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

 


Fatal Traffic Accident Patterns and Driver Impairment in British Columbia

G. William Mercer*, Ph.D. and Wayne K. Jeffery*, MSc.

*Senior Policy Analyst, B.C. Police Commission, Ministry of Attorney General, 405 - 815 Hornby St., Vancouver, British Columbia, V6Z 2E6, Canada

**Section Head, Toxicology, Royal Canadian Mounted Police, "E" Division Headquarters, 5201 Heather St., Vancouver, British Columbia, V5Z 3L7, Canada

The views expressed in this paper are those of the writers and do not necessarily represent those of any other individuals or of any organizations. Detail on the analyses presented in this paper can be obtained by writing either author.

ABSTRACT

Blood samples, driver records and accident records of 41 female and 186 male fatally injured drivers were examined. Toxicologies showed: 37% alcohol-only; 11% alcohol-and-drugs; and 9% drugs-only. The most frequently found drugs were: 48% alcohol; 13% tetrahydrocannabinol or its metabolites (THC/THCCOOH); 4% cocaine; and 5% diazepam.

To investigate the relationship between accident patterns and alcohol and drug use, a factor analysis of accident and driver records produced a 7-factor varimax solution accounting for 63% of the matrix variance. The "Male Single Vehicle" factor related positively to alcohol, the "Young Recidivists" factor related to both alcohol and THC/ THCCOOH, and the "Weather" factor related to CNS depressants. Accidents should be examined using multivariate techniques to develop more effective accident classification systems.

INTRODUCTION

One aspect of the development of the epistemology of alcohol impairment in traffic accidents is that there is an a priori assumption that if alcohol is present it played a role in the accident. A second aspect is simply reductionist: The reasoning is that the weather was good, the roads were good, there were no other vehicles involved, the driver was young and healthy, and if alcohol was found, then by process of elimination, the accident is classified as "alcohol impaired".

In order to avoid any a priori or reductionist bias, the present study examined the question of impairment and traffic accidents by first defining accident configurations without reference to attributed cause or toxicological finding, and only then examined the relationship between the accident configuration, alcohol impairment as an attributed cause, and driver toxicologies that included drugs other than alcohol.

METHOD

Police traffic accident reports, drivers' previous driving records, and full toxicological analyses of drivers' blood were examined for the 227 drivers (82% male) who died within 24 hours of a traffic accident in British Columbia between October 1, 1990 and September 31, 1991.

To determine underlying patterns of accidents, a varimax factor analysis was performed on a set of 20 variables comprised of: the eight accident environment variables; two traffic law variables; three vehicle types; number of vehicles in the accident; vehicle damage severity; number of passengers in the vehicle; and driver variables of gender, age, previous accident frequency; and two composite variables - previous "reckless driving" convictions and previous "drinking driving" convictions. These latter two measures were of drivers' previous 5-year driving conviction history and were created using factor score loadings derived from an oblique factor analysis of the nine most frequent violations drawn from a random sample of 10,000 driving records: the "reckless driving" factor loaded on exceeding the speed limit, speeding in school zone; disobeying a red light, disobeying an amber light, disobeying a stop sign, and disobeying a sign or signal; and the "drinking driving" factor loaded on impaired driving, a 24-hour suspension for impaired driving, speeding, and excessive speed for conditions (Mercer, 1987).

These 20 variables reduced to a seven factor solution accounting for 63% of the matrix variance before rotation. Criteria for the number of factors was a scree (discontinuity) test and that each factor must have an eigenvalue greater than 1 (see, for example, Rummel, 1970). Interpretation was based on loadings at +/-.30. Complete estimation factor scores were produced for the 7 factors and each accident was then classified in terms of its highest factor score.

RESULTS

Toxicologies showed: 37% alcohol-only; 11% alcohol-and-drugs; and 9% drugs-only. The most frequently found drugs were: 48% alcohol; 13% tetrahydrocannabinol or its metabolites (THC/THCCOOH); 4% cocaine; and 5% diazepam. Due to the low frequencies of drug use found in the toxicological analyses, a number of drug - "dummy" related variables were created by grouping drugs into present/absent categories: central nervous system stimulants; central nervous system depressants; tetrahydrocannabinol (THC) and/or tetrahydrocannabinol metabolite (THCCOOH); and alcohol.

From Table 1, the factors were interpreted as: Factor 1 - "Motorcycle" accident; Factor 2 - "Intersection"; Factor 3 - "Young Recidivists" (it is noteworthy that only personal characteristics, as opposed to accident environment or vehicle characteristics, loaded on this factor); Factor 4 - "Weather"; Factor 5 - "Male Single Vehicle"; Factor 6 - "Rural/Commercial Vehicle"; and Factor 7 - "Party Car".

Table 1
Factor Loadings

  VARIMAX FACTOR LOADINGS
VARIABLES F1: MTCL F2: INTERS F3: RECIDV F4: WTHR F5: MSV F6: RU/COM F7: PARTY
Env.: Area Density .05 .55 .10 -.07 .01 -.47 -.03
Env.: Road Curves .11 -.28 -.11 .02 .14 .32 .45
Env.: Intersection .03 .83 -.04 -.01 -.09 -.07 -.04
Env: Traf. Control Dev. -.02 .88 -.04 -.07 -.05 -.00 -.00
Env.: Zoned 50+km -.16 -.37 .04 -.00 -.17 .64 -.10
Env.: Road Traction Bad -.05 -.09 -.08 .88 -.02 -.01 .04
Env.: Weather Bad -.07 -.04 -.15 .84 -.08 -.02 .08
Env.: Ambient Darkness -.04 -.32 -.06 -.32 .30 -.17 .36
Vehicle: Motorcycle .86 .02 .14 -.08 -.15 -.17 -.02
Vehicle: Passenger -.90 -.08 -.01 .09 -.12 -.27 .01
Vehicle: Commercial .29 .12 -.19 -.06 .33 .67 .03
Vehicle: Damage Severe -.49 .06 .16 .00 .20 .43 .29
Number Passengers .20 .13 -.02 .18 -.09 -.00 .68
Number Vehicles -.10 .10 .11 .30 -.58 -.01 -.13
Driver: Gender (Male+) .29 -.07 .20 .06 .53 .03 .03
Driver: Age -.24 .17 -.39 .22 -.04 .24 -.48
Driver: Wear Seat Belt .26 .01 -.08 -.01 -.75 -.01 .11
Driver: Reckless History .06 .04 .80 -.02 .03 -.07 .00
Driver: DWI History -.11 .04 .73 -.00 .32 .01 -.11
Driver: Accident History .08 -.06 .63 -.19 -.14 -.01 .06

Table 2 shows a positive relationship between alcohol and drug presence and the accident factors: The Young Recidivists factor related to the presence of alcohol and THC/THCCOOH; a positive relationship between the Weather factor and the presence of CNS depressants; and a positive relationship between the Male Single Vehicle factor and the presence of alcohol. Table 3 shows that "Alcohol Impaired" as a police-reported cause was associated with accidents loading on the Young Recidivists factor and the "Male Single Vehicle factor".

Table 2
Drug Presence by Factors

DRUG: FACTOR: Young Recidivists
ALCOHOL not highest score highest factor score
No N=116 57% 26%
Yes N=108 43% 74%
chi-square = 11.29 p.=.001
THC/COOH    
No N=197 90% 77%
Yes N=37 10% 23%
chi-square = 4.57 p.=.033
  FACTOR: WEATHER
CNS DEPRSNT not highest score highest factor score
No N=209 95% 84%
Yes N=15 5% 16%
Chi-square = 5.12 p.=.024
  FACTOR: Male Single Vehicle
ALCOHOL not highest score highest factor score
No N=116 58% 24%
Yes N=108 42% 76%
chi-square = 14.47 p.=.000

Table 3
Factors by "Alcohol Impaired" as Reported Cause

CAUSE: FACTOR: Young Recidivists
ALCOHOL not highest score highest factor score
No N=155 72% 54%
Yes N=69 28% 46%
chi-square = 12.6 p.=.000
  FACTOR: Male Single Vehicle
ALCOHOL not highest score highest factor score
No N=155 73% 53%
Yes N=69 27% 47%
chi-square = 5.89 p.=.015

DISCUSSION

The findings of these toxicologies fall within the general ranges found by other researchers (e.g. Caplan et al., 1990; Donelson, 1987, Root, 1990). Also, while this relatively small sample of accidents can result in an unstable factor structure, some confidence in these analyses comes from the observation that factors 1, 2, 3, 4, 5 and 7 closely parallel factors identified in a factor analyses of over 19,000 casualty traffic accident reports and driver records (Mercer, 1988).

In terms of the classic alcohol-impaired traffic accident, the Male Single Vehicle factor comes closest: single vehicle, darkness, no seat belt, and the male driver having a previous history of impaired driving charges. It is not surprising that this factor was associated with alcohol being found in the toxicology, and the police citing "alcohol" as a cause. The Young Recidivists factor was associated with the presence of alcohol and with the presence of THC/THCCOOH, as well as with "alcohol" as a police-reported causes. However, except that the young driver had a generally poor record with reckless driving and impaired driving types of charges and a history of previous accidents, there were no demographic, environmental or vehicular variables loading on the factor. One interpretation of this pattern is that this is the sort of irresponsible, alcohol and drug using driver with an "attitude" that will have an accident wherever he/she happens to be.

The Weather factor was associated with the presence of CNS depressants and their association with this factor leads to speculation that they may have had an impairing role in the accidents.

Overall, these analyses suggest that drugs other than alcohol are contributing to fatal traffic accidents in British Columbia. Ongoing data collection is needed to monitor the levels and role of drugs other than alcohol in traffic accidents. More broadly, in order to evaluate accident frequencies and trends, a mutually exclusive classification system needs to be developed. In turn, accident countermeasures could be better developed and evaluated by targeting these identifiable, multivariate accident configurations.

REFERENCES

Caplan, Y.H., Levine, B.S., and Goldberger, B.A. Drugs in driver fatalities: a preliminary study in the state of Maryland. In Perrine, M.W. (Ed.) Alcohol, Drugs and Traffic Safety - T89. pp 824-828. Proceedings of the International Conference on Alcohol, Drugs and Traffic Safety, National Safety Council, Chicago, 1990.

Donelson, A.C. Cannabis and alcohol use among drivers and pedestrians fatally injured in traffic crashes. In Noordzij, P.C. and Roszbach, R. (Eds.) Alcohol, Drugs and Traffic Safety - T86. pp 271-274, Elsevier Science Publishers, Amsterdam, 1987.

Mercer, G.W. Frequency, Types and Patterns of Traffic Convictions and Frequency and Types of Traffic Accidents. In CounterAttack Traffic Research Papers, 1986. Vancouver, British Columbia, Ministry of Attorney General, 1987.

Mercer, G.W. Toward an empirically-derived classification of casualty traffic accident patterns. In CounterAttack Traffic Research Papers, 1987. Vancouver, British Columbia, Ministry of Attorney General, 1988.

Root, I. Drugs of abuse and fatal automobile accidents. In Perrine, M.W. (Ed.) Alcohol, Drugs and Traffic Safety - T89. pp 602-607, Proceedings of the International Conference on Alcohol, Drugs and Traffic Safety, National Safety Council, Chicago, 1990.

Rummel, R.J. Applied Factor Analysis. Evanston: Northwestern University Press, 1970.