INJURY PREVENTION/ORIGINAL RESEARCH

A Positive Blood Alcohol Concentration Is the Main Predictorof Recurrent Motor Vehicle Crash

Andrea Fabbri, MD

Giulio Marchesini, MD

Massimo Dente, MD

Tiziana Iervese, MD

Marco Spada, MD

Alberto Vandelli, MD

From the Dipartimento di Urgenza-Accettazione, Azienda Unita Sanitaria Locale di Forlı, Forlı(Fabbri, Dente, Iervese, Spada, Vandelli); and ‘‘Alma Mater Studiorum’’ Universita diBologna, Bologna (Marchesini), Italy.

Study objective: The identification of risk factors for recurrent motor vehicle crashes is the basis forprevention, but few studies have been published on predictors of recurrence. Our objective is todetermine the main variables predicting recurrent crashes in subjects attending an emergencydepartment (ED) for injuries after motor vehicle crash.

Methods: During a 5-year follow-up period, we studied 2,354 consecutive adult subjects treated in theED after a motor vehicle crash in 1998. The variables of the original event were tested for predictingrecurrence in a Cox proportional hazard model.

Results: During follow-up, 390 of 2,325 (16.8%) survivors were treated for injury after a new crash. Theoverall event rate was 34 per 1,000 subject-years. Four variables (age%32 years, male sex, nighttimecrash, and blood alcohol concentration O50 mg/dL) were identified as independent predictors ofrecurrent crash. After adjustment for sex, age, and nighttime, alcohol was the leading predictor(relative risk 3.73; 95% confidence interval 3.00 to 4.64). In the presence of the 4 variables, therecurrence rate was as high as 145 (117 to 175) events per 1,000 subject-years, and alcohol per seaccounted for more than 75% of events. In the absence of the 4 variables, the rate was as low as 11(7 to 17) events per 1,000 subject-years.

Conclusion: Alcohol was the most powerful behavioral factor predicting recurrent events in subjectstreated in an ED for injury after motor vehicle crash, along with young age, male sex, and nighttime.[Ann Emerg Med. 2005;46:161-167.]

0196-0644/$-see front matterCopyright ª 2005 by the American College of Emergency Physicians.doi:10.1016/j.annemergmed.2005.04.002

INTRODUCTIONBackground

Motor vehicle crashes represent a leading cause of morbidityand mortality, particularly in young people, carrying animmeasurable human cost, as well as an enormous burden tosociety.1-3 Chronic alcohol abuse and psychopathologicalfeatures have long been reported as specific features of high-riskcrash victims involved in the most severe events.4 A reviewof surveys done in European Union countries found that 1%to 3% of drivers drive under the influence of alcohol.2 Inaddition, a positive blood alcohol concentration is presentin 33% to 69% of fatally injured drivers, as well as in 8% to29% of drivers who were involved in a crash but not fatallyinjured.5 Pedestrians also put themselves at greater riskof road traffic injury when they consume too muchalcohol.6

Volume 46, no. 2 : August 2005

Upper limits of blood alcohol concentration are set at50 mg/dL for adult drivers in most European countries. In theUnited States, the limits vary from 80 to 100mg/dL, and a reviewof the experience in 16 US states7 found that the states withlower limits had 7% fewer alcohol-related motor vehicle crashes.

An alcohol-related emergency department (ED) visit wasshown to predict future problem drinking, alcohol-impaireddriving, recurrent injuries, and premature death.4,8 In thesehigh-risk subjects, recurrent motor vehicle crashes were associatedwith positive blood alcohol levels and longer hospital stay.9 In asurvival analysis, young age was shown to be the leading factor inrecurrent drunk-driving motor vehicle crashes; in particular, if adriver’s first drunk-and-drive offence occurred at a young age, heor she was significantly more likely to drink, drive, and crashagain.10 To our knowledge, the relationship between a single EDvisit for motor vehicle crash and later morbidity and mortality for

Annals of Emergency Medicine 161

Positive Blood Alcohol Concentration and Recurrent Motor Vehicle Crash Fabbri et al

Editor’s Capsule Summary

What is already known on this topic

People who drink alcohol and drive are more likely to beinvolved in motor vehicle crashes and engage in otherhigh-risk behaviors.

What question this study addressed

Patients who were involved in alcohol-related motorvehicle crashes were followed up throughout a 5-yearperiod to determine whether they were at an increasedrisk for recurrent crashes compared with crash victimsnot using alcohol.

What this study adds to our knowledge

Compared with individuals whose index crash did notinvolve alcohol, patients whose index crash involvedalcohol were more than 3 times more likely to beinvolved in another crash during the next 5 years.Seventy-five percent of recurrent crashes involved personswhose index crash involved alcohol.

How this might change clinical practice

Emergency physicians should be aware that patientsinvolved in alcohol-related vehicular accidents are at highrisk of recurrence and may consider actions designed tomitigate that risk.

injuries after recurrent motor vehicle crash has never been testedin the general population. In 1998, we studied all subjectsconsecutively admitted to our ED within 4 hours after motorvehicle crash.11 We found that an estimated positive bloodalcohol concentration at the crash (R50 mg/dL) was associatedwith specific features of subjects involved, types of injuries, andhigh-risk characteristics of the event.

ImportanceThe factors associated with an increased risk of motor vehicle

crash, namely, a positive blood alcohol concentration, mightalso be predictors of recurrence, identifying subjects with riskybehaviors. If a single positive blood alcohol concentration wasassociated with recurrence, emergency physicians might play acrucial role in preventing alcohol-related motor vehicle crashinjury and death.12 They could counsel subjects for drinkingand driving, when the subjects’ motivation is definitely higher.12

Brief interventions might be performed in high-risk subjectsselected on the basis of specific characteristics, with the aim todecrease the incidence of further events.13,14

Goals of This InvestigationThe objective of this study was to determine the incidence

rates of visits for a new motor vehicle crash and the mainpredictors of recurrence in our population, with specialemphasis on positive blood alcohol concentration.

162 Annals of Emergency Medicine

MATERIALS AND METHODSStudy Design

We conducted a prospective cross-sectional study of allsubjects treated in the ED of the district hospital of Forlıfrom January 1, 1998, to December 31, 1998, after a motorvehicle crash, including subjects hit by a motor vehicle whilewalking or cycling.7 The hospital (600 acute beds) serves an areaof 2,380 km2 (170,000 inhabitants). It is the only referralhospital where subjects involved in motor vehicle crashes areadmitted through the emergency transportation system. Thehospital database is directly connected with the General RegisterOffice of the district, and any previous admission to hospitalis available online.

The records of motor vehicle crash visits comprised personaldata and the main features of the crash, as previously reported.11

Informed consent for blood alcohol determination was obtainedfrom subjects or from an immediate relative when subjectswere unable to give their consent to take part in the study.Subject anonymity was kept by use of coded numbers.11 Theprotocol was carried out according to the Helsinki Declarationand approved by the local ethical committee. A projectcoordinator regularly checked the adherence to the protocol toensure the completeness of information.

SettingA systematic search of all subjects surviving the initial injury

was carried out by checking the death certificates and themedical databases of our local health district through December31, 2003, to determine their status at end of follow-up. Allvisits in the ED after an additional motor vehicle crash weremanually tracked through computerized records. The search wascarried out by crossing family name, address, date of birth,district health code, and coded reasons for visit/admittance ofall index cases as source of identification. All data wereabstracted from medical records by experienced recordabstractors who were unaware of the reasons for search andstudy protocol. The few conflicting data (7 cases) were handledby one of the study investigators (A.V.) unaware of the originalblood alcohol concentration. Subjects not revisiting the EDfor motor vehicle crash were considered recurrence free andwere censored December 31, 2003.

Selection of ParticipantsOriginal eligibility criteria were age 14 years or older, a

motor vehicle crash in the 4 hours before visit in the ED, andno history of acute or chronic alcohol abuse. Subjects withcomplex or severe injuries, transferred to extra-area traumacenters, were also included and followed up until hospitaldischarge. Of a total number of 4,637 subjects observed duringthe study period, we excluded 35 subjects (0.8%) who died atthe crash scene, 1,070 subjects admitted later than 4 hours,and 676 subjects aged younger than 14 years. Among theremaining 2,856 eligible subjects, only 502 subjects did nothave alcohol testing (87 subjects for protocol errors, 149subjects because of previous admissions for harmful drinking,

Volume 46, no. 2 : August 2005

Fabbri et al Positive Blood Alcohol Concentration and Recurrent Motor Vehicle Crash

and 266 subjects because they refused blood alcoholdetermination). The final alcohol testing rate of eligible subjectswas 82.4% (2,354 of 2,856).

Methods of Measurement and Data Collection andProcessing

At initial evaluation and at follow-up, all subjects werecategorized as pedestrians, bicyclists, motorcyclists, drivers, orpassengers in a car or a truck. For all subjects, information wasrecorded about the day of the week (workday from Monday at8 AM to Friday at 8 PM versus the weekend) and time of day(from 8 AM to 8 PM, daytime versus nighttime).

The mechanism of injury was classified on the basis of themain features of the crash. They were obtained by subjectsthemselves, emergency system transportation logs, and policereports.11 No internationally accepted classification is available;we simply classified the mechanisms in relation to vehicletype and subjects involved.

Blood samples for alcohol determination were confidentiallydrawn by either the emergency medical system personnel at thecrash scene, before any liquid infusion was given, or at arrival inthe ED. Blood alcohol was measured by ALC (Dade Bering Inc.,Newark, DE). The method has a lower detection limit of10 mg/dL and a coefficient of variation of 0.5% (intra-assay) and1.9% (interassay) at lower concentration (range 44 to 48 mg/dL)and 0.6% and 5.2%, respectively, at higher concentration(range 123 to 151 mg/dL). The theoretical blood alcoholconcentration at the crash was calculated by reverse extrapolation:the mean rate of ethanol elimination was multiplied by theestimated crash-to-sampling interval and added to the measuredblood alcohol concentration.We considered a correction factor of20 mg/dL for every hour between crash and sampling time.15

The blood alcohol limit was set at 50 mg/dL.Injuries were coded by experienced medical personnel

according to the Abbreviated Injury Scale for any body region(head/neck, face, chest, abdomen, extremities, external). TheNew Injury Severity Score was calculated as the score derivedfrom the 3 most severe injuries, regardless of body region.16 TheAbbreviated Injury Scale is systematically used in our ED tocategorize all injuries.17

Outcomes MeasuresAny revisit for injury after a further motor vehicle crash was

considered a main outcome measure (presence or causalrelationship of alcohol to the subsequent motor vehicle crashwas not examined). Readmissions for treatment of the sameinjury were excluded from analysis. In subjects readmitted morethan once for new trauma injuries, only the first event wasconsidered in the analysis.

Primary Data AnalysisMean values and SD, medians and interquartile ranges,

and frequencies were used to describe data distribution. Coxproportional hazards models were then used to obtainestimates and 95% confidence intervals for the relative risk of

Volume 46, no. 2 : August 2005

recurrence in relation to positive alcohol concentration,controlling for subjects’ characteristics. We included in theanalysis demographic data of the index case (sex, age), thecharacteristics of motor vehicle crash (dangerous mechanismof injury), time-of-day variables (nighttime versus daytime,weekend versus workday), injury severity, and blood alcoholdetermination. The assumption of proportional hazard waschecked with analytic and graphic methods (log minus log plotof the cumulative hazard function). These variables were selectedon the basis of previous reports and a reasonable relevance forassociation between injuries and alcohol. Person-time estimateswere calculated using an actuarial method. All analyses wereperformed with SPSS/PCC statistical package on a personalcomputer.18

RESULTSCharacteristics of Study Subjects

Of the 2,354 enrolled subjects in the original study, 29 diedduring hospital admission. The 2,325 survivors were followedup either until December 31, 2003, or to a new visit forrecurrent motor vehicle crash. The average follow-up was4.87G1.37 years. Of the 2,325 survivors, 59 (2.5%) died duringfollow-up, 9 (0.4%) subjects from motor vehicle crash and 24(1.0%) subjects from diseases independent of injury. Theremaining 26 subjects died after recurrent motor vehicle crash;10 (0.4%) subjects died from diseases unrelated to motorvehicle crash, and 16 subjects died from a further motor vehiclecrash. In total, 390 subjects (16.8%) were visited for 579recurrent motor vehicle crash. In particular, 99 subjects (4.3%)had 2 recurrences, whereas 36 (1.5%) had 3 or more.

Main ResultsThe main features of injured subjects in the original cohort,

grouped according to recurrence, are reported in Table 1. Apositive blood alcohol concentration was also associated withmultiple visits for recurrent motor vehicle crash in 24 subjects(68.3%). The overall event rate was 34 per 1,000 subject-yearsof follow-up (95% confidence interval [CI] 31 to 38).

The association between the 10 main crash characteristicsand recurrence of injury after road crash was tested bymultivariate logistic regression analysis (Table 2). Overall, 5 ofthe 10 variables considered were significantly associated withrecurrence. The presence of any of these variables wasconsidered under the heading ‘‘dangerous mechanism of injury’’in the final Cox proportional hazard model.

In the final Cox model, 8 variables were tested for predictionof recurrent injury (age%32 years, male sex, nighttime, andpositive blood alcohol concentration, type of vehicle, NewInjury Severity Score O9, weekday, and dangerous mechanismof injury). Only 4 variables were selected as significant, and apositive blood alcohol was the main predictor variable (relativerisk 3.73; 95% CI 3.00 to 4.64; P\.001) (Table 3). Analyticand graphic methods showed that the proportionalityassumption of the model was not violated (not reported indetail). The final model showed a good overall accuracy

Annals of Emergency Medicine 163

Positive Blood Alcohol Concentration and Recurrent Motor Vehicle Crash Fabbri et al

Table 1. Original clinical characteristics and crash features in subjects treated in the ED for recurrent motor vehicle crash in a 5-yearfollow-up.

Baseline

Characteristics

Cases,

No.

Recurrent Motor

Vehicle Crash Percentage OR (95% CI)

Sex

Males 1,372 301 21.9 2.73 (2.12 to 3.51)Females 953 89 9.3Age, y

14–20 306 52 17.0 1.0021–30 716 161 22.5 1.42 (1.00–2.00)31–40 411 90 21.9 1.37 (0.94–2.00)41–50 265 37 14.0 0.79 (0.50–1.25)51–60 190 17 8.9 0.48 (0.27–0.86)O60 437 33 7.6 0.40 (0.25–0.63)Weekday

Weekend 869 170 19.6 1.37 (1.10–1.70)Working day 1,456 220 15.1Time of day

Nighttime 714 200 28.0 2.91 (2.33–3.64)Daytime 1,611 190 11.8Blood alcohol concentration

\50 mg/dL 1,906 204 10.7 6.66 (5.23–8.48)R50 mg/dL 419 186 44.4

(area under the curve: 0.751Gstandard error 0.014), with highsensitivity in the prediction of recurrent injury (97.2; 95% CI94.9 to 98.3) but a very low specificity (16.0; 95% CI 14.4 to17.7) (Table 4).

The cumulative incidence of injury for recurrentmotor vehiclecrash over time was markedly increased in subjects with positiveblood alcohol concentration (Figure). In the presence of the 4predictive variables, the recurrence rate was as high as 145 eventsper 1,000 subject-years (95% CI 117 to 175 events). In thepresence of a positive blood alcohol concentration, the event ratewas 112 per 1,000 subject-years (95% CI 98 to 128 events)(ie, alcohol per se accounted for approximately 75% of totalrecurrence rate). In the absence of the 4 variables, theestimated rate was as low as 11 events per 1,000 subject-years(95% CI 7 to 17 events).

Table 2. Multivariate analysis of mechanism-of-injury variablesof the original crash for subjects treated for recurrent crash in a5-year follow-up.

Crash Characteristics Odds Ratio 95% CIs

Out-of-lane* 2.32 1.81–2.98Overtaking* 2.36 1.41–3.94Not moving obstacle* 1.72 1.16–2.54Ejected* 6.90 3.32–14.3Driver* 1.37 1.03–1.82Head on 1.20 0.93–1.56Rollover 1.23 0.72–2.11Bend road 1.28 0.94–1.74Out-of-town 0.94 0.72–1.21No belts 0.96 0.73–1.25

*The presence of any of the variables significantly associated with events was

entered in the final Cox model under the heading ‘‘dangerous mechanism of

injury.’’

164 Annals of Emergency Medicine

At readmission, blood alcohol was positive in 225 of 579readmitted subjects (38.9%), the median concentration being167 mg/dL (interquartile range 137 to 240 mg/dL). Thecharacteristics of injury in alcohol-positive cases were similar tothose previously found in the original cohort (not reported indetail).

LIMITATIONSIn our study, we tracked all subjects previously admitted to

the ED for injury after motor vehicle crash. The originalpopulation comprised more than 80% of eligible subjects, andhabitual alcohol drinking and refusal of blood alcohol testingwere the leading causes for incomplete testing. Excluding theseconditions underestimates the effect of alcohol on recurrent

Table 3. Cox proportional hazard model: variables predictingrecurrent motor vehicle crash in relation to clinical parametersand characteristics of the original crash.

All Cases Relative Risk* 95% CIs

Male sex 1.84 1.44–2.34Age%32 y 1.26 1.01–1.56AlcoholR50 mg/dL 3.66 2.94–4.57Nighttime 1.57 1.25–1.97Type of vehicle 1.01 0.94–1.08NISS O9 1.14 0.90–1.43Weekday 1.05 0.85–1.29Dangerous mechanism 1.20 0.88–1.64

NISS, New Injury Severity Score.

*Relative risk was calculated considering sex, nighttime, blood alcohol concen-

trationR50 mg/dL, age%32 years, nighttime NISSO9, weekday, and dangerous

mechanism as dichotomous variables.

Volume 46, no. 2 : August 2005

Fabbri et al Positive Blood Alcohol Concentration and Recurrent Motor Vehicle Crash

motor vehicle crash, which was previously demonstrated inchronic alcoholic patients.4

Our conclusions are based on the assumption that no eventswere missed at follow-up, so that censoring could be done byDecember 31, 2003. Outcome data were tested on the hospitaldatabase without searching for admissions to EDs out of ourarea district. This possibility is unlikely to modify the resultssignificantly, considering the characteristics of our health districtand the study population. Also, the number of subjects excludedfrom the analysis was kept to a minimum in the originaldatabase, in which only 8.3% of eligible subjects were notincluded.11 The possibility that subjects were involved in minormotor vehicle crashes not resulting in ED treatment or in motorvehicle crashes that resulted in either injuries to others or just inproperty damage cannot be tested within the frame of our studydesign.

Previous studies on recurrent motor vehicle crash werehospital-based or trauma center–based, including only subjectsin severe conditions. This policy increases the case-fatality ratebut may conceal the importance of moderate blood alcohollevels. In our institution, a district ED, minor injuries also areadmitted, thus allowing us to recruit a large population with alow number of events in terms of severe injury or mortality.

We obtained crash characteristics by subjects themselves,emergency system transportation logs, and police reports. Theyare difficult to categorize and largely differ in relation to thevehicle and subjects involved (bicycle, motorcycle, car), oraccording to the source of report (subjects themselves, police,emergency system personnel).19 When the crash scene isexamined, information on crash dynamics, angle of impact, andactual speed is difficult to obtain. For statistical purposes, crashcharacteristics were grouped, but they were nonethelessexcluded from the final model.

The use of reverse extrapolation to theoretical blood alcohollevels at crash introduces a potential source of error in cases ofabnormal alcohol elimination rate or very recent alcoholingestion, when alcohol levels increase with time. The elim-ination rate assumed in the formula is the minimum expected innormal-weight population.15 Again, this assumption gives apotential underestimate of alcohol on recurrent motor vehiclecrash.

Table 4. Sensitivity and specificity of variables included in theCox proportional hazard model in the prediction of recurrence ofmotor vehicle crash.

Recurrence of Motor Vehicle Crash

Yes No

Variables present 379 1625Variables absent 11 310Sensitivity, % 97.2

(95% CI 94.9–98.3)97.2

(95% CI 94.9–98.3)Specificity, % 16.0

(95% CI 14.4–17.7)16.0

(95% CI 14.4–17.7)

Volume 46, no. 2 : August 2005

The outcome measures were limited to ED visits for recurrentmotor vehicle crash. Admissions for alcohol-related problemsor other illnesses, however interesting to strengthen theassociation of alcohol and crash trauma, were not considered.

DISCUSSIONCrash recidivism represents an immediate, serious health

hazard. This study demonstrates that it is possible to predictrecidivism on the basis of sociodemographic and behavioralcharacteristics in subjects involved in motor vehicle crash. Inparticular, a positive blood alcohol concentration at the crashsignificantly increases the hazard for injuries for recurrent motorvehicle crash soon after the index admission (see Figure). Thisassociation is true for blood alcohol concentration of 50 mg/dLor greater (ie, above the legal limit allowed in most Europeancountries). By the use of this cutoff, alcohol remains the leadingfactor, being present in approximately 75% of events, faroutweighing the importance of sex, age, and time of day. Whendata were recalculated using the most common US standardof 80 mg/dL (n=254 alcohol-positive subjects involved inrecurrent motor vehicle crash), the results do not change: apositive blood alcohol concentration remains the leading factorpredicting recurrence (relative risk 3.83; 95% CI 3.04 to 4.81),together with male sex, nighttime, and young age.

Intoxicated drivers are involved in more crashes; they needmore health care resources and are at increased likelihood ofdeath after a traffic injury.20 The majority of studies focused onthe relationship between chronic alcohol abuse or alcoholismand motor vehicle crash. In a retrospective study, a singlealcohol-related ED visit was an important predictor of chronicdrinking problems, alcohol-impaired driving, and prematuredeath,8 with 37% of intoxicated subjects having at least a furtheralcohol-related ED visit in a 5-year follow-up. Acute andchronic alcohol abuse also affects the chance for recurrenttrauma and readmission to an urban trauma center.4 In an old

Figure. Incidence of recurrent injury after motor vehicle crash byhazard function in subjects with positive (upper line) andnegative (lower line) blood alcohol concentration.

Annals of Emergency Medicine 165

Positive Blood Alcohol Concentration and Recurrent Motor Vehicle Crash Fabbri et al

review study,21 although most alcohol-positive drivers wereinitially considered social drinkers, subsequent evidence sug-gested that 30% to 50% of drivers with alcohol-related crashesor citations have drinking problems. Studies on drinkingproblems among drivers admitted because of motor vehiclecrash confirm that approximately 18% meet criteria for currentalcohol dependence.22

Owing to methodologic difficulties related to control ofconfounding factors, there are few epidemiologic data on theassociation between a single positive blood alcohol concentra-tion and recurrent motor vehicle crash, despite calls for suchdata.23 To further indicate the importance of alcohol, ouralcohol-positive cases were identified by reverse extrapolation toa theoretic blood alcohol concentration at crash of 50 mg/dL orgreater, considering a mean rate of ethanol elimination of20 mg/dL per hour.15 When measured in the ED, the bloodconcentration of alcohol-positive subjects was between 10 and50 mg/dL in 178 of 425 cases (41.9%), a value below the legallimits in most European countries.2,3

Drinking drivers could be a specific target population foralcohol intervention in the ED.24 In our series, they account forapproximately 34% of subjects involved in recurrent car crashes.However, we did not limit our analysis to drivers at risk, becauseevidence suggests that intoxicated passengers injured in crashesare at equal risk for driving under the influence of alcohol.25

Thus, intoxicated drivers and passengers should be specificallytargeted for alcohol screening. Their presence in the ED is anopportunity to intervene to prevent future events. Emergencyphysicians are in a unique position to experience the scope ofthe problem and to intervene directly to prevent motor vehiclecrash and injury recidivism.12-14

Subjects admitted to the ED for alcohol-related injury maybe considered to be in a particularly favorable ‘‘teachablemoment’’ with regard to alcohol abuse.26,27 Interventionstrategies should be based on routine screening for alcohol use,with brief counseling aiming at reducing risky behaviorsinvolving alcohol and driving. All these interventions, whenconducted in whole or in part in the ED, showed a positiveeffect.13,14 Counseling takes less than 45 minutes and can beconducted during ongoing medical care such as suturing orsplinting. A multidisciplinary team approach using socialworkers, case managers, nurses, and other personnel mightbe a suitable solution to relieve the duty for busy emergencyphysicians. Nonetheless, also a brief interview (5 minutes)was shown to decrease the rate of reinjury requiring an EDvisit or readmission and to reduce the reported alcoholconsumption by an average of 15 drinks per week a year afterintervention.12

In summary, emergency physicians and the health caresystem should extend their approach to subjects after a roadcrash beyond the injury and its specific treatment and addressfactors that increase the chance of relapse with repeated motorvehicle crash and related injuries. Alcohol remains the mainpredictor of subsequent motor vehicle crash, and interventionmay reduce the burden it produces. According to our data,

166 Annals of Emergency Medicine

young male subjects who are admitted to the ED after anighttime motor vehicle crash and have a positive blood alcoholconcentration are the primary target for any therapeuticintervention.

The authors are indebted to Calogero Camma, MD,Universityof Palermo, for helpful support in statistical analyses and for helpfulcriticism and comments.

Supervising editor: Debra E. Houry, MD, MPH

Author contributions: AF conceived the study, wrote theprotocol, and coordinated data collection. AF and GM wrotethe paper. GM and AV conducted the study design, and GMconducted statistical analyses. MD, TI, and MS designed theprotocol and collected data. MD, TI, MS, and AV were involvedin the study coordination. AV conducted critical review of thepaper. All authors contributed to data interpretation andapproved the final version of the paper. AF takes responsibilityfor the paper as a whole.

Funding and support: The authors report this study did notreceive any grants or outside financial support or financialinterest.

Publication dates: Received for publication November 9, 2004.Revisions received January 24, 2005, February 2, 2005, andMarch 24, 2005. Accepted for publication April 5, 2005.Available online June 13, 2005.

Address for reprints: Andrea Fabbri, MD, Dipartimentodell’Emergenza, Ospedale Morgagni-Pierantoni,Azienda USL di Forlı, Via Forlanini 34, I-47100 Forlı, Italy;39-0-543-735151, fax 39-0-543-735163;E-mail andfabbri@libero.it.

REFERENCES1. World Health Organization. The world report on road traffic injury

prevention [World Health Organization Web site]. Available at:http://www.who.int/world-health-day/2004/infomaterials/world_report/en/summary_en_rev.pdf. Accessed January 12,2005.

2. European Commission. White paper: Office for Official Publicationson the European Communities, L-2985 Luxemburg. Available at:http://europa.eu.int/comm/energy_transport/en/lb_en.html.Accessed September 24, 2004.

3. European Commission. Directorate-General for Energy andTransport. Available at: http://europa.eu.int/comm/transport/road/library/rsap/memo_rsap_en.pdf. Accessed August 24,2004.

4. Rivara PF, Koepsell TD, Jurkovich GJ, et al. The effects of alcoholabuse on readmission for trauma. JAMA. 1993;270:1962-1964.

5. Odero WO, Zwi AB. Alcohol-related traffic injuries and fatalities inLMICs: a critical review of literature. In: Kloeden CN, McLean AJ,eds. Proceedings of the 13thInternational Conference on Alcohol,Drugs and Traffic Safety, August 13-18 1995, Adelaide, Australia.Adelaide, Australia: Road Accident Research Unit; 1995:713-720.

6. Keigan M, et al. The Incidence of Alcohol in Fatally Injured AdultPedestrians. Crowthorne: Transport Research Laboratory; 2003.TRL Report 579. Available at: www.trl.co.uk/static/dtlr/pdfs/TRL579.pdf. Accessed January 12, 2005.

7. Shults RA, Elder RW, Sleet DA, et al. Reviews of evidence regardinginterventions to reduce alcohol-impaired driving. Am J Prev Med.2001;21:66-88.

Volume 46, no. 2 : August 2005

Fabbri et al Positive Blood Alcohol Concentration and Recurrent Motor Vehicle Crash

8. Davidson P, Koziol-McLain J, Harrison L, et al. Intoxicated EDpatients: a 5-year follow-up of morbidity and mortality. Ann EmergMed. 1997;30:593-597.

9. Kaufmann CR, Branas CC, Brawley ML. A population-based study oftrauma recidivism. J Trauma. 1998;45:325-331.

10. Ferrante AM, Rosman DL, Marom Y. Novice drink drivers, recidivismand crash involvement. Accid Anal Prev. 2001;33:221-227.

11. Fabbri A, Marchesini G, Morselli-Labate AM, et al. Positive bloodalcohol concentration and road accidents: a prospective study in anItalian emergency department. Emerg Med J. 2002;19:210-214.

12. McDonald AJ 3rd, Wang N, Camargo CA Jr. US emergencydepartment visits for alcohol-related diseases and injuries between1992 and 2000. Arch Intern Med. 2004;164:531-537.

13. D’Onofrio G, Degutis LC. Preventive care in the emergencydepartment: screening and brief intervention for alcohol problemsin the emergency department: a systematic review. Acad EmergMed. 2002;9:627-638.

14. Crawford MJ, Patton R, Touquet R, et al. Screening and referral forbrief intervention of alcohol misusing patients in an emergencydepartment: a pragmatic randomised controlled trial. Lancet.2004;364:1334-1339.

15. Gersham H, Steeper J. Rate of clearance of ethanol from the bloodof intoxicated patients in the emergency department. J Emerg Med.1991;9:307-311.

16. Osler T, Baker SP, Long W. A modification of the injury severityscore that both improves accuracy and simplifies scoring. J Trauma.1997;43:922-926.

17. Fabbri A, Servadei F, Marchesini G, et al. Prospective validation of aproposal for diagnosis and management of patients attending theemergency department for mild head injury. J Neurol NeurosurgPsychol. 2004;75:410-416.

Volume 46, no. 2 : August 2005

18. Norusis MJ. SPSS/PCC4.0 Base: Statistics and AdvancedStatistics Manuals for the IBM PC/XT/AT and PS/2. Chicago, IL:SPSS Inc; 1990.

19. Muellerman RL, Mueller K. Fatal motor crashes: variations of crashcharacteristics within rural regions of different population densities.J Trauma. 1996;41:315-320.

20. Waller PF, Stewart JR, Hansen AR, et al. The potentiating effects ofalcohol on driver injury. JAMA. 1986;256:1461-1466.

21. Vingilis E. Drinking drivers and alcoholics: are they from the samepopulation? Res Adv Alcohol Drug Probl. 1983;7:299-342.

22. Soderstrom CA, Dischinger PC, Smith GS, et al. Alcoholism at thetime of injury among trauma center patients: vehicular crashvictims compared with other patients. Accid Anal Prev. 1997;29:715-721.

23. McKay MP. Commentary: you drink & drive, you lose: reachingthe target audience is not enough. Ann Emerg Med. 2004;44:156-159.

24. National Highway Traffic Safety Administration, National Center forStatistics and Analysis, Fatality Analysis Reporting System. Queriedby driver of vehicle in transport, previous DUI/DWI conviction, andpolice reported alcohol involvement. Available at: http://www-fars.nhtsa.dot.gov/finalreport.cfm?title=Trends&stateid=0&year=2002&title2=Alcohol. Accessed August 24, 2004.

25. Schermer CR, Apodaca TR, Albrecht RM. Intoxicated motor vehiclepassengers warrant screening and treatment similar to intoxicateddrivers. J Trauma. 2001;51:1083-1086.

26. Chang G, Astrachan BM. The emergency department surveillance ofalcohol intoxication after motor vehicle accidents. JAMA. 1988;260:2533-2536.

27. Longabaugh R, Minugh PA, Nirenberg TD. Injury as a motivator toreduce drinking. Acad Emerg Med. 1995;2:817-825.

Annals of Emergency Medicine 167