3.1.2. Affordability

Cost to households was a concern for all interventions. Many participants stated that with limited resources and competing priorities, a drowning reduction intervention would not be affordable for households. The home-based barriers' intervention was considered the most feasible for self-funding as it was viewed as a one-time investment, with maintenance being of negligible cost. They also noted that parents who are unable to afford services may cause problems and complain if excluded. Some participants suggested that families could pay different amounts depending on their income level, which could be pooled together to fund the program (Ref 3 in Table S2).

### 3.1.3. Community Engagement and Ownership

Consistent community engagemen<sup>t</sup> through regular meetings, showcases, theatre and household visits were identified as important to implementation success. Participants noted that program ownership should be transferred to the community over time, such as by setting up an implementation committee. Participants noted that without consistent engagement, people may fall back into previous habits and stop engaging with the program (Refs 4 and 5 in Table S2).

Community leaders and grassroots organisations' participants also discussed the importance of regular program monitoring. They stressed that communities and implementing agencies should work in partnership to ensure that interventions were being implemented and used as designed (Ref 6 in Table S2).

### 3.1.4. Resources and Skill Set

Participants also noted that geographical and infrastructure barriers such as the road quality and the connectivity of many areas were challenges. Participants suggested that local resources should be used where possible, such as bamboo from the area for barriers (Ref 7 in Table S2).

Grassroots organisations also noted that finding capable human resources was often challenging due to lower educational attainment in the region and the migration of skilled workers to the cities. Benefits and incentives would need to meet community expectations to recruit capable staff. However, the programs would provide an opportunity for women to access employment, as few jobs were available to them post high-school. Program providers may also face risks if a child was injured under their care from angry parents (Ref 8 in Table S2).

### 3.1.5. Social Class

Participants largely stated that caste and religion did not present an issue. Community member participants did not anticipate any discrimination towards potential intervention beneficiaries. However, some instances of discrimination against Muslim Anganwadi workers by Hindus, or against Hindu SHG members by Muslims, were reported by governmen<sup>t</sup> program participants during IDIs. Government program participants also stated that political party affiliation may affect cooperation and participation in interventions. Program staff from different political parties may refuse to work together or may discriminate against communities from other parties (Refs 9 and 10 in Table S2).

Some participants also noted that as Muslims were relatively economically disadvantaged and conservative, they may have less capacity or willingness to pay (Ref 11 in Table S2).

### *3.2. Intervention-Specific Considerations*

In addition to findings to guide general program implementation, specific considerations were identified for each of the WHO-recommended drowning interventions.

### 3.2.1. Home-Based Barriers

Home-based barriers were largely acceptable to communities provided certain conditions were met. Many adults noted that this method was used previously in their childhood but concerns for children's mental wellbeing stopped the practice as a lack of movement and social interaction with other children was considered detrimental. The intervention was also only considered suitable for younger children under the age of 2–2.5 years, as older children would try to climb the barriers (Ref 12 in Table S2).

The feasibility of different types of barriers varied between households. Some participants noted that families may struggle to keep door barriers and pond fencing gates closed due to regular access. Building and maintaining fencing around all ponds within 20–50 m of homes may not be feasible due

to the large number of ponds in some villages. Some community members expressed concerns over restricting children's movement in playpens which may be detrimental to their development (Refs 13 and 14 in Table S2).

For playpens, many participants noted that an adult would still need to be present to ensure safety. Participants also suggested that door barriers or fencing gates could be made lower so that adults could climb over without opening them, increasing convenience and reducing the likelihood of it being left open. One participant suggested that young children from nearby homes could be kept together in a large playpen in the middle of the homes with one adult supervising (Refs 15 and 16 in Table S2).

Some participants identified that locally trained professionals were required to build and install the barriers to maintain quality (Ref 17 in Table S2).

### 3.2.2. Childcare and Supervision-Based Programs

Childcare was largely acceptable in communities, especially as it provided parents with relief from supervision while they worked and o ffered an opportunity for children to participate in early childhood education, including for children with disabilities who often had few avenues for learning (Ref 18 in Table S2).

However, some participants were concerned for children's safety, as one adult was not considered enough supervision for a group. The region had also experienced instances of child tra fficking. Parents were also busy during the day and often restricted in their ability to pick up and drop o ff children. This issue would be exacerbated in monsoon season when roads are flooded. Parents were also concerned that young children below the age of two years old would not engage with activities and experience separation anxiety.

Participants o ffered a range of suggestions for childcare implementation. Children could be divided into groups by age so they could be engaged in age-appropriate activities (Refs 19 and 20 in Table S2).

Participants stated that more than one carer was required to look after children to ensure they remained supervised if one child had to be taken for a bathroom break. They also supported the employment of a trusted and known local woman with training for the role (Ref 21 in Table S2).

The provision of toys, activities and learning material was also required to ensure that parents and children would be interested. A gated 'community playground' was suggested to provide an outdoor play space. Pick up and drop o ff services would increase attendance. Toilet and water facilities were also required. Food provision would improve attendance as both parents and children would be more satisfied. The venue was also required to be large and secure for safe play (Refs 22 and 23 in Table S2).

The preferred hours for the childcare services varied. Many participants, especially mothers, noted that parents were busy in both the morning and afternoon but were home for lunch. They suggested a session both before and after lunch (Ref 24 in Table S2).

### 3.2.3. Swim and Rescue Training

Many participants believed that children had adequate swimming skills from informal lessons provided by parents in family ponds but were interested in rescue training. Participants acknowledged that some individuals did not have access to a pond to learn or did not have time to teach their children swimming, so classes were important for them. Children would also be motivated by the chance to participate in regional swimming competitions (Ref 25 in Table S2).

Ponds in this region are mostly privately owned and used for washing, cleaning and fishing. Some participants reported that there were no common ponds large enough for training in their communities, so private ponds were required. Seeking someone who would lend their pond may be di fficult. In addition, many ponds were unsuitable, being dirty and deep (Ref 26 in Table S2).

For quality control, participants suggested that guidelines for pond selection should be developed, covering location, cleanliness and depth criteria. Safety and rescue material should also be available, and platforms built for access to the pond. A changing room would also reduce community push-back as children would not travel home in wet clothing (Refs 27 and 28 in Table S2).

Some participants believed a trainer from outside the community would be better respected by the community, while others preferred a local who would have better relationships with the community and would be more consistently available (Refs 29 and 30 in Table S2).

### 3.2.4. First Responder Program

The first responder training was acceptable to most participants, especially parents of young children as they were interested in learning how to protect them (Ref 31 in Table S2).

However, cultural beliefs may remain a barrier to appropriate responses. During drowning events, people in the communities had previously ignored health advice from community health workers such as ASHAs and conducted traditional responses, such as calling a local village doctor or performing rituals on the water. These responses had led to delays in children receiving appropriate medical care (Ref 32 in Table S2).

### 3.2.5. Indigenous Interventions for Child Safety

A range of other intervention ideas and solutions were o ffered by participants. Many stated that awareness programs were required in parallel to drowning interventions to educate communities about the risks of drowning and ensure sustained behaviour change. Awareness activities would also seek to dispel harmful beliefs about drowning, such as on cultural post-drowning rituals that led to delays in children receiving first aid. Participants noted that other existing programs in communities with established activities, such as vaccination programs, could be leveraged for awareness activities (Ref 33 in Table S2).

Native interventions employed by communities were also identified. Some parents tied their children to their waist or to the house with rope while they worked. Others kept their children locked inside the home alone when they were away (Ref 34 in Table S2).

Other possible solutions were o ffered such as providing vans for school children or organising 'walking buses' where children would travel to school together, and teaching children to have a 'shore guard' during play time where one child kept watch from the pond's edge.

### *3.3. Use of Government Programs in Drowning Intervention Delivery*

Possible roles in the implementation of drowning interventions were identified for existing governmen<sup>t</sup> programs in communities.

### 3.3.1. ASHA Workers

ASHA workers were interested in supporting the dissemination of drowning reduction programs and were considered suitable for providing training due to their reputation as health workers. ASHA workers were already regularly visiting mothers and children up to the age of 5 years old and could encourage the use of drowning interventions and conduct checks of home-based barriers. However, some participants noted that not all ASHA workers had strong relationships with communities, where their health communications such as community meetings were now largely ignored due to fatigue with repeated advice and instructions (Ref 35 in Table S2).

ASHAs already had some skills in rescue and response. Some ASHAs expressed a desire to learn first aid more comprehensively to perform better in their roles. They were also willing to train others in their communities. However, ASHA workers stated that their work was highly unpredictable as they often responded to calls of women in labour, and so could not provide training and childcare for large blocks of time (Ref 36 in Table S2).

ASHA workers worked on an incentive-based system and expected added paymen<sup>t</sup> for services. (Ref 37 in Table S2).
