Demographic and clinical correlates of comorbid substance use disorders in psychosis: multivariate analyses from an epidemiological sample

Abstract

Background: While there has been substantial research examining the correlates of comorbid substance abuse in psychotic disorders, it has been difficult to tease apart the relative importance of individual variables. Multivariate analyses are required, in which the relative contributions of risk factors to specific forms of substance misuse are examined, while taking into account the effects of other important correlates. Methods: This study used multivariate correlates of several forms of comorbid substance misuse in a large epidemiological sample of 852 Australians with DSM-III-R-diagnosed psychoses. Results: Multiple substance use was common and equally prevalent in nonaffective and affective psychoses. The most consistent correlate across the substance use disorders was male sex. Younger age groups were more likely to report the use of illegal drugs, while alcohol misuse was not associated with age. Side effects secondary to medication were associated with the misuse of cannabis and multiple substances, but not alcohol. Lower educational attainment was associated with cannabis misuse but not other forms of substance abuse. Conclusion: The profile of substance misuse in psychosis shows clinical and demographic gradients that can inform treatment and preventive research.

Introduction

Comorbidity of substance use disorders and psychotic disorders is both common and of significant concern. People with this comorbidity tend to have substantially poorer symptomatic and functional outcomes, including increased relapses and rehospitalizations; financial, health, and legal problems; unstable housing and homelessness; and family burden (Drake and Brunette, 1998). Existing deficits and vulnerabilities of people with psychosis mean that these negative functional and symptomatic effects may often occur at relatively low levels of intake (Drake et al., 1989). A consequence of the impact of comorbidity on patients is that their service costs are much higher, on average, than people without the comorbidity (Bartels et al., 1993).

The effects on individuals, carers, and treatment services are exacerbated by the frequency of comorbidity. The lifetime prevalence of substance abuse or dependence in psychotic disorders is about 40–60% in most Western samples Cantor-Graae et al., 2001, Fowler et al., 1998, Regier et al., 1990. Within inpatient contexts or other high-intensity services, rates of current misuse are often higher than in outpatient or community samples of people with psychosis (Cantor-Graae et al., 2001), partly because of the higher probability that dual-disordered individuals will present for treatment of one or the other disorder (Berkson, 1949). Frequencies of substance use and the selection of particular substances are also subject to trends in the general community (Patkar et al., 1999), including changes in intake due to availability and price.

Together, these features of comorbidity make it important to study factors that are associated with substance misuse (i.e., abuse or dependence). This information may allow the development of improved models of problem development and more effective strategies for both prevention and treatment. Some consistency in correlates is emerging. Comorbidity of substance misuse and psychosis tends to be associated with younger age, single marital status, male gender Cantor-Graae et al., 2001, Duke et al., 2001, Salyers and Mueser, 2001, antisocial personality disorder, and conduct disorder Mueser et al., 1997, Mueser et al., 1999. However, most studies on the correlates of substance misuse have used relatively small samples and therefore most cannot incorporate multivariate predictions, leaving open the possibility that some observed relationships are due to confounding factors. A multivariate approach allows a more sophisticated and parsimonious modelling of relationships (e.g., Mueser et al., 2000). The examination of differential relationships between different types of substances is also rare. Where correlates for different substances are tested, differences in associated variables tend to emerge Mueser et al., 2000, Salyers and Mueser, 2001. For example, younger age and less education appear to be stronger factors in nonalcoholic drugs than alcohol Mueser et al., 2000, Salyers and Mueser, 2001. Just as overall usage and substance selection often differ over time and populations (Patkar et al., 1999), so do these demographic correlates change. A multivariate examination of differential relationships with substances in additional countries is required.

Such analyses require a large data set. The Low Prevalence arm of the Australian National Survey of Mental Health and Well-Being (Jablensky et al., 1999a) provided such a sample. While the main community survey had 10641 adults, only 99 of these screened positive for psychosis (Degenhardt et al., 2001). The Low Prevalence arm provided a targeted sample of 970 people with confirmed diagnoses of psychosis. It therefore allowed us to examine multivariate correlates of specific comorbid substance use disorders and psychotic disorders. We also examined variables that were associated with multiple coexisting substance use disorders.