FDA draws clinicians’ fire over alleged anticonvulsant-suicide link

In the latest round of what has become nearly a year-long simmering feud, the American Epilepsy Association, ensconced at its four-day-long annual meeting in Seattle, issued a letter to the U.S. Food and Drug Administration urging the watchdog group to reconsider its recommendation that package insert warnings be included along with prescriptions for eleven different medications used to treat seizure disorders, bipolar disorder, and migraines.
As always with package insert warnings, doctors are concerned about the potential for patient noncompliance, which in the case of epilepsy can be fatal. And in this instance, the AEA is critical of the methodology the FDA used in determining that a very small increased risk of suicidal thoughts and behaviors may accompany the use of anti-seizure medications.
After reading the press release in a news feed this afternoon, I decided to look into this further, given that I have been mostly on and sometimes off one of the drugs on the list since May. It didn’t take long for me to decide that the AEA, for more than one reason, appears to have a strong case.

First, I went back to the alert the FDA issued in January. The salient parts are below.

The U.S. Food and Drug Administration today issued new information to health care professionals to alert them about an increased risk of suicidal thoughts and behaviors (suicidality) in patients who take drugs called antiepileptics to treat epilepsy, bipolar disorder, migraine headaches, and other conditions.
An FDA analysis of suicidality reports from placebo-controlled studies of 11 antiepileptic drugs shows that patients taking these drugs have about twice the risk of suicidal thoughts and behaviors (0.43 percent), compared with patients receiving placebo (0.22 percent). This risk corresponds to an estimated 2.1 per 1,000 more patients in the drug treatment groups who experienced suicidality than in the placebo groups.
Following a preliminary analysis of data from several antiepileptic drugs that suggested an increased risk of suicidality, in March 2005 FDA requested this type of data from manufacturers of marketed antiepileptic drugs for which there were adequately designed controlled clinical trials. FDA received and reviewed data from 199 placebo-controlled studies of 11 drugs.
The analysis included 27,863 patients in drug treatment groups and 16,029 patients in placebo groups. There were four suicides among patients in the drug treatment groups and none among patients in placebo groups. There were 105 reports of suicidal thoughts or behaviors in the drug-treated patients and 35 reports in placebo-treated patients.

Note that topiramide, one of the drugs I mentioned two days ago that’s beong hailed as a possible alleviator of alcoholism, is on the FDA’s list of suspects.
First of all, though it’s been a while since I took biostatistics and don’t have either the meta-analysis or any of the studies feeding it–not that I hae the desire to study them anyway–there’s something about the methodology here, or maybe it’s just the manner in which the FDA reached a broad-brush conclusion about “anticonvulsants,” as if they are all chemically similar or exert their effects on the brain in identical ways. I mean, with an absolute risk that low (4 suicides in 27,000 people over three years), it seems that seven of those drugs would have been associated with no suicides at all. Of course, I’m assuming the studies used in the meta all examined one drug at a time (I can’t imagine a workable alternative).
Then there’s the matter of trying to tease out not how the control and placebo groups compared, but what constitutes a lot of suicidal behavior and thinking to begin with. In 2005, the overall suicide rate in the U.S. was 0.11 per 1,000. The “rates” in the placebo and control groups of the FDA’s study were 0 and 0.00014 (4 in 27,863) respectively. Now, I know none of these studies lasted an entire year, and that they just took suicides and divided them by enrollees (hence my use of quotes above). But am I missing something in discerning no rampant problem here?
I tried a little induction. Concord, New Hampshire has almost exactly the same number of people as the FDA’s meta-analysis (about 43,000). It would not take long, I’m sure, for four people to kill themselves and for 140 to admit to having bad thoughts, and that’s even if you leave both the state prison and the state hospital out, which is fair. And New England is a very low-suicide zone (cattle country and Native American territory is the worst).
I just can’t see antiseizure drugs were targeted by the FDA here; it looks like either undiluted paranoia or a scapegoating, and stinks of bilge.
Fast forward through the spring and summer, when an FDA advisory board voted down a “black box” warning to be placed on all prescribed units of the eleven indicted drugs, but voted for the inclusion of packages bearing warnings about the potentially increased risk for suicide. The letter sent this week by the AEA to the FDA is available as a downloadable PDF, and I’ve pasted it below in its entirety.
December 5, 2008
Russell Katz, MD
Director of the Division of Neuropharmacological Drug Products.
Food and Drug Administration, FDA 120
1451 Rockville Pike, Room 4037
Rockville, MD 20852
Dear Dr. Katz,
The Board of Directors of the American Epilepsy Society has examined carefully
the results of the meta-analysis conducted by your office on the potential relation
between the use of antiepileptic drugs (AED) and an increased suicidality risk.
The Board has taken this matter very seriously. I have created a task force to
examine the data very carefully, to follow on any new findings from the metaanalysis
and to recommend specific steps that clinicians and patients need to
follow to minimize any suicidality risk.
I understand that the data of the meta-analysis were reviewed by an Advisory
Board that met in June of this year at the offices of the FDA. The Board
recommended that warnings on this potential risk be included in the insert
packages of AEDs, but voted against the inclusion of a black box. The Board
members agreed with this recommendation.
I am writing this letter to express the unanimous concern of our Board’s members
regarding a potential misinterpretation by patients, families and physicians of the
warnings to be included in the package insert of AEDs. In our opinion, such
warnings may result in serious consequences with devastating results for the
following reasons:
1) The warnings may cause significant worry and confusion in patients with
epilepsy and their families, because they may misinterpret the magnitude of the
suicidal risk identified in the meta-analysis. Indeed, they may construe such risk
as being very high, when in fact the data suggested an extremely low risk (0.2%).
This may lead patients to discontinue their AEDs without discussing it with their
physician causing self-harm and including death. As you know, a recent study
has demonstrated a greater than three-fold increase in mortality risk associated
with non-compliance to AEDs (Faught E et al, Neurology 2008).
2) Physicians who do not specialize in epilepsy may be concerned about
prescribing AEDs for fear of being sued. As you know such a precedent already
exists in the way SSRIs were prescribed following the black box warning. Indeed,
a significant drop in the prescription of SSRIs in children and adolescents with
depression followed the inclusion of a black box warning in the package insert. In
addition, an increased refusal by parents to have their children take these
medications was also identified.
3) The members of the Board were also concerned about the committee’s
recommendation to extend the warning to all AEDs. While we understand their
rationale for such recommendation, (a concern that clinicians would limit their
prescription of AEDs to those not included in the meta-analysis), we think that it is inappropriate for an official warning not be supported by actual data.
Our Board recognizes the importance of screening patients with epilepsy for
comorbid psychiatric disorders and suicidal risk at the time of their initial
evaluation and throughout their treatment. Yet, our members consider that the
inclusion of a warning in the package insert of AEDs may cause more harm than
Therefore, we encourage you to seriously consider these concerns in the
formulation of the warnings you plan to include in the package inserts of AEDs.
Our Society will work very closely with our members and with other professional
societies to provide the necessary education to help clinicians screen patients for
comorbid mood, anxiety disorders and suicidal risks prior to and throughout their
treatment with AEDs. We are also eager to work with your office on this matter
for the benefit of patients.
Please do not hesitate to contact me if you have any questions. I appreciate your
attention to this matter.
Dennis Spencer, MD
American Epilepsy Society

I’ll end with this. Several of those drugs originally prescribed only for epilepsy are now first-line medications for bipolar disorder, usually in combination with something else. Give someone with BPD an excuse, even a faint one, to stop taking his or her meds, and there’s an odds-on chance it will happen. This is not a population of people you want to see undermedicated. So the FDA had better be damned confident in its actions here, because if all of this is nothing more than, say, overkill thanks to the Vioxx debacle, the organization could seemingly find a more innocuous means of establishing its rigor for all to see.

  1. #1 by nm on December 11, 2008 - 1:29 am

    I’m currently writing a paper about a similar problem with a different drug (X) for condition (Y). In that particular case there is no meta-analysis and there is essentially no scientific data to indicate a probabilty of harm greater than placebo. It’s an information vaccum that’s been filled by press release and media trolls.
    Certain vested interests have basically drummed the story up in a couple of places in the world and black box warnings have resulted. The media coverage of the alleged problem has of course created the widespread belief that drug X is incredibly bad. Because condition Y and drug X are so common there are now many many reports of adverse events. Who can tell if they are causal or not anymore as the datapool has been so befouled.

%d bloggers like this: