The authors are to be congratulated on undertaking an important pragmatic trial in a methodologically difficult area. However, there are (as the authors have stated) some potential issues in interpretation related to diagnostic heterogeneity and design. It might be helpful to have further clarification of
certain aspects of the study design and data to help in its interpretation.

1. The authors mention that interventions are available or have been previously developed for certain groups e.g. severe depression etc. It is stated that these are effective and this has led to exclusion of these groups, from this study. Is there a possibility that this compromised generalisability of the findings and is there any data on the outcome of such patients who were excluded during the study period?

2. It is appreciated that it might be difficult in a pragmatic design, but how clearly can the standard treatment and intervention treatment be differentiated? There seem to be quite wide variations within each treatment group in the intensity of treatment and amount of contact. Although a definition of adherence in the intervention group is given, is it possible that the standard treatment group received overlapping forms of treatment such as problem solving (albeit in a less formalised and structured way)?

3. I was not clear on the extent or type of co-morbid psychiatric diagnoses in the group e.g personality and substance use diagnoses. Although the numbers presented might be small, it is unclear if this might have relevance in understanding potential interactive effects between diagnostic categories on drop-out or outcome.

Competing interests: None declared