Advances in Mental Health Reseach

Implications for Practice

Ronald C. Kessler, Jamie M. Abelson, and Shanyang Zhao

Epidemiology is the study of the distribution and correlates of illness in the population. Descriptive epidemiology is concerned with the distribution of illness onset and course. Analytic epidemiology is concerned with the use of nonexperimental data to elucidate causal processes involved in illness onset and course. Experimental epidemiology is concerned with the development and evaluation of interventions aimed at modifying risk factors to prevent illness onset or to modify illness course. Most epidemiologic studies of psychiatric disorders are either descriptive or analytic. Experimental epidemiologic studies are for the most part limited to preventive interventions for children (for reviews, see Dryfoos, 1990; Hamburg, 1992), although there also are a small number of experimental epidemiologic studies of high-risk adults (for example, Howe, Caplan, Foster, Lockshin, & McGrath, 1995; Price & Vinokur, 1995). See Rothman (1986) for an in-depth introduction to epidemiology overall and Tsuang, Tohen, and Zahner (1995) for an in-depth introduction to psychiatric epidemiology.

Although descriptive psychiatric epidemiologic studies comparing admission and discharge rates to and from asylums were carried out as early as the 17th century, it was not until the early 19th century that analytic epidemiologic studies of psychiatric disorders began to appear (Hunter & Macalpine, 1963). The latter consistently documented associations that were interpreted as showing that environmental stresses, especially stresses associated with poverty, can lead to psychiatric disorders. However, most of these early studies were hampered by the fact that they either focused on treatment statistics or assessed disorders in the community by using key informants such as physicians and clergy rather than direct assessment. These research design features led to confounding of information about illness prevalence with information about help-seeking and labeling, resulting in the underestimation of the prevalence of clinically significant psychiatric disorders.

The end of World War II brought with it the beginning of modern psychiatric epidemiology. Evidence of widespread emotional problems in selective service recruits before the war and concerns about traumatic stress reactions in the wake of the war led to widespread concern about the prevalence and distribution of psychiatric disorders. A number of local and national surveys were carried out to study these matters. Unlike prewar studies, though, these new surveys were based on direct interviews with representative community samples.

The earliest of these postwar surveys were either carried out by clinicians or used lay interview data in combination with record data as input to clinician evaluations of caseness. In later studies, clinician judgment was abandoned in favor of less expensive, self-report symptom-rating scales that assigned each respondent a score on a continuous dimension of nonspecific psychological distress. Controversy surrounded the use of these screening scales from the onset, focusing on such things as item bias, insensitivity, restriction of symptom coverage, and the arbitrariness of decisions regarding the selection of caseness thresholds (for example, Dohrenwend & Dohrenwend, 1965; Seiler, 1973). Nonetheless, they continued to be the mainstay of community psychiatric epidemiology through the 1970s due to the combination of three factors. First, they were much less expensive to administer than clinician-based interviews. Second, compared with dichotomous clinician caseness judgments, continuous measures of distress dealt directly with the actual constellations of signs and symptoms that exist in the population rather than with the classification schemes imposed on these constellations by the committees that created the official diagnoses in the Diagnostic and Statistical Manual of Mental Disorders (DSM) of the American Psychiatric Association (APA). Third, the clinician-based diagnostic interviews available during this period did not have good psychometric properties when administered in community samples (Dohrenwend, Yager, Egri, & Mendelsohn, 1978).

However, there also were disadvantages of working with distress measures. Perhaps the most important of these was that there was nothing in the measures themselves that allowed researchers to discriminate between people who did and did not have clinically significant psychiatric problems. This discrimination was important for making social policy decisions regarding such things as the number of people in need of mental health services. Researchers who worked with measures of nonspecific psychological distress dealt with this problem by developing rules for classifying people with scores above a certain threshold as psychiatric cases (for example, Radloff, 1977). The precise cutpoints were usually based on statistical analyses that attempted to discriminate optimally between the scores of patients in psychiatric treatment and those of people in a community sample. However, as noted above, considerable controversy surrounded the decision of exactly where to specify cutpoints. Dichotomous diagnostic measures, in comparison, allowed this sort of discrimination to be made directly based on an evaluation of diagnostic criteria and were preferable if all else was equal. All else was not equal, however, during the first three decades after World War II. Diagnostic interviews were less than optimal because a lack of agreement existed on appropriate research diagnostic criteria and an absence of valid instruments for carrying out research diagnostic interviews.

It was not until the 1970s that the field was able to move beyond this controversy with the establishment of clear research diagnostic criteria (Feighner, Robins, & Guze, 1972) and the development of systematic research diagnostic interviews aimed at operationalizing these criteria (Endicott & Spitzer, 1978). The early interviews of this type required administration by clinicians; the interviews yielded rich data but had limited use in epidemiologic surveys because of the high costs associated with large-scale use of clinicians as interviewers. The majority of interviewers in these studies were clinical social workers. It is unsurprising, in light of the high costs and logistic complications of mounting a large-field operation using professionals of this sort as interviewers, that only a handful of such studies were carried out, were small (for example, Weissman & Myers, 1978), were based on samples that were not representative of the general population (for example, Kendler, Neale, Kessler, Heath, & Eaves, 1992), or were outside the United States in countries where the costs of clinician interviewing are much lower (for example, Dohrenwend, Levav, Shrout, & Schwartz, 1992).

Two responses to this situation occurred in the late 1970s. The first was the refinement of two-stage screening methods, in which an inexpensive first-stage screening scale can be administered by a lay interviewer to a large community sample and followed with more expensive, second-stage clinician-administered interviews for the subsample of initial respondents who screen positive along with a small subsample of those who screen negative (Newman, Shrout, & Bland, 1990). The hope was that two-stage screening would substantially reduce the costs of conducting clinician-administered community epidemiologic surveys. However, problems associated with reduced response rates resulting from the requirement of respondents participating in two interviews and the increased administrative costs associated with logistic complications in this design prevented it from being used widely in community surveys. It continues to be used in surveys of captive populations, however, such as schoolchildren in classrooms.

The second response was the development of research diagnostic interviews that could be administered by lay interviewers. The first instrument of this type was the Diagnostic Interview Schedule (DIS) (Robins, Helzer, Croughan, & Ratcliff, 1981), which was developed with support from the National Institute of Mental Health (NIMH) for use in the Epidemiologic Catchment Area (ECA) Study (Robins & Regier, 1991). Several other interviews, most of them based on the DIS, have been developed. The most widely used of these is the World Health Organization’s (WHO, 1990) Composite International Diagnostic Interview (CIDI).

The remainder of this chapter will provide a selective overview of the results regarding the descriptive epidemiology of psychiatric disorders in the United States based on recent surveys that have used the DIS or CIDI to study the prevalence and correlates of DSM-III (APA, 1980) or DSM-III-R (APA, 1987) disorders. The focus will be on lifetime prevalence, the proportion of a population who have experienced a particular disorder at some time in their lives, and recent prevalence—the proportion of a population who have experienced an episode of a particular disorder over some specified recent interval of time such as the past month or the past year. Some indirect information also will be reported on incidence—the proportion of the subpopulation without a history of a particular disorder who have a first onset over some specified interval of time—in the presentation of age of onset curves. The latter provide a graph of the cumulative population prevalences of disorders over the life course. For a more detailed discussion of these and other commonly used descriptive statistics in psychiatric epidemiology, see Zahner, Hsieh, and Fleming (1995). Although a thorough discussion of analytic psychiatric epidemiology is beyond the scope of this review, a brief sketch of current issues in this area also is included in a later section of the chapter.

Data Sources

The need for general population data on the prevalence of mental illness was recognized two decades ago in the report of the President’s Commission on Mental Health and Illness (1978). It was impossible to undertake such a survey at that time because of the absence of a structured research diagnostic interview capable of generating reliable psychiatric diagnoses in general population samples. As noted above, NIMH, recognizing this need, funded the development of the DIS (Robins et al., 1981), a research diagnostic interview that could be administered by trained interviewers who are not clinicians. The DIS was first used in the ECA study, a landmark study that interviewed over 20,000 respondents in a series of five community epidemiologic surveys (Robins & Regier, 1991). The ECA has been the main source of data in the United States on the prevalence of psychiatric disorders and utilization of services for these disorders for the past decade (Bourdon, Rae, Locke, Narrow, & Regier, 1992; Regier et al., 1993; Robins, Locke, & Regier, 1991) and is a major source of data for the review presented in this chapter.

General population reliability and validity studies of the DIS were not carried out until after the completion of the ECA data collection (Anthony et al., 1985; Helzer et al., 1985), and the results of these methodological studies showed generally low agreement between DIS classifications and the classifications independently made by clinical reinterviewers. Questions have been raised about the accuracy of the ECA results based on these methodological studies (for example, Parker, 1987; Rogler, Malgady, & Tryon, 1992). However, other analysts have noted that the validity problems in the DIS are concentrated among respondents who either fall just short of meeting criteria or just barely meet criteria and that the errors from false positives and false negatives tend to balance to produce fairly accurate total population prevalence estimates (Robins, 1985). Although this observation provides no assurance that the different errors are counterbalanced in all important segments of the population (Dohrenwend, 1995), the documentation that this is true in the population as a whole suggests that the ECA results yield useful overall prevalence data.

Another important limitation of the ECA Study for purposes of providing representative data is that it was carried out in only five areas in the country: New Haven, Connecticut; Baltimore; Durham, North Carolina; St. Louis; and Los Angeles. Although the ECA investigators used after-the-fact weighting to combine these local data into a consolidated data file that was representative of the country as a whole on the distribution of age, sex, and race, no attempt was made to adjust the sample for the distributions of such important variables as socioeconomic status or health insurance coverage. Furthermore, it was impossible to apply any type of weighting or adjustment procedure to compensate for the fact that all five ECA sites were in urban areas that contained large, university-based hospitals. Because the interviews were conducted entirely within the metropolitan areas containing these hospitals, the results reveal nothing about areas of the country that have low access to specialty mental and addictive services, including rural areas that are not contiguous to a major metropolitan area. Of the U.S. population, 20 percent live in such areas.

This problem was addressed when NIMH funded the National Comorbidity Survey (NCS) (Kessler, McGonagle, et al., 1994), a household survey of more than 8,000 respondents ages 15 to 54 that was carried out in a widely dispersed (174 counties in 34 states) sample designed to be representative of the entire United States. The NCS interview used a modified version of the DIS known as the CIDI (Robins et al., 1988). The CIDI expands the DIS to include diagnoses based on DSM-III-R (APA, 1987) criteria as well as the International Classification of Diseases (ICD-10) (WHO, 1990). WHO field trials of the CIDI have documented adequate reliability and validity for all diagnoses. (For a review, see Wittchen, 1994.) However, most of the WHO field trials were carried out in clinical samples. Previous research has shown that the estimated accuracy of diagnostic interviews is greater in clinical samples than in general population samples (Dohrenwend et al., 1978). Therefore, the same caution regarding diagnostic accuracy as noted above is needed in interpreting the results of the NCS.

Some assistance in evaluating the magnitude of the potential problems with the diagnoses based on the structured lay interviews in the ECA and NCS can be obtained by comparing those results with findings from the small number of modern epidemiologic surveys mentioned earlier in this chapter. Some of those surveys have used semistructured research diagnostic interviews such as the Structured Clinical Interview for DSM-III-R (SCID) (Spitzer, Williams, Gibbon, & First, 1990), administered by clinical interviewers either in an entire community sample or in a second-stage subsample of a larger community sample that was selected to overrepresent possible cases. The fact that these studies are few in number and generally based on small samples that are not entirely representative of the population makes it impossible to aggregate their results to generate an accurate portrait of the prevalence of mental illness in the U.S. population. Nonetheless, we refer to these studies in several places in the following review in an effort to provide a rough external comparison with the results based on the lay interview diagnoses obtained in the ECA and NCS surveys.

A final point regarding data sources concerns diagnostic coverage. Almost all the diagnoses are Axis I disorders in the DSM-III and DSM-III-R diagnostic systems. Not all Axis I disorders are covered in these surveys. However, the most commonly assessed Axis I disorders are mood disorders (major depression, dysthymia, and mania), anxiety disorders (generalized anxiety disorder, panic disorder, phobia, obsessive–compulsive disorder, and posttraumatic stress disorder), addictive disorders (alcohol abuse and dependence and drug abuse and dependence), and nonaffective psychoses (schizophrenia, schizophreniform disorder, schizoaffective disorder, delusional disorder, and brief reactive psychosis). Axis II disorders, which include the personality disorders and mental retardation, are generally not covered, although antisocial personality disorder (ASPD) and some measures of cognitive impairment are often assessed. The absence of information on personality disorders other than ASPD is a major omission but was necessitated by the fact that valid structured diagnostic interview methods to assess personality disorders did not exist at the time these surveys were carried out. That situation is changing rapidly, however, because several groups are working to develop measures of personality disorders that are appropriate for use in general population surveys (for example, Lenzenweger, Loranger, Korfine, & Neff, 1997; Pilkonis et al., 1995), and we can anticipate, based on this work, that future large-scale epidemiologic surveys will include comprehensive evaluations of personality disorders. For now, though, our review of evidence regarding the prevalences of personality disorders other than ASPD has to rely on the results of a small number of surveys from around the world that have been carried out using one of the recently developed assessment methods.

Lifetime and Recent Prevalences of Axis I DSM-III-R Disorders

We focus on results from the NCS, because it is the only nationally representative survey in the United States to have assessed the prevalences of a broad range of DSM-III-R disorders. As described in more detail elsewhere (Kessler, McGonagle, et al., 1994), the NCS is based on a national household sample of 8,098 respondents ages 15 to 54, including a supplemental sample of students living in group housing, the largest segment of the population that is not in the household population. The results in Table 1-1 show NCS/DSM-III-R prevalence estimates for the lifetime and 12-month disorders assessed in the core NCS interview. The prevalence estimates are presented without exclusions for DSM-III-R hierarchy rules.

The most common psychiatric disorders assessed by the NCS are major depression and alcohol dependence. A total of 17.1 percent of respondents reported a history of a major depressive episode in their lifetimes, and 10.3 percent had had an episode in the past 12 months. A total of 14.1 percent have a lifetime history of alcohol dependence, and 7.2 percent continued to be dependent in the past 12 months. The next most common disorders are social and simple phobias, with lifetime prevalences of 13.3 percent and 11.3 percent, respectively, and 12-month prevalences of 7.9 percent and 8.8 percent, respectively. As a group, addictive disorders and anxiety disorders are somewhat more prevalent than mood disorders. Approximately one in every four respondents reported a lifetime history of at least one addictive disorder, and a similar number reported a lifetime history of at least one anxiety disorder. Approximately one in every five respondents reported a lifetime history of at least one mood disorder. Anxiety disorders as a group were considerably more likely to occur in the 12 months prior to interview (19.3 percent) than either addictive disorders (11.3 percent) or mood disorders (11.3 percent), suggesting that anxiety disorders are more chronic than either addictive disorders or mood disorders. The prevalences of other NCS disorders are much lower. Antisocial personality disorder, which was assessed only on a lifetime basis, was reported by 2.8 percent of respondents, while schizophrenia and other nonaffective psychoses (NAP) were found among 0.5 percent of respondents. It is important to note that the diagnosis of NAP was based on clinical reinterviews using the SCID rather than on the lay CIDI interviews. As documented elsewhere (Kendler, Gallagher, Abelson, & Kessler, 1996), the prevalence estimate for nonaffective psychoses based on the CIDI was considerably higher but was found to have low validity when judged in comparison with the clinical reappraisals.

A total of 49.7 percent of the sample reported a lifetime history of at least one NCS/DSM-III-R disorder and 30.9 percent had one or more disorders in the 12 months prior to the interview. Although there is no meaningful sex difference in these overall prevalences, sex differences exist in prevalences of specific disorders. Consistent with previous research (Bourdon et al., 1992; Robins et al., 1981; Robins et al., 1991), men are much more likely to have addictive disorders and ASPD than women are, whereas women are much more likely to have mood disorders (with the exception of mania, for which there is no sex difference) and anxiety disorders than men are. The data also show, consistent with a trend found in the ECA (Keith, Regier, & Rae, 1991) that women in the household population are more likely to have nonaffective psychoses than men.

It is instructive to compare these NCS results with the results of the earlier ECA study. As noted above, the ECA was carried out in five communities in the United States, and the results were subsequently combined and weighted to the population distribution of the United States on the cross-classification of age, sex, and race in an effort to make national estimates (Regier et al., 1993). To the extent that this poststratification succeeded in adjusting for the lack of representativeness of the local samples, it should be possible to make valid comparisons between the ECA and NCS results. A limitation in doing this is that the ECA was based on an unrestricted age range of adults, whereas the NCS was based on the 15 to 54 age range. Another limitation is that the ECA diagnoses were based on DSM-III criteria (APA, 1980), whereas the NCS diagnoses were based on DSM-III-R criteria (APA, 1987). These two diagnostic systems differ substantially in a number of respects. To resolve these problems, collaborative ECA–NCS comparative analyses have been carried out in which subsamples in the 18 to 54 age range in both samples were compared using common measures that operationalize DSM-III criteria, which can be reconstructed from the NCS data, although DSM-III-R criteria cannot be reconstructed from the ECA data. The as yet unpublished results show a great deal of consistency between the two surveys, both in the prevalences of individual disorders and in the overall prevalence of having any disorder (Regier et al., in press).

Personality Disorders

Although the concept of personality disorder can be traced back to the beginnings of 19th-century psychiatry (for a review, see Tyrer, Casey, & Ferguson, 1991), it has only recently become the subject of epidemiologic research because standardized diagnostic criteria became available for the first time with the ICD-9 (WHO, 1977) and DSM-III (APA, 1980) classification systems. Unfortunately, it has proved to be difficult to develop reliable and valid measures of personality disorders (Perry, 1992; Zimmerman, 1994). Furthermore, there are a number of differences between the ICD and DSM systems (Blashfield, 1991) as well as substantial changes within each system in recent revisions (Morey, 1988) that add to the complexity of synthesizing the available epidemiologic evidence.

All classification schemes recognize three broad clusters of personality disorder, each defined by a series of traits that must be manifest habitually in a number of life domains to qualify as a disorder: (1) the odd cluster (for example, paranoid or schizoid personality disorders), (2) the dramatic cluster (for example, histrionic or borderline personality disorders), and (3) the anxious cluster (for example, avoidant or dependent personality disorders). A recent comprehensive international review of the epidemiology of personality disorders found only four fairly small community studies that assessed personality disorders in all three of these clusters using valid assessment methods (de Girolamo & Reich, 1993). These four studies yielded consistent lifetime prevalence estimates for overall personality disorders ranging from 10.3 percent to 13.5 percent. Caution is needed in interpreting these results, however, because previous research has shown that prevalence estimates vary substantially depending on whether, as in these surveys, full diagnostic criteria for personality disorders are required or respondents are counted if they manifest some traits of personality disorders on dimensional scales (Kass, 1985).

A number of community surveys have included assessments of one or more specific personality disorders without attempting to assess the full range of personality disturbances. By far the most commonly studied of these has been ASPD, which is characterized by persistent evidence of "irresponsible and antisocial behavior beginning in childhood or early adolescence and continuing into adulthood" (APA, 1987). Irritability, aggressiveness, persistent reckless behavior, promiscuity, and the absence of remorse about the effects of their behavior on others are cardinal features of ASPD. A number of epidemiologic surveys, including both the ECA and NCS, have found lifetime prevalences of ASPD averaging about 1 percent among women and 4 percent to 5 percent among men (de Girolamo & Reich, 1993; Merikangas, 1988). Much less is known about the prevalences of other individual personality disorders, although the available evidence suggests that none of them alone has a prevalence greater than about 2 percent in the general population (de Girolamo & Reich, 1993; Merikangas, 1988; Weissman, 1993).

Comorbidity

An important observation about the results in Table 1-1 is that the sum of the individual prevalence estimates across the disorders in each row consistently exceeds the prevalence of having any disorder in the last row. This means that there is a good deal of comorbidity among these disorders. For example, whereas the 49.7 percent lifetime prevalence in the total NCS sample means that 50 respondents out of every 100 in the sample reported a lifetime history of at least one disorder, a summation of lifetime prevalence estimates for the separate disorders shows that these 50 people reported a total of 102 lifetime disorders. This comorbidity is important for understanding the distribution of psychiatric disorders in the United States (Kessler, 1995). Although it is beyond the scope of this chapter to delve into the many types of comorbidity in the population, some aggregate results are important to review.

The results presented in Table 1-2 document that comorbidity is important in understanding the distribution of psychiatric disorders among people ages 15 to 54 in the United States. These results also provide an empirical rationale for more detailed examination of particular types of comorbidity. The four horizontal rows of Table 1-2 represent the number of lifetime disorders reported by respondents. The set of disorders considered here is somewhat smaller than in Table 1-1, accounting for the fact that 52 percent of respondents are estimated as never having any NCS/DSM-III-R disorder (which means that 48 percent are estimated to have one or more such disorders, which is smaller than the 49.7 percent in Table 1-1), 21 percent as having one, 13 percent as having two, and 14 percent as having three or more disorders.

Only 21 percent of all the lifetime disorders occurred in respondents with a lifetime history of just one disorder. This means that the vast majority of lifetime disorders in this sample (79 percent) are comorbid disorders. Furthermore, an even greater proportion of 12-month disorders occurred in respondents with a lifetime history of comorbidity. It is particularly striking that close to six out of every 10 (58.9 percent) 12-month disorders and nearly nine out of 10 (89.5 percent) severe 12-month disorders occurred to the 14 percent of the sample with a lifetime history of three or more disorders. These results show that whereas a history of some psychiatric disorder is common among people ages 15 to 54 in the United States, the major burden of psychiatric disorders in this sector of society is concentrated in a group of people with high comorbidity, who constitute about one-sixth of the population.

Given this evidence, it is of some interest to learn more about detailed patterns of comorbidity. The ECA investigators were the first to do this in a community sample. They documented that comorbidity is widespread; over 54 percent of ECA respondents with a lifetime history of at least one DSM-III psychiatric disorder were found to have a second diagnosis as well. Fifty-two percent of people with lifetime alcohol abuse received a second diagnosis, and 75 percent of people with lifetime drug abuse had a second diagnosis (Robins et al., 1991). Compared with respondents with no mental disorder, respondents with a lifetime history of at least one mental disorder had a relative-odds of 2.3 of having a lifetime history of alcohol abuse or dependence and a relative-odds of 4.5 of some other drug use disorder (Regier et al., 1990). Similar results were found in the NCS. Fifty-six percent of the respondents with a lifetime history of at least one DSM-III-R disorder also had one or more other disorders (Kessler, 1995). Fifty-two percent of respondents with lifetime alcohol abuse or dependence also had a lifetime mental disorder, whereas 36 percent had a lifetime illicit drug use disorder. Fifty-nine percent of the respondents with a lifetime history of illicit drug abuse or dependence also had a lifetime mental disorder and 71 percent had a lifetime alcohol use disorder.

The results in Table 1-3 show the proportions of people having each lifetime NCS/DSM-III-R disorder who reported at least one other lifetime disorder. Lifetime comorbidity is the norm, with proportions ranging from a low of 62.1 percent for alcohol abuse to a high of 99.4 percent for mania. The average proportion of comorbidity among disorders is 86.6 percent. This does not mean that 86.6 percent of people with one or more lifetime disorders have comorbidity, though, as those with comorbidity are counted multiple times in Table 1-3. Instead, 59.8 percent of the people who ever had one of the disorders considered in the NCS also had one or more other disorders.

Data on lifetime comorbidities of specific pairs of disorders in the NCS are reported in Table 1-4. Results are shown in the form of odds-ratios (ORs), statistics that describe the relationship between two disorders as the odds of having disorder A in patients who have disorder B, divided by the odds of having disorder A in patients who do not have disorder B. An OR of 1.0 means that there is no relationship between the two disorders, whereas an OR greater than 1.0 means that there is a positive relationship, and an OR less than 1.0 means that there is a negative relationship. A more detailed discussion of ORs is presented by Hillis and Woolson (1995). Diagnostic hierarchy rules were not used in making these calculations so as to avoid artificially deflating estimates of comorbidity. The exceptions are substance abuse and dependence, which are defined in such a way as to be mutually exclusive.

As shown in Table 1-4, all but four of the 118 ORs are greater than 1.0. This means that there is a positive association between the lifetime occurrences of almost every pair of disorders considered here. However, there is a great deal of variation in the size of the ORs. It is conceivable that this variation is due to random error. To determine whether this is the case, a comparison was made between the ORs presented here and ORs obtained by reanalyzing data from the ECA study. This comparison found a rank-order correlation between the two sets of ORs of 0.79 (Kessler, 1995), which demonstrates that the variation in ORs is systematic rather than random.

Several patterns related to this variation are worthy of note. First, one would expect that the relative sizes of the ORs would show the disorders of a single type to be more strongly related to each other than to disorders of another type. This is generally true. For example, mood disorders are strongly related to other mood disorders. However, the strength of these pairwise associations among mood disorders is generally stronger than within the anxiety disorders, with an average OR of 13.5 for mood disorders, compared with 6.2 for anxiety disorders. Second, most mood and anxiety disorders are strongly related to each other. In fact, pairwise associations between a mood disorder and an anxiety disorder (averaging 6.6) are generally stronger than between two anxiety disorders. Third, despite a substantial clinical literature pointing to the importance of comorbidity between mood disorders and addictive disorders (for example, Allen & Frances, 1986; Penick, Powell, Liskow, & Jackson, 1988) and between anxiety disorders and addictive disorders (for example, Chambless, Cherney, Caputo, & Rheinstein, 1987; Roy et al., 1991), they are among the weakest comorbidities in the table (averaging 2.4).

One of the main purposes of investigating comorbid disorders is to help refine definitions of syndromes and diagnoses. With this in mind, it is important to recognize that some of the strongest ORs in the table are associated with clusters that are generally recognized as disorders in their own right. The largest OR is between major depression and mania, a conjunction that represents bipolar disorder. Another strong OR is between mania and nonaffective psychosis, a conjunction that is part of the definition of schizoaffective disorder. There also are a number of strong ORs that are linked to discussions in the clinical literature of heretofore unrecognized disorders. For example, the suggestion has been made that comorbidity between major depression and mania is often due to a phasic panic–depressive illness characterized by panic, depressive, and mixed anxious–depressive phases (Akiskal, 1986). This possibility is consistent with the finding of a strong OR between panic and depression in the table.

Ages of Onset

The ECA and NCS studies both collected retrospective data on the ages of first onset of each lifetime disorder. They were consistent in showing that simple and social phobia have a much earlier age of onset than the other disorders considered here (Burke, Burke, Rae, & Regier, 1991; Magee, Eaton, Wittchen, McGonagle, & Kessler, 1996), with simple phobia often beginning during middle or late childhood and social phobia during late childhood or early adolescence. Substance abuse was found to have a typical age of onset during the late teens or early 20s. A substantial proportion of people with lifetime major depression and dysthymia also reported that their first episode occurred before age 20. Some other disorders, such as generalized anxiety disorder and mania, had later ages of onset, but the most striking overall impression from the data as a whole is that most psychiatric disorders have first onsets early in life.

Given the importance of comorbidity, a related question concerns which disorders in comorbid sets have the earliest ages of onset. The results in the first column of Table 1-5 show that there is considerable variation across disorders in the probability of being the first lifetime disorder. Simple phobia, social phobia, alcohol abuse, and conduct disorder are the only four disorders considered here where the majority of lifetime cases are temporally primary in this way. In general, anxiety disorders are most likely to be temporally primary, with 82.8 percent of NCS respondents having one or more anxiety disorders reporting that one of these was their first lifetime disorder compared with 71.1 percent of those with conduct disorder, 43.8 percent of those with a mood disorder, and 48.1 percent of those with a substance use disorder. Results in the third column of the table show the percentage of overall respondents who reported each disorder as temporally primary. Once again anxiety disorders are more likely to be temporally primary (45.3 percent of all lifetime cases) than are mood disorders (16.4 percent), substance use disorders (24.5 percent), or other disorders (19.5 percent).

Utilization of Services

Only a minority of those with a lifetime NCS/DSM-III-R disorder (42.0 percent) reported ever obtaining professional treatment for their problems. The proportions treated in the mental health specialty sector (26.2 percent) or in a substance abuse treatment setting (8.4 percent) are even smaller. Only about one-fifth of respondents who reported an episode of a disorder during the year before interview obtained any professional treatment during that year, and only about half of those people were seen in a mental or addictive disorders specialty setting. These results are consistent with those of the ECA study (Regier et al., 1993) and suggest that there is considerable unmet need for services. It is worth noting, however, that strong relationships exist between the of number and severity of disorders and the probability of obtaining professional help.

Social Consequences

The recent debates concerning the place of mental health coverage in health care reform has led to a new interest on the part of psychiatric epidemiologists in the social consequences of psychiatric disorders. A number of recent studies of this issue have documented that psychiatric disorders have substantial personal costs for the people who experience them as well as for their families and communities in terms of both finances (Kessler, Foster, Saunders, & Stang, 1995) and role functioning (Rhode, Lewinsohn, & Seeley, 1990; Wells et al., 1989; Wohlfarth, van den Brink, Ormel, Koeter, & Oldehinkel, 1993). Data from the ECA showed that people with psychiatric disorders have considerably more work loss days than others (Broadhead, Blazer, George, & Kit, 1990; Johnson, Weissman, & Klerman, 1992), a result replicated in the NCS (Kessler & Frank, 1997). NCS analyses also showed that early-onset psychiatric disorders are strongly related to subsequent teenage childbearing, school dropout, marital instability, and long-term financial adversity (Kessler et al., 1995; Kessler et al., 1997; Kessler & Forthofer, in press). These results document hidden societal costs of psychiatric disorders not only in the indirect sense of threats to our ability to maintain an educated and well-functioning citizenry and work force but also in the direct sense that the outcomes documented here are associated with increased use of entitlement programs, such as unemployment and welfare, which are paid for by all taxpayers. These costs need to be taken into consideration in policy evaluations of the societal cost–benefit ratio of comparing providing mental health treatment irrespective of ability to pay with the costs of failing to do so.

Analytic Epidemiology

In analytic epidemiology, the ultimate interest is in pinpointing potentially modifiable risk factors that can be subjected to experimental evaluation to develop interventions for prevention or amelioration of disease outcomes. Analytic epidemiology is such a fertile research arena that no single chapter could provide a thorough overview of contemporary research. However, several broad themes can be detected and critical works cited to provide a road map for the reader who is interested in pursuing a more in-depth investigation of this literature. Three of these themes are mentioned here, concerning the effects of stress, stress-modifiers, and genetic predisposition.

The first of these broad themes involves the intuition that stressful life experiences play a part in bringing about the onset of many psychiatric disorders. A useful recent overview of research on this theme can be found in Does Stress Cause Psychiatric Illness? edited by Carolyn Mazure (1995). Part of the research on this theme is concerned with the long-term effects of exposure to childhood adversities on adult psychopathology (for example, Kendler, Davis, & Kessler, 1997). Another part is concerned with the triggering effects of adult stressful life events on episode onsets of recurrent disorders (for example, Kessler, 1995). Still another part is concerned with the effects of exposure to chronic role-related stresses on such widely documented associations as the higher prevalence of depression among women than men (for example, McGrath, Keita, Strickland, & Russo, 1990) and the higher prevalence of most psychiatric disorders among lower- than among middle-class people (Dohrenwend et al., 1992).

Although this research consistently shows that stress is significantly related to psychiatric disorder, it also finds that a substantial proportion of people who are exposed to even the most severe types of stress do not develop clinically significant psychiatric problems as a result of this exposure. This observation has led to a second dominant theme in analytic psychiatric epidemiology, one that emphasizes the importance of individual differences in vulnerability to stress. Active lines of research exist on a number of presumed stress modifiers, including social support (Kessler, Kendler, Heath, Neale, & Eaves, 1992), appraisal and coping processes (Taylor & Aspinwall, 1996), social identities (Burke, 1996), and personality (Gilbert & Connolly, 1991). Much of the currently active work on interventions for stress-related psychiatric disorders operates by attempting to manipulate these sorts of stress modifiers. A third dominant theme is that genetic factors play an important part in most common psychiatric disorders, a finding that has been clearly and consistently documented in a number of epidemiologic studies based on either twin or adoption designs (Kendler et al., 1995; Tsuang & Farone, 1990). Population genetic studies of psychiatric disorders estimate such things as the proportion of variance in disorders from genetic factors (for example, Kendler et al., 1992) and the extent to which genetic variation operates as a stress modifier (for example, Kessler, Kendler, Heath, Neale, & Eaves, 1994).

Conclusion

Psychiatric disorders are highly prevalent in the general population. Although no truly comprehensive assessment of all Axis I and Axis II disorders has been carried out in a general population sample, it is almost certainly the case that such a study would find that the majority of the population has experienced at least one of these disorders. Although such a result might initially seem remarkable, it is actually easy to understand. The DSM classification system is very broad. It includes a number of disorders that are usually self-limiting and not severely impairing. It should be no more surprising to find that half the population has had one or more of these disorders than to find that the vast majority of the population has had the flu or measles or some other common physical malady at some time in their life.

The more surprising result is that although many people have been touched by mental illness at some time in their lives, the major burden of psychiatric disorder in the population is concentrated in the relatively small subset of people who are highly comorbid. This means that a pile-up of multiple disorders is the most important defining characteristic of serious mental illness, a result that points to the previously underappreciated importance of research on the primary prevention of secondary disorders (Kessler & Price, 1993). It also means that epidemiologic information about the prevalences of individual disorders is much less important than information on the prevalences of functional impairment, comorbidity, and chronicity. This realization has led to a recent interest in functional impairment in the changes in diagnostic criteria in DSM-IV (APA, 1994). This same emphasis also can be seen in the emphasis of NIMH’s National Advisory Mental Health Council (1993) on what they defined as "severe and persistent mental illness" (SPMI) and the emphasis of the Substance Abuse and Mental Health Service Administration (1993) on what they defined as "serious mental illness" (SMI). Joint methodological analyses of the ECA and NCS suggest that the one-year prevalences of DSM-III-R SPMI and SMI are approximately 3 percent and 6 percent, respectively, compared with one-year prevalences of any DSM-III-R disorder in excess of 30 percent (Kessler et al., 1996). It is likely that epidemiologic research on adult mental disorders over the next decade will focus on these serious and severe disorders rather than on overall prevalence. To the extent that the prevalences of particular disorders are emphasized in this work, it will likely be to study the underlying pathologies associated with ongoing impairment in functioning. An increased interest in the part played by personality disorders in the creation of ongoing role impairment is likely to emerge over the next decade in light of recent advances in conceptualization and measurement (Loranger, Sartorius, & Janca, 1996). There also is going to be a considerable expansion of research on the epidemiology of child and adolescent disorders, a topic that has not been covered in this chapter, because new initiatives have not yet advanced far enough to be reviewed here.

References

Akiskal, H. S. (1986). Mood disturbances. In G. Winokur & P. Clayton (Eds.), The medical basis of psychiatry (pp. 365–379). Philadelphia: W. B. Saunders.

Allen, M. H., & Frances, R. J. (1986). Varieties of psychopathology found in patients with addictive disorders: A review. In R. E. Meyer (Ed.), Psychopathology and addictive disorders (pp. 17–38). New York: Guilford Press.

American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author.

American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Anthony, J. C., Folstein, M., Romanoski, A. J., Von Korff, M. R., Nestadt, G. R., Chahal, R., Merchant, A., Brown, C. H., Shapiro, S., Kramer, M., & Gruenberg, E. M. (1985). Comparison of the lay Diagnostic Interview Schedule and a standardized psychiatric diagnosis: Experience in eastern Baltimore. Archives of General Psychiatry, 42, 667–675.

Blashfield, R. K. (1991). An American view of the ICD-10 personality disorders. Acta Psychiatrica Scandinavica, 82, 250–256.

Bourdon, K. H., Rae, D. A., Locke, B. Z., Narrow, W. E., & Regier, D. A. (1992). Estimating the prevalence of mental disorders in U.S. adults from the Epidemiologic Catchment Area Study. Public Health Report, 107, 663–668.

Broadhead, W. E., Blazer, D. G., George, L. K., & Kit, T. C. (1990). Depression, disability days, and days lost from work in a prospective epidemiologic survey. JAMA, 264, 2524–2528.

Burke, K. C., Burke, J. D., Rae, D. S., & Regier, D. A. (1991). Comparing age of onset of major depression and other psychiatric disorders by birth cohorts in five U.S. community populations. Archives of General Psychiatry, 48, 789–795.

Burke, P. J. (1996). Social identities and psychosocial stress. In H. B. Kaplan (Ed.), Psychosocial stress: Perspectives on structure, theory, life-course, and methods (pp. 141–174). San Diego: Academic Press.

Chambless, D. L., Cherney, J., Caputo G. D., & Rheinstein, B. J. (1987). Anxiety disorders and alcoholism. Journal of Anxiety Disorders, 1, 24–40.

de Girolamo, G., & Reich, J. H. (1993). Epidemiology of mental disorders and psychosocial problems: Personality disorders. Geneva: World Health Organization.

Dohrenwend, B. P. (1995). The problem of validity in field studies of psychological disorders—revisited. In M. T. Tsuang, M. Tohen, & G.E.P. Zahner (Eds.), Textbook in psychiatric epidemiology (pp. 3–22). New York: John Wiley & Sons.

Dohrenwend, B. P., & Dohrenwend, B. S. (1965). The problem of validity in field studies of psychological disorder. Journal of Abnormal Psychology, 70, 52–69.

Dohrenwend, B. P., Levav, I., Shrout, P. E., & Schwartz, S. (1992). Socioeconomic status and psychiatric disorders: The causation selection issue. Science, 255, 946–952.

Dohrenwend, B. P., Yager, T. J., Egri, G., & Mendelsohn, F. S. (1978). The psychiatric status schedule (PSS) as a measure of dimensions of psychopathology in the general population. Archives of General Psychiatry, 35, 731–739.

Dryfoos, J. G. (1990). Adolescents at risk. New York: Oxford University Press.

Endicott, J., & Spitzer, R. (1978). A diagnostic interview: The Schedule for Affective Disorders and Schizophrenia. Archives of General Psychiatry, 35, 837–844.

Feighner, J. P., Robins, E., & Guze, S. B. (1972). Diagnostic criteria for use in psychiatric research. Archives of General Psychiatry, 26, 57–63.

Gilbert, D. G., & Connolly, J. J. (1991). Personality, social skills, and psychopathology: An individual differences approach. New York: Plenum Press.

Hamburg, D. A. (1992). Today’s children: Creating a future for a generation in crisis. New York: Random House.

Helzer, J. E., Stoltzman, R. K., Farmer, A., Brockington, I. F., Plesons, D., Singerman, B., & Works, J. (1985). Comparing the DIS with a DIS/DSM-III–based physician reevaluation. In W. W. Eaton & L. G. Kessler (Eds.), Epidemiologic field methods in psychiatry (pp. 285–308). Orlando, FL: Academic Press.

Hillis, S. L., & Woolson, R. F. (1995). Analysis of categorical data: Use of the odds ratio as a measure of association. In M. T. Tsuang, M. Tohen, & G. E. P. Zahner (Eds.), Textbook in psychiatric epidemiology (pp. 55–80). New York: John Wiley & Sons.

Howe, G. W., Caplan, R. D., Foster, D., Lockshin, M., & McGrath, C. (1995). When couples cope with job loss. A research strategy for developing preventative interventions. In L. R. Murphy, J. J. Hurrell, S. L. Sauter, & G. P. Keita (Eds.), Job stress interventions (pp. 131–158). Washington, DC: American Psychological Association.

Hunter, R., & Macalpine, I. (Eds.). (1963). Three hundred years of psychiatry. Oxford, England: Oxford University Press.

Johnson, J., Weissman, M. M., & Klerman, G. L. (1992). Service utilization and social morbidity associated with depressive symptoms in the community. JAMA, 267, 1478–1483.

Kass, F. (1985). Scaled ratings of DSM-III personality disorders. American Journal of Psychiatry, 142, 627–630.

Keith, S. J., Regier, D. A., & Rae, D. S. (1991). Schizophrenic disorders. In L. N. Robins & D. A. Regier (Eds.), Psychiatric disorders in America: The Epidemiologic Catchment Area Study (pp. 33–52). New York: Free Press.

Kendler, K. S., Davis, C. G., & Kessler, R. C. (1997). The familial aggregation of common psychiatric and substance use disorders in the National Comorbidity Study: A family history study. British Journal of Psychiatry, 170, 541–548.

Kendler, K. S., Gallagher, T. J., Abelson, J. M., & Kessler, R. C. (1996). Lifetime prevalence, demographic risk factors and diagnostic validity of nonaffective psychosis as assessed in a U.S. community sample: The National Comorbidity Survey. Archives of General Psychiatry, 53, 1022–1031.

Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1992). A population based twin study of major depression in women: The impact of varying definitions of illness. Archives of General Psychiatry, 49, 257–266.

Kendler, K. S., Walters, E. E., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1995). The structure of the genetic and environmental risk factors for six major psychiatric disorders in women. Archives of General Psychiatry, 52, 374–383.

Kessler, R. C. (1995). Epidemiology of psychiatric comorbidity. In M. T. Tsuang, M. Tohen, & G.E.P. Zahner (Eds.), Textbook in psychiatric epidemiology (pp. 179–197). New York: John Wiley & Sons.

Kessler, R. C., Berglund, P. A., Foster, C. L., Saunders, W. B., Stang, P. E., & Walters, E. E. (1997). The social consequences of psychiatric disorders: II. Teenage parenthood. American Journal of Psychiatry, 154, 1405–1411.

Kessler, R. C., Berglund, P. A., Zhao, S., Leaf, P. J., Kouzis, A. C., Bruce, M. L., Friedman, R. M., Grosser, R. C., Kennedy, C., Kuehnel, T. G., Laska, E. M., Manderscheid, R. W., Narrow, B., Rosenbeck, R. A., Santoni, T. W., & Schneier, M. (1996). The 12-month prevalence and correlates of serious mental illness (SMI). In R. W. Manderscheid & M. A. Sonnenschein (Eds.), Mental health, United States, 1996 (pp. 59–70). Washington, DC: U.S. Government Printing Office.

Kessler, R. C., & Forthofer, M. S. (in press). The effects of psychiatric disorders on family formation and stability. In J. Brooks-Gunn & M. Cox (Eds.), Family risk and resilience: The roles of conflict and cohesion. New York: Cambridge University Press.

Kessler, R. C., Foster, C. L., Saunders, W. B., & Stang, P. E. (1995). Social consequences of psychiatric disorders, I: Educational attainment. American Journal of Psychiatry, 152, 1026–1032.

Kessler, R. C., & Frank, R. G. (1997). The impact of psychiatric disorders on work loss days. Psychological Medicine, 27, 861–873.

Kessler, R. C., Kendler, K. S., Heath, A., Neale, M. C., & Eaves, L. J. (1992). Social support, depressed mood, and adjustment to stress: A genetic epidemiologic investigation. Journal of Personality and Social Psychology, 62, 257–272.

Kessler, R. C., Kendler, K. S., Heath, A., Neale, M. C., & Eaves, L. J. (1994). Perceived support and adjustment to stress in a general population sample of female twins. Psychological Medicine, 24, 317–334.

Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., Hughes, M., Eshleman, S., Wittchen, H-U., & Kendler, K. S. (1994). Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: Results from the National Comorbidity Survey. Archives of General Psychiatry, 51, 8–19.

Kessler, R. C., & Price, R. H. (1993). Primary prevention of secondary disorders: A proposal and agenda. American Journal of Community Psychology, 21, 607–634.

Kessler, R. C., Rubinow, D. R., Holmes, C., Abelson, J. M., & Zhao, S. (1997). The epidemiology of DSM-II-R bipolar I disorder in a general population survey. Psychological Medicine, 37, 1079–1089.

Lenzenweger, M. F., Loranger, A. W., Korfine, L., & Neff, C. (1997). Detecting personality disorders in a non-clinical population: Application of a two-stage procedure for case identification. Archives of General Psychiatry, 54, 345–351.

Loranger, A. W., Sartorius, N., & Janca, A. (Eds.). (1996). Assessment and diagnosis of personality disorders: The International Personality Disorder Examination (IPDE). New York: Cambridge University Press.

Magee, W. J., Eaton, W. W., Wittchen, H-U., McGonagle, K. A., & Kessler, R. C. (1996). Agoraphobia, simple phobia, and social phobia in the National Comorbidity Survey. Archives of General Psychiatry, 53, 159–168.

Mazure, C. M. (Ed.). (1995). Does stress cause psychiatric illness? Washington, DC: American Psychiatric Press.

McGrath, E., Keita, G. P., Strickland, B. R., & Russo, N. F. (1990). Women and depression: Risk factors and treatment issues. Washington, DC: American Psychological Association.

Merikangas, K. R. (1988). Epidemiology of DSM-III personality disorders. In R. Michels & J. O. Cavenar, Jr. (Eds.), Psychiatry (Vol. 3, pp. 1–16). Philadelphia: Lippincott.

Morey, L. C. (1988). Personality disorders in DSM-III and DSM-III-R: Convergence, coverage and internal consistency. American Journal of Psychiatry, 145, 573–577.

National Advisory Mental Health Council. (1993). Health care reform for Americans with severe mental illness. American Journal of Psychiatry, 150, 1447–1465.

Newman, S. C., Shrout, P. E., & Bland, R. C. (1990). The efficiency of two-phase designs in prevalence surveys of mental disorders. Psychological Medicine, 20, 183–193.

Parker, G. (1987). Editorial: Are the lifetime prevalence estimates in the ECA study accurate? Psychological Medicine, 17, 275–282.

Penick, E. C., Powell, B. J., Liskow, B. I., & Jackson, J. O. (1988). The stability of coexisting psychiatric syndromes in alcoholic men after one year. Journal of Studies on Alcohol, 49, 395–405.

Perry, J. S. (1992). Problems and considerations in the valid assessment of personality disorders. American Journal of Psychiatry, 149, 1645–1653.

Pilkonis, P. A., Heape, C. L., Proiette, J. M., Clark, S. W., McDavid, J. D., & Pitts, T. E. (1995). The reliability and validity of two structured diagnostic interviews for personality disorders. Archives of General Psychiatry, 52, 1025–1033.

President’s Commission on Mental Health and Illness. (1978). Report to the President from the President’s Commission on Mental Health, Volume 1. Washington, DC: U.S. Government Printing Office.

Price, R. H., & Vinokur, A. D. (1995). Supporting career transitions in a time of organizational downsizing: The Michigan JOBS Program. In M. London (Ed.), Employees, careers and job creation: Developing growth-oriented human resource strategies and programs (pp. 191–209). San Francisco: Jossey-Bass.

Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychology Measurement, 1, 385–401.

Regier, D. A., Farmer, M. E., Rae, D. A., Locke, B. Z., Keith, B. J., Judd, L. L., & Goodwin, F. K. (1990). Comorbidity of mental health disorders with alcohol and other drug abuse. JAMA, 264, 2511–2518.

Regier, D. A., Kaelber, C. T., Rae, D. S., Farmer, M. E., Knauper, B., Kessler, R. C., & Norquist, G. S. (in press). Limitations of diagnostic criteria and assessment instruments for mental disorders: Implications for research and policy. Archives of General Psychiatry.

Regier, D. A., Narrow, W. E., Rae, D. S., Manderscheid, R. W., Locke, B. Z., & Goodwin, F. K. (1993). The de facto U.S. Mental and Addictive Disorders Service System: Epidemiologic Catchment Area prospective 1-year prevalence rates of disorders and services. Archives of General Psychiatry, 50, 85–94.

Rhode, P., Lewinsohn, P., & Seeley, J. (1990). Are people changed by the experience of having an episode of depression? A further test of the scar hypothesis. Journal of Abnormal Psychology, 99, 264–271.

Robins, L. N. (1985). Epidemiology: Reflections on testing the validity of psychiatric interview. Archives of General Psychiatry, 42, 918–924.

Robins, L. N., Helzer, J. E., Croughan, J. L., & Ratcliff, K. S. (1981). National Institute of Mental Health Diagnostic Interview Schedule: Its history, characteristics and validity. Archives of General Psychiatry, 38, 381–389.

Robins, L. N., Locke, B. Z., & Regier, D. A. (1991). An overview of psychiatric disorders in America. In L. N. Robins & D. A. Regier (Eds.), Psychiatric disorders in America: The Epidemiologic Catchment Study (pp. 328–366). New York: Free Press.

Robins, L. N., & Regier, D. A. (1991). Psychiatric disorders in America: The Epidemiologic Catchment Area Study. New York: Free Press.

Robins, L. N., Wing, J., Wittchen, H., Helzer, J. E., Babor, T. F., Burke, J. D., Farmer, A., Jablenski, A., Pickens, R., Regier, D. A., Sartorius, N., & Towle, L. H. (1988). The Composite International Diagnostic Interview: An epidemiologic instrument suitable for use in conjunction with different diagnostic systems and in different cultures. Archives of General Psychiatry, 45, 1069–1077.

Rogler, L. H., Malgady, R. G., & Tryon, W. W. (1992). Evaluation of mental health: Issues of memory in the Diagnostic Interview Schedule. Journal of Nervous and Mental Disease, 180, 215–222.

Rothman, K. J. (1986). Modern epidemiology. Boston: Little, Brown.

Roy, A., DeJong, J., Lamparski, D., Adinoff, B., George, T., Moore, V., Garnett, D., Kerich, M., & Linnoila, M. (1991). Mental disorders among alcoholics: Relationship to age of onset and cerebrospinal fluid neuropeptides. Archives of General Psychiatry, 48, 423–427.

Seiler, L. H. (1973). The 22-item scale used in field studies of mental illness: A question of method, a question of substance, and a question of theory. Journal of Health and Social Behavior, 14, 252–264.

Spitzer, R. L., Williams, J.B.W., Gibbon, M., & First, M. B. (1990). Structured Clinical Interview for DSM-III-R, Patient edition (SCID-P), Version 1.0. Washington, DC: American Psychiatric Press.

Substance Abuse and Mental Health Service Administration. (1993). Final notice establishing definitions for (1) children with a serious emotional disturbance, and (2) adults with a serious mental illness. Federal Register, 58, 29422–29425.

Taylor, S. E., & Aspinwall, L. G. (1996). Mediating and moderating processes in psychosocial stress: Appraisal, coping, resistance, and vulnerability. In H. B. Kaplan (Ed.), Psychosocial stress: Perspectives on structure, theory, life-course, and methods (pp. 71–110). San Diego: Academic Press.

Tsuang, M. T., & Farone, J. V. (1990). The genetics of mood disorders. Baltimore: Johns Hopkins University Press.

Tsuang, M. T., Tohen, M., & Zahner, G.E.P. (Eds.). (1995). Textbook in psychiatric epidemiology. New York: John Wiley & Sons.

Tyrer, P., Casey, P., & Ferguson, B. (1991). Personality disorder in perspective. British Journal of Psychiatry, 159, 463–471.

Weissman, M. M. (1993). The epidemiology of personality disorders: A 1990 update. Journal of Personality Disorders, 7, 44–62.

Weissman, M. M., & Myers, J. K. (1978). Affective disorders in a U.S. urban community. Archives of General Psychiatry, 35, 1304–1311.

Wells, K., Stewart, A., Hays, R., Burnam, M., Rogers, W., Daniels, M., Berry, S., Greenfield, S., & Ware, J. (1989). The functioning and well-being of depressed patients: Results from the Medical Outcomes Study. JAMA, 262, 914–919.

Wittchen, H-U. (1994). Reliability and validity studies of the WHO–Composite International Diagnostic Interview (CIDI): A critical review. Journal of Psychiatric Research, 28, 57–84.

Wohlfarth, T. D., van den Brink, W., Ormel, J., Koeter, M. W., & Oldehinkel, A. J. (1993). The relationship between social dysfunctioning and psychopathology among primary care attenders. British Journal of Psychiatry, 163, 37–44.

World Health Organization. (1977). Manual of the international statistical classification of diseases, injuries, and causes of death (9th rev.). Geneva: Author.

World Health Organization. (1990). Composite International Diagnostic Interview (CIDI), Version 1.0. Geneva: Author.

Zahner, G.E.P., Hsieh, C. C., & Fleming, J. A. (1995). Introduction to epidemiologic research methods. In M. T. Tsaung, M. Tohen, & G.E.P. Zahner (Eds.), Textbook in psychiatric epidemiology (pp. 23–54). New York: John Wiley & Sons.

Zimmerman, M. (1994). Diagnosing personality disorders: A review of issues and research methods. Archives of General Psychiatry, 51, 225–245.

The authors appreciate the helpful comments of Evelyn Bromet, Sheldon Danziger, Phil Leaf, and Uli Wittchen. More detailed results of the National Comorbidity Survey can be obtained by consulting the NCS homepage, http://www.hcp.med.harvard.edu/ncs.