Paper A reviewer's report

“Randomised controlled trial of computer-guided self-help for panic/phobic disorder: Hope for cutting waiting lists?”

This paper reports the outcome of a randomised controlled trial of a computer-guided behaviour therapy, clinician-delivered behaviour therapy and computer-guided relaxation for individuals with DSM-IV phobic disorders (including agoraphobia with panic disorder). Ninety-three participants were randomised, and 65 completed treatment, over 10 weeks. One and three months later, 60 and 52 individuals, respectively, were seen for follow-up assessments. Overall, significant improvements were found on all outcome measures for the two behaviour-therapy arms compared with the group who received computer-guided relaxation therapy. No statistically significant difference was found in outcome between the behaviour therapy groups, although the authors reported that the computer-guided group received 73% less clinician time than the other group.

Importance of the subject

The treatment of anxiety disorders is extremely important. As the authors point out, a large proportion of cases go untreated, in part because of the shortage of trained therapists. This paper is therefore of potential importance to a general readership. However, I would suggest that the importance of this subject might be enhanced by a more comprehensive introduction.

I think a general readership may not appreciate the significance of anxiety disorders in either their impact on individual well-being and social functioning, or in terms of their public health consequences. The latter is manifested in part by the burden these conditions place on primary care.

I worry also that the high proportion of self-referred patients may lead readers to view participants in this trial (and anxious patients in general) as the ‘worried well’. The authors need to address this in their Discussion.

Originality

This is an original piece of work. Professor Marks enjoys an international reputation for having established the effectiveness of behaviour therapy in anxiety disorders, and for having identified the processes responsible for improved outcomes. This paper is part of this on-going programme of research.

Scientific reliability

1. The research question is well-defined. The degree to which this has been answered depends on answers to a number of methodological questions.

2. The design of the study is satisfactory, despite the relatively high rate of attrition at follow-up.

3. I have major concerns about the methods used in this study. Firstly, it would be informative to know how many researchers conducted the assessments, and what training they received. Secondly, the diagnostic assessments were inadequately described, and it is not clear what instruments or criteria were used. Thirdly, the authors provide insufficient evidence of the psychometric properties of their outcome measures. They have not cited a single peer-reviewed reference to support the reliability and validity of these rating scales, which must therefore be viewed as unproven. These are major oversights.

4. The participants are adequately described, on the whole. However, it was unclear how this specialist service came to see so many self-referrals. This should be explained more clearly. The statement on p12, para 7 is inadequate in this respect.

I note that there is no information about the receipt of previous psychological treatment by the participants. While it is possible that none of these individuals had received any such treatment, this should be stated. Previous treatment experience might have had an effect on drop-out, if not outcome. This is a major oversight.

5. The presentation of the study findings should be improved. I note in particular that very few results of statistical tests are reported, and that there is an omission of confidence intervals throughout the paper. This is especially evident in the Abstract. This is a major oversight.

6. The Discussion could be shortened. The interpretation of the findings is reasonable, though it might be restructured in a slightly more organised fashion. I would suggest discussion of the methodological limitations of the study should precede discussion of the generalisability of these findings.

With respect to this, I think the authors should consider the implications of a 43% drop-out rate in the group assigned to computer-guided behaviour therapy. While this appears to be an effective intervention, aspects of this intervention may be unacceptable to the target population.

I am perplexed by the sentence on p12, para 6 that ends with “…NICE??”. I presume this is a typographical error.

7. The references are unsatisfactory, and include several unpublished papers.