306 Chapter 36
significant given a large enough trial or meta-analysis. There are several
ways of measuring how effective a treatment is. The commonest one
is effect size, which expresses change in ‘standard deviation’ units. For
example, if untreated hyperkinetic children are an average of 2.5 standard
deviations above the population mean on a measure of hyperactivity, and
if treatment with stimulant medication brings them down to an average of
1.4 standard deviations above the population mean, the effect size is said
to be 1.1 (i.e. 2.5 minus 1.4). A great advantage of this measure is that it
allows direct comparisons of diverse treatment modalities, such as medica-
tion and psychological therapies, and of diverse outcomes, such as different
measures of depression or measures of symptoms and social impairment.
Successful psychological therapies typically have effect sizes of around
0.6 to 0.8 when administered in research settings. In ordinary clinic set-
tings, however, the average effect size of psychological therapy may be less
or even zero. What is this disparity due to? The exclusion of hard-to-treat
children and families from research trials is likely to explain part of the
difference. There are many other plausible explanations too, though only
some are supported by the empirical evidence (see Box 36.2). The overall
message for mental health professionals is both sobering and optimistic. It
is sobering because if psychological therapies as currently used in everyday
settings sometimes have little or no effect, then it seems difficult to
justify the cost involved; but optimistic because three changes in emphasis
could boost effectiveness in future. These changes are: a shift in emphasis
towards behavioural and cognitive approaches; the use of specific, focused
Box 36.2Why is psychological therapy much less effective in
routine clinical practice than in research trials (following Weisz,
2006)?Probably relevant
Clinics make less use of behavioural and cognitive approaches.
Clinics rely less on specific, focused therapy methods.
Clinics are less likely to structure therapy (for example, through treatment
manuals) or monitor therapy to ensure that the therapist adheres to the
treatment plans.
Clinics treat cases with high degrees of comorbidity and families who attend
irregularly – both criteria for exclusion from many research trials.Probably irrelevant
Research studies are more recent than clinic studies.
Some research trials use subjects who are recruited volunteers rather than
referred patients.
Clinic settings are less conducive to success.
Clinicians are less effective than research therapists.
Research therapists have had special training in the methods just prior to the
intervention.
Clinics have to provide for a range of children, and a range of problems.
Clinics are less likely to provide brief interventions.