Article Text
Statistics from Altmetric.com
Mitchell, Williamson and Olivier's1 study estimated drowning rates for the Australian state of New South Wales (NSW) in 2005 based on resident population person-time exposure to swimming. The authors state (p. 261) that ‘failure to adjust injury rates for exposure to a hazard necessarily results in poor estimates of risk’ and, based on their findings, conclude (p. 264) drowning mortality rates to be ‘more than 200 times higher than equivalent exposure-adjusted rates for road traffic fatalities.’ This fact is cited in the Royal Life Saving Society's 2010 National Drowning Report2 (p. 3).
However, comparison data detailed below suggest that the reported time exposure drowning rate presents a gross overestimation of swimming risk. This is due to deficiencies in reported numerator data and denominator estimates.
The majority of unintentional drowning death cases included in the reported numerator would be unlikely to come from the population denominator (swimmers). This is a key principle for calculating population rates.3 Australia-wide unintentional drowning deaths reported for July 2004 to June 2005 identified 97 (37.5%) of 259 victims engaged in a swimming/leisure activity.4 In the following 12-month period (2005–2006), this figure was 54 (20.4%) of 265 unintentional drownings.5 NSW drowning mortality data, as a subset of national data, would be expected to follow a similar pattern. Mitchell et al1 did not report the numerator frequency for drowning deaths.
Mitchell et al1 report a rate of 90 000 drowning deaths per 10 million hours of swimming in NSW for 2005 (p. 264). The rate indicates that for every thousand hours of swimming at a NSW public pool, river, beach or other location, nine swimmers will drown. However, Sydney's Bondi beach alone caters to thousands of bathers most days during summer—yet drowning remains a relatively rare, and not daily, event. Drowning protection at this beach, given the risk of swimming indicated by the reported rate, may be explained by regular surf lifesaver patrols. Even so, the NSW population, which numbers many millions concentrated in coastal areas experiencing a mild to warm climate, will likely spend numerous hours swimming in unpatrolled locations.
The reported rate of 90 000 drowning deaths per 10 million hours of swimming must be questioned given that the annual frequency of drowning among swimmers in NSW is probably well below 100. Fortunately, a check of this rate is readily available using swimming participation data reported by the Australian Bureau of Statistics (ABS)6 for 2005–2006. These data were collected using a method and time period consistent with the exposure data relied upon by Mitchell et al.1
In the previous 12 months for Australia (2005–2006), 1 447 300 persons aged 15 years and over residing in private dwellings (9% of the national population) were estimated to have participated in organised (13%) or unorganised (87%) swimming. (The swimming participation rate for NSW was estimated by the ABS to be marginally higher at 10% of the population or 556 400 persons.) For national swimmer estimates in the 12 months prior to sample interview, 1.1 million (76%) reported swimming 53 times or more;186 300 (13%), 27–52 times; 89 500 (6%), 13–26 times and 71 500 (5%), 12 or less times. This estimate equates to a minimum of 64 565 100 swimming episodes in Australia by residents for the 12-month period.
Based on minimum participation frequencies reported by the ABS, I have estimated person-time exposure and drowning rates (using a numerator of 80 drowning deaths while swimming—the average over a 5-year period to 20065) for mean bathing-time exposures per swimming episode of 30 min. The calculation was based on conservative estimates (lowest swimming frequency by frequency group) and ignores what is likely to be millions of bathing hours undertaken by international visitors to Australia (note that drownings of international tourists were included in numerator data). This produced a rate of 24.8 drownings per 10 million hours of swimming. (Note: Mean swimming episodes at 15 min yielded a rate of 49.6 and, at 60 min, 12.4 per 10 million hours of swimming.)
Substantial differences between the person-time exposure rate reported by Mitchell et al1 and that listed above (respectively, 90 000 drowning deaths in NSW compared with 25 drowning deaths in Australia per 10 million hours of swimming) are unlikely to be explained by differences in water exposure or drowning patterns between NSW and other Australian states and territories. Perhaps Mitchell et al applied sample exposure data in the denominator without extrapolation to the population.
In comparison with reported traffic mortality time exposure rates, the time exposure rate of drowning mortality for persons exposed to swimming in Australia appears higher, but rather than being 200 times higher as reported in the study, it's more likely to be a factor below 10.
Mitchell et al1 rightly state (p. 264) that ‘in terms of policy development, under- or overestimation of the true risk of injury can lead to poor identification of priorities for developing injury prevention policies and interventions, and inadequate resource allocation.’ The reported rates emphasise the need for precision so as not to perpetuate these challenges to injury problems including drowning.
A spreadsheet with supporting data is available on request.
Footnotes
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.