Electronic Letters to:

PAPERS:
JONATHAN I. BISSON, JONATHAN P. SHEPHERD, DEBORAH JOY, RACHEL PROBERT, and ROBERT G. NEWCOMBE
Early cognitive–behavioural therapy for post-traumatic stress symptoms after physical injury: Randomised controlled trial
The British Journal of Psychiatry 2004; 184: 63-69 [Abstract] [Full text] [PDF]
*eLetters: Submit a response to this article

Electronic letters published:

[Read eLetter] CBT Interventions in Post-Traumatic Stress Disorder - Methodological Issues
Richard H. Baker, Prashant Mayur and Siddharta Dutta   (25 January 2004)

CBT Interventions in Post-Traumatic Stress Disorder - Methodological Issues 25 January 2004
Top
Richard H. Baker,
Specialist Registrar
Depts.of Psychiatry, Monash Medical Centre,Clayton VIC 3168 and Shepparton Hospital, 3630, Australia,
Prashant Mayur and Siddharta Dutta

Send letter to journal:
Re: CBT Interventions in Post-Traumatic Stress Disorder - Methodological Issues

r_baker_psychreg1{at}hotmail.com Richard H. Baker, et al.

Dear Madam, Sir

With reference to Bisson et al. (2004) we want to express our admiration for the clear, well – presented and important study. We would like, to raise, however, a few issues in regards to this publication:

- We wonder whether an ANCOVA is the optimal approach to analyse the result of this study; we felt that repeated measure ANOVAs, taking the patients as their own control, might be at least as good or even preferable as a second pathway of analysis.

- Linear regression analysis is originally designed for continuous variables, yet the IES and HADS used are scales using categorical variables. We are aware that in ignorance of the actual IES cut-off this is often ignored, but we wonder whether this might make the regression analysis less applicable and whether logistic regression analysis might be preferable.

- Although the raters in this study were excellently ‘blinded’, the approach to the participants who did not receive intervention meant that the study was clearly not double blind. Additionally, the clinicians delivering the intervention were clearly not blinded. The “gold standard of clinical research” is the randomised double-blind design, defined as “neither the patient nor the physician is aware of the treatment assignment.”(1) A series of sham CBT with the non-intervention group might be possible, but naturally the delivering clinician would know that it is only a sham treatment. Given the possible interaction with dynamic factors, like interpersonal skills, empathy and nonverbal communication in CBT, the question how much of the results in CBT are a result of these non -specific confounding factors, or the actual CBT technique as such, is not always easy to answer. Maybe in psychological interventions proper double -blinding is inherently impossible to achieve.

Reference (1) Rosner, P., Fundamentals of Biostatistics, Pacific Grove (CA), Duxbury 2000, p. 167.