communities (or representative samples of whole communities) is usually needed to answer questions about prevalence (the total number of cases in a given period of time) and incidence (the number of new cases in a population). In addition, many young people have more than one MEB disorder (Angold, Costello, and Erkanli, 1999). This comorbidity can increase the severity of a disorder (Kessler, Chiu, et al., 2005). Rates of comorbidity cannot be determined using clinic-based data, because cases seen in treatment settings are different in many ways from untreated cases (Berkson, 1946). Population-level information is needed to determine which diseases are of public health concern. It needs to encompass a wide range of disorders, including their rates of occurrence and co-occurrence and the burden they cause to individuals, their families, and the social organizations and agencies in which individuals live their lives.

The standard method of finding out how many cases of a disease exist in the community is to carry out a randomized survey of the general population. The size of a sample needed to provide precise answers to questions about the prevalence of an emotional or behavioral disorder depends on how common or rare it is. The less common the disorder, the larger the sample needed to provide a reliable prevalence estimate. For example, if a disorder occurs in 1 child in 10,000, researchers would need a population sample of at least 1 million children to find approximately 100 cases.

If a disorder produces such a high level of disability that every case comes to the attention of doctors, schools, or other agencies, then agency records can sometimes be used to estimate prevalence and even incidence. This method has been used by the Centers for Disease Control and Prevention (CDC) to estimate the prevalence of autism. In some countries, databases of inpatient and outpatient treatment are maintained and can be used to estimate treated prevalence. But many MEB disorders rarely come to the attention of doctors or teachers. Studies in the United States show that fewer than one in eight children with an MEB disorder is currently receiving treatment in the mental health or substance abuse systems, and only about one in four has ever received treatment (Burns, Costello, et al., 1995; Farmer, Burns, et al., 2003; Kataoka, Zhang, and Wells, 2002). To estimate the full burden of MEB disorders among children and adolescents, it is usually necessary to interview large community-based samples of parents and their children.

As mentioned earlier, there have been two recent surveys of mental illness in representative samples of the U.S. adult population: the NCS (Kessler, 1994), a follow-up of the same participants (NCS-2) (Kessler, Gruber, et al., 2007), and a second sample (NCS-R) a decade later (Kessler, Chiu, et al., 2005). The NCS included no one younger than 15. The NCS-R includes a sample of 10,000 adolescents (ages 13-17), but the data on this sample are not yet published. Although the United States supports several

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement