Schaffer Online Library of Drug Policy Sign the Resolution
Contents | Feedback | Search
DRCNet Home
| Join DRCNet
DRCNet Library | Schaffer Library | Drugs and Driving 

 


Crash Characteristics and Injuries of Drivers Impaired by Alcohol/Drugs

PF Waller, FC Blow, RF Maio, EM Hill, K Singer, N Schaefer

The University of Michigan Transportation Research Institute, 2901 Baxter Road, Ann Arbor, Michigan 48109-2150, USA

ABSTRACT

This study is based on a sample of 717 drivers presenting to two emergency departments for treatment of motor vehicle injury. Data were collected on presence of alcohol and drugs, demographic factors, history of alcohol and drug abuse, crash characteristics, and measures of injury. Based on analyses of blood samples drawn within six hours of the crash, alcohol was found to be the major drug associated with injury. Marijuana, cocaine, and opiates were identified in slightly over 14 percent of the drivers, but almost half of these also had elevated blood alcohol levels. Those patients testing positive for drugs but not alcohol had crashes that were very similar to those of drivers testing negative for both alcohol and drugs. These findings are not consistent with studies reporting that illicit drugs are a major factor in motor vehicle crashes.

INTRODUCTION

The role of alcohol in motor vehicle crashes has been recognized almost since the advent of motor vehicles (U.S. Department of Transportation, 1968). As early as 1936, Norway enacted legislation establishing a BAC of 0.05% as the limit for operating a motor vehicle. Still, alcohol persists as a major factor in motor vehicle injury (Cherpitel, 1993). The role of drugs other than alcohol is less clear, particularly when it comes to how extensively they are involved in real world crashes.

Several recent studies have reported drugs in patients admitted to hospital for treatment of injury. Soderstrom et al. (1989) reported the presence of marijuana in almost 35 percent of trauma victims admitted to a major trauma center, with motor vehicle crashes accounting for about two-third of the patients studied. Younger patients had a higher rate of marijuana use, and alcohol was more characteristic of motor vehicle crash victims. Other studies examining drug use in trauma victims include Lindenbaum et al. (1989), Rivara et al. (1989), and Brookhoff, et al. (1993). Drugs are reported in 38 percent to 75 percent of the patients studied. Studies limited to patients treated for motor vehicle injuries report around 40 percent testing positive for drugs other than alcohol ( Stoduto et al., 1991; Kirby et al., 1992).

Other studies report on fatally injured drivers, based on toxicological tests. Williams et al. (1985) reported on fatally injured young male drivers in four counties in California. Seventy percent tested positive for alcohol, 37 percent for marijuana, and 11 percent for cocaine. Each of 24 other drugs was identified in fewer than 5 percent of the drivers. Terhune et al. (1992) analyzed data from drivers from seven states, including operators of passenger cars, trucks, and motorcycles. Only drivers who died within four hours of the crash were included. Slightly over half the drivers tested positive for alcohol, 6.7 percent for marijuana, and 5.3 percent for cocaine.

Toxicological analyses conducted on a nationally representative sample of fatally injured truck drivers found that 33 percent tested positive for drugs of abuse (Sweedler and Quinlan, 1989; Crouch, et al., 1993). Alcohol and marijuana were the most frequently detected drugs, followed by cocaine, amphetamine, methamphetamine, and others.

Alcohol was the most frequently reported drug in emergency department patients and fatally injured drivers, except that in the sample of truck drivers, marijuana was found as often as alcohol (13% of the study sample). It should be noted that the rate of drug use, including alcohol, in fatally injured truck drivers was considerably lower than for fatally injured drivers in general. This finding is consistent with the lower rates of alcohol reported for crash-involved truck drivers in general.

There are important limitations with all of the studies described. First, the clinical studies are based on patients admitted to major trauma centers, so that only a limited portion of the injury spectrum is included. This restricted range of injury severely limits the possibility of determining how drugs may relate to traumatic injury. Second, many of the studies are based on urine screenings, and some drugs are detected days or even weeks after they were taken. Urine screenings do not provide good information on when the drugs were taken or whether the person was under the influence of the drug at the time of injury. Without more precise estimation of when the drugs were taken, it is not possible to determine the extent to which driving performance may have been affected. A third problem with many of the clinical studies is that only a relatively small proportion of eligible patients are included, and there is no explanation of how they were selected. Furthermore, there is no information on how the tested patients compare with those who were not tested. If testing is conducted only when drug use is suspected, the findings cannot be considered to be characteristic of the entire trauma population. Fourth, the studies rarely linked toxicology results to crash data to identify crash characteristics, so that it is not known how the crashes of drug positive drivers differ from those of drivers not using drugs.

The studies on fatally injured drivers are, of course, based on an even more restricted range of injury, thus severely limiting the meaning of the findings, particularly in regard to variables that may influence extent of injury. For example, if one is interested in how effective safety belts are in reducing or preventing injury, and only fatally injured drivers are considered, by definition every belted driver included is fatally injured. The many times that belt use reduced or prevented injury are not considered. However, some of the studies on fatally injured drivers include a reasonably representative sample of the fatally injured drivers of interest, particularly the study by Terhune et al.(1992) and those on the fatally injured truck drivers (Sweedler and Quinlan, 1989; Crouch et al., 1993).

The present study is based on 717 drivers of passenger cars, station wagons, vans, or pickup trucks who were treated for injury sustained in a motor vehicle crash. This study differs from previous studies in that (1) it includes the full spectrum of patients presenting to an emergency department for treatment of motor vehicle injury, both patients who are admitted and patients who are treated and released; (2) blood samples were drawn within six hours of the injury, providing information on drugs that may have affected performance at the time of the crash; (3) the extent of injury was determined by standardized injury scoring systems; (4) wherever possible, an extensive interview was conducted to determine past use and abuse of alcohol and other drugs; and, (5) all information was linked to crash data.

METHOD

Motor vehicle crash victims presenting for treatment were recruited into the study in two emergency departments, one in a large university hospital and the other in a large community hospital. Data collection extended over a 29-month period in one of the emergency departments and over a 15-month period in the other. All evening shifts (4 pm - Midnight) were covered in both hospitals during the respective data collection periods. A systematic sample of other shifts was also included from the university hospital. Excluded were patients transferred from other hospitals, patients < 18 years of age, pregnant patients, and patients who were institutionalized. Only patients brought directly to the emergency department were included.

Not every eligible patient could be included, even on those shifts that were covered. At times (particularly on icy mornings) the flow of patients was greater than could be processed by project personnel, and eligible subjects were missed. In addition, some patients refused participation. Basic data were compiled on these patients to determine how they may differ from those patients included in the study.

Data collected covered five main areas, namely, (1) demographic information, compiled from available records and the interview; (2) alcohol and drug history, based on the interview; (3) current alcohol and drug data, based on the analyses of the blood sample drawn and analyzed for alcohol, marijuana, cocaine, and opiates; (4) crash data, from the hard copies of the crash reports; and, (5) injury data, based on the crash report and hospital records.

All data were coded, and identifying information was used for purposes of record linkage only.

RESULTS

Comparison of Tested and Non-Tested Drivers

There were 278 drivers of passenger cars, trucks, vans, and station wagons who were eligible for the study but who were not tested for drugs. These were compared with the 717 drivers in the study, based on sex, age, education, race, marital status, history of alcohol or drug abuse or dependence, number of vehicles in the crash, TAD, MAIS 85, ISS 85, and hospital admission. There were no significant differences between groups found for any of these variables (p > .01).

The rest of the results are based on 717 drivers of passenger cars, station wagons, vans, and pickup trucks, representing the full range of injury. Findings are divided into four broad areas, namely Demographic Characteristics, Crash Variables, Driver Variables, and Injury Measures.

Because so many tests of significance were conducted, a strict criterion for reporting significance was applied, namely p < .001. Probabilities < .01 but > .001 are reported as trends deserving of further investigation.

Demographic Characteristics

The patient population cannot be considered representative of all such persons presenting to the emergency departments for treatment of motor vehicle injury, because not all shifts were covered in equal proportions. Evening shifts were deliberately oversampled, because they represented the greatest flow of the patients of interest. However, because all eligible patients presenting during the shifts that were covered were equally sampled, the study population is valid for making comparisons concerning the relationship between alcohol and drugs and other crash and patient characteristics.

Although it is recognized that alcohol is a drug, in this report, the term drug refers to marijuana, cocaine, or opiates.

Based on the analyses of whole blood, patients were categorized into four groups, namely, No Alcohol or Drugs, Alcohol Only, Drugs Only, and Alcohol plus Drugs. The mean time between injury and blood draw was slightly under one hour, and the maximum time was five hours, four minutes. Most of the drivers tested negative for both alcohol and drugs (N = 508, 70.85%). One hundred eight drivers (15.1%) tested positive for alcohol only, 56 (7.8%) for drugs only, and 45 (6.3%) for both alcohol and drugs. Of the drivers testing positive for drugs, marijuana was the drug most frequently identified (N = 91). Nine drivers tested positive for opiates and eight for cocaine, with a few drivers testing positive for more than one drug.

Sex, age, education, race, marital status, history of alcohol or drug use and abuse, current alcohol or drug use and abuse were examined in relation to each of the four identified groups.

Sex - The study sample as a whole had slightly more males (53.6%), and a disproportionate number of males tested positive for alcohol (p < .001). There was also a trend for such a relationship for drugs (p = .0035).

Age and Education - Age was associated with the use of drugs. Younger drivers had a higher probability of testing positive for illicit drugs (p < .001). There was also a tendency for younger drivers to be more likely to be using alcohol (p = .0011).

There was a trend for alcohol and drug use to be associated with slightly lower education level (p = .0016 for alcohol and .0011 for drugs). Drivers testing negative for both alcohol and drugs were older, so that the age difference may have accounted for some of the differences in education.

Race - Race was not associated with alcohol or drug use.

Marital Status - Drivers were classified as Married or Widowed versus Single, Separated, or Divorced. While alcohol use was not associated with marital status, drug use was, with Married/Widowed drivers less likely to test positive for drugs.

History of Alcohol and Drug Abuse - Almost one-third of the drivers (32%) had a history of alcohol abuse or dependence. Those testing positive for alcohol were significantly more likely to have such a history (77%, p < .001). There was also a tendency for those testing positive for drugs to have such a history (56%, p = .0078). Only 21.8 percent of drivers testing negative for both alcohol and drugs had a history of alcohol abuse or dependence.

Only 13.8 percent of the drivers had any history of drug abuse or dependence. However, those testing positive for drugs were more likely to have such a history (46.2%, p < .001). There was no significant relationship found for those testing positive for alcohol. Only 8.4 percent of drivers with neither alcohol nor drugs had a history of drug abuse.

Current Alcohol or Drug Abuse - About one-fifth (20.6%) of the drivers had evidence of current alcohol abuse or dependence. Not surprisingly, those testing positive for alcohol were more likely to have evidence of current abuse (72.9%, p < .001), and there was a tendency for drivers testing positive for drugs also to have evidence of current alcohol abuse or dependence (43.9%, p = .0013).

Very few drivers (5.6%) showed evidence of current drug abuse or dependence. However, drivers testing positive for drugs were more likely to show such evidence (32.3%, p < .001). No relationship was found for drivers testing positive for alcohol.

Perhaps the most interesting finding regarding alcohol and drug history is the large proportions of persons with evidence of alcohol or drug abuse or dependence who tested negative for both alcohol and drugs. Of those drivers with a history of alcohol abuse, over half (51.4%) were alcohol and drug free at the time of the injury. Likewise, of those with a history of drug abuse, 46.8 percent were alcohol and drug free. Those with evidence of current alcohol or drug problems were less likely to test negative for alcohol and drugs (32.2% of those with current evidence of alcohol abuse and 21.8% of those with current evidence of drug abuse). Still, these large proportions of persons who either have had or currently have evidence of alcohol and drug problems but who test negative at the time of injury provide encouragement for the possibility of combatting alcohol and drug use, particularly in combination with driving.

Crash Variables

Crash variables included number of vehicles in the crash (Single, Multiple), time of day (Day/Night), day of week (Weekday/Weekend), vehicle crush (Traffic Accident Damage scale, or TAD), and worst injury in the vehicle. Daytime/Nighttime crashes were defined as those occurring from 6:00 am to 5:59 pm versus 6:00 pm to 5:59 am. Weekday/ Weekend crashes were defined as 6:00 am Monday to 5:59 pm Friday versus 6:00 pm Friday to 5:59 am Monday. TAD was analyzed both as a categorical variable (TAD < 5 versus TAD >= 5) and as a continuous variable.

Number of Vehicles - Slightly over 30 percent of all the crashes were single vehicle. Drivers testing positive for alcohol were much more likely to be in single vehicle crashes (66.7%, p < .001), but drug use was not associated with this crash variable.

Day/Night - Almost 28 percent of all crashes occurred at night. Alcohol positive drivers were more likely to have nighttime crashes (62.7%, p < .001), but drug use was not significantly related to this variable.

Weekday/Weekend - Most crashes occurred during the week, with only 29 percent occurring on weekends. For drivers with neither alcohol nor drugs, 24.8 percent of crashes occurred on weekends. Those testing positive for alcohol were much more likely to have weekend crashes (45.1%, p < .001). Drug use was not related to day of week.

Vehicle Crush - Crashes were fairly evenly divided between those with TAD < 5 (48.8%) and those with TAD >= 5. There was a tendency for alcohol positive drivers to have more severe crashes (65.3% versus 46.3% for drivers with no alcohol or drugs, p = .0048). When TAD was analyzed as a continuous variable, the alcohol positive drivers had significantly more severe crashes (p < .001). Drug use was not associated with TAD.

Worst Injury in Vehicle - On the crash report, the officer indicates the level of injury sustained, using what is referred to as the KABCO scale. K indicates fatal injury; A, serious injury; B, moderate injury, C, minor injury, and O, no injury. Based on the proportion of K + A injuries, drivers were compared on the worst injury in the vehicle they were operating. Drivers testing positive for alcohol had a significantly higher proportion of K + A injuries in the vehicles they were operating (p < .001), but there were no significant relationships found for drugs.

Driver Variables

Information about the driver was compiled from the crash report, including whether the officer indicated that the driver had engaged in a hazardous action prior to the crash, whether the officer indicated that alcohol or drug use was suspected, and whether the driver was wearing a safety belt.

Hazardous Actions - Overall, more than two-thirds of the drivers had some kind of hazardous action indicated on the crash report. Alcohol positive drivers were significantly more like to be reported as having engaged in a hazardous action leading to the crash (96%), compared to 51.9 percent of crashes involving drivers testing negative for alcohol and drugs (p < .001). Drugs were not associated with this variable.

Alcohol Suspected - The investigating officers indicates on the crash report if alcohol and/or drugs are suspected. The judgment of alcohol involvement was almost always included, with only 4.6 percent of the cases having this variable not stated. Alcohol use was suspected in 19 percent of all the crashes, but the proportion varied according to the actual use of alcohol and drugs. Drivers who tested negative for both alcohol and drugs were rarely suspected of using alcohol (only 1% of the crashes). However, for drivers who tested positive for alcohol, officers correctly suspected alcohol 82 percent of the time (p < .001). There was also a trend for drug use to be associated with a higher probability of the officer suspecting alcohol (p = .0014). In addition, there was a tendency toward an alcohol/drug interaction in relation to the probability of the officer suspecting alcohol use, that is, the combination of alcohol plus drugs increased the likelihood of the officer suspecting alcohol (p = .0019).

Drugs Suspected - Officers indicated a suspicion of drug use in only four cases, less than one percent of the crashes. Two of these drivers tested negative for both alcohol and drugs, one tested positive for alcohol only, and one for alcohol plus drugs. In almost 19 percent of the cases the officer indicated that the driver was definitely not using drugs, but, in marked contrast to the alcohol judgment, in over 80 percent of the cases the officer made no judgment one way or the other. The proportion of cases in which the officer made a definite statement about non-use of drugs was almost the same for all four groups. No driver testing positive for drugs but not alcohol was suspected of drug use. It appears that, while alcohol is usually correctly detected, drugs are not readily identified by investigating officers.

Belt Use - Almost 70 percent of the drivers were reported as using seat belts, but belt use varied by alcohol and drug group. Of drivers testing positive for alcohol, 60 percent were not wearing belts (p < .001). Drug use was also significantly related to belt use, with 53.5 percent of drug positive drivers not wearing belts. There was also a trend toward an alcohol-drug interaction associated with lower belt use (p = .0038).

Injury Variables

Injury was considered on the basis of the KABCO scale from the crash report, the MAIS 85, the ISS 85, and whether the patient was admitted to hospital.

KABCO - Almost two-fifths (37.8%) of the drivers experienced serious or fatal injury (K + A) as judged by the investigating officer. Alcohol use was associated with a higher proportion of serious injury (55.3%, p < .001), but drug use was not.

MAIS 85 - The MAIS 85 was analyzed both as a categorical variables (MAIS 85 < 2 versus MAIS 85 >= 2) and as a continuous variable. In both instances there was a tendency for alcohol positive drivers to suffer more significant injury (MAIS 85 dichotomous, p = .0033; MAIS 85 continuous, p = .0017), but there were no significant relationships found for drugs.

ISS 85 - The ISS 85 was analyzed as a continuous variable. There was a tendency for both alcohol positive and drug positive drivers to be more seriously injured (alcohol, p = .017; drugs, p = .005).

Admission to Hospital - One-third of all the drivers were admitted to hospital, but this proportion varied greatly according to alcohol and drug group. Alcohol positive drivers were much more likely to be admitted (57.6%, p < .001). For drivers testing negative for both alcohol and drugs, only 26.9 percent were admitted. Drug use was not associated with alcohol admission.

Using logistic regression, the probability of hospital admission was further modeled. The effects of alcohol presence, drug presence, and history of alcohol or drug abuse or dependence were tested, taking into account other variables known to affect injury, namely, age and extent of vehicle crush (TAD). Gender was also included, because there is some evidence that females have a higher probability of injury in a given crash.

The overall model significantly predicted hospital admission (Chi square [8 df] = 121.19, p < .0001). The association of predicted probability of admission from this model and observed admission was 77.8 percent concordant, 21.9 percent discordant, and .3 percent tied. After accounting for TAD, sex, and age, alcohol presence significantly increased the probability of admission (Chi square [1 df] = 20.8, p < .0001). Drug presence and diagnostic history did not have significant effects. TAD, age, and alcohol were all significant independent predictors of probability of admission. After taking into account TAD, age, and alcohol, the presence of drugs had no effect.

DISCUSSION

Although there has been considerable speculation and even conjecture about the role of illicit drugs in traffic crashes, the studies in the literature raise more questions than they provide answers. Several studies have concluded that the role of such drugs is much greater than has heretofore been recognized. However, clinical studies have been based on patients admitted to major trauma centers, severely limiting the range of injury considered and providing no information on the extent to which drugs are present in less severe crashes. Furthermore, many studies have based drug determination upon urine screenings, so that it is not known when the drug was taken or whether the driver was under the influence of the drug at the time of the crash. In addition, many of the clinical studies ahve included only a relatively small proportion of the patients eligible for testing, and no information is provided on how the tested patients were selected or how they differ from those not tested. If testing is more likely to occur when drugs are suspected, the findings cannot be used to infer the extent of drug use in the larger patient population. Finally, the studies rarely relate the drug information to crash data, so that it is not known how drug crashes may differ from other crashes.

Several studies are based on better samples (Soderstrom, et al, 1989; Terhune, et al., 1992; Sweedler and Quinlan, 1989; Crouch et al, 1993), but still suffer from the restricted range of injury (either patients in a major trauma center or fatally injured drivers). Overall, the studies conclude that illicit drugs appear to be a major factor in motor vehicle crashes.

This study is based on a sample of patients presenting to two emergency departments for treatment of motor vehicle injury. The full spectrum of injury was included, with most patients treated and released. Only one-third of the patients were admitted to hospital. It was found that alcohol was the major drug involved in serious injury.

Based on blood analyses, illicit drugs were present in a relatively small proportion of the drivers (14.1%), and almost half of these also tested positive for alcohol. The average BAC for drivers testing positive for both alcohol and drugs was only slightly lower than that for drivers testing positive for alcohol only (0.148% versus 0.167%). Drivers testing positive for drugs but negative for alcohol comprised only 7.8 percent of the sample, and, on the whole, their crashes were much more similar to those of drivers testing negative for both alcohol and drugs.

It should be noted that the majority of drug positive drivers tested positive for marijuana. Cocaine and opiates were relatively rare, and the numbers were too small to conduct meaningful analyses of the different drug types. However, a cursory examination of the crash characteristics by drug type indicated that there were few differences, with the exception that cocaine positive drivers had more crashes involving speeding. Overall, the crashes of the drivers testing positive for drugs but not alcohol appear very similar to those of drivers testing negative for both alcohol and drugs. Their crashes were more likely to be daytime, weekday, multiple vehicle, with less severe TADs.

Drivers with a history of alcohol or drug abuse had a higher probability of testing positive for alcohol or drugs. Evidence of current alcohol or drug abuse increased this probability. However, a surprising proportion of drivers with a history or current evidence of such abuse tested negative for both alcohol and drugs.

While officers are good judges of alcohol use, they are not detecting drug use. The fact that the crashes of drug positive drivers are so similar to those of other drivers, with only the presence of alcohol making a difference, is probably a major reason for the difficulty seen for identifying drug use.

There are important limitations to this study. It is based on injuries sustained in motor vehicle crashes, and the findings cannot be generalized to other types of trauma. There is some evidence that intentional injury, particularly penetrating wounds, are more likely to involve drug use. However, even if that proves to be the case, it is important to understand whether, and to what extent, the drug effects per se contribute to the injury occurrence. The mere presence of the drug does not necessarily indicate that the drug was causal in producing the injury.

The study is also based on a limited catchment area in Southeast Michigan. While patients were drawn from a number of surrounding counties, it cannot be assumed that alcohol and drug use are representative of the rest of the nation.

While drug positive patients appeared to have crashes more similar to those of drivers with neither alcohol nor drugs, they were less likely to be using safety belts and possibly likely to suffer more severe injury (ISS 85, p = .005). However, when age, sex, TAD, and BAC were taken into account, drugs were not associated with a higher probability of hospital admission.

CONCLUSION

This study is based on a sample of drivers presenting to emergency departments for treatment of motor vehicle injury. Based on analyses of blood samples drawn within six hours of the crash, alcohol was found to be the major drug associated with injury. Marijuana, cocaine, and opiates were identified in slightly over 14 percent of the drivers, but almost half of these also had elevated blood alcohol levels. Those patients testing positive for drugs but not alcohol had crashes that were very similar to those of drivers testing negative for both alcohol and drugs.

While these findings do not confirm that drugs are a major factor in motor vehicle injury, the limitations of the study indicate the need for additional research based on the full spectrum of crashes and injury.

REFERENCES

Brookoff, D., Campbell, E.A., and Shaw, L.M. (1993) The underreporting of cocaine-related trauma: Drug Abuse Warning Network reports vs hospital toxicology tests. American Journal of Public Health, 83(3), 369-371.

Cherpitel, C.J. (1993) Alcohol and injuries: a review of international emergency room studies. Addiction, 88, 923-937.

Crouch, D.J., Birky, M.M., Gust, S.W., et al. (1993) The prevalence of drugs and alcohol in fatally injured truck drivers. Journal of Forensic Sciences, 38(6):1342-1353.

Kirby, J.M., Maull, L.I. and Fain, W. (1992) Comparability of alcohol and drug use in injured drivers. Southern Medical Journal, 85(8), 800-802.

Lindenbaum, G.A., Carroll, S.F., Daskal, I. et al. (1989) Patterns of alcohol and drug abuse in an urban trauma center: The increasing role of cocaine abuse. Journal of Trauma, 29(12), 1654-1658.

Rivara, F.P., Mueller, B.A., Fligner, C.L. et al. (1989) Drug use in trauma victims, Journal of Trauma,, 29(4), 462-470.

Soderstrom, C.A., Trifillis, A.L., Shankar, B.S. et al. (1988) Marijuana and alcohol use among 1023 trauma patients. Archives of Surgery, 123:733-737.

Stoduto, G., Vingilis, E., Kapur, B.M. et al. (1991) Alcohol and drugs in motor vehicle collision admissions to a regional trauma unit: Demographic, injury and crash characteristics. 35th Annual Proceedings, Association for the Advancement of Automotive Medicine, 235-247.

Sweedler, B.M. and Quinlan, K.E. (1989) Alcohol and drugs among fatally injured drivers of heavy trucks. Proceedings of the 11th International Conference on Alcohol, Drugs and Traffic Safety, 332-336.

Terhune, K.W., Ippolito, D.L., Hendricks, J.G. et al. (1992) The Incidence and Role of Drugs in Fatally Injured Drivers. Washington, DC:National Highway Traffic Safety Administration, Report No. DOT HS 808 065.

Williams, A.F., Peat, M.A., Crouch, D.J. et al. (1985) Drugs in fatally injured young male drivers. Public Health Reports, 100(1):19-25.