Discussion

This exploratory study was done to determine the frequency and patterns of self-reported rescues in three sex-matched and age-matched groups of Bangladeshi children 6–14 years old: (1) those who graduated from SwimSafe, a basic swimming and safer rescue programme; (2) those who were not SwimSafe graduates but learned to swim on their own (natural swimmers); and (3) those who were not SwimSafe graduates and had not yet learned to swim. Slightly less (11 717) than the target of 12 000 children in all three groups responded to the survey over both time periods. Children in group 3 (non-SwimSafe graduates and non-swimmers) only reported two rescues, which was insufficient to establish a meaningful rescue rate. For the other two groups, findings showed large numbers of rescues reported by both SwimSafe children and natural swimmers. Most reported rescues for both groups took place in ponds and ditches with rescuers entering the water. The methods and frequencies of rescue activities did not differ between SwimSafe graduates and natural swimmers.

The observation of only two rescues conducted by the non-swimming group and the similar rescue rates between the two groups able to swim suggests the ability to swim or confidence in the water environment was the common enabling factor for rescues to be attempted.

In this study, successful rescues were reported by young children at high rates in both 1-month recall and annual recall periods. The reported rates are much higher than those seen with bystander rescues in high-income countries (Justin Scarr, personal communication, Drowning Commissioner, International Life Saving Federation and CEO, Royal Life Saving Society, Australia, August 2013). The timing of the samples likely explains the large difference between the 1-month rate when annualised and the rate measured over 1 year. Times were chosen to coincide with the beginning and peak of the period when children play or swim in water. One-month rates during this 5-month period would be expected to be higher than those during the rest of the year when temperatures are cold and pond levels are low. The BHIS national survey showed drowning rates are markedly affected by seasonality. Logically, rescue rates would follow this pattern since most rescues were for bathing, which is less seasonal than swimming but also decreases in cold weather.

In rural Bangladesh, the necessarily frequent use of ponds for essential daily tasks, coupled with large family sizes and busy morning routines for mothers, means that supervision of very young children in and around water is often delegated to an older sibling. Many of the findings likely result from this cultural norm of sibling supervision. These include (1) most reported drowning victims entered the water without adult supervision; (2) in all reported rescue events, the rescuer was a child and an adult did not provide assistance nor was involved in any other way; (3) for children able to swim and reporting rescuing other children, victims were very young children (mean age 4.1 years) with an age difference between rescuer and reported victim of about 6 years (mean 5.9 years); and (4) most rescues (53.8%) occurred from 12:00 to 13:59 h, a time where most rural mothers are busiest with household chores.

The lack of a difference in frequency of reported rescues between SwimSafe graduates and children who learned swimming from their peers or parents may also be related to sibling supervision. In each group, very young children often got into trouble while bathing, requiring the supervising older sibling to conduct a rescue.

Other cultural characteristics of rural Bangladesh may also have resulted in another finding. Despite being taught safer land-based rescue techniques and to enter the water only as a last resort, SwimSafe children were equally likely to enter water to conduct a rescue. Rescuers reported a lack of sticks and branches at the pond or ditch to allow a reach rescue. In rural Bangladesh, sticks or branches are rarely available having been collected for firewood. Their absence likely left water entry the only available option.

Despite the surveys being timed to occur at periods where recreational water exposure (swimming and play in water) was maximal, only one in four (25.2%) reported rescues occurred when a victim was engaged in recreation (swimming 0.5% or playing 24.7%). Almost three out of four reported rescues (73.2%) occurred when a very young child entered the water for bathing (53.8%), fell in unintentionally (15.6%) or entered to retrieve something that had fallen in (15.6%). In large part due to the cultural norm of sibling supervision of very young children, bathing appears to be among the highest risk exposures to water for young children in rural Bangladesh.

The relationship between adult supervision and an increased risk of drowning has been shown to influence other types of unintentional child injury in high income countries (HICs) and LMICs.6 ,7 In HICs, professional services are often used to mitigate this risk (eg, provided by a teacher or lifeguard). Importantly, awareness needs to be raised in the community regarding the risk associated with young children bathing in bodies of water without adequate supervision. Further research is underway to characterise community attitudes regarding the age necessary for an older sibling to effectively supervise a young child around water.

The study suggests in-water rescue techniques should be added to the already included land-based techniques in the SwimSafe programme. Nine out of ten rescues in this study (90.2%) were in-water rescues where the child rescuer had contact with the child victim. Rescue equipment and supervising adults were not present, and the victim was either alone or accompanied by peers or a sibling. Cultural norms such as sibling supervision are resistant to change and doing so will likely require a substantial length of time.

In HIC settings, adult drowning while rescuing children has been described as ‘aquatic victim instead of rescuer syndrome’ (AVIR).8 This has been one factor in stimulating development of rescue techniques, policies and education programmes aimed at reducing the risk to the rescuer, especially when there is potential for direct contact with the victim in the water. Safety legislation in HIC settings usually requires safety equipment and professional supervision in regularly used public swimming areas. This has markedly lessened the need for a bystander rescuer to enter the water to conduct a contact rescue. Learning rescue techniques such as contact towing used to be common in basic lifesaving courses. Now many HIC lifesaving organisations restrict teaching of contact tows to more advanced learners.9 As a result, children in HICs are usually only taught land-based rescue techniques in entry-level courses.

The drowning environment in Bangladesh is very different from that in HIC settings. In this environment, the study found that children conduct in-water rescues that involve contact even when having received training in safer land-based techniques. This apparently results from both the nature of the drowning environment and a lack of reach extension devices and other safety equipment rather than an inability to learn the safer land-based technique.

Given that in-water rescue is the most commonly used technique by children in rural Bangladesh, risks associated with it should be mitigated. Teaching this may be viewed as having the potential to increase risk to the child rescuer. However, the finding that it is already the norm for conducting rescues argues for a risk reduction approach. Such techniques may be quite simple. An example would be for the rescuer to take off their t-shirt (the usual clothing worn by children) and extend that to the victim, avoiding direct contact. Furthermore, only about half of in-water rescues required swimming to make contact with the victim. Most rescues were conducted close to the bank where water is usually shallow and the older child rescuer was tall enough to reach the younger victim without swimming. Logically, these would entail less risk to the rescuer due to the ability to stand as well as the large difference in physical development found between children in early childhood and later childhood, which was the difference seen between the rescuer and the victim in this study.

This study has several limitations. One is rescue was self-reported by children without an ability to confirm whether a victim was truly at risk of fatal drowning. Given that almost a tenth (8.1%) continued to swim or play after being rescued, over-reporting may have occurred. However, almost three-quarters (73.7%) of rescue victims discontinued what they were doing, 2.2% of victims were unconscious and there was one fatal drowning. These numbers are consistent with real drowning risk.

Using two different samples of children is another limitation. The design does not allow a precise determination of either the 1-month or the annual rescue rates as the two surveys represent snapshots of different samples of children at two points of time, purposefully chosen to maximise the likelihood of encountering recreational play in water or swimming. Frequent, serial monitoring of a large cohort is required for that which is currently in process. Despite the inability of this study to precisely characterise rescue incidence rates, children are reporting water-based rescues of other children at substantial rates, even after allowing for potential over-reporting.

While rescue represents a potential risk, it also represents a potential newly recognised intervention for reducing early child drowning. Given the culture and the nature of present child supervision practices in rural Bangladesh, it is unlikely that it can be prevented through messages to parents regarding the need for increased supervision. Development of more effective risk knowledge and adult supervision interventions is an urgent priority. Until then, since the large majority of reported rescues are occurring as water rescues with direct contact, children should be taught safe water rescue techniques to minimise their risk and making it as safe as possible.