Modeling the impact of a state alcohol and traffic safety laws on motor vehicle fatalities using IV approaches, 1980-2009

Background: Approximately 1/3 of the nearly 29,000 motor vehicle fatalities in the U.S. were alcohol related in 2009. Policies adopted to reduce alcohol-impaired driving and improve traffic safety vary widely across states. Researchers seeking to understand the impact of different policy approaches to reducing fatalities have faced two important measurement issues: how to capture differences in the state policy environments over time and how to address endogenity issues in controlling for state differences in alcohol consumption. Methods: Using a unique dataset, we construct scores for each state for each year of the percent of 24 possible alcohol and traffic laws present in the state, and use an instrumental variable (IV) approach to include alcohol consumption in negative binomial regression models with state and year fixed effects. Results: Controlling for other state population factors, we find ranking in the highest quartile of state score in any given year is associated with an 11% lower rate of alcohol- related MVC fatalities (IRR= .89, CI=.84,.95)than those in the lowest quartile. Further, alcohol consumption per capita in a state, as measured by our IV, is associated with a more than twice the rate of alcohol-related MVCs in any given year (IRR=2.07, CI=1.66, 2.60), controlling for state population characteristics and differences in policies adopted. Conclusion: These findings signal the importance of measuring the comprehensiveness of state policy adoption across states, and the usefulness of IV approaches. State policymakers should focus on addressing gaps in state policies and developing new approaches to decrease alcohol consumption.