Send email for updates
Macdonald S., Hall W., Roman P. et al.
Addiction: 2010, 105, p. 408–416.
Unable to obtain a copy by clicking title? Try asking the author for a reprint by adapting this prepared e-mail or by writing to Dr Macdonald at firstname.lastname@example.org. You could also try this alternative source.
A review of 20 years of research on cannabis testing at work found that the patchy evidence permitted few strong conclusions about effectiveness. Given the lack of evidence, it seems wise for UK guidance to limit workplace testing to those who really warrant it because of their jobs, rather than advocating widespread testing as a deterrent.
Summary Reviewing 20 years of scientific literature, researchers aimed to assess the evidence behind various aspects of workplace cannabis testing. As in many other countries in Europe, in the UK workplace drug testing is limited primarily to professions where drug use might cause a significant risk to safety, such as transport or the military. By contrast, in the United States workplace drug testing is much more common (80% of Fortune 1000 companies have a drug testing programme) and aims not only to prevent risks to safety but primarily to reduce drug use by employees.
As the most commonly used and most commonly detected drug, the review was limited to cannabis, detected in just over 2% of workplace drug tests in the USA compared to compared to about 0.6% for cocaine. To examine the evidence behind workplace cannabis testing, the reviewers searched for research which could help answer the following questions:
• Do the acute effects of cannabis impair performance?
• Does cannabis withdrawal impair performance?
• Are long-term cannabis users at increased risk of job accidents?
• Are those who test positive for cannabis or who self-report use at increased risk for injuries or accidents?
• Does workplace drug testing deter drug use among employees?
• Does workplace drug testing reduce injuries or accidents?
In assessing the findings greater weight was given to well-designed studies with a lower risk of bias or inaccuracy.
Comprehensive research reviews have found that the immediate effects of cannabis can cause performance deficits, affecting perceptual and motor skills, attention, decision-making, learning and short-term memory. These effects last roughly four hours after smoking cannabis. It was judged unlikely that deficits would persist until the next day. One much cited study which found otherwise was so poorly designed that its results could not be relied on. Well designed studies using blood tests have generally found a link between cannabis use and being involved in motor vehicle crashes, and the weight of the evidence suggests that cannabis use impairs driving performance – and that it might therefore impair performance at work.
There is some evidence that heavy cannabis users who stop using suddenly may suffer mild agitation, appetite changes and disturbance to sleep. It is unlikely that these effects appreciably affect workplace performance.
Some studies have found that long-term heavy use of cannabis could impair cognitive functioning, but problems with the design of the studies make it difficult to draw firm conclusions; other studies have not found such an effect, and even if it is real, its relevance to functioning in daily life and the risk of accidents is unclear.
Many studies have tried to find out if people who either test positive or report using cannabis are more likely to have injuries or accidents compared to those who do not. The results are mixed, and even if there is a link, it is unclear whether this is an effect of cannabis use or some other factor, such as people prone to taking risks also being more likely to use cannabis; in this scenario it is their risk-taking that causes accidents, not their cannabis use. Studies based on urine testing have generally not found that drivers who have crashed or been responsible for an accident are more likely to test positive for cannabis than other drivers, and those which did find this were often undermined by methodological limitations.
It is not possible to draw firm conclusions from the mixed results of the research, but it is probable that after workplace drug testing is implemented, the proportion of employees testing positive falls. It is not known if this is because employees stop using cannabis or if over time they learn to cheat the tests.
Evaluations of whether drug testing programmes have led to fewer injuries or accidents suffer from serious methodological flaws. Typically, testing was only one of a number of safety improvements put in place at the same time, making it impossible to attribute any reduction in accidents to testing itself. Sometimes too what appears to be an impact of testing may simply reflect generally declining workplace accidents and injuries. Since few employees use cannabis, and accidents have a very wide range of causes, large samples are needed to detect what is likely to be at most a small effect.
Many of the studies of the effects of workplace drug testing suffer from design and analysis weaknesses, making it easy to draw conclusions not actually supported by the evidence. The best studies have examined the relationship between driving crashes and cannabis. In those which relied on blood tests, people who had crashed were between 1.4 and 6.6 times more likely to have used cannabis, but studies based on urine tests failed to find any association. The lesson is that blood testing is a more accurate way to assess impairment than urine testing, but this should be weighed against its greater intrusiveness. Employees should at least be given the option to request the more accurate blood test before being punished as a result of a finding from a urine test.
In the USA the primary aim of workplace drug testing is to deter drug use, but declining numbers of employees testing positive for cannabis cannot securely be attributed to drug testing. Possibly this reflects a general trend of declining drug use or more people finding better ways to ‘cheat’ the tests. Due to limitations of the research, it is also not possible to say that testing programmes reduce accidents or injuries. Employers should consider the relative merits of testing for alcohol rather than drugs, given that more people drink than use drugs, and the consequences may be worse – in particular in relation to road crashes, with which alcohol is associated much more strongly than cannabis.
commentary Taking its questions in turn, the answers from this review appear to be:
• Do the acute effects of cannabis impair performance? Yes, but only for the few hours it remains active in the body.
• Does cannabis withdrawal impair performance? Not known but unlikely to be any appreciable impact.
• Are long-term cannabis users at increased risk of job accidents? No direct evidence and findings and implications of studies of cognitive impairment are unclear.
• Are those who test positive for cannabis or who self-report use at increased risk for injuries or accidents? Unclear; findings mixed and research methodologically unsound.
• Does workplace drug testing deter drug use among employees? Probably, but could be other explanations for dips in positive tests.
• Does workplace drug testing reduce injuries or accidents? Unclear due to methodological limitations of the research.
The most repeated finding was that methodological weaknesses in the research meant few strong conclusions could be drawn, in particular about whether it is worth employers instigating drug testing programmes. But if they do, in respect of cannabis the conclusion that there may be a large difference between blood and urine test results, and that the former are a better indicator of impairment, should be noted by any organisation basing its programme on urine tests only.
Given the weakness of the evidence base, it seems wise that UK government guidance specifies that employers must have the consent of their employees to undertake drug testing, usually as part of wider health and safety policy agreed in a contract or staff handbook. The widespread use of drug testing as a preventative measure seen in the USA is not recommended in the UK, where guidance advises employers to limit testing to those who need to be tested, ensure that within this group, testing is random, and not to single out particular employees unless the nature of their jobs requires this. While there may be some leeway depending on the definition of those “that need to be tested”, it seems clear that drug testing in the UK might be appropriate for train drivers or airline pilots, where there is a clear safety risk, but not for most workers.
The featured review focused on cannabis, but possibly relevant too are studies of drug testing programmes in general, in which cannabis is likely to be the most detected drug. A rigorous review conducted for the Cochrane collaboration found just two studies of people who drive for a living which met its quality and relevance criteria. Echoing the featured review, the conclusion was that evidence was insufficient to be able to advise for or against drug and alcohol testing of occupational drivers as the sole long-term solution to preventing injuries.
This draft entry is currently subject to consultation and correction by the study authors and other experts.
Last revised 05 March 2014. First uploaded 03 March 2014
Give us your feedback on the site (two-minute survey)
Open Effectiveness Bank home page
Add your name to the mailing list to be alerted to new studies and other site updates
HOT TOPIC 2016 Don’t treat, just test and sanction
DOCUMENT 2012 Will intensive testing and sanctions displace treatment?
REVIEW 2011 A new paradigm for long-term recovery
STUDY 2009 The 24/7 Sobriety Project