This document contains text automatically extracted from a PDF or image file.

Formatting may have

been lost and not all text may have been recognized.
To remove this note, right-click and select "Delete table".
Addiction (1999) 94(8), 1155 1164
RESEARCH REPORT
Cannabis abuse and serious suicide attempts
ANNETTE ROGER T. L. MULDER
BEAUTRAIS,
2
1
PETER R. JOYCE
2
&
1Canterbury Suicide Project & 2Department of Psychological Medicine, Christchurch School of
Medicine, Christchurch, New Zealand
Abstract
Aims. To compare the relationship between cannabis abuse/dependence and risk of medically serious suicide
attempts in individuals making serious suicide attempts and randomly selected comparison subjects. Design.
Case control comparison. Setting. Cases, a general hospital; controls, the local community. Participants.
Cases were 302 consecutive individuals making medically serious suicide attempts; 1028 randomly selected
control subjects. Measurements. DSM-III-R mental disorder diagnoses; measures of socio-demographic
characteristics and childhood and family experiences. Findings. Of those making serious suicide attempts,
16.2% met DSM-III-R criteria for cannabis abuse/dependence at the time of the attempt, compared with
1.9% of comparison subjects (OR 10.3; 95%CI, 5.95 17.8, p 0.0001). Risks of serious suicide attempt
were signi cantly related to a series of socio-demographic and childhood characteristics, and to mental
disorders that were co-morbid with cannabis abuse/dependence. When the association between cannabis
abuse/dependence and suicide attempt risk was controlled for socio-demographic factors, childhood factors and
concurrent psychiatric morbidity, there was a marginally signi cant association (OR 2.0; 95%CI,
0.97 5.3, p 0.06) between cannabis abuse/dependence and serious suicide attempt risk. Conclusions.
These results suggested that much of the association between cannabis abuse/dependence and suicide attempt
risk arose because: (a) individuals who develop cannabis abuse/dependency tend to come from disadvantaged
socio-demographic and childhood backgrounds which, independently of cannabis abuse, are associated with
higher risk of suicide attempt, or (b) because cannabis abuse/dependence is co-morbid with other mental
disorders which are independently associated with suicidal behaviour. Nevertheless, the possibility remains
that cannabis abuse/dependence may make an independent contribution to risk of serious suicide attempt, both
directly and through the possible effects of cannabis abuse on risk of other mental disorders.
Introduction
The health risks of cannabis use are current
issues of concern (Addiction Research Foun-
dation/World Health Organization, 1981; Insti-
tute of Medicine, 1982; National Alcohol and
Drug Research Center, 1994; Public Health
Commission, 1996) and it has been suggested
that increasing substance use (including can-
nabis use) in the last three decades may have
contributed to rises in suicidal behaviour which
have occurred during this time (Fowler, Rich &
Young, 1986; Murphy, 1988; Rich, Fowler &
Young, 1989; Crumley, 1990; Carlson et al.,
1991; Felts, Chenier & Barnes, et al. 1992;
Correspondence to: Dr A.L. Beautrais, Canterbury Suicide Project, Christchurch School of Medicine, P O Box
4345, Christchurch, New Zealand. Tel: 00 64 3 3720 408; Fax: 00 64 3 3720 405; e-mail: suicide@chmeds.ac.nz
Submitted 31st August 1998; initial review completed 18th November 1998; trial version accepted 5th January
1999
ISSN 0965 2140 print/ISSN 1360-0443 online/99/081155 10 Society for the Study of Addiction to Alcohol and Other
Drugs
Carfax Publishing, Taylor & Francis Limited
1156 Annette L. Beautrais et al.
Kaminer, 1992; Hawton, 1994; Shaffer, et al.,
1996; Neeleman & Farrell, 1997). While associa-
tions between suicide risk and substance abuse
have been reported in many studies (e.g. Fowler
et al., 1986; Rich et al., 1989; Crumley, 1990;
Runeson, 1990; Marttunen et al., 1991; Felts et
al., 1992; 1992; Kaminer, 1992; Bukstein et al.,
1993; Hawton et al., 1993; Lesage, et al., 1994;
Shaffer et al., 1996), there appear to be no
studies which have examined the speci c rela-
tionships between cannabis abuse/dependence
and suicidal behaviour. None the less, it seems
reasonable to conjecture, on the basis of associa-
tions between substance abuse and suicide risk,
that rates of suicide and serious suicide attempt
are elevated among individuals with cannabis
abuse/dependence.
There are at least three explanations of any
linkages between cannabis abuse/dependence
and risk of serious suicide attempt. First, can-
nabis abuse/dependence may make a direct
causal contribution to suicide attempt risk. Sec-
ondly, any association between cannabis abuse/
dependence and risk of suicide attempt may be
accounted for by common risk factors which
lead, independently, to both cannabis abuse/
dependence and to risk of suicide attempt. There
is evidence to support this hypothesis since the
literature on suicide attempt risk and cannabis
abuse suggests that both outcomes may be
in¯ uenced by a series of common risk factors
including, for example, social disadvantage, poor
commitment to academic achievement, low reli-
giosity, less in¯ uence of parents relative to peers
and family history of alcohol abuse/dependence
(Jessor, Chase & Donovan, 1980; Newcomb
& Bentler, 1989; Pfeffer, 1989; Hawkins,
Catalano & Miller, 1992; Rutter & Smith, 1995;
Fergusson, Lynskey & Horwood, 1996).
Finally, associations between cannabis abuse/
dependence and suicide attempt risk may arise
from the co-morbidity of cannabis abuse/
dependence with other mental disorders, includ-
ing alcohol abuse/dependence, substance abuse/
dependence (other than alcohol and cannabis),
mood disorders and antisocial behaviours, which
are associated with suicide attempt risk.
This paper presents an analysis of the relation-
ships between cannabis abuse/dependence and
risk of serious suicide attempt, using data col-
lected as part of a large case control study of
suicidal behaviour (Beautrais et al., 1996). The
aims of the present analysis were:

(1) To estimate the association between can-

nabis abuse/dependence and risk of serious
suicide attempt.
(2) To examine the extent to which any associ-
ation between cannabis abuse/dependence
and suicide attempt risk may be explained
by common socio-demographic and
childhood risk factors associated with both
outcomes.
(3) To estimate the extent to which cannabis
abuse/dependence is associated with in-
creased risk of suicide attempt when due
allowance is made for the co-morbidity be-
tween cannabis abuse/dependence and other
aspects of concurrent psychiatric disorder.
Method
The data were gathered as part of the Canter-
bury Suicide Project, which is a case control
study of people in the Canterbury (New
Zealand) region who have committed suicide
(202 cases), people who have made medically
serious suicide attempts (302 cases) and
randomly selected comparison subjects (1028
cases).
Cases
The subjects who attempted suicide were a con-
secutive series of 302 individuals who made
medically serious suicide attempts during the
period 1 September 1991 31 May 1994, in the
city of Christchurch, New Zealand. Christchurch
city has a population of 315 000 and the Canter-
bury region has a regional population (including
Christchurch city) of approximately 430 000.
A medically serious suicide attempt was one
that required hospital admission for longer than
24 hours and met one of the following treatment
criteria:
(1) Treatment in a specialized unit, i.e. the
intensive care unit (for drug overdose), the
hyperbaric unit (for carbon monoxide
poisoning), or the burn unit.
(2) Surgery under general anaesthesia, e.g. for
tendon repair, stabbing injuries.
(3) Medical treatment beyond gastric lavage,
activated charcoal or routine neurological
observations.
Individuals who attempted suicide by methods
with a high risk of fatality, such as hanging or
gunshot, and who were hospitalized for more
Cannabis abuse and serious suicide attempts 1157
than 24 hours but did not meet the preceding
treatment criteria were also included in the
group of persons with serious suicide attempts.
The cases were identi ed by daily calls to the
emergency department, the psychiatric emer-
gency service and relevant admitting wards.
Daily checks were conducted to ensure
identi cation of eligible cases. Individuals who
met criteria for inclusion in the study were inter-
viewed in the hospital when medical and psychi-
atric staff considered it appropriate for them to
be seen. Most interviews were conducted in the
hospital, and the remainder were conducted in
the subjects’ homes or in the psychiatric hospital
to which some patients were transferred immedi-
ately after medical treatment. Each subject
was interviewed soon after the attempt to
minimize possible intervention of various factors,
including contact with family or friends and
commencement of psychiatric treatment.
Case identi cation was con ned to medically
serious suicide attempts because:
(1) from a public health perspective, medically
serious suicide attempts are a major source
of morbidity and health costs, and
(2) People who make medically serious suicide
attempts can be interviewed personally and
can be assumed to most resemble those who
die by suicide, about whom information can
be obtained only from family and friends.
Individuals who made more than one suicide
attempt who met criteria for inclusion in the
study were interviewed for the initial attempt
only.
In total, 317 individuals made serious suicide
attempts during the study period and 302 of
these participated in the study, giving a response
rate of 95.3%. Thirteen people (4.1%) refused
involvement, and two (0.6%) were not inter-
viewed because of language dif culties.
Comparison group
Comparison subjects were selected from elec-
toral rolls for the Canterbury region. An age and
gender-strati ed sample was obtained; the six
age strata were 18 24 years, 25 29 years, 30 39
years, 40 49 years, 50 59 years, 60 years and
older. The number of subjects in each age-and-
gender stratum was proportional to the known
age-by-gender distribution of the population
aged 18 years and over. In total, 1200 subjects
were selected for the comparison group and
1028 participated in the study. Ninety-three
(7.8%) refused involvement, 57 (4.8%) could
not be traced and 22 (1.8%) were unable to
adequately complete the interview because of
intellectual limitations, illness or language
problems. The estimated response rate for the
comparison subjects was therefore 85.7%.
National Electoral Roll estimates suggested
95.5% of the eligible population were enrolled
on Canterbury electoral rolls during the study
period. Each subject selected from the electoral
rolls was mailed a letter of introduction explain-
ing the study. The study interviewer then called
at the subject’s home, discussed the study and
arranged to return to the subject’s home to
conduct an interview at a convenient time.
The study was approved by the ethics commit-
tees of the Canterbury Area Health Board and
the Southern Regional Health Authority. Writ-
ten informed consent was obtained from all
study participants after the aims and procedures
of the study had been fully explained. For chil-
dren aged 16 and under, the written consent
of both the child and the parent/guardian was
obtained.
Data collection
Trained, experienced interviewers personally
conducted a semi-structured interview with each
subject in the study (suicide attempters and
comparison subjects) to construct retrospectively
a life history and to obtain information about
potential risk factors for suicide attempts. The
interview included a diagnostic section modi ed
from the Structured Clinical Interview for DSM-
III-R-patient version (Spitzer et al. 1988) to gen-
erate DSM-III-R (American Psychiatric
Association, 1987) diagnoses of selected mental
disorders. The eldwork was closely monitored,
and each interviewer met weekly with the super-
visor for debrie ng and for checking and editing
of each interview. For each person in the study a
parallel interview was conducted with a
ª signi cant otherº who knew the subject well
and was nominated by the subject. Information
gathered from the interviews with the subject
and the signi cant other was integrated in a
diagnostic conference (which always included
the principal investigator (ALB) and at least one
psychiatrist (PRJ or RTM) to produce, for each
subject, best-estimate diagnoses of mental
1158 Annette L. Beautrais et al.
disorders (according to DSM-III-R criteria) in
the month before the subject’s serious suicide
attempt. For conduct disorder and antisocial
personality disorder, a life-time history was ob-
tained. Multiple diagnoses on Axis I were per-
mitted. The test retest reliability of the
best-estimate diagnostic procedure was ascer-
tained by re-evaluation of a sample of 20% of all
subjects (both suicide attempters and compari-
son subjects). The test retest agreement was
high: the kappa coef cients (Fleiss, 1981) for the
principal diagnostic categories (mood disorders,
anxiety disorders, eating disorders, non-affective
psychosis, conduct disorder, antisocial personal-
ity disorder) ranged from 0.95 to 0.99.
The following measures were used in the pre-
sent analysis (these variables were selected from
a much larger dataset in preliminary analysis).
Cannabis abuse/dependence
Individuals who met DSM-III-R criteria for can-
nabis abuse or dependence during the month
prior to the suicide attempt (or interview) were
classi ed as having cannabis abuse/dependence.
Individuals who met DSM-III-R criteria for
polydrug dependence, who also met criteria for
cannabis abuse or dependence, were included in
the classi cation of those with cannabis abuse/
dependence.
Socio-economic status
Participant socio-economic status was measured
using the scale for socio-economic status in New
Zealand (Elley & Irving, 1976). This scale ranks
the population into six socio-economic groups
on the basis of occupation, with 1 denoting
professional or executive occupations and 6 de-
noting unskilled occupations. Subjects who were
receiving government social welfare bene ts and
students and dependent children were classi ed
as belonging to social class 6. To aid analysis, the
measure of socio-economic status was dichoto-
mized to give a measure of low socio-economic
status, de ned as membership of the lowest
socio-economic groups, classes 5 and 6.
No formal educational quali cations
Educational achievement was measured on a
four-point scale which classi ed participants

according to the highest educational quali cation

obtained: no formal educational quali cation; sec-
ondary school quali cation; trade or technical
quali cation; university degree or diploma. To aid
analysis, this measure was dichotomized by
de ning ª lack of formal quali cationsº as lack of
secondary school, tertiary or trade quali cations.
Poor parental relationship
Participants were asked to give a global rating on
a four point scale of their impression of the
quality of the martial relationship of the major
parent gures during the subject’s childhood.
For the present analysis, a dichotomous measure
of ª poor parental relationshipº was de ned if
subjects responded that their parents had got on
together ª not very wellº or ª very poorlyº .
Childhood sexual abuse
Subjects were asked whether or not, during
childhood, they had been ª physically or psycho-
logically forced by anyone to engage in any un-
wanted sexual activity, such as unwanted sexual
touching of his/her body or sexual intercourseº ,
a de nition previously used by Murphy (1985).
Subjects who responded positively to this ques-
tion were then asked further questions relating to
this activity (if they felt able to talk about the
matter). Subjects were classi ed as having a his-
tory of childhood sexual abuse if they responded
positively to the initial question, and subsequent
questioning established a history of childhood
sexual abuse.
In care during childhood
Subjects were asked if they had spent time, dur-
ing childhood, in a foster home, a children’s
home, or similar institution. Subjects were
classi ed as having been in care during child-
hood if they responded positively to this
question.
Parental alcohol problems
Subjects were asked whether they had experi-
enced, during childhood, additional family
dif culties. Subjects who responded positively to
this question were asked to describe the type of
family problems they had encountered. Subjects
were classi ed as having a history of parental
Cannabis abuse and serious suicide attempts 1159
alcohol problems during childhood if a positive
response was obtained to the initial question
and, if on further questioning, they speci ed
parental alcohol problems as a family problem.
Results
Bivariate relationship between cannabis abuse and
risk of serious suicide attempt
Individuals who made serious suicide attempts
had signi cantly higher rates of cannabis abuse/
dependence (N 49, 16.2%) than comparison
subjects (N 19, 1.9%) (odds ratio 10.3, 95%
con dence interval 5.95 17.8, p 0.0001).
The association between risk of serious suicide at-
tempt and socio-demographic and family experiences
Table 1 compares socio-demographic and child-
hood characteristics of those making serious sui-
cide attempts and comparison subjects. The
percentage of cases and comparison subjects ex-
posed to each factor is compared. The associ-
ation between suicide attempt status
(case/comparison) and each characteristic is
tested for signi cance with the
val 3.4 9.1); and being in care during child-
hood (odds ratio 6.7, 95% con dence
interval 3.9 11.3).
The associations between risk of serious suicide at-
tempt and concurrent mental disorder
Table 2 compares rates of DSM-III-R mental
disorders for those making serious suicide at-
tempts and comparison subjects. Risks of serious
suicide attempt were signi cantly (P 0.0001)
related to a range of mental disorders including
alcohol abuse/dependence (odds ratio 4.8,
95% con dence interval 3.5 6.70); other drug
(excluding alcohol and cannabis) abuse/depen-
dence (odds ratio 56.6, 95% con dence inter-
val 13.4 238.2); mood disorder (odds
ratio 46.8, 95% con dence interval 32.5
67.3); eating disorder (odds ratio 26.8, 95%
con dence interval 8.0 90.3); anxiety disorder
(odds ratio 5.8, 95% con dence inter-
val 3.9 8.5); and antisocial disorder (odds ra-
tio 9.5, 95% con dence interval 6.5 13.9).
2
test of indepen-
The joint relationship between cannabis abuse/depen-
dence. The table reports odds ratios for risk of
dence, socio-demographic and childhood characteris-
serious suicide attempt relative to a speci ed
tics and concurrent psychiatric morbidity
reference category. Because multiple compari-
Tables 1 and 2 show that, in addition to having
sons were used the Bonferroni-corrected
higher rates of cannabis abuse/dependence, indi-
signi cance level for this table, and Table 2, was
viduals who made serious suicide attempts
p 0.003. A number of socio-demographic and
tended to experience greater childhood and so-
childhood characteristics (including religious
cio-demographic disadvantage, and to have
af liation, cultural/ethnic identity, parental sep-
higher rates of psychiatric morbidity than those
aration, parental death, parental violent behav-
who did not make serious suicide attempts. The
iour and poor material and economic
higher rates of serious suicide attempts amongst
circumstances during childhood) were found to
those with cannabis abuse/dependence may be
be not signi cantly related to risk of serious
due to either their relatively disadvantaged back-
suicide attempt. In the interests of brevity, these
grounds or to their concurrent psychiatric mor-
factors were not included in Table 1.
bidity. To test these hypotheses, multiple logistic
Risks of serious suicide attempt were
regression methods were used.
signi cantly (p 0.0001) related to a series of
In the rst analysis, the association between
socio-demographic and childhood characteris-
cannabis abuse/dependence and risk of serious
tics, including low socio-economic status (odds
suicide attempt was adjusted for socio-demo-
ratio 6.6, 95% con dence interval 5.0 8.8);
graphic and childhood factors. Age and gender
lack of formal educational quali cations (odds
were included in this analysis as covariate factors
ratio 3.0, 95% con dence interval 2.3 3.9);
to take account of possible variation between
childhood sexual abuse (odds ratio 6.8, 95%
cases and comparison subjects. The results of
con dence interval 4.8 9.5); poor parental re-
this adjustment (Table 3) showed that statistical
lationship (odds ratio 4.7, 95% con dence in-
control for socio-demographic and childhood
terval 3.5 6.3); parental alcohol problems
factors substantially reduced the association be-
(odds ratio 5.6, 95% con dence inter-
tween cannabis abuse/dependence and risk of
1160 Annette L. Beautrais et al.
T

d
e

c
h

c
t

e
a

s
u

s
S

o
m

n
s

s
u

s
a

)
(

3
0

%
C

t
i

p
i

t
e

l
L

t
i

3
.

0
0

1
L

t
u

0
8

1
P

e
l

1
.

1
C

h
o

7
3

1
P

e
n

3
.

0
1
I

4
2

0
0

1
T

b
i

g
s

u
i

r
i

s
S

k
i

n
s

i
c

s
a

0
2

)
(

%
D

I
-

e
n

p
i

l
M

d
e

2
7

.
0

1
A

n
d

9
4

0
0

1
O

l
c

3
0

.
0

1
o

p
r

h
E

r
s

3
3

1
A

n
x

i
o

3
.

1
A

i
s

9
3

.
0

1
Cannabis abuse and serious suicide attempts 1161
Table 3. Cannabis abuse/dependence and risk of serious suicide attempt unadjusted and adjusted for the effects of
socio-demographic and childhood factors and for psychiatric co-morbidity
Outcome Odds 95%
ratio Con dence interval p
Unadjusted 10.3 5.95 17.8 0.0001
Adjusted for socio-demographic and childhood factors
1
3.2 1.7 6.0 0.0005
Adjusted for socio-demographic and childhood factors 2.0 0.97 5.3 0.06
and for psychiatric co-morbidity
2
1
Signi cant factors were: lack of formal educational quali cations; low socio-economic status; poor parental
relationship; childhood sexual abuse; parental alcohol problems; in care during childhood.
2
Signi cant factors were: lack of formal educational quali cations; low socio-economic status; childhood sexual
abuse, parental alcohol problems; mood disorder in prior month; substance disorder (alcohol or drug other than
cannabis) in prior month; antisocial disorder in life-time.
serious suicide attempt: Before adjustment for
these factors the odds ratio between cannabis
abuse/dependence and suicide attempt risk was
10.3 (95% con dence interval 5.95 17.8,
p 0.0001), whereas after adjustment this as-
sociation reduced to 3.2 (95% con dence
interval 1.7 6.0, p 0.0005).
To examine the role of psychiatric co-morbid-
ity in the association between cannabis abuse/de-
pendence and risk of serious suicide attempt, the
logistic regression analysis was extended to in-
clude comorbid measures of psychiatric disorder.
The results of this analysis (Table 3) showed that
when the association between cannabis abuse/
dependence and suicide attempt risk was con-
trolled for socio-demographic factors, childhood
factors and for psychiatric co-morbidity there
was a marginally signi cant association between
cannabis abuse/dependence and risk of serious
suicide attempt (odds ratio 2.0, 95%
con dence interval 0.97 5.3, p 0.06).
To examine the possibility that cannabis abuse
may act synergistically with psychiatric disorder
to potentiate risk of suicide attempt, the tted
model including psychiatric disorder was ex-
tended to include cannabis abuse/dependence by
psychiatric disorder interaction terms. No
signi cant interactions were found suggesting
that cannabis abuse and psychiatric disorder
combined additively to in¯ uence the log odds of
serious suicide attempt.
Parallel analysis using signi cant other data
A limitation of the preceding analysis is that
measures of childhood characteristics were based
on self-report, and these may be subject to recall
bias, i.e. it may be suggested that the apparently
strong association between suicide attempt risk
and psychosocial characteristics may re¯ ect a
reporting bias in which those making suicide
attempts may selectively report family and child-
hood circumstances. Within the constraints of
this design, complete control of recall bias was
not possible. However, some check on the possi-
bility of recall bias may be provided by compar-
ing the extent to which the ndings reported in
Tables 1 and 3 (based on measures obtained
from self-report data) may be replicated using
reports of childhood and family circumstances
provided from signi cant other data (see
ª Methodº ).
Using this approach, the data were reanalysed
using measures of childhood and family circum-
stances based on reports from signi cant others.
The analysis using signi cant other data pro-
duced conclusions which were very similar to
those reported in Table 3: after adjustment for
childhood circumstances and taking into account
concurrent psychiatric morbidity, the association
between cannabis abuse/dependence and risk
of suicide attempt was non-signi cant (odds
ratio 2.1, 95% con dence interval 0.85 5.0,
p 0.10). This nding reduces, although does
not eliminate, the possibility that the analysis
reported here was signi cantly contaminated by
recall bias.
The effects of age and gender on associations between
cannabis abuse/dependence and risk of serious suicide
attempt
The analysis thus far has examined the total
sample of cases and comparison subjects without
1162 Annette L. Beautrais et al.
considering subgroups de ned by age and gen-
der. Previous research (Fowler et al., 1986) has
suggested that the effects of substance abuse on
suicidal behaviour may be modi ed by factors
such as age. To examine the extent to which age
and gender modi ed the associations between
cannabis abuse/dependence and suicide attempt
risk after adjustment for other factors, the logis-
tic models reported in Table 3 were extended to
include interactions between (a) gender and can-
nabis abuse/dependence, and (b) age (under 30
years; 30 years and over) and cannabis abuse/de-
pendence. This analysis did not produce evi-
dence of interactions between gender and
cannabis abuse/dependence (log likelihood ratio
2
(LR
2
) 0.8, df 1, p 0.037) or between
gender and cannabis abuse/dependence
(LR 2 3.2, df 1, p 0.05). These ndings
suggest that the results in Table 3 apply gener-
ally to the sample being studied and did not vary
with age and gender.
Discussion
The results of the present analysis suggested that
individuals with cannabis abuse/dependence had
a higher risk of serious suicide attempt than
those without cannabis abuse/dependence. The
estimates from this study showed that those with
cannabis abuse/dependence had odds of serious
suicide attempt which were over 10 times higher
than those without such disorder.
However, further analysis showed that a very
substantial component of the association be-
tween cannabis abuse/dependence and risk of
serious suicide attempt arose because individuals
with cannabis abuse/dependence were a popu-
lation at higher risk of suicide attempt behaviour,
characterized by sociodemographic disadvantage
and disadvantageous childhood family circum-
stances. Before adjustment for sociodemographic
and family factors the odds ratio between can-
nabis abuse/dependence and serious suicide at-
tempt risk was 10.3, while after adjustment for
these factors the odds ratio reduced to 3.2.
Further analysis suggested that the remaining
association between cannabis abuse/dependence
and serious suicide attempt risk was largely ex-
plained by the co-morbidity of cannabis abuse/
dependence with other mental disorders
(notably, mood disorder; substance use disorder
other than cannabis abuse/dependence; anti-
social disorder). After control for confounding

factors and concurrent disorder the odds ratio

between cannabis abuse/dependence and serious
suicide attempt reduced to 2.0 and was mar-
ginally statistically signi cant (p 0.06). This
nding clearly raises the possibility that, even
after control for confounding and concurrent
factors, cannabis abuse/dependence may have
contributed to suicide attempt risk.
Furthermore, the contribution of mental dis-
orders to the association between cannabis
abuse/dependence and risk of serious suicide
attempt may have occurred in two ways:
(1) This association may have arisen because
cannabis abuse/dependence was co-morbid
with, or secondary to, other psychiatric dis-
orders including mood disorders, alcohol
abuse/dependence, other drug (excluding al-
cohol and cannabis) abuse/dependence and
antisocial behaviours. Such associations
could arise, for example, if those with de-
pressive or other disorders used cannabis to
alleviate their symptoms. Under these cir-
cumstances, the association between can-
nabis abuse and risk of serious suicide
attempt would be coincidental rather than
causal.
(2) Alternatively, cannabis abuse/dependence
may have contributed to risk of other mental
disorders, notably mood disorders which, in
turn, predisposed individuals to risk of seri-
ous suicide attempt. Under these conditions
cannabis abuse/dependence may make an
indirect causal contribution to risk of serious
suicide attempt with this effect being
mediated via other disorders secondary to
cannabis abuse/dependence.
These considerations suggest that while much
of the association between cannabis abuse/
dependence and suicide attempt appears to be
due to confounding factors associated with in-
creased risks of both cannabis abuse and suicide
attempt, the possibility remains that cannabis
abuse/dependence may make an independent
contribution to risk of suicide attempt, both di-
rectly and through the possible effects of cannabis
abuse on risks of other mental disorders.
There are some limitations to this study.
First, the present study is restricted to people
who made serious suicide attempts. While the
extent to which the results of this analysis may
apply to suicide attempts which are less serious
is not known, it seems reasonable to
Cannabis abuse and serious suicide attempts 1163
assume that similar but less marked trends may
apply to those who make suicide attempts of
minor medical severity.
Secondly, recall bias is a potential threat to
validity in case control psychiatric studies
(Lewis & Pelosi, 1990). In the present study,
parallel analyses using reports of childhood psy-
chosocial circumstances obtained from subjects
and from signi cant others suggested a strong
similarity between the conclusions drawn from
the two data sources. While this consistency does
not eliminate, entirely, the possibility that the
observed associations may be contaminated by
recall bias it does provide reassurance that the
ndings are not speci c to the reports of those
who make serious suicide attempts.
The ndings of this analysis are congruent
with a growing literature about cannabis use
which tends to suggest that the increased mental
health risks (including suicide attempt risk) seen
among individuals with cannabis abuse/depen-
dence may be due largely to the psychosocial and
contextual factors associated with cannabis use,
rather than occurring as a result of the causal
impact of cannabis abuse/dependence on the
individual’s susceptibility to psychiatric disorder
(Jessor et al., 1980; Fergusson et al., 1996).
None the less, the present analysis raises the
possibility that cannabis abuse may make rela-
tively small direct and indirect contributions to
risk of serious suicide attempt.
Acknowledgements
This research was funded by the Canterbury
Area Health Board, Healthlink South and a
grant from the Health Research Council of New
Zealand.
References
A
DDICTION
B
UKSTEIN
, O. G., B
RENT
, D. A., P
ERPER
, J. A. et al.
(1993) Risk factors for completed suicide among
adolescents with a lifetime history of substance
abuse: case control study, Acta Psychiatrica Scandi-
navica, 88, 403 408.
C
ARLSON
, G. A., R
ICH
, C. L., G
RAYSON
, P. & F
OWLER
,
R. C. (1991) Secular trends in psychiatric diagnoses
of suicide victims, Journal of Affective Disorders, 2,
127 132.
C
RUMLEY
, F. E. (1990) Substance abuse and ado-
lescent suicide behaviour, Journal of the American
Medical Association, 263, 3051 3056.
E
LLEY
, W. B. & I
RVING
, J. C. (1976) Revised socioeco-
nomic status for New Zealand, New Zealand Journal
of Educational Studies, 11, 25 36.
F
ELTS
, W. M., C
HENIER
, T. & B
ARNES
, R. (1992) Drug
use and suicidal ideation and behaviour among
North Carolina public school students, American
Journal of Public Health, 82, 870 872.
F
ERGUSSON
, D. M., L
YNSKEY
, M. T. & H
ORWOOD
, L. J.
(1996) The short-term consequences of early onset
cannabis use, Journal of Abnormal Child Psychology,
24, 499 512.
F
LEISS
, J. L. (1981) Statistical Methods for Rates and
Proportions , 2nd edn (New York, John Wiley &
Sons).
F
OWLER
, R. C., R
ICH
, C. L. & Y
OUNG
, D. (1986) San
Diego Suicide Study II. Substance abuse in young
cases, Archives of General Psychiatry, 43, 962 965.
H
AWKINS
, J. D., C
ATALANO
, R. F. & M
ILLER
, J. Y.
(1992) Risk and protective factors for alcohol and
other drug problems in adolescence and early adult-
hood: implications for substance abuse prevention,
Psychological Bulletin, 112, 64 105.
H
AWTON
, K., F
AGG
, J., P
LATT
, S. & H
AWKINS
, M.
(1993) Factors associated with suicide after parasui-
cide in young people, British Medical Journal, 306,
1641 1644.
H
AWTON
, K. (1994) Adolescent suicide and attempted
suicide: the importance of substance abuse, Crisis,
15, 13 15.
I
NSTITUTE OF
M
EDICINE
(1982) Marijuana and Health
(Washington, DC, National Academy Press).
J
ESSOR
, R., C
HASE
, J. A. & D
ONOVAN
, J. E. (1980)
Psychosocial correlates of marijuana use and prob-
lem drinking in a national sample of adolescents,
American Journal of Public Health, 70, 604 610.
K
AMINER
, Y. (1992) Psychoactive substance abuse and
dependence as a risk factor in adolescent-attempted
and -completed suicide, American Journal on Addic-
R
ESEARCH
F
OUNDATION
/W
ORLD
H
EALTH
tions, 1, 21 29.
O
RGANIZATION
(1981) Report of an ARF/WHO Sci-
L
ESAGE
enti c Meeting on the Adverse Health and Behavioural
Consequences of Cannabis Use (Toronto, Addiction
Research Foundation).
A
MERICAN
, A. D., B
OYER
, R., G
RUNBERG
, F. et al. (1994)
Suicide and mental disorders: a case control study
of young men, American Journal of Psychiatry, 151,
1063 1068.
P
SYCHIATRIC
A
SSOCIATION
(1987) Diagnos-
L
EWIS
tic and Statistical Manual of Mental Disorders, 3rd ed.
revised (DSM-III-R) (Washington, DC, American
Psychiatric Association).
B
EAUTRAIS
, G. & P
ELOSI
, A. J. (1990) The case control
study in psychiatry, British Journal of Psychiatry, 157,
197 207.
M
ARTTUNEN
, M. J. H
ILLEVI
, M. A., H
ENRIKSSON
, M.
, A. L., J
OYCE
, P. R., M
ULDER
, R. T., F
ER-
M. & L
ONNQVIST
GUSSON
, J. K. (1991) Mental disorders in
, D. M., D
EAVOLL
, B. J. & N
IGHTINGALE
, S.
adolescent suicide DSM-III-R Axes I and II diag-
K. (1996) Prevalence and comorbidity of mental
noses in suicides among 13 19-year-olds in Finland,
disorders in persons making serious suicide at-
Archives of General Psychiatry, 48, 834 839.
tempts: a case control study, American Journal of
M
URPHY
Psychiatry, 153, 1009 1014.
, G. (1988) Suicide and substance abuse,
Archives of General Psychiatry, 45, 593 594.
1164 Annette L. Beautrais et al.
M
URPHY
, J. E. (June 1985) Untitled news release
(available from St Cloud State University, St Cloud,
MN56301).
N
ATIONAL
D
RUG AND
A
LCOHOL
R
ESEARCH
C
ENTRE
(1994) The Health and Psychological Consequences of
Cannabis Use (Canberra, Australian Government
Publishing Service.
N
EELEMAN
, J. & F
ARRELL
, M. (1997) Suicide and sub-
stance misuse, British Journal of Psychiatry, 171,
303 304.
N
EWCOMB
, M. D. & B
ENTLER
, P. M. (1989) Substance
use and abuse among children and teenagers,
American Psychologist, 44, 242 248.
P
FEFFER
, C. R. (1989) Suicide Among Youth: perspec-
tives on risk and prevention (Washington, DC,
American Psychiatric Press).
P
UBLIC
H
EALTH
C
OMMISSION
(1996) Cannabis. The
public health issues 1995 1996 (Wellington, Ministry
of Health).

,R
ICH
, C. L., F
OWLER
R. C. & Y
OUNG
, D. (1989)
Substance abuse and suicide. The San Diego study,
Annals of Clinical Psychiatry, 1, 79 85.
R
UNESON
, B. (1990) Psychoactive substance use dis-
order in youth suicide, Alcohol and Alcoholism, 25,
561 566.
R
UTTER
, M. & S
MITH
, D. J. (Eds) (1995) Psychosocial
Disorders in Young People. Time trends and their causes
(Chichester, John Wiley & Sons).
S
HAFFER
, P. et al. (1996)
Psychiatric diagnosis in child and adolescent suicide,
Archives of General Psychiatry, 53, 339 348.
S
PITZER
, D., G
OULD
, M. S., F
ISHER
, M. &
F
IRST
, J. L., W
ILLIAMS
, J. B. W., G
IBBON
, M. B. (1988) Structured Clinical Interview for
DSM-III-R Patient Version (SCID-P, 6/1/88)
Washington, DC, American Psychiatric Press).