Objectives To examine the change in the racial disparity in drowning in Florida from 1970 to 2015 and to analyse the contextual factors associated with white, black and Hispanic drowning rates in Florida from 2007 to 2015.
Methods Our outcome variable is county-level annual drowning rates by race, ethnicity, sex and age group. We computed county-level contextual data, including emergency weather events, temperature, extreme weather, number of pools, quality of pools, coastline, swimming participation rates and prominent black competitive swim teams.
Results Between 1970 and 1990, the disparity in drowning rates between white and black males in Florida decreased dramatically. By 2005, the overall age-adjusted drowning rates converged. This convergence was most striking for those aged 10–34 and 35–64. While the gap has declined dramatically, there remains a racial disparity in drownings among those aged 10–34.
Conclusions Overall, racial disparities in drowning have disappeared in Florida. However, some disparities remain. There is a persistent disparity in drownings among those aged 10–34.
- time series
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Among people aged under 30 years, drowning is one of the leading causes of death from unintentional injury.1 There are also striking racial disparities in drowning: in 1999–2010, the black drowning rate was 1.4 times the white drowning rate for those under 30.2 Racial disparities in drowning are particularly pronounced for drownings that occur in swimming pools.3 In addition to these striking national patterns, there is an important hidden pattern: in some states, the drowning disparity has fallen dramatically. In fact, there are states where the overall age-adjusted black drowning rate has fallen so much that there is a convergence between the white and black drowning rates.
Figure 1 shows the black–white ratio of overall crude drowning rates nationally and by state in 1979 and 2015. The states shown are those states with public data in both 1979 and 2015 for the black and white drowning rates. Numerous states had large racial disparities in drowning in 1979 that have since disappeared. For example, in 1979, the black drowning rate was over twice the white drowning rate in Florida, Georgia, South Carolina and Mississippi. By 2015, there is no longer a racial disparity in drowning in those states. These states mirror the national trend—a large racial disparity in overall drowning in 1979 that closed by 2015. Other states, including Michigan, Missouri and Tennessee, had a smaller racial disparity in 1979 but continue to have a noticeable disparity in drowning in 2015.
In this paper, we examine Florida as an exemplar case where the overall white and black drowning rates have converged. Florida has a complex history with race and swimming. Prior to the passage of the Civil Rights Act, public pools and beaches in Florida were legally segregated. ‘Wade-ins’ and ‘swim-ins’ at segregated pools and beaches were a common form of protest in Florida during the civil right movement. An iconic moment of the civil rights movement occurred in St. Augustine, Florida, when a hotel owner poured acid into a pool where protesters were staging a wade-in at a segregated hotel pool.4 After the Civil Rights Act was enacted, many public pools were converted to private pools to continue de facto segregation.5 6
During the 1970s, the drowning rate in Florida among black males was over twice that of white males. In the past decades, numerous communities in Florida have undertaken initiatives to reduce drowning. The International Swimming Hall of Fame, located in Fort Lauderdale, has spearheaded efforts to reduce racial disparities in drowning through learn-to-swim programmes. Additionally, four out of the five counties with the highest black male drowning rate in the 1970s have proactively attempted to reduce drowning among children (Lee County,7 Palm Beach County,8 Seminole County9 and St. Lucie County).10 Florida also has a strong history of black competitive swimmers, including Maritza Correia McClendon11 (Tampa Tech) who was the first black woman to break an American swimming record and the first black woman to earn a place on the US Olympic swim team.
We show that the black–white disparity in male drowning rates in Florida has steadily declined since 1970. By the 1990s, the overall racial disparity in male drowning rates in Florida largely disappeared; by 2005, the overall racial disparity was completely gone. The convergence was most striking for those aged 10–34 and 35–64, although a drowning disparity does persist among those aged 10– 34. Importantly, the reduction of the racial disparity in drowning in Florida has occurred despite persistent economic disparities in Florida. Online supplementary appendix 1 shows that the economic disparities based on race have remained constant over this same period.
While the racial disparity in drowning has dramatically declined, Florida still has the third highest fatal drowning rate in the country.1 We used detailed county-level data on drowning rates and numerous contextual measures to examine if there remain different risk factors for drowning among white, black and Hispanic Floridians.
Our key outcome variable is the county-level drowning rate by race, ethnicity, sex and age. The Florida Department of Health publishes yearly death rates for each county of residence by race (‘White’ ‘Black’, or ‘Other’), sex, age and cause of death.12 Beginning in 2004, the Florida Department of Health also included ethnicity (‘Hispanic’ or ‘Not Hispanic’). We analyse the death rates for people who died from accidental drowning by submersion. This includes people who drowned in pools, oceans, boating accidents and bathtubs (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes W65–W74, V90–V99 and Y85). When the analysis is separated by age, some counties’ drowning rates were suppressed. Even when a rate was not suppressed, estimates based on small populations can be unstable.
Because the data was grouped by ICD-10 codes, we were not able to separate out boating accidents specifically—we included deaths from non-military accidents that occur in the water, air and space (V90–V99 and Y85). Similarly, we were not able to separate drownings that occurred in pools from those that occurred in other bodies of water. This drowning measure does not include deaths from floods, which are counted in a broader category of natural disasters. For analysis of all deaths from drowning, we used the age-adjusted death rate. For analysis separated by age group, we used the crude death rate.
Drowning is the result of a combination of factors—exposure to water,13 14 the safety of the swimming area, condition of the water,15 the presence of lifeguards,16 17 use of life preservers18–20 and a person’s swimming ability.21–23 Public policy often focuses on passive measures to reduce accidental exposure to water—for example, fences or pool covers that keep children out pools24–27—and swimming instruction.28 29
To examine if there are different risk factors for drowning between white, black and Hispanic Floridians, we developed measures for these different contextual elements of drowning. To examine factors that increase exposure to water, we calculated numerous measures for hot weather, number of pools, number of boats and the amount coastline. Exposure to water may have multiple counter-acting effects on the drowning rate. Historical and present-day discrimination in access to pools is a large component of racial differences in swimming ability.6 21 22 However, more pools per county is also associated with higher drowning rates.13 That is, being around more pools is associated with higher drowning rates, but limited access to pools also decreases average swimming ability.
The condition of the water is also an important factor in drowning—during a rip current, even a skilled swimmer can be pulled out far from shore.30 31 During a storm, a boater may find themselves in unexpectedly in rough water. We use the number of extreme weather events (including rip currents) and boating accidents to measure this risk factor. The safety of the swimming area is measured through the unsatisfactory pool inspection rate for each county.
Swimming ability is an important factor for reducing drowning.21–23 32 Previous work has shown a strong connection between black competitive swim teams and lower black drowning rates.33 We used the presence of notable competitive swimming programmes oriented at black swimmers to examine if these programmes have reduced the black drowning rate in certain Florida counties. We also estimated county-level prevalence of adult swimming based on self-reported exercise activities. Data on children’s enrollment in learn-to-swim programmes were not available.
Each measure is described in detail in the following sections. The online supplementary appendix 2, table 1A shows the data availability. Unfortunately, we had limited county-level data on preventative factors like pool fences, life guards and use of personal flotation devices.
We collected daily data on temperature from a National Oceanic and Atmospheric Administration (NOAA) weather station in each county from 2007 to 2015.34 We then collapsed this daily data into three measures of temperature for each county in each year: average temperature, number of days where the temperature was over 95° and maximum temperature.
Additionally, NOAA publishes a database that lists each extreme weather event that occurs, including floods and rip currents.35 We collapsed this detailed data into yearly number of extreme weather events in each county in each year.
Florida publishes the number of public pools that were inspected each year and the number of pools that passed the inspection.36 Florida does not inspect all public pools each year, so this data does not contain a census. The number of public pools in each county is listed for 2006,37 and the number of inspected pools in 2006 and the total number of pools registered at each county appraiser’s office are strongly correlated (0.90). Additionally, our analysis focuses on within-county change; this accounts for differences in the proportion of pools a county chooses to inspect. A limitation of this data is that if a county changes the proportion of public pools it inspects, it would appear as a change in the total number of public pools. This likely does not occur frequently. Thus, we used the total number of inspected pools as a proxy for total number of public pools; we also used the proportion of pools that failed the inspection as a measure of pool quality. We have data on number and quality of public pools from 2007 to 2015.
Boats and boating accidents
The Florida Fish and Wildlife Conservation Commission tracks the total number of boats owned in each county in each year as well as the number of boating accidents in each county in each year.38 This data is available from 2007 to 2015.
Time-invariant measures: swimming programmes, swimming prevalence and coastline
The previous contextual data all varied at the county and year level. This allowed us to use county fixed effects and, therefore, examined within-county variation. However, some important measures do not vary with time including the length of coastline,39 overall prevalence of swimming and strong traditions of competitive swimming programmes for black children and teens. These measures have been discussed in depth in the online supplementary appendix 3.
We first examined the total age-adjusted black and white drowning rates in Florida from 1970 to 2015. To account for the fact that black and white Floridians live in different parts of the state, we examined how the within-county difference between black and white drowning rates changes over time. Specifically, we computed the average difference within all 67 counties (i) for sex (s) in each year t:
The differences in drowning rates may obscure changes in the ratio of drowning rates.40 To examine the ratio of drowning rates, we first computed the ratio of age-adjusted drowning rates in Florida for each year. To again account for the fact that black and white Floridians live in different parts of the state, we then computed the average of the ratio for the 67 counties (i) for each sex (s) in each year t:
We then examined the crude drowning rate within age groups, because the racial disparity in drowning varies by age.2 Focusing solely on the total age-adjusted drowning disparity may miss important disparities that exist within age group. We examined the male crude drowning rate for children (0–9), teens and young adults (10–34), adults (35–64) and older adults (65+).
Finally, we examined if there are different risk factors for white, black and Hispanic Floridians. We used a fully interacted multivariable regression, where all contextual variables are interacted with an indicator for the black and Hispanic drowning rates. This shows if the white, black and Hispanic drowning rates are associated with different risk factors. For the regression analysis, we used data from 2007 to 2015, when we had measures for all contextual variables. Because the female drowning rate is consistently low with little racial disparity, the analysis of risk factors focuses on the male drowning rate.
Equation 1 regresses the crude drowning rate in county i for time t for race r on a vector of contextual variables ( ) and an interaction between indicators for black and Hispanic with all the contextual variables. This multivariable regression includes county and year fixed effects and all included contextual variables are measured at the county-year level. SEs are robust to allow for heteroskedasticity. Regressions are run separately for each age group.
We estimated an alternative multivariable regression that includes three time-invariant contextual variables in addition to the time-varying contextual variables. This approach is discussed in the online supplementary appendix 3.
Changes in racial drowning disparities between 1970 and 2015
Figure 2 (top left panel) shows the age-adjusted drowning rate by race and sex in Florida from 1970 to 2015. Since 1970, the drowning rate for black males declined from over 15 deaths per 100 000 to under four deaths per 100 000. For 2005 and later, there is no difference between the overall white and black male drowning rate. Equivalently, the lower left panel of figure 2 shows that ratio of the black–white drowning rates in Florida converges to one in 2005. Female drowning rates have consistently been lower than male drowning rates, with little racial disparity.
The Florida trends shown in figure 2 mirror national trends. Nationally, the overall black male and white male age-adjusted drowning rates were 13.3 and 7.3 deaths per 100 000 in 1979, declining to 1.8 for both black and white males in 2005. The national black–white ratio of age-adjusted drowning rates was 1.8 in 1979, declining to 1 by 2005. The decline in the racial drowning gap nationally and in Florida is due to the dramatic decline in the black male drowning rate and not an increase in the white male drowning rate.
To account for differences in where white and black people live in Florida, we examined the within-county variation in drowning rates. Figure 2 (right panels) shows the average within-county difference between the white and black drowning rates. This follows the same overall pattern found with the raw drowning rates: there was a steady decline in the racial disparity in male drowning, and this disparity disappeared around 2005. Examining the average within-county ratio of drowning rates reveals the same pattern—the within-county ratio of black–white male drowning rates converged to one in the early 2000s.
Examining the age-adjusted drowning rate may obscure remaining disparities within age group. While the racial gap in overall drowning rates has disappeared nationally, it persists among high-risk age groups.2 Figure 3 shows the crude male drowning rates within age group in Florida. Figure 4 shows the ratio of the crude drowning rates within age group. From 1970 to 1990, the racial disparity in drowning in Florida is most pronounced among those aged 10–34 and 35–64. While all age groups have reduced the racial disparity in drowning, a small racial gap persists among those aged 10–34. For male children aged 0–9, the drowning rates for both blacks and whites declined from 1970 to 2000 and thereafter the rates were nearly the same for both groups. However, figure 4 shows that for those aged 0–9, the ratio of black to white drowning rates increased from 1970 to 2000 and thereafter was sporadically close to 1.
Differences in risk factors
Table 1 reveals that the most important contextual variables in explaining male drowning rates among children aged 0–9 are the number of boats and boating accidents. For males aged 10–34, the most important contextual variable is boating accidents. There are no statistically significant (5% level) race effects for these boating variables. The only statistically significant race interactions occur among those aged 10–34 for emergency weather (increases in black drowning rates) and days over 95° (decreases black drowning rates).
We found that the overall white and black male drowning rates in Florida decreased from a dramatic disparity in 1970 to no difference by 2005. Between 2005 and 2015, the overall white and black male drowning rates have remained equal. This pattern was particularly striking for those aged 10–34 and 34– 65. Among these age groups, the black drowning rate was almost three times the white drowning rate in the 1970s. Today, there is no difference between the white and black drowning rate for those aged 35–64 in Florida. Despite the convergence in overall drowning rates, a disparity persists for ages 10 to –34.
Florida has undertaken numerous measures to reduce drowning. Among the five counties with the highest black male drowning rate in the 1970s, we found evidence that at least four have proactively addressed disparities in drowning. Lee County is home to the Lee Pitts Swim School, a nationally recognised organisation promoting swimming skills among black children.7 Palm Beach has strong competitive swimming teams with many black swimmers and was one of only two counties in Florida to consistently send a swim team to the annual DC Black History Month National Invitational. While not focused on black children, other organisations in Palm Beach County, Seminole and St. Lucie County all provide free swimming lessons for low-income children.8–10 We did not find an impact of specific programmes, but together, Florida has dramatically reduced their racial disparity in drowning.
Nationally, there has been a dramatic decline in the overall black–white ratio of drowning rates. However, there is a persistent racial disparity in drowning among young people. Some states have eliminated their overall racial disparity in drowning, while other states’ disparities continue to persist today. To begin to understand this improvement, we closely examined the change in the racial disparity in drowning in Florida overall and within age group. In Florida, the overall racial disparity in drowning narrowed from 1970 to 2005 and has remained negligible since. This improvement was largely driven by dramatic reductions in the racial disparity in drowning among those aged 10–34 and 35–64. While the racial disparities have been eliminated for all age groups combined, a disparity persists for ages 10–34. There are very few differences in risk factors between the white and black male drowning rates for any age group.
What is already known on this subject
There are striking national racial disparities in drowning among young people.
What this study adds
The black drowning rate has dropped dramatically in the past decades in some states, but not others.
We examined Florida as an exemplar case and show that overall racial disparities in drowning have vanished in Florida.
Despite the overall improvement in the drowning rate in Florida, there is still a racial disparity in drowning among those aged 10–34.
Contributors MMG performed the final data analysis and wrote the manuscript. SLM conceptualised the project and oversaw all analyses and writing. YL, DS and RM collected the data, investigated prior literature and performed initial analysis of the data.
Funding We thank the Minnesota Population Center (P2C HD041023NICHD) for funding this research through the Summer Diversity Program.
Competing interests None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data is publicly available. We will freely provide the cleaned and merged data used in this paper along with the statistical program files to reproduce all results.