Understanding Natural Disaster or Weather-Related Drowning Deaths Among Children

We used data from the NFR-CRS, a Web-based application that collects standardized, case-specific data on circumstances of child deaths reviewed by CDR teams. Detailed information on the scope and limitations of the database have been previously described.¹⁰ 

From 2005 to 2021, there were 203 869 deaths collected in the NFR-CRS among children ages 0 to 17 years. A multistep process, detailed in Fig 1, was used to identify the disaster/weather-related drowning deaths. First, text fields and narratives for all deaths that were attributed to a natural disaster- or weather-related incident were searched for “warning,” wind, storm, or flood, and 26 cases were identified. However, this question has only been part of the NFR-CRS since 2018, so it has not been asked for all deaths entered.

Second, text fields for motor vehicle crash deaths with an incident type of “child in/on vehicle that ran off the road” or “other event” were searched to identify vehicles driving into waters or being swept away (using terms flood, drown, water, rain, and storm) and were only included if related to a natural disaster or weather incident (n = 16). The drowning location and type of water were then recoded on the basis of details provided in text fields.

Third, text fields describing location, last activity, and narrative were searched for flood, warning, storm, weather, heavy, rain, wash, monsoon, hurricane, swollen, fast, and swept in any drowning deaths. In these 86 identified deaths, we recoded the drowning location for instances when users chose “other, specify” for drowning location, but the location was available as an existing response option (n = 30).

Fourth, fatalities from a cause of death of electrocution caused by lightning or other source were reviewed by 1 author for mention of drowning as a contributing cause of death. Finally, deaths that were attributed to an exposure to water hazards were reviewed by 1 author to identify any deaths not previously included in the above categories. No additional deaths were identified in these efforts.

Some of the above searches identified indirect deaths, but narratives from all drowning deaths, including those in pools, hot tubs, and bathtubs, were searched using key words flood, storm, weather, heavy, rain, wash, monsoon, hurricane, swollen, fast, and swept to identify all deaths indirectly attributable to a natural disaster/weather-related incident.

In total, we identified 130 directly attributed disaster/weather-related drowning deaths and 14 more that were indirectly attributed to such incidents. In instances where the NFR-CRS user selected motor vehicle crash as the primary cause of death or other for drowning location, this limited what additional data about the drowning were collected. Denominators vary in the results for this reason.

Frequencies and proportions were calculated to summarize the characteristics of the study cohort. The demographics and characteristics of the direct disaster/weather-related drowning group were compared with all other drowning not directly related to a disaster- or weather-related incident entered into the NFR-CRS (hereafter referred to as other drowning, n = 7488). A separate descriptive analysis was conducted for 14 deaths that were indirectly related to a natural disaster- or weather-related incident.

Statistically significant differences between a disaster/weather-related drowning and other drowning deaths were identified using Pearson’s χ² tests. Statistical significance was deemed P < .05. All analyses were carried out using the Statistical Package for Social Sciences, Version 29.

In the disaster/weather-related drowning cohort of 130 deaths, more than one-third were children ages 5 to 9 years old (35%) (Table 1). The largest proportion of decedents were white (71%), non-Hispanic (85%), and male (61%) (Table 1). Age distributions differed between the cohorts (Table 1), with older children representing a greater proportion of the 130 disaster/weather-related drowning deaths compared with the 7488 other drowning deaths (ages 5–9: 35% vs 15%; ages 10–14: 23% vs 11%; ages 15–17: 22% vs 15%, P < .001). Although infants and children ages 0 to 4 years old accounted for the highest number of deaths among the 7488 other drowning deaths (59%), this age group accounted for the smallest number of the 130 disaster/weather-related drowning deaths (19%, P < .001). Hispanic ethnicity was less common among the 130 disaster/weather-related drowning deaths (13% vs 19%, P = .007). Counts of disaster/weather-related drowning for type of disability were too small to report (n < 6) per NFR-CRS guidelines; however, all disabilities identified were either physical or cognitive/intellectual. Death scene investigations were conducted in 94% of the 130 disaster/weather-related drowning deaths, compared with 83% of the 7488 other drowning deaths (P = .008) (Table 1).

Comparing the 130 disaster/weather-related drowning and 7488 other drowning deaths, there were significant differences in incident location of the child’s home (7% vs 35%, P < .001), as well as incident location being a roadway (22% vs 1%, P < .001) (Table 1). All 130 (100%) disaster/weather-related drowning deaths occurring in sources of open water. The most frequent type of open water in the 130 disaster/weather-related drowning deaths were creeks (32%), followed by rivers (28%) (Table 2). There were significant differences in geographic locations, with 63% of the 130 disaster/weather-related drowning deaths taking place in rural or frontier locations compared with 30% of the 7488 other drowning deaths (P < .001) (Table 1).

Sixty-one percent of the 130 children who died in a disaster/weather-related drowning were supervised at the time of the incident compared with 38% of the 7488 other drowning (P < .001) (Table 1). Although differences between the frequency of rescue attempts in these incidents was not significant (62% vs 58%, P = .167), there was a significant difference in the incidence of an attempted rescuer also drowning in a disaster/weather-related drowning death (13% vs 3%, P < .001).

Thirty-five percent of the disaster/weather-related drowning deaths reported 2 or more children dying in the incident compared with 5% of the other drowning (P < .001) (Table 1); and 33% reported 1 or more adults perishing in the same drowning incident, compared with 3% of other drowning (P < .001) (Table 1). On the basis of information in text fields, 41 (32%) of disaster/weather-related drowning deaths were attributed to 14 unique incidents.

In some instances, text fields from the disaster/weather-related drowning deaths indicated that the child or their household was in the process of evacuating at the time of the incident. This count was too small to report (n < 6). In 28 disaster/weather-related drowning deaths (22%), text fields mentioned that warnings (such as road closed barriers, closures of schools because of weather, or weather service bulletins) were disregarded. Additionally, there were 7 instances (5%) where text fields indicated that areas with atypical water levels were sought out for viewing or recreation.

Unlike the disaster/weather-related drowning cohort, the majority (79%) of the deaths indirectly attributed to a natural disaster- or weather-related incident (n = 14) were among younger children (ages 0–4 years old) and the water source was most frequently a pool (86%). The most frequent manner of indirect death was when a storm such as a hurricane damaged fencing or gates, allowing the decedent to access the pool. Eight of the indirect deaths (57%) occurred in pools that were not part of the child’s residence.

Children are especially vulnerable during natural disaster- and weather-related incidents, but little is known about details surrounding disaster/weather-related drowning deaths among this group. The findings from this research are unique in using CDR data to more closely investigate circumstances of child drowning during natural disaster- and weather-related incidents in the United States.

Multiple deaths occurred in multicasualty incidents, including deaths of individuals attempting to rescue a child. The higher proportion of rescuers drowning in disaster/weather-related drowning deaths indicates that supervisors or bystanders may have ignored dangerous conditions to attempt a rescue. The smaller proportion of disaster/weather-related drowning deaths among the youngest group (ages 0–4 years) may reflect that increased supervision during natural disasters and weather-related incidents is protective for children in this age group and less so for older children.

Previous research in Australia identified peak risk for children dying in flood-related drowning among 10- to 14-year-olds and described behavioral factors contributing to risk, including older adolescents swimming in flood waters or driving into flooded remote areas.^11,12 On the basis of narratives in this study, there were multiple instances when either children or supervisors sought out flooded areas to see or experience the excess water leading to a larger proportion of deaths among children ages 5 years or older compared with other drowning. A supervisor leading children to water during a natural disaster or weather-related incident for recreating adds a more complicated layer to the problem, because it may suggest to the child that it is safe to visit these areas.

Approximately 1 in 5 (22%) disaster/weather-related drowning deaths expressly mentioned warnings such as road closures, closures of schools because of weather, or weather service bulletins being disregarded by the child or supervisor. This disregard for safety was also reflected in the motor vehicle specific data where the child was a passenger and not the driver in a majority of instances. This phenomenon has been seen in previous research, which identified that initial responses to emergency alerts often involve skepticism, disbelief, and denial, leading people to ignore warnings.¹³ 

Rural and frontier drowning locations were more common in the disaster/weather-related drowning cohort. Discouraging people from driving through floodwaters may be particularly challenging in rural areas and in instances where people are attempting to evacuate a natural disaster or weather-related incident.¹² In an Australian survey, 36% of people self-reported having driven through flood waters, with people living in rural areas even more likely to report having done so.¹² Findings suggest prioritizing tailoring emergency alert strategies to frontier areas with limited infrastructure.

Several indirectly related disaster/weather-related drowning deaths occurred during storm recovery when broken pool fences or other layers of protection were compromised. It may be that replacing these barriers was not prioritized in the wake of a hurricane or other storm when the broad damage to property was overwhelming. Prioritizing repair of barriers and layers of protection around pools and other residential bodies of water may prevent indirectly related drowning deaths of children after disasters.

The findings of this study have implications for pediatric and family medicine providers, public health practitioners, and preparedness professionals who work with children, youth, and families. Professionals can ensure that preparedness messages are tailored to adolescents and their caregivers. Messaging to adolescents and caregivers could emphasize avoiding flooded areas and swift moving water in natural disasters and weather-related incidents. Despite the fact that pediatric providers have many things to do and cover in relatively brief clinical encounters, they are uniquely positioned to educate families about emergency preparedness, including drowning, as part of routine preventive health care as recommended by the American Academy of Pediatrics,¹⁴ in addition to any other drowning prevention guidance they share, such as that from the Centers for Disease Control and Prevention.¹⁵ Their ability to address these and other issues are challenged by workforce shortages, particularly in rural areas where these deaths commonly occur.¹⁶ 

The impact of climate change should not be understated. Data from the past century show a rise in extreme precipitation events in the United States¹⁷ and climate modeling studies predict an increase in the frequency and intensity of extreme precipitation events, hurricanes,¹⁸ and floods¹⁹ as a result of global warming.²⁰ Increasing climate-related changes that may increase flooding and drowning risk underscores the need for disaster/weather-related drowning specific prevention strategies that include monitoring and mitigating environmental risk factors.

Better data drives more effective prevention strategies. Resources and funding dedicated to comprehensive standardized death scene investigations in disaster/weather-related drowning deaths could improve reporting on the circumstances of these deaths and development of tailored prevention. Standardized tools and investigation protocols could guide practice for death certifiers in making death determinations.^6,21 

Given the findings of this study, future research could inform understanding of the role of supervision in natural disasters and weather-related incidents, including its relationship to age. Rigorous testing that is informed by behavioral science could identify what types of prevention messages effectively resonate with families, adolescents, and rural residents in particular. A broad assessment of homeowner insurance policies may shed light on timelines for repair of fences and other barriers that limit access to residential pools after natural disasters and weather-related incidents. Finally, a nationally representative case registry for all drowning deaths could enable the collection of detailed demographic and circumstance data to inform prevention strategies.

This study was limited by small numbers. Additionally, jurisdictions have different case selection criteria for reviewing deaths in CDR and differences in data they share, so although data from the NFR-CRS can meaningfully inform prevention strategies, they are not population-level and cannot be used for population-level surveillance. Additionally, this analysis included a significant amount of text and qualitative analysis from many different CDR teams, so there were frequent inconsistencies that required recoding when possible. Keyword searches may not have identified all relevant case summaries because of differences in data entry such as misspellings of words or colloquial terms. There were instances where text narrative described more child deaths taking place in a single incident than were identified from the NFR-CRS.

Defining a natural disaster or weather-related incident on a national scale was difficult because there can be regional variation in average rainfall. A disastrous amount of rain in 1 location may be a normal amount in others. Efforts were made in qualitative analysis of text fields to distinguish disaster-level flooding, rain, or storms from a day with average rain when determining cohort inclusion. In light of this, retention ponds, storm drains, and canals were especially challenging to assess and could have been missed in the case selection. The consistency and quality of what was written in narrative fields can differ between jurisdictions, so the true count of deaths from natural disaster or weather-related incidents could be greater.

Given the descriptive nature of this study, we did not attempt to identify or control for potential confounding variables. Because of high proportions of missing or unknown data for both insurance status and whether caregivers were recipients of social services, this study was not able to describe socioeconomic status among the disaster/weather-related incident drowning cohort.

Disaster/weather-related drowning deaths have unique characteristics that may require tailored prevention strategies. With the addition of a new variable in 2018 to more reliably track disaster-related deaths, including those from natural disasters and weather-related incidents, the NFR-CRS may be a more informative data source to describe deaths before and after natural disaster- and weather-related incidents in years to come. Improving data quality around disaster/weather-related deaths may inform more effective prevention and preparedness planning at organizational, community, and family levels.

CDR

child death review

NFR-CRS

National Fatality Review Case Reporting System