SCHIZOPHRENIA

RESEARCH

ELSEVIER Schizophrenia Research 15 ( 1995) 277-281

Morbid risk of schizophrenia for relatives of patients with

cannabis-associated psychosis
P.K. McGuire a'*, P. Jones a, I. Harvey a, M. Williams a, p. McGuffin b, R.M. M u r r a y a
"Department of Psychological Medicine, Institute of Psychiatry and King's College Hospital,
De Crespigny Park, London SE5 8AF, UK
Department of Psychological Medicine, University of Wales College of Medicine, Cardiff, CF4 4XN, UK
Received 27 January 1994; accepted 13 July 1994

Abstract

Twenty-three patients admitted with acute psychosis who were cannabis positive on urinary screening were each
matched, with respect to sex, with two psychotic controls who screened negatively for all substances. The lifetime
morbid risk of psychiatric disorder was estimated among the first degree relatives of cases and controls, using
RDC-FH criteria to define diagnoses, and Weinberg's shorter method of age correction. The cases had a significantly
greater familial morbid risk of schizophrenia (7.1%) than the controls (0.7%), while the risks of other psychoses, and
of non-psychotic conditions were similar. The same pattern of familial risk was evident when the analysis was
restricted to patients with DSM-III schizophrenia. The data suggest that the development or recurrence of acute
psychosis in the context of cannabis use may be associated with a genetic predisposition to schizophrenia.

Keywords: Cannabis; Psychosis; Morbid Risk; Family History; (Schizophrenia)

1. Introduction 1978; Turner and Tsuang, 1990). Why only certain

Acute organic reactions to cannabis involving
psychotic symptoms occur with large doses in
normal subjects and m a y be a dose-related phe-
nomenon (Meyer, 1975). In contrast, certain
individuals, in the context of cannabis use, can
develop a psychosis in clear consciousness which
persists after the cessation of drug use. This is a
relatively unpredictable event which does not seem
to be dose related (Hollister, 1986). Similarly,
while cannabis use is relatively c o m m o n a m o n g
patients with an existing psychotic illness, some
patients seem particularly prone to an acute
psychotic relapse following its ingestion (Treffert,
* Corresponding author.
individuals might be vulnerable to such effects of
cannabis is unclear, but one possibility is that they
have a strong genetic diathesis for psychotic illness.
We tested this hypothesis in family data from a
matched case control study. We established the
lifetime morbid risk of psychotic disorders for the
first degree relatives of patients with acute psycho-
ses that developed in the context of cannabis use,
and compared these with the risk for the relatives
of substance-free psychotic controls.
2. Materials and methods
Cases and controls were drawn from two consec-
utive series of acute psychotic admissions to the

0920-9964/95/$9.50 © 1995 Elsevier Science B.V. All rights reserved

SSDI 0920-9964(94)00053-0
278 P.K. McGuire et aL/Schizophrenia Research 15 (1995) 277-281
Bethlem Royal, Maudsley and North Dulwich
hospitals, comprising 345 patients. The sample
procedures and socio-demographic characteristics
of these series have been detailed elsewhere
(Harvey et al., 1990; Jones et al., 1992). Inclusion
in the present study required unequivocal evidence
of hallucinations, delusions or thought disorder,
and that the subject's urine had been screened for
illicit substances within 48 h of admission. Drug
screening was performed by clinical staff, indepen-
dent of this study, and the investigators were
unaware whether patients had been screened at
the time they interviewed them.
Cases comprised all subjects testing positively
for cannabis (n=23), two of whom also screened
positively for opiates, and for benzodiazepines and
chlormethiazole, respectively. Urinary cannabis
was detected with an immunoassay (EMIT, CIBA)
which is sensitive to concentrations greater than
20 ng/ml. The controls were drawn from admis-
sions whose urine was negative for all substances.
Two controls were paired with each case with
respect to sex, taking the first 46 subjects from the
data file. Two thirds of both the cases (16/23) and
the controls (30/46) had previously been admitted
with psychosis, while one third in each group were
first admissions.
Subjects were interviewed within 72 h of admis-
sion. Lifetime drug use was assessed with a semi-
structured interview. DSM-III diagnoses (DSM-
IIIR was not available at the time the data were
collected) for the subjects were derived from
the Present State Examination (Wing et al., 1974),
and from information in their medical records.
The ethnicity of subjects was categorised in terms
of their country of birth (as reported by the
patient), and their parental ethnicity (as reported
by one of the parents). Socio-demographic data
were obtained with a semi-structured interview,
socio-economic class being defined according to
the criteria of Goldthorpe and Hope (1974).
Data on psychiatric morbidity among the first
degree relatives of subjects was collected from
interviews with the proband's mother. When the
mother was unavailable, which applied to 3 cases
and 6 controls, another first degree relative was
interviewed. Additional information was available
from the probands and the medical records of
probands and their relatives. The Family History
Research Diagnostic Criteria (Endicott et al.,
1975) were used to define psychiatric diagnoses
among the relatives, with the assessor blind to the
status of the proband. The morbid risks of
psychotic illnesses, depression, substance and alco-
hol abuse were calculated using Weinberg's method
of age correction. This method weights the number
of unaffected relatives according to age, with those
who have passed through the designated period of
risk without being affected (U3) making twice the
contribution of those who lie within the risk period
(U2), and those yet to enter the risk period not
being counted (U1). The morbid risk is calcu-
lated by dividing the number of affected relatives
(A) by the age-corrected number of relatives
(Bezugsziffer, BZ):
A
Morbid R i s k -
Aq- U2/2+U 3
The period of risk used for schizophrenia, bipolar
illness, schizo-affective disorder, unspecified psy-
chosis and substance abuse was 15-50 years. The
risk period for alcohol abuse was 18-50 years,
while that for major depression was 18-70 years.
3. Results
After matching for sex, the cases and controls
were similar in terms of age, socio-economic class,
and paternal socio-economic class. There were
relatively more Afro-Caribbeans among the con-
trols, but this was not statistically significant
(Table 1). Cases and controls were also similar
with respect to the distribution of DSM-III diagno-
ses, with the majority of both groups meeting
criteria for schizophrenia, and comparable propor-
tions with bipolar illness or atypical psychosis
(Table2). None of the cases or controls had
organic psychoses. Most of the cases admitted to
regular substance use, compared with a minority
of the controls, and the cases were significantly
more likely to have admitted to substance use than
the controls (Zz, trend = 12.7, p = 0.0004; Table 3).
There was a strong trend for the lifetime morbid
risk for all psychotic illnesses (RDC-FH schizo-
 P.K. McGuire et al./Schizophrenia Research 15 (1995) 27~281 279

Table 1 Table 4
Socio-demographic characteristics of cases and controls Lifetime morbid risk of psychiatric illness for first degree
relatives (> 15 years old) of all probands
Cannabis Cannabis
positive negative RDC-FH Relatives of Relatives of
(n =23) (n =46) diagnosis cannabis positive cannabis negative
cases (n = 110) controls (n = 216)
Male : female 20 : 3 40 : 6
Mean age (years) 25.2 25.8 A BZ MR(%) A BZ MR(%)
Median age (years) 24.0 25.0
Father's social class (mean) 3.2 3.3 Schizophrenia 5 70.5 7.1'* 1 135 0.7
Proband's social class (mean) 5.3 5.2 Bipolar illness 0 68 0 1 135 0.7
% Caucasian 47.8 39.2 Schizo-affective 1 68 1.5 1 135 0.7
% Afro-Caribbean 43.6 54.2 disorder
% Other 8.6 6.6 Unspecified 1 68 1.5 1 135 0.7
psychosis
All psychoses 7 72.5 9.6* 4 136.5 2.9
Table 2 Depression 2 51.5 3.9 l1 110 10.0
Distribution of DSM-III diagnoses among cases and controls Substance abuse 0 68 0 l 135 0.7
Alcohol abuse 3 54.5 5.5 5 132.5 3.8
DSM-III diagnosis Cannabis Cannabis
positive negative A, number of affected relatives; BZ, age-corrected number of
relatives; MR, morbid risk.
Schizophrenia 14 (60.9%) 29 (63.0%) **OR 10.2, (1.12-234), Fisher Exact 2-tailed p =0.02.
Bipolar disorder 2 (8.7) 8 (17.4) *OR 3.5 (0.9-14.9), Fisher Exact 2-tailed, p=0.052.
Major depression 0 0
Atypical psychosis 4 (17.4) 7 (15.2)
Other 3 (13.0) 2 (4.4)
depressive disorder was lower for the cases than
the controls, but this was not significant, and the
respective morbid risks for substance and alcohol
Table 3 abuse were similar (Table 4).
Lifetime self-reported substance use among cases and controls When the analysis was restricted to probands
with DSM-III schizophrenia, a similar pattern of
Cannabis Cannabis
positive negative familial morbid risks emerged; the risk of psychosis
was significantly greater for the cases than the
Never used drugs 0 12 (32.4%) controls (Odds Ratio=4.7, 0.98-25.2, p=0.03,
Once or lwice 5 (22.7%) 6 (16.2) Table 5), and this largely reflected a relatively
Several times 2 (9.1) 13 (35.1)
greater morbid risk for schizophrenia (Odds
Regular use 15 (68.2)* 6 (16.2)
Uncoded 1 9 Ratio = 9.05, 0.9-219, p = 0.04). The risks for other
psychoses were similar for the two groups, as were
*Z2, trend = 12.7, p = 0.0004. the risks for substance and alcohol abuse.
Depression was again commoner among the rela-
phrenia, schizo-affective disorder, bipolar illness, tives of the controls, but not to a significant extent
mania and unspecified functional psychosis) to be (Table 5).
greater for the relatives of the cases than the
controls (Odds ratio = 3.5, 95% ci 0.9-14.9, 1-tailed
p=0.052, Fisher's exact test; Table 4). While the 4. Discussion
morbid risks of bipolar, schizo-affective and
unspecified psychoses were relatively low for the These data suggest that there is a particularly
relatives of both groups, the risk for schizophrenia high morbid risk of schizophrenia among the
was almost 10 times higher for the cases than relatives of probands who develop, or relapse into,
the controls (Odds r a t i o = 10.2, 95% ci 1.12-234, psychosis in the context of cannabis use. This
l-tailed p = 0.02, Fisher's exact test). The risk for finding is difficult to attribute to potentially con-
280 P.K. MeGuire et al./Schizophrenia Research 15 (1995) 277-281

Table 5 and relatives concerned indicated that substance

Lifetime morbid risk of psychiatric illness for first degree use was not a factor in their illnesses.
relatives (> 15 years old) of probands with schizophrenia
The familial morbid risk of schizophrenia for
RDC-FH Relatives of Relatives of the substance negative controls (0.7%) was lower
diagnosis cannabis positive cannabisnegative than most estimates for schizophrenics using RDC
cases (n =71) controls (n= 153) criteria (1.4-5.8%; Kendler and Diehl, 1993),
whereas the risk for the cases (7.1%) was higher.
A BZ MR(%) A BZ MR(%)
T h e relatively low estimate for the controls may
Schizophrenia 4 45.5 8.8* 1 96 1.0 have been a function of the small sample size. It
Bipolar illness 0 43.5 0 1 96 1.0 could also have resulted from our use of the
Schizo-affective 1 44 2.3 0 95.5 0 R D C - F H method, as opposed to personal inter-
disorder views with relatives, although our unit has
Unspecified 1 44 2.3 1 96 1.0
psychosis found little difference in the sensitivity of these
All psychoses 6 45.5 13.2"* 3 97 3.1 two approaches (Sham, personal communication).
Depression 1 32 3.1 8 79.5 10.1 Another possibility is that schizophrenia is a heter-
Substance abuse 0 43.5 0 1 96 1.0 geneous condition with subtypes which differ with
Alcohol abuse 3 40 7.5 3 95.5 3.1 respect to familial morbid risk. Our controls may
A, number of affected relatives; BZ, age-correctednumber of belong to a subgroup whose morbid risk is not
relatives; MR, morbid risk. much greater than that of the general population,
**OR 4.7 (0.98-25.2), Fisher's exact test, 2-tailedp=0.03. whereas a vulnerability to cannabis-associated psy-
*OR 9.05 (0.9-219), Fisher exact test, 2-tailedp=0.04. chosis may be a marker for a group with an
especially high familial morbid risk.
founding differences between the cases and the The morbid risk of psychiatric illness among the
controls, as they were similar with respect to relatives of cannabis users has not been specifically
sociodemographic variables, ethnicity and psychi- examined before, but Andreasson et al. (1989)
atric diagnosis. Although the familial risk of psy- found similar familial rates of schizophrenia
chosis in general was also higher for the cases, this among schizophrenics who had and had not used
essentially reflected their greater risk of schizophre- cannabis, although the number of probands inves-
nia; there were no differences between cases and tigated was very small. A number of studies have
controls in terms of the risk of bipolar, schizo- examined familial psychiatric illness in multiple
affective or unspecified psychoses, nor in terms of drug users with psychosis, and have generally
non-psychotic psychiatric disorders. found similar rates of illness, including schizophre-
The high familial morbid risk of schizophrenia nia, among their relatives compared to substance-
seen in the cases could have reflected a tendency free controls (Bowers, 1972; Breakey et al., 1974;
for those with a strong diathesis for psychiatric Dixon et al., 1991). Similar 'incidence' rates for
illness to abuse drugs. However, this would require schizophrenia have also been described in the
that this effect was specific to schizophrenia, and parents of patients with LSD-associated psychoses
is not consistent with the low morbid risk of and substance-free schizophrenics (Vardy and
substance use we observed for the cases. Moreover, Kay, 1983). One study has used an age-corrected
there is little evidence to suggest that substance denominator to estimate familial risk among
users in general have a high familial morbid risk psychotic polydrug users, and this reported a
of schizophrenia. Nevertheless, future studies higher risk of affective disorders in cases than
might include a group of cannabis users with no controls, but a similar risk of schizophrenia
psychiatric morbidity, to control for this eventual- (Tsuang et al., 1982).
ity. Another possibility is that the relatives diag- Most previous work has thus studied multiple
nosed as suffering from schizophrenia actually had drug users rather than cannabis users, and esti-
drug-related psychoses. However, examination of mated rates of familial psychiatric illness rather
their casenotes and interviews with the probands than lifetime morbid risks. This may account for
 P.K. McGuire et al./Schizophrenia Research 15 (1995) 277-281
the failure of previous studies to identify the
differences we observed. However, the modest
number of subjects in our study meant that the
number of relatives affected with some forms of
psychiatric disorder was small. As a result, the
misdiagnosis of a few individuals could have had
a relatively strong influence on the estimated
morbid risks, and this is reflected in the relatively
broad confidence limits around our odds ratios.
Therefore, although the differences between cases
and controls with respect to schizophrenia were
statistically significant, the hypothesis requires
retesting in larger populations before firmer con-
clusions can be drawn.
Cannabis use has long been suspected as playing
a role in the pathogenesis of psychosis, although
the precise mechanism of its effect remains contro-
versial. Premorbid use may be associated with an
increased risk of developing schizophrenia in later
life (Andreasson et al., 1987), and cannabis use
by psychotic patients seems to increase the risk of
an acute relapse (Treffert et al., 1978; Turner and
Tsuang. 1990), and can exacerbate existing symp-
toms (Negrete et al., 1986). Our data support the
notion that cannabis use may 'trigger' the develop-
ment or re-emergence of a 'functional' psychosis
in predisposed individuals, rather than producing
a specific 'cannabis psychosis' (Thornicroft, 1990;
McGuire et al., 1994). As cannabis probably acts
on the brain through specific cannabinoid recep-
tors (Howlett et al., 1990), this effect may involve
an interaction between its efficacy at such sites and
the user's genetic diathesis for schizophrenia.
References
Andreasson, S., Allebeck, P., Engstromm A. and Rydberg U.
(1987) Cannabis and Schizophrenia: a longitudinal study of
Swedish conscripts. Lancet ii, 1483 1485.
Andreass,m, S., Allebeck, P. and Rydberg U. (1989)
Schizophrenia in users and non-users of cannabis. Acta
Psychizatrica Scand. 79, 505 510..
281
Bowers, M.B. (1972) Acute psychosis induced by psychotomi-
metic drug abuse. Arch. Gen. Psychiatry 27, 437 440.
Breakey, W.R., Goodell, H., Lorenz, P.C., et al. (1974)
Hallucinogenic drugs as precipitants of schizophrenia.
Psychol. Med. 4, 255 261.
Dixon, L.D., Haas, G., Weiden, P.J., et al. (1991) Drug abuse
in schizophrenic patients: clinical correlates and reasons for
use. Am. J. Psychiatry 148, 224-230.
Endicott, J., Andearsen, N., Spitzer, R.L. (1975) Family
History Research Diagnostic Criteria. Biometrics Research,
New York State Psychiatric Institute, New York..
Goldthorpe, J. H. and Hope, K. (1974) The social grading of
occupations- A new approach and scale. Oxford:
Clarendon Press..
Harvey, I., Williams, M., McGuffin, P. and Toone, B. K.
(1990) The functional psychoses in Afro-Caribbeans. Br.
J. Psychiatry 157, 515 522.
Hollister, L.E. (1986) Health aspects of cannabis. Pharmacol.
Rev. 38, 1 20..
Howlett, A.C., Bidaut-Russell, M., Devane, W.A., et al. (1990)
The cannabinoid receptor: biochemical, anatomical and
behavioural characterisation. Trends Neurosci. 13, 420 423.
Jones, P.B., Bebbington, P., Foerster, A., Lewis, S.W., Murray,
R., Russell, A., Sham, P.C., Toone, B.K. and S. Wilkins
(1993) Premorbid social underachievement is specific to
schizophrenia. Results from the Camberwell Collaborative
Psychosis Study. Br. J. Psychiatry 162, 65 71.
Kendler, K.S., Diehl, S.R. (1993) The genetics of schizophrenia.
Schizophr. Bull. 19, 261 286.
McGuire, P.K., Jones, P., Harvey, I., et al. t1994) Cannabis
and acute psychosis. Schizophr. Res. 13, 161 168.
Meyer, M.E. (1975) Psychiatric consequences of marijuana
use: the state of the evidence. In: Marijuana and Health
Hazards: Methodological Issues in Current Research,
pp 133-152. (Ed J.R. Tinkleberg), New York: Academic
Press..
Negrete, J. C., Knapp, W. P., Douglas, D. E. and Smith, W.
B. (1986) Cannabis affects the severity of schizophrenic
symptoms: results of a clinical survey. Psychol. Med.
16, 515 520.
Thornicroft, G. i1990) Cannabis and psychosis: Is there
epidemiological evidence for an association'? Br. J. Psychiatry
135, 1213 1215.
Treffert, D. (19781 Marijuana use in schizophrenia: a clear
hazard. Am. J. Psychiatry 135, 1213 1215.
Turner, W. M. and Tsuang, M. T. (1990) Impact of substance
abuse on the course and outcome of schizophrenia.
Schizophr. Bull. 16, 1990.
Wing, J. K., Cooper, J. E. and Sartorius, N. (1974)
Measurement and classification of psychiatric symptoms.
Cambridge University Press.