Article Text
PDF
Abstract
Objective To describe the trends of drowning mortality in Vietnam over time and to identify socioeconomic characteristics associated with higher drowning mortality at the provincial level.
Methods We analysed data from the Ministry of Health injury mortality surveillance system from 1 January 2009 to 31 December 2013. The surveillance covers more than 11 000 commune health centres in all provinces of Vietnam. For provincial population and socioeconomic characteristics, we extracted data from the National census 2009, the Population change and family planning surveys in 2011 and 2013. Multilevel linear models were used to identify provincial characteristics associated with higher mortality rates.
Results Over the 5-year period between 2009 and 2013, 31 232 drowning deaths were reported, equivalent to a 5-year average of 6246 drowning deaths. During this period, drowning mortality rate decreased 7.2/100 000 to 6.9/100 000 (p=0.035). Of six major geographical regions, Northern midland, Central highland and Mekong delta were those with highest mortality rates. In all regions, children aged 1–4 years had the highest mortality rates, followed by those aged 5–9 and 10–14 years. At provincial level, having a coastline was not associated with higher mortality rate. Provinces with larger population size and greater proportion of poor households were statistically significantly associated with higher mortality rates (p=0.042 and 0.006, respectively).
Conclusion While some gains have been made in reducing drowning mortality, child deaths due to drowning in Vietnam remain alarmingly high. Targeted scale-up of known effective interventions such as child supervision and basic survival skills are needed for reducing child mortality due to drowning, particularly in socioeconomically disadvantaged provinces.
- Drowning mortality
- epidemiology
- Vietnam
- low and middle-income country
Statistics from Altmetric.com
Introduction
In 2016, drowning led to 321 000 deaths worldwide1 and accounted for nearly 9% of the total global injury disability-adjusted life years.2 WHO estimated that 91% of global drowning deaths are in low-income and middle-income countries (LMICs). Globally, drowning rates are highest in LMICs in the WHO African, South-East Asia and Western Pacific regions.1
Vietnam is one of the countries with an extremely high burden of drowning in the Western Pacific region. It has more than 3000 km of coastline, a wide network of rivers, lakes and ponds, and is located in a region of Asia with the tropical monsoon climate.3 The levels of exposure to bodies of water pose a constant risk of drowning, particularly for children due to their insufficient levels of risk consciousness and water-safety knowledge. Frequent floods in the monsoon season and typhoons further compound the situation.
The high burden of drowning in Vietnam was recognised following the first national survey on injury in 2002. For all age groups, drowning was found to be the second leading cause of injury-related death with a rate of 22.6/100 000 population, second only to road traffic injuries (26.7/100 000). Among children aged 0–19 years, drowning was the leading cause of death (39.2/100 000).4 The extremely high rate of drowning led to significant efforts from government and international organisations to address the issue. Some of these included the introduction of national policies on injury prevention (Decision 197/2001/QĐ-TTg, Resolution 2158/2013/QĐ-TTg, Resolution 234/2016/QĐ-TTg); establishment of the national injury mortality surveillance system managed by the Ministry of Health to capture injury-related deaths, including drowning,5 in the entire primary healthcare network of more than 11 000 commune health centres; communication programmes on water safety implemented by the Ministry of Labour, War Invalids and Social Affairs6; a childhood injury prevention programme with a component for swimming lessons for school students implemented by Unicef between 2003 and 2007 in 24 selected communes in six provinces5; and swimming lessons implemented by The Alliance for Safe Children between 2006 and 2009 in Da Nang city.7
To date, there has been little evaluation of the national efforts to prevent drowning in Vietnam. Understanding the epidemiology of drowning in terms of person, place and time is essential to guide on-going and future drowning reduction programmes, both in Vietnam and in similar settings. The present study reports 5-year trends 2009–2013 of drowning mortality using data from the national injury mortality surveillance system and factors associated with higher drowning mortality at provincial level.
Method
Data and sources of data
Drowning data
Drowning mortality data from 1 January 2009 to 31 December 2013 were extracted from the Ministry of Health injury mortality reporting system (commonly known as A6 system in Vietnam). Extracted cases were accidental non-transport drowning and submersion, ICD-10 codes of W65 to W74. Data in 2013 were the latest validated available data at the time of our data request in 2017. The A6 mortality reporting system is coordinated and managed by the Ministry of Health. Mortality data are collected from more than 11 000 commune health centres covering all geographical areas of Vietnam.8 Officials in commune health centres are responsible to record basic demographic data (including age and gender) and the information on the cause of death in a mandatory book known as A6. Data are collated by the district-level health service and then forwarded to the provincial (the provincial department of health) and central level (the Ministry of Health). Health officials at commune level play an essential role in maintaining the system and in turn also use the information collected to plan health services at the commune level. A study evaluating the performance of the A6 system reported data completeness of 93.9% with a sensitivity of 75.4% and a specificity of 98.4%.9
Population data
Population data (ie, the denominators for rate of drowning estimations) were sourced from the General Statistics Office (GSO) of Vietnam. We extracted publicly available aggregated data from the National Census 200910 and the Population Change and Family Planning Survey in 201111 and 201312 as population denominator to match with the years of drowning data. In Vietnam, national census is conducted every 10 years. Between censuses, a nationally representative sample survey is conducted every 2 years to update the population size and general socioeconomic status. For the estimates of drowning mortality rates, we extracted the total numbers of people by sex and by age groups (<1 year, 1–4 years, 5–9 years, 10–14 years, 15–24 years and 25+ years) for individual provinces. For our analyses of association between drowning mortality rates and socioeconomic characteristics at provincial level, we also retrieved provincial population density (population/km2), the percentage of school students (the proportion of total number of primary and high school students to the population) and the percentage of households in poverty for all 63 provinces in Vietnam. In 2009, the poverty line for the period 2005–2010 was applied: poor households in rural area were those with an average income under VND 2.40 million per capita per year (~US$150) and poor households in urban area were those with an average income under VND 3.12 million per capita per year (~US$195).13 In 2011 and 2013, the poverty line for the period 2011–2015 was applied: poor households in rural and urban areas were those with income of less than VND 4.80 million (~US$228) and 6.00 million (~US$288) per capita per year, respectively.14
For the years 2010 and 2012 in which there was no population survey, population estimates were averages of those in 2009 and 2011 (for 2010) and averages between 2011 and 2013 (for 2012).
Statistical analyses
The age-standardised mortality rates of drowning were estimated using the direct standardisation method. We used the WHO world standardisation population as the reference population to perform the direct standardisation method.15 We employed Dobson et al’s method to estimate CI for the age-standardised mortality rates.16 The estimated mortality rates of drowning were presented by year, sex, age groups and six major geographical regions classified by the GSO.
To examine the change in the mortality rates of drowning over time in 63 provinces and the association between drowning mortality rates and socioeconomic characteristics at provincial level, we used linear mixed models. These models were used due to the hierarchical nature of our data with the mortality rate as a continuous outcome estimated multiple times between 2009 and 2013 (level 1) nested within province (level 2).17 A p value of less than 0.05 was considered statistically significant. All analyses were performed using Stata statistical package V.12 (Stata Corporation, College Station, Texas, USA).
Results
Over the 5-year period between 2009 and 2013, a total of 31 232 drowning deaths were recorded by the A6 mortality reporting system, equivalent to a 5-year average of 6246. Compared with the actual number of drowning deaths in 2009, there was a small increase in 2010 and a large increase in 2011 (more than 500 deaths). After that, the number of deaths gradually decreased in 2012 and again in 2013 (figure 1). Similarly, the age-standardised rate of drowning mortality per 100 000 increased in 2011 before declining to 6.9 in 2013. The decrease in drowning mortality rate, from 7.2 per 100 000 in 2009 to 6.9 per 100 000 in 2013, was statistically significant (p=0.035).
Number (n) and age-standardised drowning mortality.
The monthly average drowning numbers and age-standardised rates of drowning mortality are reported in figure 2. The rates were at the lowest level, between 4.2 and 5.5 per 100 000, during the winter months (November and February) and at the highest level, 9.9 per 100 000 in July, during the mid-summer in Vietnam.
Monthly average number (n) and age-standardised drowning mortality.
Drowning mortality rates by age group over time between 2009 and 2013 for males and females are shown in figure 3. In both males and females, mortality rates were highest among children aged between 1 and 4 years (25.5 and 15.4 per 100 000 in 2009 in males and females, respectively), followed by children aged between 5 and 9 years (16.6 and 7.8 per 100 000 in 2009 in males and females, respectively), and children aged between 10 and 15 years (14.2 and 7.2 per 100 000 in 2009 in males and females, respectively). The mortality rates were lowest among children aged less than 1 year (1.7 and 1.1 per 100 000 in 2009 in males and females, respectively). Over time, all age groups in males and females experienced the highest drowning mortality in 2011 then gradually decreased in 2012 and 2013. The level of decrease between 2011 and 2013 was largest in the males between 5 and 9 years (1.7 per 100 000) and females between 1 and 4 years (2.1 per 100 000).
Age-specific drowning mortality (/100 000 population) over time for males and females.
Vietnam is composed of 57 provinces and 6 centrally governed cities, which are at the same administrative level as provinces. The GSO of Vietnam further groups these provinces and cities into 6 regions, namely the Northern midland (14 provinces), the Red River delta (2 cities and 9 provinces), the Central coast (2 cities and 12 provinces), the Central highland (5 provinces), the South east (1 city and 5 provinces) and Mekong delta (1 city and 12 provinces). Figure 4 shows the trend of standardised drowning mortality over time in six GSO regions. Throughout the period between 2009 and 2013, South East was the region with the lowest drowning mortality rates. In 2009 and 2010, the region with the highest rates was Central highland; and from 2011 to 2013, it was Northern midland. While majority of regions experienced lower mortality rate in 2013 than that in 2009, South east and Northern midland were those that had mortality rate in 2013 higher than that in 2009.
Age-standardised drowning mortality (/100 000 population) in six major General Statistics Office regions in Vietnam.
The age-specific drowning mortality rates by sex between 2009 and 2013 in six GSO regions are presented in table 1. Consistent with the trend across age groups presented earlier, in all GSO regions, the mortality was lowest in the group aged less than 1 year, reached the highest level in the age group between 1–4 years and 5–9 years, and then decreased with age. Across regions, South east was the one with lowest mortality rates in all age groups and in children 0–19 years (both males and females). Central highland had the highest mortality rates in the age groups less than 1 year, 5–9 years and 10–14 years, all age groups combined and in children 0–19 years (both males and females). Mekong delta was the region with the highest mortality rate in the age group 1–4 years. For older children and adults, the drowning mortality rates were highest in Central coast region.
Age-specific drowning mortality (/100 000 population) between 2009 and 2013 by General Statistics Office (GSO) region
Table 2 presents the output of the models examining associations between the provincial age-standardised drowning mortality rate (per 100 000 population) and provincial demo-socio-geocharacteristics. There was a marginally statistically significant decreasing trend of drowning mortality over time of 0.173 per 100 000 per year (p=0.06). Provincial characteristics found to be associated with increased drowning mortality were population size and the percentage of households in poverty. For every increase of 10 000 people in the provincial population, there was an average increase of 0.018 per 100 000 in the drowning mortality rate (p=0.042). In terms of the percentage of households in poverty, for each increase of 10% of poor household in a province, there was an average increase of 1.34 per 100 000 in the drowning mortality. Population density, the percentage of school student (primary and high school) in the population and having a coast line in a province were not statistically associated with the drowning mortality in a province.
Multilevel mixed model of the provincial age-standardised drowning mortality (/100 000 population) and some provincial demo-socio-geocharacteristics
Discussion
To our best knowledge, while injury deaths are routinely collected by all commune health centres in Vietnam, this is the first time that drowning mortality has been examined over time. While only a small change, we found a statistically significant decrease in drowning mortality from the year 2009–2013. In an analysis of WHO mortality databases between 2000 and 2013, Wu et al found 20 out of 21 included countries with decreasing drowning mortality rates among children and adolescent (aged 20 years or less).18 To some extent, our finding could be an evidence of the efficacy of the efforts by the government agencies and international organisations to reduce the burden of drowning in Vietnam. Interventions implemented by international organisations (such as Unicef, The Alliance for Safe Children, Swim Vietnam, Water Safety and others) and government organisation (Ministry of Health; Ministry of Labour, Invalids and Social Affair; Water transport administration) have collectively contributed to the reduction in drowning mortality rates. While this is an encouraging outcome, at the lowest level of drowning mortality in 2013 (the age-standardised rate of 6.9 per 100 000), it was still more than three times the mortality in high-income countries (approximately 2.2 per 100 000) and higher than the drowning mortality in the WHO Western Pacific region (approximately 5.0 per 100 000).1
The high drowning mortality during summer months, between May and August, provides evidence of Vietnamese parents’ concern that summer is ‘drowning season’.19 In Vietnam, summer months are also the period that school children have their school holiday. It is not uncommon that while parents are at work, their children are more likely to play without adult supervision.3 Unsupervised play around water, poor awareness of risk and lack of water-safety skills contribute to the drowning risk.20 In addition, the time between May and August is also the time of increasing typhoons, heavy rain falls and overflow of water in many provinces in Vietnam, particularly those from the Central coast region and up towards north of Vietnam. The seasonal pattern of increase in drowning mortality rate during monsoon or summer season (between May and August) was also found in Bangladesh, India and Thailand.21–23
We found children between 1 and 14 years were at the highest risk of drowning mortality, consistent with patterns previously reported in the WHO world drowning report1 and a recent update on drowning epidemiology in Bangladesh using data from a national survey in 2013 (with highest mortality rate in children aged 1–4 years followed by those aged 5–9 and 10–14 years). Since the first national injury prevention in 2002 in Vietnam, significant efforts have been focused on this particularly age group. The Unicef childhood injury prevention programme resourced interventions to improve supervision for small children as well as swimming lesson for primary school children (aged 6 to 10 years).5 Other programmes implemented by The Alliance for Safe Children, Swim Safe Vietnam and a number of other organisations included swimming lessons for children were in this age group.7 Recently, in the national programme on injury prevention among children, in addition to survival swimming lesson for children, there are also targets to improve children and community worker’s knowledge about water safety, first aid and resuscitation skill.24 However, without more rigorous evaluation, it is not possible to ascribe cause and effect.
In fact, evaluation of drowning intervention is a critical area that WHO highlighted in their recent Implementation guide for drowning prevention.25 To date, evidence of effective drowning intervention is mainly from high-income countries. The only one study conducted in a LMIC that we are aware of is the Cost effective analysis of the Prevention of Child Injuries through Social-Intervention and Education (PRECISE) programme in Bangladesh. In this study, Rahman and colleagues found that village-based child care centres and swimming lessons are highly cost-effective interventions (US$812 per disability-adjusted life year (DALY) averted for child care centre and US$84 per DALY averted for swimming lesson).26 Further evaluation is needed to improve our understanding of whether and how drowning prevention interventions can be adapted for LMIC settings, and consequently to further support and reduce the disproportionate burden of drowning in LMICs.
The finding that the drowning mortality is higher in provinces with higher proportions of poor households is an important one. The evidence of higher risk of fatal drowning among people of lowest socioeconomic status or low income was also reported in multiple population surveys in Bangladesh.22 27 Prior research indicated that economic hardship is a factor that results in lack of adult supervision for children because parents have to work long hours; children commence work at younger ages and are exposed to the risk of drowning; and they are less likely to be able to participate in intervention programmes, such as swimming lessons, due to their work engagement.3 27 This finding highlights a need to further explore to prioritise provinces or areas with high proportion of poor household and possibilities to better engaged poor families into intervention programme on drowning prevention.
While we accessed the most comprehensive drowning mortality data available, our study is not without limitations. First, the data that we obtained from the Ministry of Health are still nearly 5 years old. However, validated data to 2013 were the latest that the ministry could provide at the time of our data request. Lack of resources within the division managing the A6 system limited their ability to provide more updated data. The timeliness of data availability is critical for monitoring and evaluation to guide future intervention activities, and the Ministry of Health needs to consider measures to improve and to maintain the workforce for the injury mortality surveillance system. The second limitation in our study is the lack of detailed data at the individual level. The A6 system limited data elements to be collected and we had used most of those including age, gender and provincial location of the drowning. In addition, A6 system only recorded the cause of death by broad external cause for injuries (for instance, ICD-10 codes of V01 to V99 for road traffic injuries, W65 to W74 for drownings, or W00 to W19 for fall-related injuries), which limit our ability to identify drowning related to natural disaster. We also had limited capacity to obtain a more comprehensive list of sociodemographic and geographical characteristics for provinces and used publicly available data from the GSO of Vietnam.
Conclusion
The decreasing drowning trend in Vietnam evident in this analysis is encouraging and it is likely the extensive efforts from government agencies and international organisations have contributed to these positive outcomes. However, the drowning mortality in Vietnam is still among the highest in the Western Pacific region and three times the mortality in high-income countries, so more work remains. If Vietnam is to achieve the UN Sustainable Development Goal of ensuring good health and well-being for all, investment to address child drowning must continue.
What is already known on the subject
The burden of drowning is extremely high in low-income and middle-income countries.
Drowning prevention has been a priority policy area in Vietnam.
What this study adds
Drowning mortality is on a decreasing trend in Vietnam.
Provinces with larger population, higher proportion of households in poverty are associated with higher drowning mortality rates.
Efforts of drowning prevention in Vietnam resulted in encouraging outcome; however, scale-up of efforts is needed for sustained benefit.
Acknowledgments
This research was a part of the ‘Burden of drowning and opportunities for drowning reduction in India and Vietnam’ project. We are grateful for Royal National Lifeboat Institution for the kind financial support for this research activity. We also gratefully acknowledge General Department of Preventive Medicine, Ministry of Health in providing drowning mortality data.
Footnotes
Contributors JJ was the lead chief investigator of the project. All authors contributed to the design of the study. HN analysed the data and drafted the manuscript. JJ, RQI and CP provided feedback and comments for revision.
Funding The study was funded by the Royal National Lifeboat Institution in a project to examine the burden of drowning and opportunities for drowning reduction in India and Vietnam.
Disclaimer The funder has no involvement in the study design; collection, analysis and interpretation of data; the writing of the manuscript; the decision to submit the manuscript for publication.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The ethics application for this study was reviewed and approved by the Hanoi School of Public Health Institutional Review Board (Protocol 017-260/DD-YTCC).
Provenance and peer review Not commissioned; externally peer reviewed.
