Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Characterising youth with callous–unemotional traits and concurrent anxiety: evidence for a high-risk clinical group

Characterising youth with callous–unemotional traits and concurrent anxiety: evidence for a... Growing evidence supports the existence of two variants of youth with high callous–unemotional (CU) traits who present with markedly different risk profiles and outcomes, with potential implications for risk assessment and treatment formulation. So far, studies have identified variants of CU youth mainly using data-driven cluster approaches based on levels of CU traits and co-occurring anxiety. Yet, the extent to which this knowledge may be translated into clinical practice is unclear. To this end, the present study employed a severity-based, cut-off approach to systematically characterise CU groups across a range of clinically informative domains, including trauma history, psychiatric symptomatology, affective functioning, attachment style and behavioural risk. Analyses were based on multi-rated data from a community sample of high-risk youths (n = 155, M = 18 years). Consistent with previous studies, we found that, whereas variants show comparable levels of antisocial behaviour, those who present with both high CU and high anxiety report more severe childhood maltreatment, psychologi- cal distress, ADHD symptomatology and behavioural risk—including substance use, suicidal ideation and unsafe sex. In addition, these youth show greater attachment insecurity and affective dysregulation, as indexed by levels of irritability and alexithymia. Together, findings indicate that (1) trauma history is a key factor that differentiates variants of CU youth high vs. low on anxiety, and (2) differences in individual functioning across variants point to the need for tailored clinical assessment tools and intervention strategies. Importantly, the present findings indicate that variants of CU youth can be meaningfully differentiated using cut-off based approaches that parallel methods used in clinical assessments. Keywords Callous–unemotional traits · Anxiety · Childhood maltreatment · Psychopathology · Adolescence Introduction callousness and shallow affect [ 2]. Compared to other anti- social youth, those with high CU traits show marked dif- In the DSM-5, callous–unemotional (CU) traits—referred to ferences in neurocognitive, emotional and behavioural as ‘Limited Prosocial Emotions’—feature as a new diagnos- functioning, including difficulties in social-information pro - tic specifier for conduct disorder, to enable the identification cessing [3], under-arousal to empathy-inducing stimuli [4], of a particularly severe subgroup of youth at increased risk disruptions in affective theory of mind [ 5], lower sensitiv- for early-onset and persistent antisocial behaviour [1]. CU ity to punishment [1] as well as alterations in brain regions traits are defined by a core set of affective features (paral - involved in emotion and learning (e.g. amygdala, PFC; [2]). leling the affective dimension of adult psychopathy), which Together, these features are thought to contribute to the more include low capacity for empathy, lack of guilt and remorse, violent, chronic and recidivistic pattern of antisocial behav- iour displayed by youth with high CU traits, and represent an important target for intervention. * Essi Viding It is also becoming increasingly clear that not all youth [email protected] with high CU traits are the same. Rather, they can present Department of Psychology, Institute of Psychiatry, King’s with different levels of co-occurring anxiety [ 6, 7]. This is College London, London, UK akin to what has been observed in adults with psychopathy Division of Psychology and Language Sciences, University [8] and is thought to reflect the existence of two variants with College London, 26 Bedford Way, London WC1H 0AP, UK potentially distinct aetiologies—a theory first put forward Department of Psychology, Royal Holloway, University by Karpman [9] and Cleckley [10] in 1941. Specifically, of London, Egham, UK Vol.:(0123456789) 1 3 886 European Child & Adolescent Psychiatry (2018) 27:885–898 CU traits accompanied by low levels of anxiety (CU−Anx suicidal ideation, thought disturbance) compared to both the variant) are thought to be associated with substantial devel- CU−Anx and control group. In another study based on male opmental genetic risk, whereas CU traits accompanied by juvenile offenders (n  = 238, age 14–19 years), Sharf and high levels of anxiety (CU+Anx variant) are thought to be colleagues [20] used a median-split approach to create the associated primarily with environmental trauma [11, 12]. same three groups (i.e. CU+Anx, CU−Anx, control group) The two variants are indistinguishable based on CU traits and found that the CU+Anx group reported greater exposure alone (i.e. they can be thought of as ‘behavioural pheno- to negative life events (especially violence exposure in the copies’), but the CU+Anx variant is associated with more home and community) as well more severe post-traumatic severe pre- [7, 13] and post-natal [14, 15] adversity, with symptoms compared to the other two groups. While promis- the most consistent evidence relating to childhood maltreat- ing, these studies have focussed exclusively on male youth ment [16–20]. Furthermore, variants have been shown to offenders, so that more work is needed to test whether sever - differ markedly in presentation across a range of domains, ity-based approaches can meaningfully differentiate variants including comorbid psychiatric symptomatology [7, 14, 15, in non-forensic, multi-gender populations across a wider 21, 22], impulsivity [19], self-control [23], empathy [24], range of clinically informative domains. personality traits [25], expression of aggression [26], nega- A further question with important clinical implications tive affect [20], emotional lability [27], emotional process- is how CU+Anx youth compare not only to their CU−Anx ing [6, 17, 28, 29], behavioural risk [15, 18] and biological counterparts, but also to youth who present with high anxi- function [13, 26, 30]. ety alone (Anxious group). Contrasting these two groups is Given that CU levels are currently used to inform risk necessary in order to clarify whether (1) CU+Anx youth assessment and treatment options with antisocial youth [1], experience a ‘double hit’ of negative outcomes associated the existence of variants may carry important implications with two relatively independent dimensions of psychopa- for clinical practice [17]. So far, studies have primarily iden- thology; or (2) whether the combination of high CU and tified variants of CU youth using state-of-the-art clustering Anxiety indexes a particularly high-risk group of youth who approaches, which are hypothesis-free and person-centred show additional vulnerabilities compared to those who pre- [6, 7, 14–18, 26–28, 30–32]. Together, these reports have sent with either CU or Anxiety alone. Because studies to been invaluable in demonstrating that individuals naturally date (both cluster-based and severity-based) have generally cluster into groups based on their on levels of CU and anxi- contrasted variants of CU youth to a single, generic com- ety—providing strong, data-driven evidence validating the parison group (i.e. not disaggregated by level of anxiety), existence of two variants of CU youth. However, cluster- it has not been possible to systematically address this ques- ing approaches are not practical in clinical settings, where tion. To our knowledge, only one study based on a com- treatment and risk assessment decisions are typically based munity sample of adolescents has compared the CU+Anx on variable-centred, severity-based thresholds. A handful group to a reference group who show comparable levels of of other studies have examined these traits continuously, as anxiety [31]. Interestingly, the authors reported that although opposed to comparing groups, in order to establish whether the Anxious group displayed lower levels of CU traits and presence of anxiety or trauma history moderates the associa- antisocial behaviour compared to the CU+Anx variant, the tion between CU traits and outcomes, such as empathy [24] groups presented similarly in other domains, such as low or emotional recognition [33]. While such an approach has self-esteem. Furthermore, the Anxious group consisted pri- the advantage of modelling the full range of scores, lending marily of girls, which may explain the failure to identify useful insights into the dimensional relationship between this subgroup in prior studies that have typically focused CU and anxiety, it is particularly difficult to implement in a on juvenile male offender samples. The study, however, did clinical setting. not compare groups on trauma history, psychiatric risk, and As an alternative, a small set of studies have shown that affective functioning—key clinical domains that need sys- simpler cut-off approaches (e.g. based on average scores) tematic investigation if we are to more fully understand the can be successfully employed to compare variants on spe- nature of the CU+Anx variant. cific outcomes, yielding results that are consistent with those derived from cluster-based approaches. For example, in a Romanian sample of incarcerated males (n = 125, age The present study 14–18 years), Rosan and colleagues [19] used the sample average score of CU traits and anxiety as a cut-off thresh - The aim of the present study was to comprehensively charac- old to classify youth as either CU+Anx, CU−Anx or a terise variants of CU traits in a community sample of high- control group low on both dimensions. The authors found risk youth. Specifically, we investigated whether variants that the CU+Anx group showed significantly higher lev - of CU youth (i.e. CU−Anx vs CU+Anx) identified using els of impulsivity and emotional dysregulation (e.g. anger, a variable-centred, median-based approach differ across: 1 3 European Child & Adolescent Psychiatry (2018) 27:885–898 887 (1) previously validated domains, including childhood mal- Procedure treatment history, (multi-rated) psychiatric symptoms, and behavioural risk markers; as well as (2) novel functional All procedures performed were in accordance with the ethical domains, including attachment style and affective function- standards of the UCL Research Ethics Committee (ID No. ing (irritability and alexithymia). To improve the specificity 2462/001) and with the 1964 Helsinki declaration and its later of any conclusions about these groups we also compared amendments or comparable ethical standards. Youth from the both CU variant groups with two clinically relevant compari- charity were introduced to the research by a member of staff, son groups a Low group (low on both CU and anxiety) and and, if interested, were provided information about the study an Anxious group (low on CU but high in anxiety). Based by one of the research team on site. As a result, all youth who on previous studies using cluster-based analyses as well as met with the researchers had shown interest in the study and those that have used severity-based cut-offs, we predicted agreed to participate. After the testing session, each partici- that, relative to youth only high in CU traits (CU−Anx), pant’s key worker completed a questionnaire booklet. A key those with high CU and high anxiety (CU+Anx) would be worker is a member of staff of the charity who is assigned to characterised by: (1) more severe experiences of childhood each client upon referral in order to assist in the delivery of maltreatment; (2) greater levels of psychological distress services as well as to provide socio-emotional and practical and psychiatric symptomatology; (3) significantly elevated support. In schools, youth initially received information during behavioural risk markers; but (4) similar levels of externalis- a brief presentation at a school assembly. Information sheets ing problems. Given the lack of prior research, no a priori and consent forms were then distributed to students who had hypotheses were made regarding associations with attach- attended the presentation. Those students who were interested ment style or affective functioning (as indexed by levels of in taking part completed the consent form and returned it to irritability and alexithymia) between variants of CU youth. the researchers. As a result, researchers met exclusively with Compared to the Anxious group, we expected that CU+Anx students who were interested in participating and had provided youth would show higher levels of externalising problems informed consent stating that they were willing to take part in (in line with previous studies [31]); however, no specific the study. After the consent forms were returned, a timetable predictions were made for maltreatment history, psychiatric was circulated by the Deputy Head of the schools to teach- risk and affective functioning, as these domains have not ers in the participants’ class year, in order to (1) select slots been previously examined with Anxious vs CU+Anx groups. that would be the least disruptive to each participant’s class schedule; and (2) identify which teachers knew each partici- pant best and thus could be asked to fill in the questionnaire booklet after the testing session had taken place. Out of the Method participants who initially consented to take part in the study, 89.6% attended the agreed time slots and completed the testing Participants session. After the testing session, the teachers most familiar with each participant completed the questionnaire booklet. Of The current sample draws from a larger study (n  =  204) note, 88% of informants (i.e. key workers/teachers) reported examining the effects of developmental adversity on indi- knowing the participant well (i.e. ‘a little’ = 12%; ‘moder- vidual functioning amongst socially deprived youth aged ately well’ = 54%; ‘very well’ = 34%). Informed consent was 16–24 years (mean age 18 years). Of note, we refer to our obtained from all participants included in the study. Testing sample as ‘youth’, as it is (1) in line with the term used by took place in a quiet room within the charity or the young international organisations (e.g. UN) to describe individu- person’s school depending on recruitment source. Participants als aged 15–24; and (2) consistent with the extant litera- from the charity were compensated for their time individu- ture on variants of CU youth, which is primarily focused on ally; however, students recruited from schools received group youth populations (e.g. [17–20, 24, 25]). Only participants compensation for school equipment or a final year party in for whom information was available for both CU traits and line with head-teacher preferences. Additional details of the anxiety were included in the present study (n = 155). These recruitment procedures are available elsewhere [34]. youth were recruited via multiple channels in order to cap- ture varying exposure to adversity, including inner-city col- Measures leges (n = 71, 46%) and a charity providing services and support to vulnerable, self-referred youth (n = 84, 54%). Socio‑demographic characteristics Of the total sample, 80% of participants were under the age of 20 years (M = 18) and 54% were females (N = 84). The Data on age, sex, ethnicity and IQ were collected from sample was ethnically diverse, with 52% Caucasian, 42% all participants. Cognitive ability was assessed using the Black, 6% ‘Other’ participants. two-subtest version of the Wechsler Abbreviated Scale of 1 3 888 European Child & Adolescent Psychiatry (2018) 27:885–898 Intelligence (WASI; [35]), with all participants scoring Symptoms of depression, anger, post-traumatic stress and within the 70–125 range. Participant postcode information dissociation were assessed using the self-report clinical was used to obtain a census-derived and area-weighted Index scales from the TSCC-A, as described above (α  =  0.84– of Multiple Deprivation (IMD; [36]) score, an aggregate 0.87). In addition, informants completed six subscales from measure of neighbourhood deprivation. Higher values indi- the DSM-IV-based Adolescent Symptom Inventory (ASI- cate older age, female gender, non-white ethnicity, higher 4; [40]) to assess symptoms of emotional and behavioural cognitive ability and greater neighbourhood deprivation. disorders, including generalised anxiety disorder (GAD), major depressive disorder (MDD), oppositional defiant dis- Indicator variables order (ODD), conduct disorder (CD), antisocial personality disorder (ASPD) and attention-deficit hyperactivity disor - Callous–unemotional traits CU traits were measured using der (ADHD). Each scale contained between 7 and 9 items the well-validated Inventory of Callous Unemotional traits (α = 0.89–0.94). Each item is rated on a 3-point scale from (ICU; [37]), based on informant ratings (i.e. teachers or key ‘not true’ to ‘certainly true’. workers, depending on recruitment site). The ICU contains 24 items rated on a 4-point scale from ‘not at all true’ to Behavioural risk Multiple domains of behavioural risk- ‘definitely true’. The items cluster into three subscales, taking were assessed based on self-reported measures. Sub- which show adequate internal reliability in our sample: stance use was assessed via the Alcohol Use Disorders Iden- callous (α  =  0.79), uncaring (α  =  0.88), and unemotional tification Test (AUDIT; [41]) and the Drug Use Disorders (α  =  0.73). The total ICU score was used to identify CU Identification Test (DUDIT; [42]). The AUDIT and DUDIT groups (α = 0.79). include 10 and 11 items, respectively, measuring substance use, harmful use and symptoms of dependence. The first Anxiety Participants completed the anxiety subscale of the items are rated on a 5-point scale ranging from ‘never’ to Trauma Symptom Checklist for Children (TSCC-A; [38]). ‘daily or almost daily’. The last two items from each scale The TSCC-A is a 44-item self-report inventory that includes are rated on a 3-point scale and are coded as 0 (‘no’), 2 5 clinical scales (anxiety, depression, post-traumatic stress, (‘yes, but not during the last year’) or 4 (‘yes, during the anger and dissociation) and 2 validity scales (under- and last year’). Cronbach’s alphas for the AUDIT and DUDIT hyper-response). Each item is rated on a 4-point scale were 0.82 and 0.90, respectively. Participants were addition- from ‘never’ to ‘almost all of the time’. Of note, although ally administered three yes/no items from the Youth Risk the TSCC-A is designed to measure common sequelae of Behaviour Survey (YRBS; [43]). The first two items asked trauma exposure, the anxiety scale makes no reference to about suicidal ideation (“During the past 12 months, did you traumatic events. Rather, items tap into unspecific symp- ever seriously consider attempting suicide”) and attempted toms of general anxiety, such as “feeling afraid something suicide (“During the past 12 months, how many times did bad may happen”, “worrying about things” and “feeling you actually attempt suicide?”; originally rated on a 5-point nervous or jumpy inside” (9 items; α = 0.86). scale from ‘0 times’ to ‘6 or more times’ but collapsed due to low frequency of youth reporting multiple suicide Maltreatment history attempts). The third item asked about sexual safety (“The last time you had sexual intercourse, did you or your partner Participants completed the Childhood Trauma Questionnaire use a condom or other contraceptive?”). Participants who (CTQ; [39]), a widely used 28-item self-report measure reported not having had sexual intercourse were excluded screening for experiences of maltreatment “while growing from analysis of this item (n = 42). up”. Items are rated on a 5-point scale from ‘never true’ to ‘very often true’ (e.g. “people in my family hit me so hard Attachment style The Experiences in Close Relationships that it left me with bruises or marks”). The CTQ comprises Inventory (ECR; [44]) was used as a self-report measure five subscales measuring emotional abuse, physical abuse, of attachment. The ECR comprises of two 18-item scales, sexual abuse, emotional neglect and physical neglect. The Anxiety (e.g. “I worry about being abandoned”; α  =  0.92) scales show acceptable internal consistency in our sample and Avoidance (e.g. “I try to avoid getting to close to oth- (α = 0.70–0.97). Higher scores represent more severe expe- ers”; α  =  0.91). Here, we analysed categorical scores of rience of childhood maltreatment. attachment style derived using a median-based approach, consistent with Bartholomew and Horowitz’s model [37]. Markers of individual functioning Participants were defined as (1) Secure, if scoring below midpoint on both scales (30% of sample); (2) Anxious, if Psychiatric symptoms Psychiatric symptomatology was above midpoint on the Anxiety scale only (16%); (3) Avoid- assessed using both self- and informant-report measures. ant, if scoring above midpoint on the Avoidant scale only 1 3 European Child & Adolescent Psychiatry (2018) 27:885–898 889 (26%), and (4) Disorganised, if scoring above midpoint on did not correlate significantly when examined globally both (28%). (r = 0.03). Of note, average levels of CU across the sample (M = 23.21; median = 22, see Table 1) were comparable to Aec ff tive functioning Affective functioning was measured those observed in previous studies that have used the ICU via self-report ratings of irritability and alexithymia. The to cluster variants of CU youth in mixed-gender samples, Affective Reactivity Index (ARI; [45]) includes six items including community (e.g. M = 23.65; [25]) and juvenile rated on a 3-point scale (‘not true’ to ‘certainly true’) and offender populations (e.g. M  = 22.33; [23]). Compared to measures irritability over the past 6 months, including state- these studies (both of which employed self-reports), CU ments such as “easily annoyed by others” and “often lose levels across the variants identified here were slightly lower temper”. Items were summed to form a total score, with ade- (i.e. M  = 31.09 compared to 32.30 in [25] and 33.24 CU−Anx quate internal consistency (α  =  0.88). The fist factor from in [23]; M  = 31.30 compared to 33.62 in [25] and CU+Anx the Toronto Alexithymia Scale (TAS-F1; [46]) was used to 36.01 in [23]). Of note, the median-split approach used here assess difficulty in the ability to identify one’s own feelings makes it possible to compare the CU+Anx group to (1) a and to distinguish them from bodily sensations signalling CU−Anx group, who shows comparable levels of CU levels emotional arousal. The scale comprises 7 items rated on a but significantly lower levels of anxiety; and (2) an Anxious 5-point scale from ‘I strongly disagree’ to ‘I strongly agree’ group, who instead shows comparable levels of anxiety but (e.g. “when I am upset, I don’t know if I am sad, frightened, significantly lower levels of CU. Therefore, the method ena- or angry”; α = 0.89). bles one to characterise similarities and differences between youth who present with both high CU and anxiety vs those Statistical analysis who present with either one alone. Step 1: defining groups Step 2: group comparisons We disaggregated CU groups using a median-split approach, Group comparisons were performed using regression mod- which resulted in four categorical groups (see Fig. 1): (1) els, which differed depending on data distribution. Overdis- ‘Low’, if scoring below midpoint on both measures of CU persed count variables (maltreatment scores and substance and anxiety (23%, n = 36); (2) ‘Anxious’, if scoring above use variables) were analysed using negative binomial regres- midpoint on anxiety only (28%, n = 43); (3) ‘CU−Anx’, sions. Chi-square and logistic regressions were used for if scoring above midpoint on CU only (23%, n = 36); and categorical data (sex, ethnicity, attachment style, suicidal (4) ‘CU+Anx’ if scoring above midpoint on both measures ideation and attempt, unsafe sex). Linear regressions were of CU and Anxiety (26%, n = 40). This approach paral- used for all other variables (age, IMD, IQ, TSCC-A, ASI lels methods used in clinical assessments, which often rely and affective functioning). For each analysis, we first report on concrete cut-offs rather than categories achieved by main effect statistics from the Omnibus test (i.e. X statistic data-driven approaches (e.g. cluster analyses). In line with for negative binomial regressions and categorical data; F previous findings (e.g. [24]), CU and anxiety measures statistic for linear regressions). Pair-wise comparisons are then reported for all significant main effects, including effect sizes for significant pair-wise contrasts (odds ratio for nega- tive binomial regressions and categorical data; Hedge’s g for linear regressions). To correct for inflated alphas result- ing from multiple comparisons we set the alpha threshold at p <0.01. Analyses were performed on SPSS package v. 21 [47]. Results Descriptive statistics for socio-demographic variables are presented in Table 1. Groups did not differ across age, eth- nicity, IQ and IMD. The ratio of males to females signifi- cantly differed across groups, X (3,155) = 15.23, p < 0.01. Over half of youth in the CU+Anx group were females compared to one third in the CU−Anx group. The number Fig. 1 Study groups, including two variants of CU youth and two comparison groups of females also differed markedly between CU− Anx and 1 3 890 European Child & Adolescent Psychiatry (2018) 27:885–898 1 3 Table 1 Group comparisons on socio-demographic variables and maltreatment history Overall sample Low CU High CU Omnibus test Pair-wise contrasts (n = 155) Low (n = 36) Anxious CU−Anx CU+Anx CU+Anx vs. CU−Anx CU+Anx vs. (n = 43) (n = 36) (n = 40) Anxious Effect size [95% CI] Effect size [95% CI] Socio-demographics 2 a † Sex (% female) 54.2 52.8 74.4 30.6 55.0 X (3,155) = 15.23, p < 0.01 OR = 2.84 [1.13, 7.14] – Ethnicity 80:65:10 27:8:1 22:17:4 17:17:2 14:23:2 X (9,155) = 16.43, ns – – Age, M (SD) 18.48 (2.11) 18.03 (2.16) 18.81 (2.16) 18.67 (2.03) 18.38 (2.07) F (3,155) = 1.04, ns – – IMD, M (SD) 28.22 (10.99) 25.01 (10.10) 29.21 (12.54) 29.31 (10.55) 29.20 (10.28) F (3,155) = 1.36, ns – – IQ, M (SD) 99.47 (10.95) 100.66 (9.70) 99.02 (12.48) 101.06 (9.14) 97.42 (11.75) F (3,155) = 0.84, ns – – Indicator variables CU traits (total), M (SD) 23.21 (9.53) 14.83 (4.35) 16.13 (4.38) 31.09 (6.03) 31.30 (6.68) – – – Callousness 4.76 (3.71) 2.42 (1.63) 2.30 (1.32) 7.33 (3.55) 7.17 (3.88) – – – Uncaring 11.78 (5.35) 6.97 (3.40) 8.56 (4.29) 15.88 (3.07) 15.96 (2.96) – – – Unemotional 6.72 (2.67) 5.50 (1.75) 5.28 (2.40) 8.00 (2.34) 8.25 (2.59) – – – Anxiety, M (SD) 6.74 (5.24) 2.42 (1.48) 10.16 (4.32) 2.72 (1.47) 10.55 (4.94) – – – Maltreatment history 2 a,b (3,155) = 24.69, p < 0.001 *** OR = 2.43 [1.47, 4.02] – Emotional abuse, M (SD) 9.90 (5.08) 7.50 (2.83) 11.04 (5.23) 8.11 (4.31) 12.45 (5.68) X 2 b Physical abuse, M (SD) 8.03 (4.88) 6.08 (1.64) 8.42 (5.28) 6.83 (3.41) 10.45 (6.32) X (3,155) = 37.61, p < 0.001 *** OR = 3.09 [1.81, 5.30] – 2 a,b Sexual abuse, M (SD) 5.97 (3.29) 5.22 (0.90) 6.55 (4.33) 5.47 (1.83) 6.47 (4.16) X (3,155) = 27.29, p < 0.001 ** OR = 2.82 [1.38, 5.76] – 2 † † Emotional neglect, M 10.50 (4.76) 8.80 (3.54) 11.67 (5.07) 9.17 (4.18) 11.95 (5.19) X (3,155) = 9.44 OR = 1.68 [1.02, 2.75] – (SD) 2 a,b Physical neglect, M (SD) 7.39 (3.45) 6.39 (2.60) 7.88 (3.85) 6.23 (2.34) 8.75 (3.98) X (3,155) = 22.05, p < 0.001 *** OR = 3.01 [1.71, 5.29] – 2 a,b Total maltreatment, M 41.79 (17.23) 34.00 (8.99) 45.58 (18.93) 35.86 (12.65) 50.07 (19.87) X (3,155) = 25.86, p < 0.001 *** OR = 2.33 [1.46, 3.08] – (SD) N.B. Analyses control for sex. Ethnicity = White:Black:Other. Omnibus test and pair-wise contrast are not performed for group-dependent variables (i.e. CU and Anxiety). Maltreatment history analysed using negative binomial regression. CU vs. Low do not differ in level of maltreatment. For the sake of clarity, tables presented only provide in-depth statistics for the contrasts of great- est interest (‘CU+Anx’ vs. ‘CU−Anx’ and ‘CU+Anx’ vs. ‘Anxious’). More detailed information about the other contrasts is available upon request OR odds ratio, IMD Index of Multiple Deprivation, CU callous–unemotional  p < 0.05, ** p < 0.01, *** p < 0.001 CU−Anx vs Anxious contrast significant at p < 0.01 CU+Anx vs Low contrast significant at p < 0.01 European Child & Adolescent Psychiatry (2018) 27:885–898 891 Anxious youth (30.6 vs. 74.4% females). As a result, all informant-rated outcomes (i.e. GAD and MDD). As pre- analyses included sex as a covariate. dicted, the two variants did not differ from one another in externalising behaviours—showing comparable symptoms Maltreatment history of conduct disorder, oppositional defiant disorder, and anti- social personality disorder. Both CU groups scored signifi- Mean levels of maltreatment across groups are shown in cantly higher on these externalising problems compared to Fig. 2. The CU+Anx group and the Anxious group reported either the Anxious or Low comparison groups. Interestingly, comparably high levels of total maltreatment, which dif- CU+Anx youth differed significantly from all other groups fered significantly from the comparably low levels reported in levels of self-reported psychological distress (i.e. anger, by the CU−Anx and Low groups (Table 1). With regard to post-traumatic stress and dissociation) as well as informant- specific forms of maltreatment, severity was greater in the rated ADHD symptomatology—with differences being mod- CU+Anx group compared to the CU−Anx group on meas- erate to large across these domains. The CU−Anx and Low ures of emotional, physical and sexual abuse as well as groups showed comparably (low) levels of psychological physical neglect (p < 0.001), with marginal differences for distress. emotional neglect (p < 0.05). Across forms of maltreatment, the CU+Anx group did not differ from the Anxious group, Behavioural risk markers while the CU−Anx group did not differ from the Low group. There was no significant main effect of group on alcohol Individual functioning use. The CU+Anx group reported higher drug use than the CU−Anx group (p < 0.001, OR = 2.17) and Anxious Differences in individual functioning are presented in group (p < 0.01, OR = 2.18). Endorsement of behavioural Table 2. At a mean level, the CU+Anx group showed the risk items across groups related to suicidality and unsafe most severe psychiatric symptoms, poorest affective func- sex are graphically presented in Fig. 3b. Significant main tioning and greatest rates of behavioural risk and disorgan- effects were found for suicidal ideation, suicide attempt ised attachment compared to than any other group. All con- and unsafe sex. In the CU+Anx group, 33.3% of partici- trasts between the CU+Anx and Low group were significant pants reported having thought of committing suicide in (p < 0.01), except for alcohol use. the past year and 22.5% attempted suicide, compared to 14.3% ideation and 11% attempt in the CU−Anx group. Psychiatric symptoms Rates of suicidal ideation and attempt within the CU+Anx group were also considerably higher than within the Anx- The CU+Anx group reported significantly higher inter- ious and Low groups. In addition, of those who had sexual nalising symptoms compared to the CU−Anx group (see intercourse, more than half (64%) in the CU+Anx group Fig. 3a), based on both self-reported (i.e. depression) and reported not using a condom or other contraceptive dur- ing their last sexual encounter, compared to 34.5% in the CU−Anx group, 27.6% in the Anxious group and 22.7% in the Low group. Attachment style Attachment style differed significantly across groups, X (9,154) = 38.10, p < 0.001. As can be seen in Fig. 3c, the most striking difference relates to the proportions of secure vs disorganised attachment across groups. The CU+Anx group were predominantly characterised by dis- organised (45%) and avoidant attachment (32%) styles, with only 7.5% showing secure attachment, the lowest proportion relative to any other group. The Anxious group were predominantly characterised by disorganised (34%) and anxious attachment (31%) styles, with 19% show- ing secure attachment. In contrast 53% and 44% of the CU−Anx and Low groups, respectively, were classified as securely attached. Fig. 2 Mean levels of childhood maltreatment severity across groups 1 3 892 European Child & Adolescent Psychiatry (2018) 27:885–898 1 3 Table 2 Group comparisons on markers of individual functioning Overall sample Low CU High CU Omnibus test Pair-wise contrasts (n = 155) Low (n = 36) Anxious CU−Anx CU+Anx CU+Anx vs. CU−Anx CU+Anx vs. Anxious (n = 43) (n = 36) (n = 40) Effect size [95% CI] Effect size [95% CI] Psychiatric symptoms  Self-report a,b   Depression, M (SD) 6.55 (5.00) 3.33 (1.98) 9.00 (4.52) 3.42 (2.93) 9.63 (5.52) F (2,155) = 39.29, p < 0.001 *** g = 1.37 [0.87, 1.87] – a,b   Anger, M (SD) 7.65 (5.72) 4.44 (3.79) 8.02 (5.19) 5.19 (4.24) 12.33 (5.79) F (2,155) = 28.13, p < 0.001 *** g = 1.38 [0.88, 1.88] ** g = 0.78 [0.33, 1.22] a,b †   PTSD, M (SD) 9.77 (6.81) 4.44 (3.78) 11.74 (6.07) 6.31 (4.56) 15.58 (5.99) F (2,155) = 53.61, p < 0.001 *** g = 1.71 [1.19, 2.24] g = 0.63 [0.19, 1.07] a,b   Dissociation, M (SD) 9.41 (6.13) 6.11 (4.37) 10.53 (5.35) 5.56 (3.62) 14.65 (5.96) F (2,155) = 36.59, p < 0.001 *** g = 1.80 [1.27, 2.34] ** g = 0.72 [0.28, 1.17]  Informant-rated b,c   GAD, M (SD) 4.21 (4.17) 1.52 (1.71) 4.16 (4.62) 4.04 (3.62) 6.82 (4.2) F (2,154) = 18.54, p < 0.001 ** g = 0.70 [0.23, 1.16] ** g = 0.59 [0.15, 1.03] b,c   MDD, M (SD) 2.78 (3.74) 0.71 (1.45) 2.65 (3.38) 2.57 (3.00) 5.08 (4.91) F (2,151) = 13.54, p < 0.001 ** g = 0.60 [0.14, 1.06] ** g = 0.58 [0.14, 1.02] a,b,c   ODD, M (SD) 2.91 (4.21) 0.56 (1.48) 1.46 (2.47) 4.06 (4.01) 5.63 (5.59) F (2,152) = 10.10, p < 0.001 – *** g = 0.97 [0.51, 1.42] b,c   CD, M (SD) 1.38 (2.79) 0.11 (0.40) 0.56 (1.10) 1.83 (2.64) 3.08 (4.28) F (2,151) = 8.18, p < 0.001 – *** g = 0.81 [0.37, 1.26] b,c   ASPD, M (SD) 2.09 (3.50) 0.26 (0.82) 0.93 (1.72) 2.77 (3.89) 4.52 (4.55) F (2,151) = 14.54, p < 0.001 – *** g = 1.05 [0.59, 1.51] b,c   ADHD, M (SD) 7.85 (9.19) 2.14 (4.40) 5.14 (8.96) 9.40 (7.47) 14.73 (9.57) F (2,152) = 21.41, p < 0.001 ** g = 0.61 [0.57, 1.48] *** g = 1.03 [0.57, 1.48] Behavioural risk markers  Alcohol use, M (SD) 4.99 (5.31) 5.14 (4.65) 4.71 (4.88) 4.34 (4.20) 5.78 (7.15) X (3,150) = 1.38, ns – – 2 b  Drug use, M (SD) 3.49 (6.91) 1.89 (4.86) 2.69 (5.68) 3.48 (6.42) 5.97 (9.47) X (3,150) = 23.08, p < 0.001 *** OR = 2.17 [1.27, 3.71] ** OR = 2.18 [1.32, 3.60] 2 b †  Suicidal ideation (%) 15.1 0 12.0 14.3 33.3 X (3,152) = 16.84, p < 0.001 – OR = 3.70 [1.17, 11.65] 2 b  Suicide attempt (%) 10.3 0 7.0 11.0 22.5 X (3,152) = 11.10, p < 0.01 – – 2 † †  Unsafe sex (%) 37.1 22.7 27.6 34.5 64.0 X (3,105) = 10.32 OR = 3.38 [1.10, 10.35] ** OR = 4.67 [1.47, 14.79] Affective functioning  Irritability, M (SD) 4.09 (3.71) 2.51 (2.67) 4.48 (4.10) 2.82 (2.54) 6.21 (3.93) F (2,150) = 12.77, p < 0.001 *** g = 0.99 [0.52, 1.47] –  Alexithymia M (SD) 14.63 (6.38) 10.39 (3.42) 16.55 (6.68) 12.88 (5.13) 18.05 (6.59) F (2,151) = 21.49, p < 0.001 *** g = 0.85 [0.38, 1.323] – N.B. Analyses control for sex. Hedge’s g guidelines for effect size: g of 0.20 = small, 0.50 = medium, 0.80 = large GAD generalised anxiety disorder, MDD major depressive disorder, ODD oppositional defiant disorder, CD conduct disorder, ASPD antisocial personality disorder, ADHD attention-deficit hyperactivity disorder, OR odds ratio  p < 0.05, ** p < 0.01, *** p < 0.001 a b c CU−Anx vs Anxious contrast significant at least at p < 0.01;  CU+Anx vs Low contrast significant at least at p < .01;  CU−Anx vs Low significant at least at p < 0.01 European Child & Adolescent Psychiatry (2018) 27:885–898 893 Fig. 3 Group differences on levels of psychiatric symptomatology, c Attachment style classification across groups. GAD generalised behavioural risk and attachment style. a Standardised mean levels of anxiety disorder, MDD major depressive disorder, ODD oppositional self-report (TSCC-A; top-half) psychological distress and informant- defiant disorder, CD conduct disorder, ASPD antisocial personality report (ASI; bottom-half) psychiatric symptomatology across groups. disorder, ADHD attention-deficit hyperactivity disorder b Percentage of endorsement of behavioural risk items across groups. Aec ff tive functioning Discussion The two variants of CU youth differed significantly on both This study systematically characterised variants of CU youth measures of affective functioning, with the CU+ Anx group in a high-risk community sample. Specifically, we compared showing higher levels of irritability (p < 0.001, g = 0.99) youth who presented with similarly high levels of CU traits, and alexithymia (p  <  0.001, g  =  0.85). In contrast, the but different levels of co-occurring anxiety (i.e. CU+ Anx vs CU+Anx group did not differ from the Anxious group on CU−Anx) on maltreatment history, psychiatric symptoma- either measure of affective functioning. The CU−Anx group tology and broad markers of individual functioning. The use showed a profile of affective functioning similar to that of of multiple informants was a key strength of our study, with the Low group. multi-rated assessments used in both construction of CU groups as well as the examination of individual functioning Post hoc power analysis domains. We highlight here three main findings. First, youth with CU+Anx were characterised by more severe histories The sample size in our study is consistent with the extant of childhood abuse and neglect compared to CU−Anx youth. literature on variants of CU youth in high-risk samples Second, while variants of CU youth did not differ on lev - (e.g. [15, 19, 22]), whereby elevated rates of develop- els of externalising problems (e.g. oppositional defiant and mental adversity and psychiatric symptomatology result conduct disorder symptoms), the CU+Anx group presented in increased power to detect effects (i.e. as opposed to with significantly elevated levels of psychological distress general population samples). Nevertheless, we performed (i.e. depression, anger, dissociation and PTSD symptoms), a post hoc analysis to ensure that we were appropriately insecure attachment, affective dysregulation and behavioural powered for the analyses undertaken. Based on post hoc risk. Third, the inclusion of an Anxious comparison group G*Power calculations, with a sample size of n = 155, four revealed widespread similarities in trauma history and indi- groups and moderate-to-large effect sizes for all outcome vidual functioning between CU+Anx youth and those low in variables, we found that achieved power exceeded 0.85 CU but high in anxiety. Generally, CU+Anx youth seemed across analyses. to experience a ‘double hit’ of negative outcomes associated 1 3 894 European Child & Adolescent Psychiatry (2018) 27:885–898 with CU on the one hand, and anxiety on the other. They also unsafe sexual behaviours and adverse health outcomes [48], showed additional vulnerabilities compared to youth who and suggest the CU+Anx group is highly vulnerable across presented with either CU or Anxiety alone, including more multiple domains. severe feelings of anger and dissociation, elevated ADHD Our exploratory measures delineated additional dif- symptoms, greater drug use, engagement in unsafe sex and ferences across variants of CU youth in areas of affective higher suicide risk. Overall, the identification of distinct pat- functioning and attachment to close others. Elevated lev- terns of co-occurring psychiatric, emotional and behavioural els of irritability and anger in the CU+Anx group are con- markers associated with variants of CU youth have impor- sistent with the notion that this variant features increased tant and immediate clinical applications for informing risk emotional expression and reactivity [6, 17]. Furthermore, assessment and treatment formulation. attachment disorganisation, an established sequel of child- hood maltreatment [49], was found to be most common in Childhood maltreatment robustly discriminates youth with CU+Anx, while CU−Anx featured predominantly between variants of CU youth a secure attachment style. To our knowledge, this is the first study to have examined current patterns of attachment styles As hypothesised, childhood maltreatment emerged as a key across CU groups. Finally, increased levels of alexithymia factor discriminating variants of CU youth, with CU+Anx observed in CU+Anx (and Anxious youth) compared to youth reporting more severe trauma histories compared to CU−Anx youth may also reflect the developmental impact the CU−Anx group across all individual forms of abuse and of childhood maltreatment on emotional arousal and func- neglect. This finding is consistent with prior research that tioning. The finding related to alexithymia warrants further examined maltreatment as a global construct (or as part of investigation, as it may offer clues as to why individuals a wider adversity measure; e.g. [6, 14, 15]). While previous with CU+Anx share behavioural features with those with studies that have compared variants of CU youth on individ- CU−Anx (in other words, the present with a ‘behavioural ual forms of maltreatment (e.g. [16–18]) have shown some phenocopy’), yet appear emotionally reactive in a way that inconsistencies regarding which precise forms of maltreat- CU−Anx are not. High levels of alexithymia are associ- ment reliably differentiate CU+ Anx and CU−Anx groups, all ated with an inability to describe and identify emotions, have reported more pervasive maltreatment experiences in rather than an inability to experience emotional arousal. the CU+Anx group—which is broadly in line with our find- This means that although these individuals may experience ings. In contrast to previous studies, we additionally com- heightened affect in response to another person’s distress, pared maltreatment profiles against two comparison groups their ability to display socially appropriate responses may (i.e. Low and Anxious). While CU+Anx youth reported be compromised, leading them to appear callous and uncar- comparable levels of abuse and neglect to youth presenting ing. The finding that CU+ Anx reported the highest levels with high anxiety but low CU (i.e. the Anxious group) the of dissociative symptoms compared to any other group CU−Anx group did not differ in maltreatment history from may lend additional support for this hypothesis, as do prior those showing low CU and low anxiety (i.e. Low group). reports of lack of emotional ‘clarity’ within this group [28]. In contrast, adults with primary psychopathy and youth with CU+Anx indexes a particularly vulnerable group CU−Anx have been shown to be typically characterised by of individuals low emotional arousal to other people’s distress [17]. Youth with CU+Anx presented with the highest mean lev- CU+Anx youth share many similarities with Anxious els of psychological distress across all domains examined, youth in line with adult data on individuals who score high on psychopathy and anxiety [8] as well as youth data on CU The inclusion of two comparison groups enabled us to groups [6, 7, 14, 15]. Additionally, the CU+Anx group was compare variants of CU youth to low CU youth who also characterised by significantly elevated behavioural risk, vary in their levels of anxiety. Interestingly, we found that including increased drug use, feelings of suicidality and Anxious youth, albeit lower in levels of externalising prob- engagement in unsafe sex. Alarmingly, one third of youth lems, reported similar levels of childhood trauma, emo- in the CU+Anx group in this high-risk sample reported hav- tional difficulties and psychological distress to CU+Anx ing seriously considered committing suicide in the past year, youth. Consequently, an important question that emerged and almost one fourth reported attempting suicide. Rates of from the present data related to why some youth with a unsafe sex were also high in the CU+Anx group, with more history of trauma presented with both high levels of CU than half of youth reporting not using a condom or other and anxiety (i.e. CU+Anx) while others only present with contraceptive during their last sexual intercourse. These fig- high anxiety (i.e. Anxious group). One possibility is that ures are disturbing given the known associations between youth with CU+Anx have additional genetic vulnerability to 1 3 European Child & Adolescent Psychiatry (2018) 27:885–898 895 externalising disorders/impulsivity, as is suggested by their Limitations substance use, suicidal ideation and sexual behaviour pro- file. It is also possible that CU+ Anx youth may be exposed The findings of present study should be interpreted in light to additional environmental risk factors relative to Anxious of several limitations. First, CU traits are a dimensional con- youth, that were not captured in the current study (e.g. bul- struct, not a taxon. As we wished to compare variants of CU lying-victimisation). Longitudinal investigations charting traits, a categorical approach provided an effective means children who have experienced maltreatment, but who come of communication and this way of characterising children from families characterised by different levels of externalis- is also directly relevant for informing clinical practice. In ing problems, could shed light into this issue. future, studies may benefit from using dimensional infor - mation to supplement categorical approaches. Furthermore, Research and clinical implications although the measure used in our study to index CU traits (i.e. the ICU) has been commonly employed in the literature The present findings highlight the need to differentiate on variants of CU youth as well as being shown to possess between variants of CU youth. Supplementing measures of good factor structure, construct and predictive validity in CU traits with an assessment of anxiety can offer important a range of populations [52–56], some concerns have been information for both clinicians and researchers. Failure to raised about aspects of its psychometric properties [57] so assess levels of anxiety among youth with high CU traits that results will need to be replicated using an independent may obscure the diverse constellations of needs and risk measure of CU traits. Second, the anxiety measure used in factors associated with subgroups of individuals presenting this study to define groups was taken from the same ques- with elevated CU traits. Equally, the current findings high- tionnaire as our self-reported outcomes of psychological light that experiences of childhood maltreatment markedly distress, which raises issues of shared-method variance. differ between variants of CU youth. In research and clinical However, it is important to note that variants of CU youth settings, developmental adversity is not always assessed con- were also found to differ on levels of internalising problems currently with CU traits in youth [14]. An increased aware- (i.e. symptoms of generalised anxiety and major depressive ness of maltreatment as a possible risk factor for CU+Anx disorder) based on ratings from independent informants (i.e. may be helpful in informing risk assessment and suitable teachers/key workers). Third, while inclusion of a measure intervention strategies. Importantly, the findings indicate of childhood maltreatment provided a temporal proxy for that focussing on conduct problems or antisocial behaviour the ee ff ct of developmental adversity on CU +Anx, the cross- alone is unlikely to discriminate between variants of CU sectional nature of the study meant that we were unable to youth, as they tend to present similarly on these domains. establish the causality of effects found. However, the consist- Youth with CU+Anx represent a high-risk clinical group ency with which childhood maltreatment has been found to characterised by more severe developmental trauma, con- differentiate between variants of CU youth across our study current psychiatric symptomatology, affective dysfunction, and that of the extant literature (e.g. [14–18]) considerably risk behaviours and suicide risk. For these youths, therapeu- adds confidence to this finding. Despite this, it is important tic approaches that include the experience of trauma in the to note that while the data seem to suggest that CU+Anx treatment formulation, such as trauma-focussed CBT and may be more environmentally driven than CU−Anx, it was similar evidence-based interventions, may be warranted. not possible to remove potential genetic confounds from our Equally, interventions addressing conduct problems in youth design (e.g. youth high in CU may be more likely to have with CU+Anx may need embedding in a wider therapeutic parents high in psychopathic traits, who are also more likely intervention addressing other internalising problems, par- to maltreat them). Genetically informative designs may be ticularly anxiety and depression. High rates of disorganised particularly effective in examining the contribution of such attachment in this group are likely to predict poor interper- influences (e.g. [58]). Fourth, while post hoc analyses con- sonal functioning, and will be relevant to the clinician chal- firmed that we were appropriately powered for all analy - lenged with establishing appropriate boundaries alongside ses undertaken, sample size limitations meant that we were an effective therapeutic alliance. Finally, risk assessments only able to enter sex as a free-standing covariate. In future, will need to pay particular attention to engagement in risky the use of larger samples will make it possible to examine behaviours (e.g. drug use) and increased risk of suicidal- whether sex moderates associations between variants of CU ity as these were strongly associated with CU+Anx. More youth and markers of individual functioning. Finally, even broadly, our findings support a growing emphasis in the field though sampled from the community, youth in our study on CU traits as a cross-disorder construct [50, 51], which came predominantly from high-risk, multi-problem families. needs to be more fully considered within the broader context As a result, further research is needed to establish the extent of different forms of psychopathology and risk behaviours to which findings may generalise to the wider population. across both research and clinical settings. 1 3 896 European Child & Adolescent Psychiatry (2018) 27:885–898 to a greater understanding of the nature and significance Future directions of variants of CU youth. The present findings point to a number of directions for Acknowledgements The authors are indebted to the young people, future research. First, longitudinal, prospective research key-workers and teachers who have taken part in this study. This is needed to gain a more mechanistic understanding of research was supported by Kids Company charity. CC is supported by processes underlying variants of CU traits in youth. Lon- the Economic and Social Research Council (grant ref: ES/N001273/1). EV and EMC were supported by UK Medical Research Council grant gitudinal studies may also help determine whether variants (MR/ K014080/1) during the writing of this article. EV is a Royal are predictive of different developmental trajectories and Society Wolfson Research Merit Award Holder. outcomes over time, particularly in relation to frequency and nature of violence, suicidality, and mental health Compliance with ethical standards problems. Indeed, efforts to map variants longitudinally are already beginning to emerge [7, 23, 26, 27]. Second, Conflict of interest On behalf of all authors, the corresponding author examining the timing of maltreatment experiences may states that there is no conflict of interest. be important for understanding how CU+Anx develops Open Access This article is distributed under the terms of the Creative and identifying whether developmental windows exist Commons Attribution 4.0 International License (http://creativecom- where the effect of maltreatment is more pronounced. mons.org/licenses/by/4.0/), which permits unrestricted use, distribu- Third, CU+Anx may represent a ‘phenocopy’ of CU−Anx, tion, and reproduction in any medium, provided you give appropriate but the origins of CU and the underlying neurocognitive credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. mechanisms for the two variants may differ. A number of studies have provided support for differences in behav - ioural performance across variants on measures of emo- tional processing and behavioural activation [6, 16, 17, References 28]. Future neurocognitive studies would benefit from direct comparisons of CU+Anx with anxious individu- 1. Frick PJ, Ray JV, Thornton LC, Kahn RE (2014) Can callous–une- als, as well as use of tasks that investigate processes that motional traits enhance the understanding, diagnosis, and treat- should be compromised in CU, but not in anxiety. Fourth, ment of serious conduct problems in children and adolescents? A given that CU traits are known to be moderately associ- comprehensive review. Psychol Bull 140(1):1 2. Blair RJR, Leibenluft E, Pine DS (2014) Conduct disor- ated with conduct problems (e.g. r = 0.54 in our study) der and callous–unemotional traits in youth. N Engl J Med it will be important in future to establish to what extent 371(23):2207–2216 these co-occurring symptoms may be driving observed 3. Frick PJ, Viding E (2009) Antisocial behavior from a devel- differences between CU+ Anx vs Anx only youth. From opmental psychopathology perspective. Dev Psychopathol 21(Special Issue 04):1111–1131. https://doi.org/10.1017/ a research perspective, it is notable that we found group S0954579409990071 differences in the ratio of males to females across variants 4. de Wied M, van Boxtel A, Matthys W, Meeus W (2012) Verbal, of CU youth in this community sample. While the group facial and autonomic responses to empathy-eliciting film clips by of youth with CU−Anx contained disproportionately more disruptive male adolescents with high versus low callous–unemo- tional traits. J Abnorm Child Psychol 40(2):211–223 boys, the CU+Anx group had a more balanced male to 5. Sebastian CL, McCrory EJ, Cecil CA, Lockwood PL, De Brito female ratio (slightly greater number of girls). Moreover, SA, Fontaine NM, Viding E (2012) Neural responses to affective the Anxious group featured predominantly females. These and cognitive theory of mind in children with conduct problems findings are in line with previous work examining variants and varying levels of callous–unemotional traits. Arch Gen Psy- chiatry 69(8):814–822 of CU youth [31]. Interestingly, another study has reported 6. Kimonis ER, Frick PJ, Cauffman E, Goldweber A, Skeem J (2012) that psychopathic personality traits are associated with a Primary and secondary variants of juvenile psychopathy differ history of trauma in young female offenders [59]. Future in emotional processing. Dev Psychopathol 24(3):1091–1103. studies should test whether the difference in sex ratio is a https://doi.org/10.1017/s0954579412000557 7. Meehan AJ, Maughan B, Cecil CA, Barker ED (2017) Interper- reliable finding and whether the experience of trauma may sonal callousness and co-occurring anxiety: developmental valid- represent a particularly potent risk factor for CU+Anx in ity of an adolescent taxonomy. J Abnorm Psychol 126(2):225 girls. Finally, research is needed to inform the develop- 8. Skeem JL, Poythress N, Edens JF, Lilienfeld SO, Cale EM (2003) ment of more tailored interventions as well as to evalu- Psychopathic personality or personalities? Exploring potential variants of psychopathy and their implications for risk assessment. ate whether the application of differing strategies may be Aggress Violent Behav 8(5):513–546. https://doi.org/10.1016/ more effective than a ‘one size fits all’ intervention. This is S1359-1789(02)00098-8 especially important given the dearth of programmes spe- 9. Karpman B (1941) On the need of separating psychopathy into cifically validated on youth with CU traits [17]. Together, two distinct clinical types: the symptomatic and the idiopathic. J Crim Psychopathol 3:112–137 studies addressing these future directions will contribute 1 3 European Child & Adolescent Psychiatry (2018) 27:885–898 897 10. Cleckley H (1941) The mask of sanity: an attempt to reinterpret Personal Individ Differ 111(Supplement C):124–133. https://doi. the so-called psychopathic personality. Mosby, St. Louis, MO org/10.1016/j.paid.2017.02.005 11. Viding E, McCrory E (2015) Developmental risk for psychopathy. 28. Bennett DC, Kerig PK (2014) Investigating the construct of In: Thapar A, Pine DS, Leckman JF, Scott S, Snowling MJ, Taylor trauma-related acquired callousness among delinquent youth: E (eds) Rutter’s child and adolescent psychiatry. Wiley, New York, differences in emotion processing. J Trauma Stress 27(4):415–422 pp 966–980. https://doi.org/10.1002/9781118381953.ch68 29. Dackis MN, Rogosch FA, Cicchetti D (2015) Child maltreatment, 12. Viding E, Kimonis E (2017) Callous–unemotional traits in chil- callous–unemotional traits, and defensive responding in high-risk dren and youth. In: Patrick C (ed) Handbook of psychopathy, 2nd children: an investigation of emotion-modulated startle response. edn. The Guildford Press, New York (in press) Dev Psychopathol 27(4pt2):1527–1545 13. Cecil CA, Lysenko LJ, Jaffee SR, Pingault JB, Smith RG, Rel- 30. Kimonis ER, Goulter N, Hawes DJ, Wilbur RR, Groer MW (2017) ton CL, Woodward G, McArdle W, Mill J, Barker ED (2014) Neuroendocrine factors distinguish juvenile psychopathy variants. Environmental risk, Oxytocin Receptor Gene (OXTR) methyla- Dev Psychobiol 59(2):161–173 tion and youth callous–unemotional traits: a 13-year longitudinal 31. Fanti KA, Demetriou CA, Kimonis ER (2013) Variants of cal- study. Mol Psychiatry 19(10):1071–1077. https://doi.org/10.1038/ lous–unemotional conduct problems in a community sample of mp.2014.95 adolescents. J Youth Adolesc 42(7):964–979 14. Tatar JR, Cauffman E, Kimonis ER, Skeem JL (2012) Victimi- 32. Lee H-S, Barraza-Villarreal A, Biessy C, Duarte-Salles T, Sly PD, zation history and posttraumatic stress: an analysis of psychopa- Ramakrishnan U, Rivera J, Herceg Z, Romieu I (2014) Dietary thy variants in male juvenile offenders. J Child Adolesc Trauma supplementation with polyunsaturated fatty acid during pregnancy 5(2):102–113. https://doi.org/10.1080/19361521.2012.671794 modulates DNA methylation at IGF2/H19 imprinted genes and 15. Vaughn MG, Edens JF, Howard MO, Smith ST (2009) An growth of infants. Physiol Genom 46(23):851–857 investigation of primary and secondary psychopathy in a state- 33. Dadds MR, Kimonis ER, Schollar-Root O, Moul C, Hawes DJ wide sample of incarcerated youth. Youth Violence Juv Justice (2017) Are impairments in emotion recognition a core feature of 7(3):172–188. https://doi.org/10.1177/1541204009333792 callous–unemotional traits? Testing the primary versus secondary 16. Kahn RE, Frick PJ, Youngstrom EA, Kogos Youngstrom J, variants model in children. Dev Psychopathol 1–11. http://dx.doi. Feeny NC, Findling RL (2013) Distinguishing primary and org/10.1017/S0954579417000475 secondary variants of callous–unemotional traits among ado- 34. Cecil CA, Viding E, Barker ED, Guiney J, McCrory EJ (2014) lescents in a clinic-referred sample. Psychol Assess 25(3):966– Double disadvantage: the influence of childhood maltreatment 978. https://doi.org/10.1037/a0032880 and community violence exposure on adolescent mental health. J 17. Kimonis ER, Fanti KA, Goulter N, Hall J (2016) Affective star - Child Psychol Psychiatry. https://doi.org/10.1111/jcpp.12213 tle potentiation differentiates primary and secondary variants of 35. Wechsler D (1999) Wechsler abbreviated scale of intelligence. juvenile psychopathy. Dev Psychopathol 29(4):1149–1160 Psychological Corporation, San Antonio 18. Kimonis ER, Fanti KA, Isoma Z, Donoghue K (2013) Maltreat- 36. IMD (2011) The English indices of deprivation 2010. IMD, ment profiles among incarcerated boys with callous–unemo- London tional traits. Child Maltreat 18(2):108–121 37. Frick PJ (2004) The inventory of callous–unemotional traits. 19. Rosan A, Frick PJ, Gottlieb KA, Fasicaru L (2015) Callous–une- Unpublished rating scale, University of New Orleans, New motional traits and anxiety in a sample of detained adolescents Orleans, LA in Romania. J Evid Based Psychother 15(1):79 38. Briere J (1996) Trauma symptom checklist for children (TSCC). 20. Sharf A, Kimonis ER, Howard A (2014) Negative life events Psychological Assessment Resources, Lutz and posttraumatic stress disorder among incarcerated boys 39. Bernstein DP, Fink L (1998) Childhood trauma questionnaire: a with callous–unemotional traits. J Psychopathol Behav Assess retrospective self-report manual. Psychological Corporation, San 36(3):401–414. https://doi.org/10.1007/s10862-013-9404-z Antonio 21. Docherty M, Boxer P, Huesmann LR, O’Brien M, Bushman BJ 40. Gadow KD, Sprafkin JN (1998) Adolescent symptom inventory-4. (2016) Exploring primary and secondary variants of psychopa- Checkmate Plus, Stony Brook thy in adolescents in detention and in the community. J Clin 41. Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M Child Adolesc Psychol 45(5):564–578 (1993) Development of the Alcohol Use Disorders Identifica- 22. Gill AD, Stickle TR (2016) Affective differences between psy - tion Test (AUDIT): WHO Collaborative Project on early detec- chopathy variants and genders in adjudicated youth. J Abnorm tion of persons with harmful alcohol consumption-II. Addiction Child Psychol 44(2):295–307 88(6):791–804 23. Goulter N, Kimonis ER, Hawes SW, Stepp S, Hipwell AE (2017) 42. Berman A, Bergman H, Palmstierna T, Schlyter F (2003) DUDIT: Identifying stable variants of callous–unemotional traits: a lon- the drug use disorders identification test manual. Karolinska Insti- gitudinal study of at-risk girls. Dev Psychol 53(12):2364–2376 tutet, Department of Clinical Neuroscience, Stockholm 24. Kahn RE, Frick PJ, Golmaryami FN, Marsee MA (2017) The 43. Eaton DK, Kann L, Kinchen S, Shanklin S, Ross J, Hawkins J, moderating role of anxiety in the associations of callous– Harris WA, Lowry R, McManus T, Chyen D (2008) Youth risk unemotional traits with self-report and laboratory measures behavior surveillance—United States, 2007. MMWR Surveill of affective and cognitive empathy. J Abnorm Child Psychol Summ 57(4):1–131 45(3):583–596 44. Wei M, Russell DW, Mallinckrodt B, Vogel DL (2007) The Expe- 25. Lee Z, Salekin RT, Iselin A-MR (2010) Psychopathic traits in riences in Close Relationship Scale (ECR)-short form: reliability, youth: is there evidence for primary and secondary subtypes? J validity, and factor structure. J Personal Assess 88(2):187–204 Abnorm Child Psychol 38(3):381–393 45. Stringaris A, Goodman R, Ferdinando S, Razdan V, Muhrer 26. Fanti KA, Kimonis E (2017) Heterogeneity in externalizing prob- E, Leibenluft E, Brotman MA (2012) The Affective Reactivity lems at age 3: association with age 15 biological and environmen- Index: a concise irritability scale for clinical and research settings. tal outcomes. Dev Psychol 53(7):1230–1241 J Child Psychol Psychiatry 53(11):1109–1117 27. Ezpeleta L, Granero R, de la Osa N, Domènech JM (2017) Devel- 46. Bagby RM, Parker JD, Taylor GJ (1994) The twenty-item Toronto opmental trajectories of callous–unemotional traits, anxiety and Alexithymia Scale—I. Item selection and cross-validation of the oppositionality in 3–7 year-old children in the general population. factor structure. J Psychosom Res 38(1):23–32 1 3 898 European Child & Adolescent Psychiatry (2018) 27:885–898 47. IBM Corp (2012) IBM SPSS statistics for Windows, version 21.0. validation of the inventory of callous and unemotional traits. IBM Corp, Armonk Assessment 17(1):44–57 48. Coyle K, Basen-Engquist K, Kirby D, Parcel G, Banspach S, Col- 55. Ciucci E, Baroncelli A, Franchi M, Golmaryami FN, Frick PJ lins J, Baumler E, Carvajal S, Harrist R (2001) Safer choices: (2014) The association between callous–unemotional traits and reducing teen pregnancy, HIV, and STDs. Public Health Rep behavioral and academic adjustment in children: further valida- 116(Suppl 1):82 tion of the inventory of callous–unemotional traits. J Psycho- 49. Cyr C, Euser EM, Bakermans-Kranenburg MJ, Van Ijzendoorn pathol Behav Assess 36(2):189–200. https://doi.org/10.1007/ MH (2010) Attachment security and disorganization in maltreat- s10862-013-9384-z ing and high-risk families: a series of meta-analyses. Dev Psycho- 56. Henry J, Pingault JB, Boivin M, Rijsdijk F, Viding E (2016) pathol 22(01):87–108 Genetic and environmental aetiology of the dimensions of Cal- 50. Herpers PC, Rommelse NN, Bons DM, Buitelaar JK, Scheepers lous–Unemotional traits. Psychol Med 46(2):405–414 FE (2012) Callous–unemotional traits as a cross-disorders con- 57. Berg JM, Lilienfeld SO, Reddy SD, Latzman RD, Roose A, Craig- struct. Soc Psychiatry Psychiatr Epidemiol 47(12):2045–2064 head LW, Pace TW, Raison CL (2013) The inventory of callous 51. Salekin RT (2017) Research review: what do we know about and unemotional traits: a construct-validational analysis in an at- psychopathic traits in children? J Child Psychol Psychiatry risk sample. Assessment 20(5):532–544 58(11):1180–1200 58. Humayun S, Kahn RE, Frick PJ, Viding E (2014) Callous–unemo- 52. Kimonis ER, Frick PJ, Skeem JL, Marsee MA, Cruise K, Munoz tional traits and anxiety in a community sample of 7-year-olds. J LC, Aucoin KJ, Morris AS (2008) Assessing callous–unemotional Clin Child Adolesc Psychol 43(1):36–42 traits in adolescent offenders: validation of the Inventory of Cal- 59. Odgers CL, Reppucci ND, Moretti MM (2005) Nipping psychopa- lous–Unemotional Traits. Int J Law Psychiatry 31(3):241–252 thy in the bud: an examination of the convergent, predictive, and 53. Essau CA, Sasagawa S, Frick PJ (2006) Callous–unemotional theoretical utility of the PCL-YV among adolescent girls. Behav traits in a community sample of adolescents. Assessment Sci Law 23(6):743–763 13(4):454–469 54. Roose A, Bijttebier P, Decoene S, Claes L, Frick PJ (2010) Assess- ing the affective features of psychopathy in adolescence: a further 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Child & Adolescent Psychiatry Springer Journals

Characterising youth with callous–unemotional traits and concurrent anxiety: evidence for a high-risk clinical group

Download PDF
14 pages

Loading next page...
 
/lp/springer-journals/characterising-youth-with-callous-unemotional-traits-and-concurrent-gLA8FJ9LYA

References (69)

Publisher
Springer Journals
Copyright
Copyright © 2017 by The Author(s)
Subject
Medicine & Public Health; Psychiatry
ISSN
1018-8827
eISSN
1435-165X
DOI
10.1007/s00787-017-1086-8
pmid
29222633
Publisher site
See Article on Publisher Site

Abstract

Growing evidence supports the existence of two variants of youth with high callous–unemotional (CU) traits who present with markedly different risk profiles and outcomes, with potential implications for risk assessment and treatment formulation. So far, studies have identified variants of CU youth mainly using data-driven cluster approaches based on levels of CU traits and co-occurring anxiety. Yet, the extent to which this knowledge may be translated into clinical practice is unclear. To this end, the present study employed a severity-based, cut-off approach to systematically characterise CU groups across a range of clinically informative domains, including trauma history, psychiatric symptomatology, affective functioning, attachment style and behavioural risk. Analyses were based on multi-rated data from a community sample of high-risk youths (n = 155, M = 18 years). Consistent with previous studies, we found that, whereas variants show comparable levels of antisocial behaviour, those who present with both high CU and high anxiety report more severe childhood maltreatment, psychologi- cal distress, ADHD symptomatology and behavioural risk—including substance use, suicidal ideation and unsafe sex. In addition, these youth show greater attachment insecurity and affective dysregulation, as indexed by levels of irritability and alexithymia. Together, findings indicate that (1) trauma history is a key factor that differentiates variants of CU youth high vs. low on anxiety, and (2) differences in individual functioning across variants point to the need for tailored clinical assessment tools and intervention strategies. Importantly, the present findings indicate that variants of CU youth can be meaningfully differentiated using cut-off based approaches that parallel methods used in clinical assessments. Keywords Callous–unemotional traits · Anxiety · Childhood maltreatment · Psychopathology · Adolescence Introduction callousness and shallow affect [ 2]. Compared to other anti- social youth, those with high CU traits show marked dif- In the DSM-5, callous–unemotional (CU) traits—referred to ferences in neurocognitive, emotional and behavioural as ‘Limited Prosocial Emotions’—feature as a new diagnos- functioning, including difficulties in social-information pro - tic specifier for conduct disorder, to enable the identification cessing [3], under-arousal to empathy-inducing stimuli [4], of a particularly severe subgroup of youth at increased risk disruptions in affective theory of mind [ 5], lower sensitiv- for early-onset and persistent antisocial behaviour [1]. CU ity to punishment [1] as well as alterations in brain regions traits are defined by a core set of affective features (paral - involved in emotion and learning (e.g. amygdala, PFC; [2]). leling the affective dimension of adult psychopathy), which Together, these features are thought to contribute to the more include low capacity for empathy, lack of guilt and remorse, violent, chronic and recidivistic pattern of antisocial behav- iour displayed by youth with high CU traits, and represent an important target for intervention. * Essi Viding It is also becoming increasingly clear that not all youth [email protected] with high CU traits are the same. Rather, they can present Department of Psychology, Institute of Psychiatry, King’s with different levels of co-occurring anxiety [ 6, 7]. This is College London, London, UK akin to what has been observed in adults with psychopathy Division of Psychology and Language Sciences, University [8] and is thought to reflect the existence of two variants with College London, 26 Bedford Way, London WC1H 0AP, UK potentially distinct aetiologies—a theory first put forward Department of Psychology, Royal Holloway, University by Karpman [9] and Cleckley [10] in 1941. Specifically, of London, Egham, UK Vol.:(0123456789) 1 3 886 European Child & Adolescent Psychiatry (2018) 27:885–898 CU traits accompanied by low levels of anxiety (CU−Anx suicidal ideation, thought disturbance) compared to both the variant) are thought to be associated with substantial devel- CU−Anx and control group. In another study based on male opmental genetic risk, whereas CU traits accompanied by juvenile offenders (n  = 238, age 14–19 years), Sharf and high levels of anxiety (CU+Anx variant) are thought to be colleagues [20] used a median-split approach to create the associated primarily with environmental trauma [11, 12]. same three groups (i.e. CU+Anx, CU−Anx, control group) The two variants are indistinguishable based on CU traits and found that the CU+Anx group reported greater exposure alone (i.e. they can be thought of as ‘behavioural pheno- to negative life events (especially violence exposure in the copies’), but the CU+Anx variant is associated with more home and community) as well more severe post-traumatic severe pre- [7, 13] and post-natal [14, 15] adversity, with symptoms compared to the other two groups. While promis- the most consistent evidence relating to childhood maltreat- ing, these studies have focussed exclusively on male youth ment [16–20]. Furthermore, variants have been shown to offenders, so that more work is needed to test whether sever - differ markedly in presentation across a range of domains, ity-based approaches can meaningfully differentiate variants including comorbid psychiatric symptomatology [7, 14, 15, in non-forensic, multi-gender populations across a wider 21, 22], impulsivity [19], self-control [23], empathy [24], range of clinically informative domains. personality traits [25], expression of aggression [26], nega- A further question with important clinical implications tive affect [20], emotional lability [27], emotional process- is how CU+Anx youth compare not only to their CU−Anx ing [6, 17, 28, 29], behavioural risk [15, 18] and biological counterparts, but also to youth who present with high anxi- function [13, 26, 30]. ety alone (Anxious group). Contrasting these two groups is Given that CU levels are currently used to inform risk necessary in order to clarify whether (1) CU+Anx youth assessment and treatment options with antisocial youth [1], experience a ‘double hit’ of negative outcomes associated the existence of variants may carry important implications with two relatively independent dimensions of psychopa- for clinical practice [17]. So far, studies have primarily iden- thology; or (2) whether the combination of high CU and tified variants of CU youth using state-of-the-art clustering Anxiety indexes a particularly high-risk group of youth who approaches, which are hypothesis-free and person-centred show additional vulnerabilities compared to those who pre- [6, 7, 14–18, 26–28, 30–32]. Together, these reports have sent with either CU or Anxiety alone. Because studies to been invaluable in demonstrating that individuals naturally date (both cluster-based and severity-based) have generally cluster into groups based on their on levels of CU and anxi- contrasted variants of CU youth to a single, generic com- ety—providing strong, data-driven evidence validating the parison group (i.e. not disaggregated by level of anxiety), existence of two variants of CU youth. However, cluster- it has not been possible to systematically address this ques- ing approaches are not practical in clinical settings, where tion. To our knowledge, only one study based on a com- treatment and risk assessment decisions are typically based munity sample of adolescents has compared the CU+Anx on variable-centred, severity-based thresholds. A handful group to a reference group who show comparable levels of of other studies have examined these traits continuously, as anxiety [31]. Interestingly, the authors reported that although opposed to comparing groups, in order to establish whether the Anxious group displayed lower levels of CU traits and presence of anxiety or trauma history moderates the associa- antisocial behaviour compared to the CU+Anx variant, the tion between CU traits and outcomes, such as empathy [24] groups presented similarly in other domains, such as low or emotional recognition [33]. While such an approach has self-esteem. Furthermore, the Anxious group consisted pri- the advantage of modelling the full range of scores, lending marily of girls, which may explain the failure to identify useful insights into the dimensional relationship between this subgroup in prior studies that have typically focused CU and anxiety, it is particularly difficult to implement in a on juvenile male offender samples. The study, however, did clinical setting. not compare groups on trauma history, psychiatric risk, and As an alternative, a small set of studies have shown that affective functioning—key clinical domains that need sys- simpler cut-off approaches (e.g. based on average scores) tematic investigation if we are to more fully understand the can be successfully employed to compare variants on spe- nature of the CU+Anx variant. cific outcomes, yielding results that are consistent with those derived from cluster-based approaches. For example, in a Romanian sample of incarcerated males (n = 125, age The present study 14–18 years), Rosan and colleagues [19] used the sample average score of CU traits and anxiety as a cut-off thresh - The aim of the present study was to comprehensively charac- old to classify youth as either CU+Anx, CU−Anx or a terise variants of CU traits in a community sample of high- control group low on both dimensions. The authors found risk youth. Specifically, we investigated whether variants that the CU+Anx group showed significantly higher lev - of CU youth (i.e. CU−Anx vs CU+Anx) identified using els of impulsivity and emotional dysregulation (e.g. anger, a variable-centred, median-based approach differ across: 1 3 European Child & Adolescent Psychiatry (2018) 27:885–898 887 (1) previously validated domains, including childhood mal- Procedure treatment history, (multi-rated) psychiatric symptoms, and behavioural risk markers; as well as (2) novel functional All procedures performed were in accordance with the ethical domains, including attachment style and affective function- standards of the UCL Research Ethics Committee (ID No. ing (irritability and alexithymia). To improve the specificity 2462/001) and with the 1964 Helsinki declaration and its later of any conclusions about these groups we also compared amendments or comparable ethical standards. Youth from the both CU variant groups with two clinically relevant compari- charity were introduced to the research by a member of staff, son groups a Low group (low on both CU and anxiety) and and, if interested, were provided information about the study an Anxious group (low on CU but high in anxiety). Based by one of the research team on site. As a result, all youth who on previous studies using cluster-based analyses as well as met with the researchers had shown interest in the study and those that have used severity-based cut-offs, we predicted agreed to participate. After the testing session, each partici- that, relative to youth only high in CU traits (CU−Anx), pant’s key worker completed a questionnaire booklet. A key those with high CU and high anxiety (CU+Anx) would be worker is a member of staff of the charity who is assigned to characterised by: (1) more severe experiences of childhood each client upon referral in order to assist in the delivery of maltreatment; (2) greater levels of psychological distress services as well as to provide socio-emotional and practical and psychiatric symptomatology; (3) significantly elevated support. In schools, youth initially received information during behavioural risk markers; but (4) similar levels of externalis- a brief presentation at a school assembly. Information sheets ing problems. Given the lack of prior research, no a priori and consent forms were then distributed to students who had hypotheses were made regarding associations with attach- attended the presentation. Those students who were interested ment style or affective functioning (as indexed by levels of in taking part completed the consent form and returned it to irritability and alexithymia) between variants of CU youth. the researchers. As a result, researchers met exclusively with Compared to the Anxious group, we expected that CU+Anx students who were interested in participating and had provided youth would show higher levels of externalising problems informed consent stating that they were willing to take part in (in line with previous studies [31]); however, no specific the study. After the consent forms were returned, a timetable predictions were made for maltreatment history, psychiatric was circulated by the Deputy Head of the schools to teach- risk and affective functioning, as these domains have not ers in the participants’ class year, in order to (1) select slots been previously examined with Anxious vs CU+Anx groups. that would be the least disruptive to each participant’s class schedule; and (2) identify which teachers knew each partici- pant best and thus could be asked to fill in the questionnaire booklet after the testing session had taken place. Out of the Method participants who initially consented to take part in the study, 89.6% attended the agreed time slots and completed the testing Participants session. After the testing session, the teachers most familiar with each participant completed the questionnaire booklet. Of The current sample draws from a larger study (n  =  204) note, 88% of informants (i.e. key workers/teachers) reported examining the effects of developmental adversity on indi- knowing the participant well (i.e. ‘a little’ = 12%; ‘moder- vidual functioning amongst socially deprived youth aged ately well’ = 54%; ‘very well’ = 34%). Informed consent was 16–24 years (mean age 18 years). Of note, we refer to our obtained from all participants included in the study. Testing sample as ‘youth’, as it is (1) in line with the term used by took place in a quiet room within the charity or the young international organisations (e.g. UN) to describe individu- person’s school depending on recruitment source. Participants als aged 15–24; and (2) consistent with the extant litera- from the charity were compensated for their time individu- ture on variants of CU youth, which is primarily focused on ally; however, students recruited from schools received group youth populations (e.g. [17–20, 24, 25]). Only participants compensation for school equipment or a final year party in for whom information was available for both CU traits and line with head-teacher preferences. Additional details of the anxiety were included in the present study (n = 155). These recruitment procedures are available elsewhere [34]. youth were recruited via multiple channels in order to cap- ture varying exposure to adversity, including inner-city col- Measures leges (n = 71, 46%) and a charity providing services and support to vulnerable, self-referred youth (n = 84, 54%). Socio‑demographic characteristics Of the total sample, 80% of participants were under the age of 20 years (M = 18) and 54% were females (N = 84). The Data on age, sex, ethnicity and IQ were collected from sample was ethnically diverse, with 52% Caucasian, 42% all participants. Cognitive ability was assessed using the Black, 6% ‘Other’ participants. two-subtest version of the Wechsler Abbreviated Scale of 1 3 888 European Child & Adolescent Psychiatry (2018) 27:885–898 Intelligence (WASI; [35]), with all participants scoring Symptoms of depression, anger, post-traumatic stress and within the 70–125 range. Participant postcode information dissociation were assessed using the self-report clinical was used to obtain a census-derived and area-weighted Index scales from the TSCC-A, as described above (α  =  0.84– of Multiple Deprivation (IMD; [36]) score, an aggregate 0.87). In addition, informants completed six subscales from measure of neighbourhood deprivation. Higher values indi- the DSM-IV-based Adolescent Symptom Inventory (ASI- cate older age, female gender, non-white ethnicity, higher 4; [40]) to assess symptoms of emotional and behavioural cognitive ability and greater neighbourhood deprivation. disorders, including generalised anxiety disorder (GAD), major depressive disorder (MDD), oppositional defiant dis- Indicator variables order (ODD), conduct disorder (CD), antisocial personality disorder (ASPD) and attention-deficit hyperactivity disor - Callous–unemotional traits CU traits were measured using der (ADHD). Each scale contained between 7 and 9 items the well-validated Inventory of Callous Unemotional traits (α = 0.89–0.94). Each item is rated on a 3-point scale from (ICU; [37]), based on informant ratings (i.e. teachers or key ‘not true’ to ‘certainly true’. workers, depending on recruitment site). The ICU contains 24 items rated on a 4-point scale from ‘not at all true’ to Behavioural risk Multiple domains of behavioural risk- ‘definitely true’. The items cluster into three subscales, taking were assessed based on self-reported measures. Sub- which show adequate internal reliability in our sample: stance use was assessed via the Alcohol Use Disorders Iden- callous (α  =  0.79), uncaring (α  =  0.88), and unemotional tification Test (AUDIT; [41]) and the Drug Use Disorders (α  =  0.73). The total ICU score was used to identify CU Identification Test (DUDIT; [42]). The AUDIT and DUDIT groups (α = 0.79). include 10 and 11 items, respectively, measuring substance use, harmful use and symptoms of dependence. The first Anxiety Participants completed the anxiety subscale of the items are rated on a 5-point scale ranging from ‘never’ to Trauma Symptom Checklist for Children (TSCC-A; [38]). ‘daily or almost daily’. The last two items from each scale The TSCC-A is a 44-item self-report inventory that includes are rated on a 3-point scale and are coded as 0 (‘no’), 2 5 clinical scales (anxiety, depression, post-traumatic stress, (‘yes, but not during the last year’) or 4 (‘yes, during the anger and dissociation) and 2 validity scales (under- and last year’). Cronbach’s alphas for the AUDIT and DUDIT hyper-response). Each item is rated on a 4-point scale were 0.82 and 0.90, respectively. Participants were addition- from ‘never’ to ‘almost all of the time’. Of note, although ally administered three yes/no items from the Youth Risk the TSCC-A is designed to measure common sequelae of Behaviour Survey (YRBS; [43]). The first two items asked trauma exposure, the anxiety scale makes no reference to about suicidal ideation (“During the past 12 months, did you traumatic events. Rather, items tap into unspecific symp- ever seriously consider attempting suicide”) and attempted toms of general anxiety, such as “feeling afraid something suicide (“During the past 12 months, how many times did bad may happen”, “worrying about things” and “feeling you actually attempt suicide?”; originally rated on a 5-point nervous or jumpy inside” (9 items; α = 0.86). scale from ‘0 times’ to ‘6 or more times’ but collapsed due to low frequency of youth reporting multiple suicide Maltreatment history attempts). The third item asked about sexual safety (“The last time you had sexual intercourse, did you or your partner Participants completed the Childhood Trauma Questionnaire use a condom or other contraceptive?”). Participants who (CTQ; [39]), a widely used 28-item self-report measure reported not having had sexual intercourse were excluded screening for experiences of maltreatment “while growing from analysis of this item (n = 42). up”. Items are rated on a 5-point scale from ‘never true’ to ‘very often true’ (e.g. “people in my family hit me so hard Attachment style The Experiences in Close Relationships that it left me with bruises or marks”). The CTQ comprises Inventory (ECR; [44]) was used as a self-report measure five subscales measuring emotional abuse, physical abuse, of attachment. The ECR comprises of two 18-item scales, sexual abuse, emotional neglect and physical neglect. The Anxiety (e.g. “I worry about being abandoned”; α  =  0.92) scales show acceptable internal consistency in our sample and Avoidance (e.g. “I try to avoid getting to close to oth- (α = 0.70–0.97). Higher scores represent more severe expe- ers”; α  =  0.91). Here, we analysed categorical scores of rience of childhood maltreatment. attachment style derived using a median-based approach, consistent with Bartholomew and Horowitz’s model [37]. Markers of individual functioning Participants were defined as (1) Secure, if scoring below midpoint on both scales (30% of sample); (2) Anxious, if Psychiatric symptoms Psychiatric symptomatology was above midpoint on the Anxiety scale only (16%); (3) Avoid- assessed using both self- and informant-report measures. ant, if scoring above midpoint on the Avoidant scale only 1 3 European Child & Adolescent Psychiatry (2018) 27:885–898 889 (26%), and (4) Disorganised, if scoring above midpoint on did not correlate significantly when examined globally both (28%). (r = 0.03). Of note, average levels of CU across the sample (M = 23.21; median = 22, see Table 1) were comparable to Aec ff tive functioning Affective functioning was measured those observed in previous studies that have used the ICU via self-report ratings of irritability and alexithymia. The to cluster variants of CU youth in mixed-gender samples, Affective Reactivity Index (ARI; [45]) includes six items including community (e.g. M = 23.65; [25]) and juvenile rated on a 3-point scale (‘not true’ to ‘certainly true’) and offender populations (e.g. M  = 22.33; [23]). Compared to measures irritability over the past 6 months, including state- these studies (both of which employed self-reports), CU ments such as “easily annoyed by others” and “often lose levels across the variants identified here were slightly lower temper”. Items were summed to form a total score, with ade- (i.e. M  = 31.09 compared to 32.30 in [25] and 33.24 CU−Anx quate internal consistency (α  =  0.88). The fist factor from in [23]; M  = 31.30 compared to 33.62 in [25] and CU+Anx the Toronto Alexithymia Scale (TAS-F1; [46]) was used to 36.01 in [23]). Of note, the median-split approach used here assess difficulty in the ability to identify one’s own feelings makes it possible to compare the CU+Anx group to (1) a and to distinguish them from bodily sensations signalling CU−Anx group, who shows comparable levels of CU levels emotional arousal. The scale comprises 7 items rated on a but significantly lower levels of anxiety; and (2) an Anxious 5-point scale from ‘I strongly disagree’ to ‘I strongly agree’ group, who instead shows comparable levels of anxiety but (e.g. “when I am upset, I don’t know if I am sad, frightened, significantly lower levels of CU. Therefore, the method ena- or angry”; α = 0.89). bles one to characterise similarities and differences between youth who present with both high CU and anxiety vs those Statistical analysis who present with either one alone. Step 1: defining groups Step 2: group comparisons We disaggregated CU groups using a median-split approach, Group comparisons were performed using regression mod- which resulted in four categorical groups (see Fig. 1): (1) els, which differed depending on data distribution. Overdis- ‘Low’, if scoring below midpoint on both measures of CU persed count variables (maltreatment scores and substance and anxiety (23%, n = 36); (2) ‘Anxious’, if scoring above use variables) were analysed using negative binomial regres- midpoint on anxiety only (28%, n = 43); (3) ‘CU−Anx’, sions. Chi-square and logistic regressions were used for if scoring above midpoint on CU only (23%, n = 36); and categorical data (sex, ethnicity, attachment style, suicidal (4) ‘CU+Anx’ if scoring above midpoint on both measures ideation and attempt, unsafe sex). Linear regressions were of CU and Anxiety (26%, n = 40). This approach paral- used for all other variables (age, IMD, IQ, TSCC-A, ASI lels methods used in clinical assessments, which often rely and affective functioning). For each analysis, we first report on concrete cut-offs rather than categories achieved by main effect statistics from the Omnibus test (i.e. X statistic data-driven approaches (e.g. cluster analyses). In line with for negative binomial regressions and categorical data; F previous findings (e.g. [24]), CU and anxiety measures statistic for linear regressions). Pair-wise comparisons are then reported for all significant main effects, including effect sizes for significant pair-wise contrasts (odds ratio for nega- tive binomial regressions and categorical data; Hedge’s g for linear regressions). To correct for inflated alphas result- ing from multiple comparisons we set the alpha threshold at p <0.01. Analyses were performed on SPSS package v. 21 [47]. Results Descriptive statistics for socio-demographic variables are presented in Table 1. Groups did not differ across age, eth- nicity, IQ and IMD. The ratio of males to females signifi- cantly differed across groups, X (3,155) = 15.23, p < 0.01. Over half of youth in the CU+Anx group were females compared to one third in the CU−Anx group. The number Fig. 1 Study groups, including two variants of CU youth and two comparison groups of females also differed markedly between CU− Anx and 1 3 890 European Child & Adolescent Psychiatry (2018) 27:885–898 1 3 Table 1 Group comparisons on socio-demographic variables and maltreatment history Overall sample Low CU High CU Omnibus test Pair-wise contrasts (n = 155) Low (n = 36) Anxious CU−Anx CU+Anx CU+Anx vs. CU−Anx CU+Anx vs. (n = 43) (n = 36) (n = 40) Anxious Effect size [95% CI] Effect size [95% CI] Socio-demographics 2 a † Sex (% female) 54.2 52.8 74.4 30.6 55.0 X (3,155) = 15.23, p < 0.01 OR = 2.84 [1.13, 7.14] – Ethnicity 80:65:10 27:8:1 22:17:4 17:17:2 14:23:2 X (9,155) = 16.43, ns – – Age, M (SD) 18.48 (2.11) 18.03 (2.16) 18.81 (2.16) 18.67 (2.03) 18.38 (2.07) F (3,155) = 1.04, ns – – IMD, M (SD) 28.22 (10.99) 25.01 (10.10) 29.21 (12.54) 29.31 (10.55) 29.20 (10.28) F (3,155) = 1.36, ns – – IQ, M (SD) 99.47 (10.95) 100.66 (9.70) 99.02 (12.48) 101.06 (9.14) 97.42 (11.75) F (3,155) = 0.84, ns – – Indicator variables CU traits (total), M (SD) 23.21 (9.53) 14.83 (4.35) 16.13 (4.38) 31.09 (6.03) 31.30 (6.68) – – – Callousness 4.76 (3.71) 2.42 (1.63) 2.30 (1.32) 7.33 (3.55) 7.17 (3.88) – – – Uncaring 11.78 (5.35) 6.97 (3.40) 8.56 (4.29) 15.88 (3.07) 15.96 (2.96) – – – Unemotional 6.72 (2.67) 5.50 (1.75) 5.28 (2.40) 8.00 (2.34) 8.25 (2.59) – – – Anxiety, M (SD) 6.74 (5.24) 2.42 (1.48) 10.16 (4.32) 2.72 (1.47) 10.55 (4.94) – – – Maltreatment history 2 a,b (3,155) = 24.69, p < 0.001 *** OR = 2.43 [1.47, 4.02] – Emotional abuse, M (SD) 9.90 (5.08) 7.50 (2.83) 11.04 (5.23) 8.11 (4.31) 12.45 (5.68) X 2 b Physical abuse, M (SD) 8.03 (4.88) 6.08 (1.64) 8.42 (5.28) 6.83 (3.41) 10.45 (6.32) X (3,155) = 37.61, p < 0.001 *** OR = 3.09 [1.81, 5.30] – 2 a,b Sexual abuse, M (SD) 5.97 (3.29) 5.22 (0.90) 6.55 (4.33) 5.47 (1.83) 6.47 (4.16) X (3,155) = 27.29, p < 0.001 ** OR = 2.82 [1.38, 5.76] – 2 † † Emotional neglect, M 10.50 (4.76) 8.80 (3.54) 11.67 (5.07) 9.17 (4.18) 11.95 (5.19) X (3,155) = 9.44 OR = 1.68 [1.02, 2.75] – (SD) 2 a,b Physical neglect, M (SD) 7.39 (3.45) 6.39 (2.60) 7.88 (3.85) 6.23 (2.34) 8.75 (3.98) X (3,155) = 22.05, p < 0.001 *** OR = 3.01 [1.71, 5.29] – 2 a,b Total maltreatment, M 41.79 (17.23) 34.00 (8.99) 45.58 (18.93) 35.86 (12.65) 50.07 (19.87) X (3,155) = 25.86, p < 0.001 *** OR = 2.33 [1.46, 3.08] – (SD) N.B. Analyses control for sex. Ethnicity = White:Black:Other. Omnibus test and pair-wise contrast are not performed for group-dependent variables (i.e. CU and Anxiety). Maltreatment history analysed using negative binomial regression. CU vs. Low do not differ in level of maltreatment. For the sake of clarity, tables presented only provide in-depth statistics for the contrasts of great- est interest (‘CU+Anx’ vs. ‘CU−Anx’ and ‘CU+Anx’ vs. ‘Anxious’). More detailed information about the other contrasts is available upon request OR odds ratio, IMD Index of Multiple Deprivation, CU callous–unemotional  p < 0.05, ** p < 0.01, *** p < 0.001 CU−Anx vs Anxious contrast significant at p < 0.01 CU+Anx vs Low contrast significant at p < 0.01 European Child & Adolescent Psychiatry (2018) 27:885–898 891 Anxious youth (30.6 vs. 74.4% females). As a result, all informant-rated outcomes (i.e. GAD and MDD). As pre- analyses included sex as a covariate. dicted, the two variants did not differ from one another in externalising behaviours—showing comparable symptoms Maltreatment history of conduct disorder, oppositional defiant disorder, and anti- social personality disorder. Both CU groups scored signifi- Mean levels of maltreatment across groups are shown in cantly higher on these externalising problems compared to Fig. 2. The CU+Anx group and the Anxious group reported either the Anxious or Low comparison groups. Interestingly, comparably high levels of total maltreatment, which dif- CU+Anx youth differed significantly from all other groups fered significantly from the comparably low levels reported in levels of self-reported psychological distress (i.e. anger, by the CU−Anx and Low groups (Table 1). With regard to post-traumatic stress and dissociation) as well as informant- specific forms of maltreatment, severity was greater in the rated ADHD symptomatology—with differences being mod- CU+Anx group compared to the CU−Anx group on meas- erate to large across these domains. The CU−Anx and Low ures of emotional, physical and sexual abuse as well as groups showed comparably (low) levels of psychological physical neglect (p < 0.001), with marginal differences for distress. emotional neglect (p < 0.05). Across forms of maltreatment, the CU+Anx group did not differ from the Anxious group, Behavioural risk markers while the CU−Anx group did not differ from the Low group. There was no significant main effect of group on alcohol Individual functioning use. The CU+Anx group reported higher drug use than the CU−Anx group (p < 0.001, OR = 2.17) and Anxious Differences in individual functioning are presented in group (p < 0.01, OR = 2.18). Endorsement of behavioural Table 2. At a mean level, the CU+Anx group showed the risk items across groups related to suicidality and unsafe most severe psychiatric symptoms, poorest affective func- sex are graphically presented in Fig. 3b. Significant main tioning and greatest rates of behavioural risk and disorgan- effects were found for suicidal ideation, suicide attempt ised attachment compared to than any other group. All con- and unsafe sex. In the CU+Anx group, 33.3% of partici- trasts between the CU+Anx and Low group were significant pants reported having thought of committing suicide in (p < 0.01), except for alcohol use. the past year and 22.5% attempted suicide, compared to 14.3% ideation and 11% attempt in the CU−Anx group. Psychiatric symptoms Rates of suicidal ideation and attempt within the CU+Anx group were also considerably higher than within the Anx- The CU+Anx group reported significantly higher inter- ious and Low groups. In addition, of those who had sexual nalising symptoms compared to the CU−Anx group (see intercourse, more than half (64%) in the CU+Anx group Fig. 3a), based on both self-reported (i.e. depression) and reported not using a condom or other contraceptive dur- ing their last sexual encounter, compared to 34.5% in the CU−Anx group, 27.6% in the Anxious group and 22.7% in the Low group. Attachment style Attachment style differed significantly across groups, X (9,154) = 38.10, p < 0.001. As can be seen in Fig. 3c, the most striking difference relates to the proportions of secure vs disorganised attachment across groups. The CU+Anx group were predominantly characterised by dis- organised (45%) and avoidant attachment (32%) styles, with only 7.5% showing secure attachment, the lowest proportion relative to any other group. The Anxious group were predominantly characterised by disorganised (34%) and anxious attachment (31%) styles, with 19% show- ing secure attachment. In contrast 53% and 44% of the CU−Anx and Low groups, respectively, were classified as securely attached. Fig. 2 Mean levels of childhood maltreatment severity across groups 1 3 892 European Child & Adolescent Psychiatry (2018) 27:885–898 1 3 Table 2 Group comparisons on markers of individual functioning Overall sample Low CU High CU Omnibus test Pair-wise contrasts (n = 155) Low (n = 36) Anxious CU−Anx CU+Anx CU+Anx vs. CU−Anx CU+Anx vs. Anxious (n = 43) (n = 36) (n = 40) Effect size [95% CI] Effect size [95% CI] Psychiatric symptoms  Self-report a,b   Depression, M (SD) 6.55 (5.00) 3.33 (1.98) 9.00 (4.52) 3.42 (2.93) 9.63 (5.52) F (2,155) = 39.29, p < 0.001 *** g = 1.37 [0.87, 1.87] – a,b   Anger, M (SD) 7.65 (5.72) 4.44 (3.79) 8.02 (5.19) 5.19 (4.24) 12.33 (5.79) F (2,155) = 28.13, p < 0.001 *** g = 1.38 [0.88, 1.88] ** g = 0.78 [0.33, 1.22] a,b †   PTSD, M (SD) 9.77 (6.81) 4.44 (3.78) 11.74 (6.07) 6.31 (4.56) 15.58 (5.99) F (2,155) = 53.61, p < 0.001 *** g = 1.71 [1.19, 2.24] g = 0.63 [0.19, 1.07] a,b   Dissociation, M (SD) 9.41 (6.13) 6.11 (4.37) 10.53 (5.35) 5.56 (3.62) 14.65 (5.96) F (2,155) = 36.59, p < 0.001 *** g = 1.80 [1.27, 2.34] ** g = 0.72 [0.28, 1.17]  Informant-rated b,c   GAD, M (SD) 4.21 (4.17) 1.52 (1.71) 4.16 (4.62) 4.04 (3.62) 6.82 (4.2) F (2,154) = 18.54, p < 0.001 ** g = 0.70 [0.23, 1.16] ** g = 0.59 [0.15, 1.03] b,c   MDD, M (SD) 2.78 (3.74) 0.71 (1.45) 2.65 (3.38) 2.57 (3.00) 5.08 (4.91) F (2,151) = 13.54, p < 0.001 ** g = 0.60 [0.14, 1.06] ** g = 0.58 [0.14, 1.02] a,b,c   ODD, M (SD) 2.91 (4.21) 0.56 (1.48) 1.46 (2.47) 4.06 (4.01) 5.63 (5.59) F (2,152) = 10.10, p < 0.001 – *** g = 0.97 [0.51, 1.42] b,c   CD, M (SD) 1.38 (2.79) 0.11 (0.40) 0.56 (1.10) 1.83 (2.64) 3.08 (4.28) F (2,151) = 8.18, p < 0.001 – *** g = 0.81 [0.37, 1.26] b,c   ASPD, M (SD) 2.09 (3.50) 0.26 (0.82) 0.93 (1.72) 2.77 (3.89) 4.52 (4.55) F (2,151) = 14.54, p < 0.001 – *** g = 1.05 [0.59, 1.51] b,c   ADHD, M (SD) 7.85 (9.19) 2.14 (4.40) 5.14 (8.96) 9.40 (7.47) 14.73 (9.57) F (2,152) = 21.41, p < 0.001 ** g = 0.61 [0.57, 1.48] *** g = 1.03 [0.57, 1.48] Behavioural risk markers  Alcohol use, M (SD) 4.99 (5.31) 5.14 (4.65) 4.71 (4.88) 4.34 (4.20) 5.78 (7.15) X (3,150) = 1.38, ns – – 2 b  Drug use, M (SD) 3.49 (6.91) 1.89 (4.86) 2.69 (5.68) 3.48 (6.42) 5.97 (9.47) X (3,150) = 23.08, p < 0.001 *** OR = 2.17 [1.27, 3.71] ** OR = 2.18 [1.32, 3.60] 2 b †  Suicidal ideation (%) 15.1 0 12.0 14.3 33.3 X (3,152) = 16.84, p < 0.001 – OR = 3.70 [1.17, 11.65] 2 b  Suicide attempt (%) 10.3 0 7.0 11.0 22.5 X (3,152) = 11.10, p < 0.01 – – 2 † †  Unsafe sex (%) 37.1 22.7 27.6 34.5 64.0 X (3,105) = 10.32 OR = 3.38 [1.10, 10.35] ** OR = 4.67 [1.47, 14.79] Affective functioning  Irritability, M (SD) 4.09 (3.71) 2.51 (2.67) 4.48 (4.10) 2.82 (2.54) 6.21 (3.93) F (2,150) = 12.77, p < 0.001 *** g = 0.99 [0.52, 1.47] –  Alexithymia M (SD) 14.63 (6.38) 10.39 (3.42) 16.55 (6.68) 12.88 (5.13) 18.05 (6.59) F (2,151) = 21.49, p < 0.001 *** g = 0.85 [0.38, 1.323] – N.B. Analyses control for sex. Hedge’s g guidelines for effect size: g of 0.20 = small, 0.50 = medium, 0.80 = large GAD generalised anxiety disorder, MDD major depressive disorder, ODD oppositional defiant disorder, CD conduct disorder, ASPD antisocial personality disorder, ADHD attention-deficit hyperactivity disorder, OR odds ratio  p < 0.05, ** p < 0.01, *** p < 0.001 a b c CU−Anx vs Anxious contrast significant at least at p < 0.01;  CU+Anx vs Low contrast significant at least at p < .01;  CU−Anx vs Low significant at least at p < 0.01 European Child & Adolescent Psychiatry (2018) 27:885–898 893 Fig. 3 Group differences on levels of psychiatric symptomatology, c Attachment style classification across groups. GAD generalised behavioural risk and attachment style. a Standardised mean levels of anxiety disorder, MDD major depressive disorder, ODD oppositional self-report (TSCC-A; top-half) psychological distress and informant- defiant disorder, CD conduct disorder, ASPD antisocial personality report (ASI; bottom-half) psychiatric symptomatology across groups. disorder, ADHD attention-deficit hyperactivity disorder b Percentage of endorsement of behavioural risk items across groups. Aec ff tive functioning Discussion The two variants of CU youth differed significantly on both This study systematically characterised variants of CU youth measures of affective functioning, with the CU+ Anx group in a high-risk community sample. Specifically, we compared showing higher levels of irritability (p < 0.001, g = 0.99) youth who presented with similarly high levels of CU traits, and alexithymia (p  <  0.001, g  =  0.85). In contrast, the but different levels of co-occurring anxiety (i.e. CU+ Anx vs CU+Anx group did not differ from the Anxious group on CU−Anx) on maltreatment history, psychiatric symptoma- either measure of affective functioning. The CU−Anx group tology and broad markers of individual functioning. The use showed a profile of affective functioning similar to that of of multiple informants was a key strength of our study, with the Low group. multi-rated assessments used in both construction of CU groups as well as the examination of individual functioning Post hoc power analysis domains. We highlight here three main findings. First, youth with CU+Anx were characterised by more severe histories The sample size in our study is consistent with the extant of childhood abuse and neglect compared to CU−Anx youth. literature on variants of CU youth in high-risk samples Second, while variants of CU youth did not differ on lev - (e.g. [15, 19, 22]), whereby elevated rates of develop- els of externalising problems (e.g. oppositional defiant and mental adversity and psychiatric symptomatology result conduct disorder symptoms), the CU+Anx group presented in increased power to detect effects (i.e. as opposed to with significantly elevated levels of psychological distress general population samples). Nevertheless, we performed (i.e. depression, anger, dissociation and PTSD symptoms), a post hoc analysis to ensure that we were appropriately insecure attachment, affective dysregulation and behavioural powered for the analyses undertaken. Based on post hoc risk. Third, the inclusion of an Anxious comparison group G*Power calculations, with a sample size of n = 155, four revealed widespread similarities in trauma history and indi- groups and moderate-to-large effect sizes for all outcome vidual functioning between CU+Anx youth and those low in variables, we found that achieved power exceeded 0.85 CU but high in anxiety. Generally, CU+Anx youth seemed across analyses. to experience a ‘double hit’ of negative outcomes associated 1 3 894 European Child & Adolescent Psychiatry (2018) 27:885–898 with CU on the one hand, and anxiety on the other. They also unsafe sexual behaviours and adverse health outcomes [48], showed additional vulnerabilities compared to youth who and suggest the CU+Anx group is highly vulnerable across presented with either CU or Anxiety alone, including more multiple domains. severe feelings of anger and dissociation, elevated ADHD Our exploratory measures delineated additional dif- symptoms, greater drug use, engagement in unsafe sex and ferences across variants of CU youth in areas of affective higher suicide risk. Overall, the identification of distinct pat- functioning and attachment to close others. Elevated lev- terns of co-occurring psychiatric, emotional and behavioural els of irritability and anger in the CU+Anx group are con- markers associated with variants of CU youth have impor- sistent with the notion that this variant features increased tant and immediate clinical applications for informing risk emotional expression and reactivity [6, 17]. Furthermore, assessment and treatment formulation. attachment disorganisation, an established sequel of child- hood maltreatment [49], was found to be most common in Childhood maltreatment robustly discriminates youth with CU+Anx, while CU−Anx featured predominantly between variants of CU youth a secure attachment style. To our knowledge, this is the first study to have examined current patterns of attachment styles As hypothesised, childhood maltreatment emerged as a key across CU groups. Finally, increased levels of alexithymia factor discriminating variants of CU youth, with CU+Anx observed in CU+Anx (and Anxious youth) compared to youth reporting more severe trauma histories compared to CU−Anx youth may also reflect the developmental impact the CU−Anx group across all individual forms of abuse and of childhood maltreatment on emotional arousal and func- neglect. This finding is consistent with prior research that tioning. The finding related to alexithymia warrants further examined maltreatment as a global construct (or as part of investigation, as it may offer clues as to why individuals a wider adversity measure; e.g. [6, 14, 15]). While previous with CU+Anx share behavioural features with those with studies that have compared variants of CU youth on individ- CU−Anx (in other words, the present with a ‘behavioural ual forms of maltreatment (e.g. [16–18]) have shown some phenocopy’), yet appear emotionally reactive in a way that inconsistencies regarding which precise forms of maltreat- CU−Anx are not. High levels of alexithymia are associ- ment reliably differentiate CU+ Anx and CU−Anx groups, all ated with an inability to describe and identify emotions, have reported more pervasive maltreatment experiences in rather than an inability to experience emotional arousal. the CU+Anx group—which is broadly in line with our find- This means that although these individuals may experience ings. In contrast to previous studies, we additionally com- heightened affect in response to another person’s distress, pared maltreatment profiles against two comparison groups their ability to display socially appropriate responses may (i.e. Low and Anxious). While CU+Anx youth reported be compromised, leading them to appear callous and uncar- comparable levels of abuse and neglect to youth presenting ing. The finding that CU+ Anx reported the highest levels with high anxiety but low CU (i.e. the Anxious group) the of dissociative symptoms compared to any other group CU−Anx group did not differ in maltreatment history from may lend additional support for this hypothesis, as do prior those showing low CU and low anxiety (i.e. Low group). reports of lack of emotional ‘clarity’ within this group [28]. In contrast, adults with primary psychopathy and youth with CU+Anx indexes a particularly vulnerable group CU−Anx have been shown to be typically characterised by of individuals low emotional arousal to other people’s distress [17]. Youth with CU+Anx presented with the highest mean lev- CU+Anx youth share many similarities with Anxious els of psychological distress across all domains examined, youth in line with adult data on individuals who score high on psychopathy and anxiety [8] as well as youth data on CU The inclusion of two comparison groups enabled us to groups [6, 7, 14, 15]. Additionally, the CU+Anx group was compare variants of CU youth to low CU youth who also characterised by significantly elevated behavioural risk, vary in their levels of anxiety. Interestingly, we found that including increased drug use, feelings of suicidality and Anxious youth, albeit lower in levels of externalising prob- engagement in unsafe sex. Alarmingly, one third of youth lems, reported similar levels of childhood trauma, emo- in the CU+Anx group in this high-risk sample reported hav- tional difficulties and psychological distress to CU+Anx ing seriously considered committing suicide in the past year, youth. Consequently, an important question that emerged and almost one fourth reported attempting suicide. Rates of from the present data related to why some youth with a unsafe sex were also high in the CU+Anx group, with more history of trauma presented with both high levels of CU than half of youth reporting not using a condom or other and anxiety (i.e. CU+Anx) while others only present with contraceptive during their last sexual intercourse. These fig- high anxiety (i.e. Anxious group). One possibility is that ures are disturbing given the known associations between youth with CU+Anx have additional genetic vulnerability to 1 3 European Child & Adolescent Psychiatry (2018) 27:885–898 895 externalising disorders/impulsivity, as is suggested by their Limitations substance use, suicidal ideation and sexual behaviour pro- file. It is also possible that CU+ Anx youth may be exposed The findings of present study should be interpreted in light to additional environmental risk factors relative to Anxious of several limitations. First, CU traits are a dimensional con- youth, that were not captured in the current study (e.g. bul- struct, not a taxon. As we wished to compare variants of CU lying-victimisation). Longitudinal investigations charting traits, a categorical approach provided an effective means children who have experienced maltreatment, but who come of communication and this way of characterising children from families characterised by different levels of externalis- is also directly relevant for informing clinical practice. In ing problems, could shed light into this issue. future, studies may benefit from using dimensional infor - mation to supplement categorical approaches. Furthermore, Research and clinical implications although the measure used in our study to index CU traits (i.e. the ICU) has been commonly employed in the literature The present findings highlight the need to differentiate on variants of CU youth as well as being shown to possess between variants of CU youth. Supplementing measures of good factor structure, construct and predictive validity in CU traits with an assessment of anxiety can offer important a range of populations [52–56], some concerns have been information for both clinicians and researchers. Failure to raised about aspects of its psychometric properties [57] so assess levels of anxiety among youth with high CU traits that results will need to be replicated using an independent may obscure the diverse constellations of needs and risk measure of CU traits. Second, the anxiety measure used in factors associated with subgroups of individuals presenting this study to define groups was taken from the same ques- with elevated CU traits. Equally, the current findings high- tionnaire as our self-reported outcomes of psychological light that experiences of childhood maltreatment markedly distress, which raises issues of shared-method variance. differ between variants of CU youth. In research and clinical However, it is important to note that variants of CU youth settings, developmental adversity is not always assessed con- were also found to differ on levels of internalising problems currently with CU traits in youth [14]. An increased aware- (i.e. symptoms of generalised anxiety and major depressive ness of maltreatment as a possible risk factor for CU+Anx disorder) based on ratings from independent informants (i.e. may be helpful in informing risk assessment and suitable teachers/key workers). Third, while inclusion of a measure intervention strategies. Importantly, the findings indicate of childhood maltreatment provided a temporal proxy for that focussing on conduct problems or antisocial behaviour the ee ff ct of developmental adversity on CU +Anx, the cross- alone is unlikely to discriminate between variants of CU sectional nature of the study meant that we were unable to youth, as they tend to present similarly on these domains. establish the causality of effects found. However, the consist- Youth with CU+Anx represent a high-risk clinical group ency with which childhood maltreatment has been found to characterised by more severe developmental trauma, con- differentiate between variants of CU youth across our study current psychiatric symptomatology, affective dysfunction, and that of the extant literature (e.g. [14–18]) considerably risk behaviours and suicide risk. For these youths, therapeu- adds confidence to this finding. Despite this, it is important tic approaches that include the experience of trauma in the to note that while the data seem to suggest that CU+Anx treatment formulation, such as trauma-focussed CBT and may be more environmentally driven than CU−Anx, it was similar evidence-based interventions, may be warranted. not possible to remove potential genetic confounds from our Equally, interventions addressing conduct problems in youth design (e.g. youth high in CU may be more likely to have with CU+Anx may need embedding in a wider therapeutic parents high in psychopathic traits, who are also more likely intervention addressing other internalising problems, par- to maltreat them). Genetically informative designs may be ticularly anxiety and depression. High rates of disorganised particularly effective in examining the contribution of such attachment in this group are likely to predict poor interper- influences (e.g. [58]). Fourth, while post hoc analyses con- sonal functioning, and will be relevant to the clinician chal- firmed that we were appropriately powered for all analy - lenged with establishing appropriate boundaries alongside ses undertaken, sample size limitations meant that we were an effective therapeutic alliance. Finally, risk assessments only able to enter sex as a free-standing covariate. In future, will need to pay particular attention to engagement in risky the use of larger samples will make it possible to examine behaviours (e.g. drug use) and increased risk of suicidal- whether sex moderates associations between variants of CU ity as these were strongly associated with CU+Anx. More youth and markers of individual functioning. Finally, even broadly, our findings support a growing emphasis in the field though sampled from the community, youth in our study on CU traits as a cross-disorder construct [50, 51], which came predominantly from high-risk, multi-problem families. needs to be more fully considered within the broader context As a result, further research is needed to establish the extent of different forms of psychopathology and risk behaviours to which findings may generalise to the wider population. across both research and clinical settings. 1 3 896 European Child & Adolescent Psychiatry (2018) 27:885–898 to a greater understanding of the nature and significance Future directions of variants of CU youth. The present findings point to a number of directions for Acknowledgements The authors are indebted to the young people, future research. First, longitudinal, prospective research key-workers and teachers who have taken part in this study. This is needed to gain a more mechanistic understanding of research was supported by Kids Company charity. CC is supported by processes underlying variants of CU traits in youth. Lon- the Economic and Social Research Council (grant ref: ES/N001273/1). EV and EMC were supported by UK Medical Research Council grant gitudinal studies may also help determine whether variants (MR/ K014080/1) during the writing of this article. EV is a Royal are predictive of different developmental trajectories and Society Wolfson Research Merit Award Holder. outcomes over time, particularly in relation to frequency and nature of violence, suicidality, and mental health Compliance with ethical standards problems. Indeed, efforts to map variants longitudinally are already beginning to emerge [7, 23, 26, 27]. Second, Conflict of interest On behalf of all authors, the corresponding author examining the timing of maltreatment experiences may states that there is no conflict of interest. be important for understanding how CU+Anx develops Open Access This article is distributed under the terms of the Creative and identifying whether developmental windows exist Commons Attribution 4.0 International License (http://creativecom- where the effect of maltreatment is more pronounced. mons.org/licenses/by/4.0/), which permits unrestricted use, distribu- Third, CU+Anx may represent a ‘phenocopy’ of CU−Anx, tion, and reproduction in any medium, provided you give appropriate but the origins of CU and the underlying neurocognitive credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. mechanisms for the two variants may differ. A number of studies have provided support for differences in behav - ioural performance across variants on measures of emo- tional processing and behavioural activation [6, 16, 17, References 28]. Future neurocognitive studies would benefit from direct comparisons of CU+Anx with anxious individu- 1. Frick PJ, Ray JV, Thornton LC, Kahn RE (2014) Can callous–une- als, as well as use of tasks that investigate processes that motional traits enhance the understanding, diagnosis, and treat- should be compromised in CU, but not in anxiety. Fourth, ment of serious conduct problems in children and adolescents? A given that CU traits are known to be moderately associ- comprehensive review. Psychol Bull 140(1):1 2. Blair RJR, Leibenluft E, Pine DS (2014) Conduct disor- ated with conduct problems (e.g. r = 0.54 in our study) der and callous–unemotional traits in youth. N Engl J Med it will be important in future to establish to what extent 371(23):2207–2216 these co-occurring symptoms may be driving observed 3. Frick PJ, Viding E (2009) Antisocial behavior from a devel- differences between CU+ Anx vs Anx only youth. From opmental psychopathology perspective. Dev Psychopathol 21(Special Issue 04):1111–1131. https://doi.org/10.1017/ a research perspective, it is notable that we found group S0954579409990071 differences in the ratio of males to females across variants 4. de Wied M, van Boxtel A, Matthys W, Meeus W (2012) Verbal, of CU youth in this community sample. While the group facial and autonomic responses to empathy-eliciting film clips by of youth with CU−Anx contained disproportionately more disruptive male adolescents with high versus low callous–unemo- tional traits. J Abnorm Child Psychol 40(2):211–223 boys, the CU+Anx group had a more balanced male to 5. Sebastian CL, McCrory EJ, Cecil CA, Lockwood PL, De Brito female ratio (slightly greater number of girls). Moreover, SA, Fontaine NM, Viding E (2012) Neural responses to affective the Anxious group featured predominantly females. These and cognitive theory of mind in children with conduct problems findings are in line with previous work examining variants and varying levels of callous–unemotional traits. Arch Gen Psy- chiatry 69(8):814–822 of CU youth [31]. Interestingly, another study has reported 6. Kimonis ER, Frick PJ, Cauffman E, Goldweber A, Skeem J (2012) that psychopathic personality traits are associated with a Primary and secondary variants of juvenile psychopathy differ history of trauma in young female offenders [59]. Future in emotional processing. Dev Psychopathol 24(3):1091–1103. studies should test whether the difference in sex ratio is a https://doi.org/10.1017/s0954579412000557 7. Meehan AJ, Maughan B, Cecil CA, Barker ED (2017) Interper- reliable finding and whether the experience of trauma may sonal callousness and co-occurring anxiety: developmental valid- represent a particularly potent risk factor for CU+Anx in ity of an adolescent taxonomy. J Abnorm Psychol 126(2):225 girls. Finally, research is needed to inform the develop- 8. Skeem JL, Poythress N, Edens JF, Lilienfeld SO, Cale EM (2003) ment of more tailored interventions as well as to evalu- Psychopathic personality or personalities? Exploring potential variants of psychopathy and their implications for risk assessment. ate whether the application of differing strategies may be Aggress Violent Behav 8(5):513–546. https://doi.org/10.1016/ more effective than a ‘one size fits all’ intervention. This is S1359-1789(02)00098-8 especially important given the dearth of programmes spe- 9. Karpman B (1941) On the need of separating psychopathy into cifically validated on youth with CU traits [17]. Together, two distinct clinical types: the symptomatic and the idiopathic. J Crim Psychopathol 3:112–137 studies addressing these future directions will contribute 1 3 European Child & Adolescent Psychiatry (2018) 27:885–898 897 10. Cleckley H (1941) The mask of sanity: an attempt to reinterpret Personal Individ Differ 111(Supplement C):124–133. https://doi. the so-called psychopathic personality. Mosby, St. Louis, MO org/10.1016/j.paid.2017.02.005 11. Viding E, McCrory E (2015) Developmental risk for psychopathy. 28. Bennett DC, Kerig PK (2014) Investigating the construct of In: Thapar A, Pine DS, Leckman JF, Scott S, Snowling MJ, Taylor trauma-related acquired callousness among delinquent youth: E (eds) Rutter’s child and adolescent psychiatry. Wiley, New York, differences in emotion processing. J Trauma Stress 27(4):415–422 pp 966–980. https://doi.org/10.1002/9781118381953.ch68 29. Dackis MN, Rogosch FA, Cicchetti D (2015) Child maltreatment, 12. Viding E, Kimonis E (2017) Callous–unemotional traits in chil- callous–unemotional traits, and defensive responding in high-risk dren and youth. In: Patrick C (ed) Handbook of psychopathy, 2nd children: an investigation of emotion-modulated startle response. edn. The Guildford Press, New York (in press) Dev Psychopathol 27(4pt2):1527–1545 13. Cecil CA, Lysenko LJ, Jaffee SR, Pingault JB, Smith RG, Rel- 30. Kimonis ER, Goulter N, Hawes DJ, Wilbur RR, Groer MW (2017) ton CL, Woodward G, McArdle W, Mill J, Barker ED (2014) Neuroendocrine factors distinguish juvenile psychopathy variants. Environmental risk, Oxytocin Receptor Gene (OXTR) methyla- Dev Psychobiol 59(2):161–173 tion and youth callous–unemotional traits: a 13-year longitudinal 31. Fanti KA, Demetriou CA, Kimonis ER (2013) Variants of cal- study. Mol Psychiatry 19(10):1071–1077. https://doi.org/10.1038/ lous–unemotional conduct problems in a community sample of mp.2014.95 adolescents. J Youth Adolesc 42(7):964–979 14. Tatar JR, Cauffman E, Kimonis ER, Skeem JL (2012) Victimi- 32. Lee H-S, Barraza-Villarreal A, Biessy C, Duarte-Salles T, Sly PD, zation history and posttraumatic stress: an analysis of psychopa- Ramakrishnan U, Rivera J, Herceg Z, Romieu I (2014) Dietary thy variants in male juvenile offenders. J Child Adolesc Trauma supplementation with polyunsaturated fatty acid during pregnancy 5(2):102–113. https://doi.org/10.1080/19361521.2012.671794 modulates DNA methylation at IGF2/H19 imprinted genes and 15. Vaughn MG, Edens JF, Howard MO, Smith ST (2009) An growth of infants. Physiol Genom 46(23):851–857 investigation of primary and secondary psychopathy in a state- 33. Dadds MR, Kimonis ER, Schollar-Root O, Moul C, Hawes DJ wide sample of incarcerated youth. Youth Violence Juv Justice (2017) Are impairments in emotion recognition a core feature of 7(3):172–188. https://doi.org/10.1177/1541204009333792 callous–unemotional traits? Testing the primary versus secondary 16. Kahn RE, Frick PJ, Youngstrom EA, Kogos Youngstrom J, variants model in children. Dev Psychopathol 1–11. http://dx.doi. Feeny NC, Findling RL (2013) Distinguishing primary and org/10.1017/S0954579417000475 secondary variants of callous–unemotional traits among ado- 34. Cecil CA, Viding E, Barker ED, Guiney J, McCrory EJ (2014) lescents in a clinic-referred sample. Psychol Assess 25(3):966– Double disadvantage: the influence of childhood maltreatment 978. https://doi.org/10.1037/a0032880 and community violence exposure on adolescent mental health. J 17. Kimonis ER, Fanti KA, Goulter N, Hall J (2016) Affective star - Child Psychol Psychiatry. https://doi.org/10.1111/jcpp.12213 tle potentiation differentiates primary and secondary variants of 35. Wechsler D (1999) Wechsler abbreviated scale of intelligence. juvenile psychopathy. Dev Psychopathol 29(4):1149–1160 Psychological Corporation, San Antonio 18. Kimonis ER, Fanti KA, Isoma Z, Donoghue K (2013) Maltreat- 36. IMD (2011) The English indices of deprivation 2010. IMD, ment profiles among incarcerated boys with callous–unemo- London tional traits. Child Maltreat 18(2):108–121 37. Frick PJ (2004) The inventory of callous–unemotional traits. 19. Rosan A, Frick PJ, Gottlieb KA, Fasicaru L (2015) Callous–une- Unpublished rating scale, University of New Orleans, New motional traits and anxiety in a sample of detained adolescents Orleans, LA in Romania. J Evid Based Psychother 15(1):79 38. Briere J (1996) Trauma symptom checklist for children (TSCC). 20. Sharf A, Kimonis ER, Howard A (2014) Negative life events Psychological Assessment Resources, Lutz and posttraumatic stress disorder among incarcerated boys 39. Bernstein DP, Fink L (1998) Childhood trauma questionnaire: a with callous–unemotional traits. J Psychopathol Behav Assess retrospective self-report manual. Psychological Corporation, San 36(3):401–414. https://doi.org/10.1007/s10862-013-9404-z Antonio 21. Docherty M, Boxer P, Huesmann LR, O’Brien M, Bushman BJ 40. Gadow KD, Sprafkin JN (1998) Adolescent symptom inventory-4. (2016) Exploring primary and secondary variants of psychopa- Checkmate Plus, Stony Brook thy in adolescents in detention and in the community. J Clin 41. Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M Child Adolesc Psychol 45(5):564–578 (1993) Development of the Alcohol Use Disorders Identifica- 22. Gill AD, Stickle TR (2016) Affective differences between psy - tion Test (AUDIT): WHO Collaborative Project on early detec- chopathy variants and genders in adjudicated youth. J Abnorm tion of persons with harmful alcohol consumption-II. Addiction Child Psychol 44(2):295–307 88(6):791–804 23. Goulter N, Kimonis ER, Hawes SW, Stepp S, Hipwell AE (2017) 42. Berman A, Bergman H, Palmstierna T, Schlyter F (2003) DUDIT: Identifying stable variants of callous–unemotional traits: a lon- the drug use disorders identification test manual. Karolinska Insti- gitudinal study of at-risk girls. Dev Psychol 53(12):2364–2376 tutet, Department of Clinical Neuroscience, Stockholm 24. Kahn RE, Frick PJ, Golmaryami FN, Marsee MA (2017) The 43. Eaton DK, Kann L, Kinchen S, Shanklin S, Ross J, Hawkins J, moderating role of anxiety in the associations of callous– Harris WA, Lowry R, McManus T, Chyen D (2008) Youth risk unemotional traits with self-report and laboratory measures behavior surveillance—United States, 2007. MMWR Surveill of affective and cognitive empathy. J Abnorm Child Psychol Summ 57(4):1–131 45(3):583–596 44. Wei M, Russell DW, Mallinckrodt B, Vogel DL (2007) The Expe- 25. Lee Z, Salekin RT, Iselin A-MR (2010) Psychopathic traits in riences in Close Relationship Scale (ECR)-short form: reliability, youth: is there evidence for primary and secondary subtypes? J validity, and factor structure. J Personal Assess 88(2):187–204 Abnorm Child Psychol 38(3):381–393 45. Stringaris A, Goodman R, Ferdinando S, Razdan V, Muhrer 26. Fanti KA, Kimonis E (2017) Heterogeneity in externalizing prob- E, Leibenluft E, Brotman MA (2012) The Affective Reactivity lems at age 3: association with age 15 biological and environmen- Index: a concise irritability scale for clinical and research settings. tal outcomes. Dev Psychol 53(7):1230–1241 J Child Psychol Psychiatry 53(11):1109–1117 27. Ezpeleta L, Granero R, de la Osa N, Domènech JM (2017) Devel- 46. Bagby RM, Parker JD, Taylor GJ (1994) The twenty-item Toronto opmental trajectories of callous–unemotional traits, anxiety and Alexithymia Scale—I. Item selection and cross-validation of the oppositionality in 3–7 year-old children in the general population. factor structure. J Psychosom Res 38(1):23–32 1 3 898 European Child & Adolescent Psychiatry (2018) 27:885–898 47. IBM Corp (2012) IBM SPSS statistics for Windows, version 21.0. validation of the inventory of callous and unemotional traits. IBM Corp, Armonk Assessment 17(1):44–57 48. Coyle K, Basen-Engquist K, Kirby D, Parcel G, Banspach S, Col- 55. Ciucci E, Baroncelli A, Franchi M, Golmaryami FN, Frick PJ lins J, Baumler E, Carvajal S, Harrist R (2001) Safer choices: (2014) The association between callous–unemotional traits and reducing teen pregnancy, HIV, and STDs. Public Health Rep behavioral and academic adjustment in children: further valida- 116(Suppl 1):82 tion of the inventory of callous–unemotional traits. J Psycho- 49. Cyr C, Euser EM, Bakermans-Kranenburg MJ, Van Ijzendoorn pathol Behav Assess 36(2):189–200. https://doi.org/10.1007/ MH (2010) Attachment security and disorganization in maltreat- s10862-013-9384-z ing and high-risk families: a series of meta-analyses. Dev Psycho- 56. Henry J, Pingault JB, Boivin M, Rijsdijk F, Viding E (2016) pathol 22(01):87–108 Genetic and environmental aetiology of the dimensions of Cal- 50. Herpers PC, Rommelse NN, Bons DM, Buitelaar JK, Scheepers lous–Unemotional traits. Psychol Med 46(2):405–414 FE (2012) Callous–unemotional traits as a cross-disorders con- 57. Berg JM, Lilienfeld SO, Reddy SD, Latzman RD, Roose A, Craig- struct. Soc Psychiatry Psychiatr Epidemiol 47(12):2045–2064 head LW, Pace TW, Raison CL (2013) The inventory of callous 51. Salekin RT (2017) Research review: what do we know about and unemotional traits: a construct-validational analysis in an at- psychopathic traits in children? J Child Psychol Psychiatry risk sample. Assessment 20(5):532–544 58(11):1180–1200 58. Humayun S, Kahn RE, Frick PJ, Viding E (2014) Callous–unemo- 52. Kimonis ER, Frick PJ, Skeem JL, Marsee MA, Cruise K, Munoz tional traits and anxiety in a community sample of 7-year-olds. J LC, Aucoin KJ, Morris AS (2008) Assessing callous–unemotional Clin Child Adolesc Psychol 43(1):36–42 traits in adolescent offenders: validation of the Inventory of Cal- 59. Odgers CL, Reppucci ND, Moretti MM (2005) Nipping psychopa- lous–Unemotional Traits. Int J Law Psychiatry 31(3):241–252 thy in the bud: an examination of the convergent, predictive, and 53. Essau CA, Sasagawa S, Frick PJ (2006) Callous–unemotional theoretical utility of the PCL-YV among adolescent girls. Behav traits in a community sample of adolescents. Assessment Sci Law 23(6):743–763 13(4):454–469 54. Roose A, Bijttebier P, Decoene S, Claes L, Frick PJ (2010) Assess- ing the affective features of psychopathy in adolescence: a further 1 3

Journal

European Child & Adolescent Psychiatry – Springer Journals

Published: Dec 8, 2017

There are no references for this article.