Chapter 2 Melanie E. Bennett, Jason Peer, and Selvija Gjonbalaj-Marovic Research indicates that a substantial percentage of the general population with a lifetime psychiatric disorder has a history of some other disorder (Kessler, 1997; Kessler et al., 1994), and more than half of patients in psychiatric treatment meet criteria for more than one diagnosis (Wolf, Schubert, Patterson, Marion, & Grande, 1988). The issue of comorbidity broadly refers to combinations of any types of psychiatric disorders that co-occur in the same individual. A diagnostic pair that has received significant attention over the past two decades is that of mental illness and substance abuse. The term dual diagnosis describes individuals who meet diagnostic criteria for an Axis I or Axis II mental disorder (or disorders) along with one or more substance use disorders. Although dual diagnosis has likely been a prevalent and persistent condition for a long time, it began to receive attention as both a clinical problem and research domain some 25 years ago. Today, we find ourselves with many years of research and thinking about the frequent association between mental and substance use disorders, how this association complicates the provision of mental health and substance abuse treatment services, and the impact of this association on all aspects of psychopathology research and clinical practice. In this chapter, we review data on rates of dual diagnosis, both generally and for specific domains of disorders, as well as discuss some of the ways in which dual diagnosis impacts the course, prognosis, assessment, and treatment of adult psychopathology. Finally, we review current research and thinking on the etiology of dual diagnosis and highlight clinical and research directions. A recurring theme throughout this chapter is the many ways that the conceptualization of dual diagnosis can change depending on definitions, diagnostic frameworks, assessment tools, and the sample being evaluated. Over time and as these domains have changed, thinking about what mental health and substance use disorders should be included within the conceptualization of dual diagnosis, as well as how to think about the impact of dual diagnosis on outcomes, has evolved and shifted. Our goal here is to review the literature on dual diagnosis, to present ideas that can challenge and perhaps expand the thinking on what constitutes dual diagnosis, and to provide consideration of how such expansion may impact thinking on psychopathology and diagnosis. There are several methodological issues to consider when reviewing the literature on dual diagnosis and evaluating its findings. These include the way that the sample under study shapes the findings, the range of methods used to assess psychiatric and substance use disorders and how different methods and measures shape study findings, and the critical impact of one’s definition of dual diagnosis on study findings. Data on the epidemiology of dual diagnosis come from both epidemiological and clinical studies, each of which has benefits and drawbacks. Several large-scale epidemiological studies examining rates of dual diagnosis in general population samples have been carried out since the mid-1980s. These studies provide representative information on rates of mental illness and substance use disorders, use structured diagnostic interviews, and generate results that are reliable and relevant to the population as a whole. Most of the information on rates of dual diagnosis comes from studies of clinical populations. Although such studies are not representative of the general population, they provide valuable information on the types of problems that are faced by individuals in treatment, as well as on the links between dual diagnosis, service utilization, the impact on illness, and treatment outcome. Importantly, individuals with multiple disorders are more likely to seek treatment, a condition known as Berkson’s fallacy (Berkson, 1949), so that estimates of the prevalence of comorbid disorders will be higher in clinical samples. Relatedly, factors such as inpatient or outpatient status and chronicity of illness may affect rates of dual diagnosis found in clinical samples. For example, research on patients with schizophrenia has found that more severely impaired inpatients are less likely to abuse substances than patients who are less ill (Mueser et al., 1990). Dual diagnosis rates have also been found to differ by setting, with hospital emergency rooms reflecting higher estimates than other settings (Barbee, Clark, Crapanzano, Heintz, & Kehoe, 1989; Galanter, Castaneda, & Ferman, 1988). In addition, several demographic variables correlate with substance abuse, and differences in these variables in clinical samples can influence prevalence rates. For example, gender and age both correlate with substance abuse: Males and those of younger age are more likely to abuse substances. Because studies of comorbidity in schizophrenia often use samples of inpatients who are more likely to be male, the comorbidity rate in schizophrenia found in research with clinical samples may be inflated, because males are both more likely to have substance use disorders and more likely to be inpatients in psychiatric hospitals. Another sample-related methodological issue involves the split between the mental health treatment system and the substance abuse treatment system, and the impact that this separation has on dual diagnosis research. The literature on dual diagnosis really includes two largely separate areas of investigation: research on substance abuse in individuals with mental illness, and research on mental illness in primary substance abusers. In order to get an accurate picture of dual diagnosis and its full impact on clinical functioning and research in psychopathology, both aspects of this literature must be examined. The methods used to determine psychiatric and substance use diagnoses can influence findings. The types of diagnostic measures used include structured research interviews, nonstructured clinical interviews, self-report ratings, and reviews of medical records. Although structured interviews are the most reliable method of diagnosis (Mueser, Bellack, & Blanchard, 1992), research with clinical samples will often employ less well-standardized assessments. Relatedly, studies measure different substances in their assessments of dual diagnosis, typically including alcohol, cocaine, heroin, hallucinogens, stimulants, and marijuana. Importantly, some substances are not typically considered in assessments of dual diagnosis. For example, nicotine is usually not considered a substance of abuse in dual diagnosis research, despite the high rates of use among individuals with both mental illness (Lasser et al., 2000) and substance abuse (Bien & Burge, 1990), as well as a growing literature that suggests that nicotine dependence has links, perhaps biological in nature, to both major depression (Quattrocki, Baird, & Yurgelun-Todd, 2000) and schizophrenia (Dalack & Meador-Woodruff, 1996; Ziedonis & George, 1997). Others have found elevated rates of psychiatric and substance use disorders in smokers (Keuthen et al., 2000). Taken together, factors such as the type of problematic substance use assessed, the measures that are used, and the specific substances that are included in an assessment all contribute to varying meanings of the term dual diagnosis. Definitions of what constitutes dual diagnosis are far from uniform. Studies often use differing definitions and measures of substance use disorders, making prevalence rates diverse and difficult to compare. For example, definitions used to determine rates of dual diagnosis vary, ranging from problem use of a substance based on the frequency of use or the number of negative consequences experienced as a result of use, to abuse or dependence based on formal diagnostic criteria. This is a particularly important issue when formal diagnostic criteria for substance use disorders are used to assess dual diagnosis. Currently the two most widely used systems for psychiatric diagnosis and classification are the Diagnostic and Statistical Manual (DSM; American Psychiatric Association [APA], 1994), used primarily in the United States, and the International Classification of Diseases (ICD) (World Health Organization [WHO], 1993), used primarily in other countries. There are several issues related to these diagnostic frameworks that are must be considered. Having different systems internationally means that studies done in different countries will use different definitions of psychiatric and substance use disorders, presenting a challenge when making comparisons across studies. Importantly, these diagnostic systems are similar but not identical, and there are important features of each that influence how substance use disorders are diagnosed that can, in turn, influence rates of dual disorders. The diagnostic criteria for alcohol use disorders from the fourth edition of DSM (DSM-IV) and the 10th edition of the ICD (ICD-10) provide a good example (see Hasin, 2003, for a complete review). Both systems include a diagnosis of alcohol dependence that requires at least three symptoms be present from a list of six (ICD-10) or seven (DSM-IV) that include both physiological symptoms such as tolerance and withdrawal as well as nonphysiological symptoms such as impaired control, giving up other important activities, and continued use in the face of physical or psychological consequences. Research has shown good reliability between DSM and ICD for dependence diagnoses (Hasin, 2003). In contrast, these systems differ with respect to diagnoses of alcohol abuse (as it is called in DSM-IV) or harmful use (as found in ICD-10). DSM-IV requires “recurrent use” that leads to at least one of four possible consequences (failure to fulfill obligations, use in situations in which it is physically hazardous, use-related legal problems, use despite recurrent social/interpersonal problems) in order to meet a diagnosis of alcohol abuse. ICD-10 is more specific in its definition of “recurrent” (“The pattern of use has persisted for at least 1 month or has occurred repeatedly within a 12-month period”) but less specific in describing those consequences that are “harmful,” requiring only “clear evidence that alcohol use contributed to physical or psychological harm” and that a “pattern of use has persisted” (WHO, 1993). Such differences in criteria have no doubt contributed to findings of lower reliability between these categories (Hasin, 2003). Moreover, these sorts of differences are especially important when considering how rates of dual disorders compare across nations and cultures (see Room, 2006, for a review). Another critical consideration is that both DSM and the ICD are changing systems. The preceding example was based on diagnostic criteria found in DSM-IV and ICD-10. However, in May 2013, a new edition of DSM, DSM-5 (APA, 2013), was published, with significant implications for dual diagnosis. DSM-5 includes new disorders (for example, disruptive mood dysregulation disorder has been added to the chapter on Depressive Disorders), changes to the symptom requirements needed to meet criteria for some disorders (for example, a diagnosis of schizophrenia now requires at least one positive symptom), and consolidation of related conditions into a single diagnostic classification (for example, autistic disorder, Asperger’s disorder, childhood disintegrative disorder, and pervasive developmental disorder not otherwise specified have been merged within Autism Spectrum Disorder). Overall, DSM-5 now has fewer total diagnoses—the revisions yielded 15 new diagnoses, 2 diagnoses that were not continued, and 28 that were combined. These sorts of changes can impact definitions and prevalence rates of dual diagnosis. For example, in DSM-5, the bereavement exclusion for major depressive disorder has been eliminated. This has the potential to yield some additional cases of major depression or to identify major depression earlier following loss. Although it is not yet clear how such a change would definitively impact rates of dual diagnosis, it is possible that this change would yield additional cases of major depression that could, in turn, yield more individuals affected by dual diagnosis. As the changes to diagnosis codified in DSM-5 begin to be used, we will be able to examine whether and how dual diagnosis is impacted. There were also several changes in the diagnosis of substance use disorders. The title of the chapter dedicated to alcohol and drug diagnoses was revised from Substance Use Disorders (DSM-IV) to Substance-Related and Addictive Disorders (DSM-5); this revision reflects several changes that will have a substantial impact both on how disorders related to alcohol and drug use are diagnosed and in the inclusion of other conditions as addictions. To begin with, whereas DSM-IV had separate diagnostic groups for substance abuse and dependence, DSM-5 combines abuse and dependence into one diagnostic category (substance use disorder) that is rated on a continuum from mild to severe, with the severity rating based on the number of criteria met (mild = two to three criteria met; moderate = four to five criteria met; severe = six or more criteria met). There also have been changes to the diagnostic criteria for substance use disorder. For example, the criterion of “recurrent legal problems” that was part of the substance abuse diagnosis in DSM-IV was not retained in DSM-5, and a new criterion—“craving or a strong desire or urge to use a substance”—was added. In addition, while in DSM-IV one or more criteria needed to be met to earn a diagnosis of substance abuse and three or more needed to be met for substance dependence, now in DSM-5, a diagnosis of substance use disorder requires that 2 criteria from a list of 11 be met. These changes will have a substantial impact on dual diagnosis. Combining abuse and dependence into one diagnostic category can have the effect of identifying what might have previously been thought of as a less problematic or separate state (abuse) as a more serious condition that is potentially related to something more serious (dependence), with rates of dual diagnosis increasing as a result. Importantly, the longer list of diagnostic criteria and the severity rating symptom are a visible indication of the thinking that mild substance-related problems are fundamentally related to more severe ones. That is, these are not separate diagnostic categories but ones that are related, different not in the makeup of the conditions but by degree. This pathway from mild to moderate to severe substance use disorder has the potential to impact dual diagnosis by including less severe substance use problems within the dual diagnosis framework. Whereas problem or heavy substance use that did not meet the threshold for a dependence diagnosis previously may not have been considered as part of a definition of dual diagnosis, now less severe expressions of substance use are considered a disorder that would be captured within the definition of dual diagnosis. The shift from conceptualizing substance use disorders from abuse versus dependence categories to a continuous dimension was informed by a growing literature of psychometric analyses of DSM-IV substance use disorder criteria. Studies using traditional factor analysis, latent trait modeling, and item response theory (IRT) analysis all point to a single continuous factor as the most parsimonious representation of substance-abuse and dependence symptom criteria. Specifically, several studies using exploratory factor analysis found that a single factor solution including both abuse and dependence symptoms had the best fit for alcohol use disorders (Borges et al., 2010; Harford, Yi, Faden, & Chen, 2009; Lagenbucher et al., 2004) and marijuana and cocaine use disorders (Lagenbucher et al., 2004). In addition, these findings remained consistent across different national contexts (Borges et al., 2010) and age groups (Harford et al., 2009; Martin, Chung, Kirisci, & Lagenbucher, 2006). Further, studies utilized an analysis of substance use criteria based on IRT and found that the substance abuse and dependence criteria did not fall on a continuum of severity in a manner consistent with the DSM-IV conceptualization of abuse and dependence typology for alcohol use disorders (Borges et al., 2010; Lagenbucher et al., 2004; Harford et al., 2009) and cocaine and marijuana use disorders (Lagenbucher et al., 2004; Martin et al., 2006). Specifically, for each item, IRT quantifies where an item falls on a latent variable representing severity of substance use disorders and the ability of each item to discriminate between subsequent levels of that latent variable. The DSM-IV typology would predict that those individuals with less severe substance use disorders would be less likely to endorse items associated with substance dependence (e.g., those items with greater severity) and that these items would be more closely related with regard to severity. However, when models were specified to order items in such a way, they tended to have poorer fit (Lagenbucher et al., 2004). Finally, these studies have found some evidence that criteria may be interpreted differently by respondents depending on age or national context (Borges et al., 2010; Lagenbucher et al., 2004). In sum, this greater psychometric scrutiny of DSM symptom criteria has led to a reconceptualization of substance use disorders as a continuum and it is likely that similar analysis in the future will lead to greater refinements and precision in diagnostic criteria. In turn, this notion of a continuum from mild to moderate to severe substance use disorder has the potential to impact definitions of dual diagnosis. Where, in the past, substance abuse diagnosis may not have been considered in dual diagnosis, we may now instead need to consider dual diagnosis as a continuous variable as opposed to a categorical one. We may need to consider the correlation between the severity of substance use disorder and specific Axis I diagnostic categories to best understand how different disorders are interrelated or interact with each other. Another important change in DSM-5 is expansion of the chapter on substance use disorders to include gambling disorder, based on a body of research that now indicates that many of the same underlying neurobiological processes that are activated and make substance use biologically reinforcing are found in problem gamblers, and that symptoms of problematic gambling are similar in many ways to those of substance use disorders (Petry, 2006). The inclusion of a non-substance-related addiction alongside substance-related ones provides an interesting set of questions. Should gambling disorders now be included in the thinking about dual diagnosis? What does this mean for (the likely increase in) prevalence rates of dual diagnosis that might accompany such a change, our understanding of the causes of dual diagnosis, and implications for assessment and treatment? If the most widely accepted definition of dual diagnosis involves meeting diagnostic criteria for an Axis I or Axis II mental disorder (or disorders) along with one or more substance use disorders, and gambling disorder is fundamentally similar in many ways to alcohol or drug use disorders, does the definition need to be broadened to include non-substance addictions? The inclusion of gambling disorder in DSM-5 and the questions this change raises for definitions of dual diagnosis illustrate a broader issue: What one identifies as a diagnosis is key to any definition of dual diagnosis. Gambling provides but one example. Another example can be seen with tobacco dependence. Although there is a large literature documenting higher rates of smoking among individuals with psychiatric disorders (Farrell et al., 2003), tobacco dependence has not traditionally been included as part of the definition of dual diagnosis. This is generally because with such high rates of smoking among those with psychiatric disorders, including it would yield extremely high rates of dual diagnosis, potentially sapping the concept of its meaning (if everyone has dual diagnosis, does it become less important?). However, with growing evidence that smoking may be related to the neurophysiology of some psychiatric disorders (Wing, Wass, Soh, & George, 2012), it may be that tobacco dependence represents an especially important substance use disorder to include as part of a definition of dual diagnosis. The greater attention to gambling disorder and tobacco dependence illustrates the ways in which changing ideas about mental health problems and addictions can impact rates of dual diagnosis. A final issue here concerns the definition of dual diagnosis itself. As noted earlier, dual diagnosis has generally been used to describe the co-occurrence within an individual of an Axis I or Axis II mental disorder (or disorders), with one or more substance use disorders. Within this definition lie many combinations of psychiatric and substance use disorders. It is important to keep in mind that the literature on dual diagnosis is, by necessity, simplified and compartmentalized by the definitions used to guide it. Over the past 25 years, several large-scale epidemiological studies of mental illness have examined rates of dual diagnosis, including the Epidemiologic Catchment Area Study (ECA; Regier et al., 1990), the National Comorbidity Survey (NCS; Kessler et al., 1994), the National Comorbidity Survey Replication (NCS-R; Kessler & Merikangas, 2004), and the National Longitudinal Alcohol Epidemiology Survey (NLAES; Grant et al., 1994). Although each study differs somewhat from the others in methodology, inclusion/exclusion criteria, and diagnostic categories assessed (see Table 2.1 for a brief description of methods for these studies), we can take several points from this literature that can contribute to our thinking about and understanding of dual diagnosis. Table 2.1 Methods of Several Major Epidemiological Studies on Dual Diagnosis First, epidemiological studies consistently show that dual diagnosis is highly prevalent in community samples. Each of these studies finds that people with mental illness are at greatly increased risk of having a co-occurring substance use disorder, and people with a substance use disorder are likewise much more likely to meet criteria for an Axis I mental disorder. For example, the Epidemiologic Catchment Area Study (ECA; Regier et al., 1990) was the first large-scale study of comorbidity of psychiatric and substance use disorders in the general population, and documented high rates of dual diagnosis among both individuals with primary mental disorders and those with primary substance use disorders. Overall, individuals with a lifetime history of a mental illness had an odds ratio of 2.3 for a lifetime history of alcohol use disorder and 4.5 for drug use disorder, a clear illustration of how those with mental illness are at substantially increased risk of having a comorbid substance use diagnosis. When examined by type of disorder, antisocial personality disorder (ASP) showed the highest comorbidity rate (83.6%), followed by bipolar disorder (60.7%), schizophrenia (47.0%), panic disorder (35.8%), obsessive-compulsive disorder (32.8%), and major depression (27.2%). Further analysis of ECA data (Helzer, Robbins, & McEvoy, 1987) found that men and women with posttraumatic stress disorder (PTSD) were 5 times and 1.4 times more likely, respectively, to have a drug use disorder as were men and women without PTSD. Substantial rates of dual diagnosis were also found in primary substance abusers (Regier et al., 1990). Overall, 37% of individuals with an alcohol disorder and 53% of those with a drug use disorder had comorbid mental illness. Further analyses (Helzer & Pryzbeck, 1988) found that among those with alcohol use disorders, the strongest association was with ASP (odds ratio = 21.0), followed by mania (OR = 6.2) and schizophrenia (OR = 4.0). Like the ECA study, the NCS and the NLAES found markedly high rates of dual diagnosis. NCS (Kessler et al., 1994) findings showed that respondents with mental illness had at least twice the risk of lifetime alcohol or drug use disorder, with even greater risk for individuals with certain types of mental illnesses. Findings were similar for primary substance abusers: The majority of respondents with an alcohol or drug use disorder had a history of some nonsubstance use psychiatric disorder (Kendler, Davis, & Kessler, 1997; Kessler, 1997). Overall, 56.8% of men and 72.4% of women with alcohol abuse met diagnostic criteria for at least one psychiatric disorder, as did 78.3% of men and 86.0% of women with alcohol dependence (Kendler et al., 1997). Moreover, 59% of those with a lifetime drug use disorder also met criteria for a lifetime psychiatric disorder (Kessler, 1997). Likewise, the NLAES (Grant et al., 1994; Grant & Harford, 1995) found that, among respondents with major depression, 32.5% met criteria for alcohol dependence during their lifetime, as compared to 11.2% of those without major depression. Those with primary alcohol use disorders were almost 4 times more likely to be diagnosed with lifetime depression, and the associations were even stronger for drug use disorders: Individuals with drug dependence were nearly 7 times more likely than those without drug dependence to report lifetime major depression (see Bucholz, 1999, for a review). Such findings clearly illustrate that rates of dual diagnosis are significant among individuals with mental illness and primary substance abusers, and that many types of psychiatric disorders confer an increased risk of substance use disorder. Overall these studies find that a psychiatric diagnosis yields at least double the risk of a lifetime alcohol or drug use disorder. A second important feature of rates of dual diagnosis is that they appear to be persistent. Examining how rates persist or change over time is important for several reasons. When the ECA and NCS findings were first published, the findings of high rates of both single and dual disorders were significant because they illustrated the many ways in which the understanding, assessment, and treatment of mental illness and substance use disorders were incomplete. The NCS in particular came under increased scrutiny, given that the rates it found for mental illness were even higher than those found by the ECA. Replications of these studies can demonstrate whether the high rates found in the first studies persist over time. In addition, since the first epidemiologic studies were conducted, DSM criteria have changed, leading to questions of how these diagnostic changes might impact illness rates. Finally, seeking treatment for mental distress, as well as use of medications for symptoms of depression and anxiety, are now more widely discussed and accepted than they were 10 to 20 years ago, and it is unclear how changing attitudes might impact rates of dual disorders. Findings from several replications of large epidemiologic studies indicate that even with changes in diagnostic criteria and attitudes about psychological distress, rates of dual disorders remain high. For example, the NCS was recently replicated in the NCS-R. The NCS-R (Kessler & Merikangas, 2004) shared much of the same methodology as the original NCS, repeated many questions from the original survey, and included additional questions to tap DSM-IV diagnostic criteria. Conducting these studies 10 years apart allows for an examination of the stability in rates of dual diagnosis, as well as how changes in assessment and diagnostic criteria impact the prevalence of dual diagnosis and other comorbid conditions. Comparisons of data from both studies illustrate the persistent nature of dual diagnosis. That is, although specific values have changed from one interview to another, the overall picture of dual diagnosis remains the same: Prevalence rates are high, and people with mental illness remain at greatly increased risk for developing substance use disorders. For example, for major depressive disorder (MDD; Kessler et al., 1996) found that 38.6% of respondents who met criteria for lifetime MDD also had a diagnosis of substance use disorder based on NCS data, whereas 18.5% of respondents who met criteria for 12-month MDD also had a diagnosis of substance use disorder. Results from the NCS-R confirmed the high prevalence rates of dual diagnosis in people with MDD: 24.0% of those with lifetime MDD also met criteria for a substance use disorder, and 27.1% of those who met criteria for 12-month MDD also met criteria for a substance use disorder (Kessler, Berglund, et al., 2005). Although the exact percentages change over time, the rates for dual MDD and substance use diagnoses remain strikingly high over the 10 years between studies. Similar comparisons can be made between the NLAES and a more recent NIAAA survey called the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC; Grant et al., 2004). The NESARC stressed the need to ensure that diagnoses of mood and anxiety disorders were independent from substance use disorders. A comparison of the two studies shows that dual mood/anxiety and substance use disorders continue to be highly prevalent in community samples. For example, in the NLAES (Grant & Harford, 1995), respondents with a past-year diagnosis of major depression had a 21.36% rate of a co-occurring alcohol use disorder, compared with 6.92% of those without 12-month major depression (OR = 3.65). Similarly, high odds ratios were found for 12-month major depression and drug use disorders (Grant, 1995). Importantly, results of the NESARC confirm the persistent association of substance use disorders and affective disorders. People who met criteria for any 12-month mood disorder were 4.5 times more likely to meet criteria for substance dependence (range of 3.4 to 6.4 for the four mood disorders assessed). People who met criteria for any 12-month anxiety disorder were 2.8 times more likely to meet substance dependence criteria (range of 2.2 to 4.2 for the five anxiety disorders assessed). Examining the results in terms of prevalence rates is similar: 19.97% of those with any 12-month mood disorder had at least one substance use disorder (SUD), and 14.96% of those with any 12-month anxiety disorder had at least one SUD. Similarly, 19.67% of those with a 12-month SUD had at least one mood disorder, and 17.71% had at least one anxiety disorder. Overall, comparisons from replications of large epidemiologic studies illustrate the persistence of dual diagnosis over time. A third issue that is highlighted in some epidemiological studies is the fact that the term and typical understanding of dual diagnosis may not accurately reflect the nature and complexity of the problem of co-occurring mental and SUDs. That is, co-occurring disorders can take many forms, and limiting attention to a particular number or combination of problems may restrict what we can learn about the links and interactions between mental illness and SUDs. As discussed previously, the term dual diagnosis has most often been used to refer to a combination of one mental illness and one SUD. However, epidemiologic studies find high prevalence rates of three or more co-occurring disorders that include but are not limited to dual mental-SUD combinations. For example, Kessler and colleagues (1994) found that 14% of the NCS sample met criteria for three or more comorbid DSM disorders, and that these respondents accounted for well over half of the lifetime and 12-month diagnoses found in the sample. Moreover, these respondents accounted for 89.5% of the severe 12-month disorders, which included active mania, nonaffective psychosis, or other disorders requiring hospitalization or associated with severe role impairment. Other data from the NCS (Kessler et al., 1996) showed that 31.9% of respondents with lifetime MDD and 18.5% of those with 12-month MDD met criteria for three or more comorbid conditions. In the NCS-R, Kessler, Berglund, et al. (2005) found that 17.3% of respondents met criteria for three or more lifetime disorders. In addition, in examining projected lifetime risk of developing different DSM disorders, these authors reported that 80% of projected new onsets were estimated to occur in people who already had disorders. In examining 12-month disorders in the NCS-R sample, Kessler, Chiu, et al. (2005) found similar results: 23% of the sample met criteria for three or more diagnoses. As these authors state, “Although mental disorders are widespread, serious cases are concentrated among a relatively small proportion of cases with high comorbidity” (Kessler, Chiu, et al., 2005, p. 617). Taken together, these findings illustrate the importance of thinking about dual diagnosis in the context of the broader picture of comorbid conditions. People with dual mental and substance use disorders may in fact meet criteria for a combination of multiple mental and substance use disorders. Such aggregation of mental and SUDs in a small proportion of people should influence conceptualizations regarding the processes underlying dual and comorbid conditions. In addition, comorbidity appears to be influenced by severity of mental illness. Using data from the NCS-R, Kessler and colleagues (2003) examined differences in rates of comorbid disorders (including but not limited to SUDs) in respondents with MDD of differing levels of severity. Specifically, respondents who met criteria for 12-month MDD were classified as showing mild, moderate, severe, or very severe symptoms, based on scores on the Quick Inventory of Depressive Symptomatology Self Report (QIDS-SR) for the worst month in the past year. As severity level increased, so did rates of comorbidity, defined as the percentage of respondents with two or more comorbid 12-month disorders, including SUDs. Specifically, 34.9% of mild, 58.0% of moderate, 77.3% of severe, and 82.1% of very severe MDD cases met criteria for two or more comorbid disorders. This finding of increased severity as number of disorders increases was also found for 12-month diagnoses (Kessler, Chiu, et al., 2005). This trend is another reminder that dual diagnosis and comorbidity labels represent heterogeneous groups of people who differ in meaningful ways that likely have significance in terms of assessment, treatment, and etiology of mental and SUDs. In sum, epidemiological studies are now able to tell us not only that dual diagnosis is highly prevalent but also that rates of dual disorders persist over time. In addition, we now have ample evidence to suggest that talking about dual disorders is actually a simplification of a complex problem in that patients often have more than two psychiatric disorders as well as use, abuse, or are dependent on multiple substances. Such findings suggest that thinking about the causes of dual disorders may need to be broadened in order to be able to explain this range of diversity among dual-disordered patients. As noted earlier, how studies define dual diagnosis has a major impact on prevalence rates. In this section we review three developments that have had an important impact on prevalence rates of dual diagnosis: including tobacco dependence when defining dual diagnosis, cross-national epidemiological studies, and examination of general population data focused on specific demographic subgroups. Tobacco is the most widely used substance of abuse among individuals with mental health diagnoses (Farrell et al., 2003) and rates of smoking for individuals with mental health disorders are 2 to 3 times higher what is found in the general population (Lawrence, Mitrou, & Zubrick, 2009). Smokers with mental health disorders have longer smoking histories and, importantly, are less likely to quit: Whereas tobacco use and dependence has been on the decline in the general population, rates have generally remained stable among smokers with mental health disorders (Dickerson et al., 2012; Secades-Villa et al., 2013; Williams, Steinberg, Griffiths, & Cooperman, 2013). Despite these high prevalence rates, tobacco dependence has not often been considered in research and thinking about dual diagnosis. There are several reasons for this. First, with such high rates of prevalence of tobacco dependence among those with mental health disorders, including tobacco dependence would serve to define a great number of people with a mental health disorder as having a dual diagnosis. Second, there is a complicated history of tolerance and even encouragement of smoking among those with mental health disorders, as many treatment providers believed that individuals would experience symptom relapse or display increased disruptive behavior if not able to smoke, ideas that have since been found to be untrue (el-Guebaly, Cathcart, Currie, Brown, & Gloster, 2002). Third, in the past, tobacco dependence was not fully considered an addiction in the same way that chronic and harmful use of alcohol or illicit drugs was seen. Indeed, for a long time, rates of smoking in the general population were similarly high and the negative consequences of smoking, because they were often far removed from the time of the actual behavior (i.e., the health effects were more likely to manifest after many years of smoking), were not thought about in the same way as the consequences of disordered drinking or illicit drug use. There are now a range of reasons to consider tobacco dependence when thinking about and studying dual diagnosis. The addictive nature of tobacco/nicotine is now widely known, and the negative health effects of tobacco dependence and other negative consequences for individuals with mental health disorders—cost, stigma, second-hand smoke—have been identified (see Bennett, Wilson, Genderson, & Saperstein, 2013, and Graham, Frost-Pineda, & Gold, 2007 for reviews). As information and knowledge about tobacco dependence has increased, rates of smoking in the general population have decreased dramatically, making tobacco dependence a decreasing problem for many people but still an alarming problem for individuals with mental health disorders. Estimates suggest that individuals with mental health disorders smoke 50% of all cigarettes consumed in this country (Lasser et al., 2000) and that individuals with mental health disorders have a disproportionately difficult time quitting smoking (see Mackowick, Lynch, Weinberger, & George, 2012 for a review). Moreover, research suggests that nicotine may have a unique connection with genetics and brain functioning in some psychiatric disorders including depression and schizophrenia (de Leon & Diaz, 2012; Wing et al., 2012), and that quitting might be more difficult for individuals with mental health disorders due to biological factors that may make smoking more reinforcing (Berg, Sentir, Cooley, Engleman, & Chambers, 2013) or withdrawal symptoms more problematic (Leventhal, Ameringer, Osborn, Zvolensky, & Langdon, 2013; Weinberger, Desai, & McKee, 2010) than what is experienced by other smokers. For these reasons, we believe a comprehensive review of dual diagnosis should include information on tobacco dependence, and have provided this in the following sections. The epidemiological studies just listed provide information on rates of tobacco dependence among those with mental health disorders. Lawrence and colleagues (Lawrence et al., 2009) used data from the NCS-R to examine rates of comorbid tobacco dependence and mental health disorders (affective, anxiety, and substance use disorders) as measured by the WHO CIDI. Among respondents with a mental health disorder in the past 12 months, 40.1% were current smokers (almost double the 21.3% smoking prevalence in adults without a mental health disorder in the past 12 months). Rates were highest among those with SUDs (63.6%), followed by individuals with affective disorders (45.1%) and anxiety disorders (37.6%). Goodwin, Zvolensky, Keyes, & Hasin (2012) examined rates of tobacco dependence among those with mental health disorders in the NESARC sample. Diagnoses of specific phobia, personality disorder, major depressive disorder, and bipolar disorder were associated with increased odds of nicotine dependence, as well as with increased risk of persistent nicotine dependence over a 3-year follow-up (Goodwin, Pagura, Spiwak, Lemeshow, & Sareen, 2011). In an examination of associations between smoking and personality disorders using the NESARC sample, Pulay and colleagues (2010) found that the nicotine dependence had the strongest associations with schizotypal, borderline, narcissistic, and obsessive-compulsive personality disorder diagnoses. The preceding review of epidemiological studies focuses on findings from studies conducted in the United States. An interesting and important complement to findings from the United States comes from findings from the WHO World Mental Health Surveys (WMHS; Kessler, Haro, Heeringa, Pennell, & Üstün, 2006), an assessment of diagnosis of mental health disorders using the WHO Composite International Diagnostic Interview (CIDI) carried out in 28 countries around the world with a sample of over 150,000 respondents. Findings from this international survey confirm results from the United States: Mental health disorders are prevalent in all regions of the world, and rates of comorbid SUDs are substantially higher among those with mental health disorders. For example, in an examination of prevalence rates and correlates of bipolar spectrum disorder in a subset of the WHO WMHS sample (n = 61,392), Merikangas and colleagues (Merikangas et al., 2011) found that 36.6% of those with a bipolar spectrum disorder met criteria for a co-occurring substance use disorder. Other findings from the WHO WMHS replicate findings reviewed above: Comorbidity of multiple mental health disorders, including diagnostic combinations that include psychiatric and SUDs, is associated with greater impairments such as higher risk for suicide (Nock et al., 2009). As more cross-national data are collected and analyzed, it will be important to expand on this work to examine if and how different processes and cultural factors influence the development and persistence of dual diagnosis. A special population that is important to consider when thinking about dual diagnosis is older adults. With people living longer, it is important to understand how rates of dual disorders persist or change over time. Findings from the NCS-R show lower rates of diagnosis of mood, anxiety, and substance use disorders for those 65 and older compared to those 18–64 (Gum, King-Kallimanis, & Kohn, 2009). Results from the NESARC (Chou & Cheung, 2013) find that although the overall prevalence of major depressive disorder is low in older adults (2.95% for the past year and 8.82% lifetime), among those with lifetime major depressive disorder, 20% met criteria for an alcohol use disorder. In line with the idea of including gambling disorder within the definition of dual diagnosis, several studies using NESARC data have found high rates of dual mental and gambling disorders among older adults (Chou & Cheung, 2013; Pilver, Libby, Hoff, & Potenza, 2013), highlighting the way in which older age may impact the traditional definition of dual diagnosis. For example, Pietrzak and colleagues (Pietrzak, Morasco, Blanco, Grant, Petry, 2006) examined over 10,000 older adults in the NESARC survey and found significantly elevated rates of SUDs (alcohol, tobacco, and drugs), as well as mood, anxiety, and personality disorders among those with disordered gambling. Dual diagnosis may play an important role in a subset of older adults who experience chronic mental health disorders. Mackenzie and colleagues (Mackenzie, El-Gabalawy, Chou, & Sareen, 2013) examined factors that may contribute to persistence of three mental health diagnoses—mood, anxiety, and substance use disorders—over time in older adults surveyed in the NESARC study. Their findings showed that although most disorders were not persistent from one assessment to a second one that was conducted 3 years later, comorbidity of another mental health disorder (including SUDs) was a significant predictor of persistence of any disorder over time. The fact that dual diagnosis is fairly common in the general population serves to highlight the even higher rates found in treatment settings. Clinical studies of dual diagnosis have assessed general psychiatric patients, patients with specific psychiatric disorders, and primary substance abusing patients. Clinical studies of dual diagnosis over the past 20 years indicate that one-third to three-quarters of general psychiatric patients may meet criteria for comorbid psychiatric and substance use disorders, depending on the diagnostic makeup of the sample and the level of chronicity represented (Ananth et al., 1989; Galanter et al., 1988; McLellan, Druley, & Carson, 1978; Mezzich, Ahn, Fabrega, & Pilkonis, 1990; Safer, 1987). Rates seem to fall in the higher end of this range for samples comprising more impaired patient populations. For example, Ananth and colleagues (1989) found that 72.0% of a sample of patients with schizophrenia, bipolar disorder, and atypical psychosis received a comorbid substance use diagnosis. Mezzich et al. (1990) conducted a large-scale assessment of dual diagnosis in more than 4,000 patients presenting for evaluation and referral for mental health problems over an 18-month period and found substantial rates of dual diagnosis among several diagnostic subsamples. The highest rates were seen among patients with severe mental illnesses such as bipolar disorder (45% diagnosed with an alcohol use disorder and 39% diagnosed with a drug-use disorder) and schizophrenia or paranoid disorders (42% and 38% were diagnosed with alcohol and other SUDs, respectively). However, dual diagnosis was also pronounced in other patient groups. Specifically, 33% of patients with major depression were diagnosed with an alcohol use disorder, and 18% were diagnosed with a drug use disorder. Among patients with anxiety disorders, 19% and 11% were diagnosed with alcohol and other substance use disorders, respectively. Rates of dual diagnosis have been extensively studied among patients with severe mental illness, including schizophrenia (Dixon, Haas, Weiden, Sweeney, & Frances, 1991; Mueser et al., 1990), bipolar disorder (Bauer et al., 2005; McElroy et al., 2001; Salloum & Thase, 2000; Vieta et al., 2000), and major depression (Goodwin & Jamison, 1990; Lynskey, 1998; Merikangas, Leckman, Prusoff, Pauls, & Weissman, 1985; Swendsen & Merikangas, 2000). Findings show that dual diagnosis is common in such samples. Mueser et al. (1990) evaluated 149 patients with schizophrenia spectrum disorders and found that 47% had a lifetime history of alcohol abuse, whereas many had abused stimulants (25%), cannabis (42%), and hallucinogens (18%). Dixon and colleagues (1991) found that 48% of a sample of schizophrenia patients met criteria for an alcohol or drug use disorder. Chengappa, Levine, Gershon, and Kupfer (2000) evaluated the prevalence of substance abuse and dependence in patients with bipolar disorder. Among patients with bipolar I, 58% met abuse or dependence criteria for at least one substance, and 11% abused or were dependent on three or more substances. In the bipolar II group, the rate of dual diagnosis was approximately 39%. Baethge and colleagues (2005) followed a group of first-episode bipolar I patients and found that about one-third of the sample had a SUD at the baseline assessment, and that patients using two or more substances showed poorer outcomes over the 2 years of the study. Bauer and colleagues (Bauer et al., 2005) conducted structured interviews with a large sample of inpatient veterans with bipolar disorder across 11 sites (n = 328) to examine rates of comorbid anxiety and substance use disorders. Results showed high rates of current (33.8%) and lifetime (72.3%) SUDs in the sample, along with a rate of 29.8% meeting criteria for multiple current disorders. Hasin, Endicott, and Lewis (1985) examined rates of comorbidity in a sample of patients with affective disorder presenting for treatment as part of the National Institute of Mental Health Collaborative Study of Depression and found that 24% of these patients reported serious problems with alcohol and 18% met diagnostic criteria for an alcohol use disorder. In an examination of patients with major depression, bipolar disorder, and controls participating in the National Institute of Mental Health Collaborative Program on the Psychobiology of Depression, Winokur and colleagues (1998) found that affective disorder patients had substantially higher rates of dual SUDs than did controls. Dual diagnosis is also common among patients with anxiety disorders. In their review of studies of dual anxiety and SUDs, Kushner, Sher, and Beitman (1990) found that rates differ by type of anxiety disorder, with social phobia (ranging from 20% to 36% rate of dual diagnosis) and agoraphobia (ranging from 7.0% to 27.0% rate of dual diagnosis) showing the highest rates of substance abuse comorbidity. Others have found a 22% rate of lifetime alcohol use disorder among patients with social phobia (Himle & Hill, 1991), a 10% to 20% rate for patients with agoraphobia (Bibb & Chambless, 1986), and up to a 12% rate of lifetime alcohol dependence among patients with obsessive-compulsive disorder (Eisen & Rasmussen, 1989). In addition, more attention is being given recently to dual substance abuse and PTSD in clinical samples. A growing literature examining this diagnostic combination finds high rates of dual diagnosis among patients with PTSD, with some findings as high as 80% (Keane, Gerardi, Lyons, & Wolfe, 1988). Research both with samples of veterans with PTSD and samples of women with assault- or trauma-related PTSD show strikingly high rates of comorbid substance abuse and dependence (see Stewart, Pihl, Conrod, & Dongier, 1998, for a review), as well as other disorders. Moreover, Breslau, Davis, Peterson, and Schultz (1997) interviewed a sample of 801 women and found that PTSD significantly increased the likelihood for later alcohol use disorder. Research with community mental health patients with several mental illnesses shows extremely high rates of co-occurring PTSD (14%–53%, usually undiagnosed; Grubaugh, Zinzow, Paul, Egede, & Frueh, 2011). Research documents high rates of dual diagnosis among those with eating disorders. Higher rates of drug use have been found in samples of individuals with eating disorders than in controls (Krug et al., 2008), and studies of clinical samples show high overall rates of alcohol and drug use disorders. For example, Grilo, White, and Masheb (2009) assessed DSM-IV lifetime and current psychiatric disorder comorbidity in patients with binge-eating disorder and found that more than 73% of respondents had at least one lifetime diagnosis and 43% had at least one current psychiatric diagnosis, with almost 25% of the sample meeting criteria for a lifetime SUD. Several studies have shown variation in rates of SUDs across the different types of eating disorders—anorexia nervosa, bulimia nervosa, binge-eating disorder—and subsets of disorders within these (Root et al., 2010). In addition, rates of comorbidity, including that with SUDs, may be associated with eating disorder severity, with those with more severe symptoms of eating disorder showing the highest rates of comorbid SUDs (Spindler & Milos, 2007). Some of the highest rates of comorbidity are found for patients with personality disorders, especially antisocial personality disorder (ASP). Studies show that comorbid ASP accelerates the development of alcoholism (Hesselbrock, Hesselbrock, & Workman-Daniels, 1986), and that 80% of patients with ASP have a history of problem use of alcohol (Schuckit, 1983). In a recent review of studies on dual SUDs and borderline personality disorder (BPD), Trull and colleagues (Trull, Sher, Minks-Brown, Durbin, & Burr, 2000) found that, across studies, more than 48% of patients with BPD met criteria for alcohol use disorders, and 38% of those with BPD met criteria for a drug use disorder. In a recent reanalysis of the NESARC data, Trull and colleagues (Trull, Jahng, Tomko, Wood, & Sher, 2010) found high rates of dual personality disorders and SUDs. More than one-quarter of those with ASP (26.65%) met criteria for drug dependence, although even higher rates of drug dependence were found for those with histrionic (29.72%) and dependent (27.34%) personality disorders. Important work is now illustrating the need to think more broadly about dual and multiple comorbidities and how these span Axis I and Axis II disorders. For example, a recent examination of data on generalized anxiety disorder from the NESARC study (Alegría et al., 2010) found a lifetime prevalence rate of generalized anxiety disorder with comorbid SUD of 2.04%. However, those with generalized anxiety disorder and comorbid SUD showed significantly higher rates of comorbidity of other psychiatric disorders than did those with generalized anxiety disorder alone, including higher lifetime rates of bipolar disorder, panic disorder, and ASP. A similar pattern was found for social anxiety in the NESARC data (Schneier et al., 2010): Respondents with both social anxiety disorder and comorbid alcohol use disorder were significantly more likely to earn diagnoses of mood, other anxiety, psychotic, and personality disorders, as well as additional SUDs and pathological gambling. Goldstein, Compton, and Grant (2010) examined rates of ASP in individuals with PTSD and how this combination of disorders affected risk of further comorbid psychiatric disorders. Compared to those with PTSD only, those with PTSD+ASP showed much higher rates of additional comorbid diagnoses. Specifically, those with PTSD+ASP met criteria for, on average, 5.7 additional lifetime Axis I diagnoses, whereas those with PTSD only met criteria for only 2.3 additional lifetime Axis I diagnoses. Rates of additional Axis II diagnosis were similar (2.5 additional Axis II diagnoses for those with PTSD+ASP versus 0.7 for those with PTSD only). Wildes, Marcus, and Fagiolini (2008) examined rates of eating disorders in individuals with bipolar disorder and found that a subset of individuals with bipolar disorder and loss of control over eating showed elevated rates of substance use disorders. In another study of individuals with bipolar disorder, Bauer and colleagues (2005) interviewed inpatients using the Structured Clinical Interview for DSM-IV and found that rates of comorbidity with SUDs were high (33.8% current, 72.3% lifetime), but that almost 30% of respondents had comorbid bipolar, SUD, and anxiety disorders. Such findings illustrate the importance of expanding our thinking regarding dual diagnosis into multiple comorbidities. As noted earlier, rates of tobacco dependence are high among individuals with mental health disorders. Studies indicate that more than 60% of adults with schizophrenia, 40% with bipolar disorder, and 30% with major depression in the United States smoke, compared to fewer than 20% in the overall adult population (Dickerson et al., 2012; Heffner, Strawn, DelBello, Strakowski, & Anthenelli, 2011). Similarly, research has found higher rates of smoking among those with a range of anxiety disorders, although directionality is unclear—there is evidence to suggest that smoking is a risk factor for panic disorder and generalized anxiety disorder (Moylan, Jacka, Pasco, & Berk, 2012). Some speculate that those with mental health disorders who smoke may experience a more severe subtype of disorder than those who do not smoke. For example, Strong and colleagues (2010) examined differences in smoking behavior between those with no history of major depressive disorder, those with a single episode, or those with recurrent major episodes in 1,560 participants in the NCS-R. Those with comorbid recurrent major depression reported more smoking, greater nicotine dependence, more comorbid mental health disorders, and greater impairment in functioning than those with no or a single episode of major depression. Saiyad and El-Mallakh (2012) studied the impact of smoking on symptoms of bipolar disorders (n = 134) and found that those with bipolar disorder who smoked reported more severe symptoms of anxiety, depression, and mania. Others have suggested that smoking may be a marker for a more chronic form of schizophrenia (Dalack & Meador-Woodruff, 1996) or a subtype of bulimia nervosa characterized by more depression and alcohol abuse (Sandager et al., 2008). Substance-abusing patients in treatment are a heterogeneous group, encompassing a range of substances and levels of severity. Nonetheless, researchers have found high rates of dual disorders across diverse samples of patients seeking substance abuse treatment (Arendt & Munk-Jorgensen, 2004; Falck, Wang, Siegal, & Carlson, 2004; Herz, Volicer, D’Angelo, & Gadish, 1990; Mirin, Weiss, Griffin, & Michael, 1991; Mirin, Weiss, & Michael, 1988; Penick et al., 1984; Powell, Penick, Othmer, Bingham, & Rice, 1982; Ross, Glaser, & Stiasny, 1988; Rounsaville, Weissman, Kleber, & Wilber, 1982; Watkins et al., 2004; Weissman & Myers, 1980). Findings of lifetime rates of psychiatric disorder range from 73.5% of a sample of cocaine abusers (Rounsaville et al., 1991) to 77% of a sample of hospitalized alcoholics (Hesselbrock, Meyer, & Keener, 1985) to 78% of a sample of patients in an alcohol and drug treatment facility (Ross, Glaser, & Germanson, 1988). Findings of current psychiatric disorder are similarly high, ranging from 55.7% of a group of cocaine abusers (Rounsaville et al., 1991) to 65% in a general substance-abusing sample (Ross, Glaser, & Germanson, 1988). Further reflecting their diagnostic heterogeneity, substance abusers in treatment experience a range of comorbid psychiatric disorders. Among the most widely studied have been affective disorders, and treatment-seeking substance abusers show high rates of both major depression (Hasin, Grant, & Endicott, 1988; Hesselbrock et al., 1985; Merikangas & Gelernter, 1990; Mezzich et al., 1990; Miller, Klamen, Hoffmann, & Flaherty, 1996; Rounsaville, Weissman, Wilber, Crits-Christoph, & Kleber, 1982; Weissman & Myers, 1980) and bipolar disorder (Strakowski & DelBello, 2000). Miller and colleagues (1996) surveyed a sample of more than 6,000 substance abuse treatment patients from 41 sites and found that 44% had a lifetime history of major depression. In a review of comorbidity of affective and substance use disorders, Lynskey (1998) found that the prevalence of unipolar depression among patients receiving treatment for substance use disorders ranged from a low of 25.8% for lifetime depression in a sample of 93 alcohol-dependent men (Sellman & Joyce, 1996) to a high of 67% meeting a lifetime diagnosis of major depression among a sample of 120 inpatients (Grant, Hasin, & Harford, 1989). Busto, Romach, and Sellers (1996) evaluated rates of dual diagnosis in a sample of 30 patients admitted to a medical facility for benzodiazepine detoxification and found that 33% met DSM-II-R criteria for lifetime major depression. Results from large studies of treatment-seeking substance abusers find that these patients show 5 to 8 times the risk of having a comorbid bipolar diagnosis (see Strakowski & DelBello, 2000, for a review). The importance of dual mental illness in substance-abusing samples lies in its link to functioning and treatment outcome. Burns, Teesson, and O’Neill (2005) studied the impact of dual anxiety disorders and/or depression on outcome of 71 patients seeking outpatient alcohol treatment. Comorbid patients showed greater problems at baseline (more disabled, drank more heavily) than did substance-abuse-only patients, a difference that persisted at a follow-up assessment 3 months later. An extensive literature documents high rates of comorbid personality disorders in primary substance abusers (Khantzian & Treece, 1985; Nace, 1990; Nace, Davis, & Gaspari, 1991), especially ASP (Herz et al., 1990; Hesselbrock et al., 1985; Liskow, Powell, Nickel, & Penick, 1991; Morgenstern, Langenbucher, Labouvie, & Miller, 1997; Penick et al., 1984; Powell et al., 1982). In their evaluation of a large sample of treatment-seeking substance abusers, Mezzich and colleagues (1990) found that 18% of those with alcohol use disorders and almost 25% of those with drug use disorders met criteria for an Axis II disorder. Busto and colleagues (1996) found that 42% of their sample of patients undergoing benzodiazepine detoxification met DSM-III-R criteria for ASP. Morgenstern and colleagues (1997) assessed prevalence rates of personality disorders in a multisite sample of 366 substance abusers in treatment. Results showed that more than 57% of the sample met criteria for at least one personality disorder. ASP was the most prevalent (22.7% of the sample), followed by borderline (22.4%), paranoid (20.7%), and avoidant (18%) personality disorders. Moreover, the presence of a personality disorder doubled the likelihood of meeting criteria for a comorbid Axis I disorder. Brooner and colleagues (Brooner, King, Kidorf, Schmidt, & Bigelow, 1997) assessed psychiatric disorders in 716 opioid abusers on methadone maintenance therapy and found that 47% of the sample met criteria for at least one disorder, with ASP and major depression being the most common co-occurring diagnoses. In addition, psychiatric comorbidity was associated with more severe substance use disorder. Kokkevi, Stephanis, Anastasopoulou, & Kostogianni (1998) surveyed 226 treatment-seeking individuals with drug dependence in Greece and found a 59.5% prevalence rate of personality disorder, with more than 60% of these patients meeting criteria for more than one personality disorder. Furthermore, those with personality disorders were at twice the risk for meeting an additional Axis I diagnosis. Findings are similar with anxiety disorders, with high rates of comorbid phobias (Bowen, Cipywnyk, D’Arcy, & Keegan, 1984; Hasin et al., 1988; Ross, Glazer, & Stiasny, 1988), panic disorder (Hasin et al., 1988; Penick et al., 1984), and obsessive-compulsive disorder (Eisen & Rasmussen, 1989) documented in substance-abusing populations. Thomas, Thevos, and Randall (1999) reported a 23% prevalence rate of social phobias in a large study of both inpatients and outpatients with alcohol dependence. Substance abusers also appear to be especially affected by PTSD (Cottler, Compton, Mager, Spitznagel, & Janca, 1992; Davis & Wood, 1999; Triffleman, Marmar, Delucchi, & Ronfeldt, 1995). In an analysis of cocaine-dependent patients in the National Institute on Drug Abuse Collaborative Cocaine Treatment Study, Najavitis and colleagues (1998) found that 30.2% of women and 15.2% of men met DSM-II-R criteria for PTSD. Recently, Back and colleagues (2000) found that 42.9% of a sample of cocaine-dependent individuals met criteria for PTSD, and Bonin and colleagues (Bonin, Norton, Asmundson, Dicurzio, & Pidlubney, 2000) found a 37.4% rate of PTSD in a sample of patients attending a community substance abuse treatment program. In sum, the literature clearly documents high rates of dual substance abuse and psychiatric disorders for a variety of psychopathological conditions and in a range of patient populations. Findings from epidemiological studies show that dual diagnosis is relatively common in the general population, and results of clinical studies illustrate the frequency of dual diagnosis among individuals in treatment. That rates of dual diagnosis are similarly high in both mentally ill and in primary substance-abusing populations serves to highlight the serious difficulties in having two separate and independent systems of care for mental illness and substance abuse (Grella, 1996; Ridgely, Lambert, Goodman, Chichester, & Ralph, 1998), because both populations of patients are quite likely to be suffering from both types of disorders. The importance of dual diagnosis lies in its negative impact on the course and prognosis of both psychiatric and SUDs, as well as its influence on assessment, diagnosis, and treatment outcome. Individuals with dual disorders show more adverse social, health, economic, and psychiatric consequences than do those with only one disorder, and they show more severe difficulties, often a more chronic course of psychiatric disorder, and a poorer response to both mental health and substance abuse treatment. In the next section, we review the ways that dual diagnosis impacts three general areas: patient functioning, clinical care, and research. Dual diagnosis has a profound impact on many domains of functioning. This section reviews the many ways that dual diagnosis affects symptoms of mental illness, course of mental illness over time, cognitive functioning, and compliance with treatment. Dual diagnosis severely impacts the severity and course of many disorders, especially among patients with serious mental illnesses such as schizophrenia, bipolar disorder, and recurrent major depression. Often these dually diagnosed individuals show a poorer and more chaotic course of disorder, with more severe symptoms (Alterman, Erdlen, Laporte, & Erdlen, 1982; Barbee et al., 1989; Hays & Aidroos, 1986; Negrete and Knapp, 1986), more frequent hospitalizations (Carpenter, Heinrichs, & Alphs, 1985; Drake & Wallach, 1989; Sonne, Brady, & Morton, 1994), and more frequent relapses than patients without co-occurring substance abuse (Linszen, Dingemans, & Lenior, 1994; O’Connell, Mayo, Flatow, Cuthbertson, & O’Brien, 1991; Sokolski et al., 1994). Haywood et al. (1995) found that substance abuse, along with medication noncompliance, was the most important predictor of more frequent rehospitalization among schizophrenia patients. Recently, Margolese and colleagues (Margolese, Malchy, Negrete, Tempier, & Gill, 2004) compared three groups of schizophrenia patients: those with current SUD, those with lifetime but not current SUD, and those with no current or history of SUD. Patients with current SUD showed more positive symptoms than both other patient groups, had higher scores on measures of depression as compared to the single diagnosis group, and were more likely than the single diagnosis group to be noncompliant with their medications. Winokur and colleagues (1998) found that patients with drug abuse and bipolar disorder had an earlier age of onset of bipolar disorder than those with bipolar disorder alone, as well as a stronger family history of mania. Nolen and colleagues (2004) rated patients with bipolar or schizoaffective disorder on severity of manic symptoms, severity of depressive symptoms, and number of illness episodes over a 1-year period (n = 258). Results showed that ratings for mania severity were associated with comorbid substance abuse. Lehman, Myers, Thompson, and Corty (1993) compared individuals with dual mental illness and substance use diagnoses to those with just a primary mental illness and found that the dual diagnosis group had a higher rate of personality disorder and more legal problems. Hasin, Endicott, and Keller (1991) followed 135 individuals with dual mood and alcohol use disorders who were originally studied as part of the National Institute of Mental Health Collaborative Study on the Psychobiology of Depression. Although most had experienced at least one 6-month period of remission of the alcohol disorder at some point during the follow-up period, most had relapsed after 5 years. Mueller and colleagues (1994) examined the impact of alcohol dependence on the course of major depression over 10 years among individuals with depression who participated in the National Institute of Mental Health Collaborative Depression Study. Those who were alcohol dependent at baseline had a much lower rate of recovery from major depression than those with major depression alone, illustrating the negative impact of alcohol use disorders on the course of major depressive disorder. Dual diagnosis is also a serious issue for patients with anxiety disorders such as PTSD (Najavitis, Weiss, & Shaw, 1997; Ouimette, Brown, & Najavitis, 1998). Overall, the combination of substance abuse and PTSD appears to be linked to higher rates of victimization, more severe PTSD symptoms in general, more severe subgroups of PTSD symptoms, and higher rates of Axis II comorbidity (Ouimette, Wolfe, & Chrestman, 1996). Saladin, Brady, Dansky, and Kilpatrick (1995) compared 28 women with both substance abuse and PTSD to 28 women with PTSD only and found that the dual diagnosis group reported more symptoms of avoidance and arousal, more sleep disturbance, and greater traumatic-event exposure than the PTSD-only group. Back and colleagues (2000) similarly found higher rates of exposure to traumatic events, more severe symptomatology, and higher rates of Axis I and Axis II disorders among cocaine-dependent individuals with PTSD as compared to those without lifetime PTSD. Moreover, evidence suggests that the combination of PTSD and cocaine dependence remains harmful over several years, with patients showing a greater likelihood of continued PTSD as well as revictimization several years after an initial substance abuse treatment episode (Dansky, Brady, & Saladin, 1998). Dual diagnosis also exerts a profound impact on overall life functioning. Patients with severe mental illnesses such as schizophrenia who abuse substances appear to be particularly hard hit in this regard (see Bradizza & Stasiewicz, 1997, for a review; Kozaric-Kovacic, Folnegovic-Smalc, Folnegovic, & Marusic, 1995). Drake and colleagues consistently have found that individuals with schizophrenia and comorbid substance abuse show substantially poorer life adjustment than do individuals with schizophrenia without substance abuse, and eat fewer regular meals (Drake, Osher, & Wallach, 1989; Drake & Wallach, 1989). Havassy and Arns (1998) surveyed 160 frequently hospitalized adult psychiatric patients and found not only high rates of dual disorders (48% of patients had at least one current substance use disorder; of these, 55.1% met criteria for polysubstance dependence) but also that dual diagnosis was related to increased depressive symptoms, poor life functioning, lower life satisfaction, and a greater likelihood of being arrested or in jail. Research similarly shows that patients with dual affective and alcohol use disorders show greater difficulties in overall functioning and social functioning than do patients with depression (Hirschfeld, Hasin, Keller, Endicott, & Wunder, 1990) or bipolar disorder (Singh, Mattoo, Sharan, & Basu, 2005). Newman, Moffitt, Caspi, & Silva (1998) examined the impact of different types of comorbidity (including but not limited to substance abuse–psychiatric disorder combinations) on life functioning in a large sample of young adults. Multiple-disordered cases showed poorer functioning than single-disordered cases in almost every area measured, including health status, suicide attempts, disruption in performance of daily activities, the number of months disabled because of psychiatric illness, greater life dissatisfaction, less social stability (more residence changes, greater use of welfare for support, greater rates of adult criminal conviction records), greater employment problems, lower levels of educational attainment, and greater reports of physical health problems. Weiss and colleagues (2005) examined the interplay between bipolar disorder and recovery from substance use disorders on a range of quality-of-life factors in a sample of 1,000 patients with current or lifetime bipolar disorder. Specifically, three groups were compared: those with no history of SUDs, those with past SUDs, and those with current SUDs. Results showed that the current-SUD group had the poorest functioning, and both SUD groups reported lower quality of life and higher lifetime rates of suicide attempts than did the non-SUD group. Moreover, the toxic effects of psychoactive substances in individuals with schizophrenia and bipolar disorder may be present even at use levels that may not be problematic in the general population (Lehman, Myers, Dixon, & Johnson, 1994; Mueser et al., 1990). Increasingly, clinicians and researchers are focusing on cognitive functioning in persons with dual disorders. There is a range of cognitive impairments associated with psychiatric disorders, particularly serious mental illness (e.g., schizophrenia, bipolar disorder). Individuals with schizophrenia spectrum disorders demonstrate cognitive impairments across a range of cognitive domains when compared to normative comparison samples (Heinrichs & Zakzanis, 1998). Although not as severe, individuals with affective disorders demonstrate similar impairments across a range of cognitive domains (Depp et al., 2007; Goldberg et al., 1993; Schretlen et al., 2007). These impairments are linked by modest to strong correlations to functional outcomes in schizophrenia (Green, 1996) and bipolar disorder (Dickerson et al., 2004; Dickerson, Sommerville, Origoni, Ringel, & Parente, 2001). There is also evidence that, in samples of individuals with primary SUDs, chronic or sustained substance use can contribute to cognitive impairment and resulting brain dysfunction (Bowden, Crews, Bates, Fals-Stewart, & Ambrose, 2001; Rogers & Robbins, 2001). Moreover, cognitive impairment has been implicated in substance-abuse-treatment outcomes in this population (Aharonovich, Nunes, & Hasin, 2003; Fals-Stewart & Schafer, 1992). Such findings suggest that substance use may exacerbate existing cognitive impairment, which may, in part, contribute to the poorer outcomes experienced by persons with co-occurring serious mental illness (SMI) and SUD. Although this recognition has prompted clinical research efforts to adapt and develop new substance abuse interventions designed to accommodate some of the cognitive and motivational impairments associated with SMI (e.g., Addington, el-Guebaly, Campbell, Hodgins, & Addington, 1998; Bellack, Bennett, Gearon, Brown, & Yang, 2006; Ziedonis & George, 1997), other research efforts have been directed toward further description and explication of the role of cognitive impairment in individuals with dual diagnosis. The majority of this work has focused on SMI samples. On the one hand, there is concern about the possible exacerbation of existing cognitive impairment resulting from substance use among individuals with dual disorders, suggesting that these individuals would demonstrate poorer cognitive functioning when compared to those without SUD. On the other hand, some data suggest that engaging in behaviors necessary to obtain access to substances requires a higher level of functioning (Dixon, 1999; Mueser et al., 1990), and, thus, these individuals with dual disorders would have better cognitive functioning than individuals without SUD. With regard to cognitive functioning, the data are in fact mixed and overall suggest that there are few differences between those with SMI who have a current or history of SUD and those who do not. In a meta-analysis of 22 studies investigating neurocognitive functioning among individuals with schizophrenia, Potvin, Joyal, Pelletier, and Stip (2008) found that there was no difference on a composite score of cognitive functioning between those with SUD and those without. Furthermore, they found few differences between groups on specific cognitive domains or specific cognitive measures. Depp et al. (2007) also found no differences in cognitive functioning among a sample of individuals with bipolar disorder with and without SUDs. In contrast, Carey, Carey, and Simons (2003), in a sample of individuals with schizophrenia spectrum and bipolar disorders, found that those with a current SUD or former SUD both demonstrated better cognitive functioning than those who had never used substances. The interpretation of these data is complicated by several substantive and methodological issues. First, there is some indication that impairment may vary depending on primary substance of abuse. Across the meta-analytic data, alcohol use was associated with poorer working memory performance, whereas cannabis use was associated with better problem solving and visual memory performance (Potvin et al., 2008). This finding is consistent with neuropsychological data from SUD samples without psychiatric diagnoses, where alcohol is associated with greater impairments than other drugs such as cocaine (Goldstein et al., 2004). However, types of substances are not always accounted for in dual disorder studies (e.g., Carey et al., 2003). Second, consistent with methodological limitations across the dual diagnosis research literature, the rigor with which samples have been characterized has been quite variable. Some studies have relied on chart diagnoses as opposed to diagnostic interview to identify SUD, few have verified drug status with urinalysis, and others have failed to characterize the severity, recency, or chronicity of substance use. With regard to the latter, analyses included in the Potvin et al. (2008) study indicated that as age increased, so did the cognitive impairment among those with SUD, suggesting that chronicity of use may be a moderating factor in cognitive functioning among dual disorders. Similarly, Carpenter and Hittner (1997) found that lifetime use of alcohol or cocaine (i.e., number of years of regular use) were the strongest predictors of cognitive impairment among a sample of individuals with mixed psychiatric diagnoses (affective and anxiety disorders) and SUDs. These latter findings raise the related question of how cognitive impairment changes over time as a result of substance use in individuals with dual disorders. Few investigations have addressed this question. Using a group comparison design with carefully characterized samples, Carey et al. (2003) found no difference in cognitive functioning between individuals with SMI and current SUD versus those with past history of SUD (defined as not meeting full criteria for the past 6 months). Peer, Bennett, and Bellack (2009) compared individuals with schizophrenia who met DSM-IV criteria for current cocaine dependence and those who met criteria for remission on a brief neuropsychological battery and found few differences. This study also included a parallel analysis of samples of individuals with affective disorders and cocaine dependence versus remission, which yielded similar results. Although these studies used rigorous diagnostic criteria to characterize the samples, they are limited by their cross-sectional nature. That is, they did not evaluate change within subjects in cognitive functioning as a result of discontinuation of substance use. At least two longitudinal studies have been conducted that address this question. A brief longitudinal study of inpatients with schizophrenia with or without current cocaine dependence at admission found few changes in cognition as a result of abstinence from cocaine over an 18-day study period (Cooper et al., 1999). Furthermore, there were few differences in cognition between groups at baseline or at follow-up. McCleery, Addington, and Addington (2006) followed 183 individuals with a first episode of psychosis over a 2-year study period and assessed cognition and substance use. Results indicated that cognition largely remained stable over time, while substance use declined over the study period. Together these findings suggest that cognitive functioning may be relatively static among individuals with dual disorders. Indeed, in the general SUD literature, longitudinal data suggest there are only slight and/or inconsistent improvements in neurocognitive functioning after a period of abstinence from substances (Bates, Voelbel, Buckman, Labouvie, & Barry, 2005; Di Sclafani, Tolou-Shams, Price, & Fein, 2002; Horner, 1999). There are at least two possible interpretations of these data: (1) given the significant cognitive impairment associated with SMI, substance use causes only minimal additional impairment; and (2) the toxic effects of substance use on cognition are not easily resolved following abstinence. In part, this research may be limited by the lack of sensitivity of the neuropsychological measures used for these particular research questions. With further advances in cognitive neuroscience, more refined measures that are more tightly linked to brain structures and functions impacted by chronic substance use will likely be developed (Rogers & Robbins, 2001). Although candidate brain structures and neurotransmitter pathways are increasingly being identified in the general SUD literature (e.g., Goldstein et al., 2004; Goldstein & Volkow, 2002), significantly more work is needed to understand the specifics of cognitive functioning in dual disorders, both with regard to preexisting impairment as well as a sequelae of chronic substance use. Substance abuse often interferes with compliance with both behavioral and psychopharmacological treatments. Lambert, Griffith, and Hendrickse (1996) surveyed patients on a general psychiatry unit in a Veterans Administration medical center and found that discharges against medical advice (AMA) were more likely to occur among patients with alcohol and/or substance use disorders. Pages and colleagues (1998) similarly assessed predictors of AMA discharge in psychiatric patients. The presence of SUD and a greater quantity and frequency of substance use were among the most important predictors. Owen and colleagues (Owen, Fischer, Booth, & Cuffel, 1996) followed a sample of 135 inpatients after discharge and found that medication noncompliance was related to substance abuse, and that this combination was significantly associated with lack of outpatient contact in the follow-up period. Specifically, those with dual diagnoses were more than 8 times more likely to be noncompliant with their medication. In a large-scale study of factors related to medication adherence in schizophrenia patients, Gilmer and colleagues (2004) found that substance abusers were less likely to be adherent to antipsychotic medication regimens than were schizophrenia patients who did not abuse substances.
The Problem of Dual Diagnosis
Methodological Issues in the Assessment of Dual Diagnosis
Sample Selection Influences Findings
Study Methods and Assessment Measures Influence Findings
There Is No Single Definition of Dual Diagnosis
Findings from Major Epidemiological Studies
Study
Years
Methods
ECA (Regier et al., 1990)
1980–1984
Surveyed more than 20,000 adults in five cities across the United States both in the community and in institutions. Trained interviewers used the Diagnostic Interview Schedule to determine DSM-III diagnoses. Included affective, anxiety, and schizophrenia-spectrum disorders.
NCS (Kessler et al., 1994)
1990–1992
Assessed 12-month and lifetime prevalence rates for a range of psychiatric disorders in more than 8,000 noninstitutionalized individuals ages 15–54 across 48 states using the Composite International Diagnostic Interview (CIDI) and based on DSM-III-R criteria.
NLAES (Grant et al., 1994)
1991–1992
Examined rates of co-occurrence of alcohol and drug use disorders and affective disorders in a general population sample. The NLAES is a household survey of more than 42,000 adults in the United States that utilized diagnostic interviews to assess DSM-IV diagnostic criteria for alcohol use disorders.
NCS-R (Kessler & Merikangas, 2004)
2001–2002
Nationally representative face-to-face household survey of more than 9,000 noninstitutionalized people ages 18 years or older. Diagnoses based on DSM-IV criteria assessed via CIDI interviews.
NESARC (Grant et al., 2004)
2001–2002
Nationally representative face-to-face survey of 43,093 noninstitutionalized respondents, 18 years of age or older, conducted by NIAAA. DSM-IV criteria for substance use disorders and nine independent mood and anxiety disorders were assessed with the Alcohol Use Disorders and Associated Disabilities Interview Schedule-DSM-IV Version (AUDADIS-IV), a structured diagnostic interview administered by lay interviewers.
Dual Diagnosis Is Highly Prevalent in Community Samples
High Prevalence Rates of Dual Diagnosis Persist Over Time
Dual Diagnosis Is Only One Part of the Comorbidity Puzzle
The Concept of Dual Diagnosis Has Expanded and Changed
Tobacco Dependence
Cross-National Epidemiological Studies
Prevalence of Dual Diagnosis in Older Adults
Findings from Studies of Clinical Samples
Dual Diagnosis in General Psychiatric Patients
Dual Diagnosis in Samples of Patients With Specific Disorders
Dual Diagnosis in Patients With Primary Substance Use Disorders
Clinical Impact of Dual Disorders
Impact of Dual Diagnosis on Functioning
Symptoms, Course of Illness, and Life Functioning
Cognitive Functioning
Treatment Noncompliance and Violence
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
