Rapid Responses to:

PRIMARY CARE: Clair Chilvers, Michael Dewey, Katherine Fielding, Virginia Gretton, Paul Miller, Ben Palmer, David Weller, Richard Churchill, Idris Williams, Navjot Bedi, Conor Duggan, Alan Lee, and Glynn Harrison Antidepressant drugs and generic counselling for treatment of major depression in primary care: randomised trial with patient preference arms BMJ 2001; 322: 772 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Improving preference measurement in primary care studies.

Andrew Martyn Thornett   (31 March 2001)

Author’s conclusions were not justified by findings

Wai-Ching Leung   (1 April 2001)

Doubtful 12 month intervention effects

Christopher Dowrick   (2 April 2001)

Depression requires a multidimensional approach to managment

Steve Williams   (3 April 2001)

Two distinct anti-depression treatments used together

Salvador Vale   (5 April 2001)

further analysis

Arthur Rifkin   (13 April 2001)

The type of treatment matters less than ensuring it is done properly and followed up

David Simpson   (30 April 2001)

Does randomization introduce bias?

Toke S Barfod   (10 May 2001)

Improving preference measurement in primary care studies. 31 March 2001
Andrew Martyn Thornett, Clinical Research Fellow Department of Psychiatry, University of Southampton, SO14 0YG Send response to journal: Re: Improving preference measurement in primary care studies.	The study by Chilvers et al. is one of a small number to support a relationship between receipt of preferred treatment and improved outcome in the treatment of depression.1 The patients who chose counselling did better than those randomised to counselling, although the 95% confidence interval reached but did not cross zero. Many patients express a preference for psychological treatments compared with medication.2 However, being allowed to choose treatment has not been shown to improve short-term outcome in primary care depressed patients treated with either antidepressants or counselling,3 or non- directive counselling, cognitive-behaviour therapy or usual general practitioner care.4 This difficulty demonstrating preference effects may be the result of the methodology used. As in the current study, preference has been defined as refusal to be randomised within a trial. However, it is possible that many patients are prepared to allow themselves to be randomised but would still prefer not to receive the treatment to which they are allocated, diluting the beneficial effects inherent in the preference arm. An alternative method would be to randomize the entire population, and then allow patients to accept or decline the allocated treatment. In this procedure, the consent process would be split in two, an initial consent to take part in the study on the understanding that a treatment will be offered but does not have to be accepted, and a second stage in which the individual accepts or declines the treatment. Those who decline remain in the study but are treated as the GP believes is clinically appropriate. This gives three groups which can be compared: the whole cohort, those who accept and those who decline randomization, and allows the effects of preference to be described in more detail. 1. Chilvers, C., Dewey, M., Fielding, K., Gretton, V., Miller, P., Palmer, B., et al. Antidepressant drugs and generic counselling for treatment of major depression in primary care: randomised trial with patient preference arms BMJ 2001; 322: 772 2. Priest, R. G., Vize, C., Roberts, A., et al (1996) Lay people's attitudes to treatment of depression. British Medical Journal, 313, 838- 859. 3. Bedi, N., Chilvers, C., Churchill, R., et al (2000) Assessing effectiveness of treatment of depression in primary care. Partially randomised preference trial. Br J Psychiatry, 177, 312-318. 4. Ward, E., King, M., Lloyd, M., et al (2000) Randomised controlled trial of non-directive counselling, cognitive-behaviour therapy, and usual general practitioner care for patients with depression. I: clinical effectiveness. British Medical Journal, 321, 1383-1388.
Author’s conclusions were not justified by findings 1 April 2001
Wai-Ching Leung, Honorary Lecturer in Public Health Medicine HPP, University of East Anglia, Norwich NR4 7TJ Send response to journal: Re: Author’s conclusions were not justified by findings	In their randomised trial of antidepressant drugs and generic counselling for the treatment of depression, Chilvers et al (1) concluded that generic counselling is as effective as antidepressants and that general practitioners should allow patients to have their preferred treatment. However, their findings do not support these conclusions. The authors based their sample size calculation on a difference in mean Beck scores of 5 points as the outcome, and found that 44 patients per arm were required for a power of 80%. This sample size was not achieved in the randomised arms. They did not calculate the sample sizes required for global outcome or remission, but they are likely to be much larger as these outcome variables are categorical. Therefore, the only finding which achieved a power of 80% was related to Beck scores in the combined randomised patients with patients expressing preference. However, as clearly shown by both general practitioner’s rating and research diagnostic criteria score in table 1, patients choosing counselling were objectively significantly less depressed than other groups although their Beck inventory scores were similar. In other words, compared to other groups, patients choosing counselling were relatively more depressed subjectively than objectively. These patients are less depressed objectively and might respond more readily to interventions compared to other groups. Therefore, the authors should not have combined randomised patients with patients who expressed preference. Further, the authors cannot conclude that generic counselling is as effective as antidepressants simply from the apparent lack of differences in Beck scores in the combined patients who expressed preference. The authors further concluded that general practitioners should allow patients to have their preferred treatment. While this recommendation might be appropriate, it does not follow from their findings. To draw this conclusion, the authors would need to compare the outcomes of patients who chose a specific treatment and were offered it with those who requested the same treatment but were offered another treatment instead. Reference 1. Chilvers C, Dewey M, Fielding K, Gretton V, Miller P, Palmer B, et al. Antidepressant drugs and generic counselling for treatment of major depression in primary care: randomised trial with patient preference arms. BMJ 2001; 322: 772 - 775.
Doubtful 12 month intervention effects 2 April 2001
Christopher Dowrick, Professor of Primary Medical Care University of Liverpool Send response to journal: Re: Doubtful 12 month intervention effects	At first glance, the study by Chilvers et al appears to provide further support for counselling in primary care. However on more detailed examination, the evidence that they present is less convincing. They argue that a comparable reduction in BDI scores between patients receiving counselling and antidepressants is evidence that both treatment modalities are effective at 12 months. However, there is a contrary - and stronger - argument that their results demonstrate that neither modality is effective at 12 months. Two recently published studies of psychosocial interventions for depression have also included the Beck Ddepression Inventory as an outcome measure, and followed subjects up for 12 months (1,2). In contrast to Chilvers et al, both studies included a control group. At 12 months, neither found differences in mean BDI scores between their intervention and control arms. The figures were very similar to this one. The ODIN study had mean baseline BDIs of 23 across its three groups, with 12 month BDIs of 13 for controls, 11 for problem solving and 15 for the group intervention (differences not significant). The London & Manchester study had mean baseline BDIs of 25 to 27 across its three groups, with 12 month mean scores of 9 for cognitive-behaviour therapy, 11 for counselling and 10 for treatment as usual (differences not significant). This study has baseline mean BDIs of 25 to 27, with 12 month mean scores between 13 and 17 (Table 2 in text version). In other words, the effects of these interventions, which were demonstrable at an earlier stage, appear to have dissipated by 12 months. What we are probably seeing in all three studies is the natural history of depressive disorders, which is in general to remit spontaneously over time. Although I am a strong advocate of psychosocial interventions for depression, I do not consider that the findings from this paper provide us with helpful evidence in support of generic counselling. (1)Dowrick C, Dunn G, Ayuso-Mateos JL et al. Problem solving treatment and group psychoeducation for depression: multicentre randomised controlled trial. BMJ 2000,321,1450-4(see on-line version 9.12.2000, Table 6) (2) Ward E, King M, Lloyd M et al. Randomised controlled trial of non -directive counselling, cognitive behaviour therapy and usual general practitioner care for patients with depression. I:Clinical effectiveness. BMJ 2000,321,1359-1422.
Depression requires a multidimensional approach to managment 3 April 2001
Steve Williams, General Practitioner & Cognitive Psychotherapist The Gart Surgery, Guisborough Send response to journal: Re: Depression requires a multidimensional approach to managment	Editor, Chivers et al describe their findings that both antidepressants and counselling are equally effective for the treatment of mild to moderate depression in a naturalistic general practice setting. They argue that patients should be given the choice between these two approaches. 1 Given the size and cost of the problem of depression both in economic terms and in human distress, this "either / or" seems both unnatural and inadequate. It is obviously important to isolate the relative importance of different interventions but there is well established evidence that a combined pharmacological and psychological approach to treatment is the most effective. 2 However limited, every consultation in general practice involves some psychological intervention. This may explain the apparent effectiveness of prescribing sub-therapeutic doses of antidepressants as a powerful cognitive intervention rather than a simply a placebo effect, moving depressed patients deeper self-beliefs from, for example "Myself as a totally worthless, useless, helpless person" to "Myself as suffering from a legitimate illness called depression and worthy of medical attention and treatment".3 However if treatment is left at a purely pharmacological level, relapse rates remain unacceptably and predictably high, with a perception that further treatment depends upon further external medical intervention, rather than upon the skills learnt and internal self perceptions changed during psychological treatment. Depression is a potent drain upon our economic and human resources and is already the greatest single cause of disability across the world. 4 If we are to reduce this oppressive and costly burden, we need to combine the rapid benefits of antidepressants with the longer term effects of psycho-social interventions, recognising that an illness with multifactorial causes requires a multidimensional approach to its management. Steve Williams general practitioner & cognitive psychotherapist The Garth Surgery, Guisborough TS14 6AT 1 Chivers C, Dewey M, Fileding K, Gretton V, Miller P, Palmer B, Weller D, Churchill R, Williams I, Navjot B, Duggan C, Lee A, Harrison G. Antidepressant drugs and generic counselling for treatment of major depression in primary care:randomised trial with patient preference arms. BMJ 2001;322:772-5. 2 Hollon S, DeRubeis R, Evans M. et al Cognitive Therapy and Pharmacotherapy for depression: singly and in combination. Archives of General Psychiatry, 1992; 49, 774-781. 3 Barnard PJ, Teasdale JD.Affect Cognition and Change: Re-modelling Depressive Thought. Lawrence Erlbaum Associates Ltd.1993. 4 Andrews G. Should Depression be managed as a chronic disease? BMJ 2001;322:419-421
Two distinct anti-depression treatments used together 5 April 2001
Salvador Vale, Direccion de Posgrado Facultad de Medicina, Universidad Autonoma de Campeche, Campeche, Mexico Send response to journal: Re: Two distinct anti-depression treatments used together	On the basis of results of their study, Chilvers and colleagues (1) propose that since both types of treatment (talking therapy and drugs) provide the same benefit in depressive patients, we can offer them, one or the other, based of patient's preference. In the related editorial, Wagner and Simon give support to the same opinion (2). However, some points need clarification. Firstly, authors seem to consider that depression represents a homogeneous group of patients. Nevertheless, evidence show us that depressive diseases represent several pathological processes that we usually group together, perhaps because of their common symptomatic course. This phenomenon is more evident in elderly patients (3) where even a reversible pseudo-dementia phenomenon can segregate some patients into a separated sub-group that does not be benefited with "talking therapies". Secondly, it is important to consider the mechanism of action of each treatments modality. Antidepressant drugs modify several brain neurotransmission systems and increases neural growth factors. Talking therapy, when successful, may reduce maladaptive behaviour and consequently, we believe, it diminishes the excessive waste of the same neurotransmitters systems and growth factors that is secondary to non- adaptive life styles. Until new research may show us that this antidepressant-type therapy combination do not increase patient's health, there is no firm rationale for not trying them simultaneously (if the informed patient accepts it). A different case occurs when two treatments act on the same pathophysiological mechanism. Consider for example, the case of trying to reduce risk factors for cardiovascular diseases by a hypothetical cholesterol reduction obtained with both fibrates and a statin simultaneously administered. Although at first sight this strategy may work, additive effects should not be granted and further research is needed before accepting possible benefits (or inexistent benefits and side effects) using this approach. On the other hand, no physician would recommend that patients who do not like to carry the low fat diet, might eat what they want because they are already taking statins. Could it be possible then, in the treatment of major depression, that the recommendation regarding elimination of one or another option based on patient's preference, may be derived from the isolated analysis of research results? Is it not the same error that occurs if physicians look only the diseased processes without attending in preference the needs of patients? If we do not consider the whole patient, the complete diseases processes and the hypothetical mechanisms of action of the proposed therapies, we may be at risk of considering only "cold statistical" analysis of the research in therapeutic strategies, with the consequent unreliable conclusions. 1.- Chilvers C, Dewey M, Fielding K et al. Antidepressant drugs and generic counseling for treatment of major depression in primary care: randomized trial with patient preference arms. BMJ 2001; 322: 772 - 775 2.- Wagner EH, Simon GE. Managing depression in primary care. BMJ 2001; 322: 746 – 747 3.- Vale S. Drug treatment for depression in patients inside rehabilitation wards. Postgrad Med J. 2000; 76:596
further analysis 13 April 2001
Arthur Rifkin Hillside Hospital Send response to journal: Re: further analysis	This study has studied the treatment of depression in the real world, which I agree is very important, but I wonder about some of the conclusions. If the questions concerns the relative efficacy of the two treatment, the randomized sample gives us the most accurate result. The authors consider their main finding in table 3, that 56% of the counseling group and 65% of the drug group had a good or moderate results. This difference is not significant, and they conclude that both treatments work. But they don't report the 95% CI for this finding. As I calculate it, for the observed difference in outcome of 9%, the CI is 29,-10, which is rather wide for the conclusion that the treatments are equivalent. For the good outcomes only, they show 25% vs 41% favoring drug treatment, which they also report as non-significant. As I calculate the CI of the observed difference of 16, it is 34,-2. Not only is this wide, but it shows the strong weight toward the drug treatment. It seems odd to conclude that these treatments are equivalent. These results came from a definition of response that, if I read the people correctly, consists of any other three possibilities; <4 on the check list of the RDC, <10 on the Beck, or "clear documentation in the GPs notes." This seems a rather fuzzy definition, especially since the Beck, as a self-rating scale, has less validity that the usual rater- driven ones. I think the best measure would come from the RDC checklist, in which the GP recorded specific symptoms. In table 4, for completers, the results favored drug treatment 53% vs 22%, a significant difference. If we add to this the finding that drug treatment worked significantly faster, it seems to me that we should emphasize drug treatment for the routine care of depressed patients. The authors leave us with the weaker conclusion that we should give the preference to drugs because counseling is a scarce resource. We should not assume that GPs dispensing drugs are drug vending machines. I'm sure a great deal of important psychological interaction occurs during "drug treatment." Yours truly, Arthur Rifkin, MD
The type of treatment matters less than ensuring it is done properly and followed up 30 April 2001
David Simpson, Consultant psychiatrist Tavistock clinic Send response to journal: Re: The type of treatment matters less than ensuring it is done properly and followed up	Editor- I am writing to express my concern and dismay that this editorial, which concludes that the type of treatment given to patients with depression 'matters less than ensuring it is done properly', obscures the very interesting discovery, of Chilvers et al published in the same edition, that not only does the type of treatment matter but that it matters specifically to those patients who choose counselling who appear to do better. This finding, which suggests that type of treatment does matter to patients and that giving those patients with a preference a choice, clearly has important clinical implications. This simple message is that patients with depression not only have views about their treatment but tend to prefer treatments which give them psychological understanding and when they get this it can help recovery. With this emphasis rather than that of the editorial which stresses the equivalent value of drug treatment for depression suggests that we really need to increase investment into the development of and research into psychological treatments for depression. The spin of the editorial obscures these important implications and restores the current position, which emphasises pharmacological treatments for mental ill health. It is of concern but no surprise that the investigators declare research funding by three major drug companies. Yours sincerely, Dr DAVID SIMPSON FRCP Edin FRCPsych Consultant psychiatrist Tavistock Clinic, London NW5
Does randomization introduce bias? 10 May 2001
Toke S Barfod, Research Fellow Dept. of Infectious Diseases, Rigshospitalet, DK-2100 Copenhagen, Denmark. Send response to journal: Re: Does randomization introduce bias?	EDITOR - Generally, participation in a randomised trial seems to benefit to the patient (1). However, the results recently presented by Chilvers et al suggest that this may depend on the type of treatment under investigation (2). Chilvers et al performed "a randomised trial with patient preference arms" (2). They compared generic counselling with pharmacological treatment of major depression in general practice, and studied whether randomisation influenced their effects. They found the two treatments equally effective and that they both worked better if they were actively chosen by the patients, than if they were assigned by randomisation. This difference was, however, only significant for the group receiving counselling (table). Chilvers et al suggest that this finding should be interpreted with caution because of the low power of the study. However, their findings raise the general question, whether some types of treatment are more influenced by randomisation than others. It seems plausible that, due to enhanced compliance and placebo effect, some treatments work better when they are chosen, than when they are assigned by randomisation. It also seems plausible that randomisation may cause less damage to the effects of simple drug therapy, than to the effects of more "fragile" therapies, which require a strong commitment from the patient or the therapist. There is, however, currently little evidence available to test this hypothesis. Some analyses suggest that observational studies find treatment effects that do not differ systematically from the effects found by randomised controlled trials (3,4). Specifically for psychotherapy, observational studies appear to find less treatment effect than do randomised trials (5). This is contradictory to the hypothesis, but may be caused by selection and intensified care in randomised trials, rather than by the process of randomisation. Randomised trials are the gold standard when we evaluate therapies. Therefore we need to be very aware of possible biases inherent in the methodology. To further investigate whether randomisation introduces bias, it has been proposed to use randomised designs that also include the patients who were not randomised (4), as Chilvers et al did. However, the patients who choose to get randomised, may differ from those who do not. Therefore, a safer approach may be the two-stage trial design, in which patients are randomised to either randomisation or choice by preference (4). Extensive searches have found no such trials (4). Whether randomisation disturbs the effects of some "fragile" therapies remains unknown. Toke Barfod research fellow. Department of infectious diseases, Rigshospitalet, DK-2100 Copenhagen. Denmark Table: Drop in Beck depression score 12 months after treatment assigned by randomisation subtracted from drop in score after treatment assigned by preference (confidence intervals) Drug therapy 3,1 (-1,8 to 7,8) Counselling 4,6 (0,0 to 9,2) References 1. Ashcroft R. Giving medicine a fair trial (editorial). BMJ 2000;320:1686 2. Chilvers C, Dewey M, Fielding K, Gretton V, Miller P, Palmer P, et al. Anitidepressant drugs and generic counselling for treatment of major depression in primary care: randomised trial with patient preference arms. BMJ 2001;322:1-5. 3. Ioannidis JPA, Haidich A, Lau J. Any casualties in the clash of randomised and observational evidence? (editorial). BMJ 2001;322:879-80. 4. MacLehose RR, Reeves BC, Harvey IM, Sheldon TA, Russel IT, Black AMS. A sytematic review of comparisons of effect sizes derived from randomised and non-randomised studies. Health Technol Assess 2000;4(34):1- 154. 5. Shadish WR, Ragsdale K. Random versus nonrandom assignment in controlled experiments: do you get the same answer?. [Review] [169 refs]. J Consult Clin Psychol 1996;64(6):1290-305.