*2.5. Analysis*

Analysis of the transcripts was completed using NVivo 12 [21]. Narrative analysis was used where key themes under each of the broad research objectives were derived. All transcripts were coded against a priori key themes based on the research questions, including the acceptability of each of the WHO drowning reduction programs, and considerations for the implementation and feasibility of using governmen<sup>t</sup> programs to deliver the programs. Subsequent sub-themes were developed under each of these based on commonalities and diversified perspectives from participants. We also triangulated di fferent sources of data by coding for the type of stakeholder and type of qualitative method (IDI, FGD, observation) to assess congruen<sup>t</sup> and di fferent perspectives across genders and participant type as well as to compare individual and community-level viewpoints [22]. The two independent reviewers (M.G. and P.K.) discussed their results and discrepancies before finalising the key findings.

Stakeholders were identified and then allocated to level of power and interest as based on Mendelow's Matrix [23]. The level of power describes the stakeholder's influence over program success, and the level of interest reflects the impacts of the program on the stakeholder. The framework was used to identify the correct engagemen<sup>t</sup> strategies for each of the stakeholders based on their framework allocation.
