The antidepressant suicide link – busted?

The Washington Post reports on the apparent jump in suicide rates since antidepressants got a black-box warning in 2004 after some reports suggested an increased suicide rate in youths after the initial prescription.

The article here (goddamn WaPo still can’t figure out how to link anyone but themselves) shows a disturbing correlation:

METHOD: The authors examined U.S. and Dutch data on prescription rates for selective serotonin reuptake inhibitors (SSRIs) from 2003 to 2005 in children and adolescents (patients up to age 19), as well as suicide rates for children and adolescents, using available data (through 2004 in the United States and through 2005 in the Netherlands). They used Poisson regression analyses to determine the overall association between antidepressant prescription rates and suicide rates, adjusted for sex and age, during the periods preceding and immediately following the public health warnings.
RESULTS: SSRI prescriptions for youths decreased by approximately 22% in both the United States and the Netherlands after the warnings were issued. In the Netherlands, the youth suicide rate increased by 49% between 2003 and 2005 and shows a significant inverse association with SSRI prescriptions. In the United States, youth suicide rates increased by 14% between 2003 and 2004, which is the largest year-to-year change in suicide rates in this population since the Centers for Disease Control and Prevention began systematically collecting suicide data in 1979.

This is disturbing. However, the I don’t agree with the current interpretation of the problem.

NIMH’s Insel said it is possible that antidepressants are lowering the risk of suicide overall, even as they increase the risk among a subset of patients. New research to be published soon examines genetic factors that may put some patients at particular risk, he added.

There is another explanation – a mixture of confirmation bias and post hoc ergo propter hoc reasoning. The main offender, yet again, is a meta-analysis.

The analysis pooling data from multiple clinical trials of anti-depressants found:

The overall risk ratio for selective serotonin reuptake inhibitors in depression trials was 1.66 (95% CI, 1.02-2.68) and for all drugs across all indications was 1.95 (95% CI, 1.28-2.98). The overall risk difference for all drugs across all indications was 0.02 (95% CI, 0.01-0.03).

Conclusion Use of antidepressant drugs in pediatric patients is associated with a modestly increased risk of suicidality.

It was a small effect observed, the CIs almost covered the RR of 1.00. The authors of the study also mentioned a prominent biases in such a study:

There has been a long-standing concern that antidepressant drugs might actually induce suicidality early in treatment. A textbook of psychiatry published more than 40 years ago, in referring to observations of depressed patients during initial treatment with tricyclic antidepressants, noted that, “With beginning convalescence, the risk of suicide once more becomes serious as retardation fades.”

This is true, for a long time doctors observed that if a patient would commit suicide while under their care, it was usually right after prescription of an anti-depressant, rather than later on.

The data that I think illuminate the problem, are in yet another AJP paper, shows a possible reason for the result.


In this figure you see a big drop after prescription of SSRI, but critically, in the first month after the prescription is given, suicidality is higher than at any other time during the course of treatment. Notice also the rate of suicidality is much higher among patients seeing a psychotherapist as opposed to a GP, don’t worry though, this is likely referrer bias rather than pscyhotherapists causing suicides.

Now think about it. Doctors don’t see patients in the months before the prescription and have knowledge of suicidality before they enter the medical system. What they see is a patient they’ve given a script for, and that suicides, if they happen, are usually right after prescription. But that doesn’t mean the SSRIs didn’t greatly reduce the overall risk. And even more likely, given the rapidity of the effect, that merely seeking help had an enormous influence on suicide risk (the study authors felt the effect on suicidality was too fast to be a result of pharmacologic intervention).

I think it will take more study to prove my hypothesis, but I suspect the problem was confirmation of an existing bias among clinicians, that was justifiable based on the time-window of their experience with depressed patients, but not reflective of the overall disease course. The data showing such a drastic jump in suicides after a substantial decrease in SSRI prescriptions is pretty shocking proof that removal of this intervention has cost lives, and that SSRIs are responsible for decreasing suicides. The long term decrease in suicide rates was attributed to SSRI prescriptions, although it has never been clear it was SSRIs or some other confounding variable. This convinces me at least, that SSRIs have had a significant impact on suicidality, and the long term decreases in suicidality have been from the introduction of these drugs.