1. Introduction
The complainant in this case is Mr Charles Medawar, a research worker and policy analyst, a specialist on matters relating to the promotion, prescribing, regulation and consumption of prescription medicines, with a particular interest in those prescribed for depression, anxiety and related psychic distress. See Curriculum Vitae and list of publications attached.

This case concerns a medicinal drug (paroxetine) mainly used to treat depression, sold in the UK under the trade name Seroxat® and in the US as Paxil®. Paroxetine was first developed by a Danish firm, but has been marketed worldwide by SmithKline Beecham (SKB) since about 1992. SKB became part of GlaxoSmithKline (GSK), also headquartered in the UK, in late 2000, shortly after this complaint was made.

The complaint relates to remarks made about paroxetine – implicitly claiming that it was relatively very safe - made by and in the name of SKB, on a US network TV documentary programme, broadcast in August 2000. One month after this broadcast, complainant wrote to the SKB representative involved (Dr David Wheadon, then based in Philadelphia) both to verify and query the sense of what had been said about the nature and risk of withdrawal symptoms from paroxetine.

The complainant pointed to a body of evidence suggesting that both the incidence and severity of withdrawal problems might be much greater than SKB had suggested in that broadcast. The complainant sent the company a letter summarising six specific concerns about the broadcast statements on paroxetine. He sent with it a monograph (Medawar, 1997) discussing the risk and implications of withdrawal and dependence problems with paroxetine. He also sent a book (Medawar, 1992) in which he described how, over the past 150 years, the belated discovery of withdrawal and dependence problems with a wide of drugs prescribed for psychic distress, most recently the benzodiazepine tranquillisers, had caused many people much distress.

Mr Medawar also drew Dr Wheadon’s attention to the much-visited Social Audit website (1998-2001), dedicated to enquiry into the nature of such risks and problems and the underlying reasons for them. It would have been known by both SKB and GSK that the complainant’s assessment of the evidence on this had prompted formal enquiries by, among others, the main UK and European regulatory bodies. These are the Committee on Safety of Medicines (CSM) and Medicines Control Agency (MCA) in the UK, and the European Medicines Evaluation Agency (EMEA) and Committee on Proprietary Medicinal Products (CPMP) in the EU.  

The complainant received no reply either to his original letter or to a follow-up sent in October 2000. Later, he enquired about the possibility of making a complaint about the Company’s refusal to respond, writing in December 2000 to the International Federation of Manufacturers Associations (IFPMA) about the scope of their Code of Marketing Practice. The complainant asked to be advised if the provisions of the Code would cover the complaint because "if so, I would want to consider making one".

After repeated requests for information about the status of this enquiry and complaint, the IFPMA notified Mr Medawar, in July 2001, that it was investigating the matter. They apologised for the delay. They did not advise that they would be referring the case to the Association of the British Pharmaceutical Industry (ABPI), of which both SKB and GSK were/are prominent members, nor that the preliminary adjudication on the case would be made by the Prescription Medicines Code of Practice Authority (COPA), a body closely linked to the ABPI. The IFPMA did not inform the complainant that they had advised both the Authority and GlaxoSmithKline that Sections 1.3 and 1.7 were "the relevant sections of the Code", under which the complaint should be considered. The complainant’s appeal is based in part on being denied the opportunity to frame a complaint in the most effective way.

The complaint was heard by a Panel comprising the three senior executive officers of COPA, a pharmacologist, solicitor and pharmacist. This Panel found that a complaint under I.3 of the IFPMA Code was inadmissible, and rejected the complaint covered by section I.7, The complainant was told he could pursue the matter by referring to the COPA Appeal Board, half of whose membership is independent of the pharmaceutical industry.

The substantive issues in this case arise in connection with broadcast remarks that were only a small excerpt of what Dr Wheadon said in a long recorded interview with Dr Snyderman of ABC-TV, and complainant acknowledges that SKB did not have editorial control over the programme. However, GSK has since confirmed that the company stands by the meaning of the following remarks (ABC-TV, 2000) - saying they are "not inaccurate": 

2. Re: the spirit of the IFPMA and related voluntary codes
The introduction to the IFPMA Code underlines that a key objective of the Federation (Statutes, Article 3) is "to promote and support continuous development throughout the pharmaceutical industry of ethical principles and practices voluntarily agreed on". This Introduction also states that: "in all its activities the pharmaceutical industry believes that high standards should be defined and respected and is convinced that, so far as marketing activities are concerned, self discipline is the process which best serves the public interest".

At the same time, the IFPMA advises that its code "is intended to provide an operational Code to be used in countries other than those in which a more demanding national code operates". Though it is the driving force in the IFPMA, the US Pharmaceutical Research and Manufacturers Association (PhRMA) has never developed a voluntary code of marketing practice, though it has been a steady and outspoken critic of excessive statutory regulation, notably by the US Food & Drug Administration (FDA). The USA is the world’s largest market for pharmaceutical products, but is believed be the only major industrialised country that has not developed a voluntary code, but only "a relatively brief statement of principles, based closely on the IFPMA Code" (IFPMA, 1997).

Complainant would wish to consider their possible relevance in this case, but has been unable to trace any reference to these principles through the PhRMA website, and no reference to them has been made by SKB, GSK or COPA in connection with this complaint. Nor has reference been made to extensive industry and company participation in voluntary standard-setting activities organised through bodies such as the World Health Organisation (WHO), CIOMS (Council for International Organizations of Medical Sciences) or the International Chamber of Commerce (ICC), nor to company-operated guidelines or procedures, intended to safeguard the quality of product information provided to health professionals and/or consumers.

Company and industry involvement in such activities underlines the relevance in this case of standards of behaviour voluntarily adopted by pharmaceutical companies in their operations in all major developed markets – including the commitments to interpret their codes of practice "in the spirit, as well as in the letter". This is consistent with the commitment to promote ethical behaviour – and especially to avoid marketing related practices that may be "such as to bring discredit upon, or reduce confidence in, the pharmaceutical industry".

Neither of these statements is explicit in the IFPMA Code, but both are clearly implicit and underlined, for example, by reference not only to the value of voluntary codes operating at national level but also to the importance of a related initiative by WHO (1988): "The Federation recognises the value of the definition by the World Health Organisation in 1998 of Ethical criteria for medicinal drug promotion. The present version of the IFPMA Code is consonant with these criteria, to the extent that they are applicable to our constituents". The following WHO criteria seem especially relevant and applicable in this case:

Article 6: "In this context ‘promotion’ refers to all informational and persuasive activities by manufacturers and distributors, the effect of which is to induce the prescription, supply, purchase and/or use of medicinal drugs"

Article 7: "…. All promotion-making claims concerning medicinal drugs should be reliable, accurate, truthful, informative, balanced, up-to-date, capable of substantiation and in good taste. They should not contain misleading or unverifiable statements or omissions likely to induce medically unjustifiable drug use or to give rise to undue risks …."

Article 8: "Scientific data in the public domain should be made available to prescribers and any other person entitled to receive it, on request, as appropriate to their requirements…".

Similar expressions of spirit are to be found in parts of the IFPMA Code, also in national voluntary codes, including those operating in Canada (where this broadcast would have been seen, and whence similar complaints might have arisen) and in Australia, New Zealand, India and the UK. For example:

In seeking to define the spirit that should always be reflected in the interpretation of the IFPMA Code, the complainant refers also to the "obligations of the industry", as defined in the preamble to the Suggested Code of Marketing Practices that was developed into the Code in use today: "The pharmaceutical industry, conscious of its special position, arising from its involvement in public health, and justifiably eager to fulfill its obligations in a free and fully responsible manner undertakes: … base the claims for substances on valid scientific evidence, thus determining their therapeutic indications and conditions of use; to provide scientific information with objectivity and good taste with scrupulous regard for truth, and with clear statements with respect to indications, contraindications, tolerance and toxicity; to use complete candour in dealings with public health officials, health care professionals and the public." (HAI, 1981)

Complainant contends that adherence to the spirit of a code is critical to the effectiveness of self-regulatory systems in general and to the IFPMA Code in particular. In this connection, the Appeal Board is asked to bear in mind also that IFPMA member organisations and pharmaceutical companies in Europe and elsewhere are now pressing hard to be allowed to promote prescription-only medicines to the general public, notably in TV commercials and advertisements in popular magazines. Fundamental to the industry’s case is the assertion that the public can trust companies to do the right thing.

Complainant contends that, in this case, interpretation of the code to the letter or spirit marks the difference between earning and deserving public and professional trust - or bringing discredit upon and reducing confidence in the pharmaceutical industry.

3. Relevance of section I.3
Responding to COPA (14 August, 2001), the company argued that neither I.3 nor I.7 of the IFPMA Code covered this complaint "as the sections are being taken out of context from the intended scope and purpose of the code, which is to cover marketing and promotion activities and not the invited participation in a television news magazine, to respond to charges against a product". The Panel duly found "that Section 1.3 could not apply because no promotional material was involved and no breach was ruled."

Complainant contends that this is an unacceptably and unduly restrictive interpretation of section I.3. He relies on the evidence given above relating to the need to uphold the spirit of the Code, also to the Explanatory Notes I.3 in the IFPMA Code. These make no reference to the need to substantiate only if "promotional material" is involved:

"Companies should deal objectively with requests for information made in good faith and should provide data which are appropriate to the source of the enquiry."

"Companies should produce data to support claims whether or not the data are in the public domain …"

The Panel appears to have to elicited no comment from GSK, either on its reasons for not responding to complainant’s enquiries, or on the significance of the evidence offered that conflicted with the company’s data. Complainant contends that this was procedurally incorrect, that it effectively prejudged the finding that the complaint was inadmissible under section I.3, and that this was prejudicial to complainant’s case.

Complainant also asks the Appeal Board to consider this ruling illogical. On the one hand, the Panel found that the broadcast remarks made by and in the name of SKB was "information" covered by the Code; on the other hand the Panel’s ruling exempts a company from any obligation to explain a refusal to substantiate information or to explain its position, even when there is some good evidence to suggest that its position is scientifically and otherwise untenable. By offering no comment on SKB’s refusal to respond, the Panel has in effect ruled that medical qualified, senior company figures are under no obligation to observe the minimum standards that would apply, say, to any product advertisement. If a sales representative in conversation with a doctor had made the same claims that SKB’s Dr Wheadon made publicly on nationwide TV, the company would have been required to substantiate the information given by the representative, on request.

Complainant will argue that, if the spirit of the IFPMA Code has any real meaning, the obligation to substantiate on reasonable request is implicit in the requirement always to provide information that is accurate, fair and not misleading – above all when communicating with patients and the general public, and especially when communicating through the mass media.

4. Quality of information provided
Further grounds for appeal arise in relation to the Company’s obligations to provide information of an appropriate quality. The Panel found that section I.7 of the Code was applicable – i.e. that in "Communications to the Public", including broadcast claims relating to the safety of a particular medicine, all information "should be accurate, fair and not misleading", also that companies "should adhere to the highest standards of accuracy".

Whether or not SKB and/or GSK should have responded to the original enquiry, complainant will argue that it was or should have been clear to the Panel that there were strong prima facie grounds for believing that the SKB assessment and characterisation of the withdrawal/dependence problem conflicted substantially with other evidence, and that significant and unprecedented numbers of patients were reporting serious problems when they tried to stop taking paroxetine. In numerous other adjudications, the Panel has considered the merits of evidence on substantive issues presented by both sides, but this was not done in this case.

It is argued that it was open to, and incumbent on, the Panel to invite GSK to outline its reasons for thinking that the broadcast statements presented a fair and balanced view, in spite of this evidence that the risks might be much greater than the Company supposed. The Panel should have asked the Company to offer some comment on the evidence given by the complainant before arriving at any conclusion about the reasonableness of the company’s response.

In the event, the company was asked to provide "appropriate data and information" to justify the broadcast statements made, but without taking any account of the conflicting evidence, identified as points 1 to 6 in the Panel’s summary of the complaint. On this basis, the Panel ruled that this justification was "not unreasonable", and that there had been no breach of the Code. Complainant will argue that the Panel’s ruling was unwarranted by the evidence available.

The company relied on data from two main sources, in support of claims relating to both the incidence and severity of withdrawal problems with paroxetine, and to the effectiveness of dose tapering to attenuate withdrawal distress. The company mainly relied on evidence from its clinical trials database and from spontaneous reports from clinicians. Complainant contends that the company’s reliance on its clinical trial database, to support its view that withdrawal reactions happen "very rarely", was and is improper and unwarranted.

To the best of the complainant’s knowledge, none of the trials on this database was specifically intended to measure the nature, incidence or severity of withdrawal. If he is mistaken on this point, it is solely because SKB/GSK failed to respond to his enquiries in September 2000: "Please could you also let me know specifically whether the Company has itself investigated (or commissioned) studies which elucidate the nature and extent of withdrawal problems?" If such studies/trials have in fact been done, the company is requested to provide details of these, and relevant findings, for consideration at the Appeal hearing.

Complainant assumes in the meantime either that such trials have not been done, or that they constitute only a minuscule fraction of trials of the database. He believes it was unreasonable to base any estimate of the incidence of withdrawal distress on the findings of trials that were neither intended nor designed to investigate this problem. To propose any estimate as precise as 7/8143 is to invite misunderstanding: it wrongly implies that the dimensions of the problem can be readily established and are well understood.

The evidence from other sources has some significant limitations, but investigators have consistently reported an incidence of withdrawal problems vastly greater than the 2/1,000 (0.2%) claimed by SKB. Typical figures are 3/6 cases (50%) reported by Barr et al (1994); 5/13 - 38.5% (Keuthen et al, 1994); 10/50 - 20% (Coupland et al 1996); and 5/12 - 41.6% (Bhaumik & Wildgust, 1996). One recent review concluded as follows:

In addition to the aforementioned study by Rosenbaum at el (1998), complainant refers to the study reported by Oehrberg and associates (1995) since the correspondent in the published paper is identified as Dr R Judge from SmithKline Beecham Pharmaceuticals, Harlow. The published paper does not indicate details of sponsorship, but it appears that another company was involved – Novo Nordisk AB, parent of the company that initially developed and patented paroxetine, before selling marketing rights to SKB. The investigators reported:

" … only 19 patients out of 55 (34.5%) who had received paroxetine reported any adverse event on discontinuation, as compared with seven out of 52 (13.5%) on placebo. Most patients reported just one adverse event, most being rated of mild or moderate severity"

The Appeal Board will see that the number of patients experiencing withdrawal reactions in this one trial is over twice the number reporting such effects among the 8,143 patients on the whole SKB clinical trials database. The point is not only that this trial must have been excluded from the Company database, but also that it clearly signals an incidence of withdrawal reactions far in excess of the 2/1,000 (0.2%) rate the Company says is typical.  

Complainant accepts that gradual reduction of dosage may attenuate withdrawal problems, but clearly it does not abolish them. Gradual tapering of dosages had been employed in the three cases reported by Barr et al; in four of the five cases reported by Keuthen et al; and "the majority of cases occurred despite slowly tapered withdrawal" in the series reported by Coupland et al. See also CADRMP, 1998; DTB, 1999, below).

Labelling changes approved by the FDA on 28 September 2000 clearly indicate that the company had come to accept that withdrawal symptoms might occur even after dose tapering:

Complainant argues that the suggestion offered by/on behalf of Dr Wheadon in the ABC-TV programme, that withdrawal problems are very rare if the drug is "appropriately discontinued" is essentially meaningless, and emphasises that the Company offers no sufficient advice on the need for dose tapering in product literature supplied to doctors and patients in the USA.

In addition, the qualification "if appropriately discontinued" takes insufficient account of actual clinical practice and apparently extensive lack of professional understanding, even of the existence of withdrawal reactions (Young & Currie, 1997), and it fails to acknowledge that discontinuation regimens may last for months, if not indefinitely.

Neither has the company provided any good evidence to support its contention that appropriate dose tapering on discontinuation is either practised or effective in assuaging virtually all withdrawal reactions, and complainant believes it does not exist: "as yet there is no controlled data to recommend its effectiveness, the length of time over which it should occur or the minimum dose that one should taper to" (Haddad, 2001). Complainant requests the company to produce such relevant evidence as it has.

Complainant contends also that the design of many trials on the SKB database (number unknown, but believed to be the large majority) would positively obscure evidence of the nature, incidence and severity of withdrawal. Reference will be made to one seminal study, conducted jointly by an employee of SKB and a then member of the Committee on Safety of Medicines, also to the analyses of that study published by the complainant. (Montgomery & Dunbar, 1993; Medawar, 1997 [pp 115-116], 1998a).

The Appeal Board is also asked to take into account that most of the patients referred to in the SKB letter of 14 September 2001 were involved in trials carried out some years ago. The exact numbers/dates are not known, but it is clear that 4,126/8143 patients were involved in phase 2 and 3 studies carried out in the 1980s. This cannot be considered "an up-to-date evaluation of evidence" within the meaning of I.3.

Clearly, the EMEA/CPMP are referring here not to any clinical trial, but to trials designed specifically to elucidate the withdrawal problem – and, as stated above, complainant knows of none that were carried out by SKB. However, it has been reported that SKB did carry out such studies on healthy volunteers, in the 1980s: "On average about half the volunteers taking part in a group of studies specifically designed to detect withdrawal problems suffered symptoms which suggest they had become physically dependent on the drug" (Boseley, 2001).

The source of this information was Dr David Healy, who had personally examined this documentation in discovery relating to a US court case. Healy (2001) reported his concerns to the Medicines Control Agency, indicating that the results of these studies showed "withdrawal syndromes occurred at a much higher rate than occur on benzodiazepines". Dr Healy has prepared for the Appeal Board a further statement relating to these studies on healthy volunteers and on their relevance in this case. See Annex 2.

In the same Guardian article (Boseley, 2001), Dr Wheadon reportedly referred to one of the major problems involved in distinguishing between withdrawal symptoms and relapse, in clinical practice:

"On withdrawal problems, Dr Wheadon said: ‘This is a very rare occurrence based on the data available. It is extraordinarily difficult to ferret out if it is a withdrawal effect or resurgence of the disease being treated’ … If somebody who stopped Paxil suddenly could not sleep, he said, ‘I challenge anyone to be able to tell me whether that is disease or discontinuation’."

If it were indeed so hard to distinguish between withdrawal symptoms and relapse, it is clearly unreasonable then to postulate that withdrawal symptoms are very rare. However, the problems involved in distinguishing between relapse and withdrawal would not have arisen in the above-mentioned healthy volunteer studies – since there is no underling pathology with healthy volunteers, and can therefore be no question of relapse.  

Moreover, the Company makes no reference to the existence of such diagnostic problems in advice and warnings issued to doctors or patients. This omission would tend to perpetuate ignorance of withdrawal problems, promote unnecessary drug treatment and expenditure, and cause unwarranted distress – clearly evident in many patient reports. There clearly is a significant risk that antidepressant withdrawal symptoms will be confused with relapse, and occasional references to this problem go back 40 years (See statement by Kramer, Klein & Fink (1961) cited in Medawar, 1998a), but there are also possibilities for distinguishing between the two.

For these reasons, complainant contends that Dr Wheadon’s broadcast assertion that the withdrawal problem "happens very rarely" represents a substantial and serious failure to provide patients with information that is "accurate, fair and not misleading".

Complainant accepts that the CPMP recommended that the "Summary of Product Characteristics" for all SSRIs should state "that the majority of withdrawal reactions are mild and self limiting" but asserts that, as an estimate of severity this is largely speculative and hugely imprecise. It also inevitably understates the problem, since withdrawal symptoms are frequently aborted by re-starting the discontinued drug or another SSRI.

Complainant refers to his reasons for thinking that the CPMP speaks with more authority than wisdom on this matter (Medawar, 1998b, 2000). He also emphasises that the CPMP assessment cannot be reconciled with the evidence relating to the severity of reported reactions from Price and colleagues at the Medicines Control Agency/Committee on Safety of Medicines (1996). They sent follow-up questionnaires to 217 doctors who had reported withdrawal reactions with paroxetine on Yellow Cards, to the MCA/CSM. They reported 21% of reactions were mild; 57.7% were moderate; and 21% were severe. Thus it would also seem true to say: "79% of respondents rated the reactions that they had reported as moderate or severe" (Haddad, 2001).

"Many discontinuation symptoms are mild and short-lasting, and manageable with reassurance and explanation. In some patients however, the symptoms are severe and make normal functioning impossible, even with slow withdrawal" (DTB, 1999)

Complainant requests the company to disclose what proportion of spontaneous reports of withdrawal reactions it has received have been classified as ‘serious’ (explaining also the definition used of that term). Complainant acknowledges that perhaps only a small minority of all patients experience serious reactions. However, even 5% of 70m is equivalent to 3.5 million "patient treatments", and it is suggested that it does users, prospective users and doctors a considerable disservice to deny or play down such risks. To characterise paroxetine withdrawal symptoms as mild and short-lived may conceivably have been ‘not inaccurate’ but was neither fair nor reasonable and was potentially very misleading.

Complainant accepts that the broadcast statements did not conflict with the FDA approved product information (drug label), but suggests this is no adequate excuse for misinforming the public and/or failing to provide sufficient information to procure effective clinical use. The company has offered no evidence to suggest that it was constrained by regulatory requirements from passing on information about withdrawal problems – including measures that might be used to reduce their incidence and/or severity - that reflected current and relevant clinical experience and which could be expected to promote more prudent and/or safer use of paroxetine. Complainant asserts that it was at all times open to the company to make proper and systematic investigations into problems associated with paroxetine withdrawal and to seek amendments to the labelling to reflect its findings – and that it was incumbent on the company to do promptly, not least in the light of its professed commitment to the principles of self-regulation.

The above outline the main grounds for this appeal. Complainant emphasises that there is much more evidence that he could and would wish to add – including numerous reports on the aforementioned website and other dedicated websites from anguished and angry patients. This further evidence has not been produced given constraints of time and the Board’s need to limit the duration of the Appeal. 

Annex One  
The following data were generated by the Uppsala Monitoring Centre (27 January 2001), which has operational responsibility for the WHO's Programme for International Drug Monitoring. The table identifies drugs on the Centre's database that have attracted most reports of withdrawal problems indicative of dependence.

Caution should be exercised in interpreting these data. They are not a reliable guide to the frequency of withdrawal reactions (EMEA/CPMP, 2000), and give only a crude indication even of the relative frequency of withdrawal reactions with different drugs – notably because they are not corrected for the volumes of each drug prescribed. For example, although paroxetine (Paxil/Seroxat) tops the list by a wide margin, venlafaxine (Effexor/Efexor) – which is ranked second – has been much less prescribed.

Nevertheless, the prominence of paroxetine in this list is consistent with reports from many other sources. For example, on the UK database of ADRs, "withdrawal reactions with paroxetine constitute a greater proportion of reports (5.1%) than with the other SSRIs (0.06-0.9%). They have been reported more often with paroxetine (0.3 reports per thousand prescriptions) than with sertraline and fluvoxamine (0.03%), and least often with fluoxetine (0.002%)." (Price et al, 1996)

In the Table, on following page, SSRI and related drugs appear in bold print and benzodiazepine tranquillisers are in italics

Statement by David Healy MD FRCPsych

I have been asked to confirm and comment on evidence relating to withdrawal reactions suggestive of physical dependence on paroxetine, observed in studies on healthy volunteers, carried out in the 1980s by SmithKline Beecham.

In the course of a recently settled legal action (Tobin v SmithKline, Wyoming, 2001), I acted as an expert witness for the Plaintiffs. As part of the discovery process I had sought and was granted access to SmithKline’s Beecham’s healthy volunteer archive at Harlow. My concern had been to scrutinise those records for details of possible agitation and suicidality in healthy volunteers taking paroxetine. These were present, but at least as striking was evidence from these studies about dependence on paroxetine. A detailed expert report was prepared for the plaintiffs’ lawyers in this case, which includes details of studies undertaken by SmithKline Beecham that fully substantiate concerns I communicated to the UK Medicines Control Agency in letters of 7^th and 19^th June 2001, the essence of which was also accurately reported in The Guardian newspaper (11 June 2001)

I regret that I am under a confidentiality order in regard to this material and am not able to disclose it to this appeal. However, I can confirm, and am prepared to testify to the substance of the points raised in the following exchange (in my testimony in Tobin v SmithKline) between Mr Charles Preuss, the attorney for SmithKline, and myself.

Healy: Yes, but there's a withdrawal syndrome from Paxil, including agitation, abnormal dreams and nightmares that comes through in spades in these healthy volunteer studies.

Preuss: You're saying Paxil is still active for three months?

Healy: In up to 80 percent of the volunteers on this drug for only two weeks produces withdrawal syndromes in these healthy volunteers. I'm saying in my clinical experience I've seen people on this drug for short periods of time and I've seen them have troubles three months later, yes.

My concerns about paroxetine extend far beyond the results of these studies on healthy volunteers. In the 1990s, after its release on to the market as an antidepressant, SmithKline Beecham put paroxetine into clinical trials – exemplified by the study reported by Montgomery & Dunbar, 1993 - that involved a randomised discontinuation design. The difficulties experienced by patients on randomisation to placebo were then interpreted by SmithKline Beecham as evidence of new illness episodes, and the company has subsequently responded to enquiries about the risk of withdrawal reactions and physical dependency, typically by stating that any such problems experienced by patients are simply a recrudescence of their original nervous problem. Basic pharmacological principles, epidemiological studies on depression, as well the evidence from their own healthy volunteer studies strongly suggest that such an interpretation of these data was and is quite unjustified.

Against this background SmithKline Beecham launched paroxetine in the UK with disclaimers on the datasheet to the effect that, as with any drug acting on the brain, some care needs to be taken on discontinuation. The data available to SmithKline before launch indicated problems occurring at a significantly greater rate and to a markedly more severe degree than any psychiatrist at the time would have had reason to expect either from an antidepressant or from such warnings.

Post-marketing surveillance surveys and other studies undertaken since have indicated much greater withdrawal problems with paroxetine than with the previous generation of tricyclic, MAOI and non-tricyclic or non-MAOI antidepressant drugs. A randomised controlled trial undertaken with funding provided by Eli Lilly (Rosenbaum et al, 1998), indicated rates of problems on discontinuation of paroxetine in over 30% of patients with many patients having multiple symptoms, including many novel and disturbing symptoms.

For Dr Wheadon and the company therefore to characterise paroxetine withdrawal reactions as very rare, transient, mild and/or virtually impossible to detect and distinguish from underlying psychiatric illness is simply an untenable position. It follows that I have real concerns about SmithKline promoting paroxetine for the prophylaxis of depressive disorders and other psychiatric illness, on the basis of data that are more sensibly and credibly explained in terms of physical dependence and withdrawal symptoms.

David Healy MD FRCPsych.

A.H. Young, A Currie, Physicians' knowledge of antidepressant withdrawal effects: a survey, J Clin Psychiatry, 1997, 58 (suppl 7), 28-30.