Introduction

Politicians have begun to recognise the importance of meeting the needs of offenders, as the long-term costs to society become increasingly apparent.

Research on offenders has found both high rates of mental health needs and learning disabilities [1–4]. While there is growing literature on the needs of offenders, much of this research has been cross-sectional in nature with few longitudinal studies. Longitudinal studies are important in helping us to understand developmental pathways and how needs may change over time. More importantly, understanding developmental trajectories may also provide opportunities for early intervention. For example, several longitudinal studies show an association between childhood psychopathology and adult mental disorders including childhood conduct disorder and the later development of antisocial personality disorder [5] and substance dependence [6]. Additionally, a population-based study in the US found that childhood psychiatric disorders were a common risk factor for later criminal behaviour and that severe and violent offences were predicted by membership of co-morbid diagnostic groups [7].

Moffitt and colleagues [8] from the Dunedin Study described two main developmental pathways for offending behaviour. The life-course persistent offenders whose antisocial behaviour begins in early childhood have many of the intrinsic risk factors (low IQ, language deficit and executive dysfunction) identified in previous research. Meanwhile, the adolescent onset group, have fewer intrinsic risk factors and their offending behaviour often ceases in adulthood. The Cambridge and Dunedin studies also found strong continuity of offenders’ needs from childhood and into adulthood including problems with mental health, relationships and employment [8, 9].

We conducted a prospective cohort study of the mental health and psychosocial needs of a group of adolescent male juvenile offenders admitted to secure care. With reference to Moffitt’s taxonomy, the study sample would be more consistent with offenders on the early onset and life-course persistent developmental pathway. The boys were initially assessed on admission and on average 3 months [10] and 2 years later [11] to ascertain whether their needs had changed.

The initial study [10] found that juvenile offenders in secure care had needs in many different areas (mental health, education and social). During the course of the admission some needs were met (education, substance misuse, self-care and diet) through increased supervision whilst in custody, while mental health needs persisted or worsened with new onset of depression, anxiety and post-traumatic stress disorder (PTSD) in some young people. The study also found that one in four offenders had a learning disability (IQ < 70) with many experiencing reading and spelling difficulties.

The offenders were subsequently reassessed 2 years later when about two-thirds of young people had been discharged back into the community [11]. Needs had fallen in some areas, particularly education and occupational needs. The mean number of needs at follow-up requiring an intervention was 3.4 (SD 3.0) compared with 8.2 (SD 2.5) prior to admission.

However, mental health needs persisted or worsened on discharge with 31 % of young people having substance misuse needs compared with 21 % on admission. Anxiety disorder including PTSD was also more common at follow-up, particularly in young people who remained in custody. Further analysis has shown that this may be partially attributed to the association with violent crime.

Many young people reoffended during the 2-year follow-up period [11]. Persistent or escalating offending behaviour at follow-up was associated with high frequency of previous offending (>5 offences), low intelligence (IQ < 80), criminal activity at an early age (<13 years) and a high score on the Psychopathy Screening Device [12], although the latter two variables more significantly predicted persistent offending.

This paper reports on a follow-up study of these boys, now adults, 6 years after their index admission. The main aim was to describe their mental health, current needs and offending behaviour. A secondary aim of the study was to assess which variables at baseline and follow-up were associated with persistent offending in early adulthood. Variables were selected based on findings from previous research [11], although we were also interested to understand the role of antisocial personality disorder and psychopathy. Finally, we aimed to explore whether childhood mental disorders predicted adult psychiatric problems in male offenders, as longitudinal studies of this population are sparse.

Method

Procedure and participants

Fifty-four offenders were interviewed at the 6-year follow-up point (56 %). The original study recruited 97 consecutive young male offenders aged 12 years to 17 years admitted to four local authority secure units within North West England [7]. Juvenile offenders admitted to LASU’s are on average younger than those admitted to other types of secure estates within the youth justice system.

At the initial interview, young people recruited to the study were asked to provide consent to be contacted for further interviews at a later date. Consent was obtained from their parent or guardian if they were aged 15 years or less. Those involved in the 6-year follow-up interviews were also re-consented for ongoing participation in the study. Ethical approval was obtained from the North West Multi-centre Research and Ethics Committee.

Offenders were contacted by letter and telephone through details provided in the initial study. Where initial contact with the young person was not successful, the National Offender Management Services (NOMS) database was used to establish whether the young person was still in contact with the criminal justice system. Contact was subsequently attempted through the local probation service. A nationally available database (births, deaths and marriages register) was used to check whether any of the young people had died during the follow-up period.

From the 97 young people who had participated in the initial study [8], 43 offenders were not interviewed at the 6-year follow-up. Of these, 33 of them were unable to be located (77 %), 9 did not respond to correspondence (21 %) and 1 offender declined to be interviewed (2 %). However, we were able to obtain some offending outcome data on a further 17 offenders through the NOMS and local probation services (n = 71; 73 %). Offenders that we were unable to collect outcome data for (n = 26) had moved and were no longer in contact with criminal justice services. One young offender had been murdered.

Measures

At the 6-year follow-up interview (time 4), offenders were assessed using a variety of adult measures and information was also obtained from self-reports about their offending behaviour and current occupational functioning.

DSM-IV Axis 1 psychiatric disorders were derived from the structured clinical interview for DSM disorders (SCID I), whilst the structured clinical interview II (SCID II) evaluated DSM-IV Axis II personality disorders. Both tools are administered in a semi-structured interview with the participant and have demonstrated good validity and reliability [13, 14].

The short form of the Camberwell Assessment of Need-Forensic Version (CANFOR-S) was used to assess the needs of the offenders [15]. The CANFOR is a needs assessment tool for adults with mental health problems who are also in contact with forensic services. The CANFOR-S has been piloted amongst mentally disordered offenders in secure care and shown to be a valid tool [16]. It is administered as a semi-structured interview with the client and identifies whether there is a need in 25 different domains. These are rated as no need (no problems in this area and not receiving any help), met need (has difficulties but receiving effective help), unmet need (has problems but not receiving any help or help received is not effective) or not applicable.

Adult psychopathy was measured using the screening version of the Hare Psychopathy Checklist-Revised [17]. The PCL-R is a well-recognised measure of psychopathy and has two parts. Part 1 assesses interpersonal and affective symptoms whilst part 2 assesses social deviance symptoms. Each of the 12 items are scored 0–2 (0, no; 1, maybe and 2, yes), producing a total score ranging from 0 to 24.

Statistical analyses

The data were analysed using Stata 11 statistical software [18]. We report on the prevalence rates for psychiatric disorders, needs and offending behaviour. Linear models were used to investigate the relationships between the chosen variables and offending behaviour. Where the outcome was binary, logistic regression was used and the results presented as odds ratios.

Results

Participant characteristics

The mean age of the 54 male offenders interviewed was 21.5 years (SD 1.4 years, range 19–25 years) and the majority were white (n = 45; 83 %). At the 6-year follow-up, many young offenders were either in custodial care (young offenders institution or prison) (n = 33; 61 %), or on a community order (n = 11; 20 %). Community orders are sentences given by courts (ranging in length from 12 h to 3 years) to be served within the community supervised by local probation services (community adult criminal justice teams). They may include a range of different requirements from supervision, attendance at particular programmes or substance misuse rehabilitation. A few offenders were living in the community, but were no longer in contact with probation services (n = 9; 16 %). Two-thirds of the sample were in neither employment nor training (n = 35; 65 %). Half of those living in the community were employed (n = 10; 48 %) or in training (n = 1; 5 %), compared with 1 in 7 offenders in custody who were accessing some form of training (n = 8; 15 %) whilst in secure care. The majority of offenders were also single (n = 47; 87 %). Whilst living in the community, the majority of young people were either living in their own place (n = 25; 46 %) or with family and friends (n = 24; 44 %).

The characteristics of offenders followed up at 6 years were compared with those lost to follow-up (Table 1). Excluding age, no significant differences were found in demographic or personal characteristics (ethnicity, IQ or history of being in care) between those who did and did not participate in the time 4 follow-up interviews. There were also no significant differences in rates of mental disorders at baseline between these groups. However, offenders who were followed up were significantly (p < 0.05) more likely to have committed an offence before their 13th birthday and have a history of frequent early offending. These latter two variables have been shown to be associated with more persistent offending [11] and subsequently the offenders followed up at 6 years in this study may represent a more persistent subgroup.

Table 1 Characteristics of offenders followed up at 6 years and those lost to follow-up
Full size table

Needs of offenders

The needs of the sample were assessed at follow-up using the Camberwell Assessment of Needs-Forensic Short Version [12] and are presented in Table 2. The greatest areas of need related to problems with daytime activities (engaging in work, training or structured leisure activities) (n = 25; 46 %), safety to others (violent or threatening behaviour) (n = 26; 48 %) and problems related to alcohol (n = 25; 46 %) and drug misuse (n = 39; 72 %). While offenders reported that some needs were met following intervention, they reported ongoing difficulties in a number of different areas. In addition to the above areas, a significant area of unmet need included problems with intimate relationships (difficulty finding a partner or maintaining a close relationship).

Table 2 Camberwell assessment of needs-forensic short version
Full size table

Mental disorder

Using the SCID-I, 15 % (n = 8) of offenders were found to meet the criteria for a psychiatric disorder, although this figure increased to 89 % (n = 48) when substance misuse was included. Six per cent (n = 3) of male offenders at follow-up, met the DSM-IV criteria for depression, whilst 9 % (n = 5) had an anxiety disorder (post-traumatic stress disorder n = 2; 4 % and panic disorder n = 3; 6 %). The majority of offenders had a substance abuse disorder (n = 47; 87 %), including alcohol abuse (n = 28; 52 %), drug abuse (n = 20; 37 %) or drug dependence (n = 22; 41 %). Fewer offenders were found to have alcohol dependence (n = 8; 15 %) and none of the offenders interviewed met the criteria for a psychotic disorder, eating disorder or somatoform disorder (Table 3). The absence of psychosis may be a sampling artefact as those with acute psychotic symptoms are more likely to accessing treatment in a hospital setting.

Table 3 Rates of mental disorder, personality disorder, psychopathy and risk
Full size table

The rate of mental illness found in this study was similar to rates found in the general UK population of adults (16 %), but lower than rates found in the prison population (45 %) following two national studies by the Office of National Statistics [19, 20]. Possible explanations for the relatively lower rate of mental illness found in this study, include the minimising of mental health symptoms and the lack of emotional literacy to communicate emotional needs in this sample of early-onset male offenders.

Personality disorder and psychopathy

Offenders were also assessed for a diagnosis of personality disorder using the SCID-II (Table 3). Seven offenders (13 %) failed to meet the criteria for a diagnosis of personality disorder, whilst almost a third (n = 17; 32 %) were found to have two or more personality disorders. The majority of the sample met the criteria for antisocial personality disorder (n = 46; 85 %), while 1 in 5 were diagnosed with paranoid personality disorder (n = 11; 20 %). Diagnosis of borderline personality disorder (n = 6; 11 %) and passive-aggressive personality disorder (n = 6; 11 %) were also common.

Table 4 displays the participant ratings on each of the individual items of the Hare Psychopathy Checklist (PCL-SV). Although offenders in this study displayed high levels of offending behaviour, psychopathy scores as measured by the PCL-SV were not high. The mean total psychopathy score in early adulthood was 10.9 (SD 3.91; range 3–21) out of a maximum score of 24 which equates to the 20th percentile (percentile rank based on forensic sample within the manual). The mean psychopathy score for part 2 (social deviance symptoms) (mean 8.28; SD 2.63) of the PCL-SV was noted to be significantly higher than for part 1 (interpersonal and affective symptoms) (mean 2.6; SD 2.50).

Table 4 Psychopathy
Full size table

Early onset and persistent antisocial behaviour are both key factors in the development of antisocial personality disorder (ASPD). It is therefore unsurprising that rates of ASPD were found to be high within this study sample. The finding that psychopathy rates were low is confirmed by previous research findings [21]. While many adults with a high psychopathy score (as measured by the PCL-R) will have a diagnosis of ASPD, only one in ten adults with ASPD are likely to meet the criteria for psychopathy [21]. This is further supported by the finding that scores on Part 2 (social deviance symptoms) of the PCL-SV were significantly higher than for Part 1 (interpersonal and affective symptoms).

Offending behaviour

Since the initial admission to secure care, the majority of the sample with outcome data had reoffended (n = 64/71; 90 %), with 27 convictions on average (SD 34.8; range 0–150). However, if it is assumed those lost to follow-up (n = 26) did not reoffend, as they were no longer in contact with criminal justice services, the rate of reoffending reduces to 66 % (n = 64/97).

The most common serious offences committed since time 1, were assault (n = 9; 17 %), burglary (n = 9; 17 %) and street robbery (n = 13; 24 %). More serious offences included murder or attempted murder (n = 2; 4 %), sexual offences (n = 1; 2 %), arson (n = 3; 6 %) and using firearms with intent (n = 3; 6 %). Since the index admission, the mean number of custodial sentences received by an offender was 4.0 (SD 4.5; range 0–23) and the average length of time served in custody was 28 months (SD 20.8; range 0–72 months). Almost half of offenders (n = 25; 46 %) reported receiving an offence-related intervention since the index admission, 6 years previously.

As in the previous study [11], offending behaviour of the subjects was classified into those who had persisted, escalated, declined or stopped. Offending was classified as escalated if there was an increase in frequency, severity or diversity of the type of offences committed (n = 43; 61 %). Persistent offenders were those who continued to offend at the same rate or type of offence (n = 18; 25 %). Meanwhile offenders were classified as declined in participants who continued to offend but at a lower frequency or severity (n = 2; 4 %). Finally, a few offenders were found to have stopped their offending (n = 7; 10 %). Offenders whose offending behaviour was classified as either persistent or escalating at follow-up had on average spent more time in custody (30.2 months compared with 14.3 months) and received more offence-related interventions than those who had reduced or stopped offending (48 % compared with 33 %).

Correlates of mental health outcomes at follow-up

We assessed for any correlation between a diagnosis of a mental health disorder at baseline with mental health outcomes at follow-up based on previous research findings (Table 5). The presence of a mood or anxiety disorder in adolescence (time 1) was not significantly associated with a similar diagnosis at 6-year follow-up. However, offenders who had evidence of drug abuse at baseline (time 1) were significantly more likely to have a diagnosis of substance misuse 6 years later (OR 7.5; 95 % CI 1.2–42.6).

Table 5 Correlates of mental health outcomes from adolescence to adulthood
Full size table

Correlates of persistent offending behaviour and risk

We analysed a number of variables known to be associated with persistent offending from previous studies (early age of onset of offending, high frequency of offending, low IQ and psychopathy) (Table 6). Unfortunately, we were unable to fully explore this relationship as few offenders in the 6-year follow-up sample had stopped offending to allow adequate comparison. However, we found that both a diagnosis of antisocial personality disorder (OR 8.6; 95 % CI 1.4–54.6) and living with family and friends (OR 1.2; 95 % CI 0.2–8.0) in early adulthood were significantly associated with persistent offending behaviour.

Table 6 Correlation between time 1 and time 4 characteristics and offending patterns at follow-up
Full size table

Discussion

The transition from adolescence into early adulthood is a period of development identified by significant change. There have been few longitudinal studies following offenders from adolescence into adulthood with assessments at multiple time points. There are a number of important findings to highlight from this study of early-onset male juvenile offenders.

First, this study found that the majority of offenders followed up at 6 years continued to offend, with only a few stopping or reducing their antisocial behaviour. This may reflect both a failure of interventions to reduce recidivism whilst offenders have been in contact with criminal justice services, as well as the persistent nature of offending behaviour in this particular sample. International studies suggest that recidivism rates for adolescent offenders are similarly high [22, 23]. These findings lead us to question whether many of the interventions and services currently provided to reduce recidivism in more persistent offenders are ineffective or at best prevent further escalation. This is of concern as financial resources are limited and cost-effectiveness is increasingly important in service provision. The experience of the research team in discussions with offenders and staff working in the criminal justice system in the UK was that interventions were not targeted but often universally provided, brief and fragmented due to poor continuity of care. For example, one offender admitted attending an anger management course five times.

Secondly, the study confirmed that adult offenders who have a history of persistent offending in adolescence have needs in multiple domains. Almost two-thirds of offenders were in neither employment nor training. The majority were also single and many had difficulties with intimate relationships. Living with family and friends was found to be significantly associated with persistent offending in adulthood. Marriage, employment and moving away have all been shown to foster resistance to offending behaviour [9]. A 7-year follow-up study in the US has highlighted the high prevalence of negative outcomes for adolescent offenders, from unemployment, substance misuse and further custodial care to a higher rate of mortality, mostly from violent causes [24].

Nine out of ten persistent adult offenders had a substance misuse disorder and substance misuse in adolescence was strongly correlated with later substance misuse in adulthood. A longitudinal study of male offenders has shown that substance misuse and depressive symptoms in adolescence were significantly predictive of offending trajectories, even after controlling for other factors [25]. Chronic high level offenders were found to be particularly at risk of developing depression and drug use in adulthood [25]. Subsequently, the relationship between antisocial behaviour and substance misuse is likely to be complex with reciprocal effects occurring over time. This finding emphasises the importance of early interventions for both adolescent offenders and non-offenders with substance misuse problems.

The study findings are also supported by a large international study. A longitudinal study of almost 2,000 individuals followed from birth to 33 years in Baltimore (USA), found that life-course persistent offenders were more likely to experience adverse physical and mental health outcomes compared with adolescent-limited peers. The authors suggest that the link between offending and health-related outcomes in adulthood are both direct and indirect, through childhood disadvantage, failed education and employment and antisocial lifestyles including high-risk behaviours [26].

Finally, with regard to continuity of mental health disorders from childhood into adulthood, this study found that childhood psychopathology did not accurately predict adult disorders, except for substance misuse. This is contrary to the findings of other longitudinal studies [27] and likely to reflect insufficient power within the study. Numerous studies have shown that conduct disorder in childhood and adolescence is strongly associated with the later development of antisocial personality disorder [28]. However, it has been suggested that it is the number and type (covert compared with overt symptoms) of conduct disorder symptoms and not just the diagnosis that increases the risk of developing antisocial personality disorder [29].

While a diagnosis of antisocial personality disorder in this sample of persistent offenders was high, rates of psychopathy in adulthood was relatively low which is consistent with findings from earlier studies [17]. This may highlight the limitations of a diagnosis of antisocial personality disorder based primarily on behavioural difficulties. The study also found that antisocial personality disorder but not psychopathy predicted offending in adulthood. This is supported by the findings of a meta-analysis that impulsive and antisocial traits of psychopathy (factor 2) are more strongly associated with antisocial conduct than affective and interpersonal traits (factor 1) [30].

Service implications

Together these findings have implications for service delivery. A recent review of services for adult offenders in the UK has highlighted a number of problems within the criminal justice system [31]. These include difficulties identifying offenders with mental health needs and learning disabilities, problems accessing appropriate treatment interventions and poor continuity of care.

The sample of offenders selected for this longitudinal study focused on boys in secure care. This is likely to represent a subgroup of early-onset offenders with multiple vulnerabilities, including local authority care. As adults, many of these offenders were persistent offenders with high rates of antisocial personality disorder and substance misuse. Many offenders also had significant difficulties in personal relationships and were living with family and friends. The latter was found to be a significant predictor of persistent offending.

While there is good evidence for the effectiveness of early interventions such as parenting programmes, antisocial behaviour can persist into adulthood despite these interventions for a subgroup of young people [28]. A multi-factorial conceptualisation of juvenile offending is now well established [28]. With this in mind, effective interventions for juvenile offending to date include; multi-systemic therapy (MST) [32], multi-dimensional foster care (MTFC) [33] and functional family therapy (FFT) [34]. Economic analysis suggests that functional family therapy is a cost-effective intervention that can improve adolescent antisocial behaviour and should be considered for older adolescents where alternatives such as parent training programmes and cognitive problem-solving skills training are ineffective [35].

Lipsey’s meta-analysis identified three primary factors which characterised programme effectiveness for juvenile offenders; a therapeutic intervention philosophy, a focus on high-risk offenders and the quality of the intervention [36]. All three factors are key features of both MST and MTFC.

Multi-systemic therapy is a multi-modal intervention where interventions are targeted at not only the young person, but also their family, school and peers which is essential following the findings of this longitudinal study. Evaluation studies of MST in a number of different countries have been promising, including a relatively low drop-out rate and benefits maintained at follow-up [32, 37]. MST has also shown to be effective for young people with substance misuse disorders [32], which is an important consideration in view of the high prevalence rate in persistent offenders. However, criticisms of MST include the requirement for a high level of therapeutic expertise, the importance of treatment fidelity as well as the cost of implementation. Therefore, while MST is unsuitable as a universal intervention for all offenders, it may be cost-effective for those at risk of more serious long-term antisocial behaviour [35].

A more cost-effective alternative with a focus on young people with substance misuse problems is multi-dimensional family therapy (MDFT). MDFT is a family-based outpatient treatment for adolescents with drug abuse problems and behavioural difficulties. It attempts to address the needs of the young person through therapy sessions with the family and wider social systems using a systemic model. MDFT has been found to be effective in studies in the US [38] and more recently in Europe, following a large multi-site randomised control trial for young people with cannabis misuse [39].

Despite increasing evidence supporting the effectiveness of specific treatment programmes, there are concerns that few high-risk offenders have access to these treatments. Problems with poorly co-ordinated commissioning and interagency working are likely to be key factors in successful implementation. However, governments and policy makers will need to address this issue urgently as existing resources are currently being used ineffectively. This study found that offending behaviour is persistent in high-risk offenders in spite of receiving custodial care and offence-related interventions.

When interpreting the results of this study there are some limitations that should be considered. This was a long-term study of male juvenile offenders and subsequently the findings cannot be generalised to female offenders. Indeed, studies would suggest that developmental pathways for offending behaviour differ between the two. Female offenders have a typically later onset of antisocial behaviour with fewer neuropsychological risk factors, similar to the adolescent onset group [40].

Through the course of the 6-year study a number of offenders were also lost to follow-up leaving a subgroup of offenders who were potentially more persistent in their offending behaviour. Therefore, the findings may not reflect the needs of less severe offenders. Additionally, the low sample size by the 6-year follow-up point may have contributed to insufficient power in establishing continuity between childhood and adult mental health problems.

Finally, self-reports of offending behaviour were used for those interviewed at the 6-year follow-up, while we were reliant on the NOMS database for those lost to follow-up. Both these methods have limitations in the accuracy of the information collected and may also have contributed to an inability to find predictive factors for persistent offending.

Further large-scale randomised controlled trials are needed to evaluate the clinical and cost-effectiveness of interventions such as MST and MDFT. Such studies should incorporate an adequate follow-up period and differentiate mediators of treatment effectiveness if we are to better understand the needs of this challenging group.