Two articles on the effect of recreational marijuana commercialization on traffic fatalities reached opposite conclusions in the June 22nd 2020 issue of JAMA Internal Medicine. Readers should understand the effect of study design differences before concluding that marijuana can be commercialized without an increase in traffic fatalities.
Kamer (doi:10.1001/jamainternmed.2020.1769) reported that traffic fatalities increased in both Colorado and Washington after commercializing recreational marijuana. But Santaella-Tenorio (doi:10.1001/jamainternmed.2020.1757) found traffic fatality increased in Colorado but not in Washington during a similar time period.
Kamer used an appropriate single relevant control group for both states. Santaella-Tenorio used a different synthetic control group for each state. Washington’s synthetic control group had an increase in traffic fatalities similar to that state, so the authors found no traffic fatality difference between Washington and its synthetic control group.
Santaella-Tenorio proposed three reasons for the different outcomes between Washington and Colorado, none of which are convincing. The authors neglected to point out the largest reason for the different outcomes – different control groups. The Washington control group was heavily weighted (83.9%) toward only two states – California and New Jersey. California enacted legal marijuana during the study period and New Jersey’s medical marijuana patient pool doubled during the study period, so it’s not surprising to see traffic fatality rate increase in those states equal to that found in Washington.
Kamer’s conclusion is sobering: if all states were to commercialize marijuana and have the same effect on their traffic fatalities as Colorado and Washington, the nation could experience an additional 6,800 traffic deaths annually. Santaella-Tenorio’s results should be ignored, even though he found 75 “excess traffic deaths” in Colorado after marijuana commercialization.