Effects of cannabis use on outcomes of psychotic disorders: systematic review

ZAMMIT Stanley, et al
Journal article citation:
British Journal of Psychiatry, 193(11), November 2008, pp.357-363.
Cambridge University Press

Researchers searched 10 relevant databases (to November 2006), reference lists of included studies and contacted experts. Thirteen longitudinal studies from 15,303 references were included. Data extraction and quality assessment were conducted independently and in duplicate. Cannabis use was consistently associated with increased relapse and non-adherence. Associations with other outcome measures were more disparate. Few studies adjusted for baseline illness severity, and most made no adjustment for alcohol, or other potentially important confounders. Adjusting for even a few confounders often resulted in substantial attenuation of results. Confidence that most associations reported were specifically due to cannabis is low. Despite clinical opinion, it remains important to establish whether cannabis is harmful, what outcomes are particularly susceptible, and how such effects are mediated. Studies to examine this further are eminently feasible.

Extended abstract:
ZAMMIT Stanley; et al.;


Effects of cannabis use on outcomes of psychotic disorders: systematic review.

Journal citation/publication details

British Journal of Psychiatry, 193(11), November 2008, pp.357-363.


Thirteen longitudinal studies are reviewed and show that cannabis use is consistently associated with increased relapse/readmission to hospital and reduced adherence to treatment. However, evidence on the associations with other outcomes is contradictory, in part because of the methodological inadequacies of most studies. Although it is generally believed by clinicians that cannabis use leads to poorer outcomes for those with psychotic disorders, this cannot be claimed with any confidence on the basis of current evidence.


Cannabis use is more common in people with psychotic disorders than in those without, and there is some evidence that it may increase the incidence of psychotic outcomes independently of the transient effects of intoxication. This review focuses specifically on longitudinal studies which, unlike the more common cross-sectional and case control studies, can provide evidence of the direction of any association between cannabis use and psychosis.


What sources were used?
The following databases were searched from inception to November 2006: BIOSIS; CINAHL (Cumulative Index to Nursing and Allied Health Literature); EMBASE (Excerpta Medica); ISI Proceedings; LILACS (Latin American and Caribbean Health Sciences); MedCarib (Caribbean Health Sciences Literature); Medline; PsycINFO; Web of Knowledge; and Zetoc (British Library database of journal and conference contents). The reference lists of relevant studies were checked, and authors and other experts were contacted for further published and unpublished studies.

What search terms/strategies were used?
The search strategy was developed by four of the authors, including an experienced librarian, using free text and MeSH (Medical Subject Heading) terms. Full details are available on request, but the general format was:

(psychosis OR schizophrenia OR hallucinations OR delusions) [plus synonyms] AND
substance abuse [plus synonyms]

The searches were limited to studies on humans, but not by language or publication status.

What criteria were used to decide on which studies to include?
Eligible studies were longitudinal studies (or case control studies nested within research of this type) in which cannabis use was measured prior to the measurement of a psychotic outcome. The disorders defined as psychotic are listed, together with primary outcomes of interest including relapse, readmission and changes in symptom scores. Studies of individuals with a dual diagnosis of psychosis and cannabis misuse/dependence were excluded.

Who decided on their relevance and quality?
The filtering of the study results is summarised in a flow diagram. Of the more than 15,000 papers delivered by the searches, 226 were acquired in full text for detailed checking against the inclusion criteria. Those that passed this stage were quality assessed by recording how potential non-causal explanations of results, especially bias and confounding, were accounted for. MOOSE (Meta-analysis of Observational Studies in Epidemiology) guidelines were followed. These processes were conducted independently, apparently by two authors, and disagreements resolved by consensus.

How many studies were included and where were they from?
Thirteen studies were reviewed. Seven were based in single countries: Australia (4), Spain (3), Canada (2), the UK (2) and Germany (1). The final study was a multi-centre trial based in North America and Western Europe. The results of the studies are summarised in a table available via http://www.bristol.ac.uk/psychiatry/research/cannabis.html. The authors also note five ‘near misses’ that all report associations between cannabis use and worse psychotic outcomes. These are also summarised in a table available via this web address, as are the results of the quality assessments of the studies.

How were the study findings combined?
The variety in the definitions of psychotic outcomes and cannabis exposure measure precluded meta-analysis, and the synthesis is thus narrative.

Findings of the review

The 13 studies reported 52 outcomes. Cannabis use was associated with a worse outcome in 14 of these, and a better outcome in seven. There was no evidence of an association in either direction for the other 31.

The use of cannabis ‘was associated fairly consistently’ with increased relapse or hospital admission, and reduced adherence to treatment. However, the results for symptom severity were mixed. Three studies provided evidence that cannabis use was associated with increased positive symptoms, and one that use was associated with reduced negative symptoms. The evidence for other outcomes was also mixed, with some studies reporting reduced quality of life, reduced productivity and a more continuous illness course, while others found positive effects on the length of hospital stays and ‘deficit schizophrenia’.

The ability to interpret the results of the review is heavily dependent on the methodological quality of the studies, especially in relation to bias and confounding. This is generally poor. Only four of the studies made any attempt to adjust for the severity of illness at baseline. Similarly, only six studies adjusted for socio-demographic characteristics such as social class and gender, and most made no adjustment for alcohol or other drug use. All of these factors are associated with poorer outcomes and are likely to lead to an incorrect estimation of the true causal association between these outcomes and cannabis use. Over-estimation is the most likely result given the fact that clinical opinion has long assumed that cannabis use has detrimental effects but only one study reported that the measurement of outcomes was masked to cannabis exposure status.

While some studies may have over-estimated the association between cannabis use and poorer psychotic outcomes, others may have under-estimated the association or failed to find it. Reasons for this could include the general failure to make repeated and detailed measurements of cannabis exposure during the follow-up period, and lack of statistical power (i.e. insufficient sample size to detect an association). None of the studies that failed to find an association between cannabis use and psychotic outcomes presented power calculations, and the authors judge that the majority of the samples were too small. Finally, the inconsistent results found in many cases might have been the result of differences in the diagnostic composition of samples: those using samples with schizophrenia or related spectrum disorders showed more consistent evidence of poorer outcomes in those using cannabis than did samples of patients with any psychotic disorder.

Authors' conclusions

‘Insufficient empirical evidence exists at present to adequately examine whether cannabis use has detrimental effects on the outcome of psychotic disorders, or to determine the pathways by which such effects are mediated, although these are important and clinically relevant questions.’

Future studies should be longitudinal in design and adequately powered, and should include: baseline measures of illness severity and other characteristics known to be associated with poorer outcomes; repeated measures of psychopathology and use of cannabis, alcohol and other substances; measures to minimise the chances of reverse causation effects; clearly defined diagnostic categories; and a more consistent approach to measuring both cannabis exposure and psychotic outcomes.

Subject terms:
outcomes, psychoses, drug misuse, disorders;
Content type:
systematic review
Register/Log in to view this article
Journal home page
ISSN online:
ISSN print:

Key to icons

  • Free resource Free resource
  • Journal article Journal article
  • Book Book
  • Digital media Digital media
  • Journal Journal

Give us your feedback

Social Care Online continues to be developed in response to user feedback.

Contact us with your comments and for any problems using the website.

Sign up/login for more

Register/login to use standard search filters, access resource links, advanced search and email alerts