A

mswoa(lseoe©

K

1

enIcsabfttte

1d

Available online at www.sciencedirect.com

Journal of Science and Medicine in Sport 12 (2009) 457–462

Original paper

Direct observation measurement of drowning risk exposurefor surf beach bathers

Damian Morgan ∗, Joan Ozanne-Smith, Tom TriggsMonash University Accident Research Centre, Australia

Received 19 January 2008; received in revised form 15 April 2008; accepted 21 April 2008

bstract

Because not all persons bathe at surf beaches, drowning rates based on resident population are likely to be underreported. To facilitateore precise drowning risk exposure data, this study aimed to develop a reliable direct observation measure of frequency and duration for

urf beach bather exposure to water, by gender and age group. Bathers were defined as persons entering the water to wade, swim or surfith equipment. Observed bathers were systematically selected entering the water in daylight hours at six patrolled or unpatrolled beachesver 10 days. Variables measured were: weather and water conditions, water entries, duration of water exposure, water exposure locationnd person factors. The dataset comprised 204 (69.6%) males and 89 (30.4%) females, with males more likely to be in an older age groupp < 0.05). Compared to females, males spent longer in the water, were more likely to use surfing equipment, and mainly used a surf zoneocated farther from the shore in deeper water (p < 0.05). Two factors were significant predictors of bathing duration (adjusted R2 = 0.45): mainurf zone occupied (based on water depth and distance from shore); and surf equipment used. The study provides new information about water

xposure for bathers at surf beaches and new methods for measuring exposure to drowning risk. The findings suggest that overrepresentationf adolescent and adult males in surf beach drowning statistics is in part a product of greater total exposure to the water plus more frequentxposure to deeper water and bathing farther from shore.

2008 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.

ion

ntd1ftctria

eywords: Bathing beaches; Drowning; Environmental exposure; Observat

. Introduction

Though accurate measures of risk exposure are required tostablish the epidemiology of injury, suitable methods haveot been developed for injury types including drowning.1

nstead, drowning epidemiology is routinely reported asrude rates based on community populations. These mea-ures probably underreport the true drowning rate and lackdequate precision to determine whether rate differencesetween population subgroups (e.g., gender or age) resultrom more frequent exposure to water or exposure to iden-ified risk factors. Addressing this knowledge gap requiresailored measures of water exposure frequency and dura-

ion applicable to subgroups, supplemented by measures ofxposure to candidate drowning risk factors.

∗ Corresponding author.E-mail address: damian.morgan@buseco.monash.edu (D. Morgan).

lmvAtb

440-2440/$ – see front matter © 2008 Sports Medicine Australia. Published by Eloi:10.1016/j.jsams.2008.04.003

Surf beach drowning accounts for 10% of unintentionalon-boating drowning in Australia.2 Based on a descrip-ive epidemiological study, the annual Australian surf beachrowning rate from July 2001 to June 2005 was 0.16 per00,000 population.3 Since a community telephone surveyor the state of Victoria suggests a significant proportion ofhe Australian population do not regularly visit beaches,4 thisrude drowning rate underestimates the rate of drowning inhe surf beach bather-exposed population. Moreover, crudeates do not explain why the ratio of male–female surf drown-ng exceeds 8:1 or why no surf beach drowning occurredmong the 0–9 years age group.3

Reasons postulated for the overrepresentation of ado-escent and adult males in recreational drowning include

ore frequent water exposures and bathing at unsuper-

ised swimming locations or high risk drowning zones.5,6

lthough self-reported data from surf beach patrons supporthese hypotheses,7 the validity of self-report data may beiased through recall and social desirability.8 Developing

sevier Ltd. All rights reserved.

4 and M

aesr

obsaceup

2

bpiEttep9p

JWpotiosrwdom4

tahfcddfls

dp

tsiazawsrsdbaarttp

twdpdSd

S(yatdmMismas

rvfsatvbbAf

58 D. Morgan et al. / Journal of Science

lternative methods to measure the frequency of bathers’xposure to water and to drowning risk factors will checkelf-report data and support explanations for crude drowningate differences between population subgroups.

The study aimed to develop a reliable method of directbservation to measure the frequency and duration of surfeach bathers’ (defined as persons entering the water to wade,wim or surf with equipment) exposure to water, by gendernd age group. Variables measured were: weather and wateronditions, water entries, duration of water exposure, waterxposure location and person factors including equipmentsed. A second aim was to model observable variables thatredict bathing duration.

. Method

Beaches are natural morphological features characterisedy wave deposited sediment.9 The pilot testing phase com-rised a sampling frame of 20 wave dominated beachesdentified by Short, situated consecutively along the South-ast Australian coastline, over a 39-day period.9 Three of

he beaches are patrolled regularly by trained lifesavers. Pilotesting determined the procedure for data collection plus anstimated population of 200,000 water entries in the sam-ling frame. A sample size of 384 water entries provided5% confidence of obtaining scores within 5% of the trueopulation.10

Data were collected during summer (December 2003 toanuary 2004) over five weekdays and five weekend days.

ithin the piloted sampling frame, six purposely selectedatrolled or non-patrolled high-use beaches were sampledn alternate days. The researcher took position close-byhe main beach access allowing unobstructed water view-ng. The viewing position changed over the day dependingn the position of the sun. Using binoculars, the first per-on entering the water (within approximately 600 m of theesearcher) in each of six 10-min periods per sampling houras tracked. Monitoring of this bather continued until watereparture. Sampling hours over each day varied dependingn the patterns of water use, number of users, and require-ents for observation breaks. The tracking target per day was

2 bathers over 7 h.Recorded weather and water conditions were sourced from

he Bureau of Meteorology.11 These were: daily maximumir and water temperatures in degrees Celsius, daily sunshineours, and predicted wave height for the area. Recordingsor each water entry were: time entered water, identifyinglothing, equipment used, gender, and estimated age. Timeeparting water was recorded on the bather’s exit. Locationata recorded were: bathing at a patrolled beach, bathing in aagged (supervised) area (patrolled beaches only), and main

urf zone used.

Based on Short and Hogan, the area used by bathers wasivided into three surf zones (Fig. 1 web based data sup-lement): the swash zone; the inner wave beaker zone; and

mai

edicine in Sport 12 (2009) 457–462

he outer wave breaker zone.12 The swash zone begins at thehoreline and extends seaward to a depth of 1 m. This zones characterised by water constantly advancing and recedingt the shore edge induced by wave action. The inner breakerone is located beyond one metre of water depth (this beingpproximately an adult’s chest height) and characterised byater moving shoreward as broken waves and then returning

eaward through rip channels. The outer breaker zone is theegion farthest from shore where ocean swell first break intourf waves, releasing significant energy. Due to increasingepth and distance from shore, the relative drowning risk forathers (assuming constant risk contribution of other personnd situation factors) is lowest at the swash zone and highestt the outer wave breaker zone.12 The three surf zones wereeadily recognisable to the trained eye at all beaches but con-inually varied in length and distance from shore accordingo the tidal level, the physical nature of the beach and therevailing wave height.

Over the 10-day period, 298 bathers were tracked acrosshe 6 beaches. During observations, five persons (1.7%)ere missed leaving the water and so excluded from theataset. Misses occurred on two particularly busy days withinatrolled (supervised) swimming areas. No drowning inci-ents were recorded by the National Coroners Informationystem or Surf Life Saving Australia at surveyed beachesuring the research period.3

Data were analysed using the Statistical Package for Socialcience.13 Estimated age was grouped into six categories0–10 years; 10–19 years; 20–29 years; 30–39 years; 40–49ears; over 49 years). Case counts, percentage and medi-ns are reported. Alpha level was set at 0.05 with two-tailedests. The Kolmogorov–Smirnov test identified non-normalistributions within groups on continuous variables (note:ain surf zone occupied assumed ordinal distribution).ann–Whitney test (U) assessed gender differences for time

n water, air temperature, wave height, sunshine hours, mainurf zone occupied, and age group with the effect size esti-ate r calculated manually (r = −0.1 small, −0.3 medium,

nd −0.5 large).14 Gender differences within surfer andwimmer subgroups were also investigated.

Kendall’s tau (τ) was chosen for non-parametric cor-elation tests of association between selected continuousariables due to the small sample size.14 Results are reportedor age group and time in water, air temperature, wave height,unshine hours, and main surf zone occupied with associ-ted effect size r2 provided by Gilpin.15 Pearson chi-squareest determined associations between gender and categoricalariables equipment used, bathing at a patrolled beach, andathing in a patrolled zone. Association size was estimatedy Cramer’s ν with interpretation from Rea and Parker.16

forementioned categorical variables were grouped to testor differences on age group using Mann–Whitney test.

Modelling followed multiple linear regression (enter

ethod) for observed independent variables theorised to be

ssociated statistically with the criterion variable (CV) timen water. These were: air temperature, wave height, sunshine

and M

hau

3

btsTTye

iwnrsct(gr

wtgttsaras

tmu

ndsetspwmzn(

Tns0otbwawa

dwdlz(Mo(loC

TF

A

01234O

O

N

D. Morgan et al. / Journal of Science

ours, being at a patrolled beach, bathing in a patrolled zone,nd main surf zone occupied, gender, age, and equipmentsed.

. Results

Table 1 (web based data supplement) lists sample days,each name and bathers tracked plus associated beach fea-ures and prevailing conditions. Males were most frequentlyampled each day, comprising 69.6% of the entire sample.he distribution of age group within gender was non-normal.he median age group was 20–29 years for males and 10–19ears for females; this difference being significant with smallffect size (Table 1).

Within both gender and age groups, variables of timen water, air temperature, wave height and sunshine hoursere non-normal distributions. Significant differences wereot found between genders for air temperature (U = 8604.5,= −0.04), wave height (U = 8393.5, r = −0.08) or sun-hine hours (U = 8462.5, r = −0.05). Similarly, no significantorrelations were found between age group and condi-ions for wave height (τ = 0.08, r2 = 0.01) or sunshine hoursτ = −0.09, r2 = 0.02). The positive correlation between ageroup and air temperature was significant (τ = 0.1, p = 0.02,2 = 0.03) but of small effect.

Bathers most frequently recorded as entering the waterere in the 10–19 years age group (Table 1). In comparison

o females, males had more frequent water entries in all ageroups except the 0–9 years. Overall, males spent a medianime of 19 min in the water compared to 11 min for females,his difference being significant and close to medium effectize. Total exposure (person-hours) were 93.84 h for malesnd 22.15 h for females equating to a relative crude exposureatio exceeding 4:1. A positive correlation was found betweenge group and time in water (τ = 0.1, p = 0.02, r2 = 0.03) withmall effect.

Compared to females, males were primarily more likelyo use a surf zone farther from the shoreline, the difference of

edium effect size (Table 2). The majority of males (60.3%)sed equipment whereas the majority of females (69.7%) did

t(rp

able 1requency of water entry and time in water by gender and age group

ge group Males Females

Time in water (min) Time in water (mi

Case count % Median Case count

–9 8 3.9 11.5 90–19 91 44.6 18 53a

0–29 31a 15.2 28 70–39 33 16.2 23 110–49 25 12.3 27 5ver 49 16 7.8 15.5 4

verall 204 100 19 89

ote: **significant at p < 0.01.a Median of age group distribution within gender on ordinal scale.

edicine in Sport 12 (2009) 457–462 459

ot (moderate association). Subgroup analysis assessed gen-er differences within surfer (bogie board or surf board) andwimming (all others) groups. For surfers, males were waterxposed for longer (U = 583.5, p < 0.01, r = −0.32) and far-her from shore (U = 542.5, p < 0.01, r = −0.32). Within thewimming group, males were farther from shore (U = 2685.0,< 0.01, r = −0.22) but did not spend more time in theater (U = 3011.0, r = −0.10). Overall, a higher percentage ofales bathed at a non-patrolled beach or outside supervised

ones at patrolled beaches, but these differences were not sig-ificant with associations negligible and weak respectivelyTable 2).

Results for age group are reported inable 2 (web based data supplement). A positive sig-ificant correlation was found between age group and mainurf zone occupied but only small effect. Both genders in the–9 age group used the shore break zone as the main zoneccupied. Age group distributions were non-normal withinhe grouping variables equipment used, being at a patrolledeach and bathing in a supervised zone. No associationsere found between age group and equipment used or being

t a patrolled beach. For those at a patrolled beach, bathersithin flagged (supervised) zones were more likely to be inyounger age group (small effect size).

Standard multiple regression (enter method) was used toetermine factors that statistically predict the CV time inater (min). Observed independent variables (IVs) for con-itions (air temperature, wave height, and sunshine hours),ocation (being at a patrolled beach, bathing in a patrolledone, and main surf zone occupied), and person factorsgender, age, and equipment used) were initially modelled.

eeting assumptions required square root transformationf the CV time in water (min) and deletion of one casea 40–49 age group surfer in the water for 135 min), fol-owing residual inspection. Only two IVs, main surf zoneccupied, and equipment used, significantly predicted theV at p < 0.05. Remaining IVs were not significant predic-

ors of the CV and as none added predicted shared varianceadjusted R2), they were removed from the analysis. Theesulting model had an adjusted R2 = 0.45; F4,287 = 59.7,< 0.01 (Table 3 web based data supplement).

Males vs. females test statistic

n)

% Median

10.1 10 By age group: U = 6946**, r = −0.2059.6 12

7.9 812.4 8

5.6 164.5 14.5

100 11 By time in water: U = 6008**, r = −0.27

460 D. Morgan et al. / Journal of Science and Medicine in Sport 12 (2009) 457–462

Table 2Surf zone, equipment used, beach type and bathing location by gender

Males Females Males vs. females test statistic

Case count % Case count %

Main surf zone occupiedShore break zone 80 39.2 67a 75.3 U = 5436**, r = −0.35Inner breaker zone 68a 33.3 18 20.2Outer breaker zone 56 27.5 4 4.5

Equipment usedNo equipment 81 39.7 62 69.7 χ2 (3, N = 293) = 29.36**, ν = 0.32Boogie board 50 24.5 18 20.2Surfboard 54 26.5 3 3.4Other floatation or swimming aids (e.g., wetsuits, flippers) 19 9.3 6 6.7

Bathing at a patrolled beachYes 121 59.3 61 68.5 χ2 (1, N = 293) = 2.24, ν = 0.09No 83 40.7 28 31.5

Bathing in a patrol zone (bathers at a patrolled beach only)Yes 69 57.0 43 70.5 χ2 (1, N = 182) = 3.12, ν = 0.13

N

4

fetatwaardtobdpdhttdrcp

Efbcprba

isrmcs

gdtwTw9scworw

ewsEvbtp

o

No 52

ote: **significant at p < 0.01.a Median of age group distribution within gender on ordinal scale.

. Discussion

This study provides new information about water exposureor bathers at surf beaches and new methods for measuringxposure to drowning risk. Direct observation of a substan-ial sample of surf beach bathers found that adolescent anddult males entered the water more frequently, were a longerime in the water, and bathed farther from shore and in deeperater relative to females. Key behavioural differences were

lso apparent between bathers in age groups from 10 to 49nd younger bathers. This evidence suggest that the over-epresentation of adolescent and adult males in surf beachrowning is in part a product of greater total exposure tohe deeper water farther from shore, although exposure tother factors such as alcohol or no supervision may exacer-ate risk.17 This study corresponds with self-reported genderifference pattern from the same population,7 where com-ared to females, males self-reported more time in the wateruring a beach visit (median time of 1-h for males and half-our for females) and being more often in water too deepo touch bottom. Nevertheless, measures of time-exposureo water reported in this study remain crude estimates ofrowning risk exposure given that factors including surf expe-ience, group behaviours, parental supervision, swimmingompetence, swim fatigue, health status, and risk or abilityerceptions were not obtained by direct observation.

Under low to moderate surf conditions prevailing in South-ast Australia, wave action causes adult bathers to lose their

ooting in depths above one metre.12 In these circumstancesathers can become caught in seaward moving rip currenthannels flowing at velocities higher than achieved by com-

etitive swimmers. Epidemiological study indicates that aips current is associated with at least 22% of recreation-ased surf beach drowning.3 Further, a study of bather’sbility perceptions found that males were more confident

ponb

43.0 18 29.5

n their swimming ability, rip current recognition, and surfwimming performance relative to females.7 Hence, the highelative rate of adolescent and adult male surf beach drowningay be explained partially by more frequent exposure to rip

urrents amplified by overconfidence among inexperiencedurf bathers.

Just over half of males sampled used surfboards or boo-ie boards designed for wave riding. Yet just 15.5% of surfrowning victims used surf craft.3 Surf craft may be a protec-ive factor due to the flotation provided or may be associatedith other protective factors such as surf bathing experience.he protection provided by bathing in or near patrolled zonesas supported by Fenner et al.’s study. Here, analysis of0 resuscitation reports for Queensland from 1972 to 1993howed the likelihood of successful resuscitation increasedloser to the patrol area.18 In the present study, no differenceas found between genders for bathing at a supervised beachr swimming in a patrolled zone. However, further study isequired to confirm this finding given the gender differencesere close to significant levels (Table 2).Two observed variables, main surf zone occupied and

quipment used, predicted 45% of the variability of time inater. The remaining variability may rest with person factors

uch as surf experience and confidence in swimming ability.ven so, this finding indicates that surf beach drowning pre-ention campaigns should target bathers intending to moveeyond the wading zone. Such campaigns should highlighthe risk of rip currents and the potential for fatigue from longeriods of exposure in the high-energy surf conditions.

This is the first published study to measure by directbservation bathing exposure and exposure to drowning risk

resented by deep water or bathing farther from shore. Directbservational studies of drowning risk are practical for otheratural water settings with identified drownings such asay beaches, lakes or rivers. New information from further

and M

aetl

stsstuwnptw

FtomcmTbwtrricahab

5

rtdaoco

P

•

•

•

A

ipRcCg

A

b2

R

D. Morgan et al. / Journal of Science

pplication of the method developed here could generatexplanations for unintentional drowning to support effec-ive countermeasures plus more precise drowning rates byocation.

Limitations of the sample, method and research designignify that the study results should be interpreted with cau-ion and recommendations be considered in this light. Theample was drawn from six beaches, located within a 100 kmtretch of coastline, within a limited range of air and wateremperatures. Although a case could be made for the pop-lation and sampling units to be characteristic of Victorianave dominated beaches during summer, representation mayot extend to other surf beaches, time periods or batheropulations.9 Representative studies are required to assesshe generalisability of surf beach bather characteristics andater behaviours found in this study.To guide further study, other limitations are noted here.

irstly, despite rigorous planning, this study was 23% underhe targeted sample size. Secondly, unknown bias may haveccurred through over sampling, with more frequent bathersore likely tracked.19 Thirdly, age estimates used here may

ontain inaccuracies, although similar estimates based onore limited observation have been reasonably precise.20

he fourth limitation is the potential exclusion of some beachather groups. For example, surfers may have entered theater from obscured rock platforms and night-time observa-

ions were not made. The final limitation concerns observereliability. Careful planning and attention was undertaken toecord accurate identifying data at water entry (e.g., identify-ng clothing and equipment used). The bather was observedontinually in the water and recorded data were cross checkedt bather exit to verify identity. Nevertheless, the potential foruman error require that future studies include inter-rater reli-bility checks using multiple observers for heavily patronisedeaches.

. Conclusion

Developing methods for gathering denominator dataequired for accurate drowning epidemiology is a challengingask. Nevertheless, more precise exposure data will assist inetermining the contribution of causal drowning risk factorsnd subsequent countermeasures. Given the global burdenf drowning, which in many countries is the second leadingause of unintentional injury, this effort is both warranted andverdue.21

ractical implications

Direct observation of surf beach bathers provided esti-

mates of bathing frequency and duration by gender andage groupings.Over-representation of adolescent and adult males in surfbeach drowning may be, in part, a product of greater

edicine in Sport 12 (2009) 457–462 461

total exposure to the water plus more frequent exposureto deeper water and bathing farther out from shore.More precise data on casual drowning risk factors willassist in the efforts to curb drowning deaths on surfbeaches.

cknowledgments

Ethical approval granted by Monash University Stand-ng Committee on Ethics in Research Involving Humans,roject no. 2001/431. National park access provided by DSEesearch Permit 10002639. Funded by the first author’s PhDandidature with the Monash University Accident Researchentre. The authors acknowledge with thanks reviewer sug-estions used to improve the manuscript.

ppendix A. Supplementary data

Supplementary data associated with this article cane found, in the online version, at doi:10.1016/j.jsams.008.04.003.

eferences

1. Pless IB, Hagel BE. Injury prevention: a glossary of terms. J EpidemiolCommunity Health 2005;59:182–5.

2. Mackie IJ. Patterns of drowning in Australia 1992–1997. Med J Aust1999;171:587–90.

3. Morgan D, Ozanne-Smith J, Triggs T. Descriptive epidemiology ofdrowning deaths in a surf beach swimmer and surfer population. InjPrev 2008;14:62–5.

4. Sweeny Research. Attitudes to water safety. Melbourne: VictorianDepartment of Justice; 2006.

5. Howland JR, Hingson R, Mangione T, et al. Why are most drowningvictims men? Sex differences in aquatic skills and behaviors. Am JPublic Health 1996;86(1):93–6.

6. Gulliver P, Begg D. Usual water-related behaviour and ‘near drowning’incidents in young adults. Aust N Z J Public Health 2005;29(3):238–43.

7. Morgan D, Ozanne-Smith J, Triggs T. Comparison of male and femaleself-reported water and drowning risk exposure at surf beaches. Aust N ZJ Public Health, submitted for publication [unpublished observations].

8. Foley KL, Manuel J, Vitolins M. The utility of self-report in medi-cal outcomes research. Evidence-Based Healthcare and Public Health2005;9:263–4.

9. Short AD. Beaches of the Victorian Coast and Port Phillip Bay. Sydney:University of Sydney; 1996.

10. Mitchell ML, Jolley J. Research design explained. Texas: HarcourtCollege; 2001.

11. Bureau of Meteorology. Victorian weather observations. Available at:http://www.bom.gov.au/weather/vic/observations.shtml. Accessed 10March 2004.

12. Short AD, Hogan CL. Rip currents and beach hazards: their impact onpublic safety and implications for coastal management. J Coastal Res1995:197–209 [special issue].

13. SPSS for Windows. Release 16.0.0. Chicago: SPSS Inc.; 2007.14. Field A. Discovering statistics using SPSS. London: Sage; 2005.15. Gilpin AR. Table for conversion of Kendall’s Tau to Spearman’s Rho

within the context of measures of magnitude of effect for meta-analysis.Educ Psychol Meas 1993;53:87–92.

4 and M

62 D. Morgan et al. / Journal of Science

16. Rea LM, Parker RA. Designing and conducting survey research. 3rded. San Francisco: Jossey-Boss; 2005.

17. Driscoll TR, Harrison JA, Steenkamp M. Review of the role of alco-hol in drowning associated with recreational aquatic activity. Inj Prev2004;10:107–13.

18. Fenner PJ, Harrison SL, Williamson JA, et al. Success of surf lifesavingresuscitations in Queensland, 1973–1992. Med J Aust 1995;163:580–3.

edicine in Sport 12 (2009) 457–462

19. Crompton JL. On-site sampling: a potential source of overestimatingvisitor use. J Travel Res 2001;3:321–2.

20. Moeller S, Berger L, Salvador JG, et al. How old is that child? Validatingthe accuracy of age assignments in observational surveys of vehiclerestraint use. Inj Prev 2002;8(248):251.

21. Smith G. The global burden of drowning. In: Bierrens JJLM, editor.Handbook on drowning. Germany: Springer-Verlag; 2006. p. 56–61.