Introduction

A well-known concept in the field of child psychology and psychiatry is the distinction between internalizing and externalizing behavior that was primarily developed by Achenbach (1966). Internalizing behavior such as anxious/depressive symptoms, social withdrawal, somatic complaints, and traumatic stress affects internal psychological environment rather than the external world (Liu 2003). There is general consensus that episodes of depression and anxiety disorders are classified as internalizing behavior. Somatoform disorders, eating disorders, and tic disorders are discussed as belonging to this category.

Externalizing behavior refers to a group of behavioral problems that manifest in outward behavior, thereby focusing on the negative impact on the environment (Liu 2003). Externalizing behavior consists of disruptive, hyperactive, and aggressive behaviors (Hinshaw 1987). The differentiation between externalizing and antisocial behavior is under discussion (Liu 2003) as some researchers view externalizing behavior to represent a less severe form of antisocial behavior, especially in young children (Shaw and Winslow 1997). Attention-deficit/hyperactivity disorders (ADHD) of the combined and hyperactive/impulsive subtypes can be classified as externalizing behavior (Liu 2003).

There is evidence that parental internalizing problems are associated with child psychosocial problems (Spijkers et al. 2013). Children with externalizing behavior problems such as conduct disorder are more likely to grow up to become delinquent as adolescents, and criminal and violent as adults (Farrington 1997). Externalization behavior is among the most frequent features of incarcerated male adolescents and male young adults (Rösler et al. 2004).

Currently, the concept of internalizing and externalizing behavior has rarely been adopted in adult psychiatry. Accordingly, there are no modifications of the conceptualization and categorization of internalizing and externalizing behavior that refer to adults. Youth Self-Report and the Young Adult Self-Report (YSR, YASR) are valid for diagnosing externalizing and internalizing disorders in children, adolescents, and young adults (Achenbach 1991, 1997). There are no standardized assessment scales for adults.

Internalizing and externalizing behavior in children and adolescents affected with ADHD is subject of multiple studies (Connor and Ford 2012). Studies including adults affected with ADHD (AADHD), however, do not focus or even name these phenomena irrespective of multiple studies for the association with personality traits and the comorbidity of adult ADHD with axis I and axis II (personality) disorders (Jacob et al. 2007).

Neuroticism (revised NEO personality inventory, NEO-PI-R, Costa and McCrae 1992) and Harm Avoidance (Tridimensional Personality Questionnaire, TPQ, Cloninger et al. 1993) are anxiety- and depression-associated personality traits that might reflect internalizing behavior (Griffith et al. 2010).

Adults affected with AADHD have significantly higher levels of the anxiety- and depression-associated personality traits Neuroticism according to NEO-PI-R (Ranseen et al. 1998; Retz et al. 2004; Jacob et al. 2007) and Harm Avoidance (TPQ) (Downey et al. 1996, 1997; Jacob et al. 2007).

Low scores on Agreeableness (NEO-PI-R) and Conscientiousness (NEO-PI-R) that are related to antisocial and criminal behavior (Ross et al. 2009) might reflect externalizing behavior. The published studies are in line with the notion that individuals affected with AADHD score significantly lower in Conscientiousness than in controls (Costa and McCrae 1992; Ranseen et al. 1998; Retz et al. 2004; Sobanski 2006). Novelty Seeking (TPQ) is associated with exploratory activity in response to novel stimulation, impulsive decision making, extravagance in approach to reward cues, and quick loss of temper and avoidance of frustration (Cloninger et al. 1993). Significantly higher scores of Novelty Seeking (TPQ) in adult ADHD are replicated (Downey et al. 1996, 1997; Jacob et al. 2007).

According to the National Comorbidity Survey Replication (NCS-R), 18.6 % of subjects with AADHD have a 12-month comorbidity of depressive episodes (Kessler et al. 2006). The 12-month comorbidity of anxiety disorders is 47.1 % in the subjects affected with AADHD according to the NCS-R (Kessler et al. 2006). Social phobia has the highest 12-month comorbidity (29.3 %) among the specific anxiety disorders in AADHD according to NCS-R (Kessler et al. 2006).

Since personality disorders (PDs) have, due to the age of onset, a much higher relevance in adults than in adolescences, no one raised the question which PDs are related to internalizing and externalizing behavior. Anxious or fearful Cluster C PDs (avoidant, dependent, and obsessive compulsive PD) share symptoms and affection of internal psychological environment with other disorders that are classified as internalizing behavior (Liu 2003). In the Epidemiologic Survey on Alcohol and Related Conditions (NESARC), the third most common PD is Cluster C obsessive–compulsive PD with 19.3 % (GP = 7.8 %) (Bernardi et al. 2012). There is multiple evidence from clinical studies that Cluster C PDs are more frequent than emotional, dramatic, or erratic Cluster B PDs in AADHD (Matthies et al. 2011; Miller et al. 2007; Williams et al. 2010). Emotional, dramatic, or erratic Cluster B PDs share outward behavior that may have a negative impact on the environment with other externalizing behavior (Liu 2003). In the NESARC, the most frequent PDs in AADHD are Cluster B borderline PD with 33.7 % (GP = 5.2 %) and Cluster B narcissistic PD with 25.2 % (GP = 5.7 %) (Bernardi et al. 2012).

In general, there is considerable evidence that parent and adolescent internalizing behavior is associated with more impaired functioning at various levels of the family system (Hughes and Gullone 2008).

The aim of this study was to examine whether internalizing and externalizing behavior occurred more often in AADHD, whether subtypes differed in this respect, and whether they impacted on psychosocial status. Therefore, we tested several hypotheses: (1) whether AADHD in general and inattentive type AADHD in particular are associated with higher scores of the anxiety- and depression-related personality traits Neuroticism (NEO-PI-R) and Harm Avoidance (TPQ). (2) Whether people affected with AADHD in general and inattentive type AADHD in particular have a high comorbidity with depression, anxiety disorders, and anxious or fearful Cluster C PDs. (3) Whether patients affected with AADHD in general and with combined and hyperactive type AADHD in particular have lower scores in Agreeableness (TPQ) and Conscientiousness (NEO-PI-R) and higher scores in Novelty Seeking (TPQ). (4) Whether these patients have higher comorbidity with emotional, dramatic, or erratic Cluster B PDs. (5) And whether both internalizing and externalizing behaviors result in lower psychosocial status in AADHD.

Methods

Participants

In- and outpatients affected with AADHD that referred to the Department of Psychiatry, Psychosomatics, and Psychotherapy, University of Wuerzburg were recruited between 2003 and 2009. The first 372 patients (173 females, 199 males; mean age 33.3 years, SD 10.3) were recruited 2003–2005 (Jacob et al. 2007). The extended sample comprises 910 patients (452 females, 458 males; mean age 34.5 years, SD = 10.2); the second wave was recruited 2006–2011 in an identical fashion.

Inclusion criteria were AADHD according to the diagnostic criteria of DSM-IV, onset before the age of 7 years via retrospective diagnosis, life-long persistence, and current diagnosis (Jacob et al. 2008). Age at recruitment was between 18 and 65 years. Probands affected with substance use disorders underwent detoxification in an inpatient setting. Exclusion criteria were as follows: the symptoms occur exclusively during the course of a pervasive developmental disorder, schizophrenia, or other psychotic disorder or symptoms are better accounted for by another mental disorder (criterion E of DSM-IV). Further exclusion criteria were as follows: IQ level below 80 (Mehrfachwahl-Wortschatz-Intelligenztest, MWT-B <13 points) (Lehrl 1977) and bipolar affective disorder (excluded due to the unsolved problems of differential diagnosis). Inclusion and exclusion criteria are not modified compared with our previously published protocol (Jacob et al. 2007).

The Ethics Committee of the University of Wuerzburg approved the study, and written informed consent was obtained from all patients after procedures and aims of the study had been fully explained.

Measures

Diagnosis of AADHD was made as a four-step procedure (Jacob et al. 2008). First, other physical and mental conditions were excluded that could explain the symptoms more adequately. Mental disorders were assessed with the structured clinical interview of axis I and axis II (SCID I/II) to exclude differential diagnoses and to detect comorbid conditions (Wittchen et al. 1997). The intellectual functioning was assessed with MWT-B (AADHD: IQ mean = 111.5, SD = 14.0) to exclude patients with mild cognitive impairment. Second, AADHD was assessed according to the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) criteria. Informative input from partners, relatives, and friends was also collected. To ensure diagnostic validity, subjects were examined by more than one experienced investigator at least at two time points. In the case of a mismatch of the results, the patient was again examined. Third, diagnosis of childhood manifestation of ADHD was retrospectively assessed in a structured clinical interview. Wender Utah Rating Scale was performed (Wender 1995). Additional information from school report cards/certificates and from parents was included if available, but were not obligatory. Fourth, anamnestic information demonstrates that the symptoms are a lifelong condition and definitely do not have an episodic course.

Internalizing and externalizing behavior was assessed by two different personality trait questionnaires (NEO-PI-R and TPQ) the structured clinical interviews of axis I (Structured Clinical Interview for DSM-IV Axis I Disorders [SCID I]; First et al. 1996) and axis II (Structured Clinical Interview for DSM-IV Axis II PDs [SCID II]; First et al. 1997) disorders.

NEO-PI-R is designed to give measures of the five domains of personality according to the personality model of Costa and McCrae (1992). TPQ follows a biological model of personality traits derived from animal research that was proposed by Cloninger et al. (1993).

Internalizing behavior is assessed by the anxiety- and depression-related personality traits Neuroticism (NEO-PI-R) and Harm Avoidance (TPQ). Higher scores for both personality traits indicate more internalizing behavior. The latter is heritable and related to high serotonergic activity (Cloninger et al. 1993). Along the concept that PDs reflect the extreme ends of normally distributed personality traits (Reif and Lesch 2003), anxious–fearful Cluster C PDs (avoidant, dependent, obsessive–compulsive, passive-aggressive, and depressive PD) were also classified as internalizing behavior. Finally, we classified mood and anxiety disorders as internalizing behavior, but excluded somatoform disorders, eating disorders, and tic disorders, because there is no general consensus for their classification as internalizing behavior.

Low scores on Agreeableness (NEO-PI-R) and Conscientiousness (NEO-PI-R), high scores on Novelty Seeking (TPQ), and comorbidity with emotional, dramatic, or erratic Cluster B PDs (antisocial, borderline, histrionic, and narcissistic PDs) are used to detect externalizing behavior. The concept of Novelty Seeking (TPQ) assumes a dopaminergic dysregulation (Cloninger et al. 1993).

Psychosocial status

Psychosocial status was assessed on the basis of a standardized biographical history of each patient (Jacob et al. 2007). The following conditions were rated with one point each (which were simply summarized to obtain the total score): family status (1 point max.): divorced, or separated, or two or more times married. Education (2 points max.): discontinued, two or more classes repeated. Occupational qualification (2 points max.): unskilled, unemployed. Additional factors were psychiatric inpatient treatment, delinquency, suicidal behavior, and aggressive behavior (one point each). This results in a psychosocial status scale (with values 0–9), where low scores indicate fewer psychosocial problems and thus a better psychosocial status (in our sample: mean 4.0, SD 1.0).

Statistical analysis

Frequencies of comorbid axis I and axis II disorders were calculated for the entire AADHD sample and separately for AADHD subtypes. Personality traits were compared with published German reference values (Ostendorf and Angleitner 2004; Weyers et al. 1998) by standardizing each patient’s score with the appropriate age-, group-, and sex-specific population mean and standard deviation and comparing the resulting standardized scores to zero by the sign test. Differences in personality traits (as measured by NEO-PI-R and TPQ values) between groups (e.g., subtypes) were tested by ANOVA. Prevalence of axis I or II disorders (both specific PDs and “any Cluster B” or “any Cluster C” PDs) was compared between all three subtypes by chi-square test. The relationship between psychosocial status and personality traits, PDs, or axis 1 disorders was investigated by a linear regression model, adjusted for age and ADHD subtype. All reported p values are nominal, uncorrected, and should be evaluated against appropriate levels of significance to account for multiple testing of several hypotheses.

Results

Internalizing behavior

AADHD subjects showed significantly higher anxiety- and depression-related personality traits Neuroticism (NEO-PI-R, p < 0.00001) and Harm Avoidance (TPQ, p < 0.00001) scores than the published German reference values (Ostendorf and Angleitner 2004; Weyers et al. 1998, Table 1). Neuroticism and Harm Avoidance scores were significantly different between AADHD subtypes (both p < 0.0001), but patients affected with inattentive type AADHD were intermediate in Neuroticism and only very slightly higher in Harm Avoidance than in patients affected with AADHD of the combined or hyperactive/impulsive type. We did not perform subscale analyses of Neuroticism (anxiety, angry hostility, depression, self-consciousness, impulsivity, and vulnerability) due to the lack of validity.

Table 1 Personality traits and internalizing behavior
Full size table

The lifetime comorbidity of depressive disorders and anxiety disorders is increased in AADHD compared with the general population (Wittchen et al. 2010; Table 2). Differences in comorbidity with depressive and anxiety disorders between AADHD subtypes were mostly not significant. Patients affected with AADHD of the inattentive type had a similar prevalence of depressive disorders as those with combined type (and higher than hyperactive/impulsive type). The prevalence of anxiety disorders was similar in patients with AADHD of the inattentive type and the hyperactive/impulsive type (and lower than combined type).

Table 2 Axis I/II disorders and internalizing behavior
Full size table

The prevalence of (internalizing) Cluster C PDs is much higher in AADHD patients than in the general population (Table 2). Avoidant Cluster C PD is the third most prevalent PD in AADHD. For most Cluster C PDs, differences between subtypes were highly significant. However, only patients affected with AADHD of the hyperactive/impulsive type had much lower comorbidity with Cluster C PDs, while those with combined and inattentive type AADHD had a very similar comorbidity with Cluster C PDs.

Externalizing behavior

People affected with AADHD show significantly lower scores of Conscientiousness (NEO-PI-R, p < 0.00001) and significantly higher scores of Novelty Seeking (TPQ, p < 0.00001) than the published German reference values (Ostendorf and Angleitner 2004; Weyers et al. 1998, Table 3), while there are no relevant differences in Agreeableness (NEO-PI-R, p = 0.89) between those two groups. Patients affected with combined and hyperactive type AADHD have lower scores in Agreeableness (TPQ), while the hypothesized differences in Conscientiousness (NEO-PI-R) and Novelty Seeking (TPQ) could not be detected. Scores in Agreeableness (TPQ), Conscientiousness (NEO-PI-R), and Novelty Seeking (TPQ) were significantly different between AADHD subtypes, but patients affected with inattentive type AADHD were intermediate in Conscientiousness and Novelty Seeking and only very slightly higher in Agreeableness than in patients affected with AADHD of the combined or hyperactive/impulsive type.

Table 3 Personality traits and externalizing behavior
Full size table

The prevalence of (externalizing) Cluster B PDs is also much higher in AADHD patients than in the general population (Table 4). The most prevalent PDs in patients affected with AADHD were narcissistic PD followed by histrionic PD (Table 3). In comparison, the comorbidity with antisocial PD was relatively infrequent. Combined and hyperactive type AADHD patients had higher comorbidity with Cluster B PDs (with exception of borderline PD) than inattentive type AADHD patients.

Table 4 Personality disorders and externalizing behavior
Full size table

Co-occurrence of internalizing and externalizing behavior

Remarkably, many AADHD patients show both internalizing and externalizing behavior. Only 249 patients (28.4 %) had no PD, while 275 patients (31.4 %) had exactly one PD and 353 patients (40.3 %) had two or more PDs. Of these, 218 patients (24.9 %) had both externalizing Cluster B and internalizing Cluster C PDs. Even more frequent was the co-occurrence of externalizing Cluster B PDs with internalizing mood or anxiety disorders, which were present in 301 patients (34.3 %).

Psychosocial status

The mean score of the psychosocial status in the whole AADHD sample was 4.0 (SD = 1.0). We do not have data of a healthy control group available to judge the effect of AADHD per se. There was no relevant difference between males and females (p = 0.7), but a small, statistically significant effect of age (p = 0.03, with younger patients having more psychosocial problems than older patients, which could plausibly be an ascertainment effect) and AADHD subtype (p = 0.05, with the hyperactive subtype having on average 0.25 psychosocial problems more than the other subtypes). We therefore adjusted all following analyses for age and subtype.

Among the disorders that can be regarded as representing externalizing behavior, the strongest effect on psychosocial score was seen for AADHD patients with antisocial or borderline PD, who reported on average 4.6 and 4.4 more serious problems than compared with 4.0 in those without such a PD (Table 5). In general, AADHD patients who had at least one Cluster B PD had worse psychosocial status of 4.1 than patients who did not have a Cluster B PD (mean of 3.9). Externalizing behavior as measured by Conscientiousness scores, on the other hand, was associated with better psychosocial status. Individuals of the same age and subtype who differ by 44 points in their Conscientiousness score (equivalent to 2 standard deviations in our sample) have on average a difference of 0.26 serious problems on the psychosocial scale.

Table 5 Psychosocial status
Full size table

Internalizing disorders on axis 1 were associated with worse psychosocial status: mood disorders in general (mean psychosocial score of 4.1) and in particular major depression (mean score of 4.1) as well as anxiety disorders (mean score of 4.2), while those without any mood disorders had a mean psychosocial score of 3.9. Among Cluster C PDs, only passive-aggressive PD leads to significantly worse psychosocial score (mean 4.4), while obsessive–compulsive PD was associated with a better psychosocial score (mean 3.8). Higher Neuroticism and Harm Avoidance scores were also associated with worse psychosocial status. A difference of 26 in Neuroticism score (again 2 SD in our sample) is associated with 0.2 more serious psychosocial problems (at identical age and subtype).

Discussion

Our data indicate that internalizing and externalizing behavior is frequent and relevant not only in childhood and adolescent manifestations of ADHD, but also in AADHD. Due to the clinical relevance of personality traits in adults, the assessment of internalizing behavior should include the anxiety- and depression-related personality traits Neuroticism (NEO-PI-R) and Harm Avoidance (TPQ). We confirm earlier findings of significantly higher scores of Neuroticism (NEO-PI-R) and Harm Avoidance (TPQ) in AADHD compared with the published German reference values (Downey et al. 1996, 1997; Jacob et al. 2007; Ranseen et al. 1998; Retz et al. 2004). However, we have to consider that almost all subscales of NEO-PI-R except for Agreeableness significantly differ in AADHD compared with controls. Further, we confirm that internalizing behavior on axis I such as mood and anxiety disorders is very common in AADHD. The co-occurrence of internalizing and externalizing disorders is common in our probands affected with AADHD. Patients with Cluster B PDs have a high comorbidity with mood and anxiety disorders and Cluster C PDs. Although an average difference of 0.5 points in the psychosocial status may seem small, in fact it is not given the crude nature of the scale: an average difference of 0.5 could, e.g., be obtained if every second patient in the PD group would be divorced as opposed to no divorces in the control group.

There is a considerable debate whether PDs are distinct disease entities or extreme variations of “normal” personality traits, following a Gaussian distribution (Reif and Lesch 2003). Interestingly, AADHD subtypes did not influence internalizing behavior in AADHD. Both internalizing and externalizing behavior is relevant from a psychosocial point of view. Neuroticism and Harm Avoidance as well as mood and anxiety disorders are associated with worse psychosocial status in our sample of adults affected with ADHD with a very similar effect size to that of externalizing Cluster B PDs.

Comorbid emotional, dramatic, or erratic Cluster B PDs are most frequent in our AADHD sample. The low comorbidity with antisocial PDs in our study reflects the recruitment in a tertiary clinical referral center. Adult ADHD subtypes are related to Agreeableness and most of the Cluster B PDs. Patients affected with hyperactive/impulsive and combined type of AADHD have an elevated comorbidity with Cluster B PDs that might reflect externalizing behavior. In particular, borderline PD and antisocial PD lowered the psychosocial status. Since all items included in this psychosocial status scale are important life events or serious problems, we consider a mean difference of 0.5 points as clinically relevant. Although an average difference of 0.5 points in the psychosocial status may seem small, in fact it is not given the crude nature of the scale: an average difference of 0.5 could, e.g., be obtained if every second 75 % of patients with a specific PD group would be divorced as opposed to only 25 % of divorces in the control group without this particular PD.

We conclude that AADHD in general is associated with both internalizing and externalizing behavior. This finding is in line with studies that show higher rates of AADHD, internalizing and externalizing behavior in incarcerated juveniles (Armistead et al. 1992; Carswell et al. 2004; Sarris et al. 2000). The affection of internal psychological environment and of the external world, i.e., the occurrence of avoidance and anger, have to be taken into account in future treatment strategies.

Inattentive subtype did not influence internalizing behavior, while combined and hyperactive type had some influence on externalizing behavior. This raises the question whether the association with both internalizing and externalizing behavior might reflect an underlying emotional dysregulation in AADHD. Factor analyses confirm the presence of emotional dysregulation that is one of the accessory symptoms of AADHD described by Wender Utah criteria (Reimherr et al. 2005). The concept of emotional dysregulation has emerged from a content point of view, which was verified post hoc in factor analyses. Neurobiological studies provide evidence for the relevance of emotional dysregulation in AADHD (Musser et al. 2013). Interestingly, medication with stimulants improves core symptoms and emotional dysregulation (Reimherr et al. 2007).

The concept of internalizing/externalizing behavior is based on formal statistical results, especially factor analyses studies (Beelmann and Schneider 2003). Internalizing and externalizing behavior is interpreted post hoc as over-controlling and under-controlling (Esser et al. 2000) and can be regarded as consequences of emotional dysregulation.

The relation between AADHD and comorbid conditions remains unsolved. Usually, the onset of ADHD is earlier than that of other axis I and axis II disorders so that it has to be discussed whether ADHD is a psychological and neurobiological vulnerability factor of these mental disorders. A lifetime history of ADHD is also associated with higher perceived stress that might labilize predisposed individuals to comorbid conditions (Bernardi et al. 2012; Meinzer et al. 2013). Results from recent studies suggest common susceptibility genes in the etiology of AADHD and comorbid conditions such as the brain-expressed GTP-binding RAS-like 2 gene (DIRAS2) and Kv channel-interacting protein 4 (KCNIP4) (Reif et al. 2011; Weißflog et al. 2012). These genes might code for common neurobiological underpinnings of AADHD and related disorders such as cell adhesion molecules (e.g., CDH13, ASTN2) and regulators of synaptic plasticity (e.g., CTNNA2, KALRN) (Lesch et al. 2008) or for the common underlying personality traits Neuroticism and Harm Avoidance such as the functional promoter polymorphism of the serotonin transporter gene (5-HTTLPR) (Landaas et al. 2010).

These data present a replication and extension of a previously published study (Jacob et al. 2007). The initial study is about prevalence of comorbid axis I and axis II disorders as well as personality traits, while the present study is about internalizing and externalizing behavior in an extended sample.

Differences in selection and sampling between our and other studies have to be taken into account. The results for comorbidity of clinical referral studies and catchment studies show distinct differences from each other accordingly. The concept of internalizing and externalizing behavior has several methodological limitations (Esser et al. 2000). Mandatory definitions and standardized psychometric assessment scales are missing in adults. Generally, internalizing behavior is characterized by the following: no typical age of onset, prevalence increases with age, and high probability of remission. However, this does not apply to personality traits and PDs. The factor analytic generation of the concept of internalizing and externalizing behavior is content free from hypothesis. The post hoc hypothesis of under-controlling and over-controlling behavior is probably convincing but not evidence based. There is a tendency to underestimate internalizing behavior compared with externalizing behavior due to the obvious relevance of the latter in interpersonal situations. The personal relevance, however, is often not taken into account.

Taken together, the differentiation of internalizing and externalizing behavior could be an interesting starting point for further clinical, epidemiological, and neurobiological research in children, adolescents, and adults affected with ADHD.