The identification of childhood precursors of adult psychiatric disorders offers the possibility of early intervention and hence prevention. In the case of antisocial personality disorder the early indicators are remarkably clear. Starting with Robins' (Reference Robins1966) classic follow-up of children referred to a clinic for conduct problems, numerous studies have shown that persistent and pervasive aggressive and disruptive behaviours seen before the age of 11 years are strongly associated with persistence of antisocial behaviours through adolescence and into adult life. As Robins described, the risk extends far beyond antisocial behaviours to unstable relationships, unreliable parenting and underachievement in education and at work (Reference Moffitt, Caspi and HarringtonMoffitt et al, 2002). This broad constellation of difficulties is reflected in DSM—IV antisocial personality disorder (American Psychiatric Association, 1994). Furthermore, children who do not have conduct problems are very unlikely to subsequently develop antisocial personality disorder (which is rare without a history of conduct problems). Conduct disorder is a specific diagnosis within DSM—IV, which requires antisocial acts seen generally in older children and adolescents. In this paper the terms ‘conduct problems’ and ‘the conduct disorders’ are used to denote serious oppositional, aggressive or antisocial behaviours whether or not they meet DSM criteria for conduct disorder.
Selective review of findings published over the past 10 years in childhood predictors of antisocial personality disorder, and consideration of issues still to be addressed in relation to early identification of individuals at risk.
From a clinical and policy perspective, the strength of the continuity from conduct problems to antisocial personality disorder is ample grounds for making strenuous efforts to prevent the appearance of aggressive and disruptive behaviours in young children, and to intervene early once they have been identified. It is not the purpose of this paper to review the evidence for the effectiveness of prevention programmes and of early interventions for conduct problems, but a brief summary highlights the need for further refinements in early identification. A small number of adequately designed randomised controlled trials of preventive programmes to reduce conduct problems have been carried out, some of which have yielded promising results. Equally, whereas there have been some significant improvements, often the effects have been quite small; and some studies have shown no benefits (Reference LeMarquand, Tremblay, Vitaro, Hill and MaughanLeMarquand et al, 2001). There is substantial support for the effectiveness of parent management training programmes in reducing overall levels of conduct problems in children (Reference Kazdin, Hill and MaughanKazdin, 2000), and for the effectiveness of stimulants where conduct problems are associated with attention-deficit hyperactivity disorder (ADHD) (Reference Swanson, Kraemer and HinshawSwanson et al, 2001). Nevertheless, there has been considerable variability in outcomes. Parent training has been found to be less effective for the higher-risk families characterised by socio-economic disadvantage, marital discord or single parent status, high parental stress and maternal unresolved loss or trauma (Reference Routh, Hill and SteeleRouth et al, 1995; Reference KazdinKazdin, 1997). Children with more severe or chronic problems or with comorbid conditions are less likely to do well (Reference Ruma, Burke and ThompsonRuma et al, 1996). Evidence of the long-term effectiveness of psychosocial treatments for conduct problems, and of stimulants for conduct problems comorbid with ADHD, is lacking.
It may be that the problem will be solved simply through better treatment techniques; however, attention to six issues in early identification may also be of value in generating ideas for the development of interventions. First, conduct problems in young children are associated with many other adverse factors such as ineffective parenting practices, discordant and unstable families, poor peer relationships and educational failure. It is important to clarify whether it is the child's disorder that requires early identification, or these associated factors or both. Second, conduct problems in childhood are generally identified on the basis of a broad cluster of behaviours. The identification of subtypes may lead to a better understanding of underlying mechanisms, and hence to improved matching of treatment to clinical needs. Third, in approximately 50% of children with early conduct problems these do not persist into adolescence and adult life. Ways of distinguishing persisters and desisters are needed. Fourth, given the intractability of behaviour problems in some young children, we need to ask whether identification at an earlier age is possible. Fifth, the adult outcomes of children who show early conduct problems and then desist, and of those whose problems start in adolescence, need to be considered. Finally, we need to attend to the adult outcomes that we are attempting to anticipate. It may be that specific antisocial outcomes have different antecedents from those of antisocial personality disorder.
What is predictive?
It is possible that, because conduct problems are associated with a wide range of adverse individual, family and social factors, the conduct problems per se are not the antecedents of antisocial personality disorder but are markers for these other difficulties that are the true antecedents. In general, the evidence supports conduct problems as true antecedents (Reference Farrington, Loeber, Van Kammen, Robins and RutterFarrington et al, 1990). For example, studies that have assessed both conduct problems and quality of peer relationships, and then followed children over several years, have consistently found that early conduct problems predict later antisocial behaviours (Reference Tremblay, Masse and VitaroTremblay et al, 1995; Reference Woodward and FergussonWoodward & Fergusson, 1999). By contrast, the role of peer relationships has been less clear. This should not, however, be interpreted to mean that the associated factors are unimportant. For example in the Dunedin Multidisciplinary Health and Development Study, violent crime at the age of 18 years was predicted by the combination of temperamental lack of control (quick to show negative emotions when frustrated, poor impulse control) and number of changes of parental figure before the age of 13 years, which probably reflected a range of family adversities (Reference Henry, Caspi and MoffittHenry et al, 1996).
Sources of heterogeneity in the conduct disorders
Longitudinal studies from childhood to adulthood have used a wide range of ways to characterise conduct problems. Generally they have made use of summary scores generated from a range of questionnaires completed by teachers and parents (Reference Farrington, Loeber, Van Kammen, Robins and RutterFarrington et al, 1990; Reference Fergusson, Lynskey and HorwoodFergusson et al, 1996; Reference Moffitt, Caspi and DicksonMoffitt et al, 1996). The consistency of the findings may suggest that it does not matter much how the problem is defined. Equally, there are pointers to potentially important kinds of heterogeneity. Children with conduct problems and hyperactivity/inattention differ from those with ‘pure’ conduct disorder in that their problems are more severe and likely to persist, and they are more likely to have neuropsychological deficits (Reference LynamLynam, 1996). Lynam (Reference Lynam1998) has argued that children with attention-deficit hyperactivity problems are ‘fledgling psychopaths’, implying that they are more likely to show in adult life the combination of callousness, superficial charm and antisocial behaviour that characterises a sub-group of adults with antisocial personality disorder. Frick and colleagues give priority to callous—unemotional traits in childhood. In a series of studies they have demonstrated that children with antisocial problems who exhibit these traits differ from other children with antisocial problems (Reference Barry, Frick and DeShazoBarry et al, 2000) in apparently having fewer verbal deficits (Reference Loney, Frick and EllisLoney et al, 1998) and in coming from families that are not characterised by dysfunctional parenting practices seen generally in the conduct disorders (Wootton et al, 1997). Children exhibiting callous—unemotional traits may also have a deficit in processing behavioural evidence of distress in others. Associations between scores assessing callous and unemotional characteristics and a reduced ability to recognise fear and sadness have been shown in young adolescents recruited in mainstream schools and children with identified emotional and behavioural problems (Reference Blair and ColesBlair & Coles, 2003; Reference Stevens, Charman and BlairStevens et al, 2001).
Loeber et al (Reference Loeber, Wung and Keenan1993) have proposed that three contrasting patterns of childhood antisocial problems reflect different pathways for different behaviour patterns: an ‘overt’ pathway characterised by bullying, followed by early fighting and proceeding to more serious violence; a ‘covert’ pathway starting with lying and stealing, and going on to more serious damage to property; and an ‘authority conflict’ pathway in which oppositional and defiant behaviours are prominent.
A further distinction, between ‘reactive’ and ‘proactive’ antisocial behaviours, cuts across this three-category typology. Reactive acts occur in response to actual or perceived threat from others, whereas proactive behaviours are initiated by the individuals (Reference Dodge and CoieDodge & Coie, 1987). Reactive aggression is thought to involve angry retaliation, in contrast to the cold unprovoked calculation of proactive aggression. Dodge et al (Reference Dodge, Lochman and Harnish1997) reported that, compared with children showing proactive aggression, ‘reactive’ children were more likely to have been physically abused, to have poor peer relationships, to have shown aggression from an earlier age and to have attention-deficit and hyperactivity symptoms. A central idea in Dodge's model is that reactive aggression is mediated by a readiness to perceive hostile intent in the actions of others. However, the evidence for this is inconsistent. At this stage these can be considered as promising subtypes that may lead to a more precise specification of mechanisms, and hence provide pointers to different kinds of intervention. Longitudinal studies to determine whether they differ in course are needed.
Who are the persisters and desisters?
We have already referred to the poor outlook of children with both conduct disorder and ADHD symptoms. On the basis of retrospective reports within a large epidemiological study, Robins & Price (Reference Robins and Price1991) found that the number of childhood antisocial problems is associated with risk of antisocial personality disorder. Studies within childhood provide some further clues regarding risk of persistence. Loeber et al (Reference Loeber, Green and Lahey2000) found that early fighting and hyperactivity predicted persistence of antisocial behaviours over a 6-year period among boys referred for conduct problems. In a prospective study of a representative general population sample from ages 7-9 years to 14-16 years, persisters had the highest levels of family adversity and lower IQ and self-esteem (Reference Fergusson, Lynskey and HorwoodFergusson et al, 1996). Children with early conduct problems that did not persist had levels of these risk factors that were intermediate between those of persisters and of children who lacked early behavioural problems. Persisters were more likely than those whose early antisocial behaviours had remitted to have a deviant peer group in adolescence. Whether this was a reflection or a cause of persistence is not clear; however, it is consistent with Sampson and Laub's argument that a key factor in determining persistence may be the presence or absence of social bonds and controls (Reference Sampson and LaubSampson & Laub, 1994).
We might suppose that, given the stability of conduct problems from the age of 3 years onwards, earlier precursors should be readily identifiable. However, the findings have been inconsistent. For example, the idea has been extensively investigated that early ‘difficult’ temperament, comprising traits such as predominantly negative emotions and ready frustration, contributes to irritable parenting, which in turn increases the risk for conduct problems. Studies using assessments of temperament based on parental reports have yielded some positive findings, but these are vulnerable to parental attributions. Recent studies have failed to demonstrate consistently that observational measures of temperament made in the first year of life predict later conduct problems (Reference Belsky, Hsieh and CrnicBelsky et al, 1998; Reference Aguilar, Sroufe and EgelandAguilar et al, 2000). Early attachment difficulties might be expected to increase the risk for later conduct problems. Here again the evidence is not convincing (Reference HillHill, 2002). It is likely that the quality of parenting in infancy is predictive of later conduct problems (Reference Belsky, Hsieh and CrnicBelsky et al, 1998) and it may be that the most promising approaches to the identification of early predictors will examine specific interactions between infant characteristics and early social experience (Reference Shaw, Owens and VondraShaw et al, 1996; Reference Belsky, Hsieh and CrnicBelsky et al, 1998).
Desisters and later onsets
We have focused so far on boys who show early conduct problems that persist into adult life. It has generally been assumed that those whose conduct problems remit have ‘recovered’. However, recent evidence from the Dunedin Study suggests that although these children are not at increased risk for antisocial outcomes, they are by no means free of difficulties (Reference Moffitt, Caspi and HarringtonMoffitt et al, 2002). At the age of 26 years they had higher rates of depression and anxiety disorders, both self- and informant-rated, and they were socially isolated, with few friends. They shared the poor educational and work records of the life-course persistent group who were antisocial as adults. Likewise, those with onset in adolescence, provisionally termed by Moffitt ‘adolescence limited’, were not free of problems by the age of 26 years. Compared with those who were not significantly antisocial in childhood or adolescence, these young men had higher rates of documented and self-reported drug and property crimes, and their informants reported more depression and anxiety symptoms.
Heterogeneity within antisocial personality disorder
Thus far in this paper the assumption has been made that the DSM-IV antisocial personality disorder category best summarises the antisocial outcomes of interest. There is little doubt that it succeeds as a broad-brush characterisation of antisocial behaviour and associated wider social dysfunction. However, it lacks specificity. In common with all DSM diagnoses, it requires the presence of a number of maladaptive behaviours or mental states identified from a larger set. Hence, the requirements can be met in numerous ways. This may limit the investigation of more specific causal factors, and so a more precise specification of the adult antisocial outcomes may be needed.
The identification of ‘psychopathic disorder’ makes the point. DSM-IV antisocial personality disorder is present in 50-80% of convicted offenders, but a much smaller group of 15-30% are judged to have characteristics such as grandiosity, callousness, deceitfulness, shallow affect and lack of remorse (Reference Hart and HareHart & Hare, 1989). These individuals are more likely than other offenders to have a history of severe and violent offences, and they may also have a distinctive deficit in interpersonal sensitivity. In a comparison of prisoners with and without psychopathic disorder, the groups did not differ in their ability to attribute correctly happiness, sadness and embarrassment to protagonists in short stories. However, in response to guilt stories, those with psychopathic disorder were more likely to attribute happiness or indifference to the protagonists (Reference Blair, Sellars and StricklandBlair et al, 1995). It has been proposed that psychopathy is associated with a failure to inhibit aggression in response to signs of distress in others, arising from a deficit in processing behavioural evidence of that distress (Reference Blair, Jones and ClarkBlair et al, 1997). There is supportive evidence that, compared with other offenders, adults with psychopathic disorder have reduced autonomic responses to distress cues (Reference Chaplin, Rice and HarrisChaplin et al, 1995; Reference Blair, Jones and ClarkBlair et al, 1997). As we saw earlier, a subgroup of children with antisocial problems who exhibit callous—unemotional traits has been identified that may parallel adults with psychopathic disorder. No studies have yet tested for continuity between child and adult psychopathic traits by following these children into adult life.
Children at risk for future antisocial personality disorder are readily identified, but evidence on the long-term effectiveness of prevention and treatment programmes is limited. Some progress has been made in identifying subgroups of children with antisocial problems in which different causal processes operate, and therefore for which there are different treatment needs. The available research does not yet tell us whether differences in the patterning, or associated features, of childhood conduct problems are predictive of distinctive adult outcomes. If is possible that this review was subject to selection bias.
Clinical Implications and Limitations
▪ The identification and treatment of conduct problems in early childhood are central to the prevention of antisocial personality disorder.
▪ The conduct disorders are heterogeneous in the patterning and course of symptoms, with implications for matching treatment to type of problem.
▪ There is considerable heterogeneity within antisocial personality disorder, so that there is a need to identify specific early indicators of particular adult antisocial outcomes.
▪ Most of the research reviewed in the article refers to antisocial personality disorder in males.
▪ Few longitudinal studies of general populations have included sufficient numbers of antisocial children to explore heterogeneity.
▪ Little is known about very early indicators of children at risk for the development of conduct problems.