Basic HTML Version
310
ALASTAIR DOBBIN ET AL.
significantly more effective than antidepressant treatment in the pref-
erence arms. The fact that groups were not randomized but self-
selected was addressed to some degree by adjusting for baseline fac-
tors in the analysis using multiple regression. The significance of the
treatment effect remained, showing that the result from the
t
test was
at least not explained by imbalance at baseline; although this does not
rule out some other form of bias. The comparison of the two random-
ized groups could not be attempted, because there were only four
cases across the groups.
The effects of hypnosis and medication were compared for each SF-36
scale adjusting for the baseline level, and significant differences were
found for general health (
p
= .012, 95% CI for effect size 5.1 to 37.9) and
vitality (
p
= .003, CI 10.3 to 46.6).
There were no adverse events with either treatment, and no admis-
sions to hospital.
D
ISCUSSION
We found that significantly more patients, 93%, (
p
< .001) preferred
self-help, self-hypnosis to medication for the treatment of depression.
In a systematic review of preferences of depressed patients offered
either psychotherapy/counseling or medication, there was always a
preference for psychotherapy/counseling over pharmacotherapy;
such preferences range from 51% to 66%, differences (percent prefer-
ring psychotherapy/counseling
minus
percent preferring antidepres-
sants) ranged from 6% to 38% (van Schaik et al., 2003). In the current
study, 93% preferred self-hypnosis and the difference was 87%. Bedi
et al. (2000), with similar recruitment procedures (PRP design),
showed a 64% preference for a face-to-face psychological intervention
(counseling); our higher figure (93%) may reflect the immediate self-
help nature of our intervention, which may be perceived by partici-
pants as more convenient, less threatening, and less stigmatizing.
Previous studies have shown that most patients do not want referrals
to a mental health specialist (McKeon & Carrick, 1991; Priest, Vize,
Roberts, & Tylee, 1996). There was also good patient recruitment and
compliance, which may have been because of the GP-care base of the
treatment. These results support the increasingly recognized view that
the PRP-study design is necessary to evaluate the first stages of com-
plex intervention treatment in primary care where the RCT design
makes patients unwilling to participate (Bedi et al., 2000; Dwight-
Johnston et al., 2001; Ward et al., 2000). The results from PRP studies
provide data on treatments in a naturalistic setting, where often, as this
study shows, patients will exercise a choice. Benchmarking this study’s
results with similar studies (Table 1) showed a strong degree of
concordance across demographic and pre- and posttreatment data,
Downloaded by [ ] at 04:44 26 March 2012