Department of Health
Market Towers 1 Nine Elms Lane London SW8 5NQ
Telephone 0171-273 0270
Facsimile 0171- 273 0293
20 November 1997

Dear Mr Medawar,


Thank you for your letter of 23 October 1997, which I received on 24 October. As requested we are treating this as a formal request for information under the Code of Practice on Access to Government information. We are concerned that you should question the competence and integrity of the MCA, and find no basis for this. The responses to each of your points are provided below:

1 . The MCA's policy on responding to requests for information is based on the Code of Practice on Access to Government Information (the Code). Our target for responding to requests for information remains 20 working days. The following changes in relation to the MCA's policy on disclosure of information have taken place since August 1996:-

You may wish to note that Ministers have asked the MCA to examine its current policies on access to information about medicines, with a view to maximising public access whilst safeguarding legitimate confidential interests (see Official Record (Hansard) 24 July 1997, Col. 726). This review is now underway, taking account of the forthcoming White Paper on Freedom of Information and in the context of European Community licensing and control procedures.

2. See response to question 4.

3. You requested specific information relating to clinical trials data about paroxetine. The Agency would normally treat such information as commercially confidential, as it was supplied by the applicant for the marketing authorisation on the understanding that it would not be released (exemptions 13 and 14 of Part 11 of the Code). However, the general presumption under the Code is that the MCA will make information available unless there are sound reasons of public interest for not releasing it. We are therefore consulting the marketing authorisation holder and will then decide what, if any, relevant information can nonetheless be disclosed. I will respond to you further on this matter as soon as I can.

4. The original data sheet for Seroxat (paroxetine) advised: "as with any psychoactive medicine it is advisable to discontinue therapy gradually as abrupt or sudden discontinuation may lead to symptoms such as disturbed sleep, irritability or dizziness". This statement was based upon the findings of pre-licensing clinical trials which provided some evidence of symptoms occurring after withdrawal, but the incidence was less than that seen with commonly used anti-depressants. The applicant was therefore requested to include a precautionary warning in the data sheet. Comments in the paper concerning the absolute risk of withdrawal reactions with any of the SSRIs aimed to explain the conflicting findings of spontaneous reporting of adverse drug reactions (ADR) and prescription event monitoring (PEM). Such arguments are conventional in the discussion section of scientific papers. Since we published our data, a recent analysis of events detected through PEM (F J Mackay et al (1997), Pharmacoepidemiology and Drug Safety 6.. 235-246) showed that withdrawal symptoms were more frequent with paroxetine than the other 3 SSR1s, with 15 reports out of 13,741 patients with follow up information at 6 months or more after first prescription (0. 1 %), and only 2 with each of the other SSRIs (out of 10,983-12,734 patients in each group). The risk of suffering symptoms on discontinuing an SSRI has been the subject of discussion in published literature. However, a consensus on the frequency of clinically important symptoms on SSRI discontinuation does not appear to have been reached. Studies in general have been relatively small. None of the published studies have included the level of exposure of patients reached by the cohorts for PEM of SSRIs (over 10,000 patients in each case).

5. I remind you that paragraph 1 of my letter of 15 September 1997 indicated that, at our request, the applicant included advice against abrupt withdrawal of paroxetine in the data sheet. Please refer to paragraph 4 for the reasons regarding the original data sheet advice. This was changed in 1993 on the basis of spontaneous reports of withdrawal reactions and the advice is now: "As with many psychoactive medicines, abrupt discontinuation should be avoided. Symptoms including dizziness, sensory disturbance (e.g. paraesthesia), anxiety, sleep disturbances (including vivid dreams), agitation, tremor, nausea, sweating and confusion have been reported following abrupt discontinuation of Seroxat. They are usually self-limiting and symptomatic treatment is seldom warranted. No particular patient group appears to be at higher risk of these symptoms; it is therefore recommended that when anti-depressant treatment is no longer required, gradual discontinuation by dose-tapering or alternate day dosing be considered". The British National Formulary incorporated advice against abrupt discontinuation in March 1994.

6. In respect of points 6-1 0 in your letter, benzodiazepines are well-recognised to be drugs which cause dependence (as defined by the WHO criteria) with abuse potential, and in this respect they are different to SSR1s, which have only been associated with withdrawal symptoms. In our view, their dependence potential was clearly indicated in the warning issued by the CSM in Current Problems 1988, which reflected the Committee's considerable concern over this matter. Product information for benzodiazepines states that dependence potential is increased with higher doses and long-term use, and indicates that withdrawal symptoms are a feature of dependence (i.e. they are not the only aspect of the problem).

7. See response to question 6.

8. Contrary to your suggestion that the MCA and CSM's view that benzodiazepines are drugs of dependence is out of line with other bodies, the views expressed above are widely accepted amongst professional bodies. Furthermore they are consistent with the view of other regulatory authorities, both in the EC and around the world. With regard to the Report of the Royal College of Psychiatrists, your quote makes it clear that dependence is mainly manifest by withdrawal symptoms. The report also discusses abuse and tolerance associated with benzodiazepines hence encompassing the WHO criteria.

9. The recognition of dependence with benzodiazepines in the mid-1980s was not based on ICD 10 criteria published in 1992 - we quoted these as the most current criteria. ICD9 and DSM included similar criteria in the late 1970's and 80's.

10. We disagree that there is negligible evidence for a qualitative difference between the experiences of patients on cessation of benzodiazepine therapy and the experiences of patients who discontinue treatment with SSRIs. I pointed out differences in my previous letter. This position is supported by views of authors of published papers. Other features of dependence with the benzodiazepines include tolerance, and abnormal drug seeking behaviour (e.g. manipulation of doctors to obtain prescriptions, alteration of prescriptions to provide larger quantities).

11. As requested I enclose a blank copy of the questionnaire referred to in the British Journal of Clinical Pharmacology paper, together with the covering letter.

We do not consider that we emphasised how mild symptoms usually were in our publication. We presented the data received. In the questionnaire survey, the majority were described as "moderate". Our presentation of the data and our comments were not biased. Describing them as biased would misrepresent the true situation.

12. We accept that few yellow card reports have been received for benzodiazepines and dependence (point 6c and 12) and recognise that the evidence leading to the recognition of dependence with benzodiazepines did not come from the yellow card scheme. Although the scheme has been very successful in identifying drug safety hazards over the years, we have never claimed that it is always effective, and we are well aware of its limitations. In this respect, the figures you quote in point 12 of your letter illustrate that dependence for benzodiazepines was very poorly identified by spontaneous reporting. In contrast the scheme identified symptoms occurring on withdrawal of SSRIs which were not associated with other features of dependence. This large number of reports for SSRIs partly reflects the higher overall reporting rate in the late 80's/90's for new medicines compared to the reporting rate in the 60's and 70's. It is also likely that our action to inform doctors of the issue in the "Current Problems" article in 1993 stimulated reporting. Publications in the literature on this subject are also likely to impact on reporting.

In summary therefore our conclusions concur with your own in that we consider that paroxetine is associated with a variety of symptoms occurring on discontinuation. However, we differ from your view in that we consider that these symptoms are not part of a dependence syndrome and that the product information reflects the evidence available. This contrasts with the well established position of benzodiazepines as drugs of dependence. We consider therefore that your criticisms of MCA/CSM's view on the matter are not well founded.

If you have a query about this letter, please contact me.

Yours sincerely,

Dr J Price
Head of Cinical Evaluation Unit


Contents page
List of MCA/CSM Correspondence