Is Cannabis Really a Gateway Drug?

You’ve probably heard someone say that cannabis is a gateway drug before. It’s become one of the most common anti-cannabis arguments and an often-cited reason that marijuana should remain illegal.

The overall principle is simple. Using weed introduces you to the world of illegal drugs, removing stigma around drugs and priming your brain to search for a better hit in the form of something like cocaine or heroin. This, anti-cannabis advocates argue, is one reason it’s bad to make marijuana legal.

As part of our series of the Best Arguments for and Against Cannabis, we’re going to delve deep into the gateway effect. We’ll look at what it means, why it looks bad for cannabis, and how people in favor of cannabis respond to the criticism.

Summary: Is There a Gateway Effect from Cannabis?

So is cannabis a gateway to other drug use? Here’s a summary of both sides of the debate:

Yes…

• The “gateway sequence” of drug use is extremely common in all western research.
• Animal research clearly shows that THC sensitizes teen brains to opioids in future.
• The gateway effect is still present when researchers control for additional variables and in some twin studies.
• Even if not physically caused by marijuana, it could be caused in other ways. For example, the psychological effect of switching to illegal drug use could lead to other drug use. Likewise, associating with dealers and users could lead to other drug use.

No…

• The gateway sequence reflects the availability of different substances, which is why it varies around the world.
• As a result, the specific sequence of substances is less relevant than the factors that sent users down that path.
• Research that controls for psychological, social and other factors find either a much smaller gateway effect or none at all.
• Common liability to addiction and similar approaches explain both the subsequent drug use and the original “gateway” use.

What is the Gateway Hypothesis?

The gateway hypothesis is the name given to the idea that cannabis use encourages people to use other, “harder” drugs like heroin or cocaine. In short, it claims that people go from no drug use, to the use of legal drugs like alcohol or tobacco, to the use of cannabis and then onto other illegal drugs.

This ultimately came from a 1975 paper from Kandel and Faust, which described a common sequence of use among high school students in New York State. The paper basically involved following a couple of different groups of teens from the fall of 1971 to the spring of 1972. It followed up again after they graduated high school. The idea was to follow these teens and see how their use of drugs evolved from the initial surveys to the one afterwards.

Generally, they found that people either stay in the same “category” of drug use or move one level either up or down. In other words, not many people move to drugs like cocaine or heroin without trying cannabis first.

Moving From Weed to “Hard Drugs” – The Basic Case for a Gateway

Many pieces of research from many places around the world confirm that marijuana use generally – but not always – precedes hard drug use. This could be explained by marijuana causing people to use harder drugs (e.g. by “priming” their brain for drug use) or by some underlying reason that makes some people more likely to smoke cannabis and more likely to use other drugs.

The gateway hypothesis is the first of these possibilities. If this is the case, then allowing or otherwise encouraging marijuana use would increase the rates of more serious drug use. So while weed is not as dangerous as heroin, what good is that if it makes people use heroin a few years down the line? This is the core of the “cannabis is a gateway drug” argument against weed. 

Evidence for a Gateway Effect

The basic evidence for a gateway effect follows the general pattern of the result from Kandel’s research (Kandel and Faust, 1975). The original study was based on a sample of teens from New York. It showed that of the people who start to use drugs, there tended to be a sequence: non-use, legal drugs, cannabis, pills, psychedelics, cocaine and heroin. Kandel points out that this isn’t always followed. In fact, around 60% of cannabis users never progressed in the population studied.

Many other studies have confirmed the key observations that led to the gateway hypothesis:

• Later research (Yamaguchi and Kandel, 1984) followed up with the same participants again. They found that the patterns continued, with some minor differences between males and females.
• Secades-Villa et al. (2015) used data from the National Epidemiological Survey on Alcohol and Related Conditions. They found that 44.7% of the 6,624 participants who started using cannabis before other drugs progressed to other drug use.
• Lynskey, Vink and Boomsma (2006) tested the idea using “discordant” twins from the Netherlands Twin Register. This is where one twin had used cannabis and the other hadn’t. The researchers found that twins who’d used cannabis early in life were 6.5 times more likely to have used hard drugs or party drugs by the end of the study. In total, 17.9% of early cannabis users progressed to further drug use.
• Taylor et al. (2017) surveyed UK teens and followed up with them at age 21 to ask about other illicit drug use. All cannabis users (regardless of how often) were more likely to be dependent on nicotine, drink problematically and use other illicit drugs, compared to non-users. With adjustments for some relevant factors, regular cannabis users were 26 times more likely than non-users to have tried other illicit drugs.
• Kirby and Barry (2012) affirm the general pattern of the gateway hypothesis. However, they point out more explicitly that alcohol should be considered the main “gateway” drug, not cannabis. However, the role of cannabis in the progression was ultimately the same in this study as in Kandel’s original work.

Why is Cannabis a Gateway to Harder Drugs? What Exactly Happens?

The mechanism underlying the gateway effect is a major topic of discussion in both the wider debate and the academic literature. Researchers have condensed the possibilities into three options:

• Trying “soft” drugs like cannabis might lead to a psychological or physical need for repeating that experience and trying stronger but similar experiences.
• Actually acquiring and using cannabis can put people into contact with dealers and users. This, in turn, could lead people to trying other drugs.
• The fact that people try cannabis without (at least obvious) negative effects undermines public health messaging around drug use. “If weed is OK, then maybe they’re wrong about other stuff too…”

It’s a little difficult to test the last two. They’re clearly true to some extent, but how relevant is hard to say. But the first has been subject to experiments.

• Ellgren et. al. (2006) gave rats doses of THC throughout adolescence (or placebo). They found that the THC-treated rats continually increased their heroin dose, while the non-treated rats stayed consistent. However, the rats did not choose to take the heroin originally.
• Cadoni et. al. (2001) took three groups of rats: giving one group THC, one morphine and the other a placebo for three days. After a two-week break, each rat randomly received either cannabis, morphine or a cannabis agonist. The researchers found that the THC-treated rats responded more to morphine and to THC in the second test, and the morphine-treated rats also responded more to THC. This suggests “cross-sensitivity” between the substances.
• Renard et. al. (2016) looked at many studies conducted on rats relevant to the gateway effect, and found that the literature generally confirms the results discussed above. THC appears to prime rats – and possibly humans, since our reward centers are similar – for opiate use in later life. Results for other drugs such as cocaine are less clear-cut, but seem similar.

Why Did They Try Weed in the First Place? The Case for Common Liability

Critics of the gateway hypothesis (Vanyukov et. al., 2012) point to a substantial gap in the logic of the explanation. Why did the individual use cannabis/alcohol/“the gateway” originally? Cannabis may make people more likely to try hard drugs, but what makes one person try cannabis and the other not try it? The gateway hypothesis has no answer, but common liability to addiction does.

The key point of common liability is that existing genetic, environmental and psychological factors make specific people more likely to drink alcohol, smoke tobacco, try cannabis and use “harder” drugs. Proponents argue that there are multiple, unconsidered variables at play, not some specific “third variable” that gateway researchers miss. While the details get quite heavy (and you can check Vanyukov’s paper for the full scoop), it’s clear that – for example – a rough home life coupled with depression and drug-using friends may make someone more likely to experiment with many types of drugs.

Evidence for Common Liability and the Limitations of the Gateway Hypothesis

There are many different types of evidence against the gateway hypothesis, including twin studies, research from different cultures and studies where more variables are accounted for. One thing to remember, though, is that common liability also explains much of the evidence used to support the gateway.

A good example of this type of research (Cleveland and Wiebe, 2008) used twins in the Add Health study of American teens, comparing the outcomes of both identical and non-identical twins based on their marijuana use. They ended up with 510 pairs of twins (285 identical and 225 non-identical). They calculated the twin-to-twin difference in their early marijuana use. The research found that while non-identical twins showed the “gateway effect” (i.e. early marijuana use predicted later drug use), the same was not true for identical twins.

The authors concluded, “Although the gateway hypothesis implies that there is something about marijuana use that, by itself, potentiates the risk of hard drug use, our results imply that there is something genetic that renders the use of both marijuana and other drugs more likely for some people.”

Other evidence for common liability and against the gateway effect includes:

• Degenhardt et. al. (2010) looked at the World Health Organization’s World Mental Health surveys, carried out in 17 countries around the world. They found that the association between “gateway” substances and later hard drug use depends on how common these substances are in the country. In Japan, for instance, only 1.6% had used cannabis. As a result, 83% of people who use harder drugs didn’t use cannabis before.
• Reddon et. al. (2018) looked at at-risk youth in Canada. They found that those who reported at least daily use of cannabis at the start of the survey were 34% less likely to start injecting drugs by follow-up. This effect was mainly due to a reduction in the chance of injecting stimulants, but there was no difference for opioid injecting. Under the gateway hypothesis, the result should be an increase in all injecting.  
• Vanyukov et. al. (2009) developed a liability scale for drug use, and put it to the test using the children of people with substance abuse disorders and both genetic and fraternal twins. The idea was to account for the variability in risk in a more reliable way than the gateway hypothesis does. The authors were successful, although the measure worked better for high risk individuals than lower-risk ones.
• McCutcheon and Watts (2017) looked at the effect of psychological “strain” on the gateway hypothesis. They used Add Health study data and found, like most research, that cannabis use predicts other drug use in future. However, when they accounted for strain, this association disappeared. Strain, however, predicted both the other drug use and the original cannabis use.

Our Take on the Gateway Hypothesis

People appear to use cannabis and then harder drugs because of cultural norms and underlying psychological, social and genetic factors that make them want to use any drugs. Calling one substance a “gateway” to another misses the much more complex and important picture that explains the original behavior as well as what happens afterwards. It is a stronger tool for political discussions than scientific ones.

---

References

Agrawal, A. et al. (2004) “Cannabis and other illicit drugs: Comorbid use and abuse/dependence in males and females,” Behavior Genetics, 34(3), pp. 217–228. https://doi.org/10.1023/b:bege.0000017868.07829.45

About the source:

• Peer reviewed? Yes, published in Behavior Genetics.
• Methodology: Data was taken from the Virginia Twin Registry. Researchers used it to test 13 models of drug comorbidity (using multiple drugs).
• Sample size: 1,191 male and 934 female twin pairs.
• Main results: The correlated morbidities model (common liability, in the language of the study) was a good fit for the data for both cannabis and other drug use. The adapted gateway model considered didn’t entirely explain the data, but there was evidence that high-risk cannabis users have a higher risk of other drug use.
• Other notes: The study had quite a limited amount of other illicit drug use, and the authors combined many different categories of drugs into a single group to enable conclusions to be drawn. However, this could have missed many important details. They also didn’t consider the age of first use of cannabis.

Bretteville-Jensen, A.L., Melberg, H.O. and Jones, A.M. (2008) “Sequential patterns of drug use initiation – can we believe in the gateway theory?,” The B.E. Journal of Economic Analysis & Policy, 8(2). https://doi.org/10.2202/1935-1682.1846

About the source:

• Peer reviewed? Yes, published in The B.E. Journal of Economic Analysis & Policy.
• Methodology: Postal survey of Norwegian 21 to 30 year olds. Aside from the basics, the authors focused on applying a multivariable probit model to account for the additional variables likely to be at play.
• Sample size: 4,561
• Main results: The gateway effect was present in the initial analysis. However, when more relevant factors were taken into account, it was substantially reduced (but still present).
• Other notes: This study wasn’t longitudinal, so depended on recall for many data points. They also note that it’s likely the model didn’t account for every relevant variable, particularly those that become more or less relevant with time.

Cadoni, C. et al. (2001) “Behavioural sensitization after repeated exposure to Δ 9 -tetrahydrocannabinol and cross-sensitization with morphine,” Psychopharmacology, 158(3), pp. 259–266. https://doi.org/10.1007/s002130100875

About the source:

• Peer reviewed? Yes, published in Psychopharmacology.
• Methodology: Male rats were given THC in increasing amounts over three days, twice per day. Equal numbers were given morphine or simply a placebo. After a 14 day break, they were given either THC, morphine or a cannabinoid agonist, and their behavior was assessed.
• Main results: Rats treated with THC showed an increased response to THC and morphine given after the break. Rats treated with morphine also showed a heightened response to THC, so there was cross-sensitization.
• Other notes: This offers a potential explanation for observations rather than establishing a gateway effect in itself.

Cleveland, H.H. and Wiebe, R.P. (2008) “Understanding the association between adolescent marijuana use and later serious drug use: Gateway effect or developmental trajectory?,” Development and Psychopathology, 20(2), pp. 615–632. https://doi.org/10.1017/s0954579408000308  

About the source:

• Peer reviewed? Yes, published in Development and Psychopathology.
• Methodology: Used data from Add Health, a nationally-representative survey of American high school students, using the three “in-home” data waves, from 1995, 1996 and 2001. They focused on same-gender monozygotic (identical) and dizygotic (non-identical) twin pairs.
• Sample size: 510 twin pairs, 285 monozygotic and 225 dizygotic.
• Main results: Differences in early marijuana use was linked to differences in hard drug use later in life, but only in dizygotic twins. In monozygotic twins, marijuana use didn’t predict later drug use. They conclude that the apparent “gateway effect” might be better conceptualized through genetic variation.
• Other notes: Since the researchers compared outcomes based on the amount of cannabis use, not “use” vs. “non-use,” in some ways this tests a slightly modified version of the gateway effect. Instead, the researchers answer, “does more cannabis use lead to more hard drug use in future?”

Degenhardt, L. et al. (2008) “Does the ‘gateway’ matter? associations between the order of drug use initiation and the development of drug dependence in the national comorbidity study replication,” Psychological Medicine, 39(1), pp. 157–167. https://doi.org/10.1017/s0033291708003425

About the source:

• Peer reviewed? Yes, published in Psychological Medicine.
• Methodology: The authors used data from the US National Comorbidity Survey Replication. They found “violators” of the gateway sequence and looked statistically what predicts violations and whether they’re associated with dependence risk.
• Sample size: 9,282 (5.2% of whom violated the gateway sequence)
• Main results: They found no link between gateway violations and later dependence overall (with minor exceptions). However, early-onset internalizing disorders (e.g. depression and anxiety) increased the chance of violation. Both internalizing and externalizing disorders (e.g. bipolar and ADHD) increased the risk of all drug dependence.
• Other notes: This study asked participants to remember the order they first used drugs. This may introduce some error through misremembering.

Degenhardt, L. et al. (2010) “Evaluating the drug use ‘gateway’ theory using cross-national data: Consistency and associations of the order of initiation of drug use among participants in the WHO world mental health surveys,” Drug and Alcohol Dependence, 108(1-2), pp. 84–97. https://doi.org/10.1016/j.drugalcdep.2009.12.001

About the source:

• Peer reviewed? Yes, published in Drug and Alcohol Dependence
• Methodology: Used data from WHO World Mental Health surveys to look at the order of the “sequence” for different countries.
• Sample size: 85,088, from 17 countries.
• Main results: They found some variation in the sequence by country. The authors write, “the “gateway” pattern at least partially reflects unmeasured common causes rather than causal effects of specific drugs on subsequent use of others.”
• Other notes: Variations in the sequence in different countries don’t necessarily show that cannabis isn’t a “gateway drug,” but it does call the hypothesis into question.

Ellgren, M., Hurd, Y.L. and Franck, J. (2004) “Amphetamine effects on dopamine levels and behavior following cannabinoid exposure during adolescence,” European Journal of Pharmacology, 497(2), pp. 205–213. https://doi.org/10.1016/j.ejphar.2004.06.048  

About the source:

• Peer reviewed? Yes, published in the European Journal of Pharmacology
• Methodology:  Four experiments in which rats were either given a cannabinoid receptor agonist or THC, before they were given amphetamines, with saline controls. The researchers tested the impact of prior cannabis use on the effect of amphetamines, looking at brain dopamine levels and locomotor activity.
• Results: Pre-treatment with the agonist or THC didn’t alter the response to amphetamines, either in terms of behavior or dopamine response. The authors conclude that the results don’t support the gateway hypothesis for cannabis to amphetamines.
• Other notes: Firstly, rats aren’t a perfect model for humans, so as always this should be taken with a pinch of salt. They also didn’t have the expected dopamine response to the cannabinoid receptor agonist, possibly because they injected it into the body cavity rather than intravenously.

Ellgren, M., Spano, S.M. and Hurd, Y.L. (2006) “Adolescent cannabis exposure alters opiate intake and opioid limbic neuronal populations in adult rats,” Neuropsychopharmacology, 32(3), pp. 607–615.  https://doi.org/10.1038/sj.npp.1301127

About the source:

• Peer reviewed? Yes, published in Neuropsychopharmacology.
• Methodology: Rats were given either THC or a placebo every third day from adolescence (from to 28 to 49 days old), and then (after being introduced to it) they were allowed to self-administer heroin in 3 hour sessions from day 57 to day 102 (young to full adulthood).
• Sample size: 12 rats.
• Main results: Rats treated with THC continued to increase their heroin dose, while other rats reached a plateau. It also was shown to alter the behavior of μ opioid receptors.
• Other notes: The authors point out that although the THC-treated rats increased their heroin dose more than the others; they didn’t start self-administering sooner than the non-treated rats.  

Fergusson, D.M. and Horwood, L.J. (2000) “Does cannabis use encourage other forms of illicit drug use?,” Addiction, 95(4), pp. 505–520. https://doi.org/10.1046/j.1360-0443.2000.9545053.x  

About the source:

• Peer reviewed? Yes, published in Addiction.
• Methodology: Longitudinal study of New Zealand children, taking place over 21 years. The authors looked at cannabis use, other illicit drug use and psychological and environmental factors before age 15.
• Sample size: 1,265
• Main results: Those who used cannabis more than 50 times per year were 59.2 times more likely to use other illicit drugs (primarily hallucinogens or prescription drugs), after adjusting for some relevant factors. 99% of other illicit drug users had used cannabis first, but 63% of cannabis users didn’t use other illicit drugs.
• Other notes: The authors point out that this study still can’t really distinguish between cannabis causing the future use and another variable being responsible for both behaviors.

Fuller, C.M. et al. (2001) “Factors associated with adolescent initiation of injection drug use,” Public Health Reports, 116(1_suppl), pp. 136–145. https://doi.org/10.1093/phr/116.s1.136

About the source:

• Peer reviewed? Yes, published in Public Health Reports.
• Methodology: Survey of injection drug users aged 15 to 30 in Baltimore, Maryland. Authors were looking for factors associated with starting to inject drugs.
• Sample size: 226
• Main results: Those who had first had sex before 14, who had smoked crack in the year before injecting, those who’d smoked marijuana and those who used condoms in the year before were more likely to have started injecting as adolescents. African Americans and past-year cocaine users were less likely to have started injecting as adolescents.
• Other notes: The sample size here is very small, and there was a potential for selection bias in the recruiting procedure.

Kandel, D. and Faust, R. (1975) “Sequence and stages in patterns of adolescent drug use,” Archives of General Psychiatry, 32(7), p. 923. https://doi.org/10.1001/archpsyc.1975.01760250115013

About the source:

• Peer-reviewed? Yes, published in Archives of General Psychiatry.
• Methodology: Two prospective, longitudinal studies about legal and illegal drug use.
• Sample size: 8,206 adolescents at time 1 (fall 1971), 7,250 remaining at time 2 (spring 1972) and 1,635 at time 3 (the year after high school graduation).
• Main results: Of the initial users or non-users, those who progress to further drug use tend to follow the pattern: non-use, legal drugs, cannabis, pills (amphetamines, primarily), to psychedelics, to cocaine and to heroin. However, “progression” of any type isn’t hugely common, with 55 and 63% of cannabis users at time 1 and time 2, respectively, remaining at this “level.”
• Other notes: The authors acknowledge that the progression is likely culturally determined, rather than related to inherent properties of specific substances. The authors also point out that those who dropped out were more likely to have used cannabis than those who remained, potentially introducing a bias.

Kirby, T. and Barry, A.E. (2012) “Alcohol as a gateway drug: A study of US 12th graders,” Journal of School Health, 82(8), pp. 371–379. https://doi.org/10.1111/j.1746-1561.2012.00712.x

About the source:

• Peer reviewed? Yes, published in the Journal of School Health.
• Methodology: Used data from Monitoring the Future 2008, focusing mainly on the relationship between alcohol use and illicit drug use.
• Sample size: 14,577
• Main results: Alcohol use is associated with increased use of other substances. For instance, of those who’d drank alcohol before, 58% had used marijuana, compared to 43.4% of the overall group.
• Other notes: This study didn’t have “order of use” data to work with, and so they can’t objectively make any conclusions about gateway sequencing. However, it does suggest that alcohol is a more important gateway drug than people think.

Lynskey, M.T., Vink, J.M. and Boomsma, D.I. (2006) “Early onset cannabis use and progression to other drug use in a sample of Dutch twins,” Behavior Genetics, 36(2). https://doi.org/10.1007/s10519-005-9023-x

About the source:

• Peer-reviewed? Yes, published in Behavior Genetics.
• Methodology: The researchers took the sample from a long-running study based on the Netherlands Twin Register, with the samples taken from the 1995 and 2000 studies.
• Sample size: The total sample was 6,228 twins, with 1,800 participants participating at both waves. There were 219 same-sex twin pairs in which one used cannabis and the other didn’t.
• Main results: The twins with early cannabis use were 6.5 times more likely to report any use of hard drugs or party drug at a later stage, compared to their non-using twin. Overall, 17.9% of early cannabis users reported any hard or party drug use later.
• Other notes: While the twin design automatically controls for many relevant factors, others – such as peer groups – may still be substantially different between the twins.

Mayet, A. et al. (2012) “Cannabis use stages as predictors of subsequent initiation with other illicit drugs among French adolescents: Use of a multi-state model,” Addictive Behaviors, 37(2), pp. 160–166. https://doi.org/10.1016/j.addbeh.2011.09.012

About the source:

• Peer reviewed? Yes, published in Addictive Behaviors.
• Methodology: Cross-sectional study of French adolescents, with retrospective reporting of drug use.
• Sample size: 29,393
• Main results: Cannabis experimenters were 21 times more likely to start using other drugs than non-users, and daily cannabis users were 124 times more likely, again compared to non-users.
• Other notes: Authors didn’t account for psychological or social variables in the analysis, so it is really a basic demonstration of the gateway sequence. Other research identifies many relevant factors beyond the previous drugs used.  

McCutcheon, J.C. and Watts, S.J. (2017) “An examination of the importance of strain in the cannabis gateway effect,” International Journal of Offender Therapy and Comparative Criminology, 62(11), pp. 3603–3617. https://doi.org/10.1177/0306624×17729433

About the source:

• Peer reviewed? Yes, published in the International Journal of Offender Therapy and Comparative Criminology.
• Methodology: Looked at participants in waves 1 and 2 of the Add Health study (as in [8, 14]), looking at cannabis use at wave 1 and other drug use at wave 2. The researchers took psychological “strain” into account, hypothesizing that any apparent “gateway” effect would be reduced when it was accounted for.
• Sample size: 12,403
• Main results: In the overall data-set, cannabis use at wave 1 was associated with other drug use before wave 2. However, when strain was accounted for, the association between cannabis use and other drug use was no longer significant. Strain, however, predicted both other drug use by wave 2 and cannabis use at wave 1.
• Other notes: Participants who’d already used other drugs were discounted from the analysis (because they already had the outcome at baseline), but these could have included many more clear-cut “gateway” cases. The authors also point out that they just said “used drugs” or “didn’t” rather than measuring how much or how often.

Melberg, H.O., Jones, A.M. and Bretteville-Jensen, A.L. (2009) “Is cannabis a gateway to hard drugs?,” Empirical Economics, 38(3), pp. 583–603. https://doi.org/10.1007/s00181-009-0280-z  

About the source:

• Peer reviewed? Yes, published in Empirical Economics.
• Methodology: Uses postal survey data of 21 to 26 year olds in Oslo, combined with drug price data from the Norwegian Institute for Alcohol and Drug Research. The researchers aimed to identify the gateway effect using a bivariate survival model with a measure of shared “frailty,” meaning factors that make people more prone to both hard and soft drug use.
• Sample size: 1,984
• Main results: The results suggest that there are two distinct groups, a smaller group of “troubled” youths (24% of the sample) where cannabis roughly doubles the risk of progression to other drugs, and a larger group where it has little impact on risk. They also note that including price changes in the analysis is crucial for accurate modeling.
• Other notes: Splitting the sample into two groups was ultimately an assumption, and although it may make sense, it’s certainly important to keep in mind.

Nkansah-Amankra, S. and Minelli, M. (2016) “‘Gateway hypothesis’ and early drug use: Additional findings from tracking a population-based sample of adolescents to adulthood,” Preventive Medicine Reports, 4, pp. 134–141. https://doi.org/10.1016/j.pmedr.2016.05.003

About the source:

• Peer reviewed? Yes, published in Preventative Medicine Reports.
• Methodology: Researchers used data from the Add Health survey, a national longitudinal survey on adolescent health. They focused on participants who were interviewed in 1994-1995 and on the three subsequent waves, up to 2007-2008. Tobacco, alcohol and marijuana use were measured at wave 1, and other drug use was considered for subsequent waves.
• Sample size: 11,194
• Main results: Using “gateway drugs” in early adolescence was associated with further drug use in older adolescence, but the relationships weren’t consistent into adulthood. However, higher depressive symptoms, for instance, were consistently associated with subsequent drug use.
• Other notes: The authors write, “Relationships between early drug use and later drug use in adulthood cannot be solely explained by the gateway hypothesis.” Note that this study doesn’t discount the gateway hypothesis playing a role either.

Prince van Leeuwen, A.L, et. al. (2011). “Can the gateway hypothesis, the common liability model and/or, the route of administration model predict initiation of cannabis use during adolescence? A survival analysisthe TRAILS study.” Journal of Adolescent Health, 48(1), 73–78. https://doi.org/10.1016/j.jadohealth.2010.05.008

About the source:

• Peer reviewed? Yes, published in the Journal of Adolescent Health.
• Methodology: Used data on teens who completed three waves of the TRacking Adolescents’ Individual Lives Survey (TRIALS), to compare the gateway hypothesis, the common liability model and the route of administration model. The study focused on the alcohol/tobacco to cannabis gateway.
• Sample size: 2,113
• Main results: The researchers found that tobacco use doesn’t increase the risk of using cannabis more than alcohol, so the route of initiation model was discarded. The study also found that those who used both alcohol and tobacco were more likely than those who used either individually, which the gateway hypothesis is “not broad enough” to explain.
• Other notes: This doesn’t really investigate cannabis as a gateway (since alcohol and tobacco were the “gateway” substances addressed), but still addresses the overall model.

Reddon, H. et al. (2018) “Cannabis use is associated with lower rates of initiation of injection drug use among street-involved youth: A longitudinal analysis,” Drug and Alcohol Review, 37(3), pp. 421–428. https://doi.org/10.1111/dar.12667

About the source:

• Peer-reviewed? Yes, published in Drug and Alcohol Review
• Methodology: Used data from the At-Risk Youth survey, from 2005 to 2015, for participants who reported that they’d used illicit drugs but had never injected.
• Sample size: 481
• Main results: Daily cannabis use was associated with lower initiation into injecting drug use, with 34% less chance of starting to inject for the weed users. They performed a sub-analysis that showed that cannabis users were less likely to start injecting stimulants, but with no difference for opioids.
• Other notes: The teens studied were identified as “at risk,” and so the group doesn’t generalize to the overall population. They also lacked information about their first use of cannabis or other drugs, which could be relevant.

Renard, J., Rushlow, W.J. and Laviolette, S.R. (2016) “What can rats tell us about adolescent cannabis exposure? Insights from Preclinical Research,” The Canadian Journal of Psychiatry, 61(6), pp. 328–334. https://doi.org/10.1177/0706743716645288

About the source:

• Peer reviewed? Yes, published in The Canadian Journal of Psychiatry.
• Methodology: Review of numerous studies into adolescent cannabis exposure using rat models, with one section focusing on the gateway hypothesis.
• Main results: The research suggests that regular THC exposure as an adolescent could create a heightened sensitivity to opiates in adulthood. There are some complexities to other results (e.g. for cocaine, males showed no effect) but overall most studies agree.  
• Other notes:  There are many gaps in knowledge on this topic, with limited understanding of the mechanisms and how it affects other drugs of abuse. As always, the results of animal studies should be treated cautiously.

Secades-Villa, R. et al. (2015) “Probability and predictors of the Cannabis Gateway Effect: A National Study,” International Journal of Drug Policy, 26(2), pp. 135–142. https://doi.org/10.1016/j.drugpo.2014.07.011

About the source:

• Peer reviewed? Yes, published in the International Journal of Drug Policy
• Methodology: Tested the gateway hypothesis using data from the National Epidemiological Survey on Alcohol and Related Conditions. They focused on participants who started using cannabis before any other drug.
• Sample size: 6,624
• Main results: 45% of participants with lifetime cannabis use progressed to other illicit drug use.
• Other notes: The presence of a “progression” is not evidence that cannabis caused the progression. In short, this does not refute the alternative hypothesis. However, it does confirm the existence of the progression.

Stringer, S. et al. (2016) “Genome-wide association study of Lifetime Cannabis use based on a large meta-analytic sample of 32330 subjects from the International Cannabis Consortium,” Translational Psychiatry, 6(3).  https://doi.org/10.1038/tp.2016.36

About the source:

• Peer reviewed? Yes, published in Translational Psychiatry.
• Methodology: Genome-wide association study, looking for genetic variants associated with cannabis use.
• Sample size: 32,330 from cohorts and 5,627 from replication samples.
• Main results: NCAM1, CADM2, SCOC and KCNT2 were associated with lifetime cannabis use. NCAM1 is also associated with cigarette smoking and other substance use risk. They also found a strong genetic correlation between cannabis use and lifetime cigarette smoking.
• Other notes: The authors note that their study didn’t have sufficient power to detect very small effects, and also that cannabis use was merely “used” or “didn’t,” with no information about frequency of use. Both of these could have obscured potential findings.

Taylor, M. et al. (2017) “Patterns of cannabis use during adolescence and their association with harmful substance use behaviour: Findings from a UK birth cohort,” Journal of Epidemiology and Community Health, 71(8), pp. 764–770. https://doi.org/10.1136/jech-2016-208503

About the source:

• Peer-reviewed? Yes, published in the Journal of Epidemiology and Community Health.
• Methodology: UK population-based birth cohort, with information on cannabis use collected each year from age 13 to 18. A postal questionnaire assessed other drug use at age 21.
• Sample size: 5,315
• Main results: All cannabis users were more likely to be dependent on nicotine, drink alcohol harmfully and use other illicit drugs compared to non-users. Regular cannabis users were around 26 times more likely to have used illicit drugs by age 21, in the fully adjusted model.
• Other notes:  It’s possible that some of the results – for example, the impact of cannabis on the likelihood of harmful drinking – are impacted by participants starting using substances at the same time (e.g. starting to drink at the same time as starting using cannabis). Authors point out that the results are consistent with both the gateway hypothesis and the common liability hypothesis.

Vanyukov, M.M. et al. (2009) “Measurement of the risk for substance use disorders: Phenotypic and genetic analysis of an index of common liability,” Behavior Genetics, 39(3), pp. 233–244. https://doi.org/10.1007/s10519-009-9269-9

About the source:

• Peer reviewed? Yes, published in Behavior Genetics.
• Methodology: Recruited two groups, one from the children of male participants in a substance abuse survey, and another group of same-sex twins. They used these to test an index for liability to drug addiction (i.e. a “common liability” framework).
• Sample size: 500 for the first group, and 232 twin pairs (21% of whom were identical).
• Main results: They found that their index is a valid and predictive measure, and is highly heritable (at least in males). It was more correlated in the higher-risk groups than the lower ones, better for predicting marijuana use than other drug use or alcohol use (especially in low risk groups). The index scores of identical twins were also greater than that of non-identical (fraternal) twins.
• Other notes: The authors note that the youth study concluded before the age at which their parents developed a substance use disorder, so some important data may have been missed.  

Vanyukov, M.M. et al. (2012) “Common liability to addiction and ‘Gateway hypothesis’: Theoretical, empirical and evolutionary perspective,” Drug and Alcohol Dependence, 123. https://doi.org/10.1016/j.drugalcdep.2011.12.018  

About the source:

• Peer reviewed? Yes, published in Drug and Alcohol Dependence.
• Methodology: A literature review of studies relevant to the gateway hypothesis and common liability to addiction model.
• Main results: The “sequence” of drug use proposed by the gateway hypothesis is variable and determined by availability, while common liability to addiction is supported by data and genetic theory.
• Other notes: Paper mainly describes the evidence for common liability, with only a comparatively short section on the gateway effect and its evidence.  

Yamaguchi, K. and Kandel, D.B. (1984) “Patterns of drug use from adolescence to young adulthood: II. sequences of progression.,” American Journal of Public Health, 74(7), pp. 668–672. https://doi.org/10.2105/ajph.74.7.668

About the source:

• Peer reviewed? Yes, published in the American Journal of Public Health.
• Methodology: Follow-up interview of the original cohort in Kandel’s landmark “gateway” study.
• Sample size: 1,325
• Main results: The findings largely corroborate Kandel’s earlier results, with alcohol, cigarettes and marijuana preceding use of other illicit drugs. They also found that cigarettes can come before marijuana for women, even without alcohol, but alcohol consistently precedes marijuana use in men.
• Other notes: Again, the fact that the use of one drug precedes use of the other doesn’t mean there is a causal link. As they write: “use of a drug at a particular stage does not invariably lead to the use of other drugs higher up in the sequence. Many youths stop at a particular stage and do not progress further.”