To the Editor: The randomised controlled trial recently reported by Blashki and colleagues does not support their hypothesis.1 The drop-out rates in both arms of the trial were very high — only 62% of general practitioners in the intervention group and 54% in the control group completed the trial. One cannot have any confidence in their conclusion that a short training course can improve GP skills in the provision of cognitive behavioural strategies (CBS). For example, what if the 38% of GPs in the intervention group who dropped out actually deteriorated in their CBS skills and therefore declined to be videotaped?

Furthermore, only 56 of 1021 GPs in Victoria were willing to enrol in the trial. The authors concluded that their findings could only be applied to GPs who have a special interest in mental health. Perhaps the low participation rate indicates another more relevant idea — that GPs have had enough of “training models” being imposed on their lives. A recent systematic review has confirmed the low impact that educational training programs have on GPs for the management of mental health problems.2

In reply: Our words in conclusion to our article were carefully chosen as: “Competency in CBS [cognitive behavioural strategies] in highly motivated GPs [general practitioners] can be improved by a brief training intervention”1 (italics added) — not that all such interventions will lead to improvements for all GPs, but that well designed and conducted training for selected GPs can do so. Research so far leaves open the possibility of large enough effect sizes for GP mental health training to be relevant to policy.2 The review cited by Kljakovic was limited in scope and noted the poor quality of studies included.3

A drop-out bias in our study, as proposed, seems most unlikely to us. Rather than being imposed, this training model was developed with GPs, by GPs and for GPs, and so might achieve better results than previous interventions studied. Funding for GP participation such as that more commonly available in drug studies might have increased participation. It is true that a great deal of training has been offered to GPs, and we hold that our study shows that such training can lead to GPs significantly improving their skills in this area. GP training should be considered within multifaceted interventions to improve primary mental health care.4