It has been well established that the alcohol ignition interlock device reduces the rate of repeat DUI while installed on vehicles. The interlock device also records and stores all driver BAC test outcomes. In a recent analysis based on 5.5 million breath tests provided by over 2000 interlock assigned DUI offenders in Alberta, it was found that the rate of fail-level BAC tests (BAC>.04%) accumulated during the first five months of interlock use, strongly predicts the likelihood of reconviction for DUI during the 12-24 months after the interlock was removed. The predictive power of this record, based as it is on actual drinking-driving behavior, was a stronger predictor than prior DUI offenses, widely regarded as the best advance indicator of future DUI. This proposal seeks support for a secondary data analysis study of the interlock records from the largest interlock programs in Canada (Quebec) and the United States (Texas). Based on approximately 20,000 interlock users who can be expected to have collectively provided over 40 million breath tests (with an estimated 50,000 BAC positive tests), this will be the most comprehensive evaluation of interlock user DUI behavior yet. The time base for each offender in each jurisdiction will be divided into three phases, before the interlock, during the interlock and after the interlock. Pre-interlock variables in the driver record, such as moving violations, prior DUI, arrest BAC, etc., will be studied as predictors/covariates. During the period of interlock monitored driving, the patterns and frequency of positive BAC tests will be compared to these other predictors to identify the best advance predictors of repeat DUI after interlock removal. The rate of BAC positive interlock tests will be adjusted for representation as a proportion of vehicle use. The post-interlock period is the best time to evaluate the impact since it is well-known that the rate of repeat offense increase rapidly after the interlock control is removed. These data will be evaluated by segmentation analyses (Exhaustive CHAID and CART) as a preliminary search strategy to identify unusual combinations of variables, and by survival analyses (Kaplan-Meier and Cox Regression) to evaluate the time to repeat. The findings will have substantial practical relevance for public policy.
| Marques, Paul R; Tippetts, A Scott; Voas, Robert B (2003) The alcohol interlock: an underutilized resource for predicting and controlling drunk drivers. Traffic Inj Prev 4 Suppl 1:5-11 |
| Marques, Paul R; Tippetts, A Scott; Voas, Robert B (2003) Comparative and joint prediction of DUI recidivism from alcohol ignition interlock and driver records. J Stud Alcohol 64:83-92 |
| Marques, Paul R; Tippetts, A Scott; Voas, Robert B (2003) The alcohol interlock: an underutilized resource for predicting and controlling drunk drivers. Traffic Inj Prev 4:188-94 |
| Marques, Paul R; Voas, Robert B; Tippetts, A Scott (2003) Behavioral measures of drinking: patterns from the Alcohol Interlock Record. Addiction 98 Suppl 2:13-9 |
