Adolescent outcomes of childhood conduct disorder among clinic-referred boys: predictors of improvement

Journal of Abnormal Child Psychology,  August, 2002  by Benjamin B. Lahey,  Rolf Loeber,  Jeffrey Burke,  Paul J. Rathouz

• E-mail
• Print
• Link

Children are said to exhibit conduct disorder (CD) if they engage in multiple antisocial behaviors, including deception, physical aggression, and violation of property rights. Because CD is highly impairing to the individual and harmful to victims, it has been the subject of many studies. Although a great deal is known about the adolescent outcomes of childhood behavior problems in general (Farrington, Loeber, & Van Kammen, 1990; Fergusson, Horwood, & Lynskey, 1995; Nagin & Tremblay, 2001; Stattin & Magnusson, 1989; Verhulst & Van der Ende, 1992), less is known about the adolescent course of conduct problems in the subgroup of children whose problem behaviors are diverse and serious enough to meet diagnostic criteria for CD. Because it may be misleading to generalize findings on the longitudinal course of samples of children who mostly exhibit minor behavior problems to children who meet criteria for CD, it is important also to study the longitudinal course of the children at the extreme end of the behavior p roblem continuum who meet criteria for CD.

Some aspects of the longitudinal course of CD have been studied extensively (Frick & Loney, 1999). In a number of studies, children with serious conduct problems were reassessed in adulthood, with these studies suggesting that (a) adults who meet criteria for antisocial personality disorder (APD) almost always exhibited CD as children, and (b) although most children with CD show occupational and social dysfunction as adults, only about one third of children with CD later meet criteria for adult APD (Bardone, Moffilt, Caspi, & Dickson, 1996; Harrington, Fudge, Rutter, Pickles, & Hill, 1991; Robins, 1966, 1978; Robins, West, & Herjanic, 1975; Storm-Mathisen & Vaglum, 1994; Zoccolillo, Pickles, Quinton, & Rutter, 1992). What is not clear, however, is whether the children with CD who do not meet criteria for adult APD recover from CD during adolescence or continue to exhibit CD through adolescence, but do not meet the specific criteria for APD during adulthood.

Three longitudinal studies of population-based samples provide preliminary evidence on the adolescent course of conduct problems among children who met diagnostic criteria for CD in childhood. In the Isle of Wight Study, 35% of 93 children who met criteria for CD at 10-11 years of age continued to meet criteria for CD at 14-15 years of age (Graham & Rutter, 1973), but in a similar German study, most (8 of 11) boys who met criteria for CD at age 8 years met criteria for CD at age 13 years (Esser, Schmidt, & Woemer, 1990). In the Ontario Child Health Study, 45% of 36 children with CD when they were 4-12 years of age, met criteria for CD again at 8-16 years (Offord, Boyle, Racine, Fleming, et al., 1992). If the results of these three studies are combined to provide a tentative "meta-analytic" estimate of the stability of childhood CD into adolescence, only 56 of 140 (40%) children given the diagnosis of CD were given the same diagnosis again during adolescence. This suggests that slightly more than half of child ren with CD cease to meet criteria for CD sometime during late childhood or adolescence.

There are at least three reasons why additional research is needed on the adolescent outcomes of CD. First, in a preliminary report covering the first 4 years of the present study (Lahey et al., 1995), we found that boys who met diagnostic criteria for CD in the initial assessment showed levels of CD behaviors in subsequent waves that fluctuated above and below the diagnostic threshold for CD over time. As a result, estimating the persistence of CD from a single follow-up assessment greatly underestimates the stability of CD. It is possible, therefore, that these previous studies of the adolescent outcome of CD, which conducted only a single follow-up assessment, underestimated the persistence of CD. Second, Robins (1966) found a linear association between the number of childhood conduct problems and adult antisocial behavior among clinic-referred boys, suggesting that it is important to understand the course of CD at the level of the number of symptoms over time, rather than focusing only on the stability of the diagnosis. Third, the previous studies of the stability of children who received diagnoses used population-based samples and used diagnostic criteria for CD that were broader than the current DSM-IV definition. Although representative samples offer important advantages in such research, these studies did not demonstrate that the youths who received the diagnosis of CD were impaired enough to warrant clinical diagnosis and treatment. As a result of these issues, more data are needed to improve our understanding of the adolescent outcome of children with clinically-significant CD. Indeed, full knowledge of the adolescent outcome of childhood CD is currently the "missing link" in our understanding of the development of serious antisocial behavior from childhood to adulthood.

There is also a pressing need to identify childhood factors that predict which children with CD will have more or less favorable outcomes during adolescence. Identifying such early predictors will improve the prognosis of childhood CD and, if the predictors reflect modifiable causal processes that maintain CD overtime, their identification may lead to improved methods of treatment. There is considerable published evidence from longitudinal studies of samples of children with conduct problems that were not limited to those who met diagnostic criteria for CD that may point to likely baseline predictors of the outcomes of childhood CD (Frick & Loney, 1999). Consistent evidence suggests that higher levels of conduct problems in childhood predict greater stability of future conduct problems (Loeber, 1982, 1991).