U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Mental Health (UK). Antisocial Personality Disorder: Treatment, Management and Prevention. Leicester (UK): British Psychological Society; 2010. (NICE Clinical Guidelines, No. 77.)

Cover
  • March 2013: Some recommendations in sections 5.3.9, 5.4.9, 5.4.14, 5.4.19, 5.4.24 and 8.2 have been removed from this guideline by NICE. August 2018: Some recommendations have been updated to link to NICE topic pages.

March 2013: Some recommendations in sections 5.3.9, 5.4.9, 5.4.14, 5.4.19, 5.4.24 and 8.2 have been removed from this guideline by NICE. August 2018: Some recommendations have been updated to link to NICE topic pages.

Cover of Antisocial Personality Disorder

Antisocial Personality Disorder: Treatment, Management and Prevention.

Show details

5INTERVENTIONS IN CHILDREN AND ADOLESCENTS FOR THE PREVENTION OF ANTISOCIAL PERSONALITY DISORDER

5.1. INTRODUCTION

The diagnostic criteria for antisocial personality disorder stipulate that there must be evidence of conduct disorder in childhood (see DSM-IV; APA, 1994). This is consistent with epidemiological and other evidence, which demonstrates an early developmental trajectory for antisocial problems and other related difficulties (see Chapter 2). These factors, taken together with the considerable pessimism that has existed regarding treatment of antisocial personality disorder in adults, and the limited evidence that has been collected demonstrating the effectiveness of such treatment, have led to an increasing focus on interventions for children and their families to prevent the development of conduct disorder and subsequent antisocial personality disorder.

As was highlighted in Chapter 2, the development of conduct or related problems in childhood and adolescence does not mean that a person will inevitably develop antisocial personality disorder. Estimates of the probability that children who develop conduct disorder or related problems will go on to develop antisocial personality disorder generally range from 40% (Steiner & Dunne, 1997) to 70% (Gelhorn et al., 2007). Despite this variation, it seems clear that preventive interventions targeting conduct disorders in children have the potential to substantially reduce antisocial personality disorder occurrence and/or severity. The reduction of the degree of distress and damage caused to children and their families as a result of a child’s chronic conduct problems is itself, of course, a worthwhile venture. The focus in this particular chapter, however, is on the longer-term implications of treating and preventing conduct disorder in children and adolescents.

This chapter will first consider risk factors associated with the development of antisocial personality disorder (see Section 5.2). This will be followed by assessing the evidence regarding the effectiveness of early interventions for antisocial and other behavioural problems and interventions targeting children at risk of developing conduct disorder and antisocial personality disorder in later childhood or adulthood. These interventions are primarily focused on risk factors related to the parent(s), rather than the child, and they require at-risk children to be identified before the emergence of symptoms, which may be in early childhood, infancy, or even during pregnancy (see Section 5.3). The chapter will then consider separately the evidence regarding particular preventive interventions (see Section 5.4), including interventions that directly target the child (for example, Kazdin, 1995), interventions addressing the parents (for example Webster-Stratton, 1990), interventions directed at families (for example Szapocznik et al., 1989) and interventions that simultaneously target families and the wider social environment (for example Henggeler et al., 1992).

5.2. RISK FACTORS

5.2.1. Introduction

Early interventions for the prevention of antisocial personality disorders are reviewed in Section 5.3. An important debate regarding public health interventions concerns whether to focus these interventions on the population as a whole (universal prevention) or on individuals more likely to develop the disorder in the future (selected and indicated prevention). Universal prevention interventions seek to shift the population distribution of the disorder as a whole with the aim that those at the extremes of the distribution will benefit from this reduction in overall incidence of the disorder in the population. In addition, as the population is the focus of the interventions those individuals with a greater risk of developing the disorder are not stigmatised (see Farrington & Coid, 2003).

In contrast, selected and indicated preventative interventions require identifying people at risk of developing the disorder and targeting them for intervention. The advantage of this approach is that those at greatest risk receive intensive intervention and therefore such an approach is more likely to be cost effective. However, there are problems associated with the impact of labelling children (as has been discussed in more detail in Chapter 2). A further difficulty is that currently there is no specific tool or measure that can identify the relatively small number of people who go on to develop antisocial personality disorder with particularly high precision (Moran & Hagell, 2001). Advances in the knowledge of risk factors may enable identification of those at greatest risk who might particularly require intervention (Hill, 2003).

Few studies have directly sought to identify risk factors for the development of antisocial personality disorder (see Farrington & Coid, 2003). However, there are a number of studies that have examined predictors of antisocial behaviour and/or offending in adulthood that are likely to be informative in evaluating the developmental pathway to antisocial personality disorder.

5.2.2. Definition and aim of review

The aim of this review is to assess risk factors for the development of antisocial personality disorder. Risk factors reviewed in this section fall into three main categories: individual (relating to the child), family (relating to the family of the child) and social (relating to the social environment of the child).

5.2.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline can be found in Table 4. Only studies with outcome data on offending and/or the proportion of participants meeting diagnostic criteria for antisocial personality disorder or conduct disorder were included. Only cohort studies with a minimum of 5 years’ follow-up period and a minimum age at follow-up of 15 years were included.

Table 4. Databases searched and inclusion/exclusion criteria for clinical evidence.

Table 4

Databases searched and inclusion/exclusion criteria for clinical evidence.

5.2.4. Studies considered4

The review team conducted a new systematic search for cohort studies that assessed the risk factors for developing antisocial personality disorder. Twenty-nine trials examining clinical outcomes met the eligibility criteria set by the GDG. All were published in peer-reviewed journals between 1989 and 2008. In addition, 22 studies were excluded from the analysis. The most common reason for exclusion was that the data were not extractable.

5.2.5. Clinical evidence for risk factors

Evidence from the important outcomes and overall quality of evidence are presented in Table 5 (further information about included studies can be found in Appendix 17).

Table 5. Study information and summary evidence table on risk factors for developing antisocial personality disorder.

Table 5

Study information and summary evidence table on risk factors for developing antisocial personality disorder.

Studies used a variety of outcomes, therefore only very broad risk factors could be combined in the meta-analysis. As expected, child behaviour problems were associated with greater risk of antisocial personality disorder outcomes at preschool (odds ratio [OR] = 1.91; 1.66, 2.19), middle school (OR = 2.56; 2.10, 3.12) and adolescence (OR = 3.05; 2.56, 3.63). Although the presence of attention deficit hyperactivity disorder (ADHD) appeared to be a slightly stronger predictor (OR = 6.22; 4.06, 9.54).

There were a variety of family risk factors reported including parenting styles, parents’ antisocial behaviour and parental disharmony/separation. These effects were all of a similar magnitude, for example, in the combined family measure in adolescence the OR was 2.50 (1.82, 3.41).

There was slightly less data on social risk factors but in a combined analysis of factors associated with social deprivation the OR was 2.39 (1.89, 3.04).

5.2.6. Clinical summary

There have been a number of studies assessing risk factors for developing offending behaviour and adult behaviour problems, and much less on receiving a diagnosis of antisocial personality disorder. Despite the relatively large number of studies with long follow-up periods it is only possible to draw very general conclusions regarding risk factors in this field.

There appears to be a number of factors associated with antisocial personality disorder including individual child factors (for example, exhibiting behaviour problems as a child, having a diagnosis or showing symptoms of ADHD), family factors (for example, parental antisocial behaviour and harsh parenting style) and social factors (for example, low socioeconomic status). However it should also be reiterated that although these factors may be associated with a greater risk of developing antisocial personality disorder, the majority of children with such risk factors will not in fact develop the disorder in adulthood.

5.3. EARLY INTERVENTIONS

5.3.1. Introduction

The primary aim of early interventions for antisocial and other behavioural problems and interventions targeting children at risk of developing conduct disorder and antisocial personality disorder in later childhood or adulthood is preventative, and as such, for the interventions to have any value, mechanisms must be in place to identify those children, and their families, who might derive benefit from them. The current ‘lingua franca’ of prevention is based on the work of Gordon (1983), popularised by the Institute of Medicine report. It differentiates between three strategies of prevention, each defined by the group they target: (1) universal, (2) selected and (3) indicated.

Universal strategies of prevention are directed at the general population. Where applicable, the term is to be preferred over the more traditional designation of ‘primary prevention’, because it specifies that the population to which the intervention is applied is not pre-selected. Universal preventive strategies may, and most often do, identify high-risk populations, but unlike selected intervention programmes, they do not seek to identify or target individuals within a population based on individual characteristics indicative of high risk. Thus the programme is delivered universally. It is the population that is at risk (and in these interventions, that risk is generally low), not the individual within the population.

Selected prevention intervention, as a category, generally overlaps with ‘secondary prevention’, although it also includes some interventions that would be considered primary preventions. These strategies are applied to people who are markedly at risk of developing the disorder or who show its very early signs. Interventions tend to focus on the reduction of risk and the strengthening of resilience. Risk is obviously higher in these selected groups. Often this is a result of a concentration of risk factors rather than the intensity of any one factor. Hence poverty, unemployment, inadequate transportation, sub-standard housing, parental mental health problems and marital conflict may come together to affect a particular child and may be addressed in preventive programmes. For example, the Elmira Project (described fully below; see Olds et al., 1997), found that an early intensive nurse home visitation intervention worked well to prevent child maltreatment in the early years and delinquency at 15 years’ follow-up, but only in the highest risk group. These individuals were identified by the mother’s age, low socioeconomic status and single parent status.

Indicated intervention, as a category, approximately mirrors the category of tertiary prevention. These interventions are aimed at specific disorder groups, and they target people in whom prodromal symptoms of the disorder are already evident but the full disorder has not yet developed. The treatment of conduct disorder, for example, can be conceptualised as an indicated intervention for antisocial personality disorder, since conduct disorder is part of the diagnostic criteria for antisocial personality disorder. It is often hard to identify an intervention as selected or indicated based on the therapeutic activity that is involved because treatment of conduct disorder can also be regarded as a selected preventive intervention, since conduct disorder can be a risk factor for antisocial personality disorder. Cognitive behavioural therapy (CBT), for example, may be used as a treatment strategy in both selected and indicated prevention interventions for antisocial behaviour problems. Also, in practice, modern intervention programmes tend to combine universal, selective and indicated prevention into complex packages (for example, Conduct Problems Prevention Research Group, 1992).

Behavioural problems affect approximately one in seven children and have in themselves major societal, economic and personal ramifications (Scott, 2007). If untreated, up to 50% of preschool children exhibiting behavioural problems will subsequently develop severe mental health disorders, disorders such as conduct disorder and oppositional defiant disorder, and depression (for example, Tremblay et al., 2004), and the social costs of non-treatment additionally encompass the various consequences that these disorders entail, such as truancy, family stress, substance misuse, delinquency and unemployment (Barlow & Stewart-Brown, 2000). The evidence in support of management approaches to behavioural problems (including individual psychotherapy and parenting programmes), will be considered in Section 5.4. Parenting programmes share many elements with prevention programmes in that both aim to reduce harsh and abusive parenting, increase warm parenting and educate parents about normal development (for example, Barlow et al., 2005). Given that treatment services are unlikely to ever be able to meet the needs of all children with behavioural problems, the prevention of these difficulties may be an appropriate first step in reducing the prevalence and/or severity of antisocial personality disorder.

There have been many thousands of studies evaluating the effectiveness and benefits of preventive interventions for conduct disorder, although fewer RCTs (Buckner et al., 1985; Durlak, 1997; Mrazek & Haggerty, 1994; Trickett et al., 1994). In general, quasi-experimental investigations produce promising findings, but in the vast majority of cases, such positive results do not stand up to more rigorous RCT tests (Olds et al., 2007). Even more disappointing is the fact that only a handful of controlled studies have followed samples for long enough to provide clear indications of whether antisocial personality disorder may be prevented through early intervention with asymptomatic children.

Current practice

Practitioners in children’s services in the UK have become increasingly interested in focusing on prevention to address emotional and behavioural problems, including conduct disorder and related problems, in children and adolescents. A major initiative, the Sure Start Local Programmes, began in 1998 to address the needs of at-risk children by targeting those children and their families. The current prevailing view is that this programme has had only limited success, and this is generally attributed to the fact that the programme was insufficiently targeted on the families with most need (Belsky et al., 2006). However, as a response to these limitations, changes were made to the programmes, including specifying services more clearly, placing greater emphasis on the child’s well-being, focusing on reaching the most vulnerable and adjusting provision to take into account family disadvantage (Melhuish et al., 2007; Belsky et al., 2006).

The most recent evaluation suggests these modifications may have had an impact on outcomes (Melhuish et al., 2008). There were improvements (small-to-medium effect sizes) in the home learning environment, families accessing services and reduced parenting risk. However, benefits of the programme for child development were of a small magnitude. There were no statistically significant effects on the naming vocabulary sub-scale of the British Ability Scale or on child negative social behaviour and small statistically significant effects on child positive social behaviour and independence (Melhuish et al., 2008). There was some evidence that improvements to parenting and family outcomes may in turn lead to improved child outcomes but this has yet to be conclusively shown.

More recently, there has been an interest in developing and implementing programmes using the model of those developed by David Olds (see above). Such programmes, targeting vulnerable parents and children, are currently being carried out and the feasibility of their use in the UK has been tested (Barnes et al., 2008; see below).

5.3.2. Definition and aim of review

The aim of this review is to assess early interventions for antisocial and other behavioural problems and interventions targeting children at risk of developing conduct disorder and antisocial personality disorder in later childhood or adulthood. Programmes under review fall into each of the three main categories of prevention discussed above (that is, universal prevention, selected prevention and indicated prevention).

5.3.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline can be found in Table 6. This narrative review is restricted to studies with follow-up data on participants at a minimum of 15 years of age and a minimum follow-up period of at least 8 years. Only studies with outcome data on offending and/or the proportion of participants meeting diagnostic criteria for antisocial personality disorder were included.

Table 6. Databases searched and inclusion/exclusion criteria for clinical evidence.

Table 6

Databases searched and inclusion/exclusion criteria for clinical evidence.

5.3.4. Studies considered

The review team conducted a new systematic search for RCTs and quasi-experimental studies that assessed the benefits and disadvantages of early interventions for preventing antisocial personality disorder.

Seven trials examining clinical outcomes met the eligibility criteria set by the GDG (McGauhey et al., 1991; Olds et al., 1997; Schweinhardt et al., 1997; Lally et al., 1988; Campbell & Ramey, 1994; Reynolds, 1991; Hawkins et al., 1991). All were published in peer-reviewed journals and books between 1988 and 2007. Fifty-four studies were excluded from the analysis. The most common reason for exclusion was an inadequate follow-up period.

5.3.5. Clinical evidence for early interventions

Programmes for parents of infants and toddlers

This section reviews studies of interventions for infants and toddlers. Typically they are targeted at parents of newborn infants and may involve interventions in the antenatal period.

The infant health and development programme

Low birth weight is a risk factor for a range of health and developmental problems. In the early 1980s, McGauhey and colleagues devised a programme consisting of home visiting, parenting groups and educationally enriched day care, the latter designed to promote exposure to increasingly complex cognitive tasks and language experiences (McGauhey et al., 1991). Nine hundred and eighty five low birth weight newborns were assigned either to this programme or to a control condition. The sample was stratified by birth weight, with a very low birth weight group comprising infants weighing less than or equal to 2,000 g and a low birth weight group comprising infants weighing between 2,001 and 2,500 g (Brooks-Gunn et al., 1994). At the most recent follow-up, when children were 18 years old, approximately two-thirds of the sample was still adhering to the assessment protocol. An intent-to-treat analysis of data from this follow-up (McCormick et al., 2006) found the intervention to have beneficial effects in the 2,001–2,500 g group but not for the lower weight sub-sample. The effects were mainly on risk behaviours and on various measures of cognitive competence.

Analysis of the costs of the programme indicated it to be a fairly costly intervention, but a cost-benefit analysis has not been conducted since savings achieved by the programme have not yet been computed (Karoly et al., 2005). The decision to adopt enhanced care arrangements for low birth weight children should await a comprehensive cost-effectiveness analysis.

Nurse home-visiting programmes

Several studies on nurse home-visiting programmes have reported significant success in providing effective developmental support. As part of the treatment programme, the mother’s concerns about being involved in a family intervention are addressed with the goal of making the treatment programme more acceptable to these mothers and facilitating treatment delivery (Olds, 2002). In the best researched programme, the Nurse-Family Partnership (NFP), the nurse’s work is directed towards a number of aims, such as improving mothers’ prenatal health-related behaviours (for example, by reducing mothers’ consumption of cigarettes, alcohol and illegal drugs), enhancing the competence of early-life care received by the child, and helping parents develop a vision for their futures, plan subsequent pregnancies, complete their educations, find work, and enhance their economic self-sufficiency. Fathers, grandmothers, and other concerned family members or friends are systematically involved in the programme, which also involves linking families with health and other services. Nurses receive detailed visit-by-visit programme guidelines to structure their work (Olds et al., 2003).

The NFP model has been tested in three separate RCTs since 1977 (Olds et al., 1997, 1998, 2002, 2004; Kitzman et al., 1997, 2000; Olds et al., 2002, 2004). The first of these studies, conducted in Elmira, New York, with a sample of 400 low income, primarily white families, collected follow-up data on families up to the point the child turned 15 (Olds et al., 1997, 1998). The other two studies, one in Memphis with a sample of 1,138 low income, primarily African American families (Kitzman et al., 1997, 2000), and one in Denver with a sample of 735 families, including a large portion of Hispanics (Olds et al., 2002, 2004), yielded data that provided, though not unequivocally, additional support for the approach, although neither study reported follow-up data beyond 6 years. High rates of adherence to the evaluation protocol were achieved in the studies, with between 81 and 86% of mothers randomised being successfully followed-up for assessment at 4 to 15 years.

Data from the 15-year follow-up of the Elmira sample (Olds et al., 1997) showed differences in rates of state-verified reports of child abuse and neglect between treatment and control groups, with families visited by nurses during pregnancy and infancy being 48% less likely to be identified as perpetrators of child abuse and neglect; for families with unmarried mothers and for low socioeconomic status families, the effect of the programme on maltreatment was increased, but if there was domestic violence in the household, the effect of the programme on maltreatment was reduced. There were also fewer arrests, convictions and days of incarceration among mothers visited by nurses. Importantly for this guideline, young people whose mothers were visited by nurses had 59% fewer arrests and 90% fewer adjudications as persons in need of supervision for incorrigible bad behaviour. They had fewer (although not quite significant statistically) convictions and violations of probation and fewer sexual partners. These and other beneficial effects of the programme were more notable in the families with the most economically deprived unmarried mothers. The impact of the programme was insufficient to cause changes in teachers’ reports of behaviour problems, school suspensions and parents’ or children’s reports of major or minor acts of delinquency (Olds et al., 1998).

The Memphis study replicated many of the initial results from the early follow-ups of the New York project (Kitzman et al., 1997, 2000). In the Memphis study, follow-up in middle childhood revealed that children in the experimental group had higher intellectual functioning and receptive vocabulary, fewer behavioural problems in the borderline or clinical range and expressed less aggression and incoherence in response to story stems compared with children in the control group (Olds et al., 2004). Nurses in the Denver trial produced effects consistent with the previous two trials (Olds et al., 2002, 2004), and testing at 4-year follow-up showed more advanced language, superior executive functioning and better behavioural adaptation in those children from the nurse-visited group whose mothers had low psychological resources than in similar children from the control group. Notably, paraprofessionals, who were also employed to deliver the programme, produced about half the effects that nurses were able to deliver.

Based on these three trials, the Washington State Institute for Public Policy estimated that for every family served by nurses, society experiences a $17,000 return on the investment (Aos et al., 2004). Thus, according to US evaluations, the NFP qualifies as an evidence-based community health programme, one that can help transform the lives of vulnerable mothers pregnant with their first children. A key element of implementation is enrolling first-time, low-income mothers early in pregnancy.

NFP is currently being implemented in ten pilot sites in England (Barnes et al., 2008). Families have been recruited through NHS systems, with age as the single inclusion criteria for expectant first-time mothers under 20 years (income data not often available) and a slightly more elaborate set of inclusion criteria applied to expectant first-time mothers between the ages of 20 and 23 years (not in employment, education or training and never employed/had no qualifications or no stable relationship with the baby’s father). In the first year, in all pilot sites, a total of 1,217 young mothers (average age 17.9 years, range 13 to 24 years), or 87% of those eligible for the programme, were successfully given treatment. Out of 7,500 nurse visits, a father was present for 1,820.

The first-year report of the evaluating team (Barnes et al., 2008) suggest that delivery of NFP programmes meeting standards for good treatment fidelity is possible in the UK. This conclusion was based on the following observations:

  1. Appropriate clients have been recruited.
  2. NFP was delivered effectively in all sites.
  3. NFP was acceptable to UK clients.
  4. NFP was acceptable also to fathers and other family members.
  5. NFP was acceptable to health visitor practitioners delivering the programme.
  6. Organisational infrastructure and support were seen as favourably impacting on successful delivery.

Initial indicators of effectiveness are promising, with many clients reporting plans to return to education, closer involvement of fathers with infants, greater confidence as parents, and engaging in activities with children likely to enhance cognitive and social development. The data so far collected on the health-related changes that have already been observed in mothers as a result of treatment participation (for example, reduced smoking) may reasonably be expected to enhance child health and reduce negative child outcomes (for example, asthma).

In England, as in the US, NFP appears to function as an important bridge to other services for the most ‘hard-to reach’. However, the history of prevention efforts make it clear that the true impact of NFP in the UK cannot be determined until a randomised UK trial has been conducted.

Preschool programmes for infants and toddlers

This section reviews studies on interventions for infants and toddlers typically at 6 months and up to 5 years of age. These interventions may involve preschool nursery programmes, educational interventions, and home visiting.

The High-Scope Perry Preschool Project

Of all preschool programmes aimed at disadvantaged children, the Perry Preschool Project is perhaps the best documented. The programme’s initial goal (Schweinhart et al., 1993) was to better equip poor minority children for school entry. It focused on poor families from a high-risk group, had low attrition rates and a follow-up to age 40. It included 30 weekly special classes each lasting 2½ hours, as well as weekly home visits by teachers. Most children participated for 2 years. Active learning and the facilitation of independence and self-esteem were the focus of the intervention. Problem-–solving skills and task persistence were also strongly encouraged. The teachers were highly skilled, supervised and had a special brief to establish good home-school integration.

In the study under review, this high-scope intervention was contrasted with two controls: a behavioural programmed learning approach and a child-centred nursery programme. The last follow-up occurred when the child reached the age of 40 years. Up to adolescence, the high-scope group fared best and the programmed learning group fared worst (Schweinhart et al., 1985). At age 19, only 15% of children in the high-scope intervention group had been classified as ‘mentally retarded’ whereas 35% of the control group had been so labelled. While over half of the children in the control groups had been arrested, only 31% of the high-scope group had ever been detained (RR = 0.6, 95% CI: 0.38, 0.95). In the follow-up to age 27, lifetime arrest rates in the high-scope group were half those of the control groups. While minor offences and drug-related arrests accounted for much of this difference, recidivist crime was also reduced in the intervention group. Overall, 33% of the control groups but less than 7% of the high-scope group had been arrested more than five times (RR = 0.21, 95% CI: 0.07, 0.58). Similar improvements were observed in teenage pregnancy rates, high school graduation, home ownership and social benefits. Cost-benefit analysis revealed that the programme saved the US taxpayer $7 for each dollar spent. This return was accrued from savings in welfare, social services, legal and incarceration expenditures (Schweinhart et al., 1993; Schweinhart & Weikart, 1993).

The last follow-up reported progress to age 40, and 112 out of 123 of the adults who had participated in the study as children were interviewed (Schweinhart, 2007). Fifty-five per cent of the comparison but only 36% of the programme group had been arrested at one time (RR = 0.65, 95% CI: 0.43, 0.98). Forty-eight per cent of the no-programme group but only 32% of the programme group were arrested for one or more drug-related crimes (RR = 0.41, 95% CI: 0.19, 0.85). Significant group differences in arrests and crimes cited at arrests appeared consistently throughout the study participants’ lifetime, but significant group differences in conviction and sentences appeared only at ages 28 to 40. Compared with the no-programme group, the programme group had significantly fewer members sentenced to prison for felonies from ages 28 to 40 (RR = 0.28, 95% CI: 0.09, 0.79).

The Syracuse University Family Development Research Programme

In the Syracuse University Programme the focus was on infant development, home-care and parenting skills (Lally et al., 1988). Home and day care centre curricula were designed to foster active initiative and participation, as well as a sense of self-efficacy. The programme involved the use of sensorimotor and language games to enhance cognitive development in the infant. In weekly home visits by paraprofessionals, the role of the parent as primary teacher for the child was emphasised. One learning game was played at each visit. Support regarding employment, referral, and family relations was also provided to parents during home visits. Transportation of parents and siblings to the child care centre for activity meetings was offered. The programme included high-quality half-day care for infants aged 6 to 15 months and full day care for those aged 15 to 60 months.

The sample was of a medium size (n = 108). There was no randomisation, and families receiving the intervention were compared with a matched comparison group, but this group was recruited only when the project children were already 3 years of age. The mean age of the mothers was 18 years, and more than 85% were single. All had low incomes and the majority were African-Americans.

The intervention continued until the infant reached the age of 5 years. A quarter (24%) of the children in the programme did not complete all 5 years of the intervention, and only 50 to 60% completed the follow-up at age 15. At follow-up, girls who had participated in the programme were found to be doing better in school than control girls based on grades, attendance, and teacher-rated self-esteem and impulse control. Boys in the two groups did not differ on measures of school performance, but for both boys and girls self-regard was more positive in the intervention group than in the control group, based on self-report measures. The rate of delinquency in the intervention group, calculated from police data, was 6%, whereas in the control group it was 22% (RR = 0.27, 95% CI: 0.09, 0.81).

There were also differences in terms of the seriousness of offences and the cost of crimes committed between the two groups. Lifetime average probation costs were calculated for the two groups, and were estimated at $186 per child in the intervention group and $1,985 per child in the control group (Lally et al., 1988).

An acknowledgement of the effect of attrition on outcome data would suggest that these results should be treated with caution. It is reasonable to speculate that delinquency rates in families who could not be located for follow-up were quite high, since, of those families who were located, those with a child involved in juvenile delinquency proved the most difficult to find.

The Abecedarian Project

The Abecedarian Project was an RCT of early childhood education for healthy infants from impoverished families living in a small US community in North Carolina (Campbell & Ramey, 1994). One hundred and eleven infants from low income high-risk families were recruited to the project between 1972 and 1977 and randomised to receive the 5-year preschool intervention from infancy to age 5 years. Both groups received nutritional supplements and social services assistance, with the experimental group also receiving an educational intervention in a child care centre during the first 5 years. The focus of the programme was on cognitive and fine motor development, social and adaptive skills, language and other motor skills, and the child care centre also encouraged an unusually high level of parental involvement and offered social support.

The two groups were re-randomised at entry to kindergarten with half of each group receiving additional home-based as well as school-based support for the first 3 years (Ramey & Campbell, 1991). Children in the experimental group obtained higher achievement test scores than control children who had no intervention in either preschool or kindergarten to second grade. The bulk of this difference appeared to be because of the preschool intervention. There was a further follow-up at ages 12 to 15 (Campbell & Ramey, 1994), where 80% of those children who were randomly assigned and 90% of those who received the assigned intervention were tested. The superiority of the experimental group was maintained and in a significant number of cases it increased. Importantly, the impact of the intervention in kindergarten to second grade did not endure.

One hundred and five participants of the study were followed up in terms of their crime records to age 21 (average age 21.4, range 18.7–23.9). Juvenile delinquency statistics were not reported but extensive data concerning criminal history were obtained. There were no differences between the groups in terms of arrests, regardless of offences, charges or convictions. The relative risk of arrest after age 16 was 1.10 (95% CI: 0.56–2.19). From this study there is no evidence to suggest that early preschool academic input addresses functions that come to impact on serious antisocial behaviour.

The Chicago Longitudinal Study of the Child-Parent Center Programme

The Chicago Longitudinal Study investigated the effectiveness of the Child-Parent Center Programme for more then 1,500 children born in 1979 or 1980. Beginning in preschool, the programme provided comprehensive services that had been administered through the public educational system. The Longitudinal Study of Children at Risk (Reynolds, 1991) examined the effects of a preschool plus a follow-through early intervention programme on later school outcomes in a sample of 1,106 economically disadvantaged families. The intervention had multiple components including parenting education, volunteering in the classroom, low staff-to-child ratios, home visitation and health and nutrition services including referrals by programme nurses. The system of intervention provided a smooth transition to school, it was in place by the age of 2 years and continued until the early grades. The teachers in the programme were well trained and well compensated. The programme was 3 hours per day, 5 days per week during the school year and also included a 6-week summer programme. Parents were expected to participate in the programme for about half a day per week through a variety of supported activities providing many opportunities for positive learning experiences in the school and the home.

The programme group consisted of 989 children and the comparison group of 550 children was drawn from alternative full day kindergarten programmes. There was no random assignment but some children could be divided into groups that were involved in child and parent centres in preschool classes, kindergarten and primary grades. Child and parent centres offered multiple services, emphasising literacy development, reduced class sizes and considerable parent support and involvement. A comprehensive analysis of this naturalistic dataset (Reynolds, 1994) indicated that follow-on from kindergarten and preschool to primary grades was essential for the achievement test superiority to be maintained to grade 5. Primary grade intervention (1 to 3 years) resulted in significant improvement in both school achievement and school adjustment. Participation in the Child-Parent Center preschool intervention was associated with significantly higher rates of school completion by age 20, lower rates of juvenile arrests for both violent and non-violent juvenile offences and lower rate of use of school remedial services (Reynolds et al., 2001).

Extended intervention for 4 to 6 years was linked to significantly lower rates of remedial education and juvenile arrests for violent offences. One thousand, three hundred and sixty eight cases, 888 programme cases and 480 controls were available for the 22- to 24-year outcome assessments and more or less the entire sample was available to obtain crime and employment data. By age 24 years the rate of incarceration for the comparison group was 25.6% compared with 20.6% in the preschool programme group (RR = 0.80, 95% CI: 0.65, 0.98). School-age intervention did not significantly affect incarceration rate (RR = 1.10, 95% CI: 0.90, 1.34). Neither preschool (RR = 0.89, 95% CI: 0.77, 1.03) nor school-age (RR = 1.10, 95% CI: 0.90, 1.34) intervention significantly affected overall rates of arrests but preschool intervention reduced both felony arrests (RR = 0.78, 95% CI: 0.62, 0.98) and felony convictions (RR = 0.79, 95% CI: 0.62, 1.00). Violent crime convictions were also marginally reduced by preschool intervention (RR = 0.71, 95% CI: 0.46, 1.10). Participation in the extended programme was associated with a 32% reduction in rates of arrests (17.9% versus 13.9%; RR = 0.77, 95% CI: 0.59, 1.00) and convictions (RR = 0.68, 95% CI: 0.45, 1.04) for violence. Also quite pertinent in the present context, the findings indicated a dramatic reduction in out-of-home placements from 8.4 to 4.5% associated with the preschool intervention (RR = 0.53, 95% CI: 0.35, 0.81), probably indicative of a reduction in maltreatment.

Regression analyses indicated that the outcomes could be explained by a combination of increased cognitive skills, positive family support, positive post-programme school experiences and increased school commitment.

It should also be noted that there is considerable correlational evidence suggesting that early and prolonged low-quality day care represents a risk factor for negative developmental outcomes (Belsky, 2001; NICHD, 2003; Belsky et al., 2007). However, there is also evidence from the Canadian longitudinal study (Cote et al., 2007) that never having non-maternal care is a risk factor for physical aggression for children of mothers with low educational levels. In this sample (the largest parenting study yet conducted) early non-maternal care (before 9 months) was associated with a very slight increase in aggression in mothers of high education level relative to children who never had non-maternal care. But this was a small effect when compared with the increase of risk associated with the absence of non-maternal care in children of mothers of low education level. The GDG acknowledges that these are complex issues that are hard to argue from correlational data. However, the GDG wishes to assert that this does not indicate that good quality non-maternal care for young children is necessarily harmful in high risk samples (for example, low educational level) as this flies in the face of extant data. In terms of creating opportunities for children of mothers with limited resources, making adequate non-maternal care available is something that statutory providers should consider delivering.

School-based projects

This section reviews studies of school-age children with a mean age of 7 years. Typically these interventions consist of a combination of teacher training, parent training and skills-based interventions for children.

Seattle Social Development Project

This was a classroom-based project beginning in the first grade and ending at sixth grade (Hawkins et al., 1991, 1992, 1995). The aim of the programme was to strengthen the child’s bonds with their family and school, thus engendering a high level of adherence to the standards set by both. Bonds were conceptualised as positive emotional feelings towards others (attachment), an investment in a social unit (commitment) and the adoption of the values of that unit (belief). The interventions included teacher training, parent training and social and emotional skills development for the child. The interventions included proactive classroom management, cooperative learning strategies and interactive teaching. There was a component for parents encouraging engagement in the child’s education and workshops in social learning principles of child behaviour management. There was a problem-solving curriculum as well as drug refusal skills training. The experimental design involved comparing experimental and control schools with both random and non-random assignment in a complex design.

Beginning in 1981, the intervention was randomly assigned among grade 1 pupils (7 years of age) in classrooms in eight public schools in high crime areas. These children were followed prospectively until 1985 when the study was extended to include grade 5 pupils (11 years of age) in ten additional schools. There were ultimately four groups: a full intervention group (n = 156; 114 available for follow-up) with an average dose of 4.13 years of intervention exposure; a late intervention group (n = 267; 256 available for follow-up) with an average exposure of 1.65 years; a parent training only group (n = 141; most recent study did not analyse this group; Hawkins et al., 2005); and a control group (n = 220; 205 available for follow-up) who received no intervention.

First results were encouraging (Hawkins et al., 1991; O’Donnell et al., 1995). Boys in the high-risk sub-sample who participated in the programme had fewer antisocial peers and appeared to be less likely to be involved in delinquency. In girls the major benefit was in a reduced likelihood of substance use. At age 18 the intervention group reported less lifetime violence, less heavy alcohol use, less school misbehaviour and improved school achievement compared with controls (Hawkins et al., 1999). The findings indicated that the postulated mediating variables were indeed influenced by the programme, even if the impact on delinquency was relatively low. There was substantial impact on sexual behaviour by age 21 including unplanned pregnancies and condom use (Lonczak et al., 2002).

Criminal behaviour was assessed in interviews as well as official records (Hawkins et al., 2005). The full intervention group was less likely to be involved in a high variety of crime (3% versus 9%, RR = 0.33, 95% CI: 0.11, 0.93), to have sold illegal drugs (4% versus 13%, RR = 0.30, 95% CI: 0.12, 0.74), to have abused substances (74% versus 82%, RR = 0.90, 95% CI: 0.80, 1.01) and to have a court record at the age of 21 (42% versus 53%, RR = 0.79, 95% CI: 0.62, 0.99). Although the effects reaching statistical significance were limited and the tests were not corrected for the possibility of Type I error, the full intervention group reported less crime or substance use across all measures indicating a relatively robust effect from the early intervention.

5.3.6. Clinical evidence summary

Early childhood interventions in the first 5 years of a child’s life tend to show links to a broad range of positive outcomes. These include higher cognitive skills, school attainment, higher earning capacity, health and mental health benefits, reduced maltreatment and, significantly for this guideline, lower rates of delinquency and crime. Early childhood interventions are quite unique in this regard—there are no other interventions, as far as the GDG was aware, that have generated such a broad set of positive outcomes. That the impact of interventions should extend beyond educational performance to criminal behaviour is hardly surprising given the well-documented relationship between educational outcomes and adult mental health and social behaviour (for example, Chevalier & Feinstein, 2006). There are also indications from a number of studies that early interventions are cost effective in providing both savings and increased well-being that exceed the original investments in the programmes (Karoly et al., 2005; Reynolds & Temple, 2006; Rolnick & Grunwald, 2003). The economic returns of early childhood interventions exceed cost by an average ratio of 6 to 1.

In contrast, the evidence for preschool interventions shows more moderate effects on later offending, with some programmes found not to be effective. A similar picture emerges with school-based interventions where, again, the evidence for effectiveness is modest and weaker than earlier interventions. The economic evidence from the US suggests that, in the long-term, early interventions may result in significant net savings in terms of reduced welfare payments and crime costs and improved future earnings (see below).

5.3.7. Health economics evidence

Three studies that evaluated the cost effectiveness of preschool programmes for infants and toddlers were included in the systematic review of the economic evidence (Nores et al., 2005; Masse & Barnett, 2002; Reynolds et al., 2002). Details on the methods used for the systematic search of the economic literature are described in Chapter 3. Evidence tables for all economic studies included in the guideline economic literature review are provided in Appendix 14.

A long-term cost-benefit analysis of the High-Scope Perry Preschool Programme followed up participants as they reached the age of 40 (Nores et al., 2005). The initial costs of the programme were compared with any long-term benefits in terms of net changes (versus no intervention) in educational attainment, lifetime earnings, criminal activity and welfare payments. From various perspectives (the individual participant, general public and a combination of both), the programme resulted in significant long-term net benefits of between $49,000 and $230,000 per participant.

Another long-term cost-benefit analysis was conducted for the Abecedarian project, which followed up participants at age 21 (Masse & Barnett, 2002). Again, initial intervention costs were compared with long-term net benefits in terms of future earnings, maternal earnings, education costs, health improvements and welfare use. The project resulted in significant long-term net benefits of $100,000 per participant.

Finally, a long-term cost-benefit analysis of the Chicago Child-Parent centre programme was undertaken for participants at age 20 (Reynolds et al., 2002). Initial intervention costs were compared with long-term net benefits in terms of education costs, child care costs, welfare payments, abuse/neglect costs and justice/crime costs. Again, from various perspectives (individual participant, taxpayer, and both), the programme resulted in significant net benefits of between $12,000 and $34,000 per participant.

5.3.8. From evidence to recommendations

The GDG considered the evidence available on early interventions and noted that the majority of the interventions were developed in non-UK settings and this raised some questions about the generalisability of the findings. However, the GDG was impressed by the consistent impact of these programmes often on quite disadvantaged families and took the view that the most effective interventions were those targeting families at risk. Existing evidence from the US indicates that early interventions may result in great cost savings for the public sector and the children’s families. Early indications from pilot studies conducted in the UK suggest that it may be feasible to deliver these programmes in the UK. The GDG also recognised that the focus on effective identification of at-risk children and their families was central to the effectiveness of these programmes. It was felt that without this focus the impact of the programmes was likely to be significantly reduced and therefore not cost effective.

5.3.9. Recommendations

Identifying children at risk of developing conduct problems and potentially subsequent antisocial personality disorder

5.3.9.1.

Services should establish robust methods to identify children at risk of developing conduct problems, integrated when possible with the established local assessment system. These should focus on identifying vulnerable parents, where appropriate antenatally, including:

  • parents with other mental health problems, or with significant drug or alcohol problems.
  • mothers younger than 18 years, particularly those with a history of maltreatment in childhood
  • parents with a history of residential care
  • parents with significant previous or current contact with the criminal justice system.
5.3.9.2.

When identifying vulnerable parents, take care not to intensify any stigma associated with the intervention or increase the child’s problems by labelling them as antisocial or problematic.

Early interventions for at-risk children

5.3.9.3.

Early interventions aimed at reducing the risk of the development of conduct problems, and antisocial personality disorder at a later age, may be considered for children identified to be of high risk of developing conduct problems. These should be targeted at the parents of children with identified high-risk factors and include:

  • non-maternal care (such as well-staffed nursery care) for children younger than 1 year
  • interventions to improve poor parenting skills for the parents of children younger than 3 years.
5.3.9.4.

Early interventions should usually be provided by health and social care professionals over a period of 6–12 months, and should:

  • consist of well-structured, manualised programmes that are closely adhered to
  • target multiple risk factors (such as parenting, school behaviour, and parental health and employment).

5.4. INTERVENTIONS FOR CHILDREN WITH CONDUCT PROBLEMS

5.4.1. Introduction

Current practice

The treatment and management of conduct disorder and related problems in the UK has significantly expanded in recent years. The NICE technology appraisal on parent-training programmes (NICE, 2006b) has had a great impact and programmes based on models developed by Webster-Stratton (Webster-Stratton et al., 1988) among others, are now widely available in the UK.

In addition, a major pilot programme of multisystemic therapy was developed in 2008, which is currently being rolled out in ten sites across the UK. The outcomes of this pilot programme, which is subject to a formal evaluation, may have a considerable influence on the development of interventions for conduct disorder.

However, other approaches that may be of potential value, such as individually-focused interventions including cognitive problem-solving skills, are underdeveloped in the UK. Similarly other interventions, which are reviewed below, such as functional family therapy, multidimensional treatment foster care, or brief strategic family therapy, are not widely available in the UK. This is a particular concern because the primary focus of parent-training programmes is with younger children in the age range of 4 to 10 years. Evidence-based programmes for adolescents, where parent-training programmes may be less effective, are not well developed. Beyond mainstream provision in the NHS by CAMHS, there are also some specialist services (for example, youth offending teams) where these programmes may serve as effective preventive interventions for antisocial personality disorder.

In addition, a substantial proportion of young people with conduct problems will be involved in the criminal justice system where they are likely to receive interventions predominantly based on a cognitive and behavioural approach similar to that provided for adults (see Chapter 7 for further details).

5.4.2. Aim of review and definitions of interventions

The review looked at a wide range of family and individual interventions focused on children. These interventions were divided into four main categories: child-focused (skills-based training for children), parent-focused (behaviour management training for parents), family-focused (seeking to change problem interactions within the family), and multi-component (targeting the family and the wider social environment). The intention at the beginning of the guideline development process was to embed the recommendations in the technology appraisal on parent-training programmes for children with conduct disorder (NICE, 2006b) in this guideline.

Definitions of child-focused interventions

Cognitive problem-solving skills training

The emphasis of this intervention is on children’s thought processes that impact on how they behave in interpersonal situations. The intervention includes:

  1. teaching a step-by-step approach to solving interpersonal problems
  2. structured tasks such as games and stories to aid the development of skills
  3. combining a variety of approaches including modelling and practice, role playing, and reinforcement (Kazdin, 2010).
Anger control training

This includes a number of cognitive and behavioural techniques similar to cognitive problem-solving skills training. However this includes training of other skills such as relaxation and social skills and a specific focus on managing anger. This is usually offered to school-age children who are aggressive (Kazdin, 2010).

Social problem skills training

This is a specialist form of cognitive problem-solving skills training that also aims to modify and expand the child’s interpersonal appraisal processes through developing a more sophisticated understanding of beliefs and desires in others and to improve the child’s capacity to regulate their emotional responses (see Fonagy et al., 2002).

Definitions of parent-focused interventions

Parent-training programmes

The main goals of parent-training programmes are to teach the principles of child behaviour management, to increase parental competence and confidence in raising children and to improve the parent/carer-child relationship by using good communication and positive attention to aid the child’s development. These programmes are structured and follow a set curriculum over several weeks; they are mainly conducted in groups, but can be modified for individual treatments. Examples of well-developed programmes are Triple P (Sanders et al., 2000a) and Webster-Stratton (Webster–Stratton, 1988). The focus is primarily on the main caregiver of the child or young person, although some programmes add a child-directed component (NCCMH, 2009).

Definitions of family-focused interventions

Structural or systemic family therapy

This is a psychological intervention derived from a model of the interactional processes in families. The intention is to help participants understand the effects of their interactions on each other as factors in the development and/or maintenance of behavioural problems. Additionally, the aim is to change the nature of the interactions so that families may develop relationships that are more supportive and have less conflict (NCCMH, 2005a).

Functional family therapy

This is a family-based psychological intervention that is behavioural in focus. The main elements of the intervention include engagement and motivation of the family in treatment, problem-solving and behaviour change through parent training and communication training, and seeking to generalise change from specific behaviours to have an impact on interactions both within the family and with community agencies such as schools (see for example Gordon et al., 1995).

Brief strategic family therapy

This is a psychological intervention that is systemic in focus and is influenced by other approaches such as structural family therapy. The main elements of this intervention include engaging and supporting the family, identifying maladaptive family interactions and seeking to promote new more adaptive family interactions (see for example, Szapocznik et al., 1989).

Definitions of multi-component interventions

Multisystemic therapy

This is the use of strategies from family therapy and behaviour therapy to intervene directly in systems and processes related to antisocial behaviour (for example, parental discipline, family affective relations, peer associations, and school performances) for children or adolescents (Henggeler et al., 1992).

Multidimensional treatment foster care

Like multisystemic therapy, multidimensional treatment foster care also uses strategies from family therapy and behaviour therapy to intervene directly in systems and processes related to antisocial behaviour (for example, parental discipline, family affective relations, peer associations and school performances), but for children or adolescents in out of home placements. This includes family therapy with the child’s biological parents and group meetings and other support for the foster parents (Chamberlain & Reid, 1998).

5.4.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline can be found in Table 7.

Table 7. Databases searched and inclusion/exclusion criteria for clinical evidence.

Table 7

Databases searched and inclusion/exclusion criteria for clinical evidence.

5.4.4. Studies considered5

The review team conducted a new systematic search for RCTs that assessed the benefits and disadvantages of psychosocial interventions for children, and related health economic evidence (see Appendices 8 and 11 respectively).

A total of 97 trials relating to clinical evidence met the eligibility criteria set by the GDG, providing data on 6,665 participants. Of these, one trial was a report from the Joseph Rowntree Foundation (Scott et al., 2006), one trial was a report of the Washington Institute of Public Policy (Barnoski, 2004), and 95 were published in peer-reviewed journals between 1973 and 2008. In addition, 117 studies were excluded from the analysis. The most common reason for exclusion was lack of relevant outcomes (further information about both included and excluded studies can be found in Appendix 15).

The included trials involved the following comparisons:

  • Parent training compared with control (36 trials)
  • Parent training plus an additional intervention for children compared with parent training (five trials)
  • Parent training plus an additional intervention for parents compared with parent training (six trials)
  • Cognitive problem-solving skills training compared with control (five trials)
  • Social skills training compared with control (five trials)
  • Anger control training compared with control (ten trials)
  • Family interventions compared with control (11 trials)
  • Multisystemic therapy compared with control (ten trials)
  • Multidimensional treatment foster care compared with control (two trials)
  • Other multi-component interventions compared with control (four trials)
  • Cognitive and behavioural interventions compared with control (eight trials)
  • Cognitive and behavioural plus other interventions compared with control (two trials).

5.4.5. Clinical evidence for interventions targeted at children

Evidence from the important outcomes and overall quality of evidence are presented in Table 8 and Table 9. Full study characteristics and forest plots can be found in Appendices 15 and 16 respectively.

Table 8. Study information table for trials of interventions targeted at children and/or for the treatment of conduct problems.

Table 8

Study information table for trials of interventions targeted at children and/or for the treatment of conduct problems.

Table 9. GRADE evidence summary for interventions targeted at children and/or adolescents with conduct problems (only important outcomes reported).

Table 9

GRADE evidence summary for interventions targeted at children and/or adolescents with conduct problems (only important outcomes reported).

For all of these cognitive skills-based interventions there were a variety of outcomes reported. Wherever possible the primary outcome extracted in the meta-analysis was from a total behaviour scale. Measures specifically related to the content of the programme were judged to be less generalisable.

Cognitive problem-solving skills training

There were five trials on cognitive problem-solving skills training. At end of treatment there was a small-to-medium effect favouring cognitive problem-solving skills training (SMD −0.35; −0.59 to −0.10) and this effect was sustained and actually improved at 1-year follow-up (SMD −0.42; −0.84 to −0.00).

Anger control

There were ten trials on anger control. Trials that only included interventions for children appeared to be more effective (SMD −0.37; −0.58 to −0.16). Interventions that included a parent intervention in addition to anger control training did not appear to be effective (SMD −0.06; −0.25 to 0.13).

Social skills training

There were five trials on social skills training. Although the effects were of a similar magnitude as above (SMD −0.35; −0.73 to 0.03), there was significant heterogeneity and confidence intervals were compatible with benefit and no benefit.

5.4.6. Clinical evidence summary for interventions targeted at children

Interventions that met the criteria of the review were mainly based on cognitive behavioural approaches. Most studies reported small-to-moderate reductions in behaviour problems. However, there was uncertainty whether the promising results on social skills and anger control interventions would translate to everyday clinical practice.

5.4.7. Health economics evidence for interventions targeted at children

No evidence on the cost effectiveness of interventions targeted at children was identified by the systematic search of the literature. Details on the systematic search of the economic literature are provided in Chapter 3.

5.4.8. From evidence to recommendations

There is some evidence for cognitive problem-solving skills training, anger control and social skills training as interventions targeted at children. The evidence for cognitive problem-solving skills training was slightly stronger with good evidence of efficacy at follow-up in children with relatively severe behavioural problems.

However, the evidence for anger control and social skills training was more limited with greater variability in effectiveness and questions about the generalisability of some outcome measures. The GDG judged that their main value may be in treating children with residual problems after cognitive problem-solving skills training, or in treating children when it is not possible to engage the family in treatment. They may also be effective in providing an alternative where children have not fully benefited from family interventions.

5.4.9. Recommendations

5.4.9.1.

Cognitive problem-solving skills training should be considered for children aged 8 years and older with conduct problems if:

  • the child’s family is unwilling or unable to engage with a parent-training programme (see Sections 5.4.14)
  • additional factors, such as callous and unemotional traits in the child, may reduce the likelihood of the child benefiting from parent-training programmes alone.
5.4.9.2.

For children who have residual problems following cognitive problem-solving skills training, consider anger control or social problem-solving skills training, depending on the nature of the residual problems.

5.4.9.3.

Cognitive problem-solving skills training should be delivered individually over a period of 10–16 weeks. Training should focus typically on cognitive strategies to enable the child to:

  • generate a range of alternative solutions to interpersonal problems
  • analyse the intentions of others
  • understand the consequences of their actions
  • set targets for desirable behaviour.
5.4.9.4.

Anger control should usually take place in groups over a period of 10–16 weeks and focus typically on strategies to enable the child to:

  • build capacity to improve the perception and interpretation of social cues
  • manage anger through coping and self-talk
  • generate alternative ‘non-aggressive’ responses to interpersonal problems.
5.4.9.5.

Social problem-solving skills training should usually be conducted in groups over a period of 10–16 weeks. Training should focus typically on strategies to enable the child to:

  • modify and expand their interpersonal appraisal processes
  • develop a more sophisticated understanding of beliefs and desires in others
  • improve their capacity to regulate their emotional responses.

5.4.10. Clinical evidence for interventions targeted at parents

Evidence from the important outcomes and overall quality of evidence are presented in Table 10 and Table 11. Full study characteristics and forest plots can be found in Appendices 15 and 16 respectively.

Table 10. Study information table for trials of interventions targeted at parents for the treatment of conduct problems.

Table 10

Study information table for trials of interventions targeted at parents for the treatment of conduct problems.

Table 11. GRADE evidence summary for trials of interventions targeted at parents for the treatment of conduct problems (only important outcomes reported).

Table 11

GRADE evidence summary for trials of interventions targeted at parents for the treatment of conduct problems (only important outcomes reported).

There were a large number of studies of parent training, with 36 trials comparing parent training with control. Parent training in behavioural management is mostly offered in groups but some of the studies were of parents offered this kind of help individually. There was a small-to-medium effect favouring parent training (SMD −0.36; −0.51 to −0.22). Heterogeneity was high in the meta-analysis (I2 = 63.3%), which is explained to some extent by age and level of risk. A subgroup analysis of the data suggests that children up to the age of 11 years appear to be more likely to respond than young people of 12 years or older (children: SMD −0.58; −0.78 to −0.39; young people: SMD −0.32; −0.64 to 0.00) although there is still overlap in confidence intervals. In addition, a subgroup analysis of the data comparing studies of children with different levels of risk (participants rated on factors such as the severity of behaviour problems and socioeconomic status) showed a smaller effect for studies that included participants at greater risk (high risk: SMD = −0.20; −0.33 to −0.07; less risk: SMD = −0.44; −0.54 to −0.33). There appears to be good evidence that adding an intervention (usually cognitive problem-solving skills training) focused on the child adds to the efficacy of parent training compared with parent training alone (SMD= −0.30; −0.51 to −0.09). There was less clear evidence for an additional benefit from adjunctive intervention focused on psychological problems in the parents (for example, CBT for depression in the mother; SMD = −0.12; −0.35, 0.11).

It is also important to note that moderators of the effectiveness of parent training have been identified (Dadds et al., 1987a; Dadds et al., 1987b). More severe and more chronic antisocial behaviour and comorbidity with other diagnoses predict reduced responsiveness to treatment, including dropouts and negative outcomes. However inattention, impulsivity and hyperactivity problems increase the size of the response. Extremely high levels of parental negativity towards the child also reduce responsiveness to the programme. Low socioeconomic status is associated with more limited outcomes, in particular if it occurs in combination with social insularity in the family. Maternal psychopathology, in particular depression and life events, has also been found to reduce the effectiveness of parent training. There are also findings indicating that single parent status, only one parent attending, marital disharmony and maternal insecurity of attachment may undermine progress but many of these associations are not found consistently across studies. Families with children in the pre-adolescent age group are more likely to drop out of treatment. The best current evidence-based programmes include modules for targeting these moderating factors; however, their use is more often supported by correlational rather then RCT data, although RCT data does provide some evidence for limited interventions such as telephone reminders (Watt et al., 2007). While the present review does not permit the GDG to make specific recommendations, in general it is desirable to include additional treatment modules in parent-training programmes that are likely to prevent the premature termination of treatment.

A number of individual parent training programmes have been evaluated and found to be effective (Nixon et al., 2003). For younger children (typically between 3 and 6 years) one of the most prominent is parent-child interaction therapy (PCIT) (for example, Schuhmann et al., 1998). For older children (typically between 5 and 12 years) the parent management training programmes developed in Oregon have also been shown to be effective (for example, Patterson et al., 1982). However, it is difficult to make comparisons of effectiveness of group versus individual administration as it is rarely a subject of tests. Overall effect sizes for individual parent-training programmes are also confounded by lack of commensurability in terms of the clinical characteristics of the sample.

5.4.11. Clinical evidence summary for interventions targeted at parents

There is a very large evidence base confirming the effectiveness of parent training in a range of populations in a number of countries. There was significant heterogeneity in the meta-analysis; subgroup analyses suggest that differences in the ages of the children and in level of risk may explain, to some extent, some of the inconsistency. Given the limited evidence for individual parent-training programmes and the lack of comparators with the stronger evidence base for group-based training programmes the GDG decided to focus the recommendations on group-based interventions.

There are also a growing number of studies assessing adjuncts to parent training. The results of the meta-analysis suggest that a cognitive problem-solving intervention targeted at the child may be effective. Adjuncts targeted specifically at the parent’s mental health problems were slightly less effective.

5.4.12. Health economic evidence for interventions targeted at parents

The only study identified by the systematic search of economic evidence that met the inclusion criteria for review was an economic analysis of parent training for children with conduct disorders (Dretzke et al., 2005) undertaken for the NICE technology appraisal (NICE, 2006b). According to the technology appraisal, parent training was found to be cost effective and was recommended for implementation in health and social care settings. Details on the methods used for the systematic search of the economic literature are described in Chapter 3. Evidence tables for all economic studies included in the guideline economic literature review are provided in Appendix 14.

Economic analysis in the NICE technology appraisal on parent-training/education programmes for children with conduct disorders (NICE, 2006b)

The NICE technology appraisal on parent-training/education programmes in the management of children with conduct disorders (NICE, 2006b) incorporated economic evidence from two de novo economic models assessing the cost effectiveness of parent-training/education programmes relative to no active intervention for this population.

The initial economic analysis (Dretzke et al., 2005) assessed the cost effectiveness of three parent-training/education programmes differing in the mode of delivery and the setting: a group community-based programme, a group clinic-based programme, and an individually delivered, home-based programme. Costs included intervention costs only; no potential cost savings to the NHS following reduction of antisocial behaviour in treated children were considered. Total costs of these three types of interventions were estimated based on a ‘bottom-up’ approach, using expert opinion alongside information from the literature in order to determine the healthcare resources required for providing such programmes. Meta-analysis of clinical data had demonstrated that there was no difference in clinical effectiveness between group-based and individually delivered programmes. According to the findings of the economic analysis, the group clinic-based programme was the dominant option among the three parent-training/education programmes, as it provided the same health benefits (same clinical effectiveness) at the lowest cost (total intervention cost per family was £629 for the group clinic-based programme, £899 for the group community-based programme, and £3,839 for the individual home-based programme).

Further analyses were undertaken to estimate the cost effectiveness of parent-training/education programmes assuming various levels of response to treatment and various levels of improvement in children’s HRQoL. According to this analysis, and after assuming an 80% uptake of such programmes, the group clinic-based programme resulted in a cost per responder of £10,060 and £1,006 at a 5% and 50% success (response) rate, respectively; and a cost per quality-adjusted life year (QALY) of £12,575 and £3,144 at a 5% and 20% improvement in HRQoL, respectively.

In contrast, provision of an individual home-based programme was demonstrated to incur a rather high cost of £19,196 per QALY gained, assuming it provided a 20% improvement in HRQoL. At lower levels of improvement in HRQoL, this figure became well above the £20,000 per QALY threshold of cost effectiveness set by NICE (The Guidelines Manual [NICE, 2006a]), rising at approximately £77,000 per QALY when a 5% improvement in HRQoL was assumed. This means that for families where individual parent training is the preferred option, for example in cases where parents are difficult to engage with, or the complexities of the family’s needs cannot be met by group-based programmes, the improvement in HRQoL of the child needs to reach at least 20% for the intervention to meet the cost-effectiveness criteria set by NICE.

The initial economic analysis was based on hypothetical rates of response and percentages of improvement in HRQoL following provision of parent-training/education programmes, as well as on a number of assumptions. Therefore, the results should be interpreted with caution, as acknowledged by its authors. On the other hand, it should be noted that estimated figures were conservative, as they did not include any potential cost savings resulting from reduction in antisocial behaviour in treated children and associated costs of its management. Despite its limitations, the analysis demonstrated that group-based parent-training/education programmes for children with conduct disorders were, as expected, substantially more cost effective than individually delivered programmes, because the two modes of delivery did not differ in terms of clinical effectiveness, while the intervention costs of group-based programmes were spread over a large number of treated families.

The additional economic analysis undertaken to support the NICE technology appraisal evaluated the cost effectiveness of the three parent-training/education programmes described above, plus an individually delivered clinic-based programme, over a time horizon of 1 year. Costs included intervention costs as the initial analysis, but they also incorporated cost savings to the NHS, education and social services following provision of parent-training/education programmes to children with conduct disorders. The analysis modelled three different health states, that is, normal behaviour, conduct problems and conduct disorders. It was found that the mean net cost of a parent-training/education programme in improving a child’s behaviour from conduct disorder to an improved state (either conduct problems or normal behaviour) was £90 for a group community-based programme, £1,380 for an individually delivered clinic-based programme, and £2,400 for an individually delivered home-based programme; the group clinic-based programme proved to be cost saving overall. These results further support the argument that group-delivered parent-training/education programmes for children with conduct disorders are most likely to be cost effective, especially when long-term benefits, such as the sustained effects of therapy and a reduction in the rates of future offending behaviour, as well as future cost savings to healthcare, education and social services, are considered.

5.4.13. From evidence to recommendations

The clinical and economic evidence clearly supports the implementation of parent-training programmes for children with conduct problems. The results suggest that the likely effect of parent-training programmes will be felt more for younger children. This suggests that there may be a need to consider augmenting programmes for older children who have not benefited with cognitive problem-solving skills interventions. These additional interventions should be focused on the child as there is little evidence that focusing interventions specifically on the parent is effective. For those children who have not benefited and/or whose parents have refused treatment, a second option would be to give consideration to specific individual cognitive problem-solving skills interventions.

5.4.14. Recommendations

5.4.14.1.

Group-based parent-training/education programmes are recommended in the management of children with conduct disorders.6

5.4.14.2.

Individual-based parent-training/education programmes are recommended in the management of children with conduct disorders only in situations where there are particular difficulties in engaging with the parents or a family’s needs are too complex to be met by group-based parent-training/education programmes.7

5.4.14.3.

For parents of young people aged between 12 and 17 years with conduct problems, consider parent-training programmes.

5.4.14.4.

Additional interventions targeted specifically at the parents of children with conduct problems (such as interventions for parental, marital or interpersonal problems) should not be provided routinely alongside parent-training programmes, as they are unlikely to have an impact on the child’s conduct problems.

5.4.14.5.

Programme providers should also ensure that support is available to enable the participation of parents who might otherwise find it difficult to access these programmes.8

5.4.14.6.

Support to enable the participation of parents who might otherwise find it difficult to access these programmes might include:

  • individual parent-training programmes
  • regular reminders about meetings (for example, telephone calls)
  • effective treatment of comorbid disorders (in particular, attention deficit hyperactivity disorder in line with ‘Attention deficit hyperactivity disorder’ NICE clinical guideline 72 [NICE, 2008).
5.4.14.7.

It is recommended that all parent-training/education programmes, whether group- or individual-based, should:9

  • be structured and have a curriculum informed by principles of social-learning theory
  • include relationship-enhancing strategies
  • offer a sufficient number of sessions, with an optimum of 8–12, to maximise the possible benefits for participants
  • enable parents to identify their own parenting objectives
  • incorporate role-play during sessions, as well as homework to be undertaken between sessions, to achieve generalisation of newly rehearsed behaviours to the home situation
  • be delivered by appropriately trained and skilled facilitators who are supervised, have access to necessary ongoing professional development, and are able to engage in a productive therapeutic alliance with parents
  • adhere to the programme developer’s manual and employ all of the necessary materials to ensure consistent implementation of the programme.
5.4.14.8.

Programmes should include problem solving (both for the parent and in helping to train their child to solve problems) and the promotion of positive behaviour (for example, through support, use of praise and reward).

5.4.14.9.

Programmes should demonstrate proven effectiveness. This should be based on evidence from randomised controlled trials or other suitable rigorous evaluation methods undertaken independently.10

5.4.15. Clinical evidence for interventions targeted at families

Evidence from the important outcomes and overall quality of evidence are presented in Table 12 and Table 13. Full study characteristics and forest plots can be found in Appendices 15 and 16 respectively.

Table 12. Study information table for trials of family interventions.

Table 12

Study information table for trials of family interventions.

Table 13. GRADE evidence summary for family interventions (only important outcomes reported).

Table 13

GRADE evidence summary for family interventions (only important outcomes reported).

Eleven trials assessed the effectiveness of family interventions. It appears that family interventions are more effective than control for reducing both behavioural problems (SMD = −0.75; −1.19 to −0.30) and offending (RR = 0.67; 0.42 to 1.07).

The heterogeneity observed in the risk of re-offending was explained by problems with therapist competence in BARNOSKI2004. A subgroup analysis found a large difference when including only competent (RR = 0.57; 0.42 to 0.78) or non-competent therapists (RR = 0.70; 0.36 to 1.38). Data from MCPHERSON1983 was not included in the analysis as data were not extractable.

The heterogeneity observed in the behaviour scales outcome appeared to be due to NICKEL2005 and NICKEL2006A. A subgroup analysis showed that substantially larger effects were reported (SMD = −1.48; −1.97 to −0.99) in these studies on reduction in drug use, compared with the other studies’ effects on total behaviour (SMD = −0.42; −0.68 to −0.15).

5.4.16. Clinical evidence summary for interventions targeted at families

There appears to be good evidence for the effectiveness of family interventions in a range of adolescents with conduct problems including offenders. In addition, quasi-experimental implementation studies confirm the effectiveness of these interventions in naturalistic settings.

5.4.17. Health economic evidence for interventions targeted at families

Systematic literature review

Two studies from the US were identified that considered the cost effectiveness of interventions targeted at families (Barnoski, 2004; Crane et al., 2005). The study by Barnoski (2004) evaluated functional family therapy for moderate to high-risk juvenile offenders (13 to 17 years). Costs of the intervention were compared with differences in recidivism rates and resulting criminal justice costs versus no intervention. Overall, functional family therapy resulted in significant net savings due to lower rates of recidivism compared with no intervention. The study by Crane and colleagues (2005) was a simple retrospective cost analysis of in-home or in-office family therapy versus no treatment for youths with conduct disorder. Over 30 months both interventions resulted in significant net savings (p < 0.0001) in terms of reduced future healthcare spending. No studies were identified that considered the cost effectiveness of family interventions in the UK. Details on the methods used for the systematic search of the economic literature are described in Chapter 3. Evidence tables for all economic studies included in the guideline economic literature review are provided in Appendix 14.

Economic modelling

Objective

The guideline systematic review and meta-analysis of clinical evidence demonstrated that provision of functional family therapy to families of adolescents with a history of offending behaviour significantly reduces the rates of future reconviction. Offending behaviour and subsequent reconviction lead to substantial costs, not only to the criminal justice system but also to victims and society in general. A cost analysis was undertaken to assess whether the costs to the NHS of providing functional family therapy to families of adolescents at risk for offending behaviour are offset by future cost savings resulting from reduction in offending behaviour (and subsequent reconviction rates) in young offenders.

Methods

Intervention examined Functional family therapy is a short-term intervention: on average 8 to 12 sessions are needed for mild problems and up to 30 hours of direct service (for example, clinical sessions, telephone calls and meetings involving community resources) for more difficult cases. For most participants, sessions are spread over a 3-month period. Functional family therapy programmes have been successfully delivered in home-based, clinic-based and school-based settings. In Washington where functional family therapy was evaluated, trained therapists had caseloads of 10 to 12 families (Barnoski, 2004). The effectiveness of therapy in reducing recidivism may be directly related to the competence of the therapist (Barnoski, 2004). Implementation of functional family therapy, therefore, focuses particularly on developing therapist competence rather than simply teaching skills.

Costs considered in the analysis A simple economic model was developed to estimate the net total costs (or cost savings) associated with provision of functional family therapy to families of adolescents at risk for offending behaviour. Adolescents with conduct disorder and/or offending behaviour have been found to incur substantial costs to the health, educational, social and criminal justice services. Scott and colleagues (2001a) estimated the public costs incurred by children with conduct disorder from 10 years of age through adulthood (by age 28) in the UK. The authors reported a total cost of £70,000 per person diagnosed with conduct disorder in childhood, compared with £7,000 for a person without any conduct problems. Criminal justice system services bore most of this cost (64%), whereas the cost to educational services reached 18% of the total cost. Foster and residential care costs amounted to 11% of the total cost, and social benefits to another 4%. Finally, the cost to the health-care services was only 3% of the total cost incurred by individuals with conduct disorder from childhood through adulthood.

NICE recommends that economic analyses of healthcare interventions adopt a NHS and personal social services (PSS) perspective (NICE, 2006a). However, in the case of adolescents with offending behaviour, the majority of incurred costs falls on the criminal justice system, education services, social and other public services. Only a small minority of costs is covered by the NHS and PSS perspective. For this reason, the economic analysis adopted a broader perspective than that of the NHS and PPS, including any costs to public services for which appropriate information was available.

The study by Scott and colleagues (2001a) illustrated the variety and magnitude of costs associated with conduct disorder and, more broadly, offending behaviour; nevertheless, little evidence exists about the potential reduction (or increase) in specific cost components resulting from provision of functional family therapy to families of young offenders. Clinical evidence has demonstrated that functional family therapy significantly reduces reconviction rates, and subsequently costs relating to crime. It is likely that provision of functional family therapy, by reducing offending behaviour, also reduces other types of cost, such as health and social care costs, as well as costs to the educational services. However, no appropriate relevant data that could inform this economic analysis were identified in the literature. For this reason, the analysis has considered only intervention costs (that is, costs of providing functional family therapy) and costs related to crime/offending behaviour of adolescents. All other categories of costs to the public sector, such as health and social care costs and costs to educational services, were conservatively assumed to be the same for adolescents receiving functional family therapy and for those not receiving the intervention, and were subsequently omitted from the analysis. This is acknowledged as a limitation of the economic analysis. However, costs relating to crime constitute the most substantial part of the costs incurred by young offenders; therefore, the economic analysis has probably considered the majority of costs associated with providing functional family therapy to families of young offenders.

Model input parameters

Clinical efficacy of functional family therapy and baseline re-offending rate in juvenile offenders

Clinical data on re-arrest rates (reflecting re-offending behaviour) associated with provision of functional family therapy were derived from three studies (ALEXANDER1973; BARNOSKI2004; GORDON1995). Meta-analysis of these data undertaken for the guideline showed that providing functional family therapy to families of adolescents with offending behaviour reduced the rate of re-arrest compared with usual care/no treatment. Of the three clinical studies included in the meta-analysis, BARNOSKI2004 reported results for participants treated by both competent and non-competent therapists, and separate results for a subgroup of participants treated by competent therapists only. Therefore, two separate guideline meta-analyses were performed: one including all efficacy data reported in the relevant literature, and one sub-analysis including efficacy data on families treated by competent therapists only. The results of the meta-analysis indicated that provision of functional family therapy by both competent and non-competent therapists reduces the re-arrest rates in adolescents with offending behaviour, but these results were non-significant at the 0.05 level (mean RR of re-arrest of functional family therapy versus control: 0.67; 95% CI: 0.42 to 1.07). In contrast, when only data on competent therapists were considered, functional family therapy was shown to significantly reduce the re-arrest rates in juvenile offenders (mean RR of re-arrest of functional family therapy versus control: 0.59; 95% CI: 0.43 to 0.79). Details on the clinical studies considered in the economic analysis are available in Appendix 15. The forest plots of the respective meta-analyses are provided in Appendix 16.

The baseline re-offending rate for adolescents with previous offending behaviour was taken from a national report containing 12-month data on re-offending for adolescents aged 10 to 17 years released from custody (either from prison, secure training centres or secure children’s homes) or commencing a non-custodial court disposal, or given an out-of-court disposal (either a reprimand or final warning) in England and Wales in 2006 (Ministry of Justice, 2008b). According to this document, the re-offending rate in this population was 38.7% over 12 months. This rate was defined by the number of offenders in the cohort re-offending at least once during the 12-month follow-up period, where the offence resulted in a conviction at court or an out-of-court disposal. The 12-month rate of adolescent re-offending following provision of functional family therapy in the economic analysis was calculated by multiplying the estimated RR of re-arrest of functional family therapy versus control by the baseline re-offending rate.

Intervention costs (costs of providing functional family therapy)

In order to calculate total intervention costs, relevant resource use was estimated and combined with respective unit costs. Resource use estimates were based on information provided in the clinical studies included in the guideline systematic review. According to these estimates, functional family therapy consisted of 12 sessions over a 90-day period lasting 1.5 hours each, delivered to groups of ten families of adolescents with offending behaviour.

The unit cost of therapists providing functional family therapy was estimated to be similar to that of clinical psychologists (Band 7). The national unit cost of clinical psychologists has been estimated at £67 per hour of client contact in 2006/07 prices (Curtis, 2007). This estimate was based on the mid-point of Agenda for Change salaries Band 7 of the April 2006 pay scale according to the National Profile for Clinical Psychologists, Counsellors and Psychotherapists (NHS, 2006). It includes salary, salary on-costs, overheads and capital overheads but does not take into account qualification costs as the latter are not available for clinical psychologists.

Based on the above resource use estimates and the unit cost of clinical psychologists, the cost of providing the Reasoning and Rehabilitation programme was estimated at £121 per adolescent with offending behaviour in 2006/7 prices.

Costs of adolescent offending behaviour

In order to estimate the annual cost resulting from repeat of offending behaviour by adolescent offenders, three types of data are needed:

  • proportion of different types of offences committed by adolescent re-offenders
  • costs associated with each type of offence
  • number of offences per adolescent re-offender per year.

Data on the proportion of different types of offences committed by adolescent re-offenders in England and Wales were derived from a national report published by the Ministry of Justice (2008b). The same document reported that the number of offences per juvenile re-offender were 3.181 over 12 months.

Regarding costs associated with each type of offence committed by adolescent offenders, these were taken from a variety of sources:

  • Costs of offences against individuals and households, such as violence including homicide, sexual offences, theft including stealing vehicles and stealing from vehicles, robbery, criminal and malicious damage and domestic burglary were taken from Home Office data (Dubourg & Hamed, 2005). This report estimated a wide range of costs associated with crime, including costs incurred in anticipation of crime, such as security expenditure and insurance administration, costs directly resulting from crime, such as stolen or damaged property, lost output, emotional and physical impact on victims, health services to victims, other victim services, as well as costs to the criminal justice system, including police services.
  • Costs of non-domestic burglary and costs of fraud and forgery were also based on Home Office data (Brand & Price, 2000). Reported costs included the same cost elements as described above.
  • The cost of motoring offences (excluding thefts from or of vehicles and drink driving) was assumed to correspond to the cost of accidents leading to damage (but not injury) as reported by the Department of Transport (2007). This cost included police costs, costs relating to insurance and administration services, and costs resulting from property damage.
  • The cost of drugs import/export/production and supply was derived from Home Office estimates of the average cost of arrest for drugs possession and supply (Godfrey et al., 2002). The same report suggested that the cost of arrest for possession of drugs (but not supply) was equal to the general cost of arrest. The latter cost was reported in the same document and was assumed to reflect the cost associated with arrest for drug possession and small-scale supply.
  • The costs of public order or riot, soliciting or prostitution, handling, and absconding or bail offences were (rather conservatively) assumed to correspond to the cost of general arrest, as reported in Godfrey and colleagues (2002). The costs of drink driving and other, not specified, offences were based on assumptions.

Costs reported in the literature were uplifted to 2007 prices using the Retail Prices Index (Office for National Statistics, 2008). The cost per offence committed by adolescent re-offenders was estimated as the mean cost of all offences weighted by the proportion of offences committed on average by an adolescent re-offender. Table 14 shows the percentage of offences committed by adolescent re-offenders, the cost of each type of offence as estimated in the literature, and the weighted average cost per offence committed by adolescent re-offenders.

Table 14. Percentage and costs of offences committed by adolescent re-offenders.

Table 14

Percentage and costs of offences committed by adolescent re-offenders.

The average cost per offence committed by adolescent re-offenders was estimated at £3,639. Since this population has been found to commit 3.181 offences over 12 months (Ministry of Justice, 2008b), the 12-month cost associated with offending behaviour is £11,576 per juvenile re-offender.

Time horizon of the analysis Of the three studies included in the relevant guideline meta-analysis of functional family therapy clinical data, two studies had a time horizon of 18 months (ALEXANDER1973 and BARNOSKI2004, with a total study population of 765 adolescents) and one study had a time horizon of 5 years (GORDON1995, with a study population of 54 adolescents). For the base-case analysis, a 2-year time horizon was chosen. However, time horizons up to 5 years were tested in sensitivity analysis, to explore the magnitude of potential savings resulting from provision of functional family therapy.

Discounting Costs incurred beyond 12 months were discounted at an annual rate of 3.5%, as recommended by NICE (NICE, 2006a).

Table 15 provides all input parameters utilised in the base-case analysis of the economic model of functional family therapy for families of adolescents at risk for offending behaviour.

Table 15. Input parameters utilised in the economic model assessing the net costs (or savings) resulting from provision of functional family therapy to families of adolescents at risk for offending behaviour.

Table 15

Input parameters utilised in the economic model assessing the net costs (or savings) resulting from provision of functional family therapy to families of adolescents at risk for offending behaviour.

Sensitivity analysis One- and two-way sensitivity analyses were undertaken to explore the robustness of the results under the uncertainty characterising some model input parameters. The following scenarios were tested in sensitivity analysis:

  • Use of the 95% CIs of the RR of re-arrest of functional family therapy versus control.
  • Exclusion of data on adolescents seen by non-competent therapists (that is, using a mean RR of re-arrest of functional family therapy versus control 0.59 with 95% CIs 0.43 to 0.79).
  • Increase of intensity of functional family therapy; 18 sessions of 2 hours each were assumed.
  • Reduction in the baseline re-offending rate for adolescents with previous offending behaviour; an annual rate of 20% was tested.
  • Extension of the time horizon of the analysis beyond 2 years; limited evidence suggested that the effect of functional family therapy in reducing the re-arrest rates in adolescents with a history of offending behaviour remained over 5 years (GORDON1995). Therefore, potential net savings accrued over 3, 4 and 5 years following provision of functional family therapy were estimated. This scenario was combined with all other scenarios described above.
  • Potential net savings accrued over 3, 4 and 5 years were also estimated assuming that the effect of the intervention was reduced over time; in this scenario the RR of functional family therapy versus control was multiplied by a factor of 1.15 for every year after 2 years from initiation of the intervention, to capture this assumed decline in the clinical effect over time, until functional family therapy had no beneficial effect over control (that is, until RR became 1).
  • Reduction in the annual cost of offending behaviour per re-offender: a conservative figure of £1,000 per offence and two offences per re-offender per year were simultaneously assumed, resulting in a total annual cost of offending behaviour of £2,000 (instead of £11,576, which was the respective estimate used in base-case analysis) – therefore, any savings expected from reduction in re-offending following provision of functional family therapy would be much lower under this hypothesis.
  • Simultaneous use of an RR of functional family therapy versus control of 0.79 (which was the upper 95% CI in the sub-analysis that included competent therapists) and a conservative annual cost of offending behaviour per adolescent re-offender of £2,000, as estimated in the previous scenario.
  • Combination of alternative time horizons between 2 and 5 years with all other scenarios described above.
Results

Base-case analysis The reduction in re-offending rates achieved by provision of functional family therapy to families of adolescents at risk for re-offending yielded cost-savings equalling £2,908 per adolescent with offending behaviour over the 2 years of the analysis. Providing functional family therapy incurs a cost of £121 per adolescent, but this cost was offset by the substantial savings from reduction in offending behaviour. Overall, functional family therapy resulted in a net saving of £2,787 per adolescent with offending behaviour over 2 years. Full results of the base-case analysis are reported in Table 16.

Table 16. Results of economic analysis assessing the net costs (or savings) resulting from provision of functional family therapy to families of adolescents at risk for offending behaviour.

Table 16

Results of economic analysis assessing the net costs (or savings) resulting from provision of functional family therapy to families of adolescents at risk for offending behaviour.

Sensitivity analysis Results of the cost analysis were robust under the different scenarios examined in sensitivity analysis. Under all scenarios, provision of functional family therapy resulted in overall net savings even under a time horizon of 2 years, with the only exception being the use of the upper 95% CI of RR of re-arrest of functional family therapy versus control taken from meta-analysis of data including non-competent therapists (this upper 95% CI had a value of 1.07 as results were non-significant at the 0.05 level). Under the most optimistic scenario of a lasting effect of 5 years, and using the lower 95% CI of the RR of re-arrest of functional family therapy versus control, functional family therapy resulted in net savings of £12,021 per adolescent with offending behaviour. Full results of sensitivity analysis are presented in Table 17.

Table 17. Results of sensitivity cost analysis of provision of functional family therapy to families of adolescents at risk for offending behaviour.

Table 17

Results of sensitivity cost analysis of provision of functional family therapy to families of adolescents at risk for offending behaviour.

Discussion – limitations of the analysis

The results of the economic analysis suggest that provision of functional family therapy to families of adolescents with a history of offending behaviour is likely to be cost saving from a broad economic perspective in the UK. Intervention costs were shown to be offset by savings from a reduction in the rates of re-arrest. The results were robust under the majority of scenarios examined in sensitivity analysis. The only exception was the use of results of meta-analysis that included data on both competent and non-competent therapists; in this case, results were statistically insignificant and use of the upper 95% CI of the RR of re-arrest of functional family therapy versus control did not produce cost savings. However, given that in the UK both clinical psychologist and family therapist training has moved towards competence-based models of training (Roth & Pilling, 2008), it is unlikely that those deemed not sufficiently competent would be involved in implementation of functional family therapy. Furthermore, under the accreditation and audit processes used in the National Offender Management Service (NOMS), poor therapists or programme tutors would not be allowed to deliver such programmes (National Offender Management Service, 2008).

Adolescents with offending behaviour incur significant costs to health, social, educational and criminal justice services, as well as to their families (Scott et al., 2001a). The economic analysis considered only intervention costs and costs relating to offending behaviour owing to lack of evidence for a difference in other types of costs between functional family therapy and no treatment. Cost data on offending behaviour were derived from several published sources reporting UK data and included, in most cases, a wide range of costs, such as costs incurred in anticipation of offending behaviour (for example, security expenditure), costs directly resulting from offending (such as costs of stolen or damaged property and emotional and physical impact on victims), costs of offering health and other services to victims, as well as costs to the criminal justice system. Although it is acknowledged that omission of educational, social and other healthcare costs constitutes a limitation of the analysis, existing evidence indicates that costs of offending behaviour are probably the most significant costs incurred by adolescents with offending behaviour. Besides, it is likely that functional family therapy, by improving adolescent behaviour, also reduces other costs incurred by adolescent offenders, such as costs falling on special educational services. If this is true, then the economic analysis has only underestimated the net savings gained from functional family therapy. Furthermore, some of the cost data on offending behaviour that were utilised in the economic analysis comprised criminal justice system costs only. The healthcare costs and emotional distress of victims, the financial and economic burden to the families of both victims and offenders, and the feelings of fear and insecurity in anticipation of crime were not considered in most documents reporting cost data on offending behaviour. Had these factors been considered, the cost savings from reduction in offending behaviour might be greater than figures reported in the analysis.

Sensitivity analysis showed that even if a more intensive functional family therapy programme were implemented, the intervention would still be cost saving. More intensive functional family therapy programmes than this described in the base-case economic analysis may be needed in more complex cases, and this would result in higher intervention costs. On the other hand, it has been shown that adolescents with a more severe history of offending behaviour are characterised by higher rates of re-offending and higher numbers of offences per year (Ministry of Justice, 2008b). Therefore, a reduction in offending behaviour in this group of adolescents would lead to greater cost savings, compared with adolescents with mild offending behaviour. Consequently, complex cases, which might require more intensive treatment, are likely to produce greater cost savings, offsetting the higher intervention costs.

The time horizon of the analysis was 2 years, according to available evidence. However, limited evidence indicates that the beneficial effect of functional family therapy may last for longer time periods (over 5 years following provision of the therapy). Consequently, net savings from functional family therapy estimated in base-case analysis are rather conservative; greater cost savings may be realised if the effect of functional family therapy lasts longer than 2 years.

Conclusion

Overall, and despite conservative estimates utilised in the economic model, provision of functional family therapy to families of adolescents at risk for offending behaviour is likely to be cost-saving. Given that functional family therapy is also an effective intervention that improves adolescent offending behaviour, functional family therapy is likely a cost-effective intervention.

5.4.18. From evidence to recommendations

The evidence suggests that a range of family interventions, including systemic and strategic family therapy, may be effective for children with conduct problems and conduct disorder. Interventions such as functional family therapy may be particularly effective for older adolescents for whom the evidence for the efficacy of parent-training programmes is weak, and are also likely to be cost effective. The evidence suggests that functional family therapy, and potentially brief strategic family therapy, should become viable alternatives to parent training for older adolescents. This requires individual clinicians to consider the relative benefits of the two, including child and adult preferences.

5.4.19. Recommendations

5.4.19.1.

If the parents are unable to or choose not to engage with parent-training programmes, or the young person’s conduct problems are so severe that they will be less likely to benefit from parent-training programmes, consider:

  • brief strategic family therapy for those with predominantly drug-related problems
  • functional family therapy for those with predominantly a history of offending.
5.4.19.2.

Brief strategic family therapy should consist of at least fortnightly meetings over a period of 3 months and focus on:

  • engaging and supporting the family
  • engaging and using the support of the wider social and educational system
  • identifying maladaptive family interactions (including areas of power distribution and conflict resolution)
  • promoting new and more adaptive family interactions (including open and effective communication).
5.4.19.3.

Functional family therapy should be conducted over a period of 3 months by health or social care professionals and focus on improving the interactions within the family, including:

  • engaging and motivating the family in treatment (enhancing perception that change is possible, positive reframing and establishing a positive alliance)
  • problem-solving and behaviour change through parent-training and communication training
  • promoting generalisation of change in specific behaviours to broader contexts, both within the family and the community (such as schools).

5.4.20. Clinical evidence for multi-component interventions

Evidence from the important outcomes and overall quality of evidence are presented in Table 18 and Table 19. Full study characteristics and forest plots can be found in Appendices 15 and 16 respectively.

Table 18. Study information table for trials of multi-component interventions for adolescents at risk of offending.

Table 18

Study information table for trials of multi-component interventions for adolescents at risk of offending.

Table 19. Evidence summary of multi-component interventions (only important outcomes reported).

Table 19

Evidence summary of multi-component interventions (only important outcomes reported).

Some researchers have combined two or more psychological and/or psychosocial interventions, provided concurrently or consecutively, in an attempt to increase the effectiveness of the intervention. For example, a course of family intervention may be combined with a module of social skills training. The combinations are various and thus these multi-modal interventions do not form a homogenous group of interventions that can be analysed together.

Ten trials on multisystemic therapy that met the inclusion criteria for the review were included. There was significant heterogeneity for most outcomes; however, there was consistent evidence of a medium effect on reduction in offending outcomes including number of arrests (SMD −0.44; −0.82 to −0.06) and being arrested (RR 0.65; 0.42 to 1.00).

The main source of heterogeneity was LESCHIED2002, which found no difference between multisystemic therapy and treatment as usual on all primary outcomes. A possible explanation is that the majority of trials of multisystemic therapy were conducted in the US by the founders Henggeler and colleagues, whereas LESCHIED2002 is a Canadian trial undertaken independently from the founders of multisystemic therapy. However, a study by OGDEN2004 on a Norwegian sample, which was also conducted independently, found positive effects for multisystemic therapy for slightly different outcomes.

Henggeler and colleagues (2006a) argue the lack of effectiveness reported in LESCHIED2002 is probably because of problems with treatment fidelity and the challenges of setting up a new multisystemic therapy service. There were differences in effectiveness between sites and the site with the lowest fidelity was also found to have the least favourable outcomes.

There were only two trials that met the inclusion criteria of the review on multidimensional treatment foster care. There was a medium effect favouring multidimensional treatment foster care (SMD = −0.55; −0.36 to −0.82).

There were three trials assessing other multi-component interventions. It was not possible to meta-analyse these studies as there were major differences in the interventions and their effectiveness, as well as very high heterogeneity (I2 = 83.9%). There was considerable variability in outcomes, with BARRETT2000 finding a large effect favouring the intervention (SMD = 1.41; −2.19, −0.63). In contrast, CAVELL2000 (SMD = 0.26; −0.25, 0.77) and FRASER2004 (SMD = −0.17; −0.60, 0.25) found no benefit for the intervention.

5.4.21. Clinical evidence summary for multi-component interventions

There is a relatively large evidence base concerning the effectiveness of multisystemic therapy. While there was significant heterogeneity, there is good evidence of efficacy for reducing offending for up to 14 years’ follow-up.

There were promising findings on the efficacy of multidimensional treatment foster care, with consistent moderate reductions in offending associated with this intervention compared with treatment as usual.

There is inconclusive evidence for the effectiveness of other multi-component interventions.

5.4.22. Health economic evidence for multi-component interventions

One study from the US was identified that considered the cost effectiveness of multi-component interventions targeted at children (Foster et al., 2006). The study evaluated the cost effectiveness of the Fast-Track intervention, a 10-year, multi-component intervention designed to reduce violence among at-risk children with conduct problems. The extra costs of the intervention programme versus no treatment were evaluated against three clinical outcomes: cases of conduct disorder averted; criminal offences avoided; and acts of interpersonal violence averted. Overall, for all three outcomes, the intervention was not cost effective at conventional willingness-to-pay thresholds. Subgroup analyses showed that the intervention was more cost effective for high-risk than low-risk children.

5.4.23. From evidence to recommendations

The evidence suggests that for children at risk of going into care, multidimensional treatment foster care is an effective intervention. For conduct disordered adolescents for whom parent training is not appropriate and who are at significant risk of offending, multisystemic therapy is an effective intervention. It is important for both of these interventions that high fidelity to the model is preserved. The limited economic evidence from a US setting suggests that multi-component interventions may only be cost effective in high-risk children.

5.4.24. Recommendations

5.4.24.1.

For young people aged between 12 and 17 years with severe conduct problems and a history of offending and who are at risk of being placed in care or excluded from the family, consider multisystemic therapy.

5.4.24.2.

Multisystemic therapy should be provided over a period of 3–6 months by a dedicated professional with a low caseload, and should:

  • focus specifically on problem-solving approaches with the family
  • involve and use the resources of peer groups, schools and the wider community.
5.4.24.3.

For young people aged between 12 and 17 years with conduct problems at risk of being placed in long-term out-of-home care, consider multidimensional treatment foster care.

5.4.24.4.

Multidimensional treatment foster care should be provided over a period of 6 months by a team of health and social care professionals able to provide case management, individual therapy and family therapy. This intervention should include:

  • training foster care families in behaviour management and providing a supportive family environment
  • the opportunity for the young person to earn privileges (such as time on the computer and extra telephone time with friends) when engaging in positive living and social skills (for example, making their bed and being polite) and good behaviour at school
  • individual problem-solving skills training for the young person
  • family therapy for the birth parents to provide a supportive environment for the young person to return to after treatment.

5.5. COORDINATION OF CARE

The primary objective of early interventions for conduct problems in childhood is to prevent the development of antisocial personality disorder in adults. However, as will be clear from the evidence above, these interventions may not always be successful, and even where a child does not go on to develop ASPD significant mental health problems may continue into adult life. It is therefore very important that healthcare professionals working with children effectively coordinate the care they provide and also ensure an appropriate transition to adult services for those children who require continuing care.

5.5.1. Recommendations

General principles when working with children and their families

5.5.1.1.

Child and adolescent mental health service (CAMHS) professionals working with young people should:

  • balance the developing autonomy and capacity of the young person with the responsibilities of parents and carers
  • be familiar with the legal framework that applies to young people, including the Mental Capacity Act, the Children Acts and the Mental Health Act.

Transition between child and adolescent services to adult services

5.5.1.2.

Health and social care services should consider referring vulnerable young people with a history of conduct disorder or contact with youth offending schemes, or those who have been receiving interventions for conduct and related disorders, to appropriate adult services for continuing assessment and/or treatment.

5.5.2. Research recommendations

Through identifying research limitations from the evidence-based reviews, the guideline development group has formulated the following research recommendations.

5.5.2.1. Effectiveness of multisystemic therapy versus functional family therapy

Is multisystemic therapy or functional family therapy more clinically and cost effective in the treatment of adolescents with conduct disorders? A large-scale RCT comparing the clinical and cost effectiveness of multisystemic therapy and functional family therapy for adolescents with conduct disorders should be conducted. It should examine the medium-term outcomes (for example, offending behaviour, mental state, educational and vocational outcomes and family functioning) over a period of at least 18 months. The study should also be designed to explore the moderators and mediators of treatment effect, which could help to determine the factors associated with benefits or harms of either multisystemic therapy or functional family therapy.

Why this is important

Multisystemic therapy and functional family therapy are two interventions with a relatively strong evidence base in the treatment of adolescents with conduct disorders, but there have been no studies directly comparing their clinical and cost effectiveness. Their use in health and social care services in the UK is increasing. Both interventions target the same population, but although they share some common elements (that is, work with the family), multisystemic therapy is focused on both the family and the wider resources of the school, community and criminal justice systems, and through intensive individual case work seeks to change the pattern of antisocial behaviour. In contrast, functional family therapy focuses more on the immediate family environment and uses the resources of the family to change the pattern of antisocial behaviour. The study should be designed to facilitate the identification of subgroups within the conduct disorder population who may benefit from either multi-systemic therapy or functional family therapy.

5.5.2.2. Interventions for infants at high risk of developing conduct disorders

Do specially designed parent-training programmes focused on sensitivity enhancement (a set of techniques designed to improve secure attachment behaviour between parents and children) reduce the risk of behavioural disorders, including conduct problems and delinquency, in infants at high risk of developing these problems? An RCT comparing parent-training programmes focused on sensitivity enhancement with usual care should be undertaken. It should examine the long-term outcomes over a period of at least 5 years, but with consideration given to the possibility of a further 10-year follow-up. The study should also be designed to explore the moderators and mediators of treatment effect that could help determine the factors associated with benefits or harms of the intervention.

Why this is important

There is limited evidence from non-UK studies that interventions focused on developing better parent–child attachment can have benefits for infants at risk of developing conduct disorder. Determining the criteria and then identifying children at high risk (usually via parental risk factors) is difficult and challenging. Even when these factors are agreed, engaging parents in treatment can be difficult. It is important that a range of effective interventions is developed to increase the treatment choice and opportunities for high-risk groups. Several interventions, such as Nurse–Family Practitioners, are being developed and trialled in the UK. It is important for this group of children to have an alternative, effective intervention.

Footnotes

4

Here and elsewhere in the guideline, each study considered for review is referred to by a study ID in capital letters (primary author and date of study publication, except where a study is in press or only submitted for publication, then a date is not used). The references for studies in this section can be found in Appendix 17.

5

Here and elsewhere in the guideline, each study considered for review is referred to by a study ID in capital letters (primary author and date of study publication, except where a study is in press or only submitted for publication, then a date is not used). The references for the studies can be found in Appendix 15.

6

This recommendation is from NICE technology appraisal 102 (NICE, 2006b).

7

Ibid.

8

Ibid.

9

Ibid.

10

Ibid.

Copyright © 2010, The British Psychological Society & The Royal College of Psychiatrists.

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Enquiries in this regard should be directed to the British Psychological Society.

Bookshelf ID: NBK55328
Contents

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (3.8M)

In this Page

Other titles in this collection

Related NICE guidance and evidence

Recent Activity

ClearTurn OffTurn On

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...