Crashes have increased by up to 6 percent in Colorado, Nevada, Oregon and Washington – all of which have legalized recreational marijuana – compared with neighboring states, new research from the Insurance Institute for Highway Safety and Highway Loss Data Institute shows.

But officials caution that it’s difficult to directly link crashes to marijuana use because testing is infrequent and inconsistent – and the amount of marijuana in a person’s system doesn’t necessarily relate to impairment.

“Despite the difficulty of isolating the specific effects of marijuana impairment on crash risk, the evidence is growing that legalizing its use increases crashes,” Insurance Institute for Highway Safety and Highway Loss Data Institute President David Harkey said in a news release.

The frequency of claims per insured vehicle year rose a combined 6 percent after the start of retail sales of recreational marijuana in Colorado, Nevada, Oregon and Washington, compared with four control states, Idaho, Montana, Utah and Wyoming, the release says.

Retail sales of recreational marijuana began in Colorado in January 2014 and in Washington in July 2014. Sales started in Oregon in October 2015 and in Nevada in July 2017.

The analysis is based on collision loss data from January 2012 through October 2017, the release says. “Analysts controlled for differences in the rated driver population, insured vehicle fleet, the mix of urban versus rural exposure, unemployment, weather and seasonality.

A separate study examined police-reported crashes from 2012 to 2016 before and after retail sales began in Colorado, Oregon and Washington, the release says. Estimates show that compared with neighboring states that didn’t legalize marijuana sales, the three states combined saw a 5.2 percent increase in the rate of crashes per million vehicle registrations.

The Insurance Institute for Highway Safety and Highway Loss Data Institute researchers compared Colorado with Nebraska, Wyoming and Utah, and they compared Oregon and Washington with Idaho and Montana, the release says. The study controlled for differences in demographics, unemployment and weather.

“The size of the effect varied by state,” the release says. “Although the study controlled for several differences among the states, the models can’t capture every single difference. For example, marijuana laws in Colorado, Oregon and Washington differ in terms of daily purchase limits, sales taxes and available options for home growers. These differences can influence how often consumers buy marijuana, where they buy it and where they consume it.”

But the 5.2 percent increase in police-reported crash rates after legalization of recreational marijuana use is consistent with the 6 percent increase in insurance claim rates estimated by HLDI.

The number of highway deaths involving Colorado drivers who had marijuana in their system grew again in 2017, but at the same time, traffic fatalities in which drivers had enough marijuana in their bloodstream to be deemed legally impaired dropped sharply, from 52 in 2016 to 35 last year, a state study shows.

The reason for this seeming contradiction: Marijuana can remain in the bloodstream for weeks, so a positive blood test may not mean a driver was stoned at the time of a deadly crash.

Gov. John Hickenlooper told The Gazette’s editorial board this month that “one of the big issues” facing the next governor will be traffic fatalities.

Last year, 648 people died in crashes across Colorado, the highest total since 2004, the Colorado Department of Transportation reported.

Hickenlooper emphasized the importance of gathering accurate data despite the expense involved.

“We need to measure whether people have marijuana in their systems,” he said.

The blood test for marijuana use costs $300 and will show someone has used marijuana in the past 30 days, whereas the test for drunken driving costs $30, Hickenlooper said.

When someone is suspected of driving drunk and having used marijuana, often authorities will only test for impairment from alcohol because its cheaper and sufficient for a DUI charge.

State officials don’t have the authority to tell local municipalities how to test drivers, Hickenlooper said.

“The state offered to pay the local municipalities – they are the ones who more often than not investigate highway fatalities, it’s not the state troopers that decide who gets a blood test,” he said. “They say, ‘We don’t care.'”