2974

current as of December 12, 2009. Online article and related content

http://jama.ama-assn.org/cgi/content/full/286/23/2974

. 2001;286(23):2974-2980 (doi:10.1001/jama.286.23.2974) JAMA

Gordon S. Smith; Penelope M. Keyl; Jeffrey A. Hadley; et al.

Population-Based Case-Control StudyDrinking and Recreational Boating Fatalities: A

Correction Contact me if this article is corrected.

Citations Contact me when this article is cited. This article has been cited 28 times.

Topic collections Contact me when new articles are published in these topic areas.

Quality of Care

http://pubs.ama-assn.org/misc/[email protected]

http://jama.com/subscribeSubscribe

[email protected]/E-prints

http://jamaarchives.com/alertsEmail Alerts

by guest on December 12, 2009 www.jama.comDownloaded from

ORIGINAL CONTRIBUTION

Drinking and Recreational Boating FatalitiesA Population-Based Case-Control StudyGordon S. Smith, MB, ChB, MPHPenelope M. Keyl, MSc, PhDJeffrey A. Hadley, PhDChristopher L. Bartley, MARobert D. Foss, PhDWilliam G. Tolbert, MAJames McKnight, PhD

MORE THAN 43 MILLIONpeople reported using amotorboat in the UnitedStates in 1994,1 andabout 800 people died in 1998 from rec-reational boating.2 Alcohol is com-monly involved in drownings and otherunintentional injury fatalities3-7 and isincreasingly recognized as an impor-tant factor in many boating fatali-ties.8,9 Data from 4 states with high test-ing rates for 1980 to 1985 suggest that51% of people involved in boating fa-talities had a blood alcohol concentra-tion (BAC) of at least 40 mg/dL, and30% had a BAC higher than 100 mg/dL.4,10 Other countries such as Canada11

and Finland12 have an even higher pro-portion of boating fatalities linked to al-cohol use.

Alcohol use while boating affects theprobability not only of ending up in thewater but also of survival once that hap-pens. Because of this apparent doublejeopardy, alcohol use may actually bemore hazardous on a boat than in othersettings, with even low BACs greatly in-creasing relative risk (RR).8,13,14 Al-though these and other studies4,8,15,16

suggest that alcohol increases the RRof dying while boating, this relation-ship has not been well quantified.

This study sought to better define therelationship between alcohol use and the

RR of death while boating. We con-ducted a large population-based case-control study of alcohol use and recre-ational boating fatality risk in 2 states,Maryland and North Carolina. Thesestates include a diversity of waterwayson which recreational boating takesplace. We sought to determine the mag-nitude of the estimated RR of dying as-sociated with alcohol use, adjusting forknown or potential risk factors fordrowning and other boating deaths. Wealso examined whether RRs were dif-ferent for passengers and operators andwhether low BACs pose a significant RR.

METHODSIdentifying and SelectingBoating Fatalities

We searched official state boating fatal-ity records and medical examiner files

in each state to identify all recreationalboating deaths classified as acciden-tal that occurred from 1990 to 1998 inMaryland and North Carolina. Onlyboating deaths that occurred from Aprilthrough October (n=403 of 502 deaths)were included in the study. Boating ac-tivity declined markedly outside thesemonths, making control interviews pro-

Author Affiliations: Johns Hopkins Center for Injury Re-search and Policy, Baltimore, Md (Drs Smith, Keyl, andHadley); Department of Emergency Medicine, JohnsHopkins School of Medicine, Baltimore (Drs Smith andKeyl); Highway Safety Research Center, University ofNorth Carolina at Chapel Hill (Dr Foss and Messrs Bar-tley and Tolbert); Pacific Institute for Research and Evalu-ation, Rockville, Md (Dr McKnight). Mr Tolbert is nowat Rho Inc, Chapel Hill, NC. Dr Smith is now also at theCenter for Safety Research, Liberty Mutual ResearchCenter for Safety and Health, Hopkinton, Mass.Corresponding Author and Reprints: Gordon S. Smith,MB, ChB, MPH, Center for Safety Research, LibertyMutual Research Center for Safety and Health, 71 Fran-klin Rd, Hopkinton, MA 01746 (e-mail: [email protected]).

Context Alcohol is increasingly recognized as a factor in many boating fatalities, butthe association between alcohol consumption and mortality among boaters has notbeen well quantified.

Objectives To determine the association of alcohol use with passengers and op-erators estimated relative risk (RR) of dying while boating.

Design, Setting, and Participants Case-control study of recreational boating deathsamong persons aged 18 years or older from 1990-1998 in Maryland and North Caro-lina (n=221), compared with control interviews obtained from a multistage probabil-ity sample of boaters in each state from 1997-1999 (n=3943).

Main Outcome Measure Estimated RR of fatality associated with different levelsof blood alcohol concentration (BAC) among boaters.

Results Compared with the referent of a BAC of 0, the estimated RR of death in-creased even with a BAC of 10 mg/dL (odds ratio [OR], 1.3; 95% confidence interval[CI], 1.2-1.4). The OR was 52.4 (95% CI, 25.9-106.1) at a BAC of 250 mg/dL. Theestimated RR associated with alcohol use was similar for passengers and operators anddid not vary by boat type or whether the boat was moving or stationary.

Conclusions Drinking increases the RR of dying while boating, which becomes ap-parent at low levels of BAC and increases as BAC increases. Prevention efforts tar-geted only at those operating a boat are ignoring many boaters at high risk. Coun-termeasures that reduce drinking by all boat occupants are therefore more likely toeffectively reduce boating fatalities.JAMA. 2001;286:2974-2980 www.jama.com

2974 JAMA, December 19, 2001Vol 286, No. 23 (Reprinted) 2001 American Medical Association. All rights reserved.


hibitively expensive and difficult. Be-cause of difficulty finding control sub-jects at night, especially in NorthCarolina, boating deaths that occurredbetween midnight and 7:00 AM in Mary-land and between 9:00 PM and 7:00 AMin North Carolina were excluded fromthe study (13.9% of eligible cases).Deaths associated with the use of sail-boats, personal watercraft (ie, jet skis),and rafts were excluded (16.1% of eli-gible cases). Deaths on sailboats are rare,and personal watercraft and rafts are dif-ferent from other boat types.2,17,18 Fatal-ity and control subjects younger than 18years (9.7% of eligible cases) were ex-cluded because the parents of potentialunderage control subjects were often notavailable to give consent. Small inlandbodies of water were excluded in Mary-land, since only 3% of eligible deathsoccurred in them and they were widelydispersed. Despite the Coast Guard defi-nition of a boating death,2 individualswho drowned while swimming from aboat were included in our study, al-though some of our analyses excludedthem.

Control Subject SelectionControl subjects were from a strati-fied random sample of boats from wa-terways in each state during the boat-ing season (April through October)from 1997 through 1999. A complexsampling design was used to ensure thatcontrol subjects were drawn from thesame locations as fatality subjects ineach state. First, the states navigablewaterways were divided according togeographic area and type of waterwayinto strata that reflected cultural and de-mographic differences (TABLE 1).Within each stratum, areas were se-lected to represent locations of boat-ing activity. Given the large differ-ences in the types of waterways andtheir distribution in Maryland andNorth Carolina, sampling procedureswere tailored for each state.

Selection of Waterwaysfor Control SurveyNorth Carolina. The state was first di-vided into 3 geographically and cultur-

ally distinct regions (coastal, midstate,and western) with historically differ-ent patterns of alcohol use.

Bodies of water were categorized intoocean/bay/sound waterways, large andmedium-sized lakes, and small lakes andrivers. The ocean/bay/sound water-ways were treated as 1 stratum, large andmedium-sized lakes in each region com-posed 3 additional strata, and all smallerlakes and rivers composed the final stra-tum (a total of 5 strata). Within theocean/bay/sound stratum,19 geographi-cally distinct areas were identified, and6 of these were randomly selected. Eachof the 12 largest lakes and 6 of the 10medium-sized lakes were randomly se-lected as sampling areas, with the num-ber of medium-sized lakes selected pro-portionate to the population within eachregion. For the small lake and river stra-tum, the state was subdivided by lati-tude and longitude. Of the 210 result-ing subdivisions, 42 (20%) were selected,with a probability proportionate to thepopulation within the region. Withineach selected subdivision, 2 areas, 1small lake and 1 river, were then ran-domly chosen from navigable water-ways, which resulted in the selection of30 small lakes and rivers across the state,for a total of 54 selected areas in NorthCarolina.

Maryland. Recreational boating inMaryland occurs primarily in 4 bodiesof water, with the majority occurringon the Chesapeake Bay and its manyriver estuaries and also on the Po-tomac River.

Deep Creek Lake and the AtlanticOcean are 2 other areas where boatingis popular, but neither had any boat-ing fatalities during the study period.Chesapeake Bay was divided into 6strata corresponding to the upper,middle, and lower sections on both thewestern and eastern sides of the bay.The Potomac River was divided into 3strata, 2 below Washington, DC, and1 nontidal part above it. Each stratumwas divided into areas that could be sur-veyed in 1 day.

On-Water ProceduresTeams of 2 interviewers visited desig-nated areas (eg, a lake, river, or regionof a bay) multiple times by boat on apredetermined schedule that includedboth weekdays and weekends. On eachvisit, interviewers moved systemati-cally around the water to ensure thatthe entire area was covered. Upon ar-riving at a designated location, inter-viewers identified up to 6 boats near-est to them and then used a die torandomly select 1 to interview. Onlystationary or slowly moving boats weresampled. These fell into 2 categories:those that were anchored, moored,drifting, or berthed, and those that werearriving at destinations in a samplingarea such as a fishing area, beach, ma-rina, or boat ramp after being underway. In Maryland only, because of thelarge size of the Chesapeake Bay, wealso used shore teams to interview boat-ers who were returning to a boat rampor marina. Boats were approached in the

Table 1. Sampling Features Taken Into Consideration in Study Design or Analysis:Boating Case-Control Study, Maryland and North Carolina

Sampling Feature Description

Strata Navigable waterways in each state were stratified according togeographic area and type of waterway, resulting in 5 strata inNorth Carolina and 9 in Maryland.

Area Within each stratum, waterways were selected to represent boatingactivity in the stratum. In the analysis, observations within eacharea were treated as clusters to account for the lack ofindependence of the observations.

Weights We sampled all operators and up to 2 passengers per boat. Weightswere calculated and used in the analysis to adjust for thisdifferential sampling between operators and passengers.

Time of day Within each stratum, interviews were conducted to cover all times ofday rather than attempting to match the time-of-day distribution ofcases. Analyses were adjusted for the confounding effects of timeof day on the relationship of BAC levels with death.

DRINKING AND RECREATIONAL BOATING FATALITIES

2001 American Medical Association. All rights reserved. (Reprinted) JAMA, December 19, 2001Vol 286, No. 23 2975


order in which they arrived at the shore-based sampling site.

The selected boat was approached andthe operator was asked to participate inthe study. The operator was inter-viewed and asked to provide details onthe boat and the boats activities in the

past hour. Next, the operator and up to2 randomly selected passengers (18years) were asked to complete a shortself-administered questionnaire that in-cluded questions on general health anddemographiccharacteristics.Last, theop-erator and the selected passengers were

asked to provide a breath sample for al-cohol testing by a handheld breatha-lyzer (CMI Intoxilyzer D-400R; CMI Inc,Owensboro, Ky). The interviewer also re-corded information about the boat, num-ber of passengers, evidence of alcoholuse, apparent sobriety of the operator,and refusals. Institutional review boardsfor the protection of human subjects atthe Johns Hopkins School of PublicHealth and the University of North Caro-lina School of Public Health approved thestudy procedures.

Adjustments in BAC forEndogenous Alcohol in FatalitiesWhen recovery of a body is delayed,decomposition can result in postmor-tem alcohol production. Rather thanexcluding the subjects that were notrecovered within 1 or 2 days afterdeath,19 we used a conservative proce-dure based on new evidence about thetime course of decomposition to adjustthose subjects BAC levels (J.A.H. andG.S.S., unpublished data, 2001). Theamount subtracted from the observedBAC started as 0 mg/dL for cases witha submersion time of 12 hours andincreased linearly to a maximum of 40mg/dL for bodies recovered after 96hours in the water. Few drowning vic-tims produce endogenous alcohol lev-els as high as 40 mg/dL, even at the long-est recovery times.20,21 Cases in whichthe body was recovered more than 1week after the incident were excluded.

Statistical AnalysisThe increased RR of fatality associatedwithBAC,after adjustment forother fac-tors, was estimated by calculating oddsratios (ORs) using logistic regressionin Stata (StataCorp, Version 6, 2000;College Station, Tex). Effects of the sam-pling design (stratification, clustering,and weighting) were accounted for inthe analysis by using the svy Stata com-mands. Each area within a stratumwherecontrolboaterswere sampled(eg,a lake, a section of river, or an area ofbay) was treated as a primary sam-pling unit or cluster (Table 1). Becausethe number of passengers in controlboats ranged from 0 to 14 but a maxi-

Table 2. Comparison of Boat and Demographic Characteristics for Fatality and ControlSubjects, Maryland and North Carolina

No. (%)

Fatality Subjects(n = 221)

Control Subjects(n = 3943)

Boat typeCabin motorboat 18 (8.1) 597 (15.1)

Open motorboat (3 m) 154 (69.7) 3118 (79.1)

Small boats* 49 (22.2) 228 (5.8)

Time of day7:00-10:00 AM 26 (11.8) 79 (2.0)

10:01 AM-noon 21 (9.5) 336 (8.5)

12:01-2:00 PM 31 (14.0) 649 (16.5)

2:01-4:00 PM 43 (19.5) 971 (24.6)

4:01-6:00 PM 27 (12.2) 853 (21.6)

6:01-8:00 PM 46 (20.8) 724 (18.4)

8:01-10:00 PM 20 (9.0) 296 (7.5)

10:01 PM-midnight 7 (3.2) 35 (0.9)

Day of the weekWeekend 124 (56.1) 2685 (68.1)

Weekday 97 (43.9) 1258 (31.9)

SexMale 204 (92.3) 2860 (73.4)

Female 17 (7.7) 1034 (26.6)

RaceBlack 52 (24.3) 132 (3.8)

Nonblack 162 (75.7) 3324 (96.2)

Occupant statusOperator 97 (43.9) 2108 (53.5)

Passenger 124 (56.1) 1835 (46.5)

Age, y18-20 14 (6.3) 89 (2.6)

21-30 45 (20.4) 634 (18.4)

31-40 45 (20.4) 1008 (29.3)

41-50 42 (19.0) 858 (24.9)

51-60 36 (16.3) 557 (16.2)

61-70 25 (11.3) 205 (6.0)

70 14 (6.3) 88 (2.6)

ActivityCruising 90 (40.7) 1858 (47.1)

Fishing 85 (38.5) 1926 (48.8)

Drifting 36 (16.3) 1763 (44.7)

Anchored 16 (7.2) 785 (19.9)

Waterskiing 6 (2.7) 315 (8.0)

Racing 6 (2.7) 11 (0.3)

Towing 2 (0.9) 12 (0.3)

*Includes all boats less than 3 m long and all canoes, kayaks, and rowboats.Numbers do not equal n because of missing values.Nonblack consists of 97% white fatality subjects and 98% white control subjects, with the remainder other nonblack

races or unknown.There may be more than 1 activity per fatality subject and control subject; thus, the percentage will not add up to 100.




mum of only 2 passengers was sampled,appropriateweightswereapplied topro-vide a valid comparison of operators andpassengers. All analyses were adjustedfor the confounding effects of time ofday (resulting from the sampling sched-ule) by including variables for time ofday in 2-hour increments. For the mainanalysis, both BAC and age were treatedas continuous variables. Higher-orderterms were considered for both vari-ables. Categories of BAC were createdonly to compare the results with find-ings from other studies.

Multiple imputations were carriedout to replace missing values for sex,race, and age (1%, 12%, and 12%, re-spectively, for controls and 15% for racefor subjects; TABLE 2). A hot-deck pro-cedure using the approximate Bayesianbootstrap method of Rubin and Schen-ker22,23 was used. Ten imputations wereperformed for each analysis. This ap-proach assumes that within each state(Maryland or North Carolina) and boattype, missing values for subgroups ofsubjects had the same distribution asknown values.

Crude analyses suggested that con-trol operators who refused to partici-pate might have had higher BACs thanparticipating operators; 5% and 2%, re-spectively, were judged to be at leastmoderately impaired. For operatorsonly we evaluated the extent to whichrefusals to give a breath sample mighthave influenced BAC RR estimates. Thehot-deck method described above wasused to impute the missing BACs, as-suming missing values for BACs had thesame distribution as those with knownBAC within each level of the interview-ers assessment of impairment.

RESULTSFatality Subjects

Of the 253 boating victims meeting in-clusion criteria, 15 (6%) were ex-cluded from the analysis because theirbodies were recovered more than 1week after death or were recovered af-ter an unknown length of time. Amongthe 238 eligible fatality subjects, 76%were recovered within 24 hours ofdeath; 11%, within 25 to 48 hours; 9%,

within 49 to 96 hours; and 4%, within97 to 168 hours. Seventeen of these sub-jects (7.1%) were not tested for BAC.Of the 221 subjects included in thestudy, 55% had a positive BAC (ad-justed for recovery time); 36% had aBAC of at least 50 mg/dL; 27%, at least100 mg/dL; 18%, at least 150 mg/dL;11%, at least 200 mg/dL; and 7%, at least250 mg/dL. Most subjects had been inopen motorboats at least 3 m long(69.7%), and the largest number ofthem died between 6:00 and 8:00 PM(20.8%; Table 2). Subjects were pre-dominantly male and nonblack, lessthan half were operators, and most were21 to 40 years of age. Eligible subjectsexcluded because of missing BAC datadid not have different demographic fac-tors. Eleven subjects (3.2%) died inrough water, which precluded safely in-terviewing control subjects in similarconditions, but because they had BACssimilar to those of other subjects, theywere kept in the study. Passengers weremore likely than operators to have apositive BAC (68% vs 48%; P.001)that was at least 100 mg/dL (37% vs27%; P=.04).

Control SubjectsThe number of boats sampled from eachof the 14 strata ranged from 75 to 504.Almost all (93%) of the operators of

boats sampled for the control surveyagreed to participate; 87% completedthe self-administered questionnaire, and86% provided a valid breath sample. Ofthose who gave a breath sample, 7.6%refused the self-administered question-naire. The interviews yielded a total of4801 potential controls (2468 opera-tors and 2333 passengers), of whom3943 provided a valid breath sampleand were included in the analysis (Table2). Boating and demographic charac-teristics of persons who provided abreath sample differed little from thatof those who refused, although thoseon open motorboats, those who wereapproached earlier in the day, femalesubjects, and younger persons weresomewhat more likely to participate.Only 17% of participants had a posi-tive BAC. Of those, 7.4% had a BAC ofat least 50 mg/dL; 3.4%, at least 100 mg/dL; 1.4%, at least 150 mg/dL; 0.6%, atleast 200 mg/dL; and 0.3%, at least 250mg/dL. These figures represent crudeunweighted distributions from a strati-fied sample and thus are not represen-tative of boaters in these areas.

Relative RiskA greater proportion of control sub-jects were in motorboats at least 3 mlong and were female, nonblack, and 21to 50 years of age (Table 2). The RR of

Figure. Relative Fatality Risk While Boating by BAC, Maryland and North Carolina

100

10

5

50

10 10050 150 200 250

BAC, mg/dL

Adj

uste

d O

dds

Rat

io,

95%

Con

fiden

ce In

terv

als

Logarithmic scale indicating odds ratio of dying relative to having a blood alcohol concentration (BAC) of0 mg/dL. Dashed lines indicate 95% confidence intervals.




death by BAC level, compared with thatof subjects with a BAC of 0 mg/dL, wasdetermined in analyses to be a second-order quadratic relationship when ad-justed for age, race, sex, occupant sta-tus, boat type, location, time of day, andweekend/weekday. Age was modeled asa third-order quadratic relationship.The ORs for dying by BAC increasedmost rapidly at lower BACs, with therate of increase leveling off at higherBACs (FIGURE). The RR of death wasincreased even at a BAC of 10 mg/dL(OR=1.3; 95% confidence interval [CI],1.2-1.4), increasing to an OR of 52.4 ata BAC of 250 mg/dL (95% CI, 25.9-

106.1; TABLE 3). When only those per-sons meeting the official Coast Guarddefinition of boating accidents wereconsidered (ie, when the 22 subjects[10%] who died while voluntarilyswimming from a boat and when con-trol subjects from boats where peoplewere swimming were excluded), therewas no significant change in the RRs offatality (Table 3).

Additional analyses were conductedby using categories of BAC and dichoto-mizing BAC at different cut points to per-mit comparisons with other studies(TABLE 4). These values have wider CIsthan estimates of RR when BAC is usedas a continuous variable.

Interactions andSensitivity AnalysesThe RR associated with BAC was notsignificantly different between opera-tors and passengers, male and femalesubjects, black and nonblack persons,persons of different ages, or differenttypes of boats.

Adjusting for the potential bias re-sulting from control subjects who de-clined to give breath samples de-creased the ORs, but the differenceswere not significant. Because subjec-tive impressions of intoxication are un-reliable, we elected to present find-ings based on actual measurements, as

has been the practice in the few stud-ies that have evaluated refusal bias.24-26

COMMENTThe most important finding in thisstudy is the strong positive associa-tion of BAC with the RR of death amongrecreational boaters aged 18 years andolder, even at BACs less than 50 mg/dL. In addition, passenger and opera-tor drinking is associated with the sameincreased RR of death, regardless ofwhether the boat is under way.

Dose-Response Effects of AlcoholThe RRs associated with alcohol use andboating fatality increase markedly as theBAC increases, from an OR of 1.3 at aBAC of 10 mg/dL to 52 at 250 mg/dL.Our finding of increased RR at lowBACs is consistent with experimentalstudies that find significant impair-ment in many safety-related tasks atBACs below 50 mg/dL.27-30

Alcohol can affect boater safety inmultiple ways, influencing both the riskof ending up in the water (or crash-ing) and chances for survival in thewater.8,13,14,29-31 Alcohol impairs bal-ance and coordination, which can in-crease the risk of falling overboardwhether a boat is under way or not. Im-paired judgment resulting from an el-evated BAC can also increase the like-lihood of being in high-risk situations,and unlike on the roadway, having a so-ber operator will not necessarily pro-tect impaired occupants. The effects ofalcohol on the probability of survivalare greater than for other injury causes31

and, once a person enters the water, in-clude an increased risk of hypother-mia and a reduced ability to keep thehead above water.8,13,14,29 Thus, a simplefall overboard can prove fatal.

Although there is substantial evi-dence for the risk of drinking and driv-ing,27,30,32-34 there is surprisingly little in-formation about the risk of drinking andother injuries, including those associ-ated with boating. Besides that re-ported here, the only study designed toestimate the risk of drinking for boat-ers was conducted at boat ramps inCalifornia. That study had a small

Table 3. Comparison of Adjusted Odds Ratios of Dying While Boating by Blood AlcoholConcentration (BAC) Point Estimates for All Study Participants vs Only Those WhoWere Not Swimming*

BAC, mg/dL

Adjusted Odds Ratio (95% Confidence Interval)

All Study Participants(n = 4164)

Excluding Swimmers(n = 3251)

0 1.0 1.0

10 1.3 (1.2-1.4) 1.4 (1.3-1.5)

20 1.7 (1.6-1.9) 1.8 (1.6-2.1)

30 2.2 (1.9-2.6) 2.5 (2.1-2.9)

40 2.9 (2.3-3.6) 3.2 (2.6-4.0)

50 3.7 (2.8-4.7) 4.2 (3.3-5.5)

80 7.1 (5.0-10.1) 8.5 (5.9-12.3)

100 10.4 (6.9-15.7) 12.8 (8.4-19.6)

150 23.0 (14.0-37.9) 27.9 (16.6-47.0)

200 39.4 (22.4-69.6) 43.9 (24.6-78.4)

250 52.4 (25.9-106.1) 49.6 (25.2-97.5)

*Adjusted for age, race, sex, occupant status, boat type, location, time of day, and weekend/weekday. The adjust-ment included weights for differential passenger selection probabilities.

Includes all boating fatality and control subjects.Excludes subjects who died after swimming off a boat and control subjects from boats where occupants swam in the

past hour.

Table 4. Adjusted Odds Ratios of DyingWhile Boating by Blood AlcoholConcentration (BAC) Ranges

Adjusted OddsRatio (95%

Confidence Interval)

BAC ranges, mg/dL0 1.01-49 2.8 (1.6-4.8)50-99 5.7 (2.9-10.8)100-149 12.0 (5.8-24.9)150 37.4 (16.8-83.0)

BAC dichotomized,mg/dL50 10.5 (6.7-16.5)80 13.9 (8.3-23.4)100 15.7 (9.0-27.5)

*Adjusted for age, race, sex, occupant status, boat type,location, time of day, and weekend/weekday. The ad-justment included weights for differential passenger se-lection probabilities.




sample size and did not control for sev-eral relevant factors such as region, timeof day, age, sex, or boat type.4 It founda crude 10.7-fold increased risk of boat-ing fatality among operators with BACshigher than 100 mg/dL, and CIs werewide (95% CI, 4.7-68.8). In this study,we found a clear dose-response rela-tionship and controlled for many po-tential confounding variables. In addi-tion to elevated RR at very low BACs,we also found a much greater RR ofdeath at higher BACs than the Califor-nia study reported. Our main analysesincluded subjects swimming or divingoff a boat, since swimming is a com-mon part of boating activities, al-though excluding them in accordancewith Coast Guard practice2 did notchange the RR.

Operators vs PassengersAlcohol use has long been a part of rec-reational boating; 30% to 40% of boat-ers surveyed report drinking while boat-ing.1,6,35-38 Many of these boaters believethat they can safely drink more when atanchor or tied up and when they are pas-sengers rather than operators.36 Cur-rent legislation concentrates entirely onalcohol use by the boat operator whilethe boat is under way, prohibiting op-eration of a boat while intoxicated, ashave many safety campaigns.8,9,15 Somehave even promoted the use of a desig-nated driver when boating, with the im-plication that passengers can drink asmuch as they like as long as the opera-tor remains sober. Although these ap-proaches initially appear attractive, theyignore the reality that passengers can putthemselves at risk regardless of the op-erators actions or alcohol use. Onlyabout half the recreational boating fa-talities could be attributed to operatorerror.8 Most boating fatalities involvedrowning; only 18% involve collisionswith other boats or objects. The major-ity of fatalities involve falling over-board, and almost half (46%) of theseoccur when the vessel is not under way.Indeed, our findings clearly indicate thatthe RR of death is similar for operatorsand passengers and increases for bothgroups as BAC increases.

Many fatalities occur in unpoweredor low-powered boats,2,8,15 and manyothers occur while boats are not in op-eration, which undermines the assump-tion that boat handling by drunken op-erators is a primary cause of boatingfatalities. Unfortunately, since boat-ing police rarely test surviving opera-tors for alcohol use, it is impossible withcurrent data to assess the role of im-paired operators in increasing the riskof death for other boaters.

Policy ImplicationsThe implicit assumption of designateddriver programsthat a passenger candrink as long as the operator remains so-beris dangerous for boaters. All per-sons on a boat have an increased RR ofmortality if they have been drinking,even at low BACs. These findings sug-gest that countermeasures directed onlyat operators of moving boats are likelyto have less impact on alcohol-relatedboating fatalities than broader efforts toaddress drinking by anyone engaged inrecreational boating.

Study LimitationsTemporal changes in drinking prac-tice among boaters could affect alco-hol risk estimates, since fatality- andcontrol-subject data were collected fordifferent years. However, throughoutthe study period BACs among sub-jects did not change significantly overtime, nor did RRs of death estimatedacross cases from 1990 to 1994 andfrom 1995 to 1998.

Although many potentially con-founding variables were taken into ac-count, we were unable to adjust forother variables that might affect risk,such as the boaters swimming ability,the operators boating skills and expe-rience, use of personal floatation de-vices, water and weather conditions,and the condition and seaworthiness ofthe boat. Use of personal floatation de-vices was low among control subjects(about 6.7% of adults in control boats),but because such use was assessed onlyat the boat level and not for individu-als, it was impossible to include it in ouranalyses. However, this study was de-

signed to look at the total RR of deathwhen subjects had been drinking, notto separately examine the influence ofBAC on the risk of falling in the water(or crashing) and surviving once in thewater. Personal floatation device useand swimming ability would have a di-rect effect only on the latter. Finally, al-though we controlled for boating ex-posure with the random selection ofcontrol subjects, some groups, such aspersons in boats that spent most of theirtime under way, may have been under-represented.

Author Contributions: Study concept and design:Smith, Keyl, Hadley, Bartley, Foss, Tolbert, McKnight.Acquisition of data: Smith, Hadley, Bartley, Foss,Tolbert, McKnight.Analysis and interpretation of data: Smith, Keyl,Hadley, Bartley, Foss, McKnight.Drafting of the manuscript: Smith, Keyl, Hadley,Bartley, Foss.Critical revision of the manuscript for important in-tellectual content: Smith, Keyl, Hadley, Bartley, Foss,Tolbert, McKnight.Statistical expertise: Smith, Keyl, Hadley, Bartley, Foss.Obtained funding: Smith, Keyl, McKnight.Administrative, technical, or material support: Smith,Hadley, Foss, Tolbert.Study supervision: Smith, Keyl, Foss, Tolbert, McKnight.Funding/Support: This study was supported by grantR29AA07700 from the National Institute of AlcoholAbuse and Alcoholism.Acknowledgment: We wish to acknowledge the as-sistance of the staff, interviewers, boat operators, andassistants at the Pacific Institute for Research and Evalu-ation (PIRE), Calverton, Md, and the University of NorthCarolina at Chapel Hill, who coordinated field data col-lection. We would also like to acknowledge the as-sistance of the Maryland Department of Natural Re-sources, the North Carolina Wildlife ResourceCommission, and the medical examiners offices inMaryland and North Carolina, without whose help thisstudy would not have been possible. We also wish toacknowledge the assistance of the University of Auck-land Injury Prevention Research Centre.

REFERENCES

1. Logan P, Sacks JJ, Branche CM, Ryan GW, Bender P.Alcohol-influenced recreational boat operation in theUnitedStates,1994.AmJPrevMed. 1999;16:278-282.2. Boating Statistics1998. US Coast Guard Web site.Available at: http://www.uscgboating.org/saf/pdf/Boating_Statistics_1998.pdf. Accessed October 20,2000.3. Smith GS, Branas CC, Miller TR. Fatal non-trafficinjuries involving alcohol: a meta-analysis. Ann EmergMed. 1999;33:659-668.4. Mengert P, Sussman E, DiSario R. A Study of theRelationship Between the Risk of Fatality and BloodAlcohol Concentration of Recreation Boat Opera-tors. Washington, DC: US Dept of Transportation, USCoast Guard; 1992. Publication CG-D-09-92.5. Howland J, Hingson R. Alcohol as risk factors fordrownings: a review of the literature (1950-1985).Accid Anal Prev. 1988;20:19-25.6. Howland J, Magione T, Hingson R, Smith G, BellN. Alcohol as a risk factor for drowning and otheraquatic injuries. In: Watson RR, ed. Alcohol, Co-caine, and Accidents. Totowa, NJ: Humana Press;1995:85-104.




7. Smith GS, Krause JF. Alcohol and residential, rec-reational, and occupational injuries: a review of theepidemiologic evidence. Annu Rev Public Health. 1988;9:99-121.8. Howland J, Smith GS, Mangione T, Hingson R, De-Jong W, Bell N. Missing the boat on drinking and boat-ing. JAMA. 1993;270:91-92.9. Office of Boating Safety. US Coast Guard Web site.Boating under the influence. Available at: http://www.uscgboating.org/saf/saf_bui.asp. Accessed Septem-ber 12, 2000.10. Hoxie P, Cardosi K, Stearns M, Mengert P. Al-cohol in Recreational Boating Accidents. Washing-ton, DC: US Coast Guard; 1988. Publication DOT-CG-D-04-88.11. Canadian Red Cross Society. National Drown-ing Report: An Analysis of Drownings and Other Wa-ter-Related Injury Fatalities in Canada for 1997. Vi-sual surveillance report. Ottawa, Ontario: The CanadianRed Cross Society; 1999.12. Lunetta P, Penttila A, Sarna S. Water traffic ac-cidents, drowning and alcohol in Finland, 1969-1995. Int J Epidemiol. 1998;27:1038-1043.13. Howland J, Mangione T, Hingson R, Levenson S,Altwicker A. A pilot survey of aquatic activities andrelated consumption of alcohol with implications fordrowning. Public Health Rep. 1990;105:415-419.14. Wright S. SOS: alcohol, drugs and boating. Al-cohol Health Res World. 1985;9:28-30.15. National Transportation Safety Board. BoatingSafety: Safety Study. Washington, DC: National Trans-portation Safety Board; 1993. Publication SS-93-01,notation 6035, PB83-917006.16. Stiehl C. Alcohol and Pleasure Boat Operators.Washington, DC: US Coast Guard; 1975. PublicationCG-D-134-75.17. Branche CM, Conn JM, Annest JL. Personal wa-tercraft-related injuries: a growing public health con-cern. JAMA. 1997;278:663-665.18. Jones CS. Epidemiology of personal watercraft-related injury on Arkansas waterways, 1994-1997:

identifying priorities for prevention. Accid Anal Prev.2000;32:373-376.19. Wintemute GJ, Teret SP, Kraus JF, Wright M. Al-cohol and drowning: an analysis of contributing fac-tors and a discussion of criteria for case selection. Ac-cid Anal Prev. 1990;22:291-296.20. Levine B, Smith ML, Smialek JE, Caplan YH. In-terpretation of low postmortem concentrations of etha-nol. J Forensic Sci. 1993;38:663-667.21. ONeal CL, Poklis A. Postmortem production ofethanol and factors that influence interpretation: a criti-cal review. Am J Forensic Med Pathol. 1996;17:8-20.22. Rubin DB, Schenker N. Multiple imputation inhealth-care databases: an overview and some appli-cations. Stat Med. 1991;10:585-598.23. Mander A. Whotdeck.ado, whotdeck.hlpa ver-sion of hotdeck that uses logistic regression. STATA,Adrian Manders Web site, Medical Research Coun-cil, Biostatistics Unit, Cambridge, United Kingdom.Available at: http://www.mrc-bsu.cam.ac.uk/personal/adrian/stata.shtml. Accessed April 3, 2001.24. Moskowitz H, Burns M, Ferguson S. Police offic-ers detection of breath odors from alcohol ingestion.Accid Anal Prev. 1999;31:175-180.25. Wells JK, Greene MA, Foss RD, Ferguson SA, Wil-liams AF. Drinking drivers missed at sobriety check-points. J Stud Alcohol. 1997;58:513-517.26. Carlson WL. Estimation of nonrespondent BACusing a priori judgement. Accid Anal Prev. 1979;11:35-41.27. Moskowitz H, Fiorentino D. A Review on theEffects of Low Doses of Alcohol on Driving-RelatedSkills. Washington, DC: National Highway TrafficSafety Administration; 2000. Publication DOTHS-809-028.28. Howland J, Rohsenow DJ, Cote J, Siegel M, Man-gione TW. Effects of low-dose alcohol exposure onsimulated merchant ship handling power plant op-eration by maritime cadets. Addiction. 2000;95:719-726.

29. McKnight AJ, Smith GS, Marques PR, Lange JE.The Effects of Alcohol Upon Human Functioning inRecreational Boating. Landover, Md. National PublicServices Research Institute, 1994: Final Report, TheU.S. Coast Guard, Grant No. 1201.91.30. Moskowitz H, Burns M, Fiorentino D, Smiley A,Zador P. Driver Characteristics and Impairments atVarious BACs. Washington, DC: National HighwayTraffic Safety Administration; 2000. Publication DOTHS-809-075.31. Li G, Keyl PM, Smith GS, Baker SP. Alcohol andinjury severity: reappraisal of the continuing contro-versy. J Trauma. 1997;42:562-569.32. Council on Scientific Affairs. Alcohol and the driver.JAMA. 1986;255:522-527.33. Borkenstein RF, Crowther RF, Shumate RP, ZielWB, Zylman R. The role of the drinking driver in traf-fic accidents: the Grand Rapids Study. Blutalkohol.1974;II(1):1-132.34. Zador PL, Krawchuk SA, Voas RB. Alcohol-related relative risk of driver fatalities and driver in-volvement in fatal crashes in relation to driver age andgender: an update using 1996 data. J Stud Alcohol.2000;61:387-395.35. Howland J, Mangione T, Hingson R, Heeren T,Bak S, and Centers for Disease Control. Alcohol useand aquatic activitiesUnited States, 1991. MMWRMorb Mortal Wkly Rep. 1993;42:675, 681-683.36. Howland J, Mangione TW, Minsky S. Percep-tions of risks of drinking and boating among Massa-chusetts boaters. Public Health Rep. 1996;111:372-377.37. American Red Cross. American Red Cross Na-tional Boating Survey: A Study of Recreational Boats,Boaters, and Accidents in the United States. Wash-ington, DC: American Red Cross and United StatesCoast Guard; 1991.38. Ciraulo DL, Smith P, Ciraulo SC. A trauma sys-tems assessment of boating safety: a comparison ofcommercial and recreational boating practices. AmSurg. 2000;66:604-607.

Experience, the universal mother of Sciences.Miguel de Cervantes Saavedra (1547-1616)




2974

Documents

Transcript of 2974