Composition VendorNewgen KnowledgeWorks (P) Ltd.PLOS Glob Public HealthplosplosgphPLOS Global Public Health2767-3375Public Library of ScienceSan Francisco, CA USA10.1371/journal.pgph.0005337PGPH-D-25-02400Research ArticleSocial sciencesLaw and legal sciencesRegulationsSocial sciencesLaw and legal sciencesCriminal justice systemLaw enforcementSocial sciencesLaw and legal sciencesLegislationBiology and life sciencesPopulation biologyPopulation metricsDeath ratesPhysical sciencesChemistryChemical compoundsOrganic compoundsAlcoholsPhysical sciencesChemistryOrganic chemistryOrganic compoundsAlcoholsEarth sciencesGeographyHuman geographyUrban geographyUrbanizationSocial sciencesHuman geographyUrban geographyUrbanizationComputer and information sciencesInformation theoryGraph theoryDirected graphsDirected acyclic graphsPhysical sciencesMathematicsGraph theoryDirected graphsDirected acyclic graphsMedicine and health sciencesHealth careEnvironmental healthSanitationMedicine and health sciencesPublic and occupational healthEnvironmental healthSanitationAn analysis of global legislation and regulation related to drowning preventionAn analysis of global legislation and regulation related to drowning preventionhttps://orcid.org/0000-0003-3497-3137EssexRyanConceptualizationFormal analysisMethodologyVisualizationWriting – original draftWriting – review & editing12*JagnoorJagnoorConceptualizationWriting – original draftWriting – review & editing12The George Institute for Global Health, Sydney, AustraliaUniversity of New South Wales, Sydney, AustraliaJosephAlexEditorSRMIST: SRM Institute of Science and Technology (Deemed to be University), INDIA* E-mail: ressex@georgeinstitute.org.au
The authors have declared that no competing interests exist.
☯ All authors contributed equally to this work.
2532026202663e0005337198202510320262026Essex, JagnoorThis is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Drowning remains a leading but neglected cause of injury-related death worldwide. While some forms of regulation, such as pool fencing, have demonstrated clear effectiveness, less is known about the broader influence of legislation and regulatory measures on drowning mortality. This study provides the first global analysis of drowning‐prevention legislative coverage, utilising secondary analysis of cross-sectional data, to explore country‐level associations between legislation/regulation and drowning rates. Utilising the World Health Organization (WHO) Global and Regional Status Reports on Drowning, alongside economic, environmental, governance, and health-system indicators, we assembled a cross-section of 127 countries. We estimated several models that included the full set of national legislation indicators and nine covariates, complemented by a random-forest analysis. Across specifications, we observed no consistent association at the country level between the presence or count of legal measures and lower drowning mortality. By contrast, structural determinants were repeatedly implicated. For example, in our main models, higher per-capita alcohol use was positively associated with drowning mortality (IRR = 1.18, 95% CI [0.99, 1.40]), while better access to safe water/sanitation was protective (IRR = 0.84, 95% CI [0.71, 0.99]). Exploratory interactions examining the influence of institutional/governance capacity suggested that public-pool fencing and, more tentatively, alcohol restrictions near water, were more protective where institutional capacity was stronger. Our findings support the idea that legalisation/regulation is likely to be most effective when embedded within broader investments in infrastructure, service delivery, and enforcement capacity.
The authors received no specific funding for this work.Data AvailabilityAll data utilised in this study is publicly available and cited in the main text. All data can be found at the following URLs: WHO Global Status Report on Drowning: https://www.who.int/teams/social-determinants-of-health/safety-and-mobility/global-report-on-drowning-prevention WHO Western Pacific Regional Status Report on Drowning: https://www.who.int/publications/i/item/9789290619581 WHO Southeast Asia Regional Status Reports on Drowning: https://www.who.int/publications/i/item/9789290221098 Varieties of Democracy (V-DEM): https://www.v-dem.net/ World Justice Project (WJP): https://worldjusticeproject.org/ Global Health Security (GHS): https://ghsindex.org/ World Banks World Data Indicators (WDI): https://databank.worldbank.org/source/world-development-indicators.Introduction
When it comes to drowning prevention, throughout the literature we can find several examples where regulation and legislation have been effective in reducing drowning mortality and associated risks. Take the example of pool fencing legislation in Australia. Major steps toward fencing private pools in Australia started to take shape in the 1990’s, two decades after childhood drowning in backyard pools was identified as an issue [1]. While this took some time to get right and while implementation varied state by state, these laws today have proven effective, with drowning deaths dropping by about 50% in children aged 0–4 years with pool fencing playing a substantial role in this decrease [2]. Elsewhere, regulation was found to be equally as effective when it came to personal flotation devices. In another study from Australia, it was found that the introduction of legislation that mandated the use of personal flotation devices amongst recreational boaters decreased drowning fatalities by 70% when looking at mortality in the periods of 5 years pre-post intervention [3]. Internationally, we find regulatory and policy efforts for those working on or around water with extensive international regulation in relation to vessel design and safety for example [4]. At the same time however, we know little about many forms of regulation and how they directly and indirectly impact drowning prevention efforts, with few studies having focused directly on the impact of legislation and regulation in this space. There are a small number of exceptions here when it comes to drowning prevention, with a recent scoping review identifying 25 papers that provided some insights into the effectiveness of legislation for a range of outcomes, from mandating lifejackets to regulating recreational space in open water [5]. One of the findings of this study was that while legislative and regulatory efforts appeared to be promising in curbing drowning deaths, their success hinged on a number of factors, like compliance, enforcement and having the appropriate expertise when it came to development and implementation. While generally, the studies in this review were often small and relied on case studies, amongst these papers were a small number of larger studies that examined the effect of legislation and regulation on a broader scale. For example, Quan, Mills [6] examined regulation across 30 United States (US) states related to open water safety. This study found that states with greater regulation generally had lower open water drowning death rates. Amongst states without any form of regulation, drowning rates were three and four times higher amongst young people and non-white residents respectively, compared to those who had all five types of regulation.
As a whole, the literature suggests that regulation, under the right conditions, can be a powerful tool to curb drowning deaths. There are still however shortcomings and gaps. We know little about the influence of regulation on drowning in certain sectors for example. The evidence that we have is also generally based on smaller and sector-specific studies, with most studies focused on discreet pieces of legislation, like those focused on pool fencing or personal flotation devices. From these studies we can begin to put together some of the causal pathways that make regulation effective, from altering the environment, to changing behaviours and norms, however, there also remains a paucity of evidence here.
One further area that has been overlooked in this space has been the global landscape of legislative and regulatory approaches related to drowning prevention. The recent World Health Organization (WHO) Global Status Report on Drowning Prevention [7], along with two earlier regional reports [8,9] provide valuable resources in this respect, in that they give some insight into the broader regulatory and legislative measures implemented by countries to curb drowning, providing data on the coverage and type of legislation/regulation implemented by countries. Using this data this study sought to add to the modest literature above, providing the first global examination of regulatory and legislative measures related to drowning prevention.
Research aims
This study therefore sought to examine 1) the extent to which legislative measures, both individually and in aggregate, were associated with drowning mortality rates 2) the specific combinations of legal, environmental, health and governance factors that predicted countries with differential drowning burdens and 3) whether strength of enforcement and institutional capacity were associated with the effectiveness of legislation/regulation. Our analysis is ecological and cross-sectional. Consequently, estimates may reflect endogeneity (e.g., countries adopting laws in response to high drowning) and measurement error; we therefore advise caution and treat findings as descriptive of country-level patterns.
Methods
We addressed the above research aims through three complementary approaches. Aims 1–2 used negative binomial regression to estimate associations between individual laws (Model M1), aggregate legislative counts (Model M2), and drowning mortality, adjusted for structural confounders including economic development, health systems, and governance. Aim 2 was further addressed using Random Forest classification to identify optimal combinations of legal environmental, and institutional predictors of drowning burden. Aim 3 employed interaction models testing whether governance capacity moderated legislative effectiveness, supplemented by regional models (M3–M5) directly measuring enforcement implementation strength.
Data sources
Data was assembled at country level for 2021, with variables collated from a range of sources. Each data source and the variables included in this analysis are included in S1 Table. A brief description is provided here. The core outcomes and explanatory variables for this study come from the Global Status Report on Drowning report [7] and the Western Pacific and South East Asian Regional Status Reports on Drowning [8,9]. Drowning rates per 100,000 population were extracted from the global status report (and reconstructed as count data; see below), along with the presence of a national drowning prevention plan, and the presence or absence of a national drowning prevention strategy and the six legislative/regulatory measures listed, including fencing requirements for private and public pools, domestic water‐transport safety regulations, life‐jacket mandates, and alcohol restrictions near waterbodies. We extracted all legislative measures and the extent to which they were enforced from the regional status reports. We also included covariates from a range of sources. Variables included a range of economic, legal, environmental, geographic and health-system related metrics that all had potential bearing on rates of drowning and the implementation and enforcement of legislation/regulation. These were extracted from the Varieties of Democracy (V-DEM) [10], World Justice Project (WJP) [ 11], Global Health Security (GHS) [12], and the World Banks World Data Indicators (WDI) [13] datasets. We utilised data from 2021 in all cases, except for the WJP where we used the 2023 rule of law index. For average temperature we utilised mean temperature form 1995–2020.
Data transformation
We derived a number of variables from the global and regional status reports.
From the global status reports, we extracted data on seven distinct legislative/regulatory domains: (1) national drowning prevention strategy, (2) disaster risk management policy specifying drowning, (3) fencing requirements for private pools, (4) fencing requirements for public pools, (5) domestic water transport safety requirements, (6) lifejacket use mandates, and (7) alcohol regulation near waterbodies. Legislative/regulatory presence was originally categorised as national/sub-national/no; we recoded this as a binary (yes/no) variable to indicate presence (at any jurisdictional level) or absence in each country. We also computed a total legislative count by summing across six of these domains (excluding the national drowning prevention strategy, which we modelled separately as it represents strategic coordination rather than specific regulatory mechanisms). This provided a total legislative measures variable scored 0-6, where higher scores indicate more comprehensive legislative coverage across key drowning prevention domains. As the regional reports included more granular data on legislative/regulatory mechanisms and the extent of their enforcement (which took the extent to which each law was enforced, rated 0-10), we opted to calculate two variables, this included the total number of legislative instruments reported (with a possible range 0-19) and a total enforcement score (with a possible range of 0-190). Specifically, the regional reports contain data across 23 specific legislative/regulatory mechanisms organized into five domains: (1) maritime safety, (2) child safety, (3) water safety competencies, (4) disaster risk reduction, (5) drowning surveillance. Total legislative count (range 0-19) was summed by observing a simple count of how many of the 19 legislative/regulatory mechanisms were reported as present in each country or territory, regardless of enforcement strength. This provided a measure of legislative breadth. Total enforcement score (range 0-190) took implementation scores on a 0-10 scale (with lower score indicating less enforcement/implementation). We summed these ratings across all 19 items to create an overall enforcement strength score. This measure captured not just whether legislation exists, but the extent to which it was actively implemented and enforced. These two measures (legislative count and enforcement score) provided complementary information, notably, a country might have many laws on the books (high count) but weak implementation (low enforcement score), or conversely, fewer laws that are strongly enforced. We examined both independently (Models M3-M4) and their interaction (Model M5) to test whether enforcement capacity moderates the association between legislative presence and drowning mortality. To ensure consistency across all variables, exposure to natural disasters (from the GHS data see below) was reverse scored, as in the original dataset lower scores meant greater exposure to natural disasters. To aid analysis where we utilised count models, we worked with country counts of drowning deaths and a population offset. As only rates were available, deaths were reconstructed (rate × population / 100,000) and rounded to the nearest integer. This rounding introduced minor approximation error, though given the scale of the counts involved, the effect on model estimates is negligible. All continuous covariates were z scored before modelling.
VariablesOutcome
The major outcome of interest in this study was drowning mortality per 100,000 population as reported in the Global Status report [7].
Key explanatory variables/policy exposures
The main outcome variables were each of the legislative/regulatory measures from the global status report; these included fencing requirements for private and public pools, domestic water‐transport safety regulations, life‐jacket mandates, and alcohol restrictions near waterbodies. We also included whether the country had national water safety plan. A derived variable was also included that summed the number of legislative instruments (including a national water safety plan). From the regional reports, two variables were included, the number of legislative instruments reported in each country and an index that represented the total strength of their enforcement. In total this provided eight explanatory variables that were included in the main models, with regional models included as supplementary material.
Covariate selection
Alongside the above outcomes and to control for confounding we assembled 24 candidate controls spanning climate and geography (average temperature, precipitation), demographics (child population share, urbanisation, tourist visits), health-system capacity (hospital beds, physicians, per-capita health spending), risk exposures (alcohol use, natural-disaster and environmental hazards), and institutional/governance quality (rule of law, regulatory enforcement, government effectiveness). To inform final covariate selection we applied elastic-net stability selection to identify the most reliable confounders. Across 500 bootstrap resamples, an elastic-net model with ten-fold cross-validation was refit, and inclusion frequency (the proportion of resamples in which its coefficient was non-zero) was recorded. The final covariates that were selected were gross domestic product at purchasing power parity (GDP PPP), average surface temperature, legislative transparency and enforcement (V-DEM), per-capita alcohol consumption (WDI), health-sector capacity, urbanisation, natural-disaster exposure, access to water and sanitation, and public-health spending (GHS). The selection of these variables was driven by the above model and existing theoretical and empirical evidence on drowning prevention [14]. Each of these variables, a description and their source is included in S1 Table.
Causal considerations and covariate classification
To formalise assumptions about confounding, we constructed a directed acyclic graph (DAG) representing hypothesized causal relationships between legislation, structural factors, and drowning mortality (S2 Fig). The DAG identified multiple sufficient adjustment sets; all included economic development and governance capacity as essential confounders. Our final control set (nine covariates) represents a sufficient adjustment set under the DAG assumptions. We classified urbanization and public health spending as potential partial mediators (economic development may influence these, which in turn affect both legislation and drowning), though both also exhibit strong confounding properties. Sensitivity analyses excluding these variables are reported in the supplement. No collider bias was identified under the DAG structure.
AnalysisMain models
To estimate the association between legislation/regulation and drowning mortality, we modelled reconstructed death counts (rate × population/ 100,000) using negative binomial regression (NB2) with a log link function and log(population/100,000) offset to accommodate over-dispersion in count data. All models used robust standard errors and reported incidence rate ratios (IRRs) with 95% CIs. We estimate three nested specifications that differ only in how legal coverage was represented; each includes the common control set (see above):
M0 — controls only: establishing a baseline relationship between the controls and drowning deaths.
M1 — controls + eight policy indicators (from the global status report): asking which specific laws matter the most. To avoid quasi-separation in small cells, indicators that were non-varying or ultra-rare in the estimation sample were excluded.
M2 — controls + number of legislative measures (from the global status report): testing whether having more laws in total was linked to lower drowning rates.
Regional variables (total laws, enforcement score, and their interaction) were available for a much smaller country subset (n = 19). This made estimates unstable. To prevent over-fitting and avoid conflating samples, we report the regional models (M3–M5) as supplementary material (see below).
Robustness and diagnostics
We subjected the main specifications to several robustness exercises designed to probe sensitivity to estimator choice, influential observations, and collinearity among the structural controls. First, we re-estimated the negative binomial models with a population offset using a Poisson pseudo-maximum likelihood (PPML) estimator with a log link, holding the sample fixed across estimators so that any differences reflected modelling rather than composition. Second, to assess the influence of outliers, we implemented an influence-trim procedure: for each model we fit a Poisson start, computed Cook’s distances, removed the top five percent of observations, and re-fit the corresponding NB2 specification on the trimmed sample. To address potential over-parameterisation and shared variance among the controls, we implemented a dimension-reduction exercise. Specifically, we performed principal components analysis (PCA) on the structural controls and replaced the original controls with the leading components required to explain approximately 80% of their variance. We then re-estimated the policy model using NB2 (with Poisson fallback where necessary), retaining the policy indicators unchanged.
We evaluated the stability of the three main NB2 specifications using diagnostics and robustness analyses tailored to count outcomes with a population offset. For each model we simulated residuals and examined tests of dispersion, zero inflation, and uniformity; we assessed multicollinearity using variance inflation factors and we tested spatial autocorrelation using Moran’s I based on k-nearest-neighbour weights derived from country centroids.
Exploratory machine-learning classification of drowning risk
To complement parametric models, we trained a global random-forest regression model using drowning mortality rate as the outcome. Predictors comprised the full set of national legislation indicators alongside the nine covariates. We used 1,000 trees with out-of-bag prediction and computed permutation importance for each feature. To quantify uncertainty in importances, we bootstrapped the rows (B ≈ 300 resamples), re-fit the forest in each resample, and summarised the distribution of importances to obtain means and percentile 95% CIs. This non-parametric approach provided a complementary ranking of legal and contextual factors associated with variation in drowning risk, without imposing linearity or additivity. It is worth noting that permutation importance reflects the predictive contribution of a given variable to the model, that is, how much model accuracy deteriorates when that variable is shuffled. It does not imply a causal relationship between the variable and the outcome.
Interaction between governance capacity and legislative measures
To examine whether the influence of each legislative/regulatory measure varied with levels of enforcement, we considered four metrics: V-DEM’s legislative transparency/enforcement and respect for rule-of-law indices, and the World Justice Project’s rule-of-law and regulatory enforcement indices. While similar, each metric measures a slightly different element of enforcement and rule of law. V-DEM’s legislative transparency/enforcement index captures the extent to which laws are made in a transparent way and whether legislation is actually implemented and enforced in practice, while the respect for rule of law index captures whether legislation is applied impartially and in a fair and transparent way. World Justice Project’s rule-of-law index reflects how well a country adheres to the rule of law across eight dimensions, including constraints on government powers, absence of corruption and fundamental rights, while the regulatory enforcement index captures how effectively and fairly regulations are implemented and enforced in practice.
For each index, we fitted a negative binomial (NB2) count model with a population offset, including the seven global legislative indicators (presence/absence of each measure, including a national drowning-prevention strategy) and the same structural covariates as in the main models. Enforcement capacity was standardised (mean 0, SD 1) and entered both as main effects and in multiplicative terms with every law (e.g., lifejacket requirement × regulatory enforcement), allowing the model to estimate the baseline association of capacity and the incremental change in a law’s incidence rate ratio (IRR) as enforcement capacity increased. To aid interpretation, we derived predicted IRRs for countries with and without the law in question at –1 SD, the mean, and +1 SD of capacity,
ResultsMain models
We estimated associations between drowning prevention legislation and mortality using negative binomial regression, adjusting for nine structural confounders including economic development, governance capacity, health systems, and environmental factors (M0-M2; N = 104–127). Confounder coefficients are reported in Table 1 for transparency but are not interpreted as causal effects. When individual laws were examined (M1, N = 104), none of the seven legislative indicators reached conventional significance; point estimates fluctuated around the null, including for private-pool fencing (IRR = 0.64, 95% CI: 0.39–1.05, p = 0.17) and water-transport safety (IRR = 0.76, 95% CI: 0.52–1.10, p = 0.15). A simple count of national legislative measures (M2, N = 127) showed no association with mortality (IRR = 1.03, 95% CI: 0.97–1.10, p = 0.35). Overall, after adjusting for structural confounders, we found no systematic evidence that the presence of specific national laws, or having more laws in total, was associated with lower drowning mortality. Results are summarized in Fig 1 and Table 1.
10.1371/journal.pgph.0005337.t001
Main NB2 models (IRR and p values).
M0
M2
M1
GDP
0.911
0.900
0.901
(0.136)
(0.141)
(0.183)
Avg temp
1.092
1.079
1.094
(0.084)
(0.079)
(0.085)
Legislative enforcement
0.950
0.958
0.970
(0.075)
(0.076)
(0.100)
Alcohol
1.178*
1.176*
1.186
(0.076)
(0.076)
(0.131)
Health-sector robustness
0.992
0.971
0.978
(0.102)
(0.097)
(0.109)
Urbanisation
1.231+
1.235+
1.179
(0.134)
(0.136)
(0.142)
Disaster exposure
1.120
1.124
1.120
(0.089)
(0.090)
(0.098)
Water & sanitation
0.838*
0.835*
0.805+
(0.073)
(0.072)
(0.093)
Public-health spend
0.847
0.851
0.882
(0.120)
(0.121)
(0.180)
National strategy
1.044
(0.234)
Disaster policy
1.104
(0.276)
Private-pool fencing
0.638
(0.199)
Public-pool fencing
1.039
(0.282)
Water-transport safety
0.759
(0.156)
Lifejacket requirement
1.279
(0.272)
Alcohol regulation near water
1.163
(0.182)
Total laws (global)
1.032
(0.035)
Num.Obs.
127
127
104
RMSE
2454.75
2433.96
2490.38
Note: + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001
Note: confounder coefficients are presented for completeness but not interpreted as causal effects.
10.1371/journal.pgph.0005337.g001
Forest plot of main NB2 models (IRR and p values).
Because the regional dataset covered 19 countries, estimates were imprecise: coefficients on total regional laws, enforcement strength, and their interaction clustered tightly around the null with very wide confidence intervals and no consistent pattern across models. Given the small N and corresponding uncertainty, we treat these as supplementary and interpret them with caution (S1 Fig and S5 Table).
Robustness and diagnostics
Results were generally stable across alternative estimators and sample perturbations. Re-estimating the three specifications with Poisson pseudo–maximum likelihood on a fixed common sample did not overturn the main findings. In NB2, alcohol remained positively associated with drowning mortality (M0 IRR = 1.236, p < .05; M2 IRR = 1.240, p < .05) and access to safe water/sanitation remained protective (M0 IRR = 0.809, p < .05; M2 IRR = 0.811, p < .05). Under PPML these effects were attenuated and imprecise, and, consistent with the main text, neither the individual laws nor the total count of laws were systematically associated with mortality (PPML M2 total-laws IRR = 1.198, n.s.). A single PPML coefficient on “Disaster policy” was positive (IRR = 2.038, p < .05) but was not reproduced in NB2 and should be read cautiously given the smaller M1 sample (N = 71) and multiple comparisons.
An influence-trimmed analysis, dropping the top 5% of cases by Cook’s D and re-fitting NB2 on the trimmed rows yielded very similar patterns. Alcohol use was again associated with increased drowning mortality (M0 IRR = 1.298, p < .01; M2 IRR = 1.297, p < .01) and water/sanitation remained protective (M0 IRR = 0.743, p < .05; M2 IRR = 0.747, p < .05). Urbanisation became more precisely positive (M0 IRR = 1.335, p < .01; M2 IRR = 1.357, p < .01), and public-health spending was clearly protective across trimmed models (IRR ≈ 0.70, p < .001). To address shared variance among controls, we replaced them with principal components explaining ~80% of their variance and re-estimated the policy model. PC1–PC3 were strongly protective (IRR = 0.778, 0.829, 0.857; all p ≤ .05), while policy indicators again clustered around the null, with at most a marginal effect for private-pool fencing (IRR = 0.590, p < .10). Finally, diagnostics indicated no major misspecification: dispersion, zero-inflation, and uniformity tests were acceptable (p = 0.377–1.000), spatial autocorrelation in residuals was negligible (Moran’s I ≈ −0.03, p ≈ 0.63–0.65), and collinearity was moderate (max VIF ≈ 8.4 in M0/M2; 12.4 in M1) and consistent with the larger policy block. All of these results are included as supplementary material (S6-S10 Tables).
We constructed a directed acyclic graph (DAG) formalizing assumed causal relationships (S1 Fig). The DAG identified our nine-covariate set as a sufficient adjustment set. Urbanization and public health spending were classified as potential partial mediators but retained as confounders based on strong confounding properties; sensitivity analyses excluding them are reported in supplements (S12 Table)
Machine-learning classification of drowning risk
To complement the parametric models, we fitted a global random-forest regression to the log drowning-mortality rate, using the full set of national legislation indicators alongside the nine structural covariates. The ranking was dominated by structural conditions. Public-health spending, access to safe water and sanitation, GDP, and urbanisation showed the largest and most precise importances (mean 0.269 [0.167, 0.422]; 0.245 [0.155, 0.366]; 0.237 [0.144, 0.361]; 0.226 [0.111, 0.354], respectively). By contrast, the national legislation indicators exhibited low importances with intervals close to zero. Taken together, this non-parametric exercise aligns with the regression results with structural determinants accounting for most of the cross-national signal in drowning mortality, while the presence/absence of individual laws contributed comparatively little predictive power at the national level. Random forest results are summarised below in Fig 2 and as supplementary material (S11 Table).
10.1371/journal.pgph.0005337.g002
Random Forest: Importance of legislation and other factors in relation to countries with lower v higher rates of drowning (with bootstrap CIs).
Interaction between governance capacity and legislative measures
We next asked whether drowning mortality was associated with country institutional/governance capacity. To examine this, we utilised four governance proxies (V-DEM legislative transparency/enforcement; V-DEM respect for rule of law; WJP regulatory enforcement; WJP rule of law). We estimated NB2 models including the law main effects and all law × capacity interactions. Across measures, interaction estimates were generally small and imprecise, clustering near zero (Fig 3 and S12 Table). There were two exceptions. First, public-pool fencing showed a negative interaction with V-DEM respect for rule of law (β = −0.859, SE = 0.329, p = 0.009), consistent with a stronger protective association where governance capacity was higher. Second, alcohol restrictions near water displayed negative interactions with WJP regulatory enforcement (β = −0.556, SE = 0.335, p = 0.097) and with V-DEM respect for rule of law (β = −0.451, SE = 0.243, p = 0.063), however did not reach significance. For disaster policy, water-transport safety, lifejacket requirements, and private-pool fencing, interaction terms were close to null across all capacity proxies (all p > 0.10). The aggregate “total laws” count also showed no systematic interaction with capacity (Table 2).
10.1371/journal.pgph.0005337.t002
Enforcement capacity × law interaction coefficients.
Est
SE
p
Est
SE
p
Est
SE
p
Est
SE
p
Law
V-DEM: transparency/enforcement
WJP: regulatory enforcement
WJP: rule of law
V-DEM: respect for rule of law
Disaster policy
0.181
0.481
0.708
0.126
0.575
0.827
0.018
0.546
0.973
-0.064
0.451
0.886
Private-pool fencing
0.475
0.507
0.349
0.723
0.716
0.313
0.787
0.691
0.255
0.376
0.553
0.497
Public-pool fencing
-0.392
0.298
0.189
-0.585
0.596
0.326
-0.701
0.549
0.202
-0.859
0.329
0.009
Water-transport safety
0.463
0.455
0.310
0.530
0.694
0.446
0.142
0.826
0.864
-0.117
0.419
0.780
Lifejacket requirement
-0.019
0.273
0.946
-0.077
0.356
0.828
-0.064
0.380
0.867
-0.166
0.273
0.544
Alcohol near water regulation
-0.384
0.242
0.112
-0.556
0.335
0.097
-0.481
0.308
0.119
-0.451
0.243
0.063
Total laws
-0.031
0.242
0.899
0.016
0.311
0.958
0.016
0.297
0.958
0.145
0.243
0.552
10.1371/journal.pgph.0005337.g003
Interaction between enforcement capacity × law.
Discussion
This study set out to test whether national legislative and regulatory measures, individually and in aggregate, were associated with drowning mortality; to see which combinations of legal, environmental, health-system and governance factors best separated higher- from lower-burden settings; and to assess whether any legal influence depended on institutional capacity and enforcement. Taken together, the results point in a consistent direction. After accounting for structural covariates, we did not detect systematic associations between the presence or number of national laws and lower drowning mortality at the ecological, country level. Instead, variation in drowning mortality was most strongly linked to broader conditions, access to safe water and sanitation, alcohol consumption, urbanisation, and related socio-economic factors. Where we did find associations, they were conditional on country governance and institutional capacity. That is, public-pool fencing and, more tentatively, alcohol restrictions near water were more protective where rule-of-law and regulatory enforcement were stronger. This aligns with case studies showing effectiveness in targeted contexts while suggesting effects do not aggregate cleanly at the country level, likely because implementation, coverage, and compliance vary. As a whole, our global cross-section shows that legislative coverage on its own was not reliably associated with lower drowning related mortality; any benefits appear to rely on the institutional and infrastructural context in which those laws were introduced and implemented.
We treat these findings with some degree of caution, given the broader literature and the limitation of this study (see below). That is, while there have been no studies to examine global legislative coverage by country, our findings stand in contrast to much of the literature, that speaks to the effectiveness of regulation and policy in relation to drowning whether this be in relation to pool fencing [15] or the regulation of personal flotation devices [16]. Notably, our results stand in contrast to other studies that have employed similarly broad designs, at least when looking at legislation at a national level [6]. It is important to be clear about what our null findings mean. The absence of a statistically significant association at the ecological, country level does not mean that laws are ineffective or policy-irrelevant. National-level aggregation masks the considerable heterogeneity in how laws are designed, implemented, and enforced across settings. These null results should therefore be read with caution, as a reflection of the limits of ecological analysis, not as evidence against legislation as a means to curb drowning. Put another way, while this study found no evidence of an association at an ecological level, it does not follow that legislation is ineffective. Our other finding helps explain this and also better aligns with the literature, that is, our finding also suggest that the effectiveness of legislation was dependent on a range of factors like governance and institutional capacity, including the implementation and enforcement of legislation [5]. These findings are important in other ways, beyond the implications for understanding the role of legislation in drowning prevention. The results also speak to the need for better data in this space, not only related to drowning mortality, but in relation to legislation/regulation and its effectiveness. This echoes the findings of a recent scoping review that, amongst other things, reached similar conclusions [5], notably around the need for greater data, not only in relation to legislative and regulatory coverage, but in relation to the qualitative elements of the regulation or legislation in question. Finally, and one thing that this analysis points out is that broader social, political and environmental factors are also influential when it comes to drowning prevention, consistent with much of the previous literature in this space [17,18]. Taken with our findings that enforcement capacity may have differential influence on various legislative measures, this directs us to the question of what (legislation/regulation) works, for whom in what circumstance. These results suggest it is important we think about legislation alongside conditions that shape drowning risk. Because structural determinants, safe water access and sanitation, alcohol use, urbanisation, and health-system capacity were found to be most influential, legal reforms are likely to be most effective when integrated alongside these considerations. For example, alcohol restrictions near water may be most effective alongside alcohol-harm reduction more generally; pool-fencing is likely going to be most effective when backed by subsidies, inspections and enforcement; transport-safety rules should be matched with training and coastguard resourcing and so on. To illustrate, pool fencing requirements have demonstrated clear effectiveness in high-income settings like Australia, where compliance is supported by inspections and public awareness. In lower-capacity settings, however, the same legislation may exist on paper without the institutional infrastructure needed to enforce it, meaning coverage and compliance remain low. Similarly, alcohol restrictions near water may reduce risk in settings with established enforcement mechanisms, but where regulatory oversight is limited, such provisions may have little practical effect. These examples, if correct, suggest that the effectiveness of legislation is not inherent to the law itself, but to the conditions in which it operates.
Limitations
At least four major limitations are worth discussing. First, the study is ecological and cross-sectional. We analysed country-level aggregates at a single point in time, so any associations we observed were averages over heterogeneous populations. The ecological fallacy applies directly here: a null or weak association at the national level does not imply that laws fail for specific groups or settings. Relatedly, we cannot establish temporal ordering. Laws may have been introduced because drowning was high, and without panel data we cannot separate reactive legislation from causal effects on later mortality. Second, measurement is coarse. The global series records the presence of legal measures but not their content, coverage, implementation, or compliance. Laws can vary widely in scope and enforceability; we lack systematic information on sub-national rules, enforcement effort, and sanctions. Our outcome is likewise an aggregate rate without age- or sex-specific detail, which limits our ability to detect targeted policy effects. Relatedly, national-level aggregation may obscure the effects of legislation that is known to be effective for specific populations or contexts. For instance, child-specific interventions like pool fencing or supervisory requirements may have measurable effects on child drowning but these would be diluted in an all-age national rate. Similarly, occupational drowning regulations targeting the fishing or maritime sectors may be effective within those populations but invisible at the level of national aggregates. This is a further reason to interpret null findings cautiously and to prioritise disaggregated data in future research. Third, power was a constraint when it came to the regional data. The regional dataset covered just 19 countries, yielding wide confidence intervals and unstable estimates, particularly for interactions. Finally, we also cannot rule out the (largely) null findings being related to the methods and/or analytic strategy we employed. These models should be read as exploratory and supportive only. More generally, while we probed robustness and diagnostics, residual confounding by unmeasured factors (like swimming and CPR skills or emergency services capacity for example) cannot be ruled out. Taken together, these limitations suggest caution in interpreting null legal effects at the national level and point to the need for longitudinal, disaggregated data on both drowning outcomes and the content, enforcement, and uptake of laws.
Conclusions
This paper provides the first global analysis of drowning prevention legislation and its association with mortality rates. While there are a number of examples that show that regulation has a role in reducing drowning, our findings underscore that these effects do not readily translate at the global level. Across 127 countries, structural determinants, income, urbanisation, alcohol use, disaster exposure, and access to safe water and sanitation, were consistently more predictive of drowning mortality than the presence or number of legislative measures. In most cases, laws themselves showed no systematic association with drowning rates, except for tentative evidence that fencing requirements for public pools and alcohol restrictions near water may reduce deaths when paired with strong institutional capacity and enforcement. The implications are twofold. First, legislation alone is unlikely to be sufficient. Legal instruments must be embedded in wider systems of governance, infrastructure, and social development if they are to deliver meaningful reductions in drowning. Second, our results highlight the pressing need for more granular, disaggregated data, notably in relation to who drowns, under what circumstances, and on the scope, coverage, and enforcement of laws. Without this, efforts to evaluate and improve legal approaches will remain constrained. Ultimately, drowning prevention cannot be advanced by legislation in isolation. Laws matter, but their effectiveness depends on the broader political, social, and environmental context in which they are introduced. Future research should move beyond documenting legal presence to interrogating what kinds of regulatory approaches work, for whom, and under what conditions.
Supporting information
Variables and data sources.
(DOCX)
Outcome summaries by model.
(DOCX)
Countries included in main models.
(DOCX)
Regions represented in main models.
(DOCX)
Summary of regional models.
(DOCX)
Robustness.
NB2 vs PPML.
(DOCX)
Robustness.
Influence-trimmed NB2 (drop top 5% Cook’s D).
(DOCX)
Countries excluded from influence trimmed NB2.
(DOCX)
Dimension reduction.
(DOCX)
Diagnostics.
(DOCX)
Random forest permutation importance.
(DOCX)
Interaction between enforcement capacity × law.
(DOCX)
Summary of regional models.
(DOCX)
Casual DAG: Drowning legislation and morality.
(DOCX)
ReferencesScottI. Prevention of drowning in home pools--lessons from Australia. Injury Control and Safety Promotion. 2003;10(4):227–36.MillerL. Analysis of unintentional drowning in Australia 2002-2022: progress, challenges, and data to inform prevention. Sydney, Australia: Australia RLSS. 2023.BugejaL, CassellE, BrodieLR, WalterSJ. Effectiveness of the 2005 compulsory personal flotation device (PFD) wearing regulations in reducing drowning deaths among recreational boaters in Victoria, Australia. Inj Prev. 2014;20(6):387–92. doi: 10.1136/injuryprev-2014-041169LiKX, WonhamJ. Maritime legislation: new areas for safety of life at sea. Maritime Policy & Management. 2001;28(3):225–34. doi: 10.1080/03088830110048880Essex R, Beri D, Jagnoor J. A scoping review of the literature on effectiveness of legislative and regulatory measures to prevent drowning. Under review.QuanL, MillsB, ChauSS, BennettE, BoltK, GomezA. Association of designated open water swim area regulations and open water drowning rates. Inj Prev. 2021;27(1):10–6. doi: 10.1136/injuryprev-2019-04346431911431World Health Organization. Global status report on drowning prevention. Geneva: World Health Organization. 2024.World Health Organization. Regional status report on drowning in South-East Asia. 2021.World Health Organization. Regional status report on drowning in the Western Pacific. 2021.Varieties of Democracy (V-Dem). https://www.v-dem.net/World Justice Project (WJP). https://worldjusticeproject.org/ 2023.Global Health Security (GHS). https://ghsindex.org/World Bank’s World Data Indicators (WDI). https://databank.worldbank.org/source/world-development-indicatorsWorld Health Organization. Global report on drowning: preventing a leading killer. 2014.ScottI. Prevention of drowning in home pools–lessons from Australia. Injury Control and Safety Promotion. 2003;10(4):227–36.MangioneT, ChowW. Changing life jacket wearing behavior: an evaluation of two approaches. J Public Health Policy. 2014;35(2):204–18. doi: 10.1057/jphp.2013.5124430805XieZ, HuangZ, RanQ, LuoW, DuW. Global burden of drowning and risk factors across 204 countries from 1990 to 2021. Sci Rep. 2025;15(1):10916. doi: 10.1038/s41598-025-95486-w40158004TanH, LinZ, FuD, DongX, ZhuS, HuangZ, et al. Change in global burden of unintentional drowning from 1990 to 2019 and its association with social determinants of health: findings from the Global Burden of Disease Study 2019. BMJ Open. 2023;13(4):e070772. doi: 10.1136/bmjopen-2022-0707723704557210.1371/journal.pgph.0005337.r001Decision Letter 0KhanUzmaAcademic Editor2026Uzma KhanUzma KhanThis is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.Submission Version0
16 Sep 2025
PGPH-D-25-02400
An analysis of global legislation and regulation related to drowning prevention
PLOS Global Public Health
Dear Dr. Essex,
Thank you for submitting your manuscript to PLOS Global Public Health. After careful consideration, we feel that it has merit but does not fully meet PLOS Global Public Health’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.
==============================
EDITOR: This paper is very detailed, important, and novel use of global data to analyse drowning mortality and its association with legislation. The contribution is significant, as there has been little comparable work at this scale. That said, a few areas could be clarified and strengthened. The methods would benefit from greater detail on the specific countries included and on any exclusions made, so that readers have a clearer sense of the study population. The list of covariates is extensive, but at times gives the impression that most available variables were included rather than only those with strong evidence for relevance; more justification for their selection would be helpful, and the appendix should ideally avoid listing variables that were not ultimately analysed. The methods and results, while rigorous, could also be written in a way that is slightly easier to follow. Finally, while I recognize the constraints of global data availability, some important factors such as swimming and CPR skills, emergency rescue services, and early warning systems are missing from the analysis. Early warning system coverage in particular may be available through other global datasets, and could be considered for inclusion. At minimum, the absence of such variables should be acknowledged more explicitly.
==============================
Please submit your revised manuscript by Oct 31 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at globalpubhealth@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pgph/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.
Please include the following items when submitting your revised manuscript:
A rebuttal letter that responds to each point raised by the editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.
Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.
We look forward to receiving your revised manuscript.
Kind regards,
Uzma Rahim Khan
Academic Editor
PLOS Global Public Health
Journal Requirements:
1. Please note that PLOS Global Public Health has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, all author-generated code must be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/globalpublichealth/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse.
2. Please upload separate figure files in .tif or .eps format. Also, remove the figures from your manuscript file but keep the legends.
For more information about figure files please see our guidelines:
3. We have noticed that you have uploaded Supporting Information files, but you have not included a list of legends. Please add a full list of legends for your Supporting Information files after the references list.
4. Please note that your Data Availability Statement is currently missing the repository name and/or the DOI/accession number of each dataset OR a direct link to access each database. If your manuscript is accepted for publication, you will be asked to provide these details on a very short timeline. We therefore suggest that you provide this information now, though we will not hold up the peer review process if you are unable.
If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.
Additional Editor Comments (if provided):
Reviewer #1:
Reviewer #2:
[Note: HTML markup is below. Please do not edit.]
Reviewers' comments:
Reviewer's Responses to Questions
Comments to the Author
1. Does this manuscript meet PLOS Global Public Health’s publication criteria ? Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe methodologically and ethically rigorous research with conclusions that are appropriately drawn based on the data presented.-->?>
Reviewer #1: Yes
Reviewer #2: No
**********
2. Has the statistical analysis been performed appropriately and rigorously?-->?>
Reviewer #1: Yes
Reviewer #2: Yes
**********
3. Have the authors made all data underlying the findings in their manuscript fully available (please refer to the Data Availability Statement at the start of the manuscript PDF file)??>
The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception. The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.-->
Reviewer #1: Yes
Reviewer #2: No
**********
4. Is the manuscript presented in an intelligible fashion and written in standard English??>
Reviewer #1: Yes
Reviewer #2: No
**********
Reviewer #1: This is an ambitious and timely article assessing the association between drowning-related legislation and mortality across countries. The topic is of clear global health relevance, and the authors’ effort to compile a cross-national dataset and triangulate multiple modelling approaches is commendable. The paper is well-written, transparent about methods, and presents extensive supplementary analyses. However, there are important methodological and interpretive issues that weaken the strength of the conclusions. In particular, sample size limitations, potential overfitting, choice of outcome modelling, and lack of clarity around causal interpretation require careful attention before publication.
Abstract
• Strengths: Clear statement of objectives and methods; multiple models (OLS, GLM, RF) reported; key findings succinctly conveyed.
• Concerns:
o Wording overstates null results (e.g., “laws do not matter”); should instead emphasize “no consistent association observed at country level.”
o Statistical detail is limited include effect sizes and 95% CIs for at least one key predictor to give context.
o Policy relevance is implied but not explicitly linked to the observed dominance of structural determinants.
o if Journal permits then add key words from MESH
Introduction
• Strengths: Good framing of drowning as a global public health issue and rationale for examining legislation.
• Concerns:
o Literature review is skewed towards descriptive burden; limited discussion of prior evaluations of policy effectiveness (e.g., natural experiments, case studies).
o Causal pathways (how laws may reduce drowning through enforcement, infrastructure, or behaviour change) need clearer articulation.
o The introduction should foreshadow limitations of ecological cross-country comparisons (endogeneity, measurement error).
Methods
• Strengths: Transparent description of data sources; multiple model specifications attempted; inclusion of robustness checks.
• Concerns:
1. Outcome modelling: OLS on rates is suboptimal; drowning deaths are count data → negative binomial or Poisson with population offset would be more appropriate. Gamma GLM is not well justified.
2. Covariate selection: Many structural predictors included simultaneously despite ~100 observations; risks overfitting. Consider penalized regression or dimension reduction (e.g., PCA).
3. Regional models (n=19): Severely underpowered for multiple predictors and interactions; these should be presented as exploratory only.
4. Law measurement: Binary presence/absence recoding risks attenuation bias; more nuanced scoring of comprehensiveness or enforcement would be preferable.
5. Causality: Cross-sectional design cannot address temporality; countries may adopt laws in response to drowning burden. This limitation should be made explicit.
6. Diagnostics: No evidence provided of checking residuals, multicollinearity (VIFs), or spatial autocorrelation. These are essential for ecological analyses.
Results
• Strengths: Clearly presented tables; coefficients and CIs reported; random forest adds triangulation.
• Concerns:
1. Interpretation: Emphasis on null law effects is too strong given wide CIs and measurement limitations.
2. Sample size: Regional models produce negative adjusted R²; interpret cautiously or drop from main text.
3. Robustness checks: Outlier exclusion changes significance of key covariates, suggesting instability. These shifts should be highlighted.
4. Consistency: Urbanisation, alcohol, and sanitation repeatedly emerge as significant predictors, this stability deserves more emphasis.
5. Interactions: Law × enforcement interactions have huge SEs and are uninterpretable at n=19; presenting them risks misleading readers.
6. Models 4-6 (n=19): The results of these models should be interpreted with extreme caution. With 10+ parameters and n=19, the models are severely overfit (as indicated by negative adjusted R² values). The coefficients are unstable and unreliable. These results should be de-emphasized in the main text and moved to the supplement with a strong caveat.
Discussion
• Strengths: Honest acknowledgment of some limitations; appropriate recognition of structural determinants.
• Concerns:
1. Overgeneralisation — concluding that “laws do not reduce drowning” is not supported; better phrasing: “we did not detect systematic associations at country level.”
2. Limited comparison to case-based or quasi-experimental evidence showing effectiveness of pool fencing or boating regulations.
3. Policy recommendations are vague; should connect observed results (structural determinants dominate) to integrated strategies combining legislation with infrastructure and enforcement.
4. Small sample and measurement issues are mentioned but downplayed; they deserve central placement.
5. Reframing Null Results: The authors appropriately avoid overstating the null findings, correctly pointing to data limitations and the potential for legislation to be effective in specific contexts. They could strengthen this by more explicitly stating that the study finds no evidence of an association at the ecological level, rather than stating legislation is ineffective. The distinction is important.
6. Limitations: The limitations section is good but should be more forceful on two points: 1) The ecological fallacy should be named and discussed as a primary limitation. 2) The severe power issue with the regional models (n=19) must be highlighted as a major constraint.
Tables and Figures
• Strengths: Informative and mostly well formatted.
• Concerns:
o Forest plots or standardized coefficients would aid comparability of effect sizes.
o Tables with >10 predictors for n=19 regions are not meaningful; consider moving to appendix.
o Random forest importance plots should include uncertainty estimates (bootstrapped importance).
Supplementary Material
• Strengths: Very transparent, showing multiple specifications, data sources, and robustness checks.
• Concerns:
o Gamma GLM results difficult to interpret; justify inclusion or simplify.
o Outlier-exclusion results should specify which countries were dropped.
o Regional analyses should be explicitly marked exploratory.
The article addresses a relevant question and shows commendable transparency. However, methodological and interpretive revisions are required:
1. Reanalyse with count-based models (negative binomial with population offset).
2. Downplay or remove unstable regional/interaction models.
3. Clarify measurement limitations of law variables.
4. Reframe conclusions to avoid overstatement and highlight structural determinants.
5. Expand on diagnostics (collinearity, spatial autocorrelation, influence analysis).
Reviewer #2: This study analyzes global legislation and regulation related to drowning prevention.
Please see my comments below.
Abstract
A scientific study should be replicable. In the abstract, the study type is unclear. Insufficient methodological details are given to the reader to understand what and how it was done?
Introduction
The introduction is written in an op-ed style. Most statements are without references. The literature search and knowledge gap are not clearly defined.
Novelty
This data is utilized from WHO report. What new information is it adding to the literature?
Methods
Insufficient details regarding the 127 countries so we cannot understand who is included in this analysis. Grouping and interpreting high-Income countries and low-income countries together statistically is not the right approach to deduce meaningful results.
Results and Discussion
The findings identify known predictors of drowning, with negative results for legislative measures, ultimately not revealing anything meaningful and novel to add to the literature
**********
what does this mean?. If published, this will include your full peer review and any attached files.). If published, this will include your full peer review and any attached files.
Do you want your identity to be public for this peer review? If you choose “no”, your identity will remain anonymous but your review may still be made public.If you choose “no”, your identity will remain anonymous but your review may still be made public.
For information about this choice, including consent withdrawal, please see our Privacy Policy .-->
Reviewer #1: Yes: Dr.Farhad AliDr.Farhad Ali
Reviewer #2: No
**********
[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]
10.1371/journal.pgph.0005337.r002Author response to Decision Letter 1Submission Version1
30 Sep 2025
Submitted filename: Response.docx
10.1371/journal.pgph.0005337.r003Decision Letter 1KhanUzmaAcademic Editor2026Uzma KhanUzma KhanThis is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.Submission Version1
23 Nov 2025
PGPH-D-25-02400R1
An analysis of global legislation and regulation related to drowning prevention
PLOS Global Public Health
Dear Dr. Essex,
Thank you for submitting your manuscript to PLOS Global Public Health. After careful consideration, we feel that it has merit but does not fully meet PLOS Global Public Health’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.
Please submit your revised manuscript by Dec 23 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at globalpubhealth@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pgph/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.
Please include the following items when submitting your revised manuscript:
A rebuttal letter that responds to each point raised by the editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.
Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.
We look forward to receiving your revised manuscript.
Kind regards,
Uzma Rahim Khan
Academic Editor
PLOS Global Public Health
Journal Requirements:
1. Please note that PLOS Global Public Health has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, all author-generated code must be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/globalpublichealth/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse.
If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.
Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.
Additional Editor Comments (if provided):
The authors response has improved the manuscript but there are additional points to be considered.
[Note: HTML markup is below. Please do not edit.]
Reviewers' comments:
Reviewer's Responses to Questions
Comments to the Author
Reviewer #1: All comments have been addressed
Reviewer #3: All comments have been addressed
**********
publication criteria ? Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe methodologically and ethically rigorous research with conclusions that are appropriately drawn based on the data presented.-->?>
Reviewer #1: Yes
Reviewer #3: Yes
**********
3. Has the statistical analysis been performed appropriately and rigorously?-->?>
Reviewer #1: Yes
Reviewer #3: No
**********
4. Have the authors made all data underlying the findings in their manuscript fully available (please refer to the Data Availability Statement at the start of the manuscript PDF file)??>
The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception. The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.-->
Reviewer #1: Yes
Reviewer #3: Yes
**********
5. Is the manuscript presented in an intelligible fashion and written in standard English??>
Reviewer #1: Yes
Reviewer #3: Yes
**********
Reviewer #1: (No Response)
Reviewer #3: This is an important study with global health implications. The authors used multiple sources of data and with a comprehensive set of structural covariates indicating major socioeconomic, environmental, and institutional determinants of drowning mortality. Also, for exposure, authors considered a variety of measures like the total count, enforcement strength, and their interaction an used multiple modeling strategies to model the relationship with exposure and outcome.
However, these are a few areas especially in design and methodology which if clarified or improved can further strengthen the paper.
Aim 3:"whether strength of enforcement and institutional capacity were influential when it came to the effectiveness of legislation/regulation." Can authors possibly rephrase it in scientific and quantifiable language?
Measurement of Variables: Can authors elaborate how did they measure the extent to which legislative measures were enforced? A clear definition of the enforcement concept is crucial for interpreting the results. Further, to enhance transparency, please provide a detailed list of the nine covariates included, categorized by their domains (e.g., economic, environmental). Additionally, please provide an expanded explanation of the scoring mechanism used to derive the total number of legislative instruments (range 0-19) and the total enforcement score (range 0-190). If possible, can authors include a table of Key explanatory variables, their data source, original form, transformation, and scoring. Can authors spell time the abbreviation the first time they are used such as GDP PPP. Also, how legislative transparency and enforcement translates into V-DEM?
Reference and Citations: Please ensure full citations and/or direct links are provided for all external datasets, including the Varieties of Democracy (V-DEM), World Justice Project (WJP), Global Health Security (GHS), and the World Bank’s World Data Indicators (WDI). Additionally, please provide specific references/citations to support the statement that the control variables were selected based on existing theoretical and empirical evidence on drowning prevention.
Analytic Strategy: Please elaborate on the specific link function used for the outcome variable, which is the drowning mortality rate per 100,000 population. Although OLS was used, which assumes a linear outcome, given that the outcome is a rate derived from count data, please justify the OLS choice over alternative models like Poisson or Negative Binomial regression. Furthermore, please detail the process for checking OLS assumptions (linearity, normality of residuals, etc.) and discuss the theoretical or empirical sources of heteroskedasticity that necessitated the use of robust variance estimators.
The stated goal 'to estimate the impact of legislation' implies a causal objective and thus the methods and language should be consistently aligned. For example, if your goal was to estimate the impact of legislation on national drowning mortality, then legislation is your main primary exposure and has to be included in all nested models. I suggest authors read on Causal Inference in Epidemiology. See this for example: Daniel Westreich, Sander Greenland, The Table 2 Fallacy: Presenting and Interpreting Confounder and Modifier Coefficients, American Journal of Epidemiology, Volume 177, Issue 4, 15 February 2013, Pages 292–298, https://doi.org/10.1093/aje/kws412 and this: https://phlr.temple.edu/sites/phlr/files/documents/CPHLR-TheoryMethods2023_NaturalExperiments.pdf.
I see that authors have multiple aims and that authors used suite of different traditional regression and machine learning analysis. Can authors identify which analysis addressed which objective? Like Exploratory machine-learning classification of drowning risk addressed which objective? How aim 2 and 3 were addressed? Also reconsider model specification again and if possible make a DAG to examine which variable are confounders (so called controls), mediator or collider? Because, if the included covariates (like GDP and urbanization) are colliders or mediators (depending on the specific causal pathway) introducing them could introduce bias or even mask exposure outcome relationship.
Results: Consider streamlining the presentation of results to focus on those that directly address the stated objectives. If authors think information in figure 1 and table 1 is overlapping, then chose to keep anyone of these in main paper and rest can go in supplementary information. Second, for the interpretation of the main model, refer to paper by Westreich referred above, only interpretating the main effect. Third, to provide essential context for the cross-national comparison, please include a table listing the included countries/geographies, alongside two key descriptive features: (1) a measure of inherent risk (e.g., proximity to water bodies, coastline length) and (2) their baseline legislative coverage (e.g., total count/score)
Interpretation/Discussion: Can authors address the possibility that the null finding of no effect of legislation on drowning could be an artifact due to methodology (design and analytic strategy) rather than conclusive evidence.
**********
what does this mean?. If published, this will include your full peer review and any attached files.). If published, this will include your full peer review and any attached files.
Do you want your identity to be public for this peer review? If you choose “no”, your identity will remain anonymous but your review may still be made public.If you choose “no”, your identity will remain anonymous but your review may still be made public.
For information about this choice, including consent withdrawal, please see our Privacy Policy .-->
Reviewer #1: No
Reviewer #3: Yes: Salima KeraiSalima Kerai
**********
[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]
10.1371/journal.pgph.0005337.r004Author response to Decision Letter 2Submission Version2
4 Dec 2025
Submitted filename: Response_auresp_2.docx
10.1371/journal.pgph.0005337.r005Decision Letter 2JosephAlexAcademic Editor2026Alex JosephAlex JosephThis is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.Submission Version2
12 Feb 2026
PGPH-D-25-02400R2
An analysis of global legislation and regulation related to drowning prevention
PLOS Global Public Health
Dear Dr. Ryan Essex,
Thank you for submitting your manuscript to PLOS Global Public Health. After careful consideration, we feel that it has merit but does not fully meet PLOS Global Public Health’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.
Please submit your revised manuscript by February 27th . If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at globalpubhealth@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pgph/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.
Please include the following items when submitting your revised manuscript:
A letter that responds to each point raised by the editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.
Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.
We look forward to receiving your revised manuscript.
Kind regards,
Alex Joseph
Academic Editor
PLOS Global Public Health
Journal Requirements:
1. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.
2. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.
Additional Editor Comments (if provided):
[Note: HTML markup is below. Please do not edit.]
Reviewers' comments:
Reviewer's Responses to Questions
Comments to the Author
Reviewer #3: All comments have been addressed
Reviewer #4: (No Response)
**********
publication criteria ? Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe methodologically and ethically rigorous research with conclusions that are appropriately drawn based on the data presented.-->?>
Reviewer #3: Yes
Reviewer #4: Yes
**********
3. Has the statistical analysis been performed appropriately and rigorously?-->?>
Reviewer #3: Yes
Reviewer #4: (No Response)
**********
4. Have the authors made all data underlying the findings in their manuscript fully available (please refer to the Data Availability Statement at the start of the manuscript PDF file)??>
The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception. The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.-->
Reviewer #3: Yes
Reviewer #4: Yes
**********
5. Is the manuscript presented in an intelligible fashion and written in standard English??>
Reviewer #3: Yes
Reviewer #4: Yes
**********
Reviewer #3: (No Response)
Reviewer #4: Great revisions so far! Almost ready for publication.
Please see below some minor revision suggestions:
-Clarify what kind of “null” this is (statistical vs. policy-relevant), why this matters, this paper is careful, but readers may still misread “no association” as “laws don’t matter." Authors can consider adding 2–3 sentences distinguishing statistical null findings at the ecological level from policy irrelevance (particularly emphasizing power, heterogeneity, and aggregation bias).
-To improve policy translation, the authors might briefly illustrate their findings with a concrete example (e.g., pool fencing in high-capacity vs. low-capacity settings, or alcohol regulation near water in differing enforcement contexts)..no new data required. Just narrative synthesis.
-Recommend authors adding a brief clarification that permutation importance reflects predictive contribution rather than causal influence, to prevent misinterpretation by non-technical readers.
-Consider highlighting earlier in the methods section that death counts were reconstructed from rates and population, and briefly noting how rounding error was handled (for method transparency and replication of the study).
-For data limitations, the authors could add few lines on how national-level aggregation may mask child-specific or occupational drowning effects where legislation is known to be effective..
**********
what does this mean?. If published, this will include your full peer review and any attached files.). If published, this will include your full peer review and any attached files.
Do you want your identity to be public for this peer review? If you choose “no”, your identity will remain anonymous but your review may still be made public.If you choose “no”, your identity will remain anonymous but your review may still be made public.
For information about this choice, including consent withdrawal, please see our Privacy Policy .-->
Reviewer #3: No
Reviewer #4: Yes: Khushbu BalsaraKhushbu Balsara
**********
[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]
10.1371/journal.pgph.0005337.r006Author response to Decision Letter 3Submission Version3
15 Feb 2026
Submitted filename: Response_auresp_3.docx
10.1371/journal.pgph.0005337.r007Decision Letter 3JosephAlexAcademic Editor2026Alex JosephAlex JosephThis is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.Submission Version3
11 Mar 2026
An analysis of global legislation and regulation related to drowning prevention
PGPH-D-25-02400R3
Dear Ryan,
We are pleased to inform you that your manuscript 'An analysis of global legislation and regulation related to drowning prevention' has been provisionally accepted for publication in PLOS Global Public Health.
Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests.
Please note that your manuscript will not be scheduled for publication until you have made the required changes, so a swift response is appreciated.
IMPORTANT: The editorial review process is now complete. PLOS will only permit corrections to spelling, formatting or significant scientific errors from this point onwards. Requests for major changes, or any which affect the scientific understanding of your work, will cause delays to the publication date of your manuscript.
If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they'll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact globalpubhealth@plos.org.
Thank you again for supporting Open Access publishing; we are looking forward to publishing your work in PLOS Global Public Health.