BMJ Mental Health and Bias

It is regrettable that BMJ Mental Health marks its transition from the Journal  Evidence-based Mental Health with the publication of a paper by O’Driscoll et al (2023) that obscures allegiance bias, by the authors simply declaring what grants they receive.  The authors work either for NHS trusts or IAPT, the former operate the latter. They use an IAPT dataset and uncritically utilise the services self-serving metric of recovery. These authors have not considered the Cochrane Risk of  Bias tool against which the study would have been judged as at high risk of bias. There is no acknowledgement of works that cast serious doubts on the Services claimed 50% recovery rate, Capobianco et al (2023), Scott (2018)

 The O’Driscoll et al (2023) paper claims that CBT may be preferred  to counselling for clients who have anxiety symptoms comorbid with depression. But the conclusions are built on sand in that:

  1. there can be no certainty that the subjects studied were depressed as there was no ‘gold standard’ diagnostic interview conducted. Instead reliance was placed on a psychometric test, PHQ-9
  2.  there can be no certainty about comorbidity because of the absence of a diagnostic interview
  3. no fidelity checks were carried out to establish whether therapists were conducting CBT or counselling. Reliance was instead placed on therapists claims.
  4.  no blind-raters were used to assess outcome
  5. there can be no certainty that the observed changes would not have happened anyway because of the absence of a credible attention control condition
  6. there can be no certainty that the observed changes were clinically meaningful or that changes endured. A 6 point improvement in the CBT group and a 5 point improvement in the counselling for depression group on the PHQ-9.
  7. the study was restricted to patients who attended 5 or more treatment sessions, but these are unrepresentative of IAPT clients. Only half of clients have 2 or more treatment sessions (defined by IAPT as ‘treatment’). The mean number of IAPT treatment sessions is 7 but the mean number of treatment sessions in the O’Driscoll et al (2023) study was 10 in counselling for depression and 11 in CBT. Further the third of IAPT clients who undergo low intensity intervention alone were excluded. Generalisation from this study is fraught with difficulties

Does the emergence of BMJ Mental Health signal the demise of evidence-based mental health? I hope not.

Capobianco, L., Verbist, I., Heal, C., Huey, D., & Wells, A. (2023). Improving access to psychological therapies: Analysis of effects associated with remote provision during COVID-19. The British journal of clinical psychology62(1), 312–324.



O’Driscoll C, Buckman JEJ, Saunders R, et al Symptom-specific effects of counselling for depression compared to cognitive–behavioural therapy BMJ Ment Health 2023;26:e300621.


Scott M. J. (2018). Improving Access to Psychological Therapies (IAPT) – The Need for Radical Reform. Journal of health psychology23(9), 1136–1147.