Unlike the situation for many physical illnesses, there are no sensitive or specific biological tests for common mental illnesses. This has resulted in less agreement about diagnostic criteria for mental than physical disorders, which, in turn, has hampered epidemiological research on the prevalence and correlates of mental disorders. The American Psychiatric Association has tried to address this problem since the 1980s by increasing the operational precision of the diagnostic criteria in its Diagnostic and Statistical Manual (DSM) of Mental Disorders. This has made it possible to develop fully structured diagnostic interviews that can be used in community epidemiological surveys of mental disorders. The survey on which this paper was based, the National Comorbidity Survey (NCS), was the first nationally representative survey of mental disorders in the U.S. to use this type of diagnostic interview.

The ability to carry out the NCS depended critically on the vision of Darrel Regier, who was at NIMH at the time, to see the need for a comprehensive, fully structured interview to assess mental disorders, and on the work of Lee Robins, from Washington University, who developed the first fully structured diagnostic interview of mental disorders. I was fortunate to be selected to build on their vision and work in carrying out the NCS.

  What, in your view, is the significance of the paper for the field?

The paper was significant in documenting two important facts that have guided a great deal of research since that time. First, we found that a very large proportion of the population—as many as half of all Americans—have met criteria for some mental disorder at some time in their life. Second, we found that the major societal burden of mental disorders is highly concentrated in the relatively small proportion of the population—in the range of 5-8% of the population in any given year—who have a history of having several comorbid mental disorders.

The first of these two results was significant because it addressed the issue of stigma that has for so long interfered with rational thinking about mental illness. The mentally ill are not some distinct set of "them" out there who are distinct from "us" sane people. Instead, the vast majority of us have been touched by some form of mental illness at some time in our lives either through personal experience or through the illness of a close loved one. In many cases these illnesses are either mild or transient or both, but they certainly should not be considered in any way foreign.

The second result was significant because it showed that serious mental illness is associated with an accumulation of emotional difficulties. Later analysis showed that most people with high comorbidity and serious mental illness begin in childhood either with an anxiety disorder or an impulse-control disorder and then accumulate other disorders over the course of adolescence. Very few of these people receive any treatment until adulthood, typically at least a decade after the onset of their first clinically significant disorder. This observation got us to thinking about the likely effects of early treatment of primary disorders in preventing the subsequent onset of comorbidity, what we refer to as primary prevention of secondary disorders. Other researchers picked up on this idea and there is now a good deal of epidemiological research, and also some treatment research, focusing on this topic in more depth.

The paper was the first one to present a number of basic facts that have subsequently been an important starting point for many different kinds of research. In addition to the two points mentioned earlier about high overall prevalence and the strong relationship between comorbidity and disorder severity, the paper presented the first nationally representative data on the prevalence of numerous individual disorders, the early ages of onset of many mental disorders, the low rates of treatment of mental disorders, and the associations of disadvantaged social position and mental disorders. Publications of subsequent research in any of these areas have generally cited this paper as one of the foundation documents that justifies their research.

The high prevalence estimates reported in the paper were initially met with a good deal of skepticism. Although subsequent clinical calibration studies showed that the estimates are accurate, this led to a deeper questioning of the accuracy of the DSM system itself. The thinking goes like this: It’s inconceivable that half the population is mentally ill. Therefore, there must be something wrong with the DSM system. The error in this thinking is that the term "mentally ill" is being taken too seriously. It wouldn’t surprise anyone if I said that 99.9% of the population had been physically ill at some time in their life. Why, then, should it surprise anyone that 50% of the population has been mentally ill at some time in their life? The reason, of course, is that we invest the term "mentally ill" with excess meaning. A number of common mental illnesses, like adjustment disorders and brief episodes of depression, are usually mild and self-limiting. Many people experience these kinds of disorders at some time in their life. This raises a legitimate question about whether all mental illnesses require treatment and, if not, why it is that conditions not requiring treatment are defined as disorders in the DSM system. This is an ongoing debate that is too big an issue to address in this forum. It is important to note, though, that the paper we’re discussing has been a lightning rod for that debate.

This paper was the first publication of the NCS. It was designed to present a broad overview of basic patterns in the data. We have subsequently gone on to publish over 150 papers that report the results of more detailed NCS analyses. I mentioned earlier, for example, that we followed up on the initial observation that comorbidity is strongly related to severity by studying detailed patterns of temporal accumulation of comorbid conditions based on retrospective age of onset reports. This line of work has resulted in an exciting new initiative in which we screen and treat eighth graders with primary anxiety disorders in an effort both to treat primary anxiety and to prevent progression to secondary comorbid depression and substance abuse. There are a number of similar threads of NCS analysis that have been spun off into important independent initiatives in this way.

In addition, we are currently in the midst of a 10-year follow-up survey of the baseline NCS respondents in conjunction with a replication of the NCS in new samples of adolescents and adults. The follow-up survey will be used to study individual-level patterns and correlates of the onset and persistence of mental disorders over the past decade. The replication survey will be combined with the initial NCS data to study population-level changes in patterns of prevalence and treatment over the decade of the 1990’s. The adolescent survey will be used to refine our understanding of psychopathology in this critical part of the life cycle, with an eye towards targeting opportunities for early intervention and treatment. Finally, the NCS replication survey is itself now being replicated in 28 different countries around the world under the auspices of the World Health Organization. A total of over 200,000 people are being interviewed in these surveys. The Institute for Social Research at the University of Michigan is the Data Collection Coordinating Center for this initiative, while my group at Harvard is the Data Analysis Coordinating Center. All these new surveys are going to be completed over the next year, at which time we will begin analysis.