More bad journalism

imagesJacob Sullum, a pro-pot writer for Reason and Fortune magazines, is at it again.

This time, he claims that marijuana’s impact on crash risk has been greatly exaggerated, based upon Rogeberg and Elvik’s (R&E) paper, soon to be published in the journal Addiction. Sullum, in his typical snarky style, writes that there’s not much evidence of a surge in traffic fatalities due to marijuana-impairment in Colorado or Washington. This shouldn’t be a surprise, since neither state bothers to measure marijuana-impairment. Blood tests, yes. But impairment, no.

Based on the R&E review, Sullum points out that the crash risk due to marijuana use is similar to drivers with a BAC of .05, which is below the .08 legal limit in the United States. He fails to note that .05 is the DUI limit used in most European countries. It’s also the DWAI limit in Colorado.

Sullum’s screeds are so unbalanced, they aren’t worth critiquing. But R&E’s paper is worth the following critique–

R&E performed a meta-analysis of 28 separate research studies. A meta-analysis is not raw research, but is a statistically valid technique that combines results of many different authors’ raw research into a single outcome. In this case, the outcome sought was the “odds ratio,” which is a measure of the increase in likelihood of someone being involved in or causing a crash because of marijuana. An odds ratio of 1.0 means there is no increase in risk, whereas an odds ratio of 2.0 means the risk is doubled.

Using two different statistical bases, R&E estimated the odds ratio to be either 1.22 or 1.36, which is lower than odds ratios of 1.92 and 2.66 from two earlier meta-analyses critiqued by R&E. All of these findings are well below typical odds ratios reported for alcohol.

For example, the odds ratio for alcohol has been shown to range between 3 and 5 when a driver’s BAC is at the legal limit of .08. Drivers arrested for DUI typically have much higher levels of BAC, such as .12, where the odds ratio is between 8 and 13, depending on the study. The odds ratio for drivers on both alcohol and marijuana is estimated to be between 14 and 20. 

R&E concluded, “Acute cannabis intoxication is associated with a statistically significant increase in motor vehicle crash risk. The increase is of low to medium magnitude.”

Their conclusion isn’t news. Nearly all researchers have reached the same conclusion, although their odds ratio numbers differ. For example, the odds ratios of the studies evaluated by R&E ranged from .22 (marijuana impairment improves driving) to 13.40 (marijuana impairment is worse than alcohol), with a median of 1.37. One researcher published 6 different odds ratios for different countries, ranging from .22 to 6.59!

Why do reported odds ratios differ so markedly from one researcher to another, and even from one country to another for the same researcher? The main reason is that most researchers rely upon measured blood levels of THC as a surrogate for measuring impairment. This in spite of the fact that no researcher in the world has claimed that tested blood levels of THC are reliable indicators of a driver’s degree of impairment. Legislators who enact THC legal limits of 5ng/ml might believe that to be the case, but there is no science to support that belief.

Assuming THC blood tests correlate well with impairment is a common mistake. Because blood tests are a good surrogate for measuring alcohol impairment, many people assume the same is true for THC. But THC is unlike alcohol chemically, biologically and metabolically. There is no reason that tests for alcohol should also work for THC. The rate of clearance of smoked marijuana’s THC from blood is so rapid that the THC level in tested blood cannot represent the typically higher THC level in the driver at the time of arrest or crash. That’s not true for alcohol, which has a slow, predictable, linear blood clearance rate. Furthermore, within minutes after smoking, the blood level of THC is much lower than the THC level in the brain. This is because THC is fat soluble, whereas alcohol is water soluble.

R&E sought “to determine whether and to what extent cannabis intoxication increases motor vehicle crash risk.” They failed to even attempt that. Rather than measuring cannabis intoxication, they measured lab results.

Does this mean that the R&E paper is a waste of ink? Perhaps. But Sullum’s review most certainly is a waste of ink.

Just because driving under the influence of marijuana is less dangerous than driving under the influence of alcohol doesn’t mean that it isn’t dangerous. Marijuana-impairment, as well as impairment by other drugs and polydrug impairment kills and maims innocent victims. There will always be attempts by the pot lobby to whitewash the facts, but science simply isn’t on their side.


DUID Articles