Author's response.

We thank Kanica Kaushal for highlighting the issue of “treatment contamination”. In the study1, contamination rate between the two study groups was assumed to be minimum or negligible, since the six villages that were selected randomly based on accessibility to PHC were not in close proximity to each other. Although the contamination rate was assumed to be minimum, we now present results of Contamination adjusted intention-to-treat (CAITT) analysis based on the study by Sussman et al2.

In this study, the number of clinic visits and weeks of treatment on antidepressant medication were significantly greater in the TI group compared to TAU group treatment even after performing intention-to-treat analysis as well as adjusted for clustering effect in the analysis. Using CAITT analysis, the effect of treatment assignment on outcome observed (number of clinic visits and weeks of treatment) was adjusted by the percentage of assigned participants who completed the treatment. Treatment completed was considered as an instrumental variable (if at all, the study groups were contaminated, we expect the percentage of patients who completed the treatment to be greater in control group also). CAITT analysis was performed using SPSS version 18.0 (SPSS Inc., Chicago II, USA). Using two stage least square analysis, the effect of treatment was significantly greater for each of the outcome in the TI group compared to the TAU group after adjusting for the instrumental variable (P<0.001). The results support our findings that subjects in the treatment intervention group were much more adherent to treatment protocol compared to the treatment as usual group as reported earlier1.