By: PETER
STURMEY
Clinicians must determine which treatment is best for which client. There are now a very large number of treatments available for all the common mental health disorders. Clinicians may well be perplexed as to which treatment to select for each particular client.
One approach to
solve this problem is to use psychiatric diagnosis to predict treatment. The terms used to describe both
pharmacological and psychological treatments often refer to diagnosis.
Psychotropic medications are called ‘anti-depressants’, ‘anti-psychotics’ and
‘anxiolytics’. Psychological treatments also often refer to diagnosis, for example
when we refer to treatment groups as ‘anxiety management groups’ or ‘support
groups for eating disorders’ and so on.
Treatment algorithms, randomized controlled trials (RCTs), reviews of the outcome
literature and reviews of evidence-based treatment, such as Cochrane reviews, National
Institute for Clinical Excellence guidelines and the APA guidelines (APA,
2005), all are organized around diagnostic categories. Many mental health
advocacy groups are also organized around
specific diagnostic groups. Thus, the notion that diagnosis predicts effective and
ineffective treatment is pervasive. This model suggests that diagnosis 1 predicts that
treatment A will be relatively effective for this diagnosis and treatment B will
be relatively ineffective for this diagnosis, and diagnosis 2 predicts that
treatment A will be relatively ineffective and treatment B will be relatively
effective for this diagnosis. Thus, we might recommend
anti-depressants for people with Major Depression, but not for a Psychotic Disorder.
Likewise, we would place people with anxiety disorders in an anxiety management
group, not in a support group for people with eating disorders. This model
is based on an interaction between diagnosis and treatment.
This model of
predicting treatment efficacy has many limitations. First, most outcome research using RCTs does not address the
question of diagnosis by treatment interaction. Rather, most RCTs merely
compare one treatment with some other procedure, such as a waiting list
control, or, more rarely, some placebo or perhaps a second treatment.
Researchers select participants to ensure that they all meet the same diagnosis. Thus, these kinds
of RCTs permit us to conclude
that treatment A may be effective for diagnosis.
They tell us
nothing about the effectiveness of this treatment for diagnosis and nothing about whether this treatment is the most effective treatment for this diagnosis. Wilson (1996) proposed a contrary
argument. He has noted that some standardized, manualized treatments for
eating disorders are highly effective. He suggested that treatment determined by
diagnosis might be highly desirable because the clinician can learn one
highly effective treatment procedure to a high
degree of proficiency. Further, there may be little room left for individualization of
treatment – which in any case might be unreliable and capricious – to improve
over this standard treatment. (Ghaderi [2006] presented some evidence to the
contrary.)
A second
limitation to this model of predicting which treatments might be effective is that response to treatment is highly
varied. RCTs emphasize statistical significance – changes that are unlikely
to be due to chance – and changes in the score of the average, but non-existent,
subject. Statistically significant results may emerge from many patterns of
response to treatment. For example, a statistically significant result might occur
if 50% of the treatment group have a large,
positive response to treatment, 25% have no response and 25% have a modest negative response to treatment, if the
experiment has a large enough number of
participants and if the dependent measures are sufficiently sensitive. A
statistically significant result may also emerge if most of the participants make a
modest improvement but one that has no practical significance for any
particular person. The average client does not exist: the clinician will never
treat this mythical person. The clinician treats specific clients. Even when there is a
strong evidence base for a particular treatment, it may be unclear at the
outset of treatment if the clinician is working with someone who will respond positively, not respond or respond negatively to this particular treatment.
A third
limitation is that clinicians frequently work with clients who have apparently already had standard, diagnosis-based
treatment and who did not respond to any meaningful degree. For example, it is
common for clinicians to work with people who have taken anti-depressants or
anxiolytic medication for many years and still have significant problems;
indeed their failure to respond to standard treatments is often the reason for
referral. Further, after standardized
psychological treatments, such as anger management, cognitive behaviour therapy for
depression and so on, a significant proportion of clients have residual problems, did
not respond or responded badly to standard treatment.
A fourth
limitation in diagnosis predicting the most effective treatment for each client is that many clients meet diagnostic
criteria for more than one diagnosis. When a clinician works with a client who
meets diagnostic criteria for Major Depression, Substance Abuse and Generalized
Anxiety Disorder, which of these three diagnoses predict the most effective treatment
for this client?
If all three predict effective treatments, in which order should the clinician implement
these treatments? Will effective treatment of the Major Depression result in a
generalized improvement in the client’s functioning, or will treatment of the
Generalized Anxiety Disorder result in the broadest spread of treatment effects?
The ability of
psychiatric diagnosis to predict the most effective treatment depends on the reliability and validity of that
diagnosis. The developers of the third edition (revised) of the Diagnostic and Statistical Manual (DSM) trumpeted its arrival as
a triumph of science (Kutchins and Kirk, 1997). The number of psychiatric diagnoses has
expanded considerably with each edition of DSM (Houts, 2002) and the developers of DSM did not
conduct reliability trials for almost all the hundreds of diagnoses in DSM-III-R
(Kutchins and Kirk, 1997). Where researchers did conduct diagnostic
trials, they were conducted after the diagnostic criteria had already been set; thus,
the results of reliability
trials did not inform
the development of the diagnostic criteria. Indeed, careful examination of the
reliability of DSM-III-R revealed that the reliability of the new diagnostic criteria
may not have been very much different from the reliability of the old
criteria (Kutchins and Kirk, 1997). In any case, this may be of limited relevance, since
the reliability of diagnosis by routine practitioners may have little to do
with the diagnostic practices of eager, well-trained government-funded researchers.
Some structured clinical interview procedures may result in quite high reliability.
However, most clinicians do not use these assessment methods routinely. In any case, the validity of these measures to differentially predict an effective treatment is
still little researched.
Clinicians may
often work with clients with rare, idiosyncratic, subclinical problems or other problems that do not meet diagnostic
criteria. In these situations too, psychiatric diagnosis may be of
limited use to predict treatment. Finally, some clinicians often feel that they have
something more to offer than skilled, but technocratic, application of
diagnostic algorithms and manualized treatment. Whether true or not, many
clinicians believe that their input into understanding the
case and designing treatment for each individual client has something to contribute to
treatment.
These limitations
to predict treatment based on diagnosis, if true, are serious. Consequently,
clinicians and professional training standards have argued that case formulation is a better way to guide
selection of the most effective treatment.
References
Peter Sturmey,
Clinical Case Formulation; Varieties of Approaches, 2009, John Wiley Ltd.
Read Also
Clinical ExpertiseEvidence-Based Practice in Psychology
Best Available Research Evidence
Definitions of case formulation
No comments:
Post a Comment