**1. Introduction**

Drowning is a leading cause of morbidity and mortality in low-and middle-income countries (LMICs) [1]. Of these deaths, 62,000 occur in India, where drowning is the foremost cause of death by injury for children aged 1–4 years [2]. Rural and remote coastal regions in LMICs present the highest risk of child drowning. Rural, forested Sundarbans region in the northern state of West Bengal is one such area. Sundarbans experiences frequent flooding, a presence of open water, poor infrastructure and poor health systems [3–5]. A recent survey in the Sundarbans found particularly high rates of drowning in children aged 1–9 years where it is likely the leading cause of death in this group [6].

The World Health Organization (WHO, Geneva, Switzerland) recommends the implementation of four effective community-based interventions in rural LMIC settings to reduce drowning in young children. These interventions are activities that may be feasibly implemented in low-resource

contexts and have been shown to reduce drowning burden [1]. These interventions are: the installation of home-based barriers controlling access to water (such as playpens and door barriers), the provision of supervised safe spaces with capable child care, teaching school-aged children basic swimming and rescue skills and training adult bystanders in rescue and resuscitation [1]. Previous research has shown that communities in the Sundarbans consider drowning a health issue [7]. Despite this perception and the high rates of drowning, there are no preventive measures implemented in the region.

Previous research and experience in the sustainable program design has shown that it is essential to understand the context, local perceptions and possible implementation-related challenges before designing and implementing community-based programs [8–10]. The identification of key stakeholders that may support or inhibit implementation must also be identified [11,12]. These stakeholders can include members of the community who can influence program engagement, as well as governmental or organisational leaders whose support and buy in is beneficial for community acceptance and access to local resources.

A key strategy that improves program sustainability is linking program goals with governmen<sup>t</sup> priorities and leveraging existing programs [13,14]. A comprehensive policy review of West Bengal and National policy found three governmen<sup>t</sup> programs that may be appropriate to build upon to implement drowning reduction programs: the integrated child development scheme (ICDS), self-help group (SHGs) schemes and the accredited social health activist (ASHA) program [15]. The federal ICDS program was introduced in 1975 and aims to provide free childcare services to children aged 3–6 years through village-based Anganwadi centres [16]. The implementation and reach of these centres are highly variable across the Sundarbans, and many centres do not provide the childcare services promised in the policy (Biswas and Chattapadhyay, 2001; Biswas et al., 2010). The quality improvement of the ICDS program has the potential to provide structured supervision, for injury prevention. The SHG scheme aims to reduce rural poverty and increase household income through the setup of self-help groups in villages, primarily with women. Some SHGs also become involved in community projects, such as the provision of midday meals in schools [17]. ASHA workers are community-based health workers who focus on child and maternal health on an incentive-based system, and have close ties with mothers [18,19]. Both SHGs and ASHAs may be leveraged in the provision of community-education such as rescue and resuscitation training and supporting families in building and maintaining home-based barriers.

We conducted the formative contextual analysis required to design a sustainable drowning reduction program for the Sundarbans, as guided by WHO recommendations. The objectives were as follows: (1) identify community perceptions and preferences towards the recommended drowning interventions; (2) explore the feasibility of leveraging ICDS, ASHA or SHG programs for the delivery of drowning reduction interventions; (3) identify contextual challenges and considerations for the design and delivery of the program; and (4) identify key stakeholders who should be engaged during the development and implementation of the program.

### **2. Materials and Methods**

We applied qualitative methods to understand the micro context in which drowning reduction interventions could be delivered in the Sundarbans. In-depth interviews (IDIs), focus group discussions (FGDs) and observations were conducted and triangulated to develop this understanding. IDIs gave insights into individual-level perspectives, and FGDs were used to identify community norms and perceptions. Observations allowed for the better understanding of governmen<sup>t</sup> program operations and systems. Qualitative methodology was guided by the Consolidated Criteria for Reporting Qualitative Research (COREQ) (see Supplementary Table S1) [20].

### *2.1. Data Collection*

Data collection was conducted in partnership with a local non-governmental organization (NGO), the Child in Need Institute (CINI). CINI has operated child and maternal health programs in rural West Bengal for the past 46 years and has extensive connections with communities and local governmen<sup>t</sup> in the Sundarbans region. The data collection of community-based participants was completed by two male data collectors recruited by CINI, managed by S.R. and R.P. who work as the programs' manager and director, respectively. The data collectors had previous experience in qualitative research in West Bengal and were trained by the researchers in the study aims and tools. One of the data collectors had experience conducting qualitative data collection in the Sundarbans and was familiar with the community. M.G. conducted English-language interviews, such as with grassroots organisations.

All data collection occurred face-to-face. IDIs and FGDs were held in locations that best suited participants, such as in community schools or Anganwadi centres. In addition to the data collectors and participants, NGO partner facilitators were present for some IDIs, FGDs and observations to lend logistical support. All IDIs and FGDs were audio recorded and lasted between 30 and 90 min. Field notes were also taken by one data collector and collated to make key point summaries of each IDI, FGD and observation on a daily basis, which was shared with the research team.

All Bengali transcripts were translated into English for analysis. No interviews were repeated. Transcripts were not returned to participants for comment due to the logistical and literacy barriers.
