|
|
| |||||||||||||||||||||||||||||||||||||||
Background Reports suggest that people who drive while intoxicated by alcohol may do so repeatedly. We hypothesized that persons arrested for driving while impaired might be at increased risk for death in an alcohol-related motor vehicle crash. To evaluate this possibility, we studied the deaths of drivers in alcohol-related motor vehicle accidents in North Carolina over a 10-year period.
Methods We compared drivers who died in motor vehicle crashes from 1980 through 1989 and who had blood alcohol concentrations of at least 20 mg per deciliter (4.3 mmol per liter), referred to as the case drivers, with those who died in crashes but had blood alcohol concentrations below 20 mg per deciliter, referred to as the control drivers. We identified case drivers and control drivers through the state Medical Examiner System. We then searched North Carolina driver-history files for the five years before each death to identify arrests for driving while impaired.
Results We linked a total of 1646 case drivers and 1474 control drivers to their driver-history files. Case drivers were more likely than control drivers to have been arrested for driving while impaired (26 percent vs. 3 percent). After we controlled for potential confounding variables, case drivers 21 to 34 years of age were 4.3 times more likely (95 percent confidence interval, 2.7 to 6.8) than control drivers to have been arrested for driving while impaired; case drivers 35 years of age or older were 11.7 times more likely (95 percent confidence interval, 6.8 to 20.1). The strength of the association appeared to increase with the number of prior arrests. Case drivers were also more likely than the general population of currently licensed drivers to have been arrested.
Conclusions Arrests for driving while impaired substantially increase the risk of eventual death in an alcohol-related crash. Aggressive intervention in the cases of people arrested for driving while impaired may decrease the likelihood of a future fatal alcohol-related crash.
The risk of a fatal crash increases rapidly as the blood alcohol concentration of a driver increases. A driver with a blood alcohol concentration of 100 mg per deciliter (22 mmol per liter) or higher is 7 times more likely to be involved in a fatal motor vehicle crash than a driver who has not consumed alcoholic beverages, and a driver with a blood alcohol concentration of 150 mg per deciliter (33 mmol per liter) or more is about 25 times more likely4.
In response to the problem of drunk driving, a number of states have enacted and enforced strict legislation designed to deter people from driving while impaired by alcohol. As a result of these efforts, nearly 1.8 million persons in the United States (approximately 1 of every 90 licensed drivers) were arrested for driving while impaired in 19863.
Descriptive studies and reports suggest that drivers who die in alcohol-related motor vehicle crashes may be more likely than other drivers to have been arrested previously for drunk driving5,6,7,8,9,10,11. We were unable to find any studies that have evaluated this association, however. We therefore conducted a case-control study to address this question and to determine whether the association with death in an alcohol-related motor vehicle crash increases with the number of arrests for driving while impaired. We hope that the results of this study will encourage the establishment of programs and policies designed to reduce the risk of death in an alcohol-related crash among drivers with prior arrests for driving while impaired.
Methods
This case-control study was conducted in North Carolina (1991 population, 6,737,00012) and made use of information available through the North Carolina Medical Examiner System and the North Carolina Division of Motor Vehicles.
Definitions
The case drivers in this study were North Carolina residents 21 years of age or older who died within one hour of an alcohol-related motor vehicle crash on a public road in North Carolina from 1980 through 1989. A death was considered related to alcohol if the driver had a reportable blood alcohol concentration at the time of death, defined by the North Carolina Medical Examiner System as 20 mg per deciliter (4.3 mmol per liter) or higher. The study included only drivers who died while driving a passenger car or light truck (gross weight <6000 lb) (2720 kg).
The control drivers were North Carolina residents 21 years of age or older who died within one hour of a non-alcohol-related motor vehicle crash on a public road in North Carolina from 1980 through 1989. A death was considered not related to alcohol if the driver had a blood alcohol concentration below 20 mg per deciliter (4.3 mmol per liter) at the time of death. The group of control drivers was also restricted to those who died while driving a passenger car or light truck. Drivers who died in non-alcohol-related motor vehicle crashes were selected as controls because of the potential confounding effects of factors that could not be measured directly, such as risk-taking behavior. Furthermore, information on other potential confounders, including marital status and seat-belt use, was available only for crash deaths.
Data Collection
We selected case drivers and control drivers through the North Carolina Medical Examiner System, which investigates all deaths in motor vehicle crashes in North Carolina and maintains a complete data base on the results of these investigations, including the results of blood alcohol testing. The medical-examiner data base also includes selected information from traffic-accident reports for all deaths in motor vehicle crashes that occur on the state's public roads. The information collected on case drivers and control drivers from the computerized medical-examiner records included demographic information (such as age, sex, race, and marital status); information on the death (place, date, and time of death, date and time of injury, and results of toxicologic testing); and information about the crash (the identifying number of the traffic-accident report, number of vehicles involved in the crash, type of vehicle, position in the vehicle of the person who died, and whether seat belts were used).
The identifying numbers of traffic-accident reports were used to link the Medical Examiner System's records for case drivers and control drivers to the computerized reports on the fatal crashes. From these reports we collected information on the type and age of the driver's vehicle and his or her North Carolina driver's license number.
We used the driver's license number to link the traffic-accident report to the driver's North Carolina driver-history file. The North Carolina Division of Motor Vehicles maintains a computerized driver-history file on every licensed driver in North Carolina. Although demographic information in the driver-history file is limited to age, race, and sex, the information on traffic violations, including arrests for driving while impaired, is extensive. Individual driver-history files for case drivers and control drivers were searched for the five years before the date of death to identify arrests for drunk driving in which the driver had a blood alcohol concentration of 100 mg per deciliter or higher (the legal limit in North Carolina), had an equivalent concentration in breath, or refused alcohol testing.
Case drivers who died in 1989 were also compared with 2359 drivers selected randomly from the general population of licensed drivers in North Carolina in 1989 who were at least 21 years of age. To identify arrests for driving while impaired in the comparison group, we searched driver-history files from January 1, 1985, through December 31, 1989. The comparison allowed us to determine whether drivers who died in alcohol-related motor vehicle crashes were more likely than the general driving population to have been arrested for drunk driving.
Statistical Analysis
We calculated crude odds ratios and 95 percent confidence intervals with use of Epi-Info statistical software (version 5.01)13. Using unconditional logistic-regression analysis, we adjusted the odds ratios for potential confounding variables. Logistic-regression models were developed in accordance with a strategy recommended previously by Kleinbaum et al.14 Briefly, we identified potential confounders through a review of published studies. We identified potential interaction terms by means of stratified analysis and evaluated these in an initial logistic-regression model that included all potential confounding variables. Finally, to increase the precision of the estimates of odds ratios, we eliminated from the logistic-regression model potential confounders that did not cause an important change in stratum-specific adjusted odds ratios. SAS statistical software was used to perform the logistic-regression procedure15.
Results
Blood alcohol concentrations were available for 83 percent of all drivers who died in a motor vehicle crash from 1980 through 1989. We identified a total of 3410 drivers who died in motor vehicle crashes with blood alcohol concentrations of at least 20 mg per deciliter and 4089 drivers who died with blood alcohol concentrations below 20 mg per deciliter. A total of 1796 (53 percent) of these alcohol-related deaths met our case definition, and 1646 (40 percent) of the non-alcohol-related deaths met our control definition. Ultimately, 1646 (92 percent) of the case drivers and 1474 (90 percent) of the control drivers were successfully linked to their individual North Carolina driver-history files. A total of 1428 (87 percent) of the 1646 case drivers had blood alcohol concentrations of 100 mg per deciliter or higher at the time of death.
Case drivers were younger than control drivers and more likely to be male (Table 1). Similar percentages of case drivers and control drivers were white. Of those for whom information on marital status was available, fewer case drivers than control drivers were married. Crashes in which case drivers died were more likely than those involving control drivers to have occurred at night (from 6 p.m. to 6 a.m.) or on weekends (from 6 p.m. Friday to 6 a.m. Monday). Among drivers for whom information on seat-belt use was available, a smaller percentage of case drivers than of control drivers wore seat belts at the time of the crash. Case drivers were also more likely than control drivers to have died in single-vehicle crashes.
|
|
|
2 years vs. 2.5 for >2 years) (Table 4). By comparison, among drivers 35 years of age or older, the strength of the association between alcohol-related death and an arrest for drunk driving increased slightly with the length of time since the most recent arrest (adjusted odds ratio, 10.6 vs. 12.8).
|
Discussion
The results of this case-control study indicate that there is a strong association between arrests for driving while impaired and alcohol-related deaths of drivers. The strength of this association appeared to increase with the number of prior arrests. After adjustment for potential confounding factors, case drivers 21 to 34 years of age were approximately 4 times as likely as controls to have had one or more arrests for driving while impaired during the previous five years; case drivers 35 years of age or older were almost 12 times as likely as controls to have been arrested one or more times. Among younger drivers, the strength of this association decreased over time. Among older drivers, in contrast, the strength of this association increased slightly over time.
This study had a number of strengths, including the completeness with which information on alcohol levels was collected and reported to the North Carolina Medical Examiner System and the large proportion of case drivers and control drivers who were successfully matched to their driver-history files, thus decreasing the likelihood of selection bias. Furthermore, the short interval specified between the time of injury and the time of death (one hour or less) improved the validity of postmortem blood alcohol measurements and decreased the likelihood that alcohol-related deaths would be misclassified as non-alcohol-related. Finally, using arrests for drunk driving rather than convictions reduced the potential effect of legal interventions, including plea bargaining, on a driver's history and provided a more sensitive and consistent indicator of the frequency of incidents of driving while impaired.
The results of this study could have been due, in part, to differences between the control drivers and the general driving population. However, the comparison of case drivers who died in 1989 with a random sample of drivers from the general driving population also showed that drivers who died in alcohol-related motor vehicle crashes were more likely than others to have had prior arrests for driving while impaired. These results support the applicability of our findings to the general driving population.
We did not attempt to control for possible differences in the length of drivers' histories. However, we believe that any resulting misclassification of histories would not have differed between case drivers and control drivers. If there were important differences, case drivers would probably be more strongly affected than control drivers, since case drivers were probably licensed for shorter periods, and the true association between alcohol-related death and prior arrests for driving while impaired would be underestimated.
A number of studies have shown that over half the drivers arrested for driving while impaired are alcoholics16; by contrast, approximately 10 percent of American adults have problems with alcohol abuse or dependence during their lifetimes17. A Swedish study found that heavy drinkers were eight times more likely to die in motor vehicle crashes than nondrinkers and that many such deaths may have been related to alcohol use18. Although we had no information on the drinking histories of the drivers in our study, it is likely that the prevalence of alcoholism among those with a history of arrests for driving while impaired was similar to that reported in other studies. An increased prevalence of alcoholism among drivers with prior arrests for driving while impaired could help to explain the strong association between these arrests and alcohol-related deaths. In other words, the stigma of an arrest for drunk driving might be expected to have less effect on the drinking and driving behavior of an alcoholic than on that of a social drinker.
If we consider arrests for driving while impaired to be a possible marker for problem drinking,16 the results of this study suggest that a history of problem drinking may be more common among younger drivers. This hypothesis is consistent with other studies that have reported an increased prevalence of problem drinking among younger persons, particularly those less than 30 years of age19. However, our findings regarding the effect of the length of time since the most recent arrest for driving while impaired on the association between prior arrests and alcohol-related deaths suggest that many of the older case drivers may have been alcoholics. Indeed, alcoholism is most common among men 35 to 50 years of age20. Taken together, these findings suggest that the differences we observed in the strength of the associations between alcohol-related deaths and prior arrests for drunk driving among drivers of different ages may be related to the natural history of alcoholism.
In commenting on the lack of attention to the risk of alcoholism among drivers injured in crashes involving alcohol, Maull et al.21 and Colquitt et al.22 argue strongly for the evaluation and treatment of these patients for alcoholism. Given the prevalence of alcoholism among persons arrested for drunk driving in general and their increased risk of dying in alcohol-related crashes, physicians should also be aware of the importance of evaluating these persons for alcoholism and referring those who are alcoholics for treatment. Although additional research is needed on the effectiveness of treatment for alcoholism in decreasing subsequent arrests for drunk driving and preventing alcohol-related crashes, the substantial proportion of alcoholics among those arrested for driving while impaired explains why legal interventions alone have not been effective in controlling drunk driving16.
The likelihood that an alcohol-impaired driver will be arrested is between 1 in 250 and 1 in 200023,24. Therefore, strategies to combat drunk driving must reach beyond the drivers who have already been arrested25. Policies that are considered effective in reducing drinking and driving include laws that deal with the availability of alcohol in general (such as raising the legal drinking age to 21 years) and with drinking and driving in particular (establishing a legal level of intoxication)26. Nevertheless, our results suggest that effective intervention when drivers are arrested for driving while impaired could reduce the number of alcohol-related deaths.
On the basis of the results of this study, we recommend aggressive intervention in cases of drivers who have been arrested for driving while impaired with blood alcohol concentrations of 100 mg per deciliter or more. Since a substantial proportion of the persons arrested for driving while impaired may be alcoholics, we further recommend that legal sanctions against drunk driving be linked with programs to identify and treat alcoholic drivers. Such programs may be particularly effective for younger drivers, whose drinking behavior is likely to be more sensitive to environmental influences20. In addition, since the association between arrests for driving while impaired and deaths increases substantially with the number of arrests, it is important to intervene after the first arrest. Such an arrest may thus present an important opportunity to decrease the risk of death from a future alcohol-related crash and to facilitate the secondary prevention of alcoholism.
Supported by a grant (H28/CCH401640011) from the Centers for Disease Control and Prevention (to the Injury Control Section of the North Carolina Department of Environment, Health, and Natural Resources).
We are indebted to Jeffrey Sacks, James Wassell, Michael Bowling, and John Butts for technical advice; to Eric Rodgman for computer programming and statistical analysis; to John Horan, Joe Sniezek, and Gwen Ingraham for critical review of the manuscript; and to Dottie Ellis for assistance in the preparation of the study protocol. This article is dedicated to the memory of Robert D. Brewer, Jr., M.D.
Source Information
From the Epidemic Intelligence Service, Division of Field Epidemiology, Epidemiology Program Office, and the National Center for Injury Prevention and Control, Centers for Disease Control and Prevention, Atlanta (R.D.B.); the Environmental Epidemiology Section (P.D.M.), Injury Control Section (T.B.C., M.J.P.), and Division of Statistics and Information Services (S.W.), North Carolina Department of Environment, Health, and Natural Resources, Raleigh; and the University of North Carolina Highway Safety Research Center, Chapel Hill (C.P.).
Address reprint requests to Dr. Brewer at the Division of Unintentional Injury Prevention, National Center for Injury Prevention and Control, Centers for Disease Control and Prevention, 4770 Buford Hwy., N.E., Mailstop K-63, Atlanta, GA 30341.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Meningitis Due to Ceftriaxone-Resistant
Lonks J. R., Durkin M. R., Meyerhoff A. N., Medeiros A. A.
Extract |
Full Text
N Engl J Med 1995;
332:893-894, Mar 30, 1995.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2010 Massachusetts Medical Society. All rights reserved. |
Previous
