NDARC Technical Report No. 93 (2000)


Psychotic disorders, such as schizophrenia and schizoaffective disorders, have a lower prevalence than other forms of mental illness, yet they impose a disproportionate personal and public health burden. One issue that has received considerable attention in recent years has been substance use among persons with psychotic disorders. There is a considerable amount of clinical research that reveals high rates of substance use and substance use disorders among this population. However, while clinical evidence suggests cause for concern, clinical samples may be prone to selection biases, and causal theories have rarely been tested.

There has been some population-level research conducted in the US. The Epidemiological Catchment Area (ECA) study, a study of representative samples of adults across 5 US sites, found elevated rates of schizophrenia among persons with alcohol and other drug use disorders. There has also been an analysis using ECA data of the relationship between drug use and having a “self-reported psychotic experience” (experiencing at least one psychotic symptom). Daily cannabis use and meeting criteria for a DSM-III alcohol use disorder were significant predictors of reporting at least one psychotic symptom during the follow-up year.

While these findings provided important population-level information concerning the risks of experiencing at least one psychotic symptom that are associated with drug use, there remains a number of issues to address. First, do the associations observed among North American populations exist in other countries? Second, is there a relationship between problematic drug use and severity of psychotic disorder? Finally, are any associations due to the confounding effects of demographic characteristics or other mental health problems?

The aims of the present paper were: a) to provide Australian estimates of the population-level association between psychotic symptoms and substance use; b) to examine patterns of association between substance use and likely cases of psychosis in the Australian general population; c) to examine the prevalence of problematic use among users according to psychosis case status; and d) to examine the association between problematic substance use and increasing scores on a psychosis screener using multivariate ordinal logistic regression (OLR), taking into account the effects of demographics, mental health, and the fact that the use of multiple drug types would be likely. OLR is a regression technique that takes into account the ordering of an ordered categorical outcome variable (in this case, number of psychotic symptoms). OLR produces an estimate of the average change in the odds for each additional point along the ordered scale (here, for each additional psychotic symptom reported).

The data used was obtained from the National Survey of Mental Health and Well-Being (NSMHWB), a stratified multistage probability sample survey of persons aged 18 years and older in the Australian population. A subset of the sample was used, comprising persons under the age of 50 years (n=6,722), because drug use and psychosis are both rare in older adults.

The psychosis screener (PS) was developed for use in the NSMHWB, and assesses the presence of characteristic psychotic symptoms. Univariate associations between “case” status on the screener and a number of drug use variables were examined: use of tobacco; use and DSM-IV abuse or dependence upon alcohol, cannabis, sedatives, stimulants and opiates; and frequency of alcohol and cannabis use. The conditional prevalence (prevalence among users) was estimated for DSM-IV alcohol and cannabis abuse and dependence, as well as for other drug use disorders (abuse or dependence on sedatives, stimulants or opiates). All prevalence estimates were weighted to conform to independent population estimates, and balanced repeated replicate weights were used to account for the survey design. Odds ratios and their 95% confidence intervals were estimated using unweighted data. A series of OLRs was carried out, in which demographic variables, mental health variables, and finally drug use variables, were entered. The significant mental health and drug use variables were retained in the final OLR model.

A total of 99 persons screened positively for psychosis (a weighted prevalence rate of 1.2% in this age group, 18 to 49 years). Regular tobacco use was much more common among persons screening positively, with 60% reporting such use, compared to 27% of non-cases. A greater proportion of cases reported regular alcohol or cannabis use. Cases were much more likely than non-cases to meet criteria for DSM-IV alcohol (24% vs. 8%) and cannabis (16% vs. 3%) use disorders. Similar patterns were observed for other drug use, with cases significantly more likely to report the use of these drugs, as well as meet criteria for abuse or dependence.

An examination of the prevalence of use disorders among users of these substances revealed that among those who reported alcohol or cannabis use, cases were much more likely than non-cases to report dependent use (25% of cases who were alcohol users versus 7% of non-cases; 41% vs. 21% for cannabis).

The OLR revealed that the association between problematic substance use and psychotic symptoms was not due to the effects of demographic variables, a measure of personality, or other mental health problems. Regular tobacco use (OR = 1.43; 95%CI 1.23, 1.66), cannabis dependence (OR = 2.00; 95%CI 1.40, 2.85), alcohol dependence (OR = 1.49, 95%CI 1.16, 1.91), and opiate abuse (OR 6.54; 95%CI 1.34, 31.81) were each significant predictors of increased numbers of psychotic symptoms.

Hence, the confounding variables considered here did not explain the relationship between problematic substance use and psychotic symptoms.Given the cross-sectional nature of the data from the NSMHWB, however, it is not possible to distinguish between other competing explanations for the associations observed here. On the basis of the evidence currently available, it seems likely that the use of drugs such as cannabis may exacerbate the symptoms of those with psychotic illnesses. Controlled outcome studies of substance abuse treatment for persons with psychosis could examine whether cessation of drug use predicts improvement in symptoms.

There are a number of implications of the present study. First, the mental health risks of problematic substance use need to be disseminated to persons who are at risk of psychotic illness, to persons who have already been diagnosed with a psychotic illness, and to persons who are heavy substance users. In particular, the risks of exacerbation of, or relapse to mental health problems need to be highlighted.

Second, more attention needs to be given to the physical health risks of heavy or problematic substance use by persons with psychotic illnesses, particularly given the high rates of tobacco and cannabis smoking. Heavy alcohol users also face increased risks of cognitive impairment and physical disease.

Finally, the use of alcohol, cannabis, sedatives, opiates and stimulants may interfere with medication, reduce treatment compliance, reduce housing stability and increase the burden upon treatment services and the family. Work is needed to address problematic substance use among persons with psychotic disorders.



Louisa Degenhardt, Wayne Hall
Date Commenced
06 Oct 2000
Resource Type
Technical Reports