Adverse Health Effects of Recreational Cannabis Use: A Review

What has research over the past two decades revealed
about the adverse health effects of recreational
cannabis use?
Wayne Hall 1,2,3
The University of Queensland Centre for Youth Substance Abuse Research and The UQ Centre for Clinical Research, Herston, Australia, 1 The National Addiction
Centre, Kings College London, London, UK 2 and National Drug and Alcohol Research Centre, University of New South Wales, New South Wales, Australia 3
ABSTRACT
Aims To examine changes in the evidence on the adverse health effects of cannabis since 1993. Methods A
comparison of the evidence in 1993 with the evidence and interpretation of the same health outcomes in 2013.
Results Research in the past 20 years has shown that driving while cannabis-impaired approximately doubles car
crash risk and that around one in 10 regular cannabis users develop dependence. Regular cannabis use in adolescence
approximately doubles the risks of early school-leaving and of cognitive impairment and psychoses in adulthood.
Regular cannabis use in adolescence is also associated strongly with the use of other illicit drugs. These associations
persist after controlling for plausible confounding variables in longitudinal studies. This suggests that cannabis use is
a contributory cause of these outcomes but some researchers still argue that these relationships are explained by
shared causes or risk factors. Cannabis smoking probably increases cardiovascular disease risk in middle-aged adults
but its effects on respiratory function and respiratory cancer remain unclear, because most cannabis smokers have
smoked or still smoke tobacco. Conclusions The epidemiological literature in the past 20 years shows that cannabis
use increases the risk of accidents and can produce dependence, and that there are consistent associations between
regular cannabis use and poor psychosocial outcomes and mental health in adulthood.
Keywords Cannabis, dependence, drug-related harms, epidemiology, health risks, mental health.
Correspondence to: Wayne Hall, The University of Queensland Centre for Youth Substance Abuse Research, Herston 4006, Australia. E-mail: w.hall@
uq.edu.au
Submitted 6 April 2014; initial review completed 21 May 2014; final version accepted 4 August 2014
WHY ARE WE CONCERNED ABOUT
RECREATIONAL CANNABIS USE?
During the past half-century, recreational cannabis use
has become almost as common as tobacco use among
adolescents and young adults. Since its use was first
reported more than 40 years ago in the United States,
recreational cannabis use has spread globally to other
developed countries and, more recently, low- and middle-
income countries [1,2].
The effects sought by cannabis users—euphoria and
increased sociability—seem to be produced primarily
by delta-9-tetrahydrocannabinol (THC) [3]. These
effects may be modulated by cannabidiol (CBD), a non-
psychoactive cannabinoid found in many cannabis prod-
ucts [3]. THC content is highest in the flowering tops of
the female cannabis plant. During the past 30 years the
THC content of cannabis has increased in the United
States from <2% in 1980 to 8.5% in 2006 [4]. THC
content has also increased in the Netherlands and prob-
ably in other developed countries [5].
Cannabis is usually smoked in a ‘joint’ or with a water
pipe (sometimes with tobacco added) because smoking is
the most efficient way to achieve the desired psychoactive
effects [3]. A dose of 2–3 mg of THC will produce a ‘high’
in occasional users who typically share a single joint with
others. Regular users may smoke up to three to five joints
of potent cannabis a day [6].
Paper presented at Through the Maze: Cannabis and Health International Drug Policy Symposium Auckland, New Zealand,
November 2013.
MONOGRAPH
bs_bs_banner
doi:10.1111/add.12703
© 2014 Society for the Study of Addiction Addiction

In epidemiological studies, ‘heavy’ or ‘regular’ canna-
bis use is usually defined as daily or near-daily use [6].
This pattern, when continued over years and decades,
predicts increased risk of many of the adverse health
effects attributed to cannabis that are reviewed below [6].
Unless stated otherwise, the remainder of this paper deals
with the adverse effects of cannabis smoking, especially
the adverse health effects of regular, typically daily, can-
nabis smoking.
OUR APPROACH TO THE LITERATURE
IN 1993
In 1993 there were very few epidemiological studies of the
health effects of cannabis. The literature was dominated
by (i) animal studies from the 1970s on the toxicity,
teratogenicity and carcinogenicity of cannabis and THC;
and (ii) human laboratory studies from the late 1970s and
early 1980s on the effects of sustained cannabis use over
7–35 days on the health of college students. There was a
small number of clinical studies of adverse health effects
in heavy cannabis users from the same period [7,8].
In the early 1990s in Australia (as elsewhere) there
were strongly polarized views on the health effects of
cannabis. The published appraisals of the limited evi-
dence were refracted through the prism of the appraisers’
preferred policies towards cannabis (decriminalization or
legalization of personal use versus intensified public edu-
cation and law enforcement campaigns to discourage
use). We adopted the following approaches to maximize
the chances that our review would be seen as credible by
advocates of these very different competing public poli-
cies towards cannabis use.
First, Nadia Solowij, Jim Lemon and I applied the
standard rules for making causal inferences about the
health effects of any drug to cannabis. That is, we looked
for: (i) epidemiological evidence of an association
between cannabis use and the health outcome in case–
control and prospective studies; (ii) evidence that reverse
causation was an implausible explanation (e.g. evidence
from prospective studies that cannabis use preceded the
outcome); (iii) evidence from prospective studies that had
controlled for potential confounding variables (such as
other drug use and characteristics on which cannabis
users differed from non-users); and (iv) clinical and
experimental evidence which supported the biological
plausibility of a causal relationship [9].
Secondly, we specified the standard of proof that we
would use in inferring that cannabis was a probable
cause of an adverse health effect; namely, evidence that
made it more likely than not that cannabis was a cause
of the adverse health effect. As we pointed out, very few
conclusions could be drawn if we demanded proof
beyond reasonable doubt. We also identified possible
adverse health effects that required further investigation,
e.g. if animal and/or human evidence indicated an asso-
ciation between cannabis use and an adverse health effect
which was biologically plausible.
Thirdly, we were prepared to infer that cannabis
could have adverse health effects when it: shared a route
of administration with cigarette smoking, e.g. respira-
tory disease, or produced similar acute effects to those
of alcohol, e.g. on driving and crash risk; and had
similar pharmacological effects to other long-acting
central nervous system (CNS) depressant drugs, e.g.
benzodiazepines.
Fourthly, we compared the probable adverse health
effects of cannabis with the known adverse health effects
of alcohol and tobacco. We aimed to do so in a way that
used the same evidential standards in drawing causal
inferences about the probable adverse health effects of all
three drugs.
In the following analysis I apply these criteria to the
more substantial research evidence that has accumu-
lated over the past 20 years on the adverse health effects
of cannabis. For each type of adverse health effect, I
(i) briefly summarize the conclusions drawn in 1993;
(ii) explain the reasons given for these conclusions; and
(iii) compare the conclusions reached in 1993 with the
inferences that may reasonably be drawn in 2013. The
review begins with acute adverse health effects, those
that may arise from a single episode of intoxication. It
then considers the adverse health and psychological
effects of regular cannabis use over periods of years and
decades.
ADVERSE ACUTE HEALTH EFFECTS
In 1993 the evidence indicated that the risk of a fatal
overdose from using cannabis was extremely small. This
remains an uncontroversial conclusion, because the dose
of THC that kills rodents is extremely high. The estimated
fatal dose in humans derived from animal studies is
between 15 [10] and 70 g [3]. This is a far greater
amount of cannabis that even a very heavy cannabis user
could use in a day [10]. There are also no reports of fatal
overdoses in the epidemiological literature [11]. There
have been case reports of cardiovascular fatalities in
seemingly otherwise healthy young men after smoking
cannabis [12] that are discussed below under ‘Cardiovas-
cular effects’ of cannabis smoking.
In 1993 we identified the following adverse acute
effects of cannabis use: (i) unpleasant experiences such as
anxiety, dysphoria and paranoia, especially among naive
users; (ii) cognitive impairment, especially of attention
and memory; (iii) psychomotor impairment that could
impair a person’s ability to drive a motor vehicle while
intoxicated; (iv) an increased risk of psychotic symptoms
2 Wayne Hall
© 2014 Society for the Study of Addiction Addiction

in high doses, especially among those with a personal or
family history of psychosis; and (v) an increased risk of
low birth weight babies, if cannabis was used during
pregnancy.
The acute adverse effects of anxiety, panic reactions
and psychotic symptoms continue to be reported, espe-
cially by naive users [6]. During the past decade there has
been an increase in the number of attendances at hospi-
tal emergency rooms in the United States in which can-
nabis is ‘mentioned’ [13]. This could reflect an increase in
acute adverse effects in naive users as the average THC
content of cannabis products has risen, an issue that is
discussed further below.
Car crash injuries and deaths
In 1993 it was clear from laboratory studies that canna-
bis and THC produced dose-related impairments in
reaction-time, information-processing, perceptual-motor
coordination, motor performance, attention and tracking
behaviour. This suggested that cannabis could potentially
cause car crashes if users drove while intoxicated, but it
was unclear whether in fact cannabis use did so. Studies
in driving simulators suggested that cannabis-impaired
drivers were aware of their impairment and compensated
for these effects by slowing down and taking fewer risks.
There were similar findings in the few studies on the
effects of cannabis use on driving on the road (see [14] for
a review).
In 1993 there were major problems in interpreting
the few epidemiological studies of cannabis use in fatal
car crashes. Most reported on cannabis metabolites,
which indicated only that cannabis had been used in the
days before the accident; they did not show that the
drivers were cannabis-impaired at the time of the acci-
dent. Moreover, in many of these studies a substantial
proportion of drivers with cannabis in their blood also
had high blood alcohol levels, making it difficult to distin-
guish between the effects of cannabis and alcohol on acci-
dent risk [9].
In the past decade, better-designed epidemiological
studies have found that cannabis users who drive while
intoxicated approximately double their risk of a car crash.
Gerberich et al. [15], for example, found that cannabis
users had higher rates of hospitalization for injury from
all causes than former cannabis users or non-users in
64 657 patients from a Health Maintenance Organiza-
tion (HMO). The relative risk (RR) of motor vehicle acci-
dents (RR = 1.96) persisted after statistical adjustment
for confounding in men. Mura et al. [16] found a similar
relationship in a case–control study of THC in the serum
of 900 people hospitalized in France with motor vehicle
injuries and 900 age- and sex-matched controls admitted
to the same hospitals for reasons other than trauma.
A meta-analysis of nine case–control and culpability
studies [17] found that recent cannabis use (indicated
by THC in blood or self-reported cannabis use) doubled
the risk of a car crash [odds ratio (OR) = 1.92 95% con-
fidence interval (CI) = 1.35, 2.73]. The risk was margin-
ally higher in: better-designed studies (2.21 versus 1.78),
in case–control rather than driver culpability studies
(2.79 versus 1.65) and in studies of fatalities rather than
injuries (2.10 versus 1.74). Very similar results were
reported in another meta-analysis [18] (pooled risk of
2.66) and in a systematic review of laboratory and epi-
demiological studies [19].
In summary, the epidemiological and laboratory evi-
dence on the acute effects of cannabis suggests strongly
that cannabis users who drive while intoxicated increase
their risk of motor vehicle crashes 2–3 times [20] as
against 6–15 times for comparable intoxicating doses of
alcohol. Cannabis use was estimated to account for 2.5%
of traffic deaths in France as against 29% for alcohol. The
risk of an accident increases substantially if cannabis
users also have elevated blood alcohol levels [19].
Reproductive effects of cannabis use
Fetal development and birth defects
In 1993 animal studies suggested that high doses of can-
nabis extract caused growth retardation and birth mal-
formations [21], but epidemiological studies did not
consistently find an increased risk of birth defects among
women who reported using cannabis during pregnancy.
It was also difficult to interpret the few studies that
reported increased rates of birth defects (e.g. [22]),
because cannabis users were more likely to smoke
tobacco and use alcohol and other illicit drugs during
pregnancy [23]. They were also less likely to seek antena-
tal care and had poorer nutrition than women who did
not use cannabis [24]. Zuckerman et al. [25] reported the
most convincing failure to find an increased risk of birth
defects in a study of a large sample of women among
whom there was a substantial rate of cannabis use that
was measured by urinalysis rather than self-report.
A meta-analysis [26] of studies in the 1980s and
1990s suggested that regular cannabis use during preg-
nancy reduced birth weight, although the effect was
smaller than that for tobacco smoking. Several large epi-
demiological studies have since reported that cannabis
use in pregnancy is associated with reduced birth
weight (e.g. [27,28]). This effect has generally persisted
after controlling statistically for other drug use (e.g.
[25,28,29]). Several of these studies also reported that
women who used cannabis had a shorter duration of
labour and an increased risk of babies small for gesta-
tional age [27].
Cannabis health effects 3
© 2014 Society for the Study of Addiction Addiction

These studies have a number of limitations. First, self-
reported rates of cannabis use during pregnancy are typi-
cally low (2–6%). Studies that have measured cannabis
use using urinalyses suggest that there is considerable
under-reporting of use, which probably attenuates asso-
ciations between cannabis use and poor birth outcomes.
Secondly, it has often been difficult to fully adjust for the
effects of major confounders such as cigarette smoking
in analyses of the effects of cannabis use on birth weight.
None the less, there is a good case on the grounds of
prudence for recommending that women should avoid
using cannabis while pregnant, or while attempting to
become pregnant.
Postnatal effects of maternal cannabis use
In 1993 a small number of studies reported increased
rates of developmental abnormalities in children born to
women who used cannabis during pregnancy, such as
developmental delays in the visual system and increased
tremor and startle shortly after birth [30]. These effects
were not reported consistently in later assessments; e.g.
some were not detected at the age of 1 month or on
ability tests at 6 and 12 months. Others were reported at
36 and 48 months, but not at 60 and 72 months [30]. As
these children entered adolescence, maternal cannabis
was associated with poorer cognitive performance. In the
Ontario study, at age 12 years, there were no differences
in full-scale IQ scores between children who were and
were not exposed to cannabis, but there were differences
in perceptual organization and higher cognitive processes
[30]. Tennes et al. [24], by contrast, found no IQ differ-
ences at 1 year between the children of users and nonus-
ers in 756 women, a third of whom used cannabis during
pregnancy.
In the past 20 years another cohort of low-income
women with higher rates of regular cannabis use [31]
has reported lower scores on memory and verbal scales of
the Stanford–Binet Intelligence Scale at age 3 in children
born to 655 low-income women (half African American
and half Caucasian) in Pittsburgh between 1990 and
1995. By age 10, maternal cannabis use at all stages of
pregnancy was associated with delinquency and problem
behaviour [32]. Cannabis-exposed children also per-
formed more poorly on reading and spelling tests and
were rated lower on academic achievement by their
teachers [33]. These findings were confirmed at age 14,
when the association between prenatal cannabis use and
poorer school performance was shown to be mediated
by the child’s lower cognitive ability, higher rates of
attentional and mood disorders and by these children ini-
tiating cannabis use before the age of 14 [34].
The behavioural effects of prenatal cannabis exposure
have been reported in only two cohort studies, and the
effects have been most consistent in the cohort of lower-
income women with higher rates of use [35]. The dose–
response relationship in one of these studies is suggestive
of a causal role for cannabis. Uncertainty remains
because of the small number of studies, the small samples
of women in each and the researchers’ limited ability to
control for the confounding effects of other drug use
during pregnancy, maternal drug use post-birth and
poor parenting. These studies have also been unable to
control for a plausible explanation of some of the effects
of maternal cannabis use, namely, genetic differences in
IQ and in the risks of conduct and substance use disor-
ders between cannabis-using mothers and their non-
using peers [35]. None the less, as with the evidence on
birth weight, it is prudent to counsel women against
using cannabis during pregnancy.
ADVERSE HEALTH EFFECTS OF
CHRONIC CANNABIS USE
Epidemiological studies of cannabis use are usually
unable to measure the doses of THC and other
cannabinoids (e.g. cannabidiol) that regular cannabis
users receive [36]. In the absence of these data, epidemio-
logical studies have defined ‘heavy’ or ‘regular’ cannabis
use as daily or near-daily use [6]. This is the pattern of use
that has been associated most consistently with adverse
health and psychological outcomes.
A major challenge in interpreting associations
between regular cannabis use and adverse health out-
comes in epidemiological studies is that regular cannabis
users differ from non-users in a variety of ways that may
reflect baseline differences in their risks of adverse out-
comes. Regular cannabis users, for example, are more
likely to use alcohol, tobacco and other illicit drugs, and
they differ from non-users in their risk-taking and other
behaviour [6]. Statistical methods of control have been
used to test the plausibility of confounding as an expla-
nation of these relationships and fixed-effects regression
has been used to test for unknown fixed differences
between users and non-users (e.g. [37]). Some research-
ers have expressed doubts about whether the first strategy
can be wholly successful [38].
Cannabis dependence
The conclusions of our 1993 review on cannabis depend-
ence provoked some scepticism. We used the DSM-III defi-
nition of cannabis dependence that included impaired
control over cannabis use and difficulty ceasing use
despite harms caused by it. DSM-III cannabis abuse
and/or dependence had been the most common type of
illicit substance use disorder identified in US mental
health surveys of the 1980s and 1990s [9]. Critics of this
epidemiological evidence argued that very few cannabis
4 Wayne Hall
© 2014 Society for the Study of Addiction Addiction