warnings for this population as well. The increased risk was smaller than for children (roughly 1.5x the risk vs. 2x the risk), and unlike the case for children, it was not statistically significant. Other independent meta-analyses found that as patients grew older, the antidepressant-induced risk of SI lessened progressively, so that for patients ages 25–64, antidepressants had no effect on suicidality, while for patients 65 and older, these drugs were actually protective, lowering the risk of suicidality by 61% (Leon A, Am J Psychiatry 2007;164:1786–1789).

How have these black-box warnings affected clinical practice over the past decade? Several studies have been published, and the consensus is that since the black-box warnings, the use of antidepressants in young people decreased significantly, and that suicide attempts increased (there was no increase in completed suicide) (Lu C et al, BMJ 2014;348). While this association cannot prove causation, given the timing it seems likely that the FDA actions caused these practice changes. Of course, just because decreased prescribing of antidepressants may lead to more suicide attempts does not mean that these medications are benign. It is possible that both antidepressant use and underuse lead to suicide risk, via different mechanisms.

Dr. John Walkup, this month’s expert, sees two factors driving the apparent increase in SI in patients in clinical trials: artifacts of research design and real effects of medications.

Research artifacts and suicidal ideation
In clinical trials of antidepressants, SI as an adverse event has typically been established based on what subjects tell researchers in response to an open inquiry. One potential source of artifact is that as young participants get more deeply involved in a clinical study, they may become more open about their symptoms, including suicidal behavior, according to Dr. Walkup. “I’ve seen this both in the clinic and in my research. A young person will say, ‘Thanks for sticking with me. I’m feeling much better. I am not suicidal now but did have those thoughts on and off since we started.’” While such comments could be discounted as pre-existing suicidal thoughts, well-trained investigators would likely conservatively report them as “new onset” SI. Ironically, this scenario might be more common when the medication is actually lifting mood and improving the patient’s overall outlook.

Another potential artifact relates to patients who discontinue medications. Suppose a young person with depression is enrolled in a 12-week-long trial. If the medication leads to improvement after 6 weeks, the patient might stop taking the medication, reasoning that it is no longer needed. However, medication discontinuation can lead to re-emergence of both depression and SI, which would be reported as an adverse event even if it occurred in the context of medication discontinuation. “In the large FDA data set, we have very limited information on how many people developed new-onset suicidal behavior after discontinuing their antidepressant,” says Dr. Walkup.

Effects of antidepressants on suicidality
Nonetheless, Dr. Walkup believes there are some pathways for the emergence or worsening of suicidal behavior during treatment with and without antidepressants. Two main mechanisms for antidepressant-induced SI in children are excessive activation/disinhibition and enhanced mood, leading to more peer group involvement.

“About 10%–15% of kids get activated by antidepressants; when that activation verges on disinhibition, one could imagine it might contribute to suicidal behavior,” says Dr. Walkup. “Yet, even though a signal for activation was found by the FDA analysis, they couldn’t identify whether it was associated with the observed suicidal behavior.” How can clinicians recognize this activation? Dr. Walkup explains that antidepressant-induced activation can include symptoms such as insomnia, increased activity level, agitation, mental restlessness, worsening anxiety, and, at the most extreme end, behavioral disinhibition. “In my experience,” says Dr. Walkup, “it’s the young kids—under 10 or 12 years old—that are much more vulnerable to activation on SSRIs. Also kids who have some kind of neurodevelopmental disturbance, such as ADHD, tic disorders, or autism spectrum, combined with anxiety, are especially vulnerable to activation.”

From a clinical perspective, Dr. Walkup starts most children who fit the risk profile on low doses of SSRIs. “Activation usually presents within the first 24–72 hours. If you start with a low dose, you may not see the activation until you get to the starting dose or the next dosage increase. If it occurs at the low dose, chances are slim that you will get to a full treatment dose. I haven’t been able to do it, no matter how slowly I titrate, which is why I tell parents to discontinue the medication entirely if they see signs of activation.”

Assuming that the medication must be discontinued because of activation, what should we do next? Psychotherapy is one option, as is trying another antidepressant, eg, a different SSRI or one of the SNRIs, such as venlafaxine or duloxetine. For children with depression and no anxiety, a dopaminergic antidepressant such as bupropion might be worth a try. For anxious kids, while there are no actual data to support such a recommendation, you might start with a sedating antidepressant like mirtazapine.

Another potential mechanism for the association of antidepressant treatment and suicidality is improvement in mood and increased energy, which may put teens in more at-risk situations. For example, increased interaction with parents could in turn lead to more parent-child conflict. Such interpersonal conflict is commonly associated with suicidal behavior. In addition, improved mood can lead teens into peer situations that backfire on them, according to Dr. Walkup. “When kids first begin to recover from depression, they seek out social interactions in a way that they hadn’t been able to before. While often positive, it can also lead to risky situations—increased involvement in drugs and alcohol, and involvement in romantic relationships. This may create more opportunity for peer conflict or romantic disappointments.” It is very important clinically to assess the patient’s emotion and behavior control before starting medication so as to anticipate whether improvement might have this effect.

Clearly, the antidepressant/suicide connection is multifactorial. Some of this association is likely not “real,” in the sense that artifacts of how research studies are conducted may yield apparent drug-induced symptoms that are actually caused by other factors. But you should be aware of ways in which medication effects may directly or indirectly complicate recovery.