Fatal car accidents involving people who had recently used pot doubled in Washington State after the drug was legalized almost four years ago, according to a report from the AAA Foundation for Traffic Safety released Tuesday.

Cannabis-involved fatalities increased from 8 percent in 2013 to 17 percent in 2014, according to AAA, which used data from Washington State’s Traffic Safety Commission’s Fatality Analysis Reporting system to track marijuana use and traffic fatalities from January 1, 2010 to December 31, 2014.

It became legal to use marijuana in Washington on December 6, 2012.

AAA combined Washington’s traffic fatality data with reports from the state’s toxicologist, which examined the presence and concentration of marijuana’s main psychoactive ingredient, THC, in the blood of drivers who had died.

About one-third of drivers who died did not have alcohol or other drugs in their blood at the time of the fatal crash — just THC, according to the study. Another third had detectable alcohol in addition to THC and the remainder of the THC-positive drivers had other drugs and/or alcohol in addition to THC.

The AAA study did not say that drivers with detectable THC at the time of the crash were impaired by the THC, nor did it analyze whether the driver that tested positive for THC was at fault in the crash.

“Trying to really pull apart how much is due to actually the cannabis … still requires additional research,” said Dr. Tom Marcotte, co-director of the Center for Medicinal Cannabis Research in San Diego. “I think a number of agencies would say we really need a few years to determine whether you’re seeing a real epidemiological effect here.”

Marcotte said marijuana use can impact drivers’ reaction times and ability to accurately estimate time. It can also result in a reduced ability to multitask. A study in the Netherlands found that drivers with THC in their systems had a more difficult time maintaining their lane position.

Drivers who are suspected of marijuana impairment are typically assessed by law enforcement, who rely on behavioral and physiological measures, as well as blood tests.

Washington is one of several states that have so-called per se limits for marijuana-impaired driving that set specific thresholds for the amount of active THC a driver can have in his system based on a blood test. THC is measured in nanograms per milliliter of blood — a system that’s similar to measuring blood alcohol content in drunk drivers.

But blood tests for THC may not be a reliable indicator of impairment.

“People can have a low level of THC in their system but still be suffering the impaired effects of cannabis,” Marcotte said. “On the other hand people can also have low levels and not be showing those effects.”

Often, blood tests require a warrant and transport of the impaired driver to a testing facility, which takes time and gives THC the chance to wear off. AAA is one of several groups calling for a two-step process to determine marijuana impairment in drivers.

AAA would like drivers to be tested for recent marijuana use. It would also like behavioral and physiological evidence to prove driver impairment from marijuana use.