|
Child depression
 Background
Traditionally, treatment has been psychological or psychoanalytical,
family therapy being most popular in the UK. However, these
are expensive treatments, some of unproven effectiveness.
Attention has therefore turned to antidepressant drugs, which
are established in trials as effective in adult major depression.
However, it is unsafe to extrapolate this evidence to young
people, since these trials excluded children, in whom the
condition certainly has a different epidemiology and may have
a different aetiology.
Randomised controlled trials are needed to decide whether
antidepressants work in child depression, since placebo responses
are common and the condition frequently gets better naturally,
making other types of evidence impossible to interpret. A
number of small RCTs of tricyclic antidepressants of variable
methodological quality has been conducted, generally reporting
non-significant trends in favour of treatment. Hazell et al
[1] recently conducted a systematic review and meta-analysis
of these trials.
Systematic review of RCTs
A comprehensive search strategy was used to identify trials
from electronic databases, and from manual searches of English
and non-English abstracts, bibliographies, and conference
proceedings.
Twelve trials were identified, all but one reporting small,
non-significant treatment benefits. Nine of the twelve studies
identified could be used in the meta-analysis; exclusion criteria
are described, but it is unclear why the authors excluded
trials where data for children and adolescents could not be
separated, since they do not present separate results for
the two groups.
Results
There was one duplicate publication. Five trials presented
their results as numbers improved (30/78 improved on treatment,
36/97 on placebo: pooled odds ratio 1.08 {95% CI 0.53 - 2.17}).
Six reported effect sizes, i.e. changes in the mean scores
of the groups. Such continuous outcomes are more difficult
to combine statistically than the above dichotomous outcomes
, especially if standard deviations are not given - the authors
had to assign this to some studies. Again, a trend towards
improvement on tricyclics was seen, (pooled effect size =
0.35 {-0.16 - 0.86}), a positive value representing improvement:
since the 95% confidence interval includes 0 no improvement),
the result is not statistically significant.
The quality of the trials was investigated using published
guidelines: the more rigorously conducted studies showed smaller
treatment effects.
There are probably some unknown "negative" studies
due to publication bias, which thus serves to strengthen the
authors' conclusions.
Conclusions
- this high quality systematic review identified twelve
trials of tricyclics, all but one weakly positive, and concludes
that tricyclics are no more effective than placebo.
- the conclusions are supported by the likely effects of
publication bias, and the less rigorous conduct of the more
positive studies. over 50% responded to placebo
- a traditional narrative review might well have concluded
tricyclics were effective, on a "small samples, no
smoke without fire" principle.
- further drug research should concentrate on other agents,
such as SSRIs
- this is a good starting point for any reader interested
in meta-analysis.
Practice points
- there is no evidence that tricyclics are more effective
than placebo for child depression and adolescents
- any possible benefits are probably outweighed by the risks
of toxicity
- these patients may well respond to non-drug strategies
Reference:
1 P Hazell, D O'Connell, D Heathcote, J Robertson, D Henry
. Efficacy of tricyclic drugs in treating child and adolescent
depression: a meta-analysis. British Medical Journal 1995
310: 897-901.
Rachel Churchill
Epidemiology Research Fellow
Institute of Psychiatry
London
David Gill
MRC / NHS R&D Health Services Research Fellow
Institute of Health Sciences
Oxford.
|
