Coleg Meddygaeth Prifysgol Cymru

University of Wales College of Medicine

hergest.bmp (2190 bytes)
Adran Meddygaeth Seicolegol / Department of Psychological Medicine
Adran Cymru y Gogledd / North Wales Department
Pennaeth Adran / Head of Department
Yr Athro /Professor Michael J. Owen
Dr. D. Healy (Cyfarwyddwr)


Dr Keith Jones, Director 1 September 2000
Medicines Control Agency
Market Towers, 1 Nine Elms Lane

Dear Dr Jones,

Thank you for your letter of the 23rd of August. Far from being reassured by it however I find a number of grounds for even more concern.

My first but perhaps least pressing concern is with the fact that as I take it from your letter the summary you sent me essentially is one that you passed on from the different companies. If this is the case I cannot see how you can be in a position to say that there aren't similar problems to those found in the Hindmarch study with either Sertraline or other SSRIs.

Secondly you characterise the Hindmarch study as small. It may be small in absolute numbers compared to some other studies but for the most part these other studies involving healthy volunteers were single dose studies. The Hindmarch study was in fact a very powerful study from a statistical point of view given its double blind nature and given its conclusive results. I'm sure your statisticians could provide you with figures for relative risk for Sertraline compared to placebo in this study.

A second point regarding this Hindmarch study is that you now, with our study, have a further study involving a similar pattern of severe adverse effects to sit with the Hindmarch study. It is my understanding that these two studies on their own would be sufficient to grant a licence to Pfizer for the use of Sertraline as a drug to produce agitation and apprehension.

In my letter of the 3rd of August I invited you to answer the question whether, given these two studies, you as the Chairman of an Ethics Committee or an Insurer for healthy volunteer studies would permit a healthy volunteer study involving Sertraline to take place without monitoring and warning. Your letter of the 23rd provides no answer to this question. Can I invite you once more to answer?

An alternative question would be how you predict the Chairs of Ethics Committees or Insurers would be likely to respond to this scenario. I would hope to be able to tell you how they respond soon. It would be useful to have your prediction on the record first.

A further problem I have is the list of studies, which you have listed in the reference section as epidemiological studies. None of these other than the Jick Study is an epidemiological study. I'm staggered that you can consider any of the others as being epidemiological studies. I'm sure you will agree that an epidemiological study should sample a very large section of the population or indicate what steps wore taken to ensure the sample that is being used is representative of the population. This is standard methodological procedure. None of the studies you've listed other than the Jick Study did this and none were designed to test the question of whether Prozac or any other SSRI can cause suicidal ideation.

The Leon study was a study involving 640 odd patients that was conceived 20 years before Prozac was launched and begun 10 years before it was launched. Only 180 odd patients got Prozac. This was not an epidemiological study and not a study to design to test whether Prozac could cause suicidal ideation.

The Warshaw and Keller study is a study of anxiety disorder patients with only .654 patients of whom again 180 odd got Prozac and the only suicide in this study was on Prozac. This is not an epidemiological study, furthermore neither the Leon nor the Warshaw and Keller study can be regarded as in any way disinterested research.

As regards the Fava and Rosenbaum study this is a study that you may or may not know has been repeatedly reanalysed by FDA officials and a range of other personnel and everybody else other than Fava and Rosenbaum find that the ratio of emergent suicidality on Prozac compared to other antidepressants in the study shows a relative risk of 3.0 or greater.

The Jick study for most epidemiologists that I have consulted provides very strong evidence implicating Prozac in not just the emergence of suicidal ideation but completed suicide.

Perhaps you have not looked at the above studies yourself and therefore are unaware of how inadequate they are for the purposes you appear to be now putting them. Perhaps you have depended on advisers. I enclose a report written by I believe an adviser to the CSM/MCA around the 1990 period when the controversy about SSRIs and suicide first emerged. This adviser was also an adviser to Eli Lilly and other SSRI producing companies around this time. The enclosed report is a paper produced for Lilly, dealing with what later came to be called the Beasley Meta-analysis/paper. I don't suppose you have a similar report on record from this period. If not I would imagine that most observers would be deeply worried by the discrepancy between the advice being given to you by your advisers on this issue compared with the advice the same advisers will have been giving to some of the companies involved. Whatever you or other observers will make of the discrepancy between the enclosed report and the advice given to you, the discrepancy between advice that the studies you cite (other than the Jick study) are epidemiological studies and what is generally held to be an epidemiological study raises questions of basic scientific literacy.

On the regulatory action you are proposing to take, can I respectfully suggest to you that this will lead to deaths. It will do so because of the following.

As phrased your additional wording says that the possibility of a suicide attempt is inherent in depression and that it it general clinical experience with all antidepressants that the risk of suicide may increase in the early stages of recovery. There is some truth to this statement.

However a great number of patients, perhaps the majority who get SSRIs either do not get them for depression or for depression that has a significant suicide risk associated with it. These are primary care nervous conditions that do not have a significant risk of suicide. They are more like the healthy volunteers in the Hindmarch study and our own study than they are like classic depressive who are at high suicide risk.

Clearly in both the Hindmarch and our own study apprehensive ruminations and suicidal preoccupations emerged in healthy volunteers. There was no hint of depression in any of those involved. There would appear to be a clear drug induced component in both these studies that provide you with sufficient grounds to include some reference to this in your proposed wording.

Unless your wording indicates that the treatments may in their own right add a further risk of emergent suicidality to whatever risk is inherent in depressive disorders themselves your advice will lead to a situation where patients who worsen an treatment will be kept on that treatment by their Genera[ Practitioners in the belief that it is only in this way that the suicide risk can ultimately be lowered. This is mistaken advice that is going to increase the rate at which patients move from emergent suicidality to suicidal acts.

But it will also do something else. By blocking off a recognition of the association between the drug and emergent suicidality, you are going to force patients who suffer from this side effect to consider themselves flawed in some way because of these developments. This in turn can be expected to have a significant negative impact on their future wellbeing, mental state and risk of suicide.

On the point you make that it is general clinical experience that this can happen with all antidepressant therapies, I would agree with you but I would also suggest that you write to Professor Ross Baldessarini at McLean Hospital. Harvard University who can give you figures on just this issue. Professor Baldessarini has been working on figures based on new drug applications to the FDA, presumably the same clinical trials that you have had available to you. From this material he has constructed relative risks for SSRIs, tricyclic antidepressants, placebo and other treatments. These risks differ dramatically and significantly between all antidepressants with the SSRIs posing a greater risk than older tricyclic antidepressants in inducing suicide attempts in the course of clinical trials. The relative risk is probably four to fivefold greater for SSRIs. Professor Baldessarini I believe would be able to provide you with the figures were you to approach him, although given the material he is working from you should have everything in-house. There is also the recently published analysis by Khan et al using the NDAs for the SSRIs showing an increased relative risk of emergent suicidality on SSRIs compared to placebo. I would have thought you would want your advice to physicians and to patients to reflect the data accurately. I do not believe the proposed advice does so. I don't intend to speculate as to why you might not wish to accurately reflect the data.

Finally please let me apologise for any infelicities there may be in the tone of this letter. Pointing out basic facts about epidemiological methodology may be advising you how to suck eggs and may be completely out of place in a letter such as this. My tone stems from the fact that I find the situation that has evolved increasingly incredible.

Yours sincerely

David Healy MD FRCPsych
Director, North Wales Department of Psychological Medicine


Uned Hergest, Ysbyty Gwynedd, Bangor, Gwynedd LL27 2PW
Ffn: (01248) 384452 Ffacs: (01248) 371397
Hergest Unit, Gwynedd Hospital, Bangor, Gwynedd LL27 2PW
Tel: (01248) 384452 Fax: (01248) 311397

The MCA acknowledged this letter on 4 September: "Dr Jones has been made aware of your letter and a response will be forwarded to you as soon as the necessary information has been gathered."


Contents page
What's New?