Author's response

We thank Kumar & Das for raising some important issues about our published article.¹ We hope that our findings do not suggest or support the better outcome in low-middle-income countries as an axiom. Our findings are, in fact, limited to the two specific contexts in which the study was carried out. Here we attempt to address each of the issues raised by them.

First, family support. We acknowledge that the measure of family support used in this study has its limitations, especially in terms of not being able to explore greater detail. The authors of the letter are correct in their interpretation of the scoring of this measure that, theoretically, a participant could end up with a score of zero, if one of the two questions had a score of zero. We had verified this as part of our analysis and none of the patients had a score of zero on either of the two questions. Therefore, the variance in the scoring of the measure was retained.

Second, family income. The authors raise an important point regarding family income adjusted to gross domestic product. Although we attempted to collect these data, unfortunately they were available infrequently and not always reliably. For some patients this meant individual income (Montréal) whereas others interpreted it as family collective income (Chennai). As a result of the unreliability of the comparative nature of the data across the two sites, we did not use it as part of our analysis. However, future studies should make an effort to do so.

Third, extrapyramidal symptoms (EPS). Given that almost invariably second-generation antipsychotic medications were used at both sites, the frequency of EPS was expectedly low. We used the prescription of anticholinergic medication as a proxy for EPS because it was collected consistently. The rate of such use was similar for patients with affective and non-affective psychotic disorders (mean rate over 24 months 9.0% and 7.5%, respectively for schizophrenia spectrum and affective psychoses, in Montréal). Only 1/15 patients with affective psychosis in Chennai received anticholinergic medication.

Fourth, patient mortality. We agree that this issue needs ‘greater emphasis’. Greater details were provided in the original submission but could not be accommodated in the final version in response to one of reviewer's comments as this was not the focus of the study. Briefly, all deaths occurred in the first 3 months of treatment and all had a diagnosis within the schizophrenia spectrum disorders.

Fifth, differences in attrition rates. The highly significant difference in attrition from treatment and the study across the two sites are presented and discussed in detail in a separate publication.²

Sixth, medication and adherence. Medications were available to all patients free of cost if they could not afford them (Chennai) and/or through a mandatory state-funded system if they did not have private insurance (Montréal). Rates of adherence to medication were similar in Montréal and Chennai (modal adherence rate 80% and 82% in Montréal and Chennai, respectively). Hence this variable was not entered into the regression analysis.

We hope that we have addressed the important questions raised by Kumar & Das.