Prevalence and co-occurrence of substance use disorders and independent mood and anxiety disorders: results from the National Epidemiologic Survey on Alcohol and Related Conditions.

Background Uncertainties exist about the prevalence and comorbidity of substance use disorders and independent mood and anxiety disorders. Objective To present nationally representative data on the prevalence and comorbidity of Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM–IV) alcohol and drug use disorders and independent mood and anxiety disorders (including only those that are not substance induced and that are not due to a general medical condition). Design Face-to-face survey. Setting The United States. Participants Household and group quarters residents. Main Outcome Measures Prevalence and associations of substance use disorders and independent mood and anxiety disorders. Results The prevalences of 12-month DSM–IV independent mood and anxiety disorders in the U.S. population were 9.21 percent (95 percent confidence interval [CI], 8.78 percent–9.64 percent) and 11.08 percent (95 percent CI, 10.43 percent–11.73 percent), respectively. The rate of substance use disorders was 9.35 percent (95 percent CI, 8.86 percent–9.84 percent). Only a few individuals with mood or anxiety disorders were classified as having only substance-induced disorders. Associations between most substance use disorders and independent mood and anxiety disorders were positive and significant (p < .05). Conclusions Substance use disorders and mood and anxiety disorders that develop independently of intoxication and withdrawal are among the most prevalent psychiatric disorders in the United States. Associations between most substance use disorders and independent mood and anxiety disorders were overwhelmingly positive and significant, suggesting that treatment for a comorbid mood or anxiety disorder should not be withheld from individuals with substance use disorders.

S ubstance use disorders and mood and anxiety disorders are widespread among the general population [1][2][3] and are associated with substantial societal and personal costs. [4][5][6][7] Furthermore, national epidemiologic surveys [1][2][3] and numerous clinical studies [8][9][10][11][12] consistently indicate that substance use disorders and mood and anxiety disorders have strong associations when considered on a lifetime basis. However, consensus has not been achieved on the meaning and implications of the lifetime association of these widespread disorders. Recent work in the general population separating past and current disorders has clarified that intoxication or withdrawal effects do not entirely account for the association, 13 as had been asserted earlier. [14][15][16][17] However, the nature of current or recent co-occurrence of substance and mood or anxiety disorder remains largely unexamined and poorly understood. Relative to lifetime disorders, current co-occurrence has much more salience in its public health and clinical implications. Thus, an important gap in knowledge about comorbidity remains.
One factor that has persistently hin dered a better understanding of the relationship between substance use dis orders and mood and anxiety disorders is diagnosis. The diagnosis of current mood or anxiety disorders among active substance abusers is complicated by the fact that many symptoms of intoxica tion and withdrawal from alcohol and other substances resemble the symptoms of mood and anxiety disorders. The diagnostic challenge among individuals with current substance use disorders has been to devise diagnostic criteria and measurement techniques that differentiate between intoxication and withdrawal symptoms and the symptoms of psy chiatric disorders. This distinction is potentially crucial for etiologic research and treatment studies.
The DSM-IV 18 represented a major departure from previous nomenclature in the importance placed on the inde pendent and substance-induced distinc tion and the clarity and specificity of the guidelines for making the distinc tion. Among individuals with substance use disorders, independent DSM-IV diagnoses of mood or anxiety disorders can be made two ways. First, the full mood or anxiety syndrome is established before substance use. Second, the mood or anxiety syndrome persists for more than 4 weeks after the cessation of intoxication or withdrawal. In contrast, substance-induced disorders are defined as those occurring only during periods of substance use (or remitting shortly thereafter). These specific diagnostic criteria provide a clearly defined situation for studying the association of substance use disorders and mood and anxiety disorders that eliminates potential diag nostic confusion arising from misdiagno sis of intoxication or withdrawal effects.
There have been recent attempts to respond to the challenge of differentiating independent and substance-induced mood and anxiety disorders in clinical samples, focusing on patients with sub- stance use disorders. [14][15][16][17] These differ entiations were based on the occurrence of substance use disorders rather than on substance use per se. In these studies, independent mood or anxiety disorders were defined as episodes occurring either before the lifetime initial onset of a sub stance use disorder or during a period of remission lasting at least 3 months. Remission was defined as abstinence. Other episodes of mood or anxiety disorders were classified as substanceinduced disorders. The distinction between independent and substanceinduced disorders in these studies is problematic in several ways. First, retrospective reports of chronological sequences occurring many years earlier may be inaccurate. Second, basing the distinction on substance use disorders rather than on periods of substance use leaves open the possibility that indepen dent psychiatric disorders occurring during periods of nondiagnosable sub stance use were missed. Third, the clinical assessment methods in these studies did not ascertain episodes of independent mood and anxiety disorders beginning during periods of drinking or drug use and persisting longer than 1 month after the cessation of use (as specified in DSM-IV), thus potentially missing further independent cases. From an epidemiologic perspective, however, the most serious problem with research on comorbidity in treated samples is that the samples of subjects do not represent the underlying populations. Avoiding this problem requires epidemi ologic methods.
To our knowledge, no epidemiologic survey has used the DSM-IV definitions of independent and substance-induced disorders to investigate comorbidity between substance use disorders and mood and anxiety disorders. The Epidemiologic Catchment Area survey, conducted in the early 1980s, based its diagnoses on the Diagnostic and Statistical Manual of Mental Disorders, Third Edition (DSM-III), 20 which had little relevance to today's diagnostic concepts, in either the criteria for substance use disorders or the characterization of the independent and substance-induced distinction. The 1990-1992 National Comorbidity Survey (NCS) 2 used

Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised
(DSM-III-R) 21 criteria. While the DSM-III-R definitions of substance use disorders were more similar to those in the DSM-IV, the handling of sub stance use disorders was quite different. The more recent 2001-2002 NCS-2 and NCS-Replication were intended to yield DSM-IV diagnoses. However, the NCS-2 and NCS-Replication assessment instruments did not differ entiate between independent and sub stance-induced disorders but, rather, asked respondents if they thought their mood or anxiety disorder was due to drinking or drug use or to a physical illness. Clearly, such opinions may differ from the intent and the specific defini tions provided in the DSM-IV.
In addition, measurement of sub stance use disorders itself has hindered examination of the independent and substance-induced distinction and its effect on the comorbidity between substance use disorders and mood and anxiety disorders in the general popula tion. In the Epidemiologic Catchment Area survey 23 and the NCS-2, 2 substance dependence was not measured as a syn drome, because clustering in time of the required number of symptoms was not assessed. In addition, the NCS-2 and NCS-Replication do not yield drug-specific diagnoses, but rather pro duce polysubstance dependence diagnoses for which dependence criteria are met for substances as a group, but not nec essarily for any specific drug. In addition, the symptoms of abuse are used as screeners for dependence, with negative responses to abuse questions leading to a skip past questions on dependence. This leads to an undercount of about one-third of the cases of dependence in the general population. 24 However, more seriously, it leads to a loss of specific types of cases, because women with dependence are much less likely to have symptoms of abuse than men. 24 Women are also the individuals most likely to have mood and anxiety disorders, so missing these cases of dependence without abuse symptoms is likely to lead to underestimates of prevalence and comorbidity. Because of the widespread prevalence of mood, anxiety, and substance use disorders and their associated disabilities and social costs, an accurate understand ing of their comorbidity is crucial to prevention and treatment. This report presents data from a major national survey designed to overcome the prob lems of previous epidemiologic surveys on comorbidity. This survey, the National Institute on Alcohol Abuse and Alcoholism's National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), 25,26 covers the comorbidity of DSM-IV substance use disorders and nine independent mood and anxiety disorders in a nationally representative U.S. sample of 43,093 respondents. To our knowledge, this is the largest comorbidity survey ever conducted. The sample size allows for accurate estimation of current comor bidity and/or rare conditions. More important, to our knowledge, NESARC is the first and only national survey to use the specific DSM-IV definitions of independent and substance-induced disorders to determine if mood, anxiety, and substance use disorders are associated even when substance-induced disorders are ruled out. Furthermore, NESARC operationalized alcohol and drug depen dence as syndromes, measured drugspecific diagnoses of dependence, and ascertained alcohol and drug dependence among all alcohol and drug users, regardless of whether they had an abuse diagnosis. The study also provides comor bidity rates separately for respondents seeking treatment for alcohol, drug, and emotional problems, because rates and patterns of comorbidity associated with the presenting complaint are most germane to practicing clinicians.

Sample
Wave I of NESARC is a nationally rep resentative face-to-face survey of 43,093 respondents, 18 years and older, con ducted by the National Institute on Alcohol Abuse and Alcoholism in 2001-2002, 25,26 The target population of NESARC is the civilian noninstitution alized population residing in the United States, including Alaska and Hawaii. The housing unit sampling frame of NESARC was the U.S. Bureau of the Census Supplementary Survey. 25 NESARC also included a group quarters sampling frame derived from the Census 2000 Group Quarters Inventory. 25 The group quarters sampling frame captures important subgroups of the population with heavy substance use patterns not often included in general population surveys. These included the military living off base, boarding houses, rooming houses, nontransient hotels and motels, shelters, facilities for housing workers, college quarters, and group homes. Hospitals, jails, and prisons were not among the group quarters sampled. The overall survey response rate was 81.0 percent, substantially higher than that of other surveys of this kind.
Black and Hispanic households were oversampled. The oversampling proce dure increased the percentage of non-Hispanic Black households in the sam ple from 12.3 percent to 19.1 percent (n = 8,245) and the percentage of Hispanic households from 12.5 per cent to 19.3 percent (n = 8,308). Black and Hispanic persons were oversampled because these subgroups have been underrepresented in previous comor bidity surveys. One sample person from each household or group quarters unit was randomly selected for interview, and young adults, ages 18 to 24, were oversampled at a rate of 2.25 times that of other members in the household.
The NESARC sample was weighted to adjust for the probabilities of selection of a sample housing unit or housing unit equivalent from the group quarters sampling frame, nonresponse at the household and person levels, the selec tion of one person per household, and oversampling of young adults. Once weighted, the data were adjusted to be representative of the U.S. population for various sociodemographic variables, including region, age, sex, race, and ethnicity, based on the 2000 Decennial Census. The sociodemographic distribution of the NESARC sample is shown in Table 1.

Substance Use Disorder Assessment
The diagnostic interview used to gener ate the diagnoses presented in this report is the National Institute on Alcohol Abuse and Alcoholism's Alcohol Use Disorder and Associated Disabilities Interview Schedule-DSM-IV Version (AUDADIS-IV), 27 a state-of-the-art structured diagnostic interview designed for use by lay interviewers. The DSM-IV diagnoses reported herein, and included in the AUDADIS-IV, were alcohol-and drug-specific abuse and dependence (excluding nicotine depen dence), major depression, dysthymia, mania, hypomania, panic disorder with and without agoraphobia, social phobia, specific phobia, and generalized anxiety disorder. Not all mood and anxiety disor ders were assessed in Wave 1 of NESARC because of time and space constraints. However, Wave 2 of NESARC will assess post-traumatic stress disorder. The AUDADIS-IV included an extensive list of symptom questions that separately operationalized DSM-IV for substance use disorders, including alcohol abuse and dependence and drug-specific abuse and dependence for eight classes of drugs, including sedatives, tranquilizers, opiates (other than heroin or methadone), stimulants, hallucinogens, cannabis, cocaine (includ ing crack cocaine), and inhalants/solvents. Consistent with the DSM-IV, 12-month (current) AUDADIS-IV diagnoses of alcohol abuse required a respondent to meet at least 1 of the 4 criteria defined for abuse in the 12-month period pre ceding the interview. The AUDADIS-IV dependence diagnoses required the respondent to satisfy at least 3 of the 7 DSM-IV criteria for dependence during the past year. The drug-specific diagnoses of abuse and dependence were derived using the same algorithm previously described for alcohol use disorders.

Mood and Anxiety Disorder Assessment
Independent and substance-induced disorders were defined for respondents who met the criteria for specific mood and anxiety disorders occurring during the past 12 months. Disorders were classified as independent if (1) the respondent abstained from alcohol and drug use in the past 12 months; (2) the episode(s) did not occur in the context of alcohol or drug intoxication or with drawal; (3) the episode(s) occurred before alcohol or drug intoxication or with drawal; or (4) the episode(s) began after alcohol or drug intoxication or with drawal, but persisted for more than 1 month after the cessation of alcohol or drug intoxication or withdrawal. Substance-induced disorders were defined as episodes that began after alcohol and/or drug intoxication and/or with drawal, but either (1) were not associated with a period of at least 1 month of abstinence or (2) did not persist for more than 1 month after the cessation of alco hol or drug intoxication or withdrawal. Respondents were classified with a 12-month independent mood or anxiety disorder if none or only some of their episodes were substance induced.
Respondents were classified with a substance-induced disorder if all of their episodes in the past 12 months were substance induced.
The prevalence of major depression reported herein ruled out bereavement. All mood and anxiety disorders due to general medical conditions also were * Data are given as odds ratios (ORs) (95% confidence interval). The ORs represent the odds of having a specific mood or anxiety disorder among individuals with a specific substance use disorder relative to the odds of having a specific mood or anxiety disorder among individuals who do not have the specific substance use disorder. ruled out. The latter were defined as those occurring during the past 12 months when the respondent was physically ill or recovering from being physically ill, with the additional requirement that a physician or other health professional confirmed that the episode was related to the respondent's physical illness or medical condition. This definition also required the onset of the mood or anxiety disorder to begin during the time of a physical illness or during recovery from it. The test-retest reliabilities of AUDADIS-IV measures of DSM-IV mood and anxiety disorders were fair to good, ranging from kappa = 0.42 for specific phobia to kappa = 0.64 for major depression. 28,30 The validity of current mood and anxiety disorders was assessed in a series of linear regres sion analyses, using the NESARC data, that examined the associations between each mood and anxiety disorder and Short-Form-12v2 54   Respondents with these current mood and anxiety disorders had significantly greater disability and social/occupational dysfunction than respondents who did not have the particular mood or anxi ety disorder. A diagnosis of hypomania was a significant predictor (p = .049) of the social functioning score.

Twelve-Month Treatment Use
The NESARC respondents were asked about 12-month treatment use separately for alcohol, drugs, and each specific mood or anxiety disorder. Alcohol treatment use was defined as seeking help for alco hol problems in the 12 months preced ing the survey, at any of the following agencies or from any of the following health professionals: human services, including family services or other social service agencies; emergency depart ments or crisis centers; alcohol specialty services, including alcohol or drug detoxification wards or clinics, outpa tient clinics, outreach programs, or day or partial patient programs; inpatient wards of a psychiatric or general hospital or community mental health facilities; alcohol or drug rehabilitation programs; halfway houses; and visits to a physi cian, psychiatrist, psychologist, social worker, or other health professional. The 12-month drug treatment use questions paralleled those of the alco hol treatment use questions, with the exception that methadone maintenance programs were also included as drug specialty services. Twelve-month treatment use was ascertained separately for each specific mood and anxiety disorder. Respondents were classified as receiving treatment in the past 12 months if they: (1) visited a counselor, therapist, physician, psy chologist, or person like that to get help for a mental disorder; (2) were a patient in a hospital for at least 1 night related to a mental disorder; (3) visited an emergency department to get help for a mental disorder; or (4) were prescribed medications for a mental disorder.

Interviewer Training and Field Quality Control
Approximately 1,800 experienced lay interviewers from the U.S. Census Bureau administered NESARC using laptop computer-assisted software that included built-in skip, logic, and consistency checks. On average, the interviewers had 5 years of experience working on Census and other health-related national surveys. The interviewers completed 10 days of extensive training. This was standardized through centralized training sessions under the direction of the National Institute on Alcohol Abuse and Alcoholism and Census headquarters staff.
Regional supervisors recontacted a random 10 percent of all respondents for quality control purposes and for verification of the accuracy of the inter viewers' performance. In addition, 2,657 respondents were randomly selected to participate in a reinterview study after completion of their NESARC interview. These interviews not only served as an additional check on survey data quality but formed the basis of a test-retest reliability study 30 of AUDADIS-IV modules introduced in NESARC. (ORs) were used to study associations between substance use disorders and independent mood and anxiety disor ders. Standard errors and 95 percent confidence intervals were estimated using a software package (SUDAAN 55 ) that uses Taylor series linearization to adjust for the design effects of complex sample surveys like NESARC.

Prevalence of Mood and Anxiety Disorders
The 12-month prevalences of indepen dent mood and anxiety disorders were 9.21 percent and 11.08 percent in the total sample, respectively ( Table 2). The prevalences of substance-induced mood and anxiety disorders among respondents with any mood or anxiety disorder in the total sample and among respondents with and without a cur rent substance use disorder were small, less than 1.0 percent. Of the approxi mately 19.3 million adults who had a current mood disorder, only 202,211 experienced episodes that were classi fied exclusively as substance induced. Similarly, among those with a current anxiety disorder (23.0 million), only a few (50,980) experienced episodes that were exclusively classified as sub stance induced. Of those respondents who were classified as having at least one current independent mood or anxiety disorder, only 7.35 percent and 2.95 percent, respectively, reported experiencing independent and substanceinduced episodes during the year pre ceding the survey.

Prevalence of Substance Use Disorders
The 12-month prevalences of any sub stance, any alcohol, and any drug use disorders were 9.35 percent, 8.46 per cent, and 2.00 percent, respectively ( Table 3). The rate of cannabis use disor der was 1.45 percent, far exceeding the rates of other drug-specific use disor ders (0.02 percent for inhalant/solvent abuse to 0.35 percent for opioid use disorders). The rates for abuse exceeded those for dependence regardless of the specific substance use disorder examined.

Co-Occurrence of Substance Use Disorders and Mood and Anxiety Disorders
The 12-month associations between substance use disorders and independent mood and anxiety disorders are shown in Table 4 in the form of ORs. The overall pattern of ORs is overwhelm ingly positive, with 84.8 percent of the disorder-specific ORs positive (i.e., >1.0) and statistically significant. All independent mood and anxiety disorders were strongly and consistently related to alcohol and drug use disorders (ORs, 1.3-13.9). Any drug abuse also was significantly related to all independent mood and anxiety disorders (ORs, 1.6-4.2). The exception to the overall pattern was the level of association between alcohol abuse and specific independent mood and anxiety disorders, which was not always significant. All the indepen dent mood and anxiety disorders were consistently more strongly related to alcohol and drug dependence than to drug abuse. Mania was more strongly related to the substance use disorders (ORs, 1.4-13.9) than any other mood or anxiety disorder. Among the anxiety disorders, panic disorder with agora phobia was most strongly associated with substance use disorders (ORs, 1.9-10.5).
As indicated by the entry in the upper left corner of Table 5, 19.67 percent of the respondents with any substance use disorder had at least one independent mood disorder during the same 12-month period. Further more, 17.71 percent had at least one independent anxiety disorder. Among respondents with any substance use disorder, 3.30 percent to 14.50 percent also had a specific mood disorder, and 1.46 percent to 10.54 percent had a specific anxiety disorder. These rates were consistently lower for abuse than for dependence and highest for any drug dependence. Respondents with substance use disorders were more likely to have major depression and specific phobia than any other mood or anxiety disorder.

Prevalence of Substance Use Disorders Among Respondents With Mood or Anxiety Disorders
Among respondents with any 12 month mood disorder, 19.97 percent had at least one substance use disorder, and among those with any 12-month anxiety disorder, 14.96 percent had at least one substance use disorder ( Table  6). Among respondents with specific mood disorders, 18.07 percent to 27.91 percent also had at least one sub stance use disorder. This was also true of 13.83 percent to 24.15 percent of the respondents with specific anxiety disorders. Prevalences were consistently lower for abuse than for dependence. Respondents with panic disorder with agoraphobia and generalized anxiety disorder were more likely than those with other mood and anxiety disorders to have a substance use disorder.

Prevalence of Substance Use Disorders Among Respondents With Mood and Anxiety Disorders Who Sought Treatment
The percentage of respondents with at least one 12-month independent mood disorder who sought treatment in the past 12 months was 25.81 percent, while the corresponding percentage for respondents with at least one independent anxiety disorder was 12.13 percent (Table 7). Treatment use was greater for those with dysthymia, major depression, and mania than for those with hypomania. Among respondents with anxiety disorders, treatment use was greater for those with panic disorder, with and without agoraphobia, and generalized anxiety disorder than for those with social and specific phobias.
Among respondents reporting specific independent mood disorders, between 18.54 percent and 30.97 percent had a comorbid substance use disorder, primarily

Prevalence of Mood and Anxiety Disorders Among Respondents With Substance Use Disorders Who Sought Treatment
Only 5.81 percent and 13.10 percent of respondents who had a 12-month alcohol use disorder or a 12-month drug use disorder, respectively, sought treatment for their particular substance use disor der during that same period (Table 8). Among those who sought treatment for an alcohol use disorder, 40.69 percent, 33.38 percent, and 33.05 percent had at least one independent mood disorder, independent anxiety disorder, or drug use disorder, respectively. Among respondents with any drug use disorder who sought treatment for that disorder, 60.31 percent had at least one indepen dent mood disorder, 42.63 percent had at least one independent anxiety disor der, and 55.16 percent had a comorbid alcohol use disorder.

Comments
The major findings of this study docu ment the extremely high rates of sub stance use disorders and independent mood and anxiety disorders in the U.S. population and confirm the strength of associations between them. The preva lence of any current independent mood disorder was 9. Similarly, about 20 percent of the individ uals with at least one current indepen dent mood disorder had a comorbid substance use disorder, while about 15 percent of the individuals with at least one 12-month independent anxiety disorder had a substance use disorder. More important, this study also demon strated that a few individuals in the general population experienced current mood (202,211 adult Americans) or anxiety (50,980 adult Americans) disor ders that were only substance induced. Of considerable clinical relevance is the finding that 40.7 percent of the individuals with a current alcohol use disorder who sought treatment during the same period had at least one current independent mood disorder, while more than 33 percent had at least one current independent anxiety disorder. Among individuals with a current drug use dis order who sought treatment, about 60 percent and 43 percent had at least one current independent mood or anxiety disorder, respectively. Similarly, among individuals with at least one current independent mood or anxiety disorder who sought treatment, about 20 percent and 16 percent, respectively, had a cur rent substance use disorder that was more likely to be an alcohol than a drug use disorder. This suggests that the predom inance of substance-induced (approxi mately 60 percent) rather than indepen dent mood or anxiety disorders found in several recent clinical studies 15-17 of substance abusers was most likely due to diagnostic methods that do not entirely conform to the DSM-IV guidelines for differentiating independent from sub stance-induced disorders. Regardless of the relative prevalence of independent and substance-induced disorders, however, substance-induced mood or anxiety disorders among individuals with sub stance use disorders are serious conditions. For example, when diagnosed carefully according to DSM-IV guidelines, sub stance-induced disorders have been shown to increase the risk for poor out come of substance dependence 56 and lifetime number of suicide attempts. 57 Additional longitudinal research is needed to examine differences in the course and prognosis of chronic substanceinduced disorders and independent mood and anxiety disorders in treated samples.
Taken together, the NESARC results provide clear and persuasive evi dence that mood and anxiety disorders must be addressed by alcohol and drug treatment specialists and that substance use disorders must be addressed by pri mary care physicians and mental health treatment specialists. These results highlight the need for all individuals in treatment to be fully assessed for the presence or absence of a range of psy chiatric disorders. Furthermore, the results underscore the importance of past and ongoing development of improved treatments for those individ uals meeting the criteria for two or more disorders. [58][59][60][61][62] Moreover, these results strongly suggest that treatment for a mood or anxiety disorder should not be withheld from those with sub stance use disorders in stable remission on the assumption that most of these disorders are due to intoxication or withdrawal. Left untreated, such mood disorders have been shown to lead to relapse of substance dependence 56 and can also be fatal, as many former sub stance abusers with severe untreated independent depression will die by suicide. Short of this ultimately adverse outcome, independent mood and anxiety disorders, particularly among individu als who have a comorbid substance use disorder, are immensely disabling. [4][5][6][7] From an etiologic perspective, this study does not resolve questions regard ing the casual mechanisms underlying the relationship between DSM-IV substance use disorders and indepen dent mood and anxiety disorders. Prospective surveys have great potential to inform us about processes associated with comorbidity and will provide the vehicles for examining the sequencing of comorbid disorder onset. NESARC was designed with this paradigm in mind, and its second wave will be fielded in 2004-2005.