Social Audit Ltd
P O Box 111 London NW1 8XG
Telephone/Fax 44 (0)171 586 7771


Dr J S Price
Head, Clinical Evaluation Unit
Medicines Control Agency
Market Towers, 1 Nine Elms Lane
London SW8 5NQ 23 October 1997

Dear Dr Price,

Thank you for your letter of 14 October. In case they have not already been informed about my enquiries, I am copying this reply to Dr Jones and Professor Rawlins. This is to underline that what concerns me relates to the competence and integrity of the MCA and CSM, rather than the impact of your paper on readers of the British Journal of Clinical Pharmacology. This is also to give notice that this is a formal request for information, under all relevant Agency standards and codes. Had you agreed to a meeting, this might have been avoided and it could have saved us both a lot of time; however, I am grateful to you for saying you would be pleased to respond to any points I want to raise. Here are a few.

1. Can I please start by asking if the MCA/CSM policies on disclosure have significantly changed over the past twelve months and, if so, please will you indicate what changes have been made and which, if any, were formally discussed by the CSM? In the meantime, I am assuming that the MCA's rules on disclosure remain based on the Code of Practice on Access to Government Information (1994) and on its Standard Operating Procedures, requiring, for example, a target response time maximum of 20-working days.

2. Please will you explain the evident contradiction between what is said in your published paper and what you say in points 1, 6 and 7 in your letter of 15th September? If evidence of withdrawal reactions in pre-licensing clinical trials of paroxetine was sufficient for the MCA/CSM to request the manufacturers to include a warning, its seems either very confused and/or misleading to suggest that withdrawal reactions from SSRIs are so rare that they barely show up in a PEM analysis and that the risk may be so low as to be "undetectable except through spontaneous reporting where drug exposure is high".

3. On the basis of data published by the US Food & Drug Administration about the numbers involved in pre-clinical testing of paroxetine, I would be inclined to think that withdrawal reactions were observed in trials in which only a few hundred patients were involved. However, please will you let me know the following, to explain the incidence observed in relation to the trials' design and size and characteristics of the patient base:

[a] Did the protocols relating to any of the pre-licensing clinical trials of paroxetine require that patients be withdrawn from the active drug in order to observe possible withdrawal symptoms, and if so, how many patients completed such trials?

[b] Did the MCA/CSM make any specific requirement of the Product License applicant to investigate the incidence and severity of withdrawal reactions and, if so, what was the duration of active drug treatment, and how many patients completed such trials ?

[c] What proportion of patients involved in all pre-licensing clinical trials experienced withdrawal symptoms, in relation to the total numbers who discontinued treatment with paroxetine during the study period, excepting those who withdrew because of problems with side effects or protocol violations?

[d] What proportion of patients involved in properly conducted, placebo-controlled, pre-licensing clinical trials experienced withdrawal symptoms, in relation to the total numbers who discontinued treatment with paroxetine during the study period, excepting those who withdrew because of problems with side effects or protocol violations?

[e] What proportion of patients involved in pre-licensing clinical trials of more than six weeks' duration experienced withdrawal symptoms, in relation to the total numbers who discontinued treatment with paroxetine during the study period, excepting those who withdrew because of problems with side effects or protocol violations?

4. In the light of the answers to question 3, does the CSM/MCA consider it safe to conclude that the incidence of withdrawal symptoms in normal clinical use of paroxetine would be lower than 1%? And if you cannot on this basis confirm your conclusion that withdrawal reactions are rare (you imply an incidence of the order of 0.001%), will you withdraw it?

5. In point 1 of your letter of 15th September, you said that the advice against abrupt withdrawal of paroxetine was included in the data sheet at the request of the MCA/CSM. Please will confirm my understanding from this, that the manufacturers had not themselves proposed inclusion of such advice in the data sheet.

6. In point 3 of your letter of 15 September, you suggest that the MCA/CSM regard benzodiazepines as drugs of dependence because features of dependence such as "craving, drug seeking behaviour and dose escalation" have been reported for benzodiazepines but not for SSRIs. In this connection, please will you confirm each of the following:

[a] the MCA/CSM has never issued any warning or other statement on benzodiazepines relating to any features of dependence other than withdrawal symptoms.

[b] no warnings relating to specific features of dependence other than withdrawal symptoms have ever appeared in any benzodiazepine data sheet

[c] over the past 30-plus years, the CSM/MCA has received negligible numbers of Yellow Card reports relating to any manifestation of dependence other than withdrawal symptoms - for most benzodiazepines less than one such report per few million prescriptions

[d] there is no evidence in the minutes of any CSM meeting of concern about any features of dependence on benzodiazepines such as "craving, drug seeking behaviour and dose escalation".

7. Would you care to cite any published report giving convincing evidence that craving, drug-seeking behaviour or dose escalation are remotely characteristic of benzodiazepine dependence in therapeutic settings?

8. Please will you indicate any fundamental disagreement you would have with the following characterisations of benzodiazepine dependence by [a] the American Psychiatric Association's 1990 Task Force on Benzodiazepine Dependency, and [b] in the Royal College of Psychiatrists 1997 paper, Benzodiazepines: Risks, Benefits or Dependence - bearing in mind also that neither paper refers to either craving or drug-seeking behaviour in connection with normal therapeutic use:

[a] "the presence of a predictable abstinence syndrome following abrupt discontinuance of benzodiazepines is evidence of the development of physiological dependence"

… "Research studies do not report development of tolerance to therapeutic effects … some clinicians have, however, observed slight increases in benzodiazepine doses over time ... These dose increases are usually small, and long-term use does not lead to significant dosage increases over time or to high dose abuse"

[b] "Dependence on benzodiazepines is mainly manifest by withdrawal symptoms on cessation … Dosage escalation in uncomplicated cases is rare."

9. Please will you confirm that recognition and designation of the benzodiazepines as drugs of dependence, in the early to mid-1980s, was based not on the ICD-10 criteria you cite in your point 2 - but pretty much solely on evidence of withdrawal symptoms following long-term use at recommended dosage levels.

10. In short, I believe the MCA/CSM's has negligible evidence for the argument you offer under points 3 and 5. Dependence on benzodiazepines is overwhelmingly characterised by the existence of a withdrawal syndrome, and they overwhelmingly resemble the SSRIs in this respect. Unless the MCA/CSM can put up good evidence in support of your proposition that "there are therefore qualitative differences to the (dependence-related) reports with SSRIs", I invite you to withdraw it.

11. Please will you send me a blank copy of both questionnaires (with guidance notes) sent to doctors who had reported withdrawal reactions with paroxetine, in the survey you and your MCA/CSM colleagues reported in the British Journal of Clinical Pharmacology? I note your point about the error in presentation of your results, yet you did not explain why you chose to emphasise how mild withdrawal symptoms usually were, when they were as often described as "severe" as "mild". Would I be wrong to use this in a forthcoming series of lectures as an example of unacceptable bias ?

12. You argue (point 3) that it would be inappropriate to attempt any quantitative comparison between benzodiazepines and SSRIs on the basis of the numbers of reports of withdrawal reactions received. I accept that there are confounding factors and that it would be futile to attempt any strict comparison - but I had never suggested this. When I wrote that the numbers of reported withdrawal reactions for SSRIs invite comparison with those reported for benzodiazepines, I had in mind the following remarks, made by one of your MCA colleagues: "Spontaneous reporting schemes cannot provide estimates of risk because the true number of cases is invariably underestimated and the denominator is not known. Yet these systems do provide some measure of frequency. It obviously matters whether the number of reports to a reaction is one or several hundred. Empirically, it seems reasonable to suggest that there is likely to be some relation between the number of spontaneous reports and the true risk of an adverse reaction but this has yet to be investigated." (Waller PC, Measuring the frequency of adverse drug reactions, Br J Clin Pharmacol (1992), 33, 249-252).

I also note that the confounding factors you specifically mention would tend to favour SSRIs in a comparison with benzodiazepines - since the latter have had much longer marketing lives, far greater usage, and have enjoyed/ suffered comparable if not greater publicity. Let me underline the point by comparing withdrawal reactions reported (to March 1997) for the most prescribed benzodiazepines, temazepam and diazepam, with those reported for paroxetine:

Introduced Ballpark No. Yellow Card reports of:
of NHS scripts withdrawal reactions dependence- all forms
Temazepam 1977 50m 5 3
Diazepam 1963 180m 20 16
Paroxetine 1991 6m 802 9

On the basis of these figures, it would be hard to avoid the following conclusions:

[a] It would seem both incomprehensible and potentially dangerous for the CSM/MCA to proceed on the basis that diazepam and temazepam should be designated drugs of dependence, but paroxetine not

[b] No significant differences exist between paroxetine and these two BDZs, in relation to dependence problems other than withdrawal symptoms. There is no credible basis for your assertion that qualitative differences between BDZs and SSRIs justify defining the former as drugs of dependence, and the latter not.

[c] The differences shown in the above table are so stark as to suggest that the main reason for your declining to compare withdrawal reactions to BDZs and SSRIs would be that the MCA/CSM recognises that, in any such comparison, the SSRIs would come out of it looking at least as troublesome as the benzodiazepines, if not much more so.

On this last point, you will appreciate that reports of withdrawal reactions to paroxetine far outnumber those for all other SSRIs (all benzodiazepines too). Nevertheless the presumption would be that some risk of dependence attaches to other SSRIs [a] taking into account that the numbers of reports of withdrawal reactions to fluoxetine and sertraline are at least double those for diazepam; and [b] bearing in mind the point I emphasised in my letter of 23 July, that the essence of the dependence problem is not that some people experience sometime severe symptoms on withdrawal, but that many more continue to take drugs in order to avoid them.

Subject to your comments, also taking into account many other questions I would wish to raise, I maintain the views expressed in my last letter - that the CSM/MCA is making yet another appalling mistake, almost a replay of the events described in my book, Power & Dependence, which I sent you three months ago. Would you not now accept that there is at least a substantial risk that dependence on SSRIs will prove at least serious as with the benzodiazepines? One way and another, your letters of 15 September and 14 October have reinforced me in this view, but I look forward to receiving your further comments on these points in the hope that the MCA/CSM will now accept the need for damage limitation.

Yours sincerely,

Charles Medawar


Contents page
List of MCA/CSM Correspondence