We read with interest the study by Wells et al ^{Reference Wells, Browne, Aguilar-Gaxiola, Al-Hamzawi, Alonso and Angermeyer1} where the important issue of adherence to treatment services has been addressed. Although the study analysed the data generated from the robust methodology of the World Mental Health Survey, which is a landmark in the field of psychiatric epidemiology, it needs to address some of the conceptual issues of treatment adherence particularly relevant to the low-/lower-middle-income countries.

Long-term follow-up and regular treatment is mostly prevalent in high-income countries that have an organised mental healthcare service. In countries having lesser mental healthcare resources, such coordinated provision of treatment is lacking. When treatment is sought from general medical services, the patient is only provided symptomatic relief and neither the provider nor the client has any knowledge about long-term follow-up. Such lack of communication between them is mostly due to deficiency of mental health infrastructure in terms of either quality or quantity. ² One may argue that traditional or non-conventional modes are the main treatment providers in such countries. But for them often the treatment proceeds on an ‘as and when required’ basis. ^{Reference Chavan, Gupta, Sidana, Arun and Jadhav3} For spiritual and religious healers the client would often be attached to them in a special bond of faith or gratitude for generations, such as in the guru-chela relationship. ^{Reference Neki4} In such situations, a question such as ‘Did you complete the full recommended course of treatment? Or did you stop before the [provider] wanted you to stop?’ seems irrelevant. We propose that a little extra effort to standardise this question across different settings would have made the methodology of Wells et al more robust.

Slightly different definitions for mental health treatment drop out have been used in previous studies. ^{Reference Edlund, Wang, Berglund, Katz, Lin and Kessler5,Reference Wang6} The authors have very rightly pointed out that this is one of the reasons for the differences between drop-out rates found in national surveys and corresponding subsamples of the present study. So, if such a ‘slightly different definition’ of drop out influences their rates in high-income countries where the determinants are less heterogeneous, we can obviously assume that its effect on the low-/lower-middle-income countries will be marked.

Although the authors have made elaborate attempts to find the predictors of drop out, they did not take into account many potentially relevant factors related to patient (e.g. stigma, functional impairment, satisfaction with treatment), professional (e.g. communication skills, clinical expertise) and service delivery (e.g. environmental obstacles). Apart from this, the fact that the centres in some countries were not representative of the whole population influenced generalisability of the study. Overall, this unique effort by the authors is praise-worthy and will go a long way in understanding the dynamics of treatment drop outs from a global perspective.