There are several reasons why two disorders might co-occur — that is, be truly comorbid (Caron & Rutter, 1991; Kessler, 1995). These are: (1) that there is a direct causal relationship between the two, with the presence of one disorder making another more likely to develop; (2) that there is an indirect causal relationship between the two, with one disorder affecting a third variable in a way that increases the likelihood of the second disorder; and (3) that there are common factors that increase the risk of both disorders. These are discussed in more detail below.

Direct causal relationship
Indirect causal relationship
Common factors

Direct causal relationship

There is a range of causal relationships that have been used to explain specific types of comorbidity between substance use problems and other mental health problems. Mental disorders have been argued to cause substance use disorders, and vice versa.

Mental health problems cause substance use problems

A plausible hypothesis of the relationship between substance use disorders and other mental health problems is that persons with mental health problems who begin to use substances to alleviate the symptoms of their illness develop problematic use as a result of over-use (Khantzian, 1985; Khantzian, 1997; Pope, 1979). A central assumption of this 'self-medication' hypothesis is that substances are used to alleviate symptoms and that specific substances will be selected for their specific effects upon mood and cognition. For example, it has been suggested that persons who are heroin dependent use heroin to ameliorate aggression and rage, while persons who are cocaine dependent use it to alleviate symptoms of depression (Khantzian, 1985).

A variation of the self-medication hypothesis has also been used to explain the relationship between schizophrenia and substance use. One such hypothesis is that persons with schizophrenia use tobacco to reduce positive symptoms such as hallucinations and delusions (Gilbert & Gilbert, 1995), and also to reduce negative symptoms such as blunted affect, apathy and anhedonia (Gilbert & Gilbert, 1995; McEvoy & Brown, 1999).

However, the evidence that specific drugs are used to 'treat' specific symptoms is less than compelling (Mueser et al., 1998). For example, self-report studies of persons with schizophrenia and substance use disorder have found very little evidence that different substances are used to alleviate specific mood states or symptoms (Dixon, Haas, Weiden, Sweeney, & Frances, 1991; Noordsy et al., 1991). Furthermore, patterns of substance use among persons with psychotic disorders tend to reflect substance availability and hence show the same patterns of substance use as are found in the general population (Hall, 1998).

The common co-occurrence of alcohol use and anxiety disorders has suggested the 'tension reduction' hypothesis (Cappell & Greeley, 1987). This hypothesis proposes that persons with anxiety disorders use alcohol to relieve anxiety or distress, and that problematic use becomes more likely (being reinforced) because alcohol becomes the means to control these negative mood states (Cappell & Greeley, 1987). This hypothesis is consistent with the acute anxiolytic effects of alcohol (Allan, 1995). However, it is less consistent with what is known about the longer-term effects of alcohol consumption. The effects of chronic alcohol use in high doses include increased anxiety (Stockwell & Bolderston, 1987; Stockwell, Hodgson, & Rankin, 1982). Studies of phobic disorders have also found that phobic anxiety is not alleviated by alcohol use (Marshall, 1997).

A more general form of the self-medication hypothesis proposes that substances are used in an attempt to relieve a variety of dysphoric moods, such as depression and anxiety, general malaise and boredom (Mueser et al., 1998). Research on self-reported reasons for substance use has provided some support for this notion (e.g. Warner et al., 1994); but it can be argued that alleviating dysphoria is simply one among many risk factors — such as poor social skills, poor social functioning and peer group influences — that increase the likelihood of both substance use and mental disorders (Mueser et al., 1998).Top of page

Substance use problems cause mental health problems

A different type of direct causal hypothesis is that substance use problems precipitate mental health problems. For example, there is evidence that some persons may develop depression that is secondary to alcohol dependence (Marc A. Schuckit et al., 1997) in the sense that it develops after alcohol dependence and is likely to remit with abstinence from alcohol (Brown & Schuckit, 1988).

There has also been considerable debate over whether cannabis use is causally related to schizophrenia (Blanchard, Brown, Horan, & Sherwood, 2000; Hall, 1998; Hall & Degenhardt, 2000; McKay & Tennant, 2000; Mueser et al., 1998; Thornicroft, 1990; Thornicroft, Meadows, & Politi, 1992). Some have argued that cannabis use can trigger a 'cannabis psychosis' (Solomons, Neppe, & Kuyl, 1990), while others have argued that its use might precipitate schizophrenia in vulnerable individuals (Andreasson, Allebeck, & Rydberg, 1987).

Comorbidity between different substance use problems has also been explained in causal terms. For instance, it has been hypothesised that the use of cannabis leads to the later use of other illicit drugs (O'Donnell & Clayton, 1982). There has been a great deal of debate about this 'gateway hypothesis'. A strong relationship exists between the use of cannabis and the later use of other illicit substances (Fergusson & Horwood, 1997; Fergusson & Horwood, 2000; Kandel & Faust, 1975; Kandel, Yamaguchi, & Chen, 1992), and it persists after statistical control for a wide range of personal, family background and environmental factors (Fergusson & Horwood, 2000). Nevertheless, it could be that other variables account for the association, which have not been considered in research to date. Alternatively, common genetic factors may play some role in increasing the likelihood of both cannabis use and other substance use, a possibility that has been given some support by twin studies (Tsuang et al., 1998). These possibilities are considered below.

Indirect causal relationship

An indirect causal relationship would exist between two comorbid disorders if one disorder had an effect upon another factor that, in turn, increased the likelihood of developing the second disorder. For example, research has shown that the presence of early-onset substance use disorders reduces the likelihood of completing high school, entering tertiary education, and completing tertiary education (Kessler, Foster, Saunders, & Stang, 1995). Difficulties encountered because of poor educational achievement might subsequently increase the likelihood of other problems, such as depression and continued substance use problems.

Similarly, persons who are alcohol dependent may be more likely to lose their jobs because of poor work performance or absenteeism. Indeed, one of the criteria for DSM-IV substance use disorders is disruption to, or failure to, complete roles such as occupational requirements (American Psychiatric Association, 1994). Unemployment could then lead to depression because of the lack of a regular income and perceived damage to their career.

Common factors

Common risk factors may well explain an association between two disorders (Caron & Rutter, 1991; Kessler, 1995; Mueser et al., 1998). If disorders are predominantly the result of a set of risk factors and these sets are the same or similar for two disorders, it may well be the case that 'comorbidity' reflects the fact that the pathways by which persons develop one disorder are the same as those by which they develop another. These common factors might be biological, personality, social and environmental, or a combination of these factors.Top of page

Biological factors

Neurotransmitter function

There is suggestive evidence that common physiological factors may explain the co-occurrence of different substance use disorders (homotypic comorbidity). This is plausible given that different substances act upon similar brain loci and upon the same neurotransmitter systems (Koob & Moal, 1997; Krishnan-Sarin, Rosen, & O'Malley, 1999; Nutt, 1997). Furthermore, some of the underlying neural substrates of mental disorders and substance use disorders are similar. There is considerable evidence that both substance use disorders and mental disorders are characterised by disturbances in monoamine neurotransmitter function (Doris, Ebmeier, & Shajahan, 1999; Iqbal & van Praag, 1995; Koob & Moal, 1997; Koob & Le Moal, 2001). Some have argued that one reason for comorbidity between alcohol use disorders and anxiety disorders may be reduced serotonin function (Tollefson, 1991).

Genetic factors

The possibility of a common genetic vulnerability to problematic use of different substances was examined in a sample of male twins (True et al., 1999; Tsuang et al., 1998). One of these studies examined the genetic and environmental contributions to illicit substance abuse of, and dependence on, cannabis, stimulants, sedatives, opiates and psychedelics (Tsuang et al., 1998). It found that while the vulnerability to dependence upon different substance types had some unique (drug-specific) genetic effects (0% for psychedelics, 5% sedatives, 9% stimulants, 11% cannabis and 38% heroin) there was a significant common genetic component. This comprised 6% of the variance for heroin use disorders, 22% for cannabis, stimulants, sedatives, and 26% for psychedelic use disorders. Analysis revealed that a 'common vulnerability' model provided the simplest explanation of the data, with around one third of the variance of this common vulnerability caused by genetic effects.

A similar analysis of alcohol and nicotine dependence (True et al., 1999) found that there was a significant common genetic vulnerability (r = 0.68) to both nicotine and alcohol dependence among male twins, with 26% of the variance in the risk for alcohol dependence shared with the genetic risk of nicotine dependence. This research needs to be replicated among female twins.

Twin studies have also provided some evidence that there are common genetic influences upon substance use disorders and mental disorders (i.e. for heterotypic comorbidity). For example, research has suggested that common genetic factors increase the risk of alcohol dependence, anxiety symptoms, and affective symptoms (Tambs, Harris, & Magnus, 1997).

A twin study of women also found that there were significant common genetic factors implicated in the comorbidity between major depression and tobacco smoking (K. Kendler et al., 1993). This study found that the heritability of liability to tobacco smoking and major depression was 55% and 48%, respectively. Analyses were conducted to examine whether there was a causal relationship between tobacco smoking in major depression, or whether common factors accounted for the association that was observed between the two. The best explanation of the co-occurrence of tobacco smoking and major depression in this sample was a common genetic factor. There was no evidence of common environmental factors. The correlation between smoking and major depression due to these genetic factors was estimated at + 0.56 (K. Kendler et al., 1993).Top of page

Individual factors

Temperament is commonly associated with substance use and mental health, particularly the trait of neuroticism. Persons scoring high on neuroticism have been characterised as more anxious, worrying, depressed and moody (Eysenck & Eysenck, 1991). Persons who are heavy substance users score higher on neuroticism than those who are not (Francis, 1996). Persons who suffer from mood and anxiety disturbances also have higher levels of trait neuroticism, and a considerable part of the liability to both mood and anxiety disorders is explained by higher levels of trait neuroticism (Andrews, 1996; Andrews, Stewart, Allen, & Henderson, 1990).

Social and environmental factors

Common genetic influences or individual factors play an incomplete part in explaining comorbidity.Twin studies have also shown that shared environmental factors increase the likelihood of both alcohol dependence and major depression among women (Tambs et al., 1997; True et al., 1999; Tsuang et al., 1998). Tsuang (1998) and colleagues found that two thirds of the common vulnerability to different types of illicit drug use disorders was explained by shared environmental factors.

This is not surprising, given that there is a wealth of evidence that a number of factors are common to both mental disorders and substance use disorders. For example, social disadvantage is more common among persons who are problematic substance users (Institute of Medicine, 1996); who meet criteria for mood disorders and anxiety disorders (Blazer, 1995; Kessler et al., 1994; Weissman, Livingston Bruce, Leaf, Flroio, & Holzer, 1991); and who meet criteria for psychotic disorders, and there is evidence to suggest that this is not merely because of social drift after developing the disorder (Mueser et al., 1998). For all these groups of disorders, studies have shown that there are higher rates of separation and divorce, and a lower likelihood that persons will be married or in a defacto relationship (Blazer, 1995; Jablensky, Sartorius, & Ernberg, 1991; Kessler et al., 1994; Weissman et al., 1991).

There is also a number of other factors that have been similarly associated with substance use disorders and with mental disorders, such as parental psychiatric illness and family dysfunction (Fergusson, Horwood, & Lawton, 1990; Fergusson, Horwood, & Lynskey, 1994; Rutter, 1987; Velez, Johnson, & Cohen, 1989). It is possible that these social factors serve to increase the apparent 'comorbidity' of mental disorders.

Kendler and colleagues (1993) also found that common genetic influences explained the co-occurrence of nicotine dependence and major depression. Another study examined this issue using data from a longitudinal study of adolescents from Christchurch, New Zealand (Fergusson, Lynskey, & Horwood, 1996). It examined the association between nicotine dependence and major depression while controlling for a large number of demographic variables, family background characteristics, and personal characteristics. It found — in apparent contrast to the Kendler study — that the co-occurrence of the two could be almost completely explained by common environmental factors, and that the most parsimonious explanation of the relationships between the two did not include a causal relationship.

While this may appear to be a contradiction of the Kendler study, it must be borne in mind that genetic and environmental factors are not independent. There is evidence, for example, of a genetic influence both upon exposure to stressful life events, and in responses to them (Kendler, 1998; Kendler et al., 1995; K. S. Kendler et al., 1993). Hence, in controlling for a large number of environmental factors, Fergusson and colleagues may well have been controlling for some of the genetic influences upon both nicotine dependence and major depression.What is clear from both of these studies, regardless of which sort of influence accounted for the comorbidity (environmental and/or genetic influences), both studies agreed in that there was no evidence that major depression caused nicotine dependence or vice versa.

A similar conclusion was reached by Lynskey and colleagues (1998) in an examination of liability to alcohol, tobacco and cannabis use using the same New Zealand cohort. This study found that the simplest explanation of the relationship between alcohol, tobacco and cannabis use was a 'common vulnerability' model of increased liability to the use of the three substances, which could be completely explained by a large number of environmental factors included in the analyses (Lynskey et al., 1998).Top of page


While at present there remains much that is not known about the causes of comorbidity, there is increasing evidence to suggest that simple causal hypotheses may not easily explain the association. There is a broad convergence of risk factors for both problematic substance use and mental disorders; a plausible hypothesis for the comorbidity between these disorders is that substance use and mental disorders (mood disorders, anxiety disorders, personality disorders and psychotic disorders) share common risk factors and life pathways. A number of longitudinal cohort and twin studies have explicitly examined this hypothesis and have concluded that common factors explain the comorbidity between alcohol, tobacco and cannabis use (Lynskey et al., 1998); dependence on different illicit drugs (Tsuang et al., 1998); alcohol and nicotine dependence (True et al., 1999); and nicotine dependence and major depression (Fergusson et al., 1996; K. Kendler et al., 1993).