Rep36 Understanding Personality Disorder

3.2 Psychological interventions in health service settings 

---

36
3.2.2 Effectiveness of therapies for
personality disorder
Decisions about clinical care should be based on
available research evidence. The evidence used
to make decisions about clinical care is based on
a hierarchy, with the best level of evidence
coming from meta-analyses and systematic
reviews (using grouped data from randomised
controlled trials), followed by single randomised
controlled trials, cohort studies and cross
sectional surveys and case controls at lower levels
of evidence. Randomised controlled trials
(RCTs) are therefore considered the ‘gold
standard’ of evidence in medicine as a whole.
This type of study design is the most appropriate
to answer highly specific questions, usually of the
form ‘Is treatment X better than treatment Y for
patients with a specific disorder?’ Without a
randomised controlled design, conclusions
cannot be definitive about the effectiveness of
one treatment compared to another. 
Seligman (1995) has argued that the
properties that make an RCT so scientifically
rigorous make it the wrong method for
evaluating psychotherapies because RCTs do not
reflect what is done in routine clinical practice.
Criticisms of RCTs, particularly from practitioners
of psychological therapies, have centred round a
number of issues (Persons & Silberschatz, 1998;
Slade & Priebe, 2001). The main issue is that
RCTs involve grouping patients, typically by
diagnosis or by a specific problem, such as
deliberate self-harm. This assumes that all people
with the same diagnosis or problem are similar,
and that by dividing them at random into groups,
individual differences between people are taken
into account. In RCTs, the patients entered into
the study are often a highly selected group that
fit specific entry criteria and many patients may
be excluded, thus reducing the degree to which
the patients are representative of the group as a
whole that may be suitable for treatment.
Randomisation of patients in itself introduces
artificiality as it does not consider what the
patient would have chosen if offered a choice of
therapies. This issue is of importance if one
considers that patients are not passive recipients
of treatment and that their level of participation
in a clinical trial can vary from full participation
to dropping out altogether, introducing an
important source of bias in the results.
Psychological treatment research using
randomised controlled designs often utilises
treatment manuals that may accurately describe
the treatment patients receive but it can be
argued that therapists in naturalistic settings
behave differently and more flexibly, adapting
the therapy to the patient, rather than following
a rigid treatment protocol. Although individuals
may have the same disorder, one individual may
have a different set of problems and
psychological issues from another.