Reply

We thank Eddleston and Weerasinghe ^{Reference Eddleston and Weerasinghe1} for their careful engagement with the SPIRIT trial and welcome the opportunity to clarify interpretation and reporting. Our paper ^{Reference Vijayakumar, Pathare, Jain, Pandit, Lohumi and Kalha2} reports that the integrated package of three interventions did not reduce the primary composite outcome (suicide attempts plus suicide deaths) at 12 months (rate ratio 0.89, 95% CI 0.62–1.27). We also reported a lower between-arm suicide death estimate during follow-up (rate ratio 0.57, 95% CI 0.32–1.02, P = 0.061), while explicitly noting that the trial was not powered to detect statistically significant differences in suicide deaths. We agree that this finding should be interpreted as a trend rather than as definitive evidence of reduced suicide mortality at 12 months. We would add, however, that suicide deaths did not increase in absolute terms in the intervention arm (44 deaths in both periods), whereas they rose markedly in the control arm (38 to 66 deaths). The adjusted estimate (incidence rate ratio 0.57) therefore reflects a meaningful difference in trajectory relative to contemporaneous changes in controls, and we believe that this trend merits explicit reporting even given the wide confidence intervals and the trial’s power limitations.

SPIRIT was designed as a pragmatic cluster randomised trial of an integrated village-level package (school mental health promotion, centrally managed pesticide storage and community health worker training). As pre-specified in the protocol, the pesticide storage component was evaluated using the same composite primary outcome (all-cause suicide/attempt) as the overall trial; therefore SPIRIT does not permit causal attribution to pesticide storage in isolation, and we stated this explicitly as a limitation. We nevertheless reported implementation details at scale: 52 of 54 intervention villages established storage facilities and 8370 lockers were provided, with high uptake (88.83% used at least once). We emphasise that SPIRIT was designed as a real-world implementation study, not a controlled efficacy trial. Effect dilution is well recognised when interventions are scaled from controlled to real-world delivery settings, and the usage metrics we report reflect adoption at this scale. We accept that more granular reporting, including distribution of withdrawals, seasonal patterns, and the duration of pesticide storage in lockers, would allow more precise evaluation of whether access was meaningfully reduced during impulsive crises; we will address this in the published record as data analysis allows. We do not interpret SPIRIT as a definitive efficacy test of pesticide storage in isolation, but as evidence about the feasibility and potential effects of a multilevel package implemented at scale.

We share concerns about measuring non-fatal attempts under stigma and limited routine data. This was the rationale for implementing a triangulated community informant surveillance system with hospital and police records, previously shown to identify substantially more suicides and attempts than official sources alone. We are also pleased to note that our comprehensive community surveillance approach has since been incorporated as the basis for the World Health Organization toolkit on community surveillance for suicidal behaviour, ³ reflecting its broader adoption and quality. We acknowledge that ascertainment may still be incomplete. Critically, however, under-ascertainment of non-fatal attempts does not materially undermine the suicide death trend finding: deaths are substantially less likely to have been missed through our triangulated approach, and the secondary finding of a trend towards reduced suicide deaths is therefore not affected by incomplete attempt ascertainment. We agree that reporting of method-specific outcomes (pesticide versus other) and more granular storage usage metrics would strengthen interpretability, and we will clarify these points in the published record as appropriate.

We agree that differential ascertainment of non-fatal attempts is a key risk in community surveillance studies. The SPIRIT protocol addressed this through standardised surveillance procedures, extensive training and retraining of community informants, and via triangulation across multiple data sources; these procedures were applied consistently throughout the study in both arms. We also note that community health worker training and community engagement may independently increase detection and reporting of non-fatal attempts, potentially making counts of suicide attempt more dependent on surveillance capacity than suicide deaths, which are inherently less subject to underreporting. We therefore interpret changes in attempt rates conservatively and foreground the suicide death estimates alongside the pre-specified limitations regarding ascertainment.

Regarding the earlier four-village study, ^{Reference Vijayakumar and Satheesh-Babu4} we acknowledge that it should not be characterised as evidence of effectiveness, given its baseline imbalances and the authors’ own caution in interpreting results. Our intent in citing this work was to demonstrate that communal pesticide storage models were feasible and acceptable in the Indian context, and that is precisely what the feasibility study demonstrated. It was not a pilot efficacy study, and we did not intend to present it as such. We will rephrase any wording that implies causal conclusions from the feasibility work, and will emphasise that SPIRIT was designed to provide more robust evidence at scale.

Finally, we welcome the commentary’s emphasis on highly hazardous pesticide (HHP) regulation and wish to be explicit: we have never advocated against bans on hazardous pesticides. Our reference to ‘significant hurdles’ in the context of India was not intended to suggest that bans are impossible: India has already banned or restricted several HHPs, including methyl parathion, phorate and phosphamidon, and monocrotophos has been restricted for use on vegetables. Our intent was to note the practical challenges in achieving comprehensive, enforced restrictions at pace. Reducing pesticide suicide deaths requires a combination of strategies, and centrally managed community storage was positioned as a pragmatic, locally governed interim intervention alongside continued strong advocacy for regulation. We support HHP bans as a central long-term policy goal and positioned storage as a complementary, not competing, approach.