We should welcome the three studies on children and adolescents in this issue (by Mannuzza et al., Myers et al., and Harrison et al.) but perhaps with some mixed feelings. It seems reasonable to expect that any issue of the American Journal of Psychiatry should have several articles about children and adolescents, who make up one-third or more of the U.S. population. Yet far fewer than one-third of the articles in this distinguished journal are about children and adolescents. That may be a sign of trouble, with multiple causes and sequelae. It is only a partial explanation to say that good child and adolescent studies are often even harder, longer, and more expensive to do than adult studies.

The three papers are all admirable. Their numbers of subjects vary from fairly large (N=104) to huge (N=74,008). All three had federal funding (the National Institute of Mental Health; the National Institute on Alcohol and Alcohol Abuse, the Department of Veterans Affairs, and the National Institute on Drug Abuse; and the Substance Abuse and Mental Health Services Administration). Two came from university/hospital/clinic work; one from a state human services department in cooperation with most of that state's schools (!). The three are geographically spread (California, New York, and Minnesota). All three rely on questionnaires and/or predominantly structured interviews. All three depend heavily on DSM-IV diagnostic categories, and all tend to involve computer-play with correlations of largely descriptive behavioral categories.

Many important areas and ways of thinking about youngsters are not present in these—or any three—papers. Here, as often happens in recent research, I particularly miss depth, psychology and psychodynamics, family, society, biopsychosocial integration, and development. Comparing two points in time is fine, but development, a central clinical concept in understanding all youngsters, is not just the passage of time. The papers use statistical association models, currently rather dominant in psychiatric research, rather than models, however tentative and testable, involving deeper study of individual people and broader questions of biopsychosocial causality and meaning. The very word "model" can get co-opted and bent, as when Myers et al. offer as "the model that best fit the data" a purely statistical effort: "(–2 log likelihood χ²=134.26; model χ²=12.50, p<0.01; goodness of fit index=123.58, df=130)."

Some potentially large qualms about the validity of DSM-IV, especially for children and adolescents, and of self-report questionnaires are expressed, but the basic framework of the three studies remains standard DSM-IV and questionnaires.

Not a single child psychiatrist is an author or co-author of any of these three articles. That fact seems striking, and probably unfortunate from the point of view of cooperation, practical long-term validity, solidity, subtlety, clinical relevance, and biopsychosocial integration.

The study by Myers et al. calls attention to 15- and 20-year-olds with some fairly common behavioral problems. It illustrates advantages and limitations of DSM-IV and behavioral studies and may seem superfluous to many clinicians, or even somewhat tautological, in ways born of defining the world by DSM-IV. It treats conduct disorder, antisocial personality disorder, and substance abuse as discrete and solid things and is interested in "progression." Many clinicians, however, would see the predominant issues as concerning teenagers with probably complex, overlapping, imperfectly defined behavioral descriptive clusters as well as integrally related biological, psychological, familial, and social components that, not particularly surprisingly (due presumably to factors outside or beneath the view of a behavioral focus), persist.

Such studies are, in a way, conversations held between DSM-IV and DSM-IV.

Male-female differences are notable but hardly examined. The study by Myers et al. could be summarized as saying that more deviant behavior and pretreatment drug use before age 16 best predicts more similar pathology at age 20—i.e., "progress" from conduct disorder to antisocial personality disorder. Note that conduct disorder is nearly wholly a diagnosis made before age 18; antisocial personality disorder is wholly a diagnosis made after age 18. The study compares people at age 16 with themselves at age 20 (i.e., before versus after age 18) and finds that lots of people with conduct disorder and a history of substance abuse at age 15.9—61%—have antisocial personality disorder at age 20.

The study by Mannuzza et al. follows a cohort of 104 white boys with attention deficit-hyperactivity disorder (ADHD) into young adulthood (age 24 or, in a previous, parallel study of 103 boys, age 25). Looking thoughtfully at young adults, the authors find that, as opposed to several other possibly expectable areas of pathology, young adult antisocial personality and nonalcohol substance abuse were significantly more prevalent in the probands than in comparison subjects with no history of ADHD. It somewhat surprisingly finds that ADHD was rare (4%) in young adult men who had had childhood ADHD. Since only three cohorts of such youngsters have been followed into adulthood (albeit very young adulthood) and reported (two by these authors), and since adult ADHD is a fairly lively topic in the current psychiatric literature, it seems a pity that the authors did not make more of the apparent, dramatic drop in ADHD among both of their proband groups from age 18 to age 24 or 25.

Mannuzza et al. suggest that the three follow-up studies so far suggest that "there is no `true' or universal rate of adult ADHD." Perhaps. Or, more modestly, perhaps the rate is small, and we have not yet found what it is, despite this excellent study.

The study by Harrison et al. is memorably large: 74,008 students adequately completed questionnaires. Such size creates considerable power, even if limitations include shallowness, even some narrowness, behavioral focus, blending of different categories (use of all substances was blended into generic "substance" use), reliance on questionnaires, and reliance on self-reporting in teenagers (in a realm where self-delusion, peer pressure, wishes to conform, wariness of adult truth-seeking, etc., are not unlikely to create systematic distortions). The authors do consider some important differences between teens and adults as well as other limitations of their survey (e.g., the self-report problems and the fact that the most serious substance abusers may have been left out). Even if the bits of data are imperfect, however, such large studies retain significant strength. Apparently ignoring caffeine and nicotine (alas), the authors offer as one conclusion that alcohol is by far the most common substance used by high schoolers, with marijuana a very distant second. Another conclusion may have enough clout to change something: the authors found that DSM-IV's adult-derived criteria for substance abuse and substance dependence are probably less useful, at least for teenagers, than would be the simpler spectrum of criteria and total number of symptoms they propose.

Overall, these three interesting papers do not define child or adolescent psychiatry, but they can remind us that energetic involvement of child and adolescent clinicians in research and writing remains a major challenge and opportunity in psychiatry.

Address reprint requests to Dr. Hartmann, 147 Brattle St., Cambridge, MA 02138.