Article Text

Download PDFPDF

Does teaching children to swim increase exposure to water or risk-taking when in the water? Emerging evidence from Bangladesh
  1. Tom Stefan Mecrow1,
  2. Michael Linnan2,
  3. Aminur Rahman1,
  4. Justin Scarr3,
  5. Saidur Rahman Mashreky4,
  6. Abu Talab4,
  7. A K M Fazlur Rahman4
  1. 1International Drowning Research Centre—Bangladesh (IDRC-B), Dhaka, Bangladesh
  2. 2The Alliance for Safe Children (TASC), Atlanta, USA
  3. 3Royal Life Saving Society Australia (RLSSA), Sydney, Australia
  4. 4Centre for Injury Prevention Research, Bangladesh (CIPRB), Dhaka, Bangladesh
  1. Correspondence to Dr Aminur Rahman, International Drowning Research Centre—Bangladesh (IDRC-B), House—B 162, Road—23, New DOHS, Mohakhali, Dhaka 1206, Bangladesh; aminur{at}


Background SwimSafe, a basic swimming and safe rescue curriculum, has been taught to large numbers of children in Bangladesh. Teaching swimming potentially increases risk if it increases water exposure or high-risk practices in water. This study compares water exposure and risk practices for SwimSafe graduates (SS) with children who learned swimming naturally.

Methods Interviewers obtained detailed water exposure histories for the preceding 48 h from 3936 SS aged 6–14 and 3952 age-matched and sex-matched children who had learned swimming naturally. Frequencies of water exposure and water entries for swimming or playing were compared.

Results There were 9741 entries into water among the 7046 participants in the 48 h prior to interview. About one-third (31.2%) had no water entries, one-tenth (10.5%) entered once, half (49.2%) entered twice and a tenth (9.1%) entered three or more times. Proportions of children in each group were similar. About 99.5% of both groups only entered the water for bathing. For those entering to swim or play, the mean number of entries was similar (SS 1.63, natural swimmer (NS) 1.36, p=0.40). Swimming or playing alone in the water was rare (1 SS, 0 NS).

Conclusions Most water exposure for children is for bathing. Less than 1% swam or played in the water during the 48 h recall period (0.6% SS, 0.4% NS). Learning swimming in SwimSafe did not increase water exposure nor did it increase water entry for playing or swimming compared with children who learned to swim naturally.

  • swim learning
  • recreational water exposure
  • SwimSafe
  • child drowning
  • drowning prevention
  • Bangladesh

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.



Drowning is a leading cause of child death in both high-income countries (HICs) and low-income and middle-income countries (LMICs).1 While a leading cause of death in both, the rates of drowning are much higher in LMICs. For example, the 2003 Bangladesh Health and Injury Survey (BHIS) found drowning was the leading single cause of child death from 1 to 17 years of age with a rate of 27.3/100 000 children 1–17 years.2 This is 10–20 times the rate for HICs.3 ,4

Until recently, evidence did not exist that swimming ability protected children in LMICs from drowning. In 2005, the Centre for Injury Prevention and Research Bangladesh (CIPRB), the Bangladesh Swimming Federation, The Alliance for Safe Children and the Royal Life Saving Society Australia (RLSSA) developed SwimSafe. SwimSafe incorporates basic swimming (25 m distance) with safe rescue and basic water safety skills for children 4–12 years old.5 It was shown to reduce fatal drowning in children in rural Bangladesh by 93% (RR=0.072, 95% CI 0.017 to 0.307) in a large cohort trial between 2006 and 2011.6 The demonstration that SwimSafe protects against drowning raises the possibility of it being one component of a national drowning prevention programme.

The population aged 4–14 years in Bangladesh, who are at highest risk of drowning and are likely participants in a swimming intervention such as SwimSafe, number well over 40 million children.7 A programme potentially covering this number of children requires evidence of safety before moving to full scale. One potential safety issue is whether children who participate and learn basic swimming skills then increase their overall exposure to water. Another is whether children taught basic swimming may increase risk-taking when in water (eg, swimming or playing in water alone), even if overall water exposures are not increased. In either case, this may increase the risk of drowning. While a theoretical risk, the actual risk is not yet known.8 With over 40 million children potentially participating, if only 1% exhibit these behaviours, hypothetically it could place over 400 000 children at increased risk. This study was undertaken to provide data on frequencies of exposure to water and risk-taking behaviours in recreational water exposure of SwimSafe graduates (SS) compared with children who learned swimming naturally without exposure to SwimSafe.


The study used repeated cross-sectional surveys to compare exposure to water and risk in recreational water exposure among rural graduates of SwimSafe with age-matched and sex-matched children who learned swimming naturally with no exposure to SwimSafe. Increased risk in recreational water exposure was defined as (1) increased frequency of swimming or playing in water or (2) swimming or playing alone in water.

Study period

Two surveys were conducted at different times of the season when swimming is a normal activity for children. The first was in March–April 2011 when ponds and rivers fill up and the weather becomes warm and frequent swimming begins. The second was in September–October 2011, late in the swimming season when swimming and water play was reported to be highest by rural residents.


In both surveys, 2000 SS aged 5–14 were selected using simple random sampling from the approximately 200 000 graduates of the rural SwimSafe programme. Each SS was matched with a control child who had learned swimming naturally and had no exposure to SwimSafe. Matching was done on area of selection, age and sex. A total of 7046 paired SS and natural swimmers (NS) had usable data.

Data collection

A pre-tested questionnaire was used to collect data regarding entry into water during the previous 48 h. The 48 h recall period was chosen to maximise detailed recollection of each episode in order to properly assign risk of a particular episode. Trained interviewers conducted face-to-face interviews with the children following informed consent from parents. Data were collected on the number of water entries in the previous 48 h; the type of water body for each entry; the main reason for each water entry and the person(s) accompanying the child on each entry.

Data entry and analysis

Data analysis was done with Stata V.11. Conditional logistic regression was used to examine the matched pair data for significant differences in proportions between the groups, and the matched pairs t test was used to evaluate differences in means between groups.



There were 7046 children with completed interviews from the two sampling periods from the rural subdistrict of Raiganj. The proportion of male and female participants was similar in SS and NS children in each age group (table 1).

Table 1

Distribution of children by method of learning swimming, age and sex

Number of all entries into water

There were a total of 9608 entries into water from the 7046 participants in the 48 h prior to the interview. About one-third (31.2%) had no water entries, one-tenth (10.5%) entered once, half (49.2%) entered twice and a tenth (9.1%) entered three or more times. Proportions of children in each group were similar in frequency of water entry (table 2).

Table 2

Distribution of SwimSafe (SS) and natural swimmers (NS) by number of entries

Reason for entry into water

About 99% of both groups (99.3% SS, 99.6% NS) only entered the water for bathing and <1% entered the water for playing or swimming (0.6% SS, 0.4% NS) (table 3).

Table 3

Reasons for each entry into water by method of learning swimming

The results that follow only pertain to water entry for playing and/or swimming and exclude other exposures that were almost entirely bathing. Over the 48 h recall period, the number of children in both groups who played or swam in the water was quite small. There were 19 SS children who played or swam in the water, with 16 entries to play and 15 entries to swim. There were 11 NS children who played or swam in the water, with 12 entries to play and 3 to swim.

Number of entries into water to swim or play by age and sex

The number of entries in the paired samples was too few for reliable statistical calculation for differences between the SS and NS groups (table 4).

Table 4

Mean number of recreational water exposures by method of learning swimming and age

Location of entry

Most recreational entries (84.8%) took place in ponds (28 entries, 60.9%) and ditches, low areas normally dry that fill up with large amounts of water in the rainy season (11 entries, 23.9%). Four entries were in rivers (8.7%) and two entries were in canals (4.3%).

Accompanying person for recreational water entry

The majority of recreational water entries for both groups were accompanied by friend (90.3% SS and 73.3% NS) as seen in table 5. There were too few recreational water entries done alone to determine statistical difference between SS and NS groups.

Table 5

Accompanying person for each recreational water exposure


This study examined the frequency of water exposure, exposure to water for recreation and risk-taking behaviours while swimming or playing in water for young children in rural Bangladesh. It compared these in children who had learned to swim in the SwimSafe programme with similar children who had learned swimming naturally. Almost all water exposure for children in both groups came from bathing (hygiene-related). Less than 1% of either cohort swam or played in the water during the 48 h period studied. There were no differences in overall water exposure, recreational water exposure for playing or swimming, and the high-risk behaviour of playing or swimming alone.

The study shows recreational water exposure is a very small fraction of all children's daily water exposure in the study population. While commonly thought that children in rural areas of Bangladesh spend large amounts of time playing, most children 4 years and older have responsibilities for domestic chores, assisting in the fields, helping with animals, supervising siblings and often enrolling in school. As a result, they do not have a large amount of unstructured time to play. The survey results are consistent with this, showing water entry for recreation is not a daily event for most, even when the weather is warm and ponds and other water bodies are full.

The need to enter water for essential daily activities has meant that the majority of children in Bangladesh learn swimming informally, often from a young age. Thus, Bangladeshi children do have water exposure through playing and swimming, although at much lower rates than water exposure for bathing. Learning swimming skills informally is a cultural norm for rural children. An anthropological study on natural acquisition of swimming ability in rural Bangladesh showed it is a structured process that begins in infancy and continues through early childhood. Some children are able to swim from 3 years of age.9 The BHIS national survey found that a large majority of children, both rural and urban, have naturally acquired swimming ability by the end of childhood.2 Over three-quarters of children in Bangladesh learn to swim by age 17 and engage in recreational water play or swimming throughout the length of childhood.

The study examined whether recreational water exposure that does occur does so with increased risk. Solitary, unaccompanied play or swimming in water presents such an increased risk. A child who gets into trouble while alone has no one to provide help to prevent them from drowning. One goal of SwimSafe was to educate caretakers and children to the risk of this practice and reduce it. This rarely occurred for either group. Future research will be needed to see whether this is due to the limited sample size.

The methodology employed in this study to examine risk and water exposure should be considered in the context of the limited but growing international literature on risk exposure measures for drowning.10 The alternative measure of time in water, for example, relies on observational methods that may be suitable for well-resourced studies, particularly in high-income settings where fatal drowning rates are low. In such settings, it is useful to precisely discriminate high-risk groups as they may vary greatly from the overall population risk. In LMICs, especially in rural settings, proximity to water is universal across virtually all population subgroups. In such high-exposure environments with their subsequent high rates of drowning and paucity of public health resources, the overall population risk is suitable to inform policy and develop appropriate prevention measures.

Observational methods were not practical in this study. The self-reported nature of the data from young children is a limitation. Children were asked to report the main reason for water entry, and it is possible that multiple behaviours occurred for each entry. Since recreational water exposure infrequently involved the caretaker, it was not possible to validate the self-report of the children. Similarly, the length of each recreational exposure was unable to be determined as young children do not have watches and the study involved self-report and not observation by others. Finally, the self-reported swimming ability for the naturally acquired swimming cohort was not tested through observation. The reporting methodology has been standardised.11 It has been validated by comparison of stated swimming ability and measured swimming ability in the same study populations (unpublished data, summary available on request).

An important limitation in this study is power. It is likely that if increased water exposure or risk-taking while in the water occurs as a result of learning to swim, it only occurs at low levels. Indirect evidence for this can be seen in the nested case–control study in the BHIS national survey. The study showed lack of swimming ability was associated with higher risk of drowning (OR=4.5, 95% CI 1.25 to 19.4).2 The lower drowning risk in children able to swim suggests learning to swim does not lead to large increases in high-risk behaviours that result in drowning. If such behaviours were frequent, drowning rates would be higher among children with swimming ability, reflecting the increased risk. At the level of discrimination in this study with pooled samples of 7046 children at two different points in the swimming season, children learning to swim through SwimSafe do not appear to have increased exposure to water or high-risk behaviours when in the water. However, that does not exclude these could occur in a small subgroup of children. Continued sampling among the expanded SwimSafe cohort and controls will be necessary to provide the levels of confidence needed when hundreds of thousands or millions of children are to be taught.

Another constraint is the short duration of recall. The 48 h recall period was chosen so that a child would be able to correctly recall detailed characteristics of each water exposure when guided by a trained interviewer. It is necessary to interview the child as many water exposures take place outside the knowledge of the caretaker. In fact, this study showed that many were solitary and caretaker interviews would not contain these. The short duration of recall provided detailed information on exposures that occurred, but may have reduced the likelihood of hearing about relatively infrequent exposures or risk-taking. While the study attempted to address this by having two different times of sampling, it is possible relatively infrequent behaviours would need more or longer sampling intervals before a high level of confidence could be reached.

This study does not address the question of whether children who have not learned to swim have greater water exposure or engage in risk behaviours such as playing alone in the water compared with children who have learned to swim. This is the subject of research in process and will be addressed in a future publication.

Importantly as well, this study only looked at rural children in Bangladesh. Children in other cultures in different LMICs may have different risk behaviour profiles. SwimSafe is being implemented in Thailand and Vietnam, and similar risk studies will be needed in these settings in order to address this question.

Despite these limitations, it appears that in Bangladesh participation in the SwimSafe programme does not result in an increased risk of water exposure or risk-taking when in the water for recreational purposes compared with children who learned swimming naturally. As scale increases further, continued vigilance with expanded surveys will be necessary to ensure that participation does not place children at increased risk as an unintended adverse consequence of the programme.

What is already known on the subject

  • Drowning is a leading cause of death in children in rural Bangladesh.

  • High rates of fatal drowning relate to the daily exposure to water bodies in close proximity to the child's home.

  • SwimSafe, a basic swimming and safe rescue curriculum, is effective in preventing child drowning in rural Bangladesh.

What this study adds

  • The most common water exposure for children in rural Bangladesh is for bathing. Recreation such as swimming or playing in water is much less frequent.

  • Children who graduate from SwimSafe do not have increased exposure to water compared with children who learn to swim naturally.

  • Children who graduate from SwimSafe swim and play in water at the same rates as children who learn to swim naturally.


We thank the Australian Government Agency for International Development, Royal Life Saving Society Australia and The Alliance for Safe Children for their support in the study. We gratefully acknowledge the local government officials, community leaders and the community people of Raiganj who participated in the study. We also acknowledge with thanks reviewer suggestions used to improve the manuscript, particularly Joan Ozanne-Smith, Julie Gilchrist, Steve Beerman and Joost Bierens.



  • Contributors TSM, AR and SRM participated in the design, implementation and supervision of field work and analysis, and wrote the initial draft of the paper. SRM and AT were involved in instruments development, supervised field work, participated in analysis of data and contributed to the manuscript. JS advised on project design and implementation, and contributed to the manuscript. ML and AKMFR conceived the study, participated in the design, supervised throughout and contributed to the manuscript. ML and AKMFR are guarantors.

  • Funding Australian Government Agency for International Development (AusAID).

  • Competing interests None.

  • Ethics approval Ethical Review Committee, Centre for Injury Prevention and Research, Bangladesh (CIPRB).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The data of the study are stored in the Statistics Department of the IDRC-B. AR and AT have access to the data set.