|Social Audit Ltd|
|P O Box 111 London NW1 8XG|
|Telephone/Fax 44 (0)171 586 7771|
|Mrs. Elizabeth Hopkins, Post Licensing Division|
|Medicines Control Agency|
|Market Towers, 1 Nine Elms Lane|
|London SW8 5NQ||
23 March 2000
Dear Mrs Hopkins,
I enclose a copy of a full-page advertisement for Seroxat (paroxetine, SmithKline Beecham), published in General Practitioner, 17 March 2000. If advertisements like this comply with the Advertising Regulations 3A(2), I clearly have a lot to learn about the MCA's interpretation of regulations and the adequacy of advertising control.
This is the fourth occasion in the last few months on which I have referred an advertisement to the MCA when the same important point of principle seems to arise. To what extent is it permissible to make bold headline claims for none too gentle drugs, on the basis of rather tentative small-print findings? Is it reasonable (and likely to encourage rational prescribing) to make sweeping generalisations on the basis of relatively isolated observations, when these would be of little or no real benefit to the vast majority of people exposed to the drug?
The claim in question here is that paroxetine "Lifts mood quicker than fluoxetine". It is cited to two papers, whose authors both chose their words much more carefully than the headline claim in the advertisement might suggest:
"Conclusion - Paroxetine and fluoxetine showed similar efficacy, with a greater effect on associated anxiety symptoms and possibly an earlier onset of action for paroxetine. This early result should be confirmed in further studies " (De Wilde et al., 1993)
"Paroxetine appears to produce an earlier onset of anxiolytic effect than fluoxetine, but this result needs to be interpreted in the context of multiple comparisons and prior similar results in the published literature." (Chouinard et al, 1999)
The 1993 study involved only 37 paroxetine patients, plus 41 on fluoxetine, and 15 patients withdrew before the final assessment at six weeks. The authors reported that six patients withdrew because of adverse events, although "no serious adverse events occurred during the study". On the other hand, they reported that "twenty-three adverse events were rated severe in 9 paroxetine-treated patients [and] 25 adverse events were rated as severe in 11 fluoxetine-treated patients". The significance of the main claim in the advertisement seems not much clearer either. This is because:
The study involved patients suffering from depression, but the headline claim is based on an assessment from a sub-scale relating to anxiety symptoms.
There was no difference in efficacy for depression (Hamilton-21 scale) between the fluoxetine or paroxetine treated groups
No statistically significant differences between treatment groups were detected using the MADRS scale for depression at any of the assessments (weeks 1, 3, 4 and 6).
"The results of the HSCL, which is a patient self-rating scale, were not in accordance with the physician rating scale, showing an early response for fluoxetine at week 1, and a trend in favour of paroxetine from week 3 onwards".
The small but statistically significant reduction in anxiety symptoms with paroxetine-treated patients that was found at week 3, was not seen in the assessments at weeks 1, 4 or 6.
The statistically significant difference observed, using the CGI scale, was found at the week 4 assessment - not at weeks 1, 3 or 6.
In short, although "this study seems to show that the efficacy of paroxetine and fluoxetine is very similar", the advertisement emphasises exactly the opposite.
Data-dredging is the main problem with the second study too; in this case the paper points to two or three statistically significant differences among the 66 different efficacy measures reported. These suggested superiority of paroxetine over fluoxetine [a] on two measures at Week One, and [b] on another measure at Week Two. This is not the stuff headline claims should be based on. However, to their collective (if not unanimous) credit, the authors acknowledge this. See Abstract:
"Limitations: Differences observed between the two drugs in antianxiety effects were limited to two measures of anxiety among several others"
"Discussion: The data indicate that paroxetine and fluoxetine have comparable antidepressant and anxiolytic efficacy. Paroxetine appears to produce an earlier improvement in agitation and psychic symptoms compared with fluoxetine"
I should also like to draw to the attention of the MCA/CSM the really important findings in this study (Chouinard et al., 1999). This may not be one for you, but perhaps it will concern you enough to want to pass it on; if so, will you please let me know to whom? I am assuming that someone at the MCA/CSM will be looking into the evidence you have from Dr David Healy (about akathisia and the evidence linking it, in extreme cases, to suicidal and murderous behaviour)? If not, perhaps this is one for Dr. Jones.
The real meat in this study is barely alluded to in the Abstract. It appears in section (3.3) on Emergent Anxiety. Here the authors describe what they found when looking very closely for evidence of symptoms of anxiety that showed up only after starting drug treatment - the implication being that the anxiety resulted from the drug treatment itself. It makes pretty alarming reading.
Of the 100-odd patients on each drug, 40 on paroxetine and 33 on fluoxetine failed to complete the 12-week study, the main reasons being an adverse reaction or that the drug didn't work. Among them, were the (?) 11 patients on paroxetine and (?) 12 patients on fluoxetine found to have experienced emergent anxiety - both somatic anxiety (e.g. akathisia) and psychic anxiety (mental agitation, distress).
"For somatic anxiety, 7% and 4.1% of paroxetine- and fluoxetine-treated patients, respectively, experienced emergent new somatic anxiety. Treatment differences were not statistically significant. Three patients in the paroxetine group reported emergent new psychic anxiety symptoms, but no fluoxetine-treated patients did".
Despite the emptiness of the Abstract, there is an unusually clear and detailed account of the 15 "clinically significant adverse events" in this study. Imagine how SmithKline Beecham (paroxetine) and Eli Lilly (fluoxetine) would react - and what the MCA might be up against - if they saw this kind of small print turned into headlines:
Four paroxetine- and two fluoxetine-treated patients were withdrawn for psychiatric reasons. In the paroxetine group, the psychiatric reasons for withdrawal were: severe anxiety and moderate agitation (n = l); insomnia and depression (n=1); suicidal tendencies with a definite suicidal plan (n = 1); and increased suicidal tendencies and deterioration of depression (n = 1). In the fluoxetine group, the psychiatric reason for withdrawal was for suicidal tendencies (n = 2).
In addition, six patients experienced significantly disabling or incapacitating adverse effects. In the paroxetine group (n = 3), the disabling/incapacitating effects were: depersonalization, constipation, impaired concentration, dizziness and abnormal vision (n = 1); pharyngitis (n = 1); and suspected hypomania (n = 1). Two of these patients (depersonalisation, suspected hypomania) were withdrawn. In the fluoxetine group (n = 3) the disabling/incapacitating effects were: rash/pruritus with tachycardia (n = 1); multiple nosebleeds (n = 1); and elevated TSH and detection of microsomal antibodies (n = 1). Two of these patients (rash/pruritus, multiple nosebleeds) were withdrawn from the study.
I can't help thinking that the CSM/MCA would be taking leave of its senses if it did not insist on the need for cautious prescribing of such drugs. To permit advertisements like this is playing with fire.
CLICK HERE TO READ ON