 |
Contents page |
 |
What's New? |
Department of Psychological Medicine
Hergest Unit
Ysbyty Gwynedd
BANGOR
Gwynedd LL57 2PWTel : (01248) 384452
Fax : (01248) 371397
June 27th 2003
Professor Ian Weller,
Chair, MHRA SSRI Review Committee.
Market Towers
1 Nine Elms Lane
London SW8 5NQ
Dear Ian
This letter responds to some of the issues raised at the June 20th meeting. It seems appropriate to copy Dr June Raine and Richard Brook in this, and I would be very happy for you to circulate this to all Committee members. When it comes to liasing further with some of the Committee members on issues raised, I would envisage copying you, Dr Raine and Richard Brook in on all such correspondence and would be happy for this further correspondence to be made available to other Committee members. The thought of so much paperwork flying around will encourage me at least to keep my interventions as brief and as infrequent as possible.
A key point raised in the meeting, and flagged by you as central to how the Committee plans to proceed, was the need to collate all clinical trial data germane to the issue of suicidality, and in particular placebo controlled RCT data. As someone who has argued for some time that it’s quite likely that each of the companies has sufficient placebo controlled data on file at this point, across indications (including smoking cessation, obesity, OCD and other disorders), to resolve the issue of specific suicide risks in terms of each of the drugs on the market, I fully support your intentions in this area.
Seeking out this data seems central for another reason also. One of the worrying features of the SSRI suicide / dependence controversy has been how companies have managed to portray the problem as a choice between patient or clinician-led "anecdotes" on the one side and scientific evidence on the other. Proponents of the view that the drugs may cause a problem have been faced with a choice between these two sets of evidence and have been portrayed as picking the weaker of the two sets.
As the problem has evolved, the MHRA has appeared to be faced with an invidious choice between denigrating the value of what patients have to say or deserting the values of science in a manner that might conceivably leave the agency open to legal action.
This forced choice is however more apparent than real, and perhaps stems from an ability of pharmaceutical companies to manufacture doubt in addition to other products they market. In endorsing patient accounts, and the early clinical descriptions of the problem, the apparent initial failure of RCT evidence to support the existence of a problem never led me to the view that the usual rules of science should be discarded. I have always expected that if the correct trials were undertaken, a very clear signal endorsing both patient and clinician accounts would come through in both the epidemiological and RCT evidence. The astonishing fact is that notwithstanding all the problems inherent in the current RCT and epidemiological evidence, the suicide risk shows up there remarkably clearly.
While it appears that we all agree that collating placebo controlled trial data is of importance, of even better value would have been trials specifically designed to address the question of suicidality. Having left you a trial design at the last meeting, drawn up by Lilly in 1990, it is clear that such trials could have been undertaken. (I attach a further copy with this letter).
As a second best, standard trials could have included the scale specifically designed for Lilly to capture the phenomenon of emergent suicidality (copy attached). There really is no good reason why large numbers of trials of any antidepressants conducted post-1990 could not have used such a scale.
In our current position, which involves having to rely on trials not designed to shed light on this issue, two sets of problems emerge, one of which arises in principle, and another, which stems from the conduct of the relevant trials
The problem that arises in principle from current RCTS of SSRIs and risks of suicide is whether it is possible to construct an appropriate relative risk or odds ratio for suicidal acts from standard trials, when the drug in question may both lower the risk in some patients and increase it in others. As I understand it the issue was somewhat simpler in children where there may have been no lowering of risk to offset against the increased risk.
An extreme version of this problem arises in the case of older agents, such as Imipramine, which in severe hospitalised depression can lower the suicide risk, while at the same there is a respectable clinical pedigree stating that Imipramine can trigger suicidality. This leads to a situation where I can assemble a set of trials for this drug yielding a relative risk of suicidality on Imipramine less than 1.0, but yet would argue the drug can cause suicidality. I would argue that relative risk indices in this case point to the frequency of risk against benefit rather than the likelihood of actual causality in principle.
This scenario yields a conundrum for anyone who wants risks and benefits balanced in one simple odds ratio. The methodological difficulties are further compounded by the evidence that some of these drugs may produce suicidality as a withdrawal effect. While grappling with relative risks of treatments that both increase and lower risk, is not unheard of – as in the case of vaccines - no-one to the best of my knowledge has ever had to grapple with the issue of constructing an appropriate index of relative risk in a situation where the data point to both on and off effects from treatment.
Despite these complexities, fortunately, I think the data that I have seen is sufficiently clear-cut to indicate there are substantial grounds for concern, which leads me to the second point.
I think it will not be sufficient for your Committee to simply ask the companies for their tabulations of the data for 5 sets of reasons.
First there is the issue of suicidal acts being coded under placebo, when these did not in fact happen on placebo in the conventional sense. I spent some time developing this point on June 20th; this point is indicative of the kind of problem you face but is not the entire problem.
(I attach as appendix 1 to this letter material from the depositions of Martin Brecher, the FDA official responsible for the Safety Review of paroxetine, and of David Wheadon, a medical officer for GSK, which confirm the "miscoding". These depositions offer some indications as to how such a problem might in fact have arisen in a manner that might not have been deliberately deceptive).
Second, although constrained by confidentiality orders, I am at liberty to let you know that there are other serious problems with company tabulations of the data quite aside from the categorisation of washout suicidal acts under the heading of placebo. Many suicidal acts for instance simply have not been coded as suicidal acts. They have been coded under terms such as failure of response to treatment rather than an adverse effect of treatment, and where listed under adverse event, the patient may show up coded as emotionally labile rather than suicidal or even coded as having dropped out of the study for a primary adverse event of nausea when in fact there has also been a suicidal act.
Third in some trials up to 50% of patients were on concomitant medication. Company memoranda and published articles mention this as one way to handle akathisia, leading to the extraordinary situation that companies were actively managing in trials a problem that they then claim these trials reveal does not exist.
Fourth, investigators were also permitted to lower the dose of treatment in many of these trials at signs of agitation.
Fifth, there was a widespread failure to follow-up patients who dropped out of these studies. Neither you nor I can have any confidence that a significant number of patients who dropped out did not go on to suicide.
I have at least 4 large boxes of patient records from Lilly trials of Prozac, which bring these points out in some detail. I would be happy to forward these to you along with the deposition testimony, expert report and trial testimony of Dr Nancy Lord who has analysed the Lilly studies in detail. I do not however think these Lilly studies were unique.
Thus while I applaud the MHRA’s wish to resolve this issue on the basis of scientific data, I think you need to be aware that a proportion of the data tabulated by the companies will not qualify under minimal trade description standards as scientific data. At the very least you need to inspect the raw data yourself. If I can assist in this process, I would be happy to help in anyway that seemed appropriate.
Trials Sought
In the course of the meeting I was asked for the reference number for a placebo controlled trial conducted on Seroxat in recurrent brief depression, which gave clearly higher rates of suicidal acts on Seroxat. I am still chasing this, but need to make you aware that ongoing confidentiality orders may make it impossible for me to give you any trial numbers. For exactly this reason I took care not to specify the protocol numbers of any healthy volunteer trials in the submission I made to you.
But I can forward the slides used by Dr David Baldwin to present some of the data from this trial at an ECNP meeting in September 1999. Dr Baldwin as an expert member of the former committee reviewing this issue is better placed to give you further details on the reference number of the trial. Again you need to see the raw data from this study rather than depend on pre-tabulated data.
I suggested you should also seek details of a comparable study using Prozac, conducted by Dr Montgomery commencing in 1990. Reading the reports to Lilly from 1990 left with you for the June 20th meeting, it would seem that Dr Montgomery rather than Lilly may well have initiated this trial and the company accordingly might argue that they do not have the data. However, if this is the case, Dr Montgomery should have the data, and as a former member of the CSM, I would have thought he would be happy to hand it over.
On the issue of whether lower doses of these agents would be safer, it would be worth getting details of a study on once a week Prozac reported by Dr Montgomery in 1986, all details of which seem to me to have vanished. Even the abstract from this presentation has vanished from the meeting proceedings.
Absolute Risk
The question was also raised as to the absolute risk run by patients. As I recall it, figures were offered indicating four million prescriptions were issued for Seroxat last year and it was estimated that four million scripts may translate into as many as 1 million people.
The epidemiological and RCT trial evidence would suggest the rates of suicides on SSRIs over and above the rate that might be contributed by depression of 100 per 100,000 patients. This would mean 1,000 suicides per million patients.
To establish an absolute risk, however, will need a figure for the number of patients starting on SSRI treatment in any one year rather than the figure for the absolute numbers of patients on treatment. I would expect that well over half of the patients on treatment, on Seroxat in particular, will be on chronic treatment in either a voluntary or involuntary capacity.
This would leave several hundred thousand patients starting an SSRI per annum. Figures like this would point to several hundred suicides per annum.
Conventionally, it is said that 5% of the population in any year may be depressed of which it is thought that less than half are picked up and are on treatment with antidepressants. Of these, a significant proportion of people put on treatment with antidepressants will not take any antidepressants prescribed. Taking these factors into account, it seems unlikely that many more than 200-300,000 start on treatment in any one year.
There may of course also be a further contribution from the use of SSRIs for a range of conditions other than depression.
A figure of 300 suicides per year would mean approximately 2 per coroner per year and 1 for every consultant psychiatrist every 10 years or so and a much less frequent problem again for General Practitioners. The rate of suicide attempts however may be up to 10 times higher than this, and the rate of treatment induced agitation/intolerance roughly 10 times higher again, making the less serious aspects of this problem very visible at the general practice level.
As mentioned the actual lowering of suicide risk in the 40% of those tolerating SSRIs is likely to be relatively minor as these are patients who are not at a substantial risk of suicide, while the increase in suicide risk in those 5% who drop out of trials for agitation will be substantially raised from the baseline low risk such patients will have had prior to treatment.
There were a number of other issues raised at the meeting, such as the role of RCT evidence in the determination of cause and effect as well as the epidemiology of suicide in schizophrenia and suicide in general, which are perhaps best addressed to individuals at the meeting, but at present I have no list of who was there and must depend on you to ensure the appropriate persons get this correspondence.
Yours sincerely
Dr David Healy
Director of the North Wales Department of Psychological MedicineCC.
Dr June Raine, Director of Post-Licensing Division, MHRA
Sarah Wark, Post-Licensing Division, MHRA
Richard Brook,Chief Executive, MIND,15-19 Broadway,London, E15 4BQAppendix 1
I have already left you:
a. A small part of November 1989 NDA paroxetine submission to the FDA.
b. The May 10, 1991, revised suicide document GSK submitted to FDA, officially entitled "Suicidal Ideation and Behavior: An Analysis of the Paroxetine Worldwide Clinical Database" and dated "April 29, 1991."
e. Dr. Brecher's FDA "Safety Review" of paroxetine dated June 19 1991.
f. Dr. Brecher's FDA "Efficacy Review" of paroxetine dated June 3, 1991.
The Total Test Population for Paroxetine Clinical Trials was 4668 and this remains constant across all submissions. These break down as follows:
Paroxetine 2963
Active Drug 1151
Placebo 554
A. 1989 data: Paroxetine 49 Suicide Events; Placebo 1 Suicide Event; With 49 events out of 2963 patients, this is a "frequent’ suicide event rate per FDA standards, rate of 1.65%
(1) Paroxetine Attempted Suicides 42
With 49 events out of 2963 patients, this is a "frequent’ suicide rate of 1.65%
Sourcing: "42" attempted suicides are documented in two places. One is in Table XI.21 ("Attempted Suicides and Overdoses—Worldwide Data") submitted in the original NDA. The 2nd place "42" is documented is in Brecher’s 1991"Safety Review", page 24 under "Suicide Attempts" (e.g. "42 of the 59 suicide attempts (71.8%) were made by patients on paroxetine who comprised 63.5% of the patients in the Phase II-III trials.")
(2) Paroxetine Completed Suicides 7
Sourcing: Brecher’s "Safety Report" page 23, gives "7" paroxetine suicides under the paragraph "Deaths—Suicide"
(3) Placebo Attempted Suicides 1
Sourcing: The original Nov., 89 NDA Table XI.21 ("Attempted Suicides and Overdoses —Worldwide Data"), Page is dated "PAR Safety summary—10 Nov 1989, page 206" Under "attempted suicides" and "placebo," a placebo patient population of "N=554" is indicated with "3*" listed below. The asterisk (*) is keyed at the bottom of the data with "* 2 overdoses during placebo run-in period."
Dr Brecher Deposition:
Dr. Brecher in his deposition transcript from March 13, 2002, pages 210 and 211, answered these questions posed by attorney for the plaintiffs Farber:
Q: "is it scientifically legitimate to count a suicidal act occurring during wash-out and run-in to the placebo count?
A. "No, because everybody got placebo." (Page 210 Line 22)
Q: "So it's scientifically illegitimate way to count, correct? (Lines 23-24)
A. Yeah. (Page 211, Line 1)
(a gap, then in later questioning…)
Q. "You recall from the sheet we just went over that there appeared to be one legitimate suicide that was not wash-in and run-out, correct? (Page 212 lines 18-20)
A. "Yes." (line 21)
Deposition of David Wheadon of GSK on issue of counting washout suicides against placebo.
Q: So in this whole period, we only have 2 placebo suicides, correct?
A: That is correct.
Q: Now, got back to the final page, which is Appendix 1, we haven’t looked at that page before, it has got a number of SB020 at the back. And at the bottom, we have an asterisk, and it says both—I will read it. ‘Suicides were committed during the placebo wash-out phase of an active control study. These acts were committed two days and seven days prior to the baseline evaluation.’ What does that mean?
A: That means that the suicides were committed while the patients were on placebo in the placebo run-in or wash-out phase, which way you want to describe it.
Q: That was my question. Is wash-out and run-in the same?
A: Yes.
Q: Why were they counted? If it didn’t occur during the baseline period, why were they counted at all?
A: Because it happened during the control phase of the trial.
Q: Was that consistent with protocol, that these – maybe I’m operating under a mistaken assumption. I thought data that occurred during these wash-out, lead-in phases were not counted in the data.
A: That is not correct. The way this data was collected – and I need to correct myself. It wasn’t because they happened during the control phase of the trial. The way these data were collected, any suicide that occurred during the course of observation of a clinical trial was included in this data base…So –just a minute—Placebo run-ins were counted as placebo, because they were on placebo. There were also suicides that occurred in patients that were on Paxil, for example, in long term extension, where there was no placebo, there was no active comparator, so the full breadth of the experience was accommodated in this particular analyses.
Q: I agree, that’s what happened, that’s exactly what the data represents, but my question is whether that is a valid scientific measuring device for conducting clinical trials.
A: Absolutely.
Q: Well, you have rules for collecting data, right? You had rules for establishing a wash-out phase. I mean that concept --
A: You have to recall, Mr. Farber, the protocol for these studies were for different purposes. They were for treating depression. The protocol for establishing an answer to the question of the occurrence of suicides on placebo or drug or active comparator did included that as I described, and that is collecting all suicides that occurred during observation in a clinical trial, regardless of whether it happened in placebo run-in, during randomization or during extension, open label extension, where people were on Paxil, and they, in some cases, knew they were on Paxil.
Q: And you consider this a valid scientific practice?
A: Absolutely.
(Wheadon deposition transcript P404 L17 through P407 L17)
B. The 1990 paroxetine data.
(1) The 1990 paroxetine suicide data appears to be a resubmission of the original 1989 suicide report. This followed an October 3, 1990 telephone call to GSK’s ("SKB") Thomas Donnelly from Dr. Brecher requesting additional suicide data from that originally submitted – in particular additional data on the number of patients who experienced progressively worsened suicidal symptoms while on paroxetine treatment as judged by the "Hamilton" scale for depression. According to the 10/3/90 internal GSK memo written by Donnelly on the content of the conversation, Dr. Brecher prefaced his request with comments on the "public relations" aspect of the issue, and that the FDA did not view the (suicide) issue as "a real issue." On March 13, 2003, when shown the10/3/90 internal memo, Dr. Brecher disavowed certain aspects of the memo, but did acknowledge conveying the basic data request to Donnelly. Dr. Brecher acknowledged that the additional data request was based on the ongoing SSRI controversy involving Prozac and the associated issue raised by Dr. Martin Teicher’s critical article of Prozac occurring in the February, 1990, issue of "American Journal of Psychiatry." Dr. Brecher acknowledged that his duties as the FDA’s "Safety Review" officer for Paxil, was further heightened by his necessity to evaluate 350 unexplained Prozac suicides, reports of which had been submitted to him at the FDA.
FDA’s Requirements on July 18, 1990. It was the FDA’s position on July 18, 1990, as promulgated by Dr. Leber, that Eli Lilly in order to remove the suicide cloud from Prozac would be required to conduct one of the following trials: "(1) a case control retrospective study,".. "(2) cohort study…," or "(3) best would be a large blind prospective study." (July 18, 1990, Eli Lilly Memorandum of Leigh Thompson)….
On March 29, 1991, in response to Dr. Leber’s direction, Eli Lilly Submitted a Draft Protocol of a Prozac "rechallenge" protocol and clinical trial to the FDA for the purpose of testing the suicide hypothesis. Lilly psychiatrist Charles Beasley was a principal drafter of the protocol. (David Wheadon, MD, deposition of October 19, 2000. Dr. Wheadon, a psychiatrist, was an official at Eli Lilly during the Prozac controversy, and in early 1992, was recruited by SmithKline Beecham. He remains employed by GSK (SKB et al)
Q: "Do you recall that Eli Lilly drafted and submitted to the FDA a rechallenge protocol?
A: "I recall that there was a discussion of a rechallenge protocol, yes." (page 45 line 15-18).
Q: "Do you remember this document being drafted and submitted to the FDA in the spring of 1991?"
A: "I recall this document that you put in front of me being discussed. I don’t recall exactly when it was submitted to the FDA, no." (P47 L11-16) Q: Do you have any reason to doubt the date on the first page of March 29, 1991?"
A: "That’s the date on the front page, yes." (P47 17-19)
Q: Now, Charles Beasley actually drafted the protocol, didn’t he?"
A: "I don’t recall exactly who drafted the protocol, but Doctor Beasley was probably the person—one of the people responsible in drafting it, yes." (P47 L20-25).
No SSRI Suicide Trial Ever Conducted. Despite the FDA directive to Eli Lilly by Dr. Leber on July 18, 1990 to conduct an SSRI suicide trial, and the submission of the "rechallenge" protocol on March 29, 1991 by Lilly, no such trial was ever conducted to determine whether SSRI’s cause suicidality in any segment of the population.
(Dr.Wheadon Deposition
October 19, 2000)
Q: So neither of these companies…(Lilly and GSK)…, to your knowledge, has ever conducted a prospective randomized control trial to determine whether or not these drugs cause some patients to become violent or suicidal; isn’t that true?
A: I’m not aware of such, no.
(Wheadon transcript P42 L19-24)
Q: Okay, now, is it true that the Lilly clinical trials were not intended to address the issue of suicidality?
A: The trials were designed – these trials were designed to address the issue of depression, of which suicidality may or may not be a part of.
Q: Is it true that the Lilly clinical trials were not intended to address the issue of suicidality?
A: In terms of the issue of emergence of suicidality as to relation to treatment, they were not designed with that question in mind, no.
Q: And the same can be said of the SmithKline Beecham clinical trials on Paxil; isn’t that true?
A: That’s correct. (Page 111 L22 through P112 L14)
The 10/3/90 telephone call placed to GSK’s Thomas Donnelly by Dr.Brecher indicated GSK should respond by additional suicide data by the "end of November."
24 Q. No, okay. All right. So this
Martin Brecher, M.D. - Mr. Farber 267
1 happened in October 3, 1990, you told him to --
2 that the FDA wanted a new report on the data from
3 -- on Hamilton and to look for signs of suicidality
4 in the database.
5 Now, do you know when they submitted
6 that report that you had asked for on October 3rd?
7 A. No, I don't remember the exact date or
8 even the week or month.
9 Q. But you had indicated that by the end
10 of '90 that you were essentially completed with the
11 safety and efficacy review?
12 A. To the best of my recollection.
13 Q. Uh-humm. I'm wondering why you would
14 have been essentially completed with that safety
15 review when you hadn't yet obtained their report
16 that you'd asked for in October 3rd?
17 A. I don't think the report would have
18 been complete without it. So if I said -- so I --
19 I think that I must have received it within the
20 time frame specified.
21 And the reason I said -- said that I
22 thought I was finished by nineteen -- end of 1990
23 was that I remember there was a gap of -- a little
24 bit of a gap of time where I wasn't working on
Martin Brecher, M.D. - Mr. Farber 268
1 Paroxetine prior to my leaving. And I also
2 remember that there was a time when I thought I was
3 finished and then the electronic submission came in
4 and I worked with that.
5 So I don't -- to my recollection or my
6 understanding of my role with the Paroxetine safety
7 review is not critical whether I was in fact
8 finished by December of 2000 -- of 1990.
9 My recollection is that when I -- that
10 in order -- when the time came that I thought
11 the -- the paper safety review was largely finished
12 would have been after I had received this data.
13 Q. This data you requested in October?
14 A. Correct.
15 Q. But if you had completed essentially
16 your safety review at the end of the year, you were
17 missing -- well, let me ask you this: Was this an
18 important component of the safety of Paroxetine at
19 that time?
20 A. Yes, it -- it is and was. And my --
21 my sense of having completed, as I remember it now,
22 my sense of having completed the Paroxetine safety
23 review based on the paper documents should have
24 included that requested material and that -- and
Martin Brecher, M.D. - Mr. Farber 269
1 the -- given that I think I asked that he -- to
2 receive it by the end of November is consistent
3 with my earlier recollection that I had finished
4 the paper assessment by the end of December.
5 Q. Okay. The paper assessment but that
6 obviously did not include or would not include this
7 particular report?
8 A. Yes, it would have. When I was
9 talking about the paper assessment, I'm meaning to
10 distinguish the documents submitted in hard copy
11 form from the electronic database that I
12 subsequently reviewed.
13 And so when I speak about the paper
14 document, I would be referring to that submission
15 as well, namely to the submission regarding
16 suicidality.
17 Q. So you believe right now, although
18 you're not 100 percent certain that that particular
19 report in paper form reached you about near the end
20 of December or thereabouts?
21 A. Near the end of November about.
22 Q. Okay. Are you aware that there -- I
23 don't have the report in front of me because I
24 haven't got it yet, but are you aware that the
Martin Brecher, M.D. - Mr. Farber 270
1 suicide ideation report that you requested was not
2 signed out of SmithKline Beecham until May 10th of
3 '91?
4 MR. DAVIS: Object to the form of the
5 question. Assumes facts not in evidence.
6 Asks the witness to speculate.
7 A. Well, my recollection is that I had
8 information prior to that and -- and as -- it was
9 in my report so I -- I got it before I left.
10 I don't recall a last minute rush to
11 complete the data based on the suicidality
12 information. So my recollection is that I had had
13 it earlier, but if I'm mistaken then so be it.
14 Q. Okay.
15 A. In answer to you -- just one other
16 point, in answer to your -- to the implied
17 contradiction, I don't -- I don't know whether they
18 had sent me -- they may have sent me material prior
19 to that May 10th date but again, I'm -- it's to the
20 best of my recollection.
21 Q. By -- by may have sent you, are you
22 interpreting or assuming something we talked to
23 earlier about a draft or interim?
24 MR. DAVIS: Object to the form of the
Martin Brecher, M.D. - Mr. Farber 271
1 question. Calls for the witness to
2 speculate.
3 A. Let me only say that I do not
4 recall -- I recall being -- having the sense of
5 being finished by the end of 1990 as I said.
6 Q. Yeah.
7 A. I do not recall needing that
8 suicidality piece or just finishing it just at the
9 end. Perhaps it occurred that way. That's not the
10 way I remember it.
11 Q. Okay. You remember that it came in in
12 late December or near the end of the year?
13 A. At the end -- my recollection is I got
14 it around the time that I asked for it.
15 Q. Okay.
16 A. And I can also say that had I
17 requested something in October and not having
18 received it until May especially knowing that I was
19 leaving, I probably would have remembered that they
20 had hung me up. So I -- I'm pretty sure that I
21 received data from them prior to May of '91.
22 Q. Okay. Now, in addition to the Ham --
23 HamD issue here of developing suicidality, the
24 suicide report that they forwarded you was more
Martin Brecher, M.D. - Mr. Farber 272
1 comprehensive than that, wasn't it?
2 A. I refreshed my memory by looking at
3 the suicidality section in my report. I don't
4 recall what precisely they sent me.
5 Q. Okay. Let's look at -- back to your
6 safety report. I hope I have it in front of me,
7 your safety review of June of '91.
8 And I want you, if you will, you can
9 get it, Doctor, to put Exhibit 15 next to your
10 safety review.
11 Okay. When you get done, let me know.
12 Let's go to your safety review page
13 25. Do you see the top table that says table 13?
14 A. Uh-humm.
15 Q. And you see the numbers up there under
16 that database, the same numbers we talked about
17 earlier, the 2,963, 554 and 1,151; do you see that?
18 A. Yes.
19 Q. All right. Get -- you'll notice in --
20 let's just take placebo, attempted suicides on page
21 25, middle column there, attempted suicides we've
22 now got six under placebo?
23 A. Yes.
24 Q. Now, look at back at Exhibit 15 and
Martin Brecher, M.D. - Mr. Farber 273
1 look up on the top, we've got the same population
2 2,963, 1,151 and 554; where did the six suicides
3 come from, six attempted suicides under placebo?
4 A. I don't know.
5 Q. Is there anywhere in your table 13
6 that indicates, whether attempted or actual
7 suicides, that we have a placebo wash-out/run-in
8 issue?
9 A. I'm just going to ask for a moment to
10 review this section.
11 Q. Take as long as you want.
12 A. Okay.
13 Q. Isn't it true that table 13 at the top
14 of page 25 on your safety review is an exact
15 duplicate of the amended report that SmithKline
16 Beecham sent to you in response to your request in
17 October?
18 MR. DAVIS: Object to the form of the
19 question.
20 A. I don't recall -- I don't recall what
21 SmithKline sent me. I assume that this, that table
22 13 was taken from the data that they submitted to
23 me.
24 Q. You're absolutely correct.
Martin Brecher, M.D. - Mr. Farber 274
1 And you see from the same population
2 back in Exhibit 15 that there is -- according to
3 this, on suicide attempts there's -- both of them
4 are run-in and this -- we discussed that other
5 issue and as to the actual suicides of wash-out and
6 run-in, they, two here, that you've previously
7 testified that if they occurred during wash-out or
8 run-in, it would be an illegitimate computation of
9 data?
10 MR. DAVIS: Object to the form of the
11 question. Asks the witness to speculate
12 about how he conducted his analysis without
13 providing him all the documents in order
14 to -- for him to make that determination.
15 BY MR. FARBER:
16 Q. Of course the sheet you're looking at
17 now only talks about attempted suicides, not
18 actual. You'll have to go to that exhibit that had
19 the two actual suicides on it, that three-page
20 exhibit. What was -- there was a three-page
21 exhibit I had that had the one we discussed and
22 then the two run-in; do you remember that exhibit?
23 Let's get it.
24 A. I think it's 16.
Martin Brecher, M.D. - Mr. Farber 275
1 Q. It's Exhibit 16. And we see from
2 Exhibit 16 as to patient number 7119.009 and
3 7119.062 that both of these suicides occurred
4 during the wash-out period, run-in period?
5 MR. DAVIS: Object to the form of the
6 question.
7 A. What I -- what I see in Exhibits 15
8 and 16 is one attempted suicide in the
9 postrandomization period.
10 Q. Period, for all suicide events?
11 A. Now, that's what's present in Exhibits
12 15 and 16. I don't -- I can't tell you how they
13 arrived at the number six.
14 Q. Okay. Thank you.
15 Now, back to my thing of the small
16 numbers swinging a great deal of results one way or
17 the other if the placebo numbers are low, the
18 placebo numbers were much lower than the Paxil
19 numbers, were they not, 554 to 2,963?
20 A. Correct.
21 Q. And we're changing from one event
22 under placebo under the document you just talked
23 about to two plus six which is eight, eight placebo
24 suicides are reported in that document, are they
Martin Brecher, M.D. - Mr. Farber 276
1 not, that they submitted to you; two actual and six
2 attempts?
3 A. That's correct.
====================================================
CLICK HERE FOR INDEX
OF DR. HEALY'S CORRESPONDENCE
|
Contents page |
|
What's New? |