3416 EVALUATING THE EFFECTIVENESS OF REMOTE ALCOHOL MONITOR INTERVENTIONS IN REDUCING MOTOR VEHICLE CRASHES INVOLVING DRUNK DRIVING IN TRAFFIC SYSTEMS IN THE UNITED STATES, TEXAS, AND CALIFORNIA Amber D. Elkins | elkins@tamu.edu
Texas A&M University: College of Veterinary Medicine & Biomedical Sciences, Department of Veterinary Pathobiology, 4467 TAMU, College Station, TX, USA 77843 ORCID |
http://orcid.org/0000-0002-2240-6656 Texas A&M University: Dwight Look College of Engineering, Department of Industrial & Systems Engineering, 3131 TAMU, College Station, TX, USA 77843
G. Wade McDonald | gwm762@mail.usask.ca
University of Saskatchewan: College of Graduate and Postdoctoral Studies, Department of Computer Science, 110
Science Place, Saskatoon, SK, Canada S7N 5C9 ORCID
| http://orcid.org/0000-0003-4763-3414 Dennis M. Gorman | gorman@sph.tamhsc.edu
Texas A&M University: School of Public Health, Department of Epidemiology & Biostatistics, 1266 TAMU, College Station, TX, USA 77843 ORCID
| http://orcid.org/0000-0001-9833-8744 Eva M. Shipp | e-shipp@tti.tamu.edu
Texas Transportation Institute, Center for Transportation Safety, 1266 TAMU, College Station, TX, USA 77843 ORCID |
http://orcid.org/0000-0002-4034-8031 Robert C. Wunderlich | r-wunderlich@tti.tamu.edu
Texas Transportation Institute, Center for Transportation Safety, 1266 TAMU, College Station, TX, USA 77843
Mark A. Lawley | malawley@tamu.edu
Texas A&M University: Dwight Look College of Engineering, Department of Industrial & Systems Engineering, 3131 TAMU, College Station, TX, USA 77843 ORCID |
http://orcid.org/0000-0003-3925-2806 Each year, motor-vehicle crashes (MVCs) are responsible for more than 1.2 million fatalities and over 50 million injuries worldwide and are the leading cause of accidental deaths in the US. Despite the large contribution of alcohol-use to MVCs, little is known about how to maximize the effectiveness of interventions alone or in combination. This can result in wasted money and other resources and, worse, in lives lost. Our long-term goal is to estimate the cost and the effectiveness of policy-based interventions aimed to reduce MVC fatalities and injuries related to driving under the influence of alcohol (DUI), to address the gaps in our understanding of the mechanisms and processes through which these interventions reduce DUI-related MVC injuries and fatalities, and to create an optimal portfolio of interventions for preventing DUI-related accidents for use by state and local governments and other relevant stakeholders. One intervention of interest for reducing drunk driving offenses being proposed for use by states involves using remote alcohol monitoring devices (RAMs), such as an ankle bracelet that conducts transdermal readings by sampling alcohol vapor just above the skin for insensible perspiration and provides continuous sobriety monitoring of the person wearing the device (24/7 monitoring).
Analysis. Funded through the Center for Transportation Safety at the Texas A&M Transportation Institute, we developed a System Dynamics (SD) array model to simulate the dynamics of DUI-related MVC injuries and fatalities in traffic systems in the United States, Texas, and California with the objective of identifying the causal mechanisms, potential leverage points, and effects of policy-based interventions mandating RAMs based on the previous studies:
- RAM device effects on first-time and repeat offenders,
- RAM offenders recidivated at higher rates (but not significantly),
- RAM offenders who did recidivate, did so at a later time, and
- Risk of a driver causing a DUI-related MVC based on prior offenses (“the offender lethality assumptions”).
The model examined the effects of these studies alone, in combination, and in various calibrations in each location and their implications for DUI-related morbidity and mortality and drunk driving rates. We verified and validated for the ability to reproduce historic trends using secondary data analysis, estimates, and relevant literature to calibrate/parameterize variables. The model started 2010-07-01 and ended 2031-07-01, using a daily time step and Census interval distribution years.
Different offender lethality assumptions were used to calibrate models testing intervention effects per location, running each assumptions set 250 times, for 1000 total realizations and then compared by total DUI deaths between scenarios. Sensitivity analysis involved applying normal distributions to calibrated variables with a mean base value of the variable and standard deviation of 25% of the base value (Monte Carlo) and performed 1000 model runs with each of the offender lethality assumptions.
Results. Using the Shapiro-Wilk test for normality, the results were not normally distributed (W = 0.98713, p-value = 0.02428). Using the Wilcoxon signed-rank test to compare scenarios, the median difference between total DUI deaths with vs. without intervention was not significantly greater than zero in any scenario examined. Model results suggest RAM devices did not significantly decrease drunk driving mortality in any geographic location or scenarios.
Keywords: public health; system dynamics; traffic; transportation; alcohol; policy
