[Top] [All Lists]

[e-drug] Antidepressants for children banned in the UK (cont'd)

E-drug: Antidepressants for children banned in the UK (cont'd)

I agree with Leo that over confident criticism by others can damage
the reputation of organisations like mine: Healthy Skepticism that
strive to only criticise when there is good justification.

However I believe that whilst uncertain the case against
antidepressants for children (aged less than 19) is strong enough
to justify banning them unless someone can produce better
evidence showing that the benefits exceed the harms for at least
some children.

We should remember that there are usually wide confidence
intervals around the frequency of adverse effects because they are
usually less common than beneficial effects.

(For those who are not confident of their understanding of
confidence intervals I will try to provide a quick explanation. In a
randomised controlled trial the difference in the outcomes for the
treatment group and the control group are due to three main
factors: 1) the effect of the treatment, 2) random variations
between the groups and 3) bias. Confidence intervals give an
indication of how much of the difference may have been due to
random variation ie noise rather than signal. 95% confidence
intervals provide a prediction of the limits within which the
difference would fall in 95% of identical trials if the trial was
repeated many times.)

It is normal for humans to take less notice of confidence intervals
than we should and think that the results in a sample closely
represent the "real effect" in the population. Psychologists call this
representativeness bias.

I have not seen the data for suicide associated with
antidepressants for children. However the information available
suggests that the confidence intervals include no effect and a
harmful effect that is large enough to worry about.

This should be evaluated in the context of how effective the drugs

I am a member of a group that has looked closely at the trials of
antidepressants for children. The confidence intervals include no
effect and at best a small benefit only. The possible small benefit is
only seen with doctor's rating scales not patients' rating scales.
Consequently it is possible that those "benefits" arise from bias eg
unblinding because of awareness of adverse effects plus doctors'
hopes not shared by their patients. The widely held belief that there
is controlled trial evidence of efficacy has arisen from spin on a few
of many endpoints measured. The widely held belief in
effectiveness based on clinical experience can be explained by the
high recovery rates seen in the placebo control groups of the trials.
I don't know if the high recovery rates result from spontaneous
recovery or a placebo effect or a non-drug effect of seeing a doctor
or something else or a combination of such factors.

There is another possible harm that can not be detected by
placebo controlled trials. If people believe that the drugs are a
good treatment then there may be less support for the non-drug
treatments that are probably much more effective eg cognitive
behavioural therapy, supportive individual therapy and social skills

I conclude that it is possible but unlikely that antidepressants do
more good than harm for some children. It is more likely that they
do more harm than good. The uncertainty justifies a ban unless
someone can produce better evidence of a better harm benefit



Dr Peter R Mansfield
Research Fellow, Department of General Practice, University of
NHMRC Public Health Postgraduate Scholarship 250465
Director, Healthy Skepticism Inc
Improving health by reducing harm from misleading drug
Flinders University Convocation Medal 2003
34 Methodist St, Willunga SA 5172 Australia
ph/fax +61 8 8557 1040

To send a message to E-Drug, write to: e-drug@healthnet.org
To subscribe or unsubscribe, write to: majordomo@healthnet.org
in the body of the message type: subscribe e-drug OR unsubscribe e-drug
To contact a person, send a message to: e-drug-help@healthnet.org
Information and archives: http://www.essentialdrugs.org/edrug

<Prev in Thread] Current Thread [Next in Thread>