Introduction
Callous–unemotional (CU) traits are thought to represent the core component of psychopathy, and include symptoms such as lack of feeling guilty, lack of empathy, being very egocentric, showing callous use of others for one’s own gain, and lacking normal emotionality, especially in showing a lack of anxiety [
13]. CU traits were incorporated in the DSM-5 [
2] as a specifier for conduct disorder (CD). The specifier consists of four criteria of which at least two should be met to index a more severe form of CD. CU traits in children with conduct problems (CP) have been reported to imply increased levels of aggressive behaviors, worse prognosis, and treatment refractoriness [
14]. Increasingly, research shows neurobiological underpinnings of psychopathy, in which reduced amygdala responsiveness to distress cues results in reduced empathic response to distress of other individuals (as captured by the callous–unemotional component of psychopathy). In addition, dysfunction in the ventromedial prefrontal cortex and striatum results in deficient decision making and reinforcement learning (as captured by the impulsive–antisocial component of psychopathy; for a comprehensive review, see [
5]).
The majority of research on CU traits has been conducted in youths with CP. However, little is known about the presence of CU traits in disorders other than CD and about their implications for severity of these disorders and functional adaptation [
20,
32]. In previous research, CU traits were not associated with quality of life (QoL) in a clinical sample of youths with CD [
22]. Impairment in youths with CP showed either no [
19,
22,
29,
30] or a positive [
15,
25] relationship with CU traits. In community samples high CU traits were associated with more global impairment [
10,
34,
43], not only in the CD subsample, but in the no CD/high CU subsample as well [
34]. There are indications that impairment in the presence of attention-deficit/hyperactivity disorder (ADHD) symptoms may be moderated by CU traits [
6,
44]. However, no studies reported yet on QoL in specific non-CD diagnoses.
Our cross-sectional study extends previous research, by examining associations between CU traits and non-CD diagnoses in a large clinical sample and by investigating relations between CU traits and QoL across non-CD disorders. We addressed the following questions: (1) what are the odds for scoring high on CU traits when being diagnosed having a psychiatric disorder other than CD?, (2) what is the relation between high CU traits and QoL in non-CD diagnoses?
Discussion
This study investigated whether CU traits should be seen as a cross disorder phenomenon that also occurs outside CD, and whether high CU traits outside CD affect functioning in important domains as reflected in a measure of QoL. In our large clinical sample, the odds for high CU traits were found to be significantly increased in ASD and DBD-NOS/ODD, while the odds for high CU traits were found to be significantly decreased in anxiety/mood disorders. For ADHD and other diagnoses, the odds for high CU traits were not significantly increased. A new and important finding is that, in all diagnostic groups, high CU traits were associated with lower QoL, and explained a significant proportion of variance (4–13 %) in QoL beyond effects of age and gender. In contrast, education level of child or parents and the police contacts were unrelated to QoL.
These findings support and extend earlier studies reporting on CU traits being present outside CD [
40] and on the negative impact of CU traits across disorders [
7]. In contrast to previous research in youths with CP (which refers to a broader concept than our DBD-NOS/ODD diagnosis), we found QoL to be significantly decreased in the high CU group compared to the low CU group. Our DBD-NOS/ODD only sample was small, however, findings remained in the same direction when controlling for comorbidity. Thus, our results underscore the importance of considering CU traits as a specifier for ODD as well [
20]. Nevertheless, as only one scale and only one source of information has been used to detect CU traits, we need to keep in mind that parents might rate ‘Has your child been able to pay attention’ (Kidscreen-27 item) similar as being ‘concerned about schoolwork’ (ICU item).
In the ADHD group, our findings contrast with previous results showing that CU traits moderate functional impairment in those with low and moderate levels of ADHD symptoms [
6]. In our sample, CU traits were related to QoL in those with a diagnosis of ADHD, all of whom had at least moderate or high levels of ADHD symptoms. This is an important finding because CU traits may moderate treatment response in ADHD as well. Effects of behavioral therapy were found to be less in the presence of high CU traits compared to those with low CU traits [
16‐
18,
42], which might be related to the component of punishment in the treatment program [
31]. Methylphenidate was found to show a positive effect on CU traits [
4] and on CP [
42]. However, it is unclear whether the effect of CU traits in these ADHD samples was confounded by comorbidity with ODD or CD. Therefore, we controlled for comorbid DBD-NOS/ODD and were able to show CU traits having incremental value in predicting QoL in youths with ADHD, over and beyond CP. It is important to note that this effect existed independently of the finding that on average there was no increased OR for CU traits in ADHD. Thus, CU traits may be an independent predictor of treatment response regardless of diagnosis.
Previous studies showed that high scores on CU traits in youths with ASD may, at least in part, reflect theory of mind (ToM) deficits due to impaired empathic response to distress cues [
38], cognitive empathy deficits [
21,
41], and significantly decreased medial prefrontal cortex responses during ToM tasks in youths with ASD compared to youths with CP and high CU traits, and typically developing youths [
33]. Thus, high CU traits in ASD may be due to different underlying cognitive and brain mechanisms than high CU traits in CD. As we found a significantly increased OR for CU traits in ASD in this clinical sample, and high CU traits negatively impacted upon QoL, it is important to further unravel the relationship between CU traits and ASD.
This study confirmed previous studies that found a negative relationship between anxiety and CU traits (for a review, see e.g., [
14]). The relationship between high CU traits and mood disorders has been investigated scarcely (see e.g., [
20]). We found high CU traits within anxiety/mood disorders to be related to significantly lower QoL. In this diagnostic group, this effect existed independently of the finding that on average there was a decreased OR for CU traits. However, when controlling for comorbidity, this effect disappeared, which might be due to the relative small number of participants in the high CU group (9 versus 39 in the low CU group). Also important to note is the fact that it is unknown whether CU traits in anxiety/mood disorders represent the same underlying construct as in youths with CP. Parents might, for example, be rating anhedonia rather than the lack of emotion, or social avoidance rather than hiding one’s emotions. Further investigation of our finding is therefore needed in either anxiety and mood disorder only groups, as well as further research into the prognosis and treatment results of high CU traits in the presence of anxiety/mood disorders.
The main strength of this study was that it focused on the relation between CU traits and QoL in clinically established non-CD diagnoses in a large psychiatric outpatient sample. However, an important limitation is the lack of control groups, such as either (a large group of) youths with disruptive behavior disorders and normal controls. A potential limitation is the fact that there is no established cut-off score for the ICU. Although it still is difficult to define which is the best way to establish cut-off scores regarding the ICU [
23], our results give important information about the distribution of high CU traits across non-CD disorders. Furthermore, the fact that we relied on parent-reported assessment of CU traits and of QoL may mean that shared method variance may have inflated correlations between these two variables. However, regarding the individuals items, there is virtually no overlap, which also emerged from collinearity analyses. Nevertheless, future studies might adopt a more comprehensive multi-informant multi-measure approach to assess CU traits, to be able to address issues of potential informant and instrument bias. Also, future research might benefit a structured interview to establish CU traits, such as the Clinical Assessment of Prosocial Emotions [
11] which provides DSM diagnoses of CU traits. However, this is a clinician-rated measure and published only recently, and not available at time of our data collection. However, its value has to be established yet. Similarly, much of the analyses depend on clinical diagnoses that were not made with a structured and well-established clinical interview (e.g., the K-SADS). Nevertheless, our diagnostic procedures were similar to those to reach best estimate clinical diagnoses and as such, thought to be more reliable than the use of structured interviews only [
27]. Not having addressed intelligence as a covariate in our analyses is also a limitation. As we did not gather information regarding intelligence in a standardized way, we applied a broad definition of intelligence in terms of identifying individuals with borderline intellectual functioning (IQ below 85). Although our results stayed similar when excluding those with an estimated IQ below 85, previous findings about the relationship between psychopathic traits and intelligence in conduct disorder have been contrasting [
1]. Therefore, it could be noteworthy to specifically examine the impact of IQ on CU traits in non-CD disorders. Also, we did not investigate the temporal relation between CU traits and stressful life events. Given the discussion regarding secondary psychopathy in which traumatization is seen as possible moderating CU traits [
28], it may be important to investigate whether traumatized children show CU behavior due to their traumatization (‘emotional numbness’), rather than having a truly underlying high CU phenotype. As we did not gather longitudinal data, we have no information that addresses a developmental perspective of CU traits, and as such its possible malleability through time [
35].
Several unanswered questions remain for future research. One such question is: what is the mechanism through which high CU traits affect QoL in non-CD disorders? The relationship between CU traits and non-CD disorders is poorly understood. It is unknown how peers perceive youths with high CU traits, and there may be other ways in which social relationships are compromised. Also we do not know whether there is equifinality or multifinality in the causes of high CU traits in non-CD disorders compared to those causes in CD. High CU traits may be related to decreased problem solving skills and thus predispose to less help-seeking behavior or being less help receptive and thus to decreased QoL. Although age was not found to be a significant moderator or covariate, it remains unclear whether CU traits in these diagnostic groups may have different meanings at different ages. As such, many issues for further research remain.