| |
Earlier this week an ACNP Group chaired by doctors John
Mann and Graham Emslie issued a "preliminary report" purporting to exonerate
the SSRI medications on the charge of drug-induced suicidality in children
and adults. Surprise, surprise1. The timing of this release,
i.e. less than two weeks before an FDA Panel is supposed to convene in
Washington to hear testimony2 on this very issue, is suspicious
at best.
But perhaps this is just my own, unhidden bias3
and cynicism speaking. Shouldn't the FDA panel and the numerous journalists
who are now beginning anew to investigate this issue have access to the
most current information possible? I demur. And agree. They should have
such access. But a "preliminary report" without the backup data hardly
provides the kind of robust data that legitimate scientific inquiry and
debate requires. It is, to be blunt, a form of "executive summary". The
Cliff's Note version if you will. Indeed, we can almost hear the sophistry,
spoken in a beguiling whisper, "you don't need to bother to wait for the
real report or look at the data. Here's the answer you need to know."
Scientific articles have "executive summaries" of their
own. They are called "abstracts". But legitimate, peer-reviewed scientific
literature requires publication of the accompanying data. Why? So the
rest of the scientific community can corroborate - or, challenge if appropriate
- the authors' conclusions. In the case of peer reviewed journal articles,
there is, at least, data which can be examined and re-evaluated. This
is quite important, as we shall see. But the ACNP "Preliminary Report"
is far more pernicious. That data won't be available for months, i.e.
not until after the flurry of public attention which is now focused on
the impending FDA hearing.
Cynic? Let the objective reader consider three short examples
from the past, on this very question of SSRI induced suicidality, and
decide for himself/herself.
The first example is a 1988 article by Muijen and colleagues,
which had been accepted for publication a scant four months after Prozac
was launched in this country4. A bit of history is necessary
to put it into perspective. Eli Lilly and Company, the maker of Prozac,
was caught flat-footed in late January of 1990. A little birdie gave them
an advance copy of a February 1990 article by prominent Harvard psychopharmacologists
Martin Teicher and Jonathon Cole5, along with nurse Carol A.
Glod, which raised the troubling specter of Prozac-induced suicidality,
and which predicted the frequency of this phenomenon or its precursors
at approximately 3.5 to 5%. The company needed to act fast or, as its
Chief Scientist Leigh Thompson wrote, "Lilly could go down the tubes."
The article raised problems on several fronts. One was
public relations. Lilly immediately implemented the three-pronged PR strategy
which its spin doctors have wielded ever since with great success: (1)
blame the "Disease" not the "Drug"; (2) blame the Church of Scientology,
which it branded as a "cult"; and, if all else fails, (3) blame "greedy
trial lawyers".6
An equally important problem was marketing, or, to use
pharma-speak, "detailing." What would Lilly's sales force tell the doctors
who were writing prescriptions for Prozac as if it was as benign as aspirin?
This is where Muijen enters the picture. In February 1990, the sales force
was admonished that they should not initiate any discussion
about the Teicher and Cole article or this issue. Not. That's N
O T! However, they needed an anchor-to-windward, i.e. some way
to deal with any doctor who had read the article or heard of the issue.
And they needed it now7. So, Lilly provided each of the salesmen
with copies of the Muijen piece.
The "executive summary", i.e. abstract, was reassuring:
Significantly more patients on fluoxetine improved than patients on placebo.
. . . Suicidal feelings were reduced to a greater degree on fluoxetine
than on mianserin and placebo at weeks 6 and 7.
Sounds good. Hopefully, the prescribing docs, who were
overwhelmingly non-psychiatrists anyway, would not bother to read past
the abstract. But a careful examination of the data within the article
presents an entirely different side. A troubling story. Nearly half (12/26)
of the patients who were "randomized" to Prozac dropped out of the study.
Of the Prozac patients who remained, 23 out of 26 had moderate to severe
side effects, including some of the very side effects which are the harbingers
of suicidality8. And the side effects were growing over time.
Even more troubling was the fact that Muijen and colleagues apparently
excluded 4 patients who had been randomized to Prozac but who withdrew
from the study during the first two weeks from their efficacy analysis.
And, most significantly, 29 of these 4 took massive overdoses
of Prozac which was so severe that it required hospitalization. Lilly,
which was labeling all suicide attempts at that time as "overdose", would
be hard-pressed to deny that these were suicidal gestures. By comparison,
no patients who were randomized either to placebo or the comparator drug,
Mianserin, experienced such suicidality.
Which message would you find to be important? The "abstract"
or the "data"?
Our second example comes from the centerpiece of Lilly's
July 1990 submission to the FDA regarding the suicide issue. Someone at
Lilly obviously realized that Muijen would not bear up to careful
scrutiny. They needed something stronger. And, so, they chose Fava
and Rosenbaum's survey of 27 psychiatrists regarding their experiences
in prescribing Prozac and other comparator drugs to 1017 depressed patients10.
Once again, the abstract and the text of the article contained the reassuring
verbiage: "None of these patients, however, reported intense suicidal
thoughts of the degree described in the previously reported six cases
[by Teicher & Cole]." But "a reanalysis of their data", by none other
than Dr. Mann and his ACNP colleagues, " comparing all fluoxetine [Prozac]-
treated cases with all nonfluoxetine-treated cases indicated a higher
rate of treatment emergent suicidality in the fluoxetine group."11
FDA epidemiologist Dr. David Graham also reanalyzed the Fava and Rosenbaum
data and wrote: "[T]he actual data from this retrospective chart-review
study do raise some potential questions. . . . Treatment-emergent suicidality
was more frequent among 'fluoxetine alone' than 'tricyclics with or without
lithium' patients. The relative risk of suicidality was 3.3 (95% CL 0.9,
12.2), p - 0.07." This analysis corroborated Teicher
& Cole!
Again. Which would you believe? Verbiage from a man who
had already told Lilly in private that looking for the "Teicher phenomenon12"
by culling through preexisting data was like trying to find the "needle
in the haystack13" ? Or a reanalysis of the actual data by
two separate groups of scientists.
The third and final example regarding the perils of executive
summaries comes via juxtaposition of the ACNP's 1993 Consensus Statement
with the subsequent sworn testimony of its principal author, Dr. J. John
Mann. In the Consensus Statement, Dr. Mann wrote for the third time in
so many years that there well might be a "small vulnerable subpopulation"
of patients who were at risk for antidepressant induced suicidality14.
And he recommended further prospective studies, specifically focused on
the question. And warnings. And, yet, he also tempered his remarks with
an industry oriented message supposedly gleaned from an independent review
of data concerning Paxil. In the Consensus Statement, the ACNP claimed
that it had examined unpublished data provided to them by the manufacturer
of the drug concerning 2963 patients on Paxil. From that data, they concluded
that no problem regarding treatment emergent suicidality or its precursors,
e.g. akathisia, mania, and psychosis, appeared15.
During the trial of Tobin v. SmithKline Beecham16,
I had the opportunity to question Dr. Mann about the ACNP's review of
the data, and, to his credit, he fessed up that "To be perfectly honest,
I can't recall how much of the statistical raw data we received at the
time that we put these numbers together. . . . It is conceivable that
we got summary tables like the kind from which these data were abstracted."
I then got to confront Dr. Mann with the actual data concerning these
patients. I showed him charts where the company had described the causal
relationship between Paxil and these very side effects, specifically including
aggression, hostility, suicidality and Paxil as being either "definite"
or "probable." I then asked whether it offended him to be in the position
of testifying on behalf of the company and seeing now, for the first time,
in open court, documents which directly contradicted his own writings
concerning this group of patients and which undermined his testimony.
SmithKline objected on the grounds that it was "argumentative" and the
judge sustained it. And, so, I never got to learn from Dr. Mann himself
about whether he was or was not offended.
Let the reader decide for himself/herself. When Dr. Mann
and friends at the ACNP tell us, less than two weeks before the scheduled
FDA hearing, that we should accept their conclusions in this "Preliminary
Report" at face value, what do you want to do? Read the Executive Summary"
Or see the data?
Case Notes:
1. Dr. Mann has served in recent years as an expert witness for
both Pfizer and SmithKline Beecham on these very issues. Dr. Emslie has long-standing
ties with the pharmaceutical industry.
2. If you want to call the 3 minute vignette limit on each
speaker "testimony".
3. The vast majority of my clients are the families of these
victims of drug induced suicidality. They are part of a group of patients championed
by Dr. J. John Mann himself in the early '90's, and labeled by him as a "small
vulnerable subpopulation" of patients at risk. See www.JusticeSeekers.com.
4. Muijen, et.al., A comparative clinical trial of fluoxetine,
mianserin and placebo in depressed outpatients, Acta Psychiatr. Scan 1988: 78:384-90.
The article hales from London. And, although it contains none of the attribution
language that is now being required, given the fact that it reports on a clinical
trial, it was almost certainly sponsored by Lilly.
5. Teicher, Glod & Cole, Emergence of Intense Suicidal Preoccupation
During Fluoxetine Treatment, Am J. Psychiatry 147:2 (Feb. 1990)
6. The principal architect of this stratagem was Mitch Daniels
who, among other things, coached Dr. Thompson to ignore reporters' questions
and stay on message. See deposition testimony of Mitchell Daniels in Espinoza
v. Eli Lilly.
7. In time Lilly would create its own body of company-generated
or sponsored literature to cite in response to Teicher and Cole. Moreover, it
even neutralized the authors of critical articles, supportive of concern, by
consulting them (as in the case of John Mann) or even hiring them as expert
witnesses (as in the case of Anthony Rothschild).
8. See Teicher, Glod & Cole's subsequent, 1993 article, which
lists 9 different, "biologically plausible" , clinically reasonable precursor
conditions. Teicher, Glod and Cole, Antidepressant Drugs and the Emergence of
Suicidal Tendencies, Drug Safety 8 (3): 186-212 (1993).
9. This was not 2 out of 1000. It was 2 out of 26, or a whopping
7.69%.
10. Fava & Rosenbaum, Suicidality and Fluoxetine: Is There
a Relationship?, J. Clin. Psychiatry, Vol. 53:3, p. 108 (1991).
11. Mann, et.al., ACNP Communication, Suicidal Behavior and
Psychotropic Medication, Neuropsychopharmacology 1993 - Vol. 8, No. 2 [hereinafter
"ACNP Consensus Statement"].
12. In other writings, e.g. my "Open Letter" in the wake of
the FDA's first action on this matter, posted on www.JusticeSeekers.com, I have
pointed out that, according to Lilly, the phenomenon was very specific, and
it had at least 3 out of 4 key attributes. I have also shown that somewhere
between 10 and 100 Lilly scientists, inside and outside of the company, recommended
a Rechallenge Protocol on patients who satisfied these elements.
13. Dr. Rosenbaum's actual words, as captured by Lilly scrivener
Dr. Charles Beasley, were "data problems needle in haystack." Realizing the
severe limitations of searching for treatment emergent suicidality via post-hoc
reviews of data sets, Dr. Rosenbaum's recommendation was, "if exists won't find
don't look at means find then rechallenge."
The title of my own paper to be submitted to the FDA Panel
will be "Needle in the Haystack". It will be posted as well on our website,
www.JusticeSeekers.com I will show, from Lilly's own internal documents and
the deposition testimony of its key people regarding this issue, that the kind
of data massaging which the FDA Panel is focused on doing may well be sending
the entire scientific community on a wild goose chase.
14. The first such writing was his 1991 article with his colleague
Dr. Kapur. Mann & Kapur, The Emergence of Suicidal Ideation and Behavior During
Antidepressant Phyarmacotherapy, Arch. Gen. Psychiatry, 48: 1027 (1991). The
"executive summary" of that article was not left alone to the abstract. Rather,
he began with a colorful quote from Sir Robert Hutchinson which captured his
central message: "From making the cure of a disease more grievous than its endurance,
Good Lord deliver us." Ironically, Dr. Mann has now become the chief apologist
for the SSRI industry on this point. There is a word to describe that transition.
The word is chutzpah!
15. One of the data-cooking techniques which was employed to
achieve this result was the conversion of SSRI suicidality into "patient years".
It is a clever way to mask a phenomenon which occurs on the SSRI drugs, mainly
during the early period of drug therapy.
16.The federal judge in that case concluded, both before and
after the trial, that the evidence which we presented linking Paxil to homicide
and suicide was "scientifically reliable" and legally admissible. After hearing
that evidence, as well as SmithKline Beecham's rebuttals thereto - including
the testimony of J. John Mann, M.D. - the jury found that Paxil "can cause some
people to become homicidal and/or suicidal". |
