Faulty Statistics of Marijuana Driver Fatalities

The U.S. House of Representative’s Committee on Oversight and Government Reform held five hearings on marijuana titled Mixed Signals and one hearing specifically on drugged driving titled Planes, Trains, and Automobiles: Operating While Stoned. (See Mixed Signals: Part 1Part 2Part 3, and Part 4). At the first Mixed Signal hearing the then-Deputy Director and now Director of the Office of National Drug Control Policy (ONDCP) , Michael Botticelli, claimed twenty-nine percent of all positive drug-tested drivers in fatal motor vehicle accidents had cannabinoids in their system. During the second hearing, Deputy Administrator of the Drug Enforcement Administration (DEA), Thomas M. Harrigan, told Congressman John Mica (R-FL) that “28 percent of fatally injured drivers test positive for marijuana.” At the “Operating While Stoned” hearing, Congressman John C. Fleming (R-LA) stated that “Colorado drivers testing positive for marijuana were involved in 28 percent of fatal drug-related vehicle crashes. That number increased to 56 percent by 2011.”

The statistics presented by the officials greatly misrepresent the number of marijuana fatally-injured drivers. The statistic is not correct cited by Botticelli, Harrigan cited the wrong percent and said it incorrectly, and Felming cited correctly but it misrepresents, as do all, the percentage of fatally-injured drivers who had cannabinoids in their system. In addition to the discrepancy in the statistics being provided in congressional hearings by administration officials and politicians, it is unclear the proportion of fatally-injured drivers who tested positive for marijuana were actually impaired (under the influence of THC) or had left over non-psychoactive marijuana metabolites.

Marijuana can stay in a person’s system for days, weeks, or months. A person’s height, weight, tolerance, activity level, and other lifestyle factors can affect how long cannabinoids can remain detectable by a drug test. Non-psychoactive marijuana metabolites will result in positive blood, urine, sweat, oral, and hair follicle test results long after impairment. Usually, drug tests measures for the left over non-psychoactive elements, not if you are currently under the influence of the psychoactive elements of marijuana. An ONDCP White Paper on drugged driving illustrates the complexity and impossibility “to establish an impairment level for cannabis because the relationship between the concentration of THC and marijuana metabolites in blood, urine and oral fluids.”

To better understand marijuana’s influence on a person’s ability to safely operate a motor vehicle, there has to be an accurate cannabis-related test to demonstrate the individual behind the wheel had been using marijuana at the time of the accident and to create a limit on what constitutes impairment. After trail and error, researchers developed an accurate test to create a limit on alcohol impairment and driving ability. In the 1990s, grant-funded research persuaded states to lower BAC levels to eight percent. Twenty-one states – Arizona, Colorado, Delaware, Georgia, Illinois, Indiana, Iowa, Michigan, Minnesota, Montana, Nevada, North Carolina, Oklahoma, Ohio, Pennsylvania, Rhode Island, South Dakota, Utah, Virginia, Washington and Wisconsin – have per se laws on the books that specify any or a specific amount of drugs in your system is forbidden; some states have created a limit at two nanograms – Ohio and Nevada – to five nanograms – ColoradoMontana, and Washington – of cannabinoids per milliliter of blood. Around fifteen have zero tolerance illegal drug per se laws which means any level of illegal substances in your body is driving under the influence, even if you are not impaired and have not used the substances in the past week. Developing an accurate picture of fatally-injured drivers who test positive for marijuana is difficult because of the lack of comprehensive data collection (such as an exact amount of drug in a person’s system at the time of the accident, and if the drug level constitutes the person was under the influence).

The Department of Health and Human Services (HHS) Substance Abuse and Mental Health Services Administration (SAMHSA) conducts an annual report on substance use and mental health titled National Survey on Drug Use and Health (NSDUH), giving researchers a wealth of data for studying substance use and mental health trends. According NSDUH annual reports, over a five year period (2005-2009), 205.8 million individuals drove under the influence of an illegal or legal substance. There were 51.1 million drivers who drove under the influence of drugs and 154.7 million drivers who drove while under the influence of alcohol. The drugged drivers averaged 10.2 million and alcoholic drivers averaged 30.9 million drivers or about 41 million people in one year drive under the influence. These numbers can be inflated and repeated if the person took the survey each year and if a person responded to the question of driving under the influence of both drugs and alcohol. During this same period, according to F.B.I. Uniform Crime Reports, there were 7,183,716 arrests for driving under the influence of alcohol and illegal drugs (1.44 million people each year).

The ONCDP Statistic: Botticelli and Harrigan

One study released by ONDCP – the study they used – calls into question the veracity of the statistics cited by Harrigan and Botticelli. The report evaluated the Department of Transportation (DOT), National Highway Traffic Safety Administration’s (NHTSA) Fatality Analysis Reporting System (FARS) statistics on driver fatalities. The report viewed five years (2005-2009) of information. During this time, 127,461 drivers died in motor vehicle accidents. 49,745 people (40 percent) were not tested. 77,716 drivers were tested, in which 20,161 drivers tested positive for drugs, 48,817 tested negative, 8,732 tested inconclusive, and 4,954 drivers tested positive for cannabinoids – about 991 drivers per year. The statistic cited by Harrigan and Botticelli views the 20,161 positive drug-tested drivers. They left out over 50,000 fatally-injured drivers who were not tested and nearly 50,000 drug-tested drivers.

Depending how the overall cannabinoid prevalence percentage is calculated you can get various answers, twenty-five percent of positive drug-tested, fatally-injured drivers test for cannabinoids, six percent of tested drivers test positive for cannabinoids, and four percent of all driver fatalities tested positive for cannabinoids. Harriagan and Botticelli cited the higher percentages only looking at those who tested positive, not including other fatally-injured drivers who were not tested and who tested negative and inconclusive.

ONDCP Data on Cannabinoid Prevalence of All Driver Fatalities, Drug-Tested Driver Fatalities, and Positive Drug-Tested Driver Fatalities

Year Total Drivers Drug Tested Drivers Positive, Drug Tested Drivers Cannabinoid Positive Drivers
Total Total Total Total % of all driver fatalities % of drug-tested % of drug-tested positive drivers
2005 27,491 15,363 3,710 839 3.1% 5.5% 22.6%
2006 27,348 16,193 4,018 987 3.6% 6.1% 24.6%
2007 26,570 16,676 4,214 1,012 3.8% 6.1% 24.0%
2008 24,254 15,683 4,267 1,117 4.6% 7.1% 26.2%
2009 21,798 13,801 3,952 999 4.5% 7.2% 25.3%
Total 127,461 77,716 20,161 4,954 3.9% 6.4% 24.6%

The Rocky Mountain High Intensity Drug Trafficking Area Statistic: Congressman Fleming

Congressman Fleming used similar limited data as Harrigan and Botticelli to get the point across that more drivers are using marijuana before operating motor vehicles. He specifically said, “Colorado drivers testing positive for marijuana were involved in 28 percent of fatal drug-related vehicle crashes. That number increased to 56 percent by 2011.” He obtained the statistic from a report by Rocky Mountain High Intensity Drug Trafficking Area (RMHIDTA), a component of the White House’s National Drug Strategy. (They released three reports on marijuana: two on the legalization of marijuana (Volume 1 and Volume 2) and one on medical marijuana.) At the table below, the Colorado Department of Transportation gathered data on driver fatalities. From 2003 – 2013, 7,995 driver died in motor vehicle accidents. Only 3,584 drivers (45 percent) were tested for drugs. 896 tested positive for drug, while the remaining 2,688 tested negative, inconclusive, or another error category. 376 drivers tested positive for cannabinoids. Overall, 4.7 percent of fatally-injured Coloradan drivers test positive for cannabinoids and 10.5 percent of drug-tested, fatally-injured Coloradan drivers test positive for cannabinoids with current data.

The RMHIDTA statistic is calculated the same as the ONDCP statistic: dividing the number of cannabinoid-positive, fatally-injured drivers by the number of drug-positive, fatally injured drivers. When calculating the cannabinoid prevalence statistic RMHIDTA left out over 4,400 fatally-injured drivers who were not tested and nearly 2,700 drug-tested drivers.  Again, this statistic, calculated by RMHIDTA and cited by Fleming, left out the fatally-injured drivers who were not tested and fatally-injured drivers who were tested, but the result was negative, inconclusive, or another error category.

Colorado Department of Transportation (CODOT) Fatality Analysis Reporting System 2003 – 2013 Data of Cannabinoid Prevalence in All Driver Fatalities, Drug-tested Driver Fatalities, and Positive, Drug-tested Driver Fatalities

Year Total Drivers Total Drug-Tested Drivers Total Positive, Drug-Tested Drivers Total Cannabinoid-positive, Drug-Tested Drivers
Total % of all drivers % of all Drug tested Drivers % of all Positive, Drug-Tested Drivers
2003 876 360 123 40 4.57% 11.11% 32.52%
2004 951 226 43 20 2.10% 8.85% 46.51%
2005 847 392 85 29 3.05% 7.40% 34.12%
2006 721 355 75 21 2.91% 5.92% 28.00%
2007 789 438 81 23 2.92% 5.25% 28.40%
2008 712 353 72 31 4.35% 8.78% 43.06%
2009 653 291 82 37 5.67% 12.71% 45.12%
2010 600 292 79 42 7.00% 14.39% 53.16%
2011 587 301 93 52 8.86% 17.28% 55.91%
2012 632 288 78 36 5.70% 12.50% 46.15%
2013 627 288 89 36 5.74% 12.50% 40.45%
Total 7,995 3,584 896 376 4.70% 10.49% 41.96%

From 2006 to 2011, RMHIDTA used Colorado’s DOT FARS data to find out the number of fatalities, but in 2012 they, as detailed in their second report, expanded their search using different agencies boosting the number of reported cannabinoid-positive, driver fatalities. If RMHIDTA kept their methodology consistent overtime, then the statistics might not be as dramatic. For their first report, RMHIDTA used CODOT data, but for second report they abandoned CODOT data. Driver fatalities increased from 21 in 2006 to 36 in 2012 or decreased from 40 in 2003 to 36 in 2012. But with RMHIDTA reports you get 32 drivers in 2006 to 70 drivers in 2012. I did my own research with the FARS data; I found an increase of 2 drivers, 33 drivers in 2006 to 35 drivers in 2012. Depending how the person looks at the information and cites it will give a different picture.

Colorado cannabinoid positive results
Colorado Department of Transportation, “Drugged Drivers Involved in Fatal Crashes,” https://www.codot.gov/safety/alcohol-and-impaired-driving/druggeddriving/assets/fatal-data-drug.pdf; Rocky Mountain HIDTA, The Legalization of Marijuana In Colorado: The Impact, Volume 1, August 2013, http://www.rmhidta.org/html/FINAL%20Legalization%20of%20MJ%20in%20Colorado%20The%20Impact.pdf; and Rocky Mountain HIDTA, The Legalization of Marijuana in Colorado: The Impact, Volume 2, August 2014, http://www.rmhidta.org/html/2014%20Legalization%20of%20Marijuana.pdf. I gather my data from the DOT’s NHTSA’s FARS data collection – http://www-nrd.nhtsa.dot.gov/CMSWeb/index.aspx.

Key Drugged Driving Elements Missing at Hearings: Prior Drug Use vs. Drug impairment and Alcohol Involvement

A report by NHTSA specifies a key element which is not disseminated throughout the hearings or testimonies. They wrote, “[i]t is important to note that drug involvement means only that drugs were found in the driver’s system. Drug involvement does not imply impairment or indicate that drug use was the cause of the crash.” Another study found “the prevalence of nonalcohol drugs reported in this study should be interpreted as an indicator of drug use, not necessarily a measurement of drug impairment.”

Another point not mentioned during the hearings by these two administration officials and conservative Congressman is that fifty-seven percent of fatally-injured drivers, who had cannabinoids in their system, had alcohol present. By my calculations, slightly over dhalf of the cannabinoid-positive Coloradan drivers had alcohol in their system as well. Colombia University Mallman School of Public Health researchers found that if a person is operating a motor vehicle under the influence of alcohol, an individual is thirteen times more likely to be involved in a car accident, while under the influence of marijuana an individual is two times more likely to be involved in an accident in comparison of a sober driver. If the person has alcohol and another drug it is twenty-three times more likely for a person to be involved in a wreck. Driver fatalities resulting in only cannabinoid prevalence is not as prevalent as alcohol.

Featured image
Source: German Lopez, “The Risk of Alcohol, Marijuana, and Other Drugs, Explained,” Vox Media, February 25, 2015, http://www.vox.com/2015/2/25/8104917/drug-dangers-marijuana-alcohol.

Recently, the Government Accountability Office (GAO) published a report in February 2015 entitled Drug-Impaired Driving: Additional Support Needed for Public Awareness Initiatives. The GAO report highlighted the limit data of self-reported, impaired-driving arrests, toxicology results, and crash data. Within in toxicology data collection, the GAO found “a lack of separation of data from driving under the influence (DUI) arrests, underreported instances of drug-impaired driving, decentralized reporting, and a lack of standardization in drug testing.” To be more detailed:

  • Separtation of Data
    • [S]tate arrest data does not currently separate data on drug-impaired driving and alcohol-impaired driving cases across law enforcement agencies
  • Underreported Instances
    • [D]rug-impaired driving may be underreported as drivers impaired by both alcohol and drugs will likely be tested and prosecuted only for alcohol impairment because, according to officials from NHTSA and six of the seven selected states, evidence collection and prosecution are much easier for alcohol-impaired driving.
    • Officials from NHTSA and six states said that in general, if a person suspected of impaired driving has a BAC over 0.08, the individual is not tested further for the presence of drugs, regardless of whether drug impairment is also suspected. As a result, drivers impaired by both drugs and alcohol may not be reported accurately in arrest data, contributing to a lack of knowledge about the number of drivers impaired by both drugs and alcohol.
  • Decentralized Reporting
    • Among the selected states, data on DUID arrests are generally collected by local law enforcement agencies, and one of the seven selected states collects statewide data on DUIDs in a centralized database.
    • In five of the seven selected states, toxicology data are maintained by individual state and local law enforcement agencies and toxicology labs (including private and public labs), with no centralized database. For example in California, DUID testing varies by jurisdiction and can be completed by one of 22 private labs or 6 public labs at the local or state level.
    • Vermont and Washington have centralized results of all DUID drug tests, which are conducted by a single laboratory for each state.
  • Lack of Standardization
    • Officials from three of the seven selected states, as well as representatives from SOFT, stated that a lack of standardization among labs means that test results from different labs cannot necessarily be compared. For example, labs do not have uniform reporting-level cutoffs for drugs (the level at which a drug is reported as present). Therefore, for the same sample, one lab may report the sample as positive for the presence of a drug while another lab may report the sample as negative because the amount present is below the reporting level cutoff for the second lab.
    • According to officials from HHS, while there are federal standards for forensic toxicology testing for federal agencies and states may establish standards for forensic testing, there are currently no federal laboratory certification requirements for forensic laboratories conducting toxicology testing for state and local law-enforcement agencies.

I do not think anyone would deny that marijuana use impairs driving abilities just as alcohol and other substances impairs driving abilities. Under the influence of alcohol is more dangerous for personal and public safety, while “[d]rivers under the influence of marijuana retain insight in their performance and will compensate where they can, for example, by slowing down or increasing effort. As a consequence, THC’s adverse effects on driving performance appear relatively small,” as a 1993 NHTSA study indicates. The study concluded:

  • Marijuana smoking which delivers THC up to a 300 pg/kg dose slightly impairs the ability to maintain a constant headway while following another car
  • A low THC dose (100 μg/kg) does not impair driving ability in urban traffic to the same extent as a blood alcohol concentration (BAC) of 0.04 g%
  • Drivers under the influence of marijuana tend to over-estimate the adverse effects of the drug on their driving quality and compensate when they can; e.g. by increasing effort to accomplish the task, increasing headway or slowing down, or a combination of these
  • Drivers under the influence of alcohol tend to under-estimate the adverse effects of the drug on their driving quality and do not invest compensatory effort
  • Is not possible to conclude anything about a driver’s impairment on the basis of his/her plasma concentrations of THC and THC-COOH determined in-a single sample.

But the 1999 NHTSA study found worse marijuana impairment because the researchers “believe that these differences are attributable to the groups’ respective experience with THC smoking and to driving under the influence of THC. The present group was less experienced and probably had not developed the same degree of behavioral tolerance as their predecessors.”

The prohibition of marijuana and other drugs creates a system where officials and researchers look to find if a person had drugs in their system rather than if the person had been under the influence of drug use at the time of the crash. Since our nation strictly prohibited drugs in the 1970s, there has been no research in the area to understand the socalled illegal substances better. You cannot push the issue under the rug thinking drug use and abuse will disappear on its own if you have a structure strictly against drug use and implementing supply-side initiatives. Eliminating drug use and abuse is idealistic – it will not happen – it is not bounded rationality, human nature, or history. Drug use will always be around, it has for thousands of years for many purposes from social to medical to religious. It will be around for another thousand years. We have to learn to live with drugs. It is possible to reduce drug use, abuse, misuse, dependency, overdose, and other associated harms within a public-health approach of treatment, education, and prevention than in a criminalization and stigmatization approach that only creates alienation from friends and family, embarrassment for being arrested and being a socalled “drug addict”, a second-class citizen, greatly decreases the chance of employment, diminishes family and  community stability, and increases distrust among the law enforcement community.

The ONDCP and RMHIDTA or essentially the White House National Drug Control Policy has to stop twisting the truth and change their statistic, so the administration officials and politicians use the correct numbers. But, the statistics presented by Botticelli, Harrigan, and Fleming are correct, in terms of the methodology, limited data, and variables – the numbers are being misrepresented and conflated and it does not represent the overall data as explain above. The problem is the dissemination of the information to the politicians and other people who have not done the research, and the message will come out that nearly thirty percent of fatally-injured drivers were under the influence at the time of the crash rather than the truth.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s