The Relationship Between Personality Inventory for DSM (PID-5) Domains and Disruptive Behavior Disorders

Previous research has demonstrated a close relationship between personality and externalizing pathology (EP) (1). Studies exploring this relationship have conceptualized personality from the perspective of the Five-Factor Model of Personality (FFM), and personality disorders (PDs) as defined in the Diagnostic and Statistical Manual (DSM) classification system. Studies utilizing adult samples have reported a close relationship between EP and PDs. For example, Attention-Deficit and Hyperactivity Disorder (ADHD) and PDs are often highly correlated with each other, with co-morbidity rates between 10 and 75% in different studies (2). In particular, antisocial, borderline, paranoid, avoidant and narcissistic PD have demonstrated consistent and substantial co-occurrence with ADHD (3–5). In addition, retrospective research and longitudinal follow-up studies report high rates of antisocial and borderline PD in adult subjects who had ADHD in childhood (2,6,7). Similar to ADHD, a close relationship between conduct disorder (CD) and personality disorder (i.e. ASPD, BPD) has been documented in several studies (8,9). Despite less empirical evidence compared to ADHD and CD, childhood Oppositional Defiant Disorder (ODD) has also been shown to be related to personality disorders in adulthood (i.e. BPD) (10).

However, several empirical questions about the relationship between EP and PDs remain. Limitations of the categorical diagnostic model for PDs in section II of the DSM-5 make it difficult to fully understand the links between PD and ADHD (11). Section II of the DSM-5 defines PDs as distinct categories that are dissimilar to each other and other mental disorders (12). This categorical model has been criticized by researchers and clinicians who note that it is restricted by excessive diagnostic cooccurrence, incompetent coverage of maladaptive personality functioning, and discretionary margins for diagnosis, as well as insufficient scientific support (13–16). In light of these discussions, it was proposed that a dimensional trait model would improve the assessment of personality pathology. In particular, the dimensional FFM (17) received the most empirical support (16,18,19) in this regard. Ultimately, the DSM-5 retained the categorical diagnostic model for PDs in Section II but also defined an alternative diagnostic system for PDs in Section III which included a dimensional model of pathological personality traits developed in the context of the FFM (12). According to the proposed alternative model in section III, PDs are characterized by impairments in personality functioning (Criterion A) and pathological personality traits (Criterion B). Criterion A states that an individual must demonstrate functional impairment in the domains of self (identity or self-direction) and interpersonal (empathy or intimacy) functioning, and Criterion B states that an individual must possess maladaptive personality traits. There are 25 trait facets specified in Criterion B of the alternative model for personality disorders (AMPD) and these facets are organized under 5 higher-order domains. These trait domains include antagonism, psychoticism, disinhibition, negative affectivity, and detachment (12). To assess these 5 trait domains and 25 facets described in the AMPD, the DSM-5 personality disorder workgroup developed and published a 220-item self-report measure, the Personality Inventory for DSM-5 (20). The maladaptive trait domains of PID-5 have been demonstrated to broadly correspond to FFM domains as expected (21–23). To address the need for improved clinical utility of the AMPD and the PID-5, the APA also published a 25-item, brief form of the PID-5.

In light of the limitations of the categorical approach to assessing PDs, several studies have examined the relationship between EP and dimensionally measured personality traits, reporting a close relationship between EP and several domains of the FFM (1). Although results have varied across studies, lower extraversion, agreeableness and conscientiousness, and higher neuroticism are most consistently reported to be related to ADHD in adult studies (24–27). Similarly, externalizing behaviors are related to higher neuroticism and lower agreeableness and conscientiousness scores in previous studies (28–30). However, the relationship between EP and the maladaptive trait model proposed in the AMPD has not been studied in detail and all prior research on this subject has been conducted with adults. Smith et al. (2017) examined the presence of categorical PDs, as defined by section II of the DSM, and the presence of pathological personality traits, as defined by the AMPD and the PID-5, in adults with a history of an ADHD diagnosis. Using self-report scales of current and childhood ADHD symptoms, the authors found that in addition to several PDs, current and childhood ADHD symptoms were correlated with almost all facets and domains of the PID-5 (31). Further, Smith and Samuel investigated whether the PID-5 facets predicted current ADHD symptoms using linear regression analyses and found that distractibility was the only PID-5 facet that predicted current ADHD symptoms (32).

The relationship between maladaptive personality traits of AMPD and internalizing/externalizing disorders has also been investigated in an undergraduate student sample (33). Externalizing behaviors (i.e. aggression, alcohol and substance abuse, antisocial behavior) were found to correlate with almost all PID-5 trait domains, however, the antagonism and disinhibition domains were the most consistently correlated. Additionally, regression analyses revealed that only the disinhibition and antagonism domains predicted externalizing symptoms (33).

A significant gap in the literature examining the relationship between EP and the AMPD is that no studies have utilized an adolescent sample - a developmental period during which personality organization maturates and personality pathology often onsets (34,35) and emotional and behavioral problems increase (36). Against this background, we aimed to explore the relationship between EP and the trait domains of the AMPD using the PID-5-BF. We aimed to examine the correlations between the PID-5 domains and disruptive behavior disorder (DBD) symptoms as an indicator of the clinical severity of EP. The term DBD originated in the DSM-4-TR and denotes a group of diagnoses (i.e. ADHD, ODD and CD) which can be used as an indicator of externalizing symptoms (37). Although this definition does not exist in the current classification system of the DSM-5, we used it to capture the symptoms of these three disorders and serve as an indicator of overall EP in the current study. Considering previous studies on this topic in samples of adults, and the description of the trait domains in the alternative model, we hypothesized that DBDs would be most significantly related to the antagonism and disinhibition domains of the PID-5-BF.

The high correlation rates observed between externalizing disorders and the antagonism and disinhibition domains of the AMPD in previous studies have raised questions about the discriminant validity of criterion B of the AMPD in distinguishing between personality pathology and externalizing disorders. The trait facets that describe the antagonism and disinhibition domains in the AMPD are very similar to the diagnostic criteria for DBDs. For example, the descriptions of the manipulativeness, deceitfulness and callousness facets have very similar features to CD symptoms (i.e. often lies to obtain goods or favors or to avoid obligations, has been physically cruel to people or to animals); questions on the PID-5 which make up the hostility facet are very similar in content to ODD symptoms (i.e. often argues with authority figures, often actively defies or refuses to comply with requests from authority figures or with rules, often deliberately annoys others); and questions of impulsivity, distractibility and risk-taking on the PID-5 are very similar to ADHD symptoms (i.e. often easily distracted by extraneous stimuli, often blurts out an answer before a question has been completed, often has difficulty waiting for his or her turn, often interrupts or intrudes on others) described in the DSM-5. Considering these content similarities and the high correlations between DBDs and PID-5 domains reported in previous studies, we aimed to test whether the antagonism and disinhibition domains provided incremental utility in predicting subjects’ total number of overall symptoms (as an index of severity of psychopathology), beyond the subjects’ DBD symptoms. We hypothesized that these domains of the PID-5 would not make a significant contribution to the prediction of the total symptoms of subjects over DBD symptoms.

We examined the relations between the PID-5 domains and DBDs using bivariate and multivariate approaches. Using a bivariate approach, the antagonism, disinhibition and psychoticism PID-5 domains demonstrated significant correlations with DBD scores (Table 1). Using a multivariate approach, hierarchical regression models found that the antagonism and disinhibition domains of the PID-5 did not demonstrate any incremental validity over total DBD symptoms and CBCL scores in predicting overall psychopathology (p>0,05) (Tables 2 and 3).

This study investigated the relationship between DBD symptoms and PID-5 domains in an adolescent inpatient sample, revealing that the antagonism, disinhibition and psychoticism trait domains are highly correlated with DBD symptoms. In addition, hierarchical regression analyses revealed that the antagonism and disinhibition domains - domains that are highly correlated with DBD in previous research as well as the current study - did not make an incremental contribution to the prediction of total symptom scores beyond that provided by DBD symptom scores. This is the first study investigating the relationship between DBD and maladaptive trait domains, as contained in the alternative model for personality disorders of DSM-5, in adolescents.

Our findings demonstrated that antagonism and disinhibition are highly correlated with DBD symptoms. Smith et al. investigated the relationship between ADHD symptoms and PID-5 domains and facets in an undergraduate student sample reporting that all PID-5 domains were correlated with self-report of current ADHD symptom severity, with disinhibition showing the strongest correlation (r=.66) (32). Similarly, another study reported that the antagonism and disinhibition domains showed the highest correlation with externalizing behaviors (i.e. aggression, alcohol misuse, substance use and antisocial behavior), although almost all domains showed correlations with externalizing symptomology to some extent (33). In addition to studies employing AMPD measures, FFM studies also align with our findings. Agreeableness and conscientiousness (which have been proposed as the opposite of antagonism and disinhibition domains of the alternative model respectively) (40) have been shown to be highly and negatively related to externalizing behaviors/symptoms (26,41). Similar findings have been demonstrated in studies utilizing adolescent samples, with agreeableness and conscientiousness significantly associated with several externalizing behaviors (i.e. vandalism, aggression and drug use) (42). This literature indicates that our findings align with prior adult and adolescent research showing that antagonism and disinhibition domains are closely linked to externalizing pathology.

In addition to disinhibition and antagonism, our findings showed that DBD scores were also correlated with the psychoticism domain of PID-5. This finding is consistent with the study of Smith et al. which found that although all domains of the PID-5 were significantly correlated with ADHD scores, the psychoticism domain was among the most highly correlated domains (r=.47) (32). Sleep et al. reported similar results, finding that the psychoticism domain was highly correlated with all externalizing behaviors measured including aggression, alcohol misuse, substance use and antisocial behavior (33). The somewhat unexpected strong correlation between psychoticism and externalizing pathology, in previous research as well as the current study, might be related to a lack of acceptable discriminant validity demonstrated by the psychoticism scale, as stated in previous research (32). Supporting this explanation, existing research indicates that the psychoticism facets correlate very highly with each other and have the highest correlations with other domains of the PID-5 (43). Furthermore, psychoticism has been found to be positively correlated with a large number of conceptually unrelated constructs including depression, anxiety, externalizing and antisocial behaviors, substance use and health concerns (44,45). Thus, the findings obtained from our study and previous similar ones support the proposal that PID-5 psychoticism relates to nearly all forms of psychopathology and may lack discriminant validity. This will be an important area for ongoing research on the construct validity of the PID-5.

Taken together, these findings are consistent with personality psychopathology spectrum approaches (see, e.g., (46) for a review; (31,33,47)), which form the basis for the Hierarchical Taxonomy of Psychopathology (HiTOP; (48)). HiTOP integrates personality pathology into hierarchically organized spectra along with all other forms of psychopathology, tying internalizing, externalizing, and psychotic spectra to the temperamental dimensions that underlie them (49). The HiTOP consortium claims that their model finally gives personality the prominent role it deserves in psychiatric nomenclature (50). However, if maladaptive traits assess nothing more than internalizing and externalizing behaviors, the question arises as to whether the assessment of maladaptive personality traits has any value beyond what is already captured in measures of DBD (or anxiety and depression on the internalizing dimension). To this end, we assessed whether the maladaptive traits of disinhibition and antagonism provided additional explanatory utility beyond the variance explained by DBD symptoms, with total psychiatric severity as the outcome. Results indicated that the Antagonism and Disinhibition domains did not make an incremental contribution to total symptom scores over and above the contribution of DBD scores. This finding is intriguing and raises some concerns about the discriminative validity of these PID-5 domains for externalizing disorders. We would expect both the antagonism and disinhibition trait domains to be related to a variety of psychiatric symptoms, as maladaptive personality traits and psychopathology are known to be strongly related. It would then appear that assessment of either maladaptive personality traits or DBD may be redundant. The HiTOP consortium would argue for the total elimination of the categories as represented in the DBDs, to be replaced by an externalizing dimension at one level of the hierarchy and the maladaptive trait domains of antagonism and disinhibition at another. In this model, all psychopathology is personality, and symptoms of psychopathology are reconceptualized as personality traits. This study cannot provide insight into whether such a reorganization of DBD symptoms would provide improved clinical utility; however, our findings do confirm redundancy and a need for clarification as to whether removing symptoms of psychopathology in favor of personality traits would provide better patient outcomes and a more rigorous conceptualization of psychopathology.

The findings of this study should be regarded cautiously due to several limitations. Instead of doing our research on members of the general community, we chose a sample of psychiatric inpatients. The sample was selected from a heterogeneous and relatively treatment-resistant inpatient population, which limits the generalizability of the data to the general population or uncomplicated cases. In addition, we did not adjust for previous medical conditions or prescription medications that may have influenced the results. Additionally, even though the utilization of an interview tool for assessing psychiatric symptoms (DISC-IV) and a self-report tool for maladaptive traits (PID-5-BF) may be seen as a strength of the study (i.e. reducing method variance), the use of different methodologies for the assessment of psychopathology and personality may have introduced unknown bias. Despite these limitations, the current study provides incremental knowledge on the ongoing debates regarding the usefulness of categorially defined disorders (such as DBDs) as well as the usefulness of reformulating our diagnostic system as a variation in personality traits.