Introduction

Injuries are a leading cause of death and disability for children and young people worldwide and a significant public health challenge. The landmark World Report on Child Injury Prevention (Peden 2008) highlighted low levels of awareness and political commitment as barriers towards preventing child injury. Whilst ending preventable deaths for children under five is a stated priority in the Sustainable Development Agenda (United Nations 2015), contemporary World Health Organization (WHO) data recorded more than 500,000 injury-related deaths in 2019 in children under 15 years, including by fatal drowning (World Health Organization 2019). Children and young people are overrepresented in fatal and non-fatal drowning statistics, and children under five years are at particular risk (Tyler et al. 2017; Franklin et al. 2020; Peden et al. 2021). Whilst drowning affects children globally, there are differences in risk factors and responses to prevent drowning as well as drowning burden (Tyler et al. 2017; Peden et al. 2021). The social, ecological and commercial determinants of health may influence drowning distribution worldwide (Leavy et al. 2023a). For example, 2019 data reported more than 40,000 children under 15 years fatally drowned in low- and middle-income countries (LMIC), compared with 1322 drowning deaths in higher-income countries (HIC) (World Health Organization 2019). Official figures are a likely underrepresentation of the actual burden as official data excludes climate-event-related drowning.

The publication of the 2014 Global Report on Drowning reinforced drowning as an underserved public health issue and called for broad prevention efforts encompassing community action, policy and legislation and research and integration with other public health agendas (World Health Organization 2014). The associated Implementation Guide recommended advancing drowning prevention action via data and well-designed studies (World Health Organization 2017). Despite these calls, to our knowledge, there are only two systematic reviews of public health interventions to prevent child drowning, both published by Australian research teams in 2015 (Wallis et al. 2015) and 2016 (Leavy et al. 2016). Those reviews and more recent systematic reviews on child (Taylor et al. 2020) and adult drowning more broadly (Peden et al. 2018; Cenderadewi et al. 2020) highlight a preponderance of descriptive and epidemiological studies that articulate the problem and present narrative commentary with possible solutions. Well-designed public health interventions with recommendations for scale-up remain largely absent from the literature.

Promisingly, more recent literature points to increasing publication of drowning prevention research (Scarr and Jagnoor 2021), likely bolstered by recent global initiatives, including the United Nations (UN) General Assembly Resolution on Global Drowning Prevention (United Nations 2021), 76th World Health Assembly Resolution on Accelerating Action for Global Drowning Prevention (World Health Organization 2023), establishment of the Global Alliance for Drowning Prevention and planning for a Global Status Report on Drowning Prevention and Global Strategy for Drowning Prevention. These initiatives have provided the impetus for greater research and attention to factors influencing effective drowning prevention strategies (Leavy et al. 2023a).

In 2016, the research team systematically reviewed the peer-reviewed literature to examine child-focused public health drowning prevention interventions (Leavy et al. 2016). That review revealed 15 high-, middle- and low-income setting studies. Interventions were skewed towards education and information with inconsistent behavioural theory and formative evaluation application. The authors called for using contemporary, evidence-informed health promotion planning and evaluation processes to develop multi-strategic interventions guided by behavioural theory. Fifteen years since the publication of the World Report on Child Injury Prevention (Peden 2008) and a decade on from the World Drowning Report (World Health Organization 2014), it is critical to determine what lessons have been learnt about what works, why, and for whom and whether these lessons are effectively translated to the benefit of the global drowning prevention community. This updated review identifies and assesses public health interventions to prevent drowning amongst children. It aimed to assess the global scope of literature and examine the contributions to the evidence since the last review was published.

Methods

Methods are summarised below and used the process established in previous reviews (Leavy et al. 2016, 2023b) and protocols (Crawford et al. 2014). The review followed the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines (Page et al. 2021).

Inclusion criteria

Included articles were peer-reviewed, published in English between 2011 and 2023 and described evaluated primary prevention interventions to prevent child drowning consistent with health promotion intervention levels (Nutbeam and Muscat 2021): individual, and/or group and/or population. Included papers were categorised by health promotion strategy (behavioural – defined as actions that individuals take concerning drowning prevention (Nutbeam and Muscat 2021); or socio-ecological – defined as part of the social, physical and policy environment (Stokols 1996)), and activity type (e.g. education, environmental, regulatory). Consistent with the UNICEF Convention on the Rights of the Child (UNICEF 1989), we defined a child as zero to 18 years. Interventions targeting children and adults were included where the impacts on or for children were clearly articulated. Articles that described clinical or therapy trials, medical interventions or conditions, risk factors, formative programme development, protocols, and technical testing of technology used in interventions were excluded.

Outcome measures

Outcome measures were drowning rates, changes in water safety behaviour or behavioural intention, drowning awareness, knowledge, attitudes, water safety policy and legislation, changes to the environment and/or water safety skills.

Search strategy

Consistent with the previous review (Leavy et al. 2016), 15 databases were searched: PubMed, JSTOR, CINAHL, EMBASE, ERIC, ProQuest, PsycINFO, ScienceDirect, Scopus, Global Health, Web of Science, Current Contents, Wiley Online Library, Medline and Sport Discus. The following keywords and associated MESH terms were used:

  1. 1.

    drown* adjacent to (prevent* or safety) within 3 words [MESH: Drowning/ pc [Prevention & Control]

  2. 2.

    interven* OR evaluat* OR ‘best practice’ OR ‘good practice’ OR ‘best practise’ OR ‘good practise’ OR ‘health promot*’ OR ‘public health’ OR polic* OR program* OR research OR prevent* OR education OR curriculum [MESH: Health Education/ and Preventive Health Services/]

Article selection, screening and quality appraisal

Articles were exported into Endnote X9 (Clarivate 2020), duplicates were removed, and the remaining articles were downloaded into Rayyan.AI (Ouzzani et al. 2016) for collaborative screening. Titles and abstracts were screened against the inclusion criteria by two research assistants. Excluded publications were reviewed, and one research assistant cross-checked 100% of the included studies. Full-text articles meeting the inclusion criteria were retrieved and reviewed by three researchers, reviewer consensus resolved any discrepancies. The reference lists of included articles were reviewed to confirm that relevant studies had not been overlooked. Two research assistants conducted quality appraisal using the MetaQAT framework for relevancy, reliability, validity and applicability (Rosella et al. 2016). A final sample of n = 43 studies was included. The complete search process is presented in Fig. 1 below.

Fig. 1
figure 1

Systematic review flow chart (Page et al. 2021)

Full size image

Data extraction, analysis and synthesis

Data were extracted into a standardised data extraction form. Results present pooled data on age, gender and key findings. Results are presented in a descriptive-numerical format with an accompanying narrative.

Results

Setting, duration and evaluation design

The review included 43 studies (Table 1) from 15 countries. The majority (n = 27) (Girasek 2011; Lawson et al. 2012; Moran et al. 2012; Wintemute et al. 2013; Franklin et al. 2015; Franklin and Peden 2017; Ramos and Anderson 2017; Wilks et al. 2017; Glassman et al. 2018; Matthews and Franklin 2018; Olaisen et al. 2018; Barcala-Furelos et al. 2019; Sandomierski et al. 2019; Button et al. 2020; Forde et al. 2020; Karatrantou et al. 2020; McCallin et al. 2020; Quan et al. 2020; Araiza-Alba et al. 2021; Petrass et al. 2021; Smith et al. 2021; Willcox-Pidgeon et al. 2021; Calverley et al. 2022; Casten et al. 2022; Love-Smith et al. 2022; Tipton et al. 2022; Zhang et al. 2022) were from HICs; eight (Solomon et al. 2013; Shen et al. 2015; Silva et al. 2016; Turgut et al. 2016; Zhu et al. 2017; Farizan et al. 2020a, b; Moura et al. 2021) were from upper middle income countries (UMICs), and the remainder (n = 8) (Ramos et al. 2018; Alonge et al. 2020; Hossain et al. 2020a,b; Foulds et al. 2021; Ashraf et al. 2022; Bhuiyan et al. 2022; Gupta et al. 2022; Khan et al. 2023) from LMICs. The United States (US) (n = 10) (Girasek 2011; Lawson et al. 2012; Wintemute et al. 2013; Ramos and Anderson 2017; Glassman et al. 2018; Olaisen et al. 2018; Forde et al. 2020; McCallin et al. 2020; Quan et al. 2020; Love-Smith et al. 2022); Australia (n = 9) (Franklin et al. 2015; Franklin and Peden 2017; Wilks et al. 2017; Matthews and Franklin 2018; Araiza-Alba et al. 2021; Petrass et al. 2021; Willcox-Pidgeon et al. 2021; Calverley et al. 2022; Casten et al. 2022) and Bangladesh (n = 6) (Alonge et al. 2020; Hossain et al. 2020a, b; Foulds et al. 2021; Ashraf et al. 2022; Bhuiyan et al. 2022; Gupta et al. 2022) were the geographical locations most commonly represented (see Fig. 2). Two studies were reported from Brazil (Silva et al. 2016; Moura et al. 2021); Canada (Sandomierski et al. 2019; Smith et al. 2021); China (Shen et al. 2015; Zhu et al. 2017); Malaysia (Farizan et al. 2020a, b; New Zealand (Moran et al. 2012; Button et al. 2020) and the United Kingdom (UK) (Tipton et al. 2022; Zhang et al. 2022). One study was reported from Greece (Karatrantou et al. 2020), Grenada (Solomon et al. 2013), Pakistan (Khan et al. 2023), Spain (Barcala-Furelos et al. 2019), Turkey (Turgut et al. 2016) and Vietnam (Ramos et al. 2018), respectively. The 43 papers were published across 27 journals, most commonly the International Journal of Aquatic Research and Education (n = 8), the Health Promotion Journal of Australia (n = 3), the International Journal of Environmental Research and Public Health (n = 3) and the International Journal of Injury Control and Safety Promotion (n = 3).

Table 1 Review of child drowning prevention interventions (n = 43)
Full size table
Fig. 2
figure 2

Global distribution of included studies (n = 43)

Full size image

Intervention study designs included three mixed-methods studies (Forde et al. 2020, Foulds et al. 2021, Gupta et al. 2022), four qualitative studies (including three observational studies) (Wintemute et al. 2013, Franklin et al. 2015, Matthews and Franklin 2018) and a content analysis (Smith et al. 2021). The remaining study designs (n = 36) were quantitative. Just under half were pre–post studies (n = 20) (Lawson et al. 2012; Solomon et al. 2013; Turgut et al. 2016; Franklin and Peden 2017; Wilks et al. 2017; Olaisen et al. 2018; Ramos et al. 2018; Sandomierski et al. 2019; Alonge et al. 2020; Button et al. 2020; McCallin et al. 2020; Quan et al. 2020; Moura et al. 2021; Petrass et al. 2021; Calverley et al. 2022; Love-Smith et al. 2022; Tipton et al. 2022; Zhang et al. 2022; Khan et al. 2023). The remaining designs were controlled trials (Girasek 2011; Moran et al. 2012; Franklin et al. 2015; Shen et al. 2015; Zhu et al. 2017; Hossain et al. 2020a,b); quasi-experimental (Silva et al. 2016; Glassman et al. 2018; Barcala-Furelos et al. 2019; Farizan et al. 2020a, b; Araiza-Alba et al. 2021); cross-sectional (Ramos and Anderson 2017; Willcox-Pidgeon et al. 2021; Casten et al. 2022) and longitudinal (Ashraf et al. 2022; Bhuiyan et al. 2022) studies. There were two female-gender-specific studies (Girasek 2011; Silva et al. 2016), including a US study by Girasek (2011) assessing the impact of a video on pool-owning pregnant women’s motivation to participate in cardiopulmonary resuscitation. Twenty studies targeted only children (Wintemute et al. 2013; Franklin et al. 2015; Shen et al. 2015; Turgut et al. 2016; Ramos and Anderson 2017; Wilks et al. 2017; Zhu et al. 2017; Olaisen et al. 2018; Ramos et al. 2018; Barcala-Furelos et al. 2019; Alonge et al. 2020; Button et al. 2020; Forde et al. 2020; Karatrantou et al. 2020; Araiza-Alba et al. 2021; Moura et al. 2021; Willcox-Pidgeon et al. 2021; Calverley et al. 2022; Tipton et al. 2022; Zhang et al. 2022); 17 studies targeted parents/caregivers only (Girasek 2011; Moran et al. 2012; Silva et al. 2016; Franklin and Peden 2017; Glassman et al. 2018; Sandomierski et al. 2019; Farizan et al. 2020b; Hossain et al. 2020a, b; McCallin et al. 2020; Quan et al. 2020; Foulds et al. 2021; Smith et al. 2021; Ashraf et al. 2022; Bhuiyan et al. 2022; Casten et al. 2022; Love-Smith et al. 2022; Khan et al. 2023); the remaining studies targeted both children and parents/caregivers (n = 6) (Lawson et al. 2012; Solomon et al. 2013; Matthews and Franklin 2018; Farizan et al. 2020b; Petrass et al. 2021; Gupta et al. 2022). The age range varied from zero to 17 years for interventions aimed towards children, whilst the ages of parents/guardians were reported as between 18 and 55 years. The duration of drowning interventions ranged from 10 min (Zhu et al. 2017) to 11 years (Franklin and Peden 2017) whilst sample sizes ranged from 31 children in Brazil in a study to assess learn-to-swim programmes (Moura et al. 2021) to 306,752 Facebook users in a study by Smith and colleagues (Smith et al. 2021) from British Columbia which sought to analyse social media posts in response to a campaign targeting boating safety. Ethical approval was reported for 34 studies (Girasek 2011; Moran et al. 2012; Solomon et al. 2013; Wintemute et al. 2013; Shen et al. 2015; Silva et al. 2016; Franklin and Peden 2017; Ramos and Anderson 2017; Wilks et al. 2017; Zhu et al. 2017; Matthews and Franklin 2018; Olaisen et al. 2018; Ramos et al. 2018; Sandomierski et al. 2019; Alonge et al. 2020; Button et al. 2020; Farizan et al. 2020a, b; Hossain et al. 2020a,b; Karatrantou et al. 2020; Quan et al. 2020; Araiza-Alba et al. 2021; Moura et al. 2021; Petrass et al. 2021; Smith et al. 2021; Willcox-Pidgeon et al. 2021; Ashraf et al. 2022; Bhuiyan et al. 2022; Calverley et al. 2022; Casten et al. 2022; Gupta et al. 2022; Love-Smith et al. 2022; Tipton et al. 2022; Khan et al. 2023).

Drowning prevention strategies

Two studies (Wintemute et al. 2013; Forde et al. 2020) reported on interventions targeting individuals, 28 studies targeted groups (Girasek 2011; Lawson et al. 2012; Moran et al. 2012; Solomon et al. 2013; Shen et al. 2015; Silva et al. 2016; Turgut et al. 2016; Ramos and Anderson 2017; Wilks et al. 2017; Zhu et al. 2017; Glassman et al. 2018; Olaisen et al. 2018; Barcala-Furelos et al. 2019; Sandomierski et al. 2019; Button; Button et al. 2020; Farizan et al. 2020a, b; Hossain et al. 2020a, b; Karatrantou et al. 2020; McCallin et al. 2020; Araiza-Alba et al. 2021; Moura et al. 2021; Petrass et al. 2021; Calverley et al. 2022; Love-Smith et al. 2022; Tipton et al. 2022; Zhang et al. 2022; Khan et al. 2023) and the remainder (n = 13) targeted populations (Franklin et al. 2015; Franklin and Peden 2017; Matthews and Franklin 2018; Ramos et al. 2018; Alonge et al. 2020; Quan et al. 2020; Foulds et al. 2021; Smith et al. 2021; Willcox-Pidgeon et al. 2021; Ashraf et al. 2022; Bhuiyan et al. 2022; Casten et al. 2022; Gupta et al. 2022). Individual-level strategies included the use of educational strategies such as swimming skills to prevent drowning in children with and without disabilities (Forde et al. 2020). An example of a group-level strategy included a one-day training programme delivered by surf lifesavers to a group of primary school students to increase beach safety knowledge and awareness (Wilks et al. 2017). Population-level interventions included campaigns such as in the study by Casten and colleagues; who reported on a water safety mass-media campaign to promote supervision of children in and around waterways (Casten et al. 2022). Most (n = 34) of the included studies utilised behavioural strategies (Table 2). Behavioural strategies were reported as educational, including swimming lessons (Franklin et al. 2015; Olaisen et al. 2018; Karatrantou et al. 2020; Willcox-Pidgeon et al. 2021; Zhang et al. 2022), the use of educational materials (Turgut et al. 2016; Zhu et al. 2017; Glassman et al. 2018; Matthews and Franklin 2018; Farizan et al. 2020a, b), training (Solomon et al. 2013; Wilks et al. 2017) or media or technology (Hossain et al. 2020a,b; Smith et al. 2021; Casten et al. 2022). For example, the study by Hossain and colleagues investigated the feasibility of mobile phones and SMS text messaging services to deliver child drowning prevention messaging in low-income settings. Eight studies (Wintemute et al. 2013; Matthews and Franklin 2018; Alonge et al. 2020; McCallin et al. 2020; Quan et al. 2020; Ashraf et al. 2022; Bhuiyan et al. 2022; Khan et al. 2023) utilised a mixture of socioecological and behavioural strategies, including a study by Alonge and colleagues (Alonge et al. 2020) which assessed the impact of creches and playpens along with staff training. Two studies utilised socioecological strategies only (Franklin and Peden 2017; Zhu et al. 2017); for example, a regulatory study by Franklin and Peden (28) reviewed the effectiveness of pool fencing legislation.

Table 2 Overview of studies by intervention level and strategy type (n = 43)
Full size table

Behavioural theory and formative research

Just over one-third of the included studies (33%; n = 14) reported using behavioural theory or models (Girasek 2011; Shen et al. 2015; Franklin and Peden 2017; Ramos and Anderson 2017; Zhu et al. 2017; Glassman et al. 2018; Matthews and Franklin 2018; Olaisen et al. 2018; Sandomierski et al. 2019; Quan et al. 2020; Araiza-Alba et al. 2021; Smith et al. 2021; Love-Smith et al. 2022; Zhang et al. 2022). Seven studies used the Health Belief Model (HBM) (Girasek 2011; Shen et al. 2015; Zhu et al. 2017; Glassman et al. 2018; Olaisen et al. 2018; Sandomierski et al. 2019; Araiza-Alba et al. 2021). One applied the HBM to testimonial-based videos (Shen et al. 2015) and another to a brief informational presentation (Love-Smith et al. 2022). Other theories included the Social Cognitive Theory (Love-Smith et al. 2022); Theory of Planned Behaviour (Sandomierski et al. 2019; Love-Smith et al. 2022), Social Marketing Theory (Quan et al. 2020; Smith et al. 2021); PRECEDE-PROCEED (Quan et al. 2020), Reasoned Action (Ramos and Anderson 2017), Theory of Health, Knowledge, Trust and Action (Zhang et al. 2022), Protection Motivation Theory (Sandomierski et al. 2019), Transtheoretical Model (Matthews and Franklin 2018), Hierarchy of Controls (Silva et al. 2016), Cognitive Theory of Multimedia Learning (Araiza-Alba et al. 2021), Social-ecological Framework (Olaisen et al. 2018) and Interest Theory (Araiza-Alba et al. 2021). Eight studies (19%) used formative research to inform the design or implementation of interventions or to profile the target audience. Methods included literature review (Farizan et al. 2020b; Petrass et al. 2021; Smith et al. 2021), consultation (Matthews and Franklin 2018; Quan et al. 2020; Farizan et al. 2020b), pilot testing (Zhu et al. 2017; Love-Smith et al. 2022) and framework review (Olaisen et al. 2018; Petrass et al. 2021). For example, one study by Quan and colleagues used focus group findings to inform the development and evaluation of their community water safety campaign (Quan et al. 2020).

Intervention outcomes

Outcomes described included changes in water safety awareness, attitudes, knowledge, behaviours and skills, drowning rates and environmental, policy and regulation changes. All studies reported some intervention effect. The majority (72%, n = 31) measured changes across multiple outcomes (Girasek 2011; Lawson et al. 2012; Moran et al. 2012; Solomon et al. 2013; Wintemute et al. 2013; Shen et al. 2015; Silva et al. 2016; Franklin and Peden 2017; Ramos and Anderson 2017; Wilks et al. 2017; Zhu et al. 2017; Glassman et al. 2018; Sandomierski et al. 2019; Alonge et al. 2020; Button et al. 2020; Forde et al. 2020; Karatrantou et al. 2020; McCallin et al. 2020; Quan et al. 2020; Farizan et al. 2020b; Araiza-Alba et al. 2021; Petrass et al. 2021; Smith et al. 2021; Ashraf et al. 2022; Calverley et al. 2022; Casten et al. 2022; Gupta et al. 2022; Tipton et al. 2022; Zhang et al. 2022). Fourteen studies (32%) reported on changes in water safety awareness (Lawson et al. 2012; Silva et al. 2016; Wilks et al. 2017; Zhu et al. 2017; Glassman et al. 2018; Alonge et al. 2020; Farizan et al. 2020a, b; Quan et al. 2020; Smith et al. 2021; Calverley et al. 2022; Casten et al. 2022; Tipton et al. 2022; Zhang et al. 2022) and 17 studies (40%) reported on changes in water safety attitudes (Girasek 2011; Solomon et al. 2013; Franklin and Peden 2017; Ramos and Anderson 2017; Zhu et al. 2017; Sandomierski et al. 2019; Button et al. 2020; Hossain et al. 2020a, b; Karatrantou et al. 2020; McCallin et al. 2020; Farizan et al. 2020b; Araiza-Alba et al. 2021; Petrass et al. 2021; Smith et al. 2021; Casten et al. 2022; Gupta et al. 2022; Love-Smith et al. 2022). Twenty-nine studies (67%) assessed various knowledge outcomes, with the majority (n = 15) specific to children’s water safety knowledge levels (Solomon et al. 2013; Shen et al. 2015; Turgut et al. 2016; Ramos and Anderson 2017; Wilks et al. 2017; Ramos et al. 2018; Barcala-Furelos et al. 2019; Button et al. 2020; Forde et al. 2020; Araiza-Alba et al. 2021; Foulds et al. 2021; Calverley et al. 2022; Tipton et al. 2022; Zhang et al. 2022) and ten to parental/guardian water safety knowledge levels (Girasek 2011; Moran et al. 2012; Silva et al. 2016; Glassman et al. 2018; Sandomierski et al. 2019; Farizan et al. 2020a, b; McCallin et al. 2020; Quan et al. 2020; Casten et al. 2022). A quasi-experimental study by Glassman and colleagues (Glassman et al. 2018) that utilised a social marketing intervention found increased parental/guardian knowledge and self-efficacy of water safety. Ten studies (23%) reported on various water safety behavioural changes (Wintemute et al. 2013; Shen et al. 2015; Zhu et al. 2017; Matthews and Franklin 2018; Sandomierski et al. 2019; Farizan et al. 2020a; McCallin et al. 2020; Quan et al. 2020; Casten et al. 2022; Gupta et al. 2022) such as intent to act (Zhu et al. 2017; Casten et al. 2022); personal floatation device (PFD) usage rates (Wintemute et al. 2013), use of lifeguarded open water sites (Quan et al. 2020), dissemination of water safety educative materials (McCallin et al. 2020) and parental supervision at public swimming pools (Matthews and Franklin 2018). Zhu and colleagues evaluated the influence of geo-specific posters on perceived threat and intention to avoid drowning ‘hotspots’ but found no statistically significant behavioural outcomes (Zhu et al. 2017). Thirteen studies (30%) measured changes in water safety skills (Moran et al. 2012; Franklin et al. 2015; Shen et al. 2015; Turgut et al. 2016; Wilks et al. 2017; Olaisen et al. 2018; Farizan et al. 2020a; Forde et al. 2020; Karatrantou et al. 2020; Petrass et al. 2021; Willcox-Pidgeon et al. 2021), with some studies comparing skill level between biological factors such as sex (Franklin et al. 2015; Olaisen et al. 2018; Petrass et al. 2021; Willcox-Pidgeon et al. 2021) and age (Franklin et al. 2015; Olaisen et al. 2018). One study by Araiza-Alba and colleagues investigated using a new innovative virtual reality tool to teach children water safety skills, reporting higher engagement and enjoyment amongst children who learnt using the intervention tool than children who learnt using traditional mediums (Araiza-Alba et al. 2021). Three studies (7%) assessed changes in drowning rates specific to creches and/or playpens use (Alonge et al. 2020), home swimming pool fencing regulation (Franklin and Peden 2017) and compliance within daycare centres (Ashraf et al. 2022). Other outcomes included environmental changes (Alonge et al. 2020; Quan et al. 2020; Ashraf et al. 2022; Bhuiyan et al. 2022; Khan et al. 2023) and increased water safety policies and regulations (Wintemute et al. 2013).

Study limitations

Most included studies reported limitations (86%; n = 37); six did not (Turgut et al. 2016; Barcala-Furelos et al. 2019; Karatrantou et al. 2020; Petrass et al. 2021; Bhuiyan et al. 2022; Tipton et al. 2022). Limitations included: sample size and generalisability related to representativeness and power (Girasek 2011; Moran et al. 2012; Franklin et al. 2015; Silva et al. 2016; Glassman et al. 2018; Matthews and Franklin 2018; Button; Button et al. 2020; Farizan et al. 2020a, b; McCallin et al. 2020; Moura et al. 2021; Smith et al. 2021; Willcox-Pidgeon; Peden et al. 2021; Casten et al. 2022; Gupta et al. 2022; Zhang et al. 2022); study design related to randomisation and controls (Franklin et al. 2015; Wilks et al. 2017; Zhu et al. 2017; Sandomierski et al. 2019; Alonge et al. 2020; Araiza-Alba et al. 2021; Foulds et al. 2021; Moura et al. 2021; Calverley et al. 2022; Love-Smith et al. 2022); measurement and data collection related to reliability and validity (Girasek 2011; Lawson et al. 2012; Wintemute et al. 2013; Shen et al. 2015; Franklin and Peden 2017; Glassman et al. 2018; Olaisen et al. 2018; Forde et al. 2020; McCallin et al. 2020; Quan et al. 2020; Willcox-Pidgeon; Peden et al. 2021; Ashraf et al. 2022; Gupta et al. 2022; Love-Smith et al. 2022; Zhang et al. 2022); statistical and analytical issues; e.g. missing data or analysis choice (Wintemute et al. 2013; Franklin and Peden 2017; McCallin et al. 2020; Willcox-Pidgeon et al. 2021; Khan et al. 2023); intervention fidelity (Moran et al. 2012; Franklin et al. 2015; Silva et al. 2016; Franklin and Peden 2017; Ramos and Anderson 2017; Wilks et al. 2017; Zhu et al. 2017; Matthews and Franklin 2018; Ramos et al. 2018; Farizan et al. 2020a; Forde et al. 2020; Araiza-Alba et al. 2021; Calverley et al. 2022; Khan et al. 2023) including programme design and dose; outcome measures (Moran et al. 2012; Franklin et al. 2015; Matthews and Franklin 2018; Ramos et al. 2018; Forde et al. 2020) including duration; contextual factors (Moran et al. 2012; Solomon et al. 2013; Shen et al. 2015; Matthews and Franklin 2018; Olaisen et al. 2018; Sandomierski et al. 2019; Farizan et al. 2020a, b; Forde et al. 2020; Quan et al. 2020; Willcox-Pidgeon et al. 2021; Calverley et al. 2022; Khan et al. 2023) including language and cost; ethical considerations (Sandomierski et al. 2019; Gupta et al. 2022) including iatrogenic effects; external validity (Girasek 2011; Solomon et al. 2013; Silva et al. 2016; Zhu et al. 2017; Glassman et al. 2018; Matthews and Franklin 2018; Olaisen et al. 2018; Alonge et al. 2020; Button; Button et al. 2020; Hossain et al. 2020a, b; Quan et al. 2020; Ashraf et al. 2022) relating to generalisability and real-world relevance; study bias (Moran et al. 2012; Franklin et al. 2015; Alonge et al. 2020; Casten et al. 2022; Gupta et al. 2022; Love-Smith et al. 2022; Khan et al. 2023) related to selection and attrition; and reporting (Ramos et al. 2018).

Author-reported recommendations

Authors provided recommendations in all except one (Smith et al. 2021) of the included studies (n = 42). Recommendations were broadly categorised as methodological (n = 19) (Lawson et al. 2012; Solomon et al. 2013; Shen et al. 2015; Silva et al. 2016; Wilks et al. 2017; Zhu et al. 2017; Glassman et al. 2018; Ramos et al. 2018; Sandomierski et al. 2019; Button et al. 2020; Forde et al. 2020; Hossain et al. 2020a, b; McCallin et al. 2020; Moura et al. 2021; Ashraf et al. 2022; Bhuiyan et al. 2022; Calverley et al. 2022; Gupta et al. 2022; Khan et al. 2023), service delivery (n = 32) (Girasek 2011; Lawson et al. 2012; Solomon et al. 2013; Wintemute et al. 2013; Franklin et al. 2015; Silva et al. 2016; Franklin and Peden 2017; Ramos and Anderson 2017; Wilks et al. 2017; Glassman et al. 2018; Matthews and Franklin 2018; Barcala-Furelos et al. 2019; Sandomierski et al. 2019; Alonge et al. 2020; Farizan et al. 2020a, b; Hossain et al. 2020a,b; Quan et al. 2020; Araiza-Alba et al. 2021; Foulds et al. 2021; Moura et al. 2021; Petrass et al. 2021; Willcox-Pidgeon; Peden et al. 2021; Ashraf et al. 2022; Bhuiyan et al. 2022; Calverley et al. 2022; Casten et al. 2022; Gupta et al. 2022; Love-Smith et al. 2022; Tipton et al. 2022; Zhang et al. 2022; Khan et al. 2023), policy (n = 3) (Franklin and Peden 2017; Alonge et al. 2020; Tipton et al. 2022) or research (n = 32) (Lawson et al. 2012; Moran et al. 2012; Solomon et al. 2013; Franklin et al. 2015; Shen et al. 2015; Silva et al. 2016; Turgut et al. 2016; Ramos and Anderson 2017; Zhu et al. 2017; Glassman et al. 2018; Matthews and Franklin 2018; Olaisen et al. 2018; Ramos et al. 2018; Sandomierski et al. 2019; Alonge et al. 2020; Button et al. 2020; Farizan et al. 2020a, b; Forde et al. 2020; Hossain et al. 2020a, b; Karatrantou et al. 2020; Araiza-Alba et al. 2021; Foulds et al. 2021; Moura et al. 2021; Petrass et al. 2021; Willcox-Pidgeon et al. 2021; Ashraf et al. 2022; Bhuiyan et al. 2022; Calverley et al. 2022; Love-Smith et al. 2022; Tipton et al. 2022; Zhang et al. 2022; Khan et al. 2023). Examples of methodological recommendations related to the duration of interventions (Zhu et al. 2017), cultural tailoring (Solomon et al. 2013), modification of study design (Sandomierski et al. 2019) or validity and reliability of instruments; for example, in the study by Lawson and colleagues (Lawson et al. 2012). Service delivery recommendations commonly proposed implications related to swimming lesson delivery (Willcox-Pidgeon; Peden et al. 2021), scaling up education interventions (Farizan et al. 2020b) and using traditional media such as in the study by Casten and colleagues (2022). Of note is that few studies made policy recommendations. Those related to pool fencing (Franklin and Peden 2017), advocacy for a more significant quantum of funding to scale-up services (Alonge et al. 2020) and mandatory swimming lessons, such as in the study by Tipton and colleagues (Tipton et al. 2022). Most papers made recommendations related to research, which included the need for large trials (Olaisen et al. 2018), longitudinal programme effects (Calverley et al. 2022), or the economic feasibility of self-taught home kits for parents to learn child CPR, such as in the study by Moran and colleagues (Moran et al. 2012).

Discussion

Drowning is one of the ten leading causes of death amongst children worldwide (World Health Organization 2021). Despite this, as highlighted by the UN Resolution, drowning remains ‘largely unrecognised relative to its impact’ (World Health Organization 2023). This paper presented a contemporary perspective on child drowning, updating findings from a 2016 systematic review (Leavy et al. 2016). It remains one of few reviews with a focus on public health fatal and non-fatal drowning prevention interventions targeting children across the globe.

The review noted a threefold increase in studies (from 15 to 43 articles) published between 2011 and 2023 reporting on public health interventions to prevent child drowning. A recent bibliometric assessment of drowning prevention research highlighted significant growth in epidemiologic studies that report on the burden and risk factors for drowning and noted the lack of evidence for interventions (Scarr and Jagnoor 2021). Balancing the research focus between epidemiological and intervention studies is critical to identifying drowning patterns and determining effective prevention measures to inform programmes and policy (Scarr and Jagnoor 2021). Like the previous 2016 review, studies from HICs predominated, noted in the broader literature (Tyler et al. 2017). However, there was a modest increase in the number of studies from LMICs, possibly reflecting the global investment in research from these regions (Jagnoor et al. 2020), including the establishment of the Centre for Injury Prevention and Research in Bangladesh (Hyder et al. 2014; Leavy et al. 2023a). Other reviews of drowning, including one focused on LMICs, have also recognised the association between drowning and development (Tyler et al. 2017; Scarr and Jagnoor 2022), highlighting issues such as limited public health infrastructure and outreach programmes creating difficulties in implementing and evaluating prevention interventions (Tyler et al. 2017; World Health Organization 2023). Drowning prevention requires greater visibility within agendas targeting the Sustainable Development Goals (SDGs), and reciprocally, the drowning prevention and water safety sectors must better integrate sustainable development goals in their action to increase drowning interventions and policy in LMICs. Accordingly, supporting greater reporting of evaluated public health interventions from LMICs is a critical call to action.

Reported strategies broadly reflect recommendations of global and jurisdictional strategies and plans, including the global Preventing Drowning: Implementation Guide (World Health Organization 2017) (for example, provide safe places away from water for preschool children, with capable childcare and train bystanders in safe rescue and resuscitation) and the Australian Water Safety Strategy 2030 (Australian Water Safety Council 2021) (for example, swimming and water safety lessons for preschool-aged children, media campaigns targeting parents and carers to raise awareness and promote prevention strategies, and swimming pool fencing legislation). However, consistent with the previous review (Leavy et al. 2016), interventions that utilised behavioural strategies at the group (interpersonal) level were the most commonly reported with a focus on health education. Whilst the literature acknowledges health education as an essential method for public health behavioural change, its overreliance may diminish the achievement of programmatic goals. The impact of health promotion programmes may be amplified by applying a more comprehensive approach and utilising a range of strategies (Nutbeam and Muscat 2021) to acknowledge better the influence of the broader social, ecological and commercial determinants on health. Of interest was some increase in environmental and technological strategies, including the use of virtual reality (VR) to teach children water safety skills in Australia (Araiza-Alba et al. 2021). This type of intervention did not feature at all in the previous review; however, the use of VR may represent an opportunity to enhance skill acquisition, whilst smartphone apps may provide safety and early warning messages for parents of young children (Hossain et al. 2020a, b). Interventions using technology are a burgeoning area of drowning prevention research.

Awareness and knowledge were commonly measured in this review. As these are short to intermediate outcomes considered antecedents to behaviour change, this is unsurprising and will continue to feature as reported outcomes, particularly in the context of short-term pilot interventions. There is a call for the sustainability of funding and to scale up what works in line with the aspirations of the UN resolution. Just over one-third of studies used theory to underpin intervention design, an increase from 20% in the previous review. However, formative research appeared to decline, only used in one in five studies in the current review compared with one-third in the previous decade. Behavioural theory and planning frameworks are key drivers of successful public health programmes (Leavy et al. 2016). Accordingly, a recommendation for their more fulsome and consistent application remains relevant. The recommendation to include theory and formative research remains unchanged, together with funding for long-term intervention delivery and evaluation.

In this updated review, limitations were broadly consistent with findings from previous reviews of child drowning (Wallis et al. 2015; Leavy et al. 2016). Of interest are relatively few papers reporting on implementation, which has implications for replication and adaptation of programme design and evaluation (Della Bona et al. 2023). In addition, the lack of qualitative studies may limit contextual insights into behavioural outcomes. These factors may limit the tailoring and contextualisation of interventions for specific populations or settings. Previous research has found few published evaluations of interventions with populations who may be at greater risk for drowning, such as migrants or First Nations peoples (Willcox-Pidgeon et al. 2020). These two areas require greater development in the drowning literature. Whilst most included studies in the review reported recommendations, few related to policy. This finding supports calls from the UN resolution to establish appropriate and proportionate regulations across sectors (World Health Organization 2023), ensuring the UN resolution is a catalyst for action that will help position child drowning prevention within broader global health and sustainable development agendas (Scarr et al. 2022). As with gaps in reporting on limitations, few recommendations reflected implementation considerations. A recent review of implementation strategies used in drowning prevention echoes this finding (Della Bona et al. 2023). Promisingly, compared with the previous reviews, there was an increased focus on economic evaluation and funding, cited in the literature as critical to scaling up and sustaining interventions (Jagnoor et al. 2020).

Strengths and limitations

This review has several strengths: an expansive search covering 15 databases, consistency with previously published reviews (Leavy et al. 2016) and protocols (Crawford et al. 2014); use of MetaQAT (Rosella et al. 2016) to provide public health relevant quality appraisal; use of multiple reviewers; and inclusion of qualitative and quantitative studies to capture a wide range of interventions. There are, however, some limitations to note: a focus on studies published in the English language and exclusion of the grey literature. It may be that key interventions were excluded as they were published in a language other than English or by a community-based organisation as an online report. We recognise that their contributions may have been excluded. No meta-analyses were conducted owing to the heterogeneity across the studies.

Conclusion

This updated review of the peer-reviewed literature published over the past decade noted an almost threefold increase in the number of papers describing the evaluation and outcomes of public health interventions to prevent fatal and non-fatal drowning amongst children. Despite this increase, interventions remain primarily educational, focusing on behavioural strategies. This is despite more recent evidence that points to a range of broader socio-ecological factors that influence drowning and its prevention. Consistent with the broader health promotion literature, more comprehensive and systems-level interventions that tackle the upstream determinants of child drowning are necessary. Attending to the factors that influence implementation and providing greater support for practitioners to publish their findings on what works, why and in what context is recommended (Alonge et al. 2020).