The Contribution of Alcohol and Other Drugs Among Fatally Injured Drivers in Quebec: Some Preliminary Results

This study presents some preliminary results regarding the contribution of alcohol and other drugs in fatal crashes in Quebec. The data comes out of two sources. First, coroner, forensic laboratory and police accident records were matched for 482 fatally injured drivers of passenger vehicles deceased between April 1999 and November 2001. Among those 482 fatally injured drivers, both blood and urine samples were obtained in 354 cases (73.4%). Second, two roadside surveys were conducted in August 1999 and 2000. Representative of the Quebec driving population, the survey sample was distributed proportionately to the number of fatal crashes per time of day (eight 3-hour periods) and day of the week (seven days). During both daytime and nighttime, a total of 11,952 drivers participated in the two roadside surveys among which 11,574 provided a breath sample (96.8%), 8,177 a saliva sample (84.6% when requested: 8,177/9,671) and 5,931 a urine sample (49.6%). The data collected allowed two different analyses: case-control (alcohol : blood/breath, other drugs : urine/urine) and responsibility (case-case approach) that compares drug cases to drug-free cases. Drugs under scrutiny included alcohol, cannabis, cocaine, benzodiazepines, opiates, barbiturates, amphetamines and PCP. Results of case-control analyses show the following odds ratios for each drug alone [with 95% confidence intervals]: Alcohol: 51-80 mg%: 3.7 [1.6-8.3], > 80mg%: 39.2 [25.560.2]; Cannabis: 2.2 [1.5-3.4]; Cocaine : 4.9 [1.4-17.4]; Benzodiazepines : 2.5 [1.4-4.3]. Other drugs (opiates, barbiturates, amphetamines, PCP) were detected less frequently but significantly increased risks were calculated for all cases (regardless of the presence of another drug) for amphetamines and PCP. For all drugs including alcohol, polyusage is systematically associated with an elevated risk. Responsibility analyses corroborate those results although odds ratios are always less high and sometimes not statistically significant, which could be explained by the limitations inherent to that methodology (lack of statistical power).