Do Psychopathic Traits Assessed in Mid-Adolescence Predict Mental Health, Psychosocial, and Antisocial, Including Criminal Outcomes, Over the Subsequent 5 Years?

Abstract

Objective:

To determine whether psychopathic traits assessed in mid-adolescence predicted mental health, psychosocial, and antisocial (including criminal) outcomes 5 years later and would thereby provide advantages over diagnosing conduct disorder (CD).

Method:

Eighty-six women and 61 men were assessed in mid-adolescence when they first contacted a clinic for substance misuse and were reassessed 5 years later. Assessments in adolescence include the Psychopathy Checklist—Youth Version (PCL-YV), and depending on their age, either the Kiddie-Schedule for Affective Disorders and Schizophrenia for School-Aged Children or the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (SCID). Assessments in early adulthood included the SCID, self-reports of psychosocial functioning, aggressive behaviour, and criminality and official criminal records.

Results:

The antisocial facet score positively predicted the number of anxiety symptoms and likelihood of receiving treatment for substance use disorders (SUDs). Lifestyle and antisocial facet scores negatively predicted Global Assessment of Functioning scores. By contrast, the interpersonal score and male sex independently and positively predicted the number of months worked or studied, as did the interaction of Lifestyle × Sex indicating that among men, but not women, an increase in lifestyle facet score was associated with less time worked or studied. Interpersonal and antisocial scores positively predicted school drop-out. Antisocial facet scores predicted the number of symptoms of antisocial personality disorder, alcohol and SUDs, and violent and nonviolent criminality but much more strongly among males than females. Predictions from numbers of CD symptoms were similar.

Conclusions:

Psychopathic traits among adolescents who misuse substances predict an array of outcomes over the subsequent 5 years. Information on the levels of these traits may be useful for planning treatment.

Most adolescents who misuse substances present with mental disorders that onset prior to substance misuse, most commonly CD.^1,2 CD is associated with an elevated risk of persistent antisocial behaviour, substance misuse, school drop out, unwanted pregnancy,³ criminal convictions,⁴ anxiety, depression,^5,6 and ASPD.⁷ Some children and adolescents with CD show elevated levels of psychopathic traits. These children show more severe conduct problems than other children with CD, including aggressive behaviour and an elevated risk of persistent criminal offending.^8,9 They show distinct impairments of cognition and emotion processing similar to adults with psychopathy. Emerging evidence suggests that these traits are present as early as age 5.¹⁰

Children with CD and high levels of psychopathic traits are relatively insensitive to punishment and tend to behave inappropriately regardless of negative consequences.^11,12 One RCT has shown that children with conduct problems and high levels of psychopathic traits fail to respond to time out,¹³ and other studies propose that they require specific interventions that target their empathetic deficits.^10,14 Similarly, adolescents with conduct problems and psychopathic traits are difficult to engage in treatment and respond poorly to offender rehabilitation programs.^15,16 Thus the presence of psychopathic traits among adolescents engaging in antisocial behaviour may indicate the need for a specific treatment approach. Most services for adolescents engaging in substance misuse, to the best of our knowledge, do not assess psychopathic traits, even though the importance of this differential diagnosis is now recognized by experts.¹⁷ Our study aimed to determine whether psychopathic traits in mid-adolescence predicted mental health, psychosocial, and antisocial (including criminal) outcomes 5 years later and would thereby provide advantages over the usual procedure of diagnosing CD.

Clinical Implications

• Training clinicians working with adolescents who misuse substances to use the PCL-YV may contribute to better understanding of adolescents.

• Psychopathic traits assessed in adolescence were associated with anxiety, treatment use, and substance misuse in the subsequent 5 years.

• The number of CD symptoms present before age 15 was also associated with these outcomes.

Limitations

• The small sample size may have affected predictions of antisocial or criminal behaviour among females.

• No information was available about whether treatments in the follow-up period were involuntary, under civil or criminal court orders.

• Our study was of a treated sample of adolescents and results may not generalize to all teenagers misusing substances.

The PCL-YV¹⁸ was designed to assess psychopathic traits among adolescents. Factor analysis identifies 4 factors: interpersonal—impression management, grandiose sense of self-worth, pathological lying, and manipulation for personal gain; affective—lack of remorse, shallow affect, callous or lack of empathy, and failure to accept responsibility; lifestyle—stimulation seeking, parasitic orientation, lacks goals, impulsivity, and irresponsibility; and antisocial—poor anger control, early behaviour problems, serious criminal behaviour, serious violations of conditional release, and criminal versatility.¹⁹ In a meta-analysis of 21 studies, 15 of them focusing exclusively on males, the PCL-YV or the PCL-Revised¹⁹ for adults predicted general and violent recidivism, more strongly among men than women.²⁰ Other studies of adolescents have shown that the PCL-YV predicted violent and nonviolent criminal convictions among males and weakly or not at all among females,^21,22 and violent behaviour.^23,24 We found only one study²³ examining PCL-YV predictions after controlling for CD and it showed that the PCL-YV continued to predict future violent offending among adolescent males.

Few studies have examined associations between psychopathic traits in adolescence and outcomes other than aggressive behaviour and criminality. The results of these studies are generally inconsistent, for example, those on internalizing problems^21,25 and suicidal behaviour.²⁶ A recent study²⁷ showed that among adolescent boys both CD and psychopathic traits predicted substance misuse 3 years later, while among girls only CD was associated with subsequent substance misuse.

Our study aimed to determine whether psychopathic traits assessed by the PCL-YV in mid-adolescence among girls and boys attending a clinic for substance misuse problems predicted mental health, psychosocial, and antisocial (including criminal) outcomes 5 years later and would thereby provide advantages over the usual procedure, that of diagnosing CD. Four questions were addressed:

1. Do scores for psychopathic traits assessed in mid-adolescence predict mental health (depression symptoms, anxiety symptoms, suicide attempts, treatment of substance abuse, treatment of other mental disorders, or number of admissions to psychiatric wards), psychosocial functioning (GAF score, time spent working or studying, school dropout, or having a child at a young age), and antisocial including criminal behaviour (symptoms of ASPD, AUDs, SUDs, self-reported aggressive behaviour, convictions and [or] self-reports for violent and nonviolent crimes) 5 years later?

2. Does the number of CD symptoms prior to age 15 predict these mental health, psychosocial, and antisocial (including criminal) outcomes 5 years later?

3. Do psychopathic traits predict any of these mental health, psychosocial, and antisocial (including criminal) outcomes 5 years later after controlling for CD symptoms?

4. Do predictions differ among females and males?

Method

Participants

In 2004, a sample of 99 females and 81 males, aged, on average, 16.8 years (SD 1.9) who contacted the only substance misuse clinic in a large urban area in Sweden, participated in a study.²⁸ Five years later, 86 women and 61 men from the original sample completed interviews. There were no significant differences between those who participated in the follow-up and those who did not regarding sex, PCL-YV total and facet scores, and the number of CD symptoms.

Measures in Mid-Adolescence

The PCL-YV¹⁸ is a 20-item rating scale measuring psychopathic traits in adolescents. Each item is scored 0 (consistently absent), 1 (inconsistently present), or 2 (consistently present). The PCL-YV was rated by clinical psychologists trained to use this instrument, and, as recommended based on the interview and clinical files. Interrater reliability was calculated on 29 clients using intraclass correlation: total score 0.86; interpersonal facet score 0.61; affective facet score 0.74; lifestyle facet score 0.67; and antisocial facet score 0.87.

Participants aged 17 and younger were assessed using the Kiddie-Schedule for Affective Disorders and Schizophrenia for School-Aged Children—Present and Lifetime Version.²⁹ Fifteen participants were rated independently by a second clinician and interrater reliability was high (CD and ODD, κ = 0.82). Participants aged 18 or older were interviewed using the SCID I and II.^30,31 Twelve participants were rated independently by a second clinician and interrater reliability was high (CD and ODD, κ = 0.82).²⁸

Measures in Early Adulthood

Ex-clients were assessed using the SCID I and II.^30,31 Twelve participants were rated independently by a second clinician and interrater reliability was high (for example, CD, κ = 0.82 and AUD, κ = 0.83). Symptoms were summed to provide total scores for depression, anxiety, AUD, drug abuse and (or) dependence, and ASPD.

Suicide attempts were scored absent (0) or present (occurred 1 or more times).

Information about treatment was collected with the life history calendar,³² from the SCID I,³⁰ and from medical records.

The number of admissions to psychiatric wards from the initial assessment was collected from Patientregistret (a national register of visits to Swedish hospitals).

The GAF³³ assessed psychological, social, and occupational functioning. GAF scores range from 1 to 100.

The life history calendar³² documented work and studies. The number of months worked or studied of more than 30 hours per week from the initial assessment was defined as an outcome.

School drop out was coded as present if less than 7 years of school were completed.

Participants reported whether or not they had children at a young age.

The MacArthur Community Violence Instrument³⁴ was administered by interview, to report on all types of physically aggressive behaviours toward others in the past 6 months. Item scores were summed to provide a total score.

Eleven questions were asked about violent crimes over the past year. Information about the number of violent crime convictions was extracted from official Swedish criminal records. Item scores and number of convictions were summed.

Fifteen questions were asked about nonviolent crimes over the past year. Information about the number of convictions for nonviolent offences was extracted from official Swedish criminal records. Item scores and number convictions were summed.

Procedure

When adolescents first contacted the clinic, they were invited to participate in the study. After a complete description of the study, and answers to their questions, the adolescent and each parent formally signed consents to complete interviews and questionnaires, and to allow the research team access to their clinical files, criminal, and social insurance records. Interviews were then scheduled with adolescents and each parent separately. For participation in the study, adolescents received a gift certificate worth SEK500 for a department store and a cinema ticket.

Five years later, the ex-clients were asked to again participate in the study. Those who agreed signed formal consents to complete interviews and questionnaires and to allow the research team to access their clinical files, criminal, health, and social insurance records, and provided a sample of saliva for DNA extraction. Interviews were conducted, on average, 5.6 years (SD 0.85) after the initial assessment. For participation in the follow-up, the ex-clients received a gift certificate worth SEK500 for a department store.

Both the initial and the follow-up study were approved by the Karolinska Institute Research Ethics Committee Nord (Swedish registration number [DNR] 03–543) and Regionala etikprövningsnämnden in Stockholm (DNR 2008/1934–31/3).

Statistical Analyses

Characteristics of female and male participants were compared using Student t tests and chi-square statistics. Binary correlations were calculated between predictor variables (PCL-YV facet scores and the number of CD symptoms) and outcome variables using Kendall tau (eTable 1). Regression models were calculated to identify independent associations of predictors with the dependent variables. Given the number of predictors, the risk of type I errors was high and was reduced by grouping predictors into blocks. Multiple regression analyses were computed to examine outcomes of continuous variables: symptoms of depression and anxiety, admissions to a psychiatric ward, months worked or studied, symptoms of ASPD, AUDs, SUDs, incidents of aggressive behaviour, convictions or self-reports of violent and nonviolent crime, and GAF scores. Logistic regression models were calculated for dichotomous outcomes: suicide attempts, treatment for substance abuse, other mental health treatment, school drop-out, and having a child at a young age. Owing to outliers on the dependent variables, 1 participant was removed from all of the analyses, and 6 participants from analyses of the number of admissions to psychiatric wards. The first series of models estimated the independent associations of PCL facet scores, with outcomes and took account of sex. By specifying the 4 facet scores as predictors, the overlap between the facets was controlled, thereby providing an independent estimate of the association of each facet with each outcome. Model I included PCL facet scores. Model II included facets scores, sex, and an interaction term of the facet scores with sex. Interaction terms were included one by one. A final model included significant predictors from models I and II. Similar analyses estimated the associations of the number of CD symptoms with outcomes. Finally, one model was calculated to determine whether the PCL facets that were predictive of outcomes in the final model would remain after considering the number of CD symptoms. To examine interaction terms, the following variables were dichotomized (low: 0 to 2, high: 3 or higher): antisocial facet; lifestyle facet, and the number of CD symptoms. Independent Student t tests were calculated to estimate significance of differences.

Results

Characteristics of the Sample

As shown in Table 2, women had significantly lower PCL total, interpersonal, affective, and antisocial scores, and number of CD symptoms than men. Within each domain of functioning, sex differences were observed. Women presented more depressive and anxiety symptoms, and more reported having children, but women had fewer symptoms of ASPD and SUDs, and fewer violent and nonviolent crimes than men. Among males, 50.8% had committed violent crimes and 93.4% nonviolent crimes, and among females, 20.9% had committed violent crimes and 88.4% nonviolent offences.

Table 2.

Characteristics of the clients

Characteristic	Womena n = 86	Mena n = 61	Difference
PCL total score	12.36 (5.88)	15.86 (8.05)	t = 2.86, df = 103.5, P = 0.005
Interpersonal facet score	2.70 (1.42)	3.21 (1.73)	t = 1.975, df = 145, P = 0.05
Affective facet score	2.36 (1.89)	3.22 (2.06)	t = 2.625, df = 145, P = 0.01
Lifestyle facet score	4.22 (1.73)	4.90 (2.33)	t = 1.91, df = 105.3, P = 0.06
Antisocial facet score	2.11 (1.80)	3.56 (2.67)	t = 3.680, df = 97.5, P < 0.001
Number of CD symptoms	2.85 (2.98)	4.10 (3.47)	t = 2.227, df = 116.7, P = 0.02
Mental health
Number of depression symptoms	4.31 (3.18)	2.07 (2.54)	t = 4.757, df = 142.8, P < 0.001
Number of anxiety symptoms	5.23 (5.74)	2.23 (3.85)	t = 3.795, df = 144.6, P < 0.001
Suicide attempts, %	8.1	6.6	χ2 = 0.002, df = 1, P = 0.97; n = 147
Treatment for substance misuse, %	49.4	45.9	χ2 = 0.063, df = 1, P = 0.80; n = 146
Treatment for other mental disorders, %	71.8	68.9	χ2 = 0.039, df = 1, P = 0.84; n = 146
Number of admissions to psychiatric wards	0.74 (1.37)	0.67 (1.47)	t = 0.262, df = 132, P = 0.79
Psychosocial functioning
GAF score	66.36 (11.31)	64.49 (11.93)	t = 0.925, df = 145, P = 0.36
Dropped out of school, %	3.5	6.7	χ2 = 0.241, df = 1, P = 0.62; n = 146
Months worked or studied	30.34 (18.15)	28.69 (19.52)	t = 0.526, df = 145, P = 0.60
A child, %	29.1	13.3	χ2 = 4.144, df = 1, P = 0.04; n = 146
Antisocial or criminal behaviour
Number of ASPD symptoms	0.50 (1.07)	1.59 (2.02)	t = 3.849, df = 83.96, P < 0.001
Number of alcohol disorder symptoms	2.08 (2.74)	2.13 (2.90)	t = 0.106, df = 145, P = 0.92
Number of drug disorder symptoms	1.63 (2.86)	2.64 (3.16)	t = 2.022, df = 145, P = 0.04
Self-reported aggressive behaviour	1.05 (1.63)	1.41 (2.01)	t = 1.165, df = 111.79, P = 0.25
Number of violent criminality	0.47 (1.22)	2.03 (3.05)	t = 3.775, df = 73.83, P < 0.001
Number of nonviolent criminality	2.80 (2.68)	8.03 (13.77)	t = 2.928, df = 63.23, P = 0.005

^aMean (SD) unless otherwise stated

Do psychopathic traits assessed in mid-adolescence predict mental health, psychosocial functioning, and antisocial including criminal behaviour 5 years later?

Table 3 presents the results of the final regression model, using PCL-YV facet scores in mid-adolescence, and interactions of these scores with sex, to predict mental health, psychosocial functioning, and antisocial (including criminal) behaviour 5 years later. (eTable 4 for results of initial models). The antisocial score positively predicted the number of anxiety symptoms and likelihood of receiving treatment for SUDs.

Table 3.

Results of regression analyses using PCL-YV scores assessed in mid-adolescence and interaction terms by sex to predict mental health, psychosocial functioning, and antisocial outcomes 5 years later

Predictor	Mental health						Psychosocial functioning
	Number of anxiety symptoms		Treatment for SUD		Other mental health treatment		GAF		Months worked or studied		Dropped out of school
	βa	P	OR (95% CI)	P	OR (95% CI)	P	βa	P	βa	P	OR (95% CI)	P
Final model r2	0.04						0.25		0.12
Sex									0.35	0.03
Interpersonal					1.24 (0.95–1.62)	0.11			0.24	0.007	0.49 (0.25–0.96)	0.04
Lifestyle							−0.26	0.01	−0.006	0.96
Antisocial	0.19	0.02	1.24 (1.07–1.44)	0.005	1.20 (0.99–1.46)	0.06	−0.29	0.004			2.07 (1.35–3.17)	0.001
Lifestyle × Sex									−0.55	0.006
Predictor
	Antisocial or criminal behaviour
	Number of symptoms of						Score for		Number of convictions or self-reports
	ASPD		AUD		SUD		Aggressive behaviour		Violent crimes		Nonviolent crimes
Final model r2	0.59		0.17		0.20		0.17		0.31		0.24
Sex	−0.25	0.07	−0.49	0.005			−0.25	0.05	−0.04	0.73	−0.09	0.42
Interpersonal	0.17	0.04	−0.08	0.51
Affective			−0.19	0.14
Lifestyle	0.09	0.31
Antisocial	0.35	<0.001	0.39	0.001	0.44	<0.001	0.15	0.26	0.19	0.11	0.13	0.28
Interpersonal × Sex	0.24	0.10	0.25	0.26
Lifestyle × Sex	0.28	0.06	0.27	0.16
Antisocial × Sex							0.41	0.02	0.42	0.009	0.44	0.009

^aStandardized β

Lifestyle and antisocial facet scores negatively predicted GAF scores. By contrast, the interpersonal score and male sex independently and positively predicted the number of months worked or studied, as did the interaction of Lifestyle × Sex, indicating that, among men, but not women, an increase in lifestyle facet score was associated with less time worked or studied. Men with high, compared with low, lifestyle score worked or studied significantly (t = 2.76, df = 58, P = 0.008) fewer months, high mean 18.24, (SD 17.63) and low mean 32.98 (SD 19.02). Among women, there was no difference in time worked or studied between those with high and low scores. Interpersonal and antisocial scores positively predicted school drop out.

All 6 antisocial outcomes in early adulthood were predicted by psychopathic traits assessed in mid-adolescence. The number of ASPD symptoms was predicted by interpersonal and antisocial scores. The number of symptoms of AUDs was predicted by male sex and the antisocial score, while the number of symptoms of SUDs was predicted only by the antisocial score. Aggressive behaviour was predicted by male sex and the interaction Antisocial × Sex, while violent and nonviolent crimes were predicted by the interaction Antisocial × Sex. These significant interaction terms, plus follow-up Student t tests, indicate that high antisocial facet scores, both among men and women, are associated with aggressive behaviour, violent and nonviolent crimes, much more strongly among men than women (online eResults).

Do the number of CD symptoms assessed in mid-adolescence predict mental health, psychosocial functioning, and antisocial (including criminal) behaviour 5 years later?

Table 5 presents the results of the final model predicting outcomes by CD Symptoms and Symptoms × Sex (online eTable 6 for results of initial models). The number of CD symptoms positively predicted the number of anxiety symptoms, as did the interaction of CD Symptoms × Sex. Women with high, as compared with low, numbers of CD symptoms presented significantly with more anxiety symptoms, high mean 7.00 (SD 6.26), and low mean 4.02 (SD 5.06) (t = 2.44, df = 84, P = 0.02). Among men, no difference in anxiety symptoms was detected for those with high and low number of CD symptoms. The number of CD symptoms positively predicted the numbers of admissions to psychiatric wards. CD symptoms negatively predicted GAF score and positively predicted school drop out.

Table 5.

Results of regression analyses using the number of CD symptoms assessed in mid-adolescence and interaction terms by sex to predict mental health, psychosocial functioning, and antisocial outcomes 5 years later

Predictor	Mental health				Psychosocial functioning
	Number of anxiety symptoms		Admissions to psychiatric wards		GAF		Months worked or studied		Dropped out of school
	Standardized β	P	Standardized β	P	Standardized β	P	Standardized β	P	OR (95% CI)	P
Final model r2	0.18		0.06		0.15		0.04
Sex	−0.16	0.16			0.11	0.35	0.11	0.36
Number of CD symptoms	0.43	<0.001	0.21	0.01	−0.29	0.01	−0.06	0.59	1.35 (1.08−1.69)	0.009
CD Symptoms × Sex	−0.30	<0.001			−0.17	0.24	−0.14	0.18
Predictor	Antisocial or criminal behaviour
	Number of symptoms of						Score for		Convictions or self-reports
	ASPD		AUD		SUD		Aggressive behaviour		Violent crimes		Nonviolent crimes
Final model r2	0.49		0.09		0.13		0.13		0.28		0.24
Sex	−0.05	0.54	−0.10	0.41	0.003	0.98	−0.18	0.13	−0.02	0.86	−0.07	0.54
Number of CD symptoms	0.34	<0.001	0.29	0.01	0.22	0.05	0.09	0.43	0.09	0.39	0.09	0.38
CD Symptoms × Sex	0.46	<0.001	0.05	0.76	0.17	0.26	0.37	0.01	0.49	<0.001	0.47	0.001

As expected, CD symptoms positively predicted the number of symptoms of ASPD, as did the interaction term CD Symptoms × Sex, indicating, and confirmed by follow-up Student t tests that, among men, the association was much stronger than among women. The number of CD symptoms positively predicted the number of symptoms of AUDs and SUDs. Only the interaction CD Symptoms × Sex predicted aggressive behaviour, and the number of violent and nonviolent crimes. Follow-up Student t tests confirmed that men with high, as compared with low, numbers of CD symptoms obtained significantly higher scores for aggressive behaviour, and significantly more convictions or self-reports of violent and nonviolent crimes. Among women, those presenting high and low numbers of CD symptoms did not differ on scores for aggression or violent crimes, but they did acquire significantly more nonviolent offences.

Table 7 presents a summary of the results of the final regression models reported above that predicted mental health, psychosocial functioning, and antisocial behaviour using the PCL-YV facet scores in one series of analyses and the number of CD symptoms in another. PCL facets and CD symptoms predicted indices of antisocial behaviour similarly, while the PCL facets predicted 2 mental health and several psychosocial outcomes.

Table 7.

Summary of PCL-YV scores and CD symptoms assessed in mid-adolescence that predicted outcomes 5 years later

Outcome	PCL-YV facet scores				Number of CD symptoms
	Interpersonal	Affective	Lifestyle	Antisocial
Mental health
Depression
Anxiety					women stronger
Suicide attempts
Treatment for substance misuse
Other treatment
Inpatient admissions
Psychosocial functioning
GAF score
Time worked or studied			men stronger
School drop out
Child at young age
Antisocial or criminal behaviour
ASPD					men stronger
AUD
SUD
Aggressive behaviour				men only	men only
Violent crimes				men stronger	men only
Nonviolent crimes				men stronger	men stronger

Positive prediction;

negative prediction

Are psychopathic traits a better predictor of outcomes after 5 years than the number of CD symptoms?

Table 8 presents the results of the regression model in which PCL facet scores were entered in addition to the number of CD symptoms. No mental health outcomes were predicted. One psychosocial outcome (GAF score) was negatively predicted by the lifestyle and antisocial facets. The number of ASPD symptoms was positively predicted by the lifestyle and antisocial facets in addition to CD symptoms. The number of symptoms of AUDs and SUDs were predicted only by the antisocial facet scores, while aggressive behaviour and violent and nonviolent crimes were only predicted by the interaction Antisocial × Sex.

Table 8.

Results of regression analyses predicting mental health, psychosocial functioning, and antisocial or criminal behaviour outcomes by psychopathic traits accounting for CD symptoms

Predictor	Mental health		Psychosocial functioning
	Numbers of anxiety symptoms		GAF
	Standardized β	P	Standardized β	P
Sex	0.05		0.25
Interpersonal
Lifestyle			−0.25	0.01
Antisocial	0.07	0.54	−0.26	0.03
Lifestyle × Sex
Number of CD	0.16	0.19	−0.04	0.71
Predictor
	Antisocial or criminal behaviour
	Number of symptoms						Score for		Convictions or self-reports
	ASPD		AUD		SUD		Aggressive behaviour		Violent crimes		Nonviolent crimes
Final model r2	0.55		0.12		0.20		0.17		0.31		0.26
Sex							−0.25	0.05	−0.04	0.73	−0.10	0.41
Interpersonal	0.28	<0.001
Antisocial	0.40	<0.001	0.26	0.03	0.44	<0.001	0.10	0.52	0.11	0.45	0.003	0.98
Antisocial × Sex							0.42	0.02	0.43	0.008	0.46	0.007
Number of CD	0.22	0.008	0.11	0.34	0.01	0.93	0.06	0.62	0.10	0.32	0.16	0.13

Discussion

Our study sought to determine the use of the PCL-YV in predicting mental health, psychosocial, and antisocial outcomes in early adulthood among females and males who had sought treatment for substance misuse in mid-adolescence. In general, the PCL-YV provided added value over and above a diagnosis of CD. Multivariate analyses were computed to determine the independent associations of each of the PCL-YV facets to outcomes. The antisocial facet predicted anxiety symptoms, substance misuse treatment, GAF scores, school drop out, and all antisocial outcomes, including ASPD, symptoms of AUDs and SUDs, aggressive behaviour, and violent and nonviolent criminality. Additionally, the interpersonal factor was associated with time worked or studied, school drop out, and ASPD, while the lifestyle factor was associated with GAF scores and time work or studied. Notably, the affective facet, often referred to as the core of psychopathy, was not associated with any outcome. Thus the PCL-YV facet scores assessed in mid-adolescence predicted a broad array of outcomes 5 years later.³⁵

The number of CD symptoms assessed in mid-adolescence also predicted outcomes 5 years later, including anxiety symptoms, inpatient admissions, GAF score, school drop out, and all antisocial outcomes. Adolescents with the highest levels of antisocial behaviour, indicated by either high antisocial facet scores or the number of CD symptoms, were more likely than those with lower levels to receive treatment. The data did not allow further investigation of these associations. It could be that clinicians in this clinic for adolescent substance misusers selected the most severely antisocial youth for treatment. However, previous studies of adolescents and adults report that psychopathic traits are associated with a failure to engage and benefit from treatment.^15,16,36 It could be that these adolescents with high scores for manipulation and other psychopathic traits use mental health services for their own ends, such as negotiating with police not to charge them if they go into treatment, or detoxing in hospital to use less drugs to get high. These findings warrant further study.

Both the PCL-YV facets and CD symptoms predicted antisocial outcomes more strongly among males than females, suggesting that, among adolescent females, other factors are contributing to maintaining, and perhaps extending, antisocial behaviour. Both the antisocial facet and CD symptoms predicted anxiety symptoms more strongly among females than males. This result concurs with findings from other studies showing that both ASPD and CD are associated with elevated rates of anxiety disorders.^37,38 Thus these comorbid disorders may be relatively long-standing by adolescence and present a challenge to clinics focused on substance misuse.

The assessment of CD symptoms in mid-adolescence provides a wealth of information about the onset and development of antisocial behaviour, and as our study and others have shown, it predicts future antisocial (including criminal) behaviours, poor psychosocial functioning, and behaviours such as dropping out of school that have long-term consequences. The PCL-YV facets also predict these same behaviours, and may provide additional information that is clinically useful. While maladaptive personality traits generally decline from mid-adolescence to early adulthood,^39,40 emerging evidence suggests that psychopathic traits are relatively stable, and consequently will continue to be associated with anxiety, poor psychosocial functioning, and antisocial (including criminal) behaviours in adulthood. Based on both an extensive interview and clinical and criminal files, the PCL-YV assesses pathological lying and manipulation. Knowing about these attributes allows clinicians to better judge the usefulness of adolescents’ self-reports. If levels of these and other psychopathic traits are high, skepticism on the part of clinicians may also be warranted regarding pronouncements of future projects, including participation in treatment. Unrealistic and grandiose plans, and a lack of future goals, are specifically assessed with the PCL. All of these characteristics are important for developing a treatment program likely to reduce substance misuse and for implementing measures to promote engagement with treatment. Further, studies of children presenting psychopathic traits indicate that they require different interventions from those that are effective with other children with CD and their parents.^14,41 One RCT has shown that children with CD and high levels of psychopathic traits learn from reward, not time out,^13,14,42 and proposals suggest that interventions focus on their insensitivity to others.¹⁰ Given the evidence that psychopathic traits are relatively stable by mid-adolescence, and that in our study the affective facet was not associated with any outcomes, research is needed to determine whether adolescents misusing substances who, present medium to high levels of these traits, would benefit from a specific treatment approach.

Our study included a small sample and multiple predictors that may have limited the validity of predictions. All participants sought treatment and thus findings may not generalize to community samples. The lack of information on patients for whom treatment was compulsory limits understanding of the sample.

Conclusion

The PCL-YV may provide useful information to clinicians treating adolescents engaging in substance misuse as it predicted mental health, psychosocial, and antisocial (including criminal) behaviour into early adulthood. Future RCTs are needed to determine whether specific treatment approaches would benefit adolescents presenting these traits.

Abbreviations

ASPD

antisocial personality disorder

AUD

alcohol use disorder

CD

conduct disorder

DSM

Diagnostic and Statistical Manual of Mental Disorders

GAF

Global Assessment of Functioning

ODD

oppositional defiant disorder

PCL-YV

Psychopathy Checklist—Youth Version

RCT

randomized controlled trial

SCID I

Structured Clinical Interview for DSM-IV Axis I Disorders

SCID II

Structured Clinical Interview for DSM-IV Axis II Disorders

SUD

substance use disorder