Socioeconomic differences in swimming ability among children in Malmö, southern Sweden: Initial results from a community-level intervention

Aims: To investigate to what extent socioeconomic differences in swimming abilities persist among children in the city of Malmö, Sweden, after a community-level swimming intervention programme in public primary schools. Methods: A compulsory swimming education programme was launched in 2014 in second grade (at age 8) in all public primary schools in Malmö, Sweden. Data for the present study on sociodemographic conditions and self-reported swimming ability in fourth grade (age 10) were used for the last birth cohort unexposed (n = 1695) and the first birth cohort exposed (n = 1773) to the intervention. Results: The swimming ability was 78 and 77%, respectively, in the pre- and post-intervention cohorts. Significantly lower self-reported swimming ability was found both pre- and post-intervention among children with support activities in school, with parents born outside Europe, North America and Australia, with manual working, unemployed or studying parents and in children enrolled in schools with socioeconomic index below median. Conclusions: The findings do not suggest that sociodemographic differences in swimming ability have decreased in the first birth cohort exposed to the community-level intervention in Malmö. Striking differences in self-reported swimming ability were noted when the children reached the fourth grade both pre- and post-intervention with marked lower abilities in socially disadvantaged groups. Monitoring of swimming abilities should continue for the present, and similar interventions aimed at reducing inequalities among children. Efforts to increase water comfort at preschool age ought to be considered.

third leading cause of injury-related deaths (including suicide) among children aged 0-14, with on average 5.2 drowning cases each year during 2012-2016 [6]. Drowning rates are linked to socioeconomic status (SES) and country of origin. A case analysis of all children's drowning deaths occurring in Sweden 1998Sweden -2007 showed that children with immigrant background, particularly from the Middle East and Iran, were overrepresented [7].
Children develop individually and might not be ready to learn to swim at the same age. However, there are studies indicating that swimming skills can be taught at preschool ages and swimming lessons in children even as young as 1-4 years old may reduce the risk of drowning [8][9][10]. In a study from Australia the optimal readiness period to learn to swim was identified to be between 5 and 6 years of age [11]. This study also showed that the children learned to swim 10 metres front crawl at approximately the same mean age of 5.5 years, irrespectively of whether they started to take lessons at 2, 3 or 4 years of age. A Danish review identified the ideal age to learn to swim to be 8 years of age, given that the child at the age of 5-7 has participated in water activities and has become comfortable in the water [12]. The Swedish National Agency for Education reports that children who receive swimming education already in preschool or in first grade (at age 7) have a higher probability to pass a swimming ability test in sixth grade compared to children who start their swimming education in second grade or later [13].
In Sweden swimming education is part of the primary school curriculum. In grades 1-3 when the children are 7-9 years old, the activities focus on making the children feel comfortable in the water, and in grades 4-6 on swimming ability. In grades 6 and 9 the children must pass a swimming ability test to pass the physical education and health course [14]. To pass the test the child has to swim continuously for 200 metres, of which at least 50 metres on backstroke, which is in line with the common Nordic swimming ability definition. Subsequently, if a child lacks swimming ability in sixth grade this will have a negative impact on school grades. If the child still cannot swim in ninth grade this will have a negative impact on the chances to enter upper secondary school programmes due to incomplete school grades.
Although swimming is part of the school curriculum low swimming ability among children has been reported from several municipalities in Sweden, especially in areas with a high proportion of families with a low SES and immigrant background [15]. Data from the city of Malmö from 2011 showed that the swimming ability among children in grade 5 varied markedly between areas [16]. As an example, in one of the socioeconomically disadvantaged areas (Rosengård) the swimming ability was 27% compared to 93% in the socioeconomically highest-ranked area (Limhamn-Bunkeflo). In order to decrease these differences in swimming ability, the city of Malmö launched, in 2014, a community-level swimming intervention programme among school children. The aim of the present study was to investigate to what extent the socioeconomic differences in swimming abilities persist in the first cohort exposed to this intervention programme. The study was approved by the regional ethical board in Lund, Sweden, and is part of BlueHealth (www.blue-health2020.eu), a project that has received funding from the European Union's Horizon 2020 research and innovation programme. This project aims to explore the benefits to human health and well-being associated with interacting with blue space (i.e. water) across Europe [17].

Swimming education in schools before intervention
The organization of the swimming education in Sweden is not regulated in the school curriculum or elsewhere, and the education therefore varies between different municipalities and schools. In Malmö, each school administration paid and organized their swimming education separately up to 2014. This led to differences in swimming education delivered to the children depending on priorities made by each school. The overall picture of how the schools organized their swimming education before 2014 is fragmented, as each school contacted swimming halls directly to reserve timeslots in order to arrange swimming lessons.

Community-level intervention
To improve the swimming abilities and make them more equal the city administration of Malmö in 2014 launched a compulsory swimming education programme, Skolsim ('School Swim') in second grade in all public primary schools, comprising about 3000 children each year. This community-level intervention was organized by the Leisure and Recreation Services Department and payed for by the Primary School Department, i.e. not by the schools themselves. The main aim of the programme was to make the children comfortable in the water, which is a prerequisite for learning how to swim. The programme consisted of a three-week swimming education intervention including 15 swimming lessons of 45 min. Each class with about 20-30 children was divided into three groups depending on prior swimming skills. Three educated swimming teachers then provided the appropriate swimming education to each group (one teacher per group).

Data collection
Health examinations are offered to all children in preschool and in grades 4, 7 and 8 in primary schools in the city of Malmö. The main objective of these health examinations is to be supportive for the children and to maintain or improve their wellbeing so that they can participate in and fulfil school tasks. Starting in the school year 2015/2016 anonymized data from the health examinations have been entered into the Pupil Health Database (ELSA) by school nurses in all public schools. Public schools constitute about 75% of all schools in Malmö. Background information from two questionnaires, one administered to the child and one to a parent, is also added into the database after permission from the parents. Data for the present study were retrieved regarding health status, family and socioeconomic situation. Self-reported swimming ability was obtained from the questionnaire statement 'I can swim 200 metres' to which the child answered 'yes' or 'no'. We used the particular socioeconomic index for each school that is routinely calculated by Statistics Sweden based on the following information about the school children: sex, the time of arrival in Sweden if immigrated, education of the parents, whether the parents are receiving financial aid, whether the child lives with one or both parents.

Study population
All children in second grade in a public primary school in Malmö, either in 2013/14 that was the last year before the swimming intervention started, or in 2014/15 that was the first year with the intervention, were eligible for the present study. Data from the health examinations in the fourth grade for these two cohorts were available in the Pupil Health Database (n=2062 and 2180, respectively). Children born outside Sweden (n = 708; 338 in the pre-intervention and 370 in the post intervention group) were excluded from the analysis to minimize the risk of including newly arrived immigrant children who had not taken part in the intervention two years earlier. An additional 62 children (26 pre-intervention and 36 postintervention) were excluded from the analysis because of missing data on swimming ability. Also, four children (three pre-intervention and one post-intervention) attending a school for autistic children were excluded, leaving 3468 children in the final study cohort, 1695 in the pre-intervention and 1773 in the post-intervention cohort. Both cohorts were similar with respect to sociodemographic variables (Table I).

Statistical analysis
We used logistic regression to assess the effect of the intervention on the binary outcome variable selfreported swimming ability (1 = able to swim 200 metres, 0 = not able), both in single-predictor logistic regression models and in multivariable models with adjustment for the sociodemographic variables on the individual and school levels listed in Table 1. Odds ratios (ORs) below one in the logistic regression analyses thus imply decreased odds of reporting ability to swim 200 metres. All variables were categorized in the regression analyses as presented in Table  I. Logistic regression analyses were also conducted in the pre-and post-intervention cohort separately to further investigate if the intervention influenced the associations between the sociodemographic variables and swimming ability. The statistical analyses were performed in STATA 13.1 (StataCorp, USA).

Intervention uptake and attendance
The school year 2014/15 was the first year of the intervention. This year the intervention covered all secondgrade classes in all public primary schools, in total 2906 children; 84% of which attended at least 12 out of 15 (80%) of all swimming lessons. This attendance rate varied between 60 and 97% across schools.

Tested swimming ability in connection with the intervention
No data on tested swimming ability at start of the intervention were available. However, at the end of the three-week intervention the children were tested concerning comfort in water and swimming ability. A child was assessed to be comfortable in the water if he/she could 'streamline' for 5 seconds, meaning to float with straight arms and head under water for 5 seconds. The proportion of children assessed to be comfortable in water was 91%, varying between 59 and 100% across schools. Swimming ability directly after the intervention according to the Nordic standard definition was assessed to be 28% among all children, varying 0-67% across schools.
The swimming ability was lower (20%) among children who had an attendance rate of less than 80% compared to the children who attended 80% or more of the lessons (swimming ability of 30%).

Estimated intervention effect in grade 4
The overall self-reported swimming ability in grade 4, i.e. two years after intervention or control education, was 78% in the pre-intervention cohort and 77% in the post-intervention cohort (Table I). All sociodemographic variables except sex were associated with swimming ability in single-predictor models (Table II, Model type I). No significant difference in swimming ability was found between the pre-and post-intervention cohorts, neither in unadjusted nor in multivariable analyses. Swimming ability in the multivariable analysis was lower among children with support Table I. Sociodemographic variables and self-reported swimming ability among children of grade 4 (age 10-11) in public primary schools in the city of Malmö, stratified according to exposure to control education (pre-intervention cohort) or the community-level swimming intervention Skolsim (post-intervention cohort). All values are presented as percentages, if not stated otherwise.

Discussion
Our evaluation of the community-level swimming programme Skolsim does not suggest that sociodemographic differences in swimming ability have decreased in the first birth cohort exposed to the intervention. Striking differences in self-reported swimming ability were noted when the children reached the fourth grade both pre-and post-intervention with marked lower abilities in socially disadvantaged groups.
A major strength of the present study was the size of two cohorts pre-and post-intervention with data on sociodemographic conditions both at the individual and the school level. The intervention uptake was relatively high, and so was participation in the health examinations entered in the Pupil Health Database, which limit the risk of selection bias in the presented results. An important limitation was that only data on self-reported, i.e. not tested or documented in e.g. school grades, swimming abilities are currently available. Previous studies have found that young adults have a tendency to underestimate their distance swimming skills when compared to actual swimming competence [18,19], but it is unclear to what extent these results can be generalized to our cohorts. It is however reasonable to assume that any misclassification of the swimming ability was non-differential with respect to the intervention, resulting in bias towards the null. Another limitation of the study was that information about how the swimming education was organized before the intervention is limited, as this was the responsibility of each school. We therefore lack measures on the number of additional swimming lessons that were given with the intervention, and also do not know to what extent the content of lessons were changed.
The present study showed that country of birth and occupation of the parents were strongly related to swimming ability among the children. The results further suggested that swimming ability was lower among children enrolled in or in need of support activities in school. These findings can be used to target groups of children with greater needs of support in their swimming training. Additional free swimming lessons mainly targeting these groups of children at different ages from preschool and onwards might reduce differences in swimming ability. Markedly lower swimming ability was seen among children enrolled at the schools with socioeconomic index below median, also after controlling for sociodemographic variables on the individual level. This finding suggests that the context and the general school environment is important also for the success of swimming education and that more directed interventions towards the schools with lower socioeconomic index could be beneficial.
Being comfortable in the water is a prerequisite to learning how to swim, and water comfort can already be taught at preschool age [8][9][10]. Starting swimming lessons at an earlier age is likely not only to improve overall swimming abilities but would also reduce the number of years with elevated risk of drowning. In 2017 the city of Gothenburg, Sweden, launched a large swimming education programme (Simlyftet) to reduce inequalities in swimming ability [20]. Based on the knowledge that swimming education is most effective at an early age, swimming lessons in the Gothenburg intervention have been introduced for all children already in preschool (6 years) as well as in grade 2. Further, all children who cannot swim in grades 5-9 in Gothenburg are also offered free swimming lessons. It is conceivable that an important reason why the Skolsim intervention in Malmö has so far not achieved its goals is that it is currently lacking an effort to achieve water comfort in children already in preschool. If water comfort is achieved at an earlier age, the aim of swimming lessons provided by the school in the second grade could be more directed towards learning how to swim.
The swimming intervention in Malmö is ongoing, and it is therefore highly relevant to continue to monitor intervention effects, at age 10 but also at age 12 when the children receive grades in physical education and health course (where swimming ability is prerequisite in order to pass the course). Structured evaluations of other on-going interventions, such as the comprehensive swimming education programme in Gothenburg, Sweden [20], should also be given priority. Lastly, it is also of interest to investigate correlates of swimming difficulties with respect to health and school situation, and to conduct more in-depth studies among both children and parents in order to better understand how improved and more equal swimming abilities can be achieved.

conflict of interest
The authors declare that there is no conflict of interest.

Funding
This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 666773. The project has also received financial support from the Swedish Research Council (grant no. 340-2013-5474).