Callous-unemotional traits, cognitive functioning, and externalizing problems in a propensity-matched sample from the ABCD study

Abstract

Background: Many studies show that both callous-unemotional (CU) traits (e.g., low empathy, lack of guilt) and cognitive difficulties increase risk for externalizing psychopathology across development. However, other work suggests that some aggression (e.g., relational, proactive) may rely on intact cognitive function, which could vary based on the presence of CU traits. Moreover, no prior research has adequately accounted for common risk factors shared by CU traits, cognitive difficulties, and externalizing problems, which confounds conclusions that can be drawn about their purported relationships. The current study addressed these knowledge gaps by leveraging rigorous propensity matching methods to isolate associations between CU traits and different dimensions of cognitive function and externalizing problems.

Methods: Associations between CU traits, cognitive functioning, and externalizing outcomes were tested within dimensional (n = 11,868) and propensity-matched group-based (n = 1,224) models using data from the Adolescent Brain Cognitive Development Study®, with rigorous statistical control for shared sociodemographic risk factors. Cross-sectional outcomes were parent-reported symptoms of conduct disorder (CD), oppositional defiant disorder (ODD), and attention deficit hyperactivity disorder (ADHD). Longitudinal outcomes were child-reported overt and relational aggression.

Results: CU traits were uniquely related to more parent-reported CD, ODD, ADHD symptoms, as well as more child-reported aggressive behaviors. Effects of cognitive difficulties were domain specific and were not consistent across dimensional and propensity matched models. There was minimal evidence for divergent associations between CU traits and externalizing outcomes as a function of cognition (i.e., no moderation).

Conclusions: Rigorous control for sociodemographic factors within propensity-matched models establish CU traits as a robust and unique risk factor for externalizing psychopathology, over and above difficulties with cognitive functioning.

Keywords: aggressive behavior; callous‐unemotional traits; cognitive function; externalizing disorders.

Figure 1

Main effects of CU traits, general ability, executive function, and learning and memory on externalizing and aggressive behavior in dimensional models. (A) Dimensional CU traits were related to higher levels of ADHD (β = .29, p < .001), CD (β = .40, p < .001) and ODD (β = .42, p < .001) symptoms. Lower general ability was related to more ADHD (β = −.10, p < .001) and CD symptoms (β = −.08, p < .001). Lower executive function was related to more ADHD (β = −.09, p < .001), CD (β = −.03, p < .01), and ODD symptoms (β = −.03, p < .01). Lower learning and memory were also related to more ADHD (β = −.08, p < .001), CD (β = −.04, p < .001), and ODD symptoms (β = −.03, p < .01). (B) Dimensional CU traits were related to more child‐reported overt (β = .09, p < .001) and relational (β = .06, p < .001) aggression. Higher general ability was related to higher levels of relational aggression (β = .09, p < .001). Lower executive function was related to higher overt aggression (β = −.04, p < .001). Higher learning and memory was related to higher relational aggression (β = .02, p < .05) and lower overt aggression (β = .−02, p < .05)

Figure 2

Interaction between CU traits and general cognition predicts CD symptoms. Simple slopes plotted at mean levels (no CU traits), 1 SD above the mean, and 2 SD above the mean, calculated with an online computational tool (Preacher et al., 2006). Lower levels of general cognition were more strongly related to CD symptoms for children with the highest levels of CU traits (B = −0.42, SE = .13, p < .01), than for children with lower levels (B = −0.31, SE = .08, p < .001) or no CU traits (B = −0.20, SE = .03, p < .001)

Figure 3

Propensity score matching procedures generate CU+ and CU− groups that do not differ significantly on any sociodemographic or methodological variables. (A) Matching procedures meant that absolute standardized mean differences between groups were rendered <.10. (B) Matching on covariates improved similarities between distributions of propensity scores between matched and unmatched groups

Figure 4

Main effects of CU traits group, general ability, executive function, and learning and memory on externalizing and aggressive behavior in propensity matched models. (A) Propensity Matched CU traits group was related to higher levels of ADHD (β = .40, p < .001), CD (β = .46, p < .001), and ODD symptoms (β = .55, p < .001). Lower general ability was related to higher levels of ADHD (β = −.10, p < .01) and CD symptoms (β = −.09, p < .01). Lower executive function was related to higher levels of ADHD symptoms (β = −.08, p < .05). Lower learning and memory was related to lower levels of ADHD (β = −.09, p < .01), CD (β = −.06, p < .05), and ODD symptoms (β = −.08, p < .01). (B) Propensity Matched CU traits group was related to more (β = .16, p < .001) and relational (β = .13, p < .05) aggression

Figure 5

Effect sizes for the association between CU traits and externalizing problems were significantly larger in size for the propensity‐matched group‐based versus dimensional models for child‐reported overt aggression and parent‐reported CD, ODD, and ADHD symptoms. Significant difference was determined by the overlap of independent confidence intervals (Cumming, 2009)