An analysis of the labs that test legal marijuana in Washington found that some appear to be more friendly to industry than others in their results, leading to calls for stricter state oversight.

Share story

Some state-certified marijuana labs testing for microbes such as E.coli and mold appear more friendly to pot merchants than others, according to an analysis by a Woodinville data scientist.

Four labs rejected none of the pot they tested over a three-month period last year, according to the analysis by Jim MacRae. Four other labs failed more than 12 percent of samples tested over the same time, with two labs rejecting 44 percent of samples for microbes.

“It’s almost impossible for that to happen,” said David Lampach, co-founder of Steep Hill Labs in Tukwila, of the disparity.

After sifting through public records for moisture, microbial, residual solvent and potency tests required by the state, MacRae sorted the lab results into three categories. He found three labs scored as “friendly” to the industry, six landed in the “kinda friendly” group, and five others appeared more neutral. (There were 14 certified labs, but two have gone out of business.)

This week, save 75% on select subscriptions.

He has not named the labs in the analysis published on his blog. Instead he identifies them by letter-code. Before naming them he is seeking industry feedback. MacRae, who has applied for pot-store licenses, argues that accurate lab tests are crucial to legalization because safety and quality assurance is a chief advantage legal pot merchants have over illegal dealers.

MacRae’s findings are being applauded by some lab officials and others who believe stricter state oversight is overdue. “That was a great tool he put out,” said Dani Luce, CEO of GOAT Labs in Vancouver. “But it should have gone to the Liquor and Cannabis Board, not the public.”

The Liquor and Cannabis Board (LCB) knows about MacRae’s analysis, said agency spokesman Brian Smith. MacRae is testifying about his findings at an agency board meeting Wednesday in Olympia.

“We are aware that some labs have higher rates of approvals than others. We are looking into it and don’t want to go into what our intentions are at this point,” Smith said.

Lampach believes the LCB is about to tighten lab oversight, something agency officials said would happen last year. A clampdown couldn’t come too soon, he said. Steep Hill, which also operates labs in California and Colorado, is considering pulling out of Washington, he said, because it can’t compete with labs that rubber-stamp results.

“It’s entirely right to look at labs because they’re incentivized to provide favorable results,” said John Davis, CEO of medical-marijuana dispensaries in Seattle and an applicant for retail pot-shop licenses. Accurate tests become more important as the state aims to bring medical patients into its retail stores this year. (Washington does not require pesticide testing.)

Labs shouldn’t be blamed for all suspicious results, some say. Labs can only test the samples provided by growers, and if those are cherry-picked, doctored or otherwise unrepresentative of a crop, that isn’t a lab’s fault.

Luce said some growers have even asked her to recommend labs that would give more favorable results than her lab.

Some in the industry have challenged MacRae’s findings in comments to his blog. On moisture tests, for instance, a result above 15 percent is a failing grade. MacRae found an abundance of results between 14.5 and 15 percent, but a tiny amount over 15 percent.

Commenters noted that growers can get a rough measure of moisture before taking samples to a lab and wouldn’t bring samples they believe to be over 15 percent. That could explain the dramatic drop in tests with scores above 15 percent.

But MacRae reasons that the concentration of results just below 15 percent is statistically unusual. Dr. Michelle Sexton, who was the chief science officer at a Kirkland lab that went out of business, agrees.

“You expect a bell-shaped response curve and it’s not what he’s seeing,” Sexton said. “Jim is on to some very important things. He’s well-trained. He has the means and methods for looking at data that reveals trends.”

In tests for residual solvents, such as butane, used in producing hash oil, MacRae found five labs that did not fail any tests in the third quarter of 2015, but one that failed 14 percent of samples.

On potency, he found four labs averaged at least 20.5 percent in total cannabinoids, which include key chemicals such as THC. The other labs all averaged between 15.6 and 19.5 percent.

“If a lab consistently produces higher THC results, I think growers would be more interested in using them,” said Kristi Weeks, policy counsel for the state Department of Health (DOH).

Under state law, LCB regulators don’t oversee labs to the same degree they do licensed pot businesses, which LCB agents inspect and test through “secret shopper” investigations employing underage buyers. Violations by licensees have led to warnings, fines, suspensions and cancellations of licenses.

Instead the LCB uses a third party, the RJ Lee Group, to certify the labs and then periodically recertify them after an audit. That process has shortcomings, some say.

Luce of GOAT Labs calls it a joke.

 

Related video: Testing cannabis

Cameron Miller explains what Bellevue laboratory The WERC Shop looks for in cannabis testing. Miller, the lab’s director of analytical services and quality assurance, runs up to 100 samples a day. Read more. (Corinne Chin / The Seattle Times)    

Labs are certified for having the proper personnel, equipment and methods. But what’s lacking, according to lab executives and experts, are proficiency tests. Such tests would give labs samples of products about which key contents are known. The labs would then test them and if their results were not close enough to the known totals, then those labs’ certification might be in jeopardy.

Sexton, who said her lab failed for lack of capital and business, believes the state should standardize practices and equipment for all labs. Until then, she said, testing data amounts to “garbage in, garbage out.”

Proficiency tests alone aren’t enough, Luce said, because a lab could pass those periodic checks and still go back to delivering biased results to customers.

“I fully support proficiency tests, but they’re not the end-all,” she said.

Steep Hill’s Lampach said proficiency tests should be supplemented by secret-shopper tests. And Lampach said labs that fail state oversight should be decertified. “It should be one (violation) and done. This is no time for games,” he said, as Washington’s legal marijuana experiment is being watched around the country.

MacRae said it’s very likely he will eventually name the labs in his analysis. But he’s not sure if he will do so publicly or just reveal their identities to state regulators.

“Naming names at this point personalizes it and potentially hurts business for labs,” he said. “The issue of bias and whether it exists is what’s important.”