Half of young drivers who died in car crashes in nine United States were under the influence of either alcohol or marijuana, or both, according to a recent study.

Researchers at Columbia University's Mailman School of Public Health found that policy changes could influence substance use among adolescents and young adults.

For the study, researcher Katherine M. Keyes and colleagues analyzed more than 7,000 fatal accidents involving drivers between the ages of 16 and 25 from 1999 to 2011 who died within one hour of the crash in California, Connecticut, Hawaii, Illinois, New Hampshire, New Jersey, Rhode Island, Washington State and West Virginia. These nine states all routinely perform toxicological tests on the blood or urine specimens of drivers who die in car crashes. Information was drawn from the Fatality Analysis Reporting System, a census of fatal traffic crashes occurring within the U.S. More than half of the crashes (54.7 percent) for the period between 1999 to 2011 occurred in California.

Overall, 50.3 percent of the deceased tested positive for alcohol, marijuana or both. Of these, 36.8 percent were under the influence of alcohol, 5.9 percent used only marijuana and 7.6 percent used both substances.

The researchers further tested whether there were any changes in patterns of alcohol and marijuana use among those aged 21 years and older who were legally allowed to consume alcohol, versus those less than 21. It was found that alcohol consumption indeed increased by 14 percent, but such prominent changes in marijuana use were not seen. Dr. Keyes does note that marijuana use decreased among those aged 21 years and older who used this drug alone. After reaching age 21, use of alcohol in combination of marijuana increased slightly.

"Policies related to the use of substances in the United States remain in flux; the rapid changes in marijuana use policy are a good example of this," said Keyes, who led the study. "It's imperative to know whether there will be unintended consequences of changes in policies, including increases or decreases in harm related to other substances that are not the focus of the policy."

The findings are detailed in the journal Injury Epidemiology.