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


Dr Susan Wood, Director, Post Licensing Division
Medicines Control Agency
Market Towers, 1 Nine Elms Lane
London SW8 5NQ

30 September 1998

Dear Dr Wood,

As I expect you know, Mr Alder recently sent me a copy of the note, prepared by the Post-Licensing Division on 19 December 1997, described as "a summary and brief commentary of the issues relating to SSRIs and withdrawal reactions/dependence raised by the publication of 'The Antidepressant Web' and the work of the MCA to review the evidence."

In this letter, I should like to outline why I regard this an inadequate and unfair assessment, and to raise (in italics) several questions arising, as formal Code of Practice requests. I thought the assessment unworthy of a legally designated "competent authority", especially one which enjoys (and in other respects deserves) high international recognition.

1. Does this note represent the best assessment of the paper (ADWEB) prepared by the MCA before 19 December 1997?

2. If YES, at what later date was any document (or part thereof) prepared by the MCA which was a significantly more thorough analysis than the one circulated with my paper in December last year?

3. Whether YES or NO, please will you supply me with any documents (or parts thereof) prepared to date, which purport to be scientific evaluations, which in your view (as head of Division) significantly advance understanding of (any of) the main areas of risk my paper discussed?

4. If so, please could you also explain why there has been no indication given to me that any more substantial evaluation of my paper had been prepared?

5. Please may I also know whether the person primarily responsible for this paper was also significantly involved in preparing the MCA's 2 December press statement, as supplied to The Times? If so, please provide fullest details of who was involved and how.

The MCA's December paper summarised my main concern as follows: "Medawar … claims that SSRIs are drugs of dependence. To support these claims, (he) makes comparisons between the number of spontaneous reports of withdrawal reactions and dependence associated with benzodiazepines and dependence and SSRIs on the UK ADROIT database". This clearly implies that this was either the only evidence, or the strongest evidence, I had for suggesting the risk of dependence was real. This is neither fair nor true and denigrates my professional ability and reputation.

6. Can you please provide me with any documentary evidence to show that the MCA has ever specifically considered or evaluated possible risks arising from or in connection with: lack of research and/or evidence into the magnitude of the risk of SSRI dependence?

7. Ditto: the persistent and sharp increases in consumption of SSRI antidepressants in England and elsewhere in the UK?

8. Ditto: the nature and extent of patient and other Internet reports and communications about the extent and/or significance of withdrawal reactions to SSRIs?

9. Ditto: the evidence of dosage escalation with and/or other manifestations of tolerance to the effects of SSRIs?

10. Ditto: the (circa 1992+) change in the official meaning of the word "dependence" - or any other aspect of the semantics of the word "dependence" and/or whether and how this might have affected users understandings of risk?

11. Ditto: the numbers of people likely to continue taking antidepressant drugs on a long-term basis essentially because they could not withdraw from them?

12. Ditto: the possibility that antidepressant withdrawal symptoms had not been adequately detected/reported in any studies relied on by the MCA?

13. Ditto: the probable incidence in any specified population of SSRI users of people who would have problems with withdrawal?

14. Ditto: the steadily increasing incidence of 'depression' and whether this might in part be linked to the unrecognised and unwanted consequences of drug treatment?

14. Ditto: any possible linkage between drug efficacy (real or perceived) and dependence?

15. Ditto: problems resulting from poor judgement or other regulatory inadequacy in assessing and/or communicating the risk of dependence with benzodiazepines?

16. Ditto: the lack of awareness of many doctors that antidepressant withdrawal symptoms existed, and were likely to affect significant numbers of users.

17. Ditto: the possibility of a therapeutic (normal dose) dependence with SSRIs to compare with the risk with benzodiazepines.

18: Ditto: the great amount of publicity about the nature and prevalence of 'depression' and the need to treat it?

19. Ditto: the recommendations for long-term use introduced in 1992, also taking into account lack of understanding of the long-term effects of SSRIs?

20. Ditto: the extent to which recommendations for gradual withdrawal might minimise or avoid problems with withdrawal

21. Ditto: the relationship between risk of dependence (not withdrawal) and drug half-life?

22. Ditto: the significance of yellow card reports about features of dependence other than withdrawal symptoms for SSRIs, and/or of any follow up enquiries by the MCA about such case reports?

The 'Background' section of the MCA 12/97 summary of my paper seeks to explain away any resemblance between BDZ and SSRI withdrawal and/or dependence, and concludes that "the use of spontaneous reporting data to compare these two groups of drugs is inappropriate". But there is no evidence of the reasoning behind this basic assumption: the summary simply states as fact that there have been many reports about SSRI withdrawal reactions and few about BDZs. Without explanation, it simply says that reports about BDZs were few and far between, unlike reports about SSRIs - and from this seems to deduce that comparisons should not be made. Where is the logic of this?

The problem with the MCA analysis is the supposition: "Despite the known extent of benzodiazepine dependence this was poorly identified by spontaneous reporting." All the evidence suggests otherwise. Between 1963 and 1980, even the Committee on Review of Medicines had no idea of the extent of benzodiazepine dependence, and believed it to be negligible. Probably until about 1985, most doctors either ignored or failed to understand the problem of therapeutic (normal dose) dependence with BDZs. By the time most doctors did understand such problems, it would have been tedious, superfluous, unhelpful and even rather embarrassing to send in a yellow card to the CSM. Accordingly, very few were sent - the evidence suggests something of the order of one actual report for every 10,000 reportable cases.

Over the same period, doctors also reported very few cases of comparable withdrawal phenomena with non-SSRI antidepressants, and the evidence unequivocally shows that doctors were mostly unaware of this problem and/or mistakenly regarded such symptoms as signs of relapse. This again suggests it is mistaken to argue that there were very low reporting rates in spite of high levels of awareness.

Against this background, I would expect the huge (relative and absolute) increase in reports of withdrawal problems for SSRIs to be related mainly to the following:

The MCA's 12/97 summary also mentions the question of reports of dependence other than those relating to withdrawal symptoms - a point underlined in the MCA's 2 December press release: "SSRIs … have not been associated with the other characteristics of dependence … (but) dependence with benzodiazepines may include other features such as craving, drug seeking behaviour and dose escalation (tolerance)". This conveniently (and unacceptably) overlooks the fact that the CRM found negligible evidence of such features of dependence following its "systematic review of the benzodiazepines" in 1980 - ie 20 years after the introduction of the first of these drugs.

In contrast to normal dose dependence, such features of dependence as craving, drug-seeking behaviour or dosage escalation would be hard to miss and eminently 'reportable'. Therefore the MCA analysis needs to explain why few cases were reported with BDZs - and why comparable numbers have been reported for SSRIs, in spite of far lower volumes of consumption.

The MCA 12/97 summary clearly suggests that no reports of these other features of dependence for SSRIs existed around 1995, when Price et al (1996) were preparing their review: "Assessment of the UK spontaneous reporting data at that time did not identify other features of dependence associated with reports on withdrawal reactions with SSRIs."

23. Please will you confirm, if possible with documentary evidence, that no such reports for SSRIs had been received - at least not in numbers which bear comparison with the numbers of such reports available today?

In any case, it might help to avoid all the problems caused by the tampering with the goalposts of definition if one thinks in terms of two distinct kinds of dependence problem - one characterised by drug consumption in a typical therapeutic setting (Type A), the other much closer to Skid Row (Type B). My paper was overwhelmingly concerned with Type A, what used to be called 'normal dose dependence'. Regardless of any question relating to Type B dependence with either BDZs or SSRIs, Type A dependence affected hundreds of thousands of BBZ users in the UK, some miserably, and my concern is that the same thing is going to happening with SSRIs.

The MCA's review started ten months ago and the indications are - to judge from the decision taken by the CPMP Pharmacovigilance Working Party - that it was completed at the latest by June. The terms of the "warning" then adopted by the CPMP - apparently largely at the instigation of the MCA - suggest the MCA continues to believe, after months of research, that the problem of SSRI withdrawal is slight, and the dependence risk nil.

24. Will you please confirm that the MCA's review has in fact been completed, indicating also when its conclusions will be published - and the analysis open to peer review?

I apologise for spelling out all these queries in the form of a long list of Code requests. I regret this, but after more than 12 months of often fruitless and very time-consuming correspondence, I could see no alternative. Having said that, I hope you appreciate that all I'm after is some straightforward and intelligible answers - so I'd be quite happy if you can find some other way of coming to these concerns which improves on all the formal Code stuff. I look forward to hearing from you.

Yours sincerely,
Charles Medawar

cc Mr Roy Alder


Contents page
List of correspondence with MCA/CSM