Department of Health
Market Towers 1 Nine Elms Lane London SW8 5NQ
TEL:  020 7273 0400
FAX:  020 7273 0675 .Our Ref: OG/0233
20 August 2002

Dear Mr Medawar,

RE: ADROIT guidance on interpretation of Yellow Card Data

Thank you for your letters of 2 and 4 August in relation to the ADROIT guidance on interpretation of Yellow Card data and paroxetine and withdrawal symptoms. This letter responds to the points made in your letter of 4 August. We are currently looking into the points raised in your letter of 2 August and I will respond separately to these.

I have answered the points as numbered in your letter of 4 August.

1. Thank you for hyperlinking the ADROIT data to the guidance on interpretation. We will certainly take on board your comments about the length and readability of this document. We are very aware of the difficulties associated with the interpretation of Yellow Card data and are working to develop better forms of communication of these data.

2. You question the statement ‘The reporting of a reaction does not necessarily mean that the drug caused the problem and therefore these reactions should not be used as a list of side effects .for the medicine to which it refers’.

We do not consider that this implies that Yellow Card data are meaningless, as you suggest. Reporters are asked to report suspected adverse reactions regardless of any doubts that they may have about causality and therefore it would be misleading to suggest that every adverse reaction reported is definitely causally associated with the drug in question. The value of a spontaneous reporting scheme is in its ability to monitor all drugs in normal clinical practice and to pick up drug safety issues at an early stage. Spontaneous reporting schemes have limitations, which are well recognised and have to be taken into account when the data are interpreted.

We do not consider that this statement would lead the press to believe that the reports sent to MCA/CSM would over-estimate the problem. The unknown extent of underreporting of suspected adverse reactions is well known and is reflected in the notes for guidance. Underreporting is discussed further in the response to paragraph 7 of your letter.

3. Your letter and attachment will form part of the data package to be considered by CSM.

4. The briefing for the Pharmaceutical Journal (PJ) was provided by staff of the Post-Licensing Division through the Department of Health Press Office as is normal practice for press briefings. The names of the individuals involved are being withheld under exemption 12 of the Code of Practice on Access to Government Information (Privacy of an Individual). The written briefing provided to the Pharmaceutical Journal by the Department of Health Press Office is attached at Annex 1.

5. You have asked for an explanation of the meaning of the data in Annex 1 of my last letter. The table is simply a list of the 20 drugs associated with the greatest number of reports of withdrawal reactions received through the Yellow Card Scheme, and the number of reports received for each throughout the life time of the drug. Further interpretation of the data would require information on the usage of each drug over the time period of question and whether there were any factors that may have stimulated or reduced the level of reporting for each drug over the relevant period of time. Even with this information, these data cannot be used to accurately calculate the incidence of an adverse reaction.

6. The data available on withdrawal reactions with SSRIs have increased since the 1996 paper by Price et al. The most recent publication from MCA/CSM on the subject of withdrawal reactions associated with SSRIs is that in the September 2000 edition of Current Problems in Pharmacovigilance. In relation to severity, this article provides the results of a detailed review of the spontaneously reported cases and states that ‘Whilst the withdrawal symptoms reported were generally not serious, there have been isolated reports of more serious symptoms on withdrawal such as severe electric shock sensations, vertigo and manic reactions.’

In relation to the frequency of withdrawal reactions this was again reviewed in detail by CSM and MCA and by CPMP. The principles for Summary of Product Characteristics wording on withdrawal reactions for SSRIs which were developed by the CPMP state that: ‘Strong evidence, which would allow definitive statements about the frequency of withdrawal reactions with the different SSRIs, is not available. Any such statements should not be based on the frequency of spontaneous reports, as withdrawal reactions with certain SSRIs are likely to be subject to significant underreporting. Any statements relating to the frequency of withdrawal reactions should be based on the results of clinical trials where these are available.’

If you have concerns that any promotional activity by pharmaceutical companies is misleading, please send us the relevant material and we will investigate your concerns.

7. You ask for verification of the statement that ‘it has been estimated from various surveys that only 10-15% of serious ADRs are reported’. I attach a list of references for the surveys in question at Annex 2. Although the figure of 10-15% is much cited as the percentage of adverse reactions reported, the true extent of underreporting is unknown and is likely to vary considerably between different drugs and different reactions. A more recent study carried out by the Drug Safety Research Unit and published in the Lancet (Heeley et al, 2001) may be of interest. It is likely that reporting for adverse reactions that occur on withdrawal of medication is lower than for acute reactions as withdrawal reactions are less likely to be recognised as being due to the medicine.

8. You ask for the best estimate of the level of underreporting of withdrawal reactions to antidepressants, including SSRIs. I am afraid that any such estimate would be meaningless. The studies that have been conducted with SSRIs and other antidepressants in which withdrawal symptoms have been recorded have varied considerably in their results depending on the study design and how the data were collected. We do not currently have a robust estimate of the frequency of withdrawal reactions with SSRIs.

9. You question the quote in the PJ which is based on the following paragraph from the notes for guidance:

‘Numerical comparisons should not be made between reactions associated with different drugs on the basis of the data in these prints alone. Comparisons can be misleading unless they take account of variations in the level of reporting, the extent of use of the drugs and a number of other confounding variables.’

This statement is designed to emphasise that numbers of reports of a particular reaction for different drugs, are not adjusted for the length of time the drug has been on the market, the extent usage of the drug or for any other of a number of factors that affect the reporting of suspected adverse reactions through the Yellow Card Scheme.

You ask Do the CSM/MCA think people would be mistaken to infer from the table that something has gone badly wrong, if five of the top six drugs on this list are SSRIs with paroxetine ahead by far?’

The large number of spontaneous reports of withdrawal reactions associated with paroxetine and to a lesser extent with other SSRIs, is what prompted the MCA, CSM and the CPMP to review this issue in detail. As mentioned in the previous letter we fully recognise the public health importance of withdrawal reactions with SSRIs and do not seek to underplay the issue.

If you have a query about this letter, please contact me. If you are unhappy with our decision, you may ask for it to be reviewed. That review will be undertaken by a senior member of the Agency who has not previously been involved in your request. If you wish to pursue that option please write to the Information Centre, 10th Floor, Medicines Control Agency, at the above address quoting reference OGO2/33. After that, if you remain dissatisfied, you may ask a Member of Parliament to make a complaint on your behalf to the Ombudsman (known officially as the Parliamentary Commissioner for Administration) who may decide to conduct his own investigation.

Yours sincerely
Director, Post-Licensing Division


Annex 1

Issue of antidepressant withdrawal reactions

The Committee on Safety of Medicines (CSM) reviewed the specific issue of dependence and withdrawal problems with paroxetine and other SSRIs and related antidepressants in 1998. From detailed review of all available data, the CSM concluded that all SSRIs cause withdrawal reactions, however they are not drugs of dependence.

[It is generally accepted that withdrawal reactions on stopping a drug are not sufficient, or necessary, for a diagnosis of drug dependence. Other features such as tolerance (requiring increased doses of the drug to produce the same effect) and drug seeking behaviour are required for this diagnosis.]

The CPMP (Committee for proprietary Medicinal Products) also conducted a similar review the following year and concluded that the available evidence did not suggest that SSRIs were drugs of dependence, and published their position on this issue in a paper Which is available on the European Medicines Evaluation Agency (EMEA) website. The product information (Summary of Product Characteristics and Patient Information Leaflet) for paroxetine and other selective serotonin reuptake inhibitors contain warnings about withdrawal symptoms. Prescribers have been reminded of the need to taper dosing when withdrawing from treatment on two occasions via ‘Current Problems in Pharmacovigilance’ in 1993 and 2000. In addition the British National Formulary carries a clear warning on this issue.

As was stated in the letter to Charles Medawar from Dr June Raine of the Medicines Control Agency on 3rd July 2002. The MCA and CSM will review and assess data recently received from the MIND patient reporting scheme. However the MCA does not currently systematically collect reports from patients through the yellow card scheme. Patients should be encouraged to consult their doctor if they suspect they are suffering from an adverse drug reaction in order that they receive appropriate advice and treatment without delay.

The issue of withdrawal reactions and dependence associated with paroxetine and other SSRIs is an important public health issue and an expert meeting involving CSM psychiatrists is planned for later this year to further review the issue of withdrawal reactions and the adequacy of current labelling.

Use of MCA data in Social Audit letter:
The list of medications found on Charles Medawar’s "Dear Doctor letter" were supplied by the MCA and are from the UK Yellow Card system. This information is available on written request to any member of the public. However when supplying this information it is always accompanied with an explanation on how to interpret the data which would have been supplied to Mr Medawar, but which he has failed to include. This information explains that:

1. The reporting of a reaction does not necessarily mean that the drug caused the problem and therefore these reactions should not be used as a list of side effects for the medicine to which it refers.

2. Many factors have to be taken into account in assessing causal relationships including temporal association, concomitant medication and the underlying condition. Reporting rates are influenced by the seriousness of the suspected reaction, their ease of recognition, the extent of use of a particular drug and may be stimulated by promotion and publicity about a drug. Numerical comparisons should not be made between reactions associated with different drugs on the basis of this information. Comparisons can be misleading unless they take account of wide a number of factors.

This "dear doctor letter" was not written in conjunction with or knowledge of the MCA or CSM.


Annex 2

Speirs, C J et al (1984). Demography of the UK adverse reactions register of spontaneous reports. Health Trends, 16, 49-52.

Inman WHW & Vessey MP (1968). Investigations of deaths from pulmonary, coronary and cerebral thrombosis and embolism in women of child-bearing age. Br Med J, 3, 746-748.

Inman WHW (1977). Study of fatal bone marrow depression with special reference to phenylbutazone and oxphenybutazone. Br. Med. J. 1, 1500-1503.

Langman M (1986) Peptic ulcer complications and the use of non-aspirin non-steroidal anti-inflammatory drugs. Adverse Drug Reactions Bulletin, 120, 448-451.

Lumley et al (1986) The under-reporting of adverse reactions seen in general practice. Pharm Med., 1, 205-212

Heeley et al (2001) Prescription-event monitoring and reporting of adverse drug reactions. Lancet 2001, 358, 1272-73.

Contents page
List of correspondence with MCA/CSM