Background
The long-term prognosis of girls with severe conduct problems treated in mandatory closed treatment institutions is poor [
1]. Adolescent girls diagnosed with a disruptive behaviour disorder (DBD) show negative outcomes in adulthood, such as early pregnancy, social isolation, personality disorders, unemployment, psychiatric co-morbidity and substance abuse [
2,
3]. Current treatments are not always effective or focused on females. Understanding the specific characteristics and etiopathology of female DBD may foster specific interventions for females.
Disruptive behaviour disorder has been linked to attenuated activation of the main stress regulation systems: the Hypothalamic–pituitary–adrenal axis (HPA-axis) and the autonomic nervous system (ANS) [
4]. The link between DBD and these systems is explained by the low arousal theory. According to this theory, individuals expressing conduct problems are characterized by low arousal levels, due to the lack of a physiological stress response, which may lead to individuals not fearing the negative consequences of their behaviour [
5]. Alternatively, low arousal may lead to sensation-seeking behaviour in order to increase the unpleasant low arousal to normal levels [
6]. Indeed, several studies in males demonstrate reduced levels of HPA-axis and ANS activity in samples with DBD or externalizing behaviour [
4]. For example, Popma and colleagues [
7] studied a sample of delinquent male adolescents (aged 12–14 years) and revealed that adolescents with DBD had a lower cortisol awakening response compared to controls without DBD. Regarding the ANS, a consistent finding in males with externalizing behaviour is a decreased heart rate (HR, a measure of both parasympathetic and sympathetic activity) and pre-ejection phase (PEP, which is a measure of sympathetic activity) in resting condition, and a heightened respiratory sinus arrhythmia (RSA, a measure of parasympathetic activity), also in resting condition(e.g. [
8‐
10]). These studies provide support that low arousal, reflecting fearlessness or sensation seeking, may be a neurobiological correlate in adolescent males with externalizing behaviour.
However, research on stress regulation systems in relation to conduct problems in girls is sparse, mainly because of the low prevalence of female DBD [
11]. The relatively small amount of research conducted in girls with externalizing behaviour problems provides inconclusive results. Pajer [
12] studied non-referred adolescent females aged 15–17 years with conduct disorder and found diminished salivary cortisol levels in girls with DBD compared to controls, similar to the findings in boys. Likewise, Platje et al. [
13] found decreased cortisol levels in girls from the general population aged 15–17 years with externalizing behaviour problems. In contrast, the study of Dorn et al. [
14] found no significant associations between low arousal and conduct problem in girls aged 6–11 years, as their cortisol output was similar to those in healthy controls. Furthermore, with regard to the ANS, the relationship between DBD and low arousal in females remains disputable. A meta-analysis in children and adolescents aged from 3 to 18.5 years by Ortiz and Raine [
4], suggests that low resting heart rate is diagnostically specific for both males and females with antisocial behaviour. Despite newer studies, more inconsistent findings are added to the literature regarding ANS activity and female DBD. A more recent study of Beauchaine and colleagues [
15] found that aggressive girls show similar autonomic response patterns to stress as normal control girls do. Also Aults et al. [
16] demonstrate that in aggressive adolescents mean age 12.4 years, from the general population, females show different autonomic reactivity than boys.
Possible explanations for the inconclusive results, besides the sparse studies of females, are large differences in sample characteristics, such as age, research population, setting, heterogeneity of quantifying conduct problems and different assessments of stress—regulation system parameters [
17]. As suggested by Beauchaine [
15], an important possible explanation is the presence of co-morbid internalizing disorders in female aggressive behaviour, as post-trauma psychopathology. Post-trauma psychopathology, including post-traumatic stress disorder (PTSD), has been linked to hyperresponsivity of the ANS, and this hyperresponsivity may “normalize” ANS functioning in the aggressive subgroup [
15]. The inconclusive findings in literature on the relation between externalizing behaviour and functioning of stress regulation systems could therefore result from ignoring comorbid post-trauma psychopathology. Girls with conduct problems have substantially higher prevalence rates of PTSD than boys with DBD [
18‐
20]. However, the prevalence of trauma exposure does not differ between boys and girls with conduct problems, the difference relays in the type of trauma. Females are 3–10 times more frequently the victim of sexual abuse, which is often accompanied by physical and emotional abuse; girls therefore are more often the victims of poly-traumatization [
19,
20]. Hamerlynck and colleagues [
21] studied a sample of detained girls aged 12–18 years, in Dutch juvenile justice institutions, and found that 21% of the girls with severe aggression also demonstrated post-traumatic stress symptoms. Moreover, a positive correlation was found between the number of traumatic experiences and extend of aggressive behaviour. This suggests that trauma exposure and subsequent post-trauma symptoms in girls are related to aggressive behaviour, a core feature of DBD. When investigating the stress-regulation system in samples diagnosed with PTSD, a common finding is decreased basal activity of stress-regulation systems, but often in combination with hyperresponsivity to stress [
22]. Although acute stress causes increased activity of the HPA-axis, which results in elevated cortisol levels [
23,
24], chronic or frequent stress leads to sensitization of the HPA axis. In the case of chronic stress, negative feedback mechanisms cause a shift of internal predetermined levels [
25], which results in reduced physiological function at rest and hyperreactivity to stressful situations [
24]. Indeed, King et al. [
26] demonstrated significant lower morning saliva cortisol levels in a group of sexually abused young girls (aged 5–7 years) compared to a control group of community children. A review of PTSD in children and adolescents age ranging from 6.4 to 15.9 years demonstrated alteration in the sympathetic ANS system, which results in elevated HR in samples with PTSD [
27]. El-Sheikh and Hinnant [
28] revealed that girls who experienced more relational stress over time demonstrated decreased RSA while at rest and higher RSA reactivity to stress compared to boys. As such, the study confirmed that stress-regulating systems in girls may respond differently to chronic stress than those of boys.
Thus, the relation between decreased activity of the stress regulation systems and externalizing behaviour problems is well established in males; however, this relation is less clear in females and post-trauma psychopathology may influence this relation. Therefore, the present study aims to investigate the relation between the main stress-regulating systems and externalizing behaviour in girls and, subsequently, the extent to which traumatic stress symptoms mediate this relation. We hypothesize that female adolescents with conduct problems report more post-trauma symptoms and no difference will be found between their stress regulation system and that of female adolescents without conduct problems and post-trauma symptoms.
Discussion
In this study, the relation between the two main neurobiological stress-regulating systems and conduct problems was investigated in a sample of adolescent females admitted to a mandatory closed treatment institution. Subsequently, the possible mediating role of post-trauma symptoms in this relation was tested. The findings confirmed that girls with DBD express higher rates of post-trauma symptoms than those without DBD. Furthermore, a direct positive relation between ANS activity and externalizing behaviour problems in female adolescents was found, while this was not present for HPA-axis activity. Finally, while post-trauma symptoms had a strong effect on externalizing behaviour problems, these symptoms had no mediating effect on the relation between the HPA-axis and ANS activity, or on externalizing behaviour problems in girls.
The finding that female adolescents with conduct problems express higher rates of post-trauma symptoms concurs with results from previous studies on this topic (e.g. [
18‐
20]). Moreover, in the present study, post-trauma symptoms were positively related to externalizing symptoms, specifically the post-trauma sub dimensions: anger, depression, dissociation and sexual concerns. These results are consistent with previous findings of girls in juvenile justice institutions [
21] find that at least 80% of female adolescents treated in juvenile justice institutions experienced one or more traumatic life event. Likewise, they also found a relation between traumatic life events and aggressive behaviour. These results have important clinical implications for the treatment of girls with externalizing behaviour problems in closed treatment settings. As post-trauma symptoms are a frequent finding in these girls and, as trauma exposure also relates to conduct problems, accurate assessment and specialized interventions for trauma symptoms are needed.
We did not find a decreased activity of stress regulation systems, i.e. HPA-axis activity or ANS activity, in girls with externalizing behaviour problems. Instead, externalizing behaviour symptoms correlated to an increased activity of the ANS system, expressed in a high HR and low PEP [
43]. A possible explanation for this finding may be that previous research on this topic examined male samples in non-residential settings. In the meta-analysis of Raine and Ortiz [
4] on the relation between ANS activity and externalizing behaviour problems, only 8 of the 40 studies included female participants. Five of the studies that included females found a relation between reduced heart rate and disruptive behaviour disorder, while all these studies used samples from the general population. The remaining three studies, which did not find any relation between ANS and externalizing behaviour problems, were performed within clinical settings [
44].
Previous research that reported a decreased HPA-axis activity in girls with externalizing behaviour problems was conducted in non-clinical settings [
12,
13]. The current findings indicate that low arousal may not be the underlying etiopathology for externalizing behaviour in severe clinical samples of females with DBD, such as our sample from a mandatory closed treatment institution. It is possible that the low arousal theory [
5,
6] only accounts for the specific sub forms of externalizing behaviour.
In our sample, externalizing behaviour problems were associated with more comorbid post-trauma symptoms. However, post-trauma symptoms do not mediate the relation between ANS activity and externalizing problems. It is known that individuals who have experienced traumatic events react with aggressive behaviour to threat-based stimuli [
20]. This form of aggression is impulsive and is accompanied by hyperarousal of the stress system. It is, furthermore, linked to early traumatic life experiences [
45,
46]. This may be reflected in the current results, in which heightened activity of the ANS system and a high level of post-trauma symptoms were found. Proactive aggression, however, is non-impulsive—rather, it is calculated [
47]. Core features of pro-active aggression are high levels of callous unemotional traits and hypo arousal of the stress system [
48‐
52]. Additionally, it is linked to life-long, persistent anti-social behaviour, with an onset in early youth [
53‐
55]. Future research investigating the low arousal theory should take these different forms of externalizing behaviour (proactive and reactive aggression), as well as post-trauma symptoms into account.
The findings in this study should be interpreted in the context of certain limitations. First, it should be noted that the sample providing ANS measurements was significantly smaller than the initial sample at the start of the study. Likewise, the exclusion of a substantial number of CAR measurements due to artefacts in the saliva collection could have influenced the results. The CAR is influenced by the menstruation cycle, anticonception use and puberty status. In this study, we reported the contraceptive use and whether the girls were menstruating at the time of sampling. However, the differences in contraceptive use or menstruation cycle vary too greatly to be taken into account. Furthermore, smoking and medication use has its influence on the cortisol and heart rate levels [
41]. Acute effects of smoking on heart rate and cortisol measures were ruled out by instructing girls not to smoke within an hour before testing, data from girls who did smoke against our instructions were excluded from the analyses. However, possible long-term effect of regular smoking on heart rate and cortisol cannot be ruled out. Finally, we were not able to correct for medication use, due to imprecise collecting of medication use because we were fairly interested in cortisol containing medication.
On the other hand, t test analyses were performed with sufficient power to draw conclusions. Unfortunately, no power analysis is performed for the path analysis. Second, the study had a heterogeneous sample with an age range from 12 to 18 years. Adolescence is characterized by major behavioural and biological changes, also in the HPA-axis functioning and its relation with externalizing behaviour [
13]. Future research should therefore focus on a more homogenous sample with regard to age and pubertal development, or it should perform subsequent sampling in girls during their adolescence. Lastly, the present study investigated the effect of post-trauma symptoms; however, the subscale post-traumatic stress symptoms revealed nothing significant. This can be clarified, since the post-traumatic stress symptoms measured by the TSCC was initially designed to measure sexual abuse and single traumatic events [
33]. However, repetitive or complex traumatic events, such as neglect, have a higher prevalence rate and can alter a person’s psychobiological development in critical periods [
56]. In addition, experiencing complex trauma can lead to complex PTSD, which differs in symptomology from PTSD [
57]. Subsequently, the timing of the onset of the traumatic experience influences the HPA-axis, as recent trauma is related to increased cortisol output.