Comorbidity is defined as the co-occurrence of one or more disorders in the same child or adolescent either at the same time or in some causal sequence (Kessler, 1995; Ollendick & King, 1994). In relation to internalising disorders, data from clinical samples points to a high overlap between these disorders and substance use disorders, independent of whether the referred problem is a substance use (Regier et al., 1990) or an internalising disorder (Bibb & Chambless, 1986). However, the frequency and nature of this comorbidity can be highly variable in substance misuse groups, ranging from acute internalising disorders at referral that appear secondary to the substance use disorder and quickly remit in treatment, leaving the "pure" substance use problem to run its course, to longstanding internalising disorders that may underlie the substance use disorder. Contamination by referral issues thus makes clinical studies unsuitable for obtaining community estimates of the comorbidity between substance use and internalising disorders and researchers must turn to epidemiological studies. Two of the most up-to-date and comprehensive of these were the Epidemiological Catchment Area study and the National Comorbidity Survey in the United States (see Kessler, 1995). These surveys were consistent in showing that the lifetime comorbidity odds-ratio of having both an internalising and a substance use disorder ranged from approximately 2.5 to 3.5. Thus, one has approximately three times the chance of suffering a substance use disorder if one has an internalising disorder, and vice versa, compared to a disorder-free person. These odds-ratios are means collapsed across specific mood and anxiety disorders and substance use disorders. They would be considerably higher if calculated according to the presence of any type of internalising disorder, and may be higher for social phobia and panic/depression in particular.
Externalising disorders have also been demonstrated to be strongly and consistently associated with substance use disorders (Glantz, Weinberg, Miner, & Colliver, 1999). For instance, results from the US National Comorbidity Survey (Kessler et al., 1996) indicate that nearly 60% of those with a lifetime diagnosis of conduct disorder also had at least one lifetime diagnosis of an addictive disorder. The same study estimates that comorbidity between Antisocial Personality Disorder and substance use disorders is even higher (83.6%), while accounts of the rates of comorbidity between Attention Deficit Disorder (ADD) and substance use disorders is less clear, possibly due to the complex relationship between ADD and a range of conduct disorder cluster behaviours.
It is also possible that the pathways to substance use disorders through internalising problems and externalising problems are interweaving. Recent research has shown that anxiety and depression may feature in externalising problems in young people far more significantly than has traditionally been acknowledged. For example, measures of attention deficit problems are highly confounded by the presence of anxiety problems (Perrin & Last, 1992), and internalising problems can enhance externalising problems through adolescence (Loeber, Russo, Stouthamer-Loeber, & Lahey, 1994). Unfortunately, longitudinal studies that simultaneously consider early internalising and externalising disorders as predictors of later substance use disorders are not available, and should be a research priority.
Given the presence of some shared risk factors between substance use, internalising and externalising disorders, and the prevalence of comorbidity amongst these disorders, a risk reduction approach to substance abuse prevention would predict that interventions that target the shared risk factors and/or comorbid conditions could help to reduce the incidence of substance use disorders. However, the design of specific preventive interventions would depend upon the nature of the causal links between other mental disorders and substance use disorders. Possible mechanisms of comorbidity are discussed next.
Causal models of comorbidity
Traditional approaches to the prevention of substance use disorders
Prevention of substance use disorders through comorbid pathways
Top of page
Causal models of comorbidityKessler and Price (1993) have proposed a model of four potential causal links between comorbid disorders, each of which has implications for the design of joint preventive efforts. Firstly, one type of disorder may lead directly to another. Thus, the abuse of certain drugs (cocaine, psychostimulants) can directly produce panic symptoms. For the purposes of this chapter, it is difficult to conceive of internalising or externalising disorders directly causing substance use disorders. Secondly, comorbidity can occur due to indirect effects of one disorder on another. Thus, social fears may lead directly to the abuse of drugs as a self-medication strategy. While evidence in this regard is limited to descriptive clinical studies, it is highly likely that this direct path is characteristic of a substantial proportion of those with substance use disorders. However, it should be noted that the reverse has been noted, whereby substance use disorders exacerbate anxiety and depression, at least in the short term. Thirdly, one disorder may be associated with contexts that potentiate the likelihood of another. Thus, disruptive behaviour problems and conduct disorder may lead to exposure to deviant peer groups that increase risk for substance use disorders. Depression may lead to an erosion of social networks that potentiates isolation and thus solitary drug taking. Fourth, comorbid conditions may share common causes. That is, problems may develop on a trajectory with each of the comorbid conditions representing different developmental stages of this trajectory. Recent research has demonstrated that generalised anxiety and depression share a genetic vulnerability (Kendler, 1996). As has been argued in this paper, it is possible, given the occurrence of shared risk factors, that anxiety, affective and conduct disorders represent earlier problems in the development of substance use disorders.
Clearly, Kessler and Price's (1993) model reflects a putative structure that is unlikely to be so distinctive in reality. Thus, two comorbid disorders may share some common causal variables, as well as having indirect effects on each other, and influencing contexts that serve to exacerbate or diminish the other disorder. Further, their model has very different implications at clinical versus population/ epidemiological levels. Patterns of inter-causality will differ from person to person, and clinicians have long been in the habit of sorting out the causal sequences of anxiety, depression, and substance abuse as a treatment guide to working with the individual client. At the population level, and thus with regard to preventive interventions aimed at large populations, any one causal pathway will explain only part of the variance in comorbidity. However, this may be enough to justify its influence on the design of large-scale community interventions.
No studies specifically designed to look at developmental causal sequences linking substance use with internalising disorders could be located. Perhaps the closest study in the literature comes from Catalano et al, (1996) who showed that a 'social development' model that emphasises social competence through late childhood and adolescence was the best predictor of substance use disorders in the late teen years. However, there exists enough indirect evidence to make some useful speculations. First, apart from transient anxiety or depression directly resulting from the abuse of specific substances, such disorders tend to precede substance use disorders developmentally. Secondly, it should be noted that in terms of comorbidity within internalising disorders, several studies have shown that anxiety problems typically precede and are risk factors for depressive disorders, although the reverse has not been found (Angst, Vollrath, Merikangas, & Ernst, 1990; Cole, Peeke, Martin, Truglio, & Seroczynski, 1998; Hagnell & Graesbeck, 1990). Thirdly, anxiety disorders and their early signs can be identified in childhood and many emerge as clear disorders in late childhood and early adolescence. Depression is relatively rare before middle adolescence and shares its initial onset period with substance use disorders, that is, in the teen years. Thus, it is likely that a pathway through anxiety disorders, depression, and then substance use disorders represents one pathway to substance use disorders that characterises many sufferers. Consequently, early intervention for internalising disorders, in particular the early signs of anxiety problems, may hold potential for reducing substance use problems in the community.
The causal pathway linking substance use disorders and externalising disorders has been more fully researched and documented in the literature. There exists a relatively clear developmental trajectory for substance use disorders that begins with early child behaviour problems, conduct problems, and attention deficit problems, high sensation seeking, and social adversity. For example, Reinherz, Giaconia, Carmola-Hauf, Wasserman and Paradis (2000) studied data from 360 respondents followed prospectively over a 17 year period to determine factors that would predict drug disorders in early adulthood. It was found that child behaviour problems such as hyperactivity, poor concentration, aggression, and hostility displayed at age six were predictive of substance disorders for both males and females at age 21. Other factors at age six predicting later substance problems included low socio-economic status, being born to young parents, and having a larger family size. In addition, a study by Windle (1990) found that antisocial behaviour in early adolescence predicted substance problems in late adolescence. Thus, based on the above evidence, it would be predicted that early intervention for childhood externalising problems could also hold potential for reducing the incidence of substance use disorders in the community.
In the following sections, implications of the comorbidity between substance use, internalising and externalising disorders for prevention efforts for substance use disorders will be considered. Traditional efforts to prevent substance use disorders will be briefly reviewed within a risk factor model. Prevention strategies based on the ideas presented in this paper regarding the interrelationships between substance use, internalising, and externalising disorders will then be addressed.
Top of page
Traditional approaches to the prevention of substance use disordersUntil quite recently, principal attempts to prevent substance use disorders in the community have focused on two of the risk factors identified for such disorders (outlined previously) — societal norms/laws and social influence (association with people who use drugs) (Botvin, 2000; Hawkins et al., 1992). With regard to societal norms/laws, prevention programs have been designed to manipulate the supply and availability of substances, change the legal consequences of substance use, and educate consumers about the adverse consequences of drug use (Hawkins et al., 1992).To address social influence issues, programs have been developed to teach young people social skills for resisting peer pressure to use substances (Hawkins et al., 1992). Unfortunately, prevention efforts targeted at these two risk factors have been shown to be ineffective (e.g., Bangert-Drowns, 1988). In fact, in some instances young people have been shown to increase their interest in substances following participation in these programs (e.g., Stuart, 1974).
The prevention programs based on psychoeducation and social skills training have been criticised for failing to change the developmental context experienced by children and youth (Hawkins et al., 1992). Given the literature already reviewed in this paper, it is clear that there are a variety of individual, familial, and interpersonal risk factors that can combine to produce a substance problem. A number of the early environmental risk factors that produce risk for substance use disorders are also risk factors for internalising and externalising disorders in young people, and these latter disorders themselves can be risk factors for the development of substance problems. It has been argued here that there are multiple pathways to substance use disorders that can begin with internalising and externalising disorders in children. Thus, it would seem reasonable to expect that effective prevention of substance use disorders could involve early intervention and prevention efforts for these comorbid conditions and associated risk factors, thus potentially interrupting the developmental trajectory for substance misuse. It is to this topic that we now turn.
Prevention of substance use disorders through comorbid pathwaysAs outlined above, a risk-reduction approach to prevention would predict that preventive interventions for both common mental disorders in young people hold promise for reducing the incidence of substance use disorders. Preventive interventions are categorised by either of two common systems. The traditional model examines prevention from the perspective of onset of disorder (Caplan, 1964). In this model, prevention can be implemented at three levels. The first level, primary prevention, intercedes prior to the onset of a disorder in order to reduce the likelihood of development of psychopathology. Secondary prevention is implemented once problems have been identified, but before these problems become severe. Finally, tertiary prevention involves treatment of current disorders with the aim of shortening the duration of the disorder and preventing relapse.
A second and subsequent model organises prevention initiatives based upon sample catchment boundaries (Mrazek & Haggerty, 1994). Within this model, a prevention program aimed at reaching a broad section of the community and applied to all individuals is considered a universal prevention program. An example would be a parent program to improve coping skills in parents and children. Alternatively, an indicated prevention specifically targets individuals who are at high risk for a disorder such as anxiety. A child who is behaviourally inhibited could be considered 'at risk' for anxiety. Thirdly, a selected prevention program targets people who are considered to be high-risk status based upon group membership, rather than individual characteristics. With respect to anxiety, this could include individuals who have been exposed to a natural disaster. With respect to conduct disorder, this could include children from low SES families. This review will discuss programs in terms of universal, indicated, and selected prevention, as at present this is the most widely used model.
There are advantages and disadvantages associated with the use of different types of intervention. An advantage of universal programs is that no selection procedures are needed and thus stigmatisation is unlikely to result. However, such programs are likely to be more expensive from both a financial and a human resource perspective. Importantly, and of ethical concern, without careful and thoughtful design, a universal program risks the possibility of doing harm to healthy people. Shochet and O'Gorman (1995) have argued that a guiding principle of any intervention must be to quarantine harm. Especially in initial trials when outcomes of prevention initiatives remain uncertain, it is imperative that, above all, people are not worse off as a result of participating in the program. For example, concern is often expressed about possible iatrogenic effects of suicide prevention programs when applied universally to young people.Top of page
Indicated or selected programs target those individuals most likely to be in need of assistance, thus optimising the use of financial and human resources. Additionally, indicated or selected programs increase the probability of identifying and intervening with individuals who otherwise may have gone unnoticed and progressed to a more severe level of dysfunction. Within some contexts, indicated and selected programs are termed 'early intervention', especially if some level of dysfunction already exists within the sample. However, the selection procedures associated with selected and indicated programs carry the risk of stigmatising or labeling individuals.
A number of criteria for developing prevention programs have been formulated by Simeonsson (1994), beginning with clear understanding of risk, protective factors, and characteristics of the targeted population. These factors inform the formulation of the prevention program. The design of choice is a randomised-controlled trial within a longitudinal study. Finally, adequate monitoring of the implementation and evaluation of the outcomes of the prevention program provides a guide for future development.