Problem solving treatment and group psychoeducation for depression: multicentre randomised controlled trial
BMJ 2000; 321 doi: https://doi.org/10.1136/bmj.321.7274.1450 (Published 09 December 2000) Cite this as: BMJ 2000;321:1450
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Having read the article "Problem solving treatment and group psycho-
education for depression: multicentre randomised controlled trial" and its
subsequent rapid responses I think we must wonder indeed who it is that is
ready for the funny farm. Life, with all its ups and downs, is not a
medical condition. I think anyone who thinks it is might be in need of
treatment themself.
Wai-Ching Leung's calculations in his rapid response were, I think,
bordering on the ridiculous, and that is being polite. My advice to him
would be: get into the real world and get a life!
Competing interests: No competing interests
I would like to congratulate Professor Dowrick and his team for
having completed a difficult study. While other letters have dealt with
concerns and potential problems with statistical "errors", my concerns lie
elsewhere.
The authors indicate that the main aims of the study were to establish the
acceptability and the effect on caseness of two psychological
interventions for depressed adults. A case is made for the effectiveness
of both of these approaches but that problem solving has the edge on both
counts.
I think it is difficult to be confident with either of these conclusions
for the following reasons.
1. Ratings do not appear to have been performed by independent,
blinded raters.
2. It would seem likely that any treatment that is offered in one's
own home(problem solving)will score more highly on acceptability,
particularly when compared with a treatment involving travel and enforced
social discourse(the group psycho-educational treatment. People exhibit an
understandable preference for the least effortful treatment options and
depressed people, in particular, are apprehensive about travelling away
from home, meeting others and are sensitive to potential rejection. This,
of course, is a problem for the psycho-educational format in general but
why didn’t the research design ensure parity between the treatment groups?
Problem solving is a treatment modality that transfers well to a psycho-
educational group format. In equalising the treatment modalities in this
way one could feel more confident about the veracity of the reported
differences in acceptability. Furthermore acceptability was assessed by
"comparing the proportion of participants who refused or failed to attend
an initial session, those who discontinued an intervention, and who
completed their assigned intervention". There are at least two problems
with this approach to measuring acceptability. One can imagine how
difficult it might be for people who are depressed to disengage from
therapy when the therapist visits them at home. The opportunity to "fail"
to attend for appointments is removed making the need for assertive
responses from the patient more crucial in opting out of therapeutic
contact. Depressed people are not noted for their ability to be assertive.
Patients disengage from therapy for a variety of reasons. In Dundee we
conducted a pilot of the 12-session version of the Coping with Depression
Course(CWD). Patients who failed to complete the course were sent a
questionnaire and asked to indicate what prompted non-attendance. Our
expectation was that disengagement would be related to finding the CWD
course unacceptable or too onerous in terms of time and effort. Our survey
indicated that this held true for fewer than 40% of those who disengaged.
The other 60% indicated a range of reasons not linked to acceptability of
the treatment modality in any way.
3. While there are reports in the literature about variants of the
CWD course the most widely used format in treating already depressed
adults has been the format which comprises twelve sessions over 10 weeks.
In the CWD course, Lewinsohn and his colleagues provide a clear and
cohesive treatment package readily accessible to mental health
professionals. It is not clear why the authors have chosen to use an
unpublished, and as far as I am aware, an unvalidated shortened version.
Furthermore the CWD course is designed to help people who are depressed
learn skills to become less depressed. The Munoz prevention of depression
version may well not be appropriate for individuals already depressed.
Which elements of the full version have been left out? It would be
important to know this in order to make some judgement about the
comparison with Problem Solving, a treatment modality that is well tried
and reasonably tested.
4. Measures of receipt of medication are not the same as measures of
class of antidepressant, optimal dose prescription, adherence, etc.. A
potential confounding factor may be that more people in one group received
more effective antidepressant medication than in the other. How can we be
sure that systematic bias is not at play here?
5. The inclusion criteria based on ICD 10 and DSM IV classifications
of depressive disorders included dysthymia, adjustment disorders and
bereavement reactions. This inclusive set of criteria may well be another
source of confounding factors within the design. People with dysthymia are
recognised as difficult to treat, people with adjustment reactions recover
once the crisis has passed and people with bereavement reactions recover
at different rates in relation to personality traits, the passing of time
and support from their families. Data on the distribution of these
diagnostic categories between the two groups would have been helpful. I am
sure the authors will say that random allocation will have ensured even
distribution of these factors between groups, but did it? The process of
minimisation in randomisation might have helped here.
Establishing the efficacy and field utility of psychological
treatments for depressive disorders in primary care or community settings
should be given high priority. Psycho-educational formats can be effective
and are one way of doing this. Having used the Problem Solving and Coping
with Depression formats with individuals meeting diagnostic criteria for
depressive disorders and witnessed effectiveness and acceptability first
hand, I am pleased this team have taken the steps to test these treatment
approaches in the field using the template of a randomised controlled
trial methodology. Given the observations outlined above, the conclusions
reached by the authors may be overstated.
Mr. John Swan, Mr Bob MacVicar and Miss Eleanor Sorrell,
Researchers,
Department of Psychiatry, University of Dundee, Ninewells Hospital and
Medical School, Dundee DD1 9SY
Competing interests: No competing interests
Dear Editor
I am a bottom-line person. What does this study say to me? We know
that problem-solving helps (so does solution-based therapy)as does psycho-
education (although it sounds rather terrifying to me)- I thought we knew
this already.
We should not also forget the value (which may not achieve
statistical significance - but every little bit is worth having?)that
service users place on being more involved in their care and having the
opportunity to express their preference of treatment (even assuming that
it is available). Depression Alliance, like other charities, operates a
nationwide service of self-help groups and is developing its self-
management programme to this end.
There is no reason why everyone even has to be funnelled through the
statutory sector, although it is equally important that the voluntary
sector is adequately funded, its people sufficiently trained and resourced
and not used as a 'dumping ground'. Many of these techniques are easily
taught and learned and can be provided in a wide range of environments
with a wide range of technologies.
We must be careful not to lose the 'human touch' in all this whilst
enabling people to become less dependent on the medical system by skilling
up our communities again. After all - how long before we have the evidence
-base for the therapeutic effect of an unconditional hug and do we really
need such evidence?
It is to be hoped that this research will influence the development
of psychological services in primary care, where the briefest element is
currently the intervention, and the longest the wait for it.
Dr Chris Manning
Co-Chair PriMHE
Vice-chair Depression Alliance
www.primhe.org
Competing interests: No competing interests
Further to Dr Gale’s rapid response (1), my calculations for the
confidence intervals for the data on the total for problem solving at 6
months are different:-
For ARR:-
SE = SQRT (p1*q1/n1 + p2*q2/n2) = SQRT (0.4255*0.5745/94 +
0.5918*0.4082/98) = 0.07117
Hence, 95% CI is 0.1663 +or- (1.96*0.07117)
i.e. 0.166 (95%CI 0.027 to 0.306)
For NNT:-
The 95%CI is therefore 6.01 (95%CI 3.3 to 37.3)
The 95% CI for relative risk is more complicated, and involves
natural logarithms and exponential. I obtained the following result:-
RR = 1.39 (95%CI 1.04 to 1.85)
All these measures show that the results relating to the problem-
solving total at 6 months are significant.
References
1. Gale G. Relative risk confidence intervals not significant. eBMJ
http://www.bmj.com/cgi/eletters/321/7274/1450#EL5 (published 13 Dec
2000)
2. Dowrick C, Dunn G, Ayuso-Mateos JL, Dalgard OS, Page H, Lehtinen V
et al. Problem solving treatment and group psychoeducation for depression:
multicentre randomised controlled trial BMJ 2000; 321: 1450
Competing interests: No competing interests
We thank Professor Dmitriou for pointing out an inconsistency in our
Table 5. He is correct in inferring that we have inadvertantly published
relative risks rather than the intended odds-ratios. We apologise for
this simple error. The figures in Table 6, however, are correctly
identified as odds-ratios.
Dr Leung suggests that we should have rounded up our NTT (for example
6.01 should have been reported as 7 rather than 6). We think that this
would have been a bit perverse. These NNT figures are only a rough and
ready guide to interpretation, for the lay reader.
Dr Gale's calculations and inferences would appear to be based on the
simple figures we present in Table 5. We would refer him, and other
readers, to the detailed text on our statistical methods, and in
particular to Table 6, which presents our logistic regression estimates of
treatment effects.
Competing interests: No competing interests
I have read the paper and responses with interest. By my calculations
the relative risk is 1.39 (95% CI 0.68- 2.82). The ARR is 0.166 (0.113-
0.218) and thus NNT of 6.0 (4.5-8.8). If this is correct, then I suggest
that there is no statistically significant treament effect.
Competing interests: No competing interests
I agree with Professor Dimitriou (1) that several errors exist in
table 5 of the full version of the paper by Dowrick et al (2) comparing
problem solving treatment and group psychoeducation for depression.
The main source of errors is mistaking relative risks for odds
ratios. The quoted “odds ratio” at the bottom of the table were in fact
relative risks. The second source of errors is the failure to round the
calculated NNT up (rather than down) to the next whole number.
For example, the results for the total for problem solving at 6
months should be calculated as follows
Odds ratio = (58/40) / (40/54) = 1.96
Relative risks (RR) = (58/98) / (40/94) = 1.39
Absolute risk reduction (ARR) = (58/98) – (40/94) = 0.1663
NNT = 1/0.1663 = 6.01 = 7 (rounded up)
Similarly, the other results should be:-
Total for problem solving at 12 months:- odds ratio 1.04; RR 1.01;
NNT122
Total for depression prevention at 6 months:- odds ratio 1.76; RR
1.34; NNT 8
Total for depression prevention at 12 months:- odds ratio 0.76;
RR0.89; NNH 15
I hope the authors would clarify whether the results in Table 6 are
actually odds ratios or relative risks.
Reference
1. Dimitriou EC. Odds and NNTs confusion. eBMJ
http://www.bmj.com/cgi/eletters/321/7274/1450#EL1 (published 10 Dec 2000)
2. Dowrick C, Dunn G, Ayuso-Mateos JL, Dalgard OS, Page H, Lehtinen
V et al. Problem solving treatment and group psychoeducation for
depression: multicentre randomised controlled trial BMJ 2000; 321: 1450
Competing interests: No competing interests
Dear Sirs,
Thank you for giving us access to such an interesting article, but it
appears that there is some confusion with regard to Table 5 and the ORs
and NNTs of both Totals [problem solving and depression prevention].
Maybe I am wrong but if the ARRs are 17% [NNT=6] and 14% [NNT=7]
then the ORs must be 1,96 and 1,76 respectively. If the ORs are 1,39 and
1,31 then the AARs must be 7% [NNT=14] and 6,5% [NNT=15] respectively.
Sincerely yours
E.C.Dimitriou, M.D.,
Professor of Psychiatry
Competing interests: No competing interests
methodological issues
We thank John Swan and his colleagues for their detailed critique
(21.12.00) of our paper. We respond to each of their points in turn.
1. It is not possible in studies of this nature to conceal group
allocation from either participants or raters, as we noted our paragraph
on assignment. However our raters were independent of the intervention. In
addition, the deployment of self-rated instruments, such as the BDI and SF
-36, offers balance to any possible rater bias that might ensue.
2. The issue of acceptability is important. As we stated in our
discussion, we agree with Swan et al that differential accessibility may
well have been an important factor here. We also agree that
discontinuation - or disengagement - is not a major problem for the
depression prevention course. However we do not agree that problem solving
should therefore have only been offered to subjects away from their own
homes. We offered each intervention in the way that was most accessible
for subjects. For a group intervention, this inevitably means providing an
alternative site. We see this as an important pragmatic difference between
individual- and group- based psychosocial interventions.
3. The psycho-education intervention has many different variants, and
the 8 session version has been tried and tested previously. We would refer
Swan et al, and other interested readers,to our references 15 & 16,
particularly the meta-analysis by Cuijpers. Our version was modified
slightly, to increase the amount of contact time and the content devoted
to social support. It was approved by the authors, and training for our
facilitators was provided by the leading European exponent of the
intervention.
4. Few subjects receiving effective levels of antidepressant
medication (as defined acording to British National Formularly criteria,
and there were no differences in proportions between the three study arms.
5. We deliberately chose a wide definiton of depressive disorders, to
enhance the pragmatic validity of the trial. There were no differences in
the distribution of diagnostic categories between the three study arms.
The proportion of subjects with adjustment disorders was too small to have
an effect on the overall outcome of the study. We agree that dysthymia is
difficult to treat. By including this important sub-group, we may
therefore have understated - rather than overstated - the potential
benefit for these interventions, if delivered exclusively to subjects with
'mainstream' depressive episodes or disorders.
Competing interests: No competing interests