Dimensional Model of Personality Disorder Incorporated Into ICD-11

More than four decades after first being proposed for DSM by Allen Frances, M.D., a dimensional model for personality disorder assessment and diagnosis will become part of an official classification of mental disorders—the International Classification of Mental Disorders, 11th Revision (ICD-11), when it becomes official in 2022.

A hybrid dimensional-categorical model of personality pathology was developed for DSM-5, called the Alternative DSM-5 Model for Personality Disorders (AMPD), but it was placed in Section III of DSM-5. The DSM-IV chapter on personality disorders was left unchanged in Section II of DSM-5.

In this article, I will introduce readers to the new ICD-11 model and compare it with the AMPD. There are striking similarities, but also some critical differences. I will also briefly review the uneven empirical bases for each model. Finally, see the end of this article for links to papers that illustrate how each model is likely to be applied to patients who might have a personality disorder or have at least some clinically relevant personality problems.

Although the ICD-11 model is described as dimensional, it is in reality a hybrid dimensional-categorical model with some resemblance to the AMPD. By a dimensional-categorical hybrid, I mean that patients are evaluated on a dimensional scale, but categorical judgments are made by the application of diagnostic thresholds akin to how hypertension is defined by cut points on scales of systolic and diastolic blood pressures. For the ICD-11, there is a scale of the severity of personality pathology that ranges from “personality difficulty” to “mild,” “moderate,” and “severe” personality disorder, which represent the only personality disorder “diagnoses” in the system. The designation of “personality difficulty” (subthreshold for a personality disorder), as well as the three levels of personality disorder, can all be amplified by these trait specifiers: negative affectivity, detachment, dissociality, disinhibition, or anankastia. A categorical “borderline pattern specifier” is also available, but no other diagnostic subtypes of personality disorder are included in the ICD-11 model.

The DSM-5 AMPD was developed nearly 10 years before the ICD-11 model, although it was not in final form until 2012. Because severity of personality pathology has been widely recognized as its most important aspect, associated with concurrent impairment and predicting many future negative outcomes, the AMPD has at its core the dimensional Level of Personality Functioning Scale (LPFS). Impairment in self (identity, self-direction) and interpersonal functioning (empathy, intimacy) is rated on a 5-point scale from 0 (little or no impairment) through 4 (extreme impairment). A score of 2, or moderate impairment, indicates the presence of a personality disorder, based on a sample of 337 patients, with maximum sensitivity and specificity.

The AMPD also includes a five-domain, 25-facet pathological personality trait rating system to describe the myriad stylistic ways personality pathology may be expressed. AMPD definitions of negative affectivity, detachment, and disinhibition are equivalent to their ICD-11 counterparts. ICD-11 dissociality mirrors AMPD antagonism, and a domain of psychoticism is included in the AMPD to capture schizotypal personality disorder, which is not classified as a personality disorder in the ICD. Anankastic traits (that is, compulsivity) in the ICD-11 model are conceptualized as low disinhibition in the AMPD. Two critical differences between the two approaches to personality traits are that ICD-11 describes personality only at the broad five-domain level, while the AMPD has a more fine-grained, 25-trait facet method. Also, whether rated by clinical interview or assessed using the self-report Personality Inventory for DSM-5 (PID-5), traits in the AMPD model are measured dimensionally according to the degree that each describes the patient, while the ICD-11 trait modifiers are essentially categories that either describe the patient or do not.

Another major difference between the two systems is that the AMPD retains six specific personality disorders for which empirical support or clinical utility was deemed sufficient: antisocial, avoidant, borderline, narcissistic, obsessive-compulsive, and schizotypal. ICD-11 has only the borderline pattern specifier, which was added to the model reluctantly by its creators after strong negative feedback from some professional quarters. The six personality disorders in the AMPD are defined by criteria including disorder-specific impairments in personality functioning (“A criteria”) and selected personality trait facets (“B criteria”) that were found empirically to be most correlated with the DSM-IV personality disorders they were intended to replace. Diagnostic algorithms for each personality disorder also were determined to maximize correspondence with DSM-IV personality disorders, minimize co-occurrence with other personality disorders, and maximize relationships to functional impairment. If a pattern of personality pathology does not correspond to one of the six personality disorders prototypes, a diagnosis of PD-trait specified (PD-TS) may be assigned. The ICD-11 borderline pattern is defined by the nine DSM-IV borderline criteria with a threshold for diagnosis of five, as in DSM-5 Section II.

There is evidence to support the validity of hybrid models of personality pathology in that both indicators of disorder and personality traits have been shown to increment each other in predicting important outcomes. Personality experts preferred an unspecified hybrid model over either purely dimensional or categorical models as a replacement for DSM-IV in 2007. In 2019, a repeat survey of experts continued to prefer a hybrid model, this time the AMPD. Since the publication of DSM-5 in 2013, the AMPD has been widely embraced by large segments of the academic personality field in countries around the world.

A comprehensive review of the research literature published in 2019 found 237 research publications on the AMPD and concluded that the AMPD severity ratings and maladaptive traits had acceptable interrater reliability, high internal consistency and consistent latent structures, substantial convergence with a broad range of clinically relevant indicators, and some evidence of incremental validity over existing PD categories.

Few studies have addressed the ICD-11 model, and many of these are based on archival data and use earlier iterations of the model. As a consequence, the ICD-11 developers have had to rely on data based on the AMPD and the well-established five-factor model of personality, with large unfilled gaps in knowledge about the actual final ICD-11 model. For example, the final ICD-11 personality severity measure has not been the subject of any reliability studies, despite its complex format, which combines 15 to 20 personality disorder features into the rating, and research on the five ICD-11 trait domain definitions has been inconclusive in confirming the proposed five-domain structure.

Although these dimensional-categorical models (especially the AMPD) have spurred much new research on personality pathology, they are intended to replace the flawed categorical approaches of DSM-IV/DSM-5 Section II and the ICD-10 for clinical purposes. Both models have been rated as having greater clinical utility than the standard approaches to describing, communicating about, and treating patients for personality pathology. The AMPD has been shown to predict important clinical needs and outcomes, such as impaired functioning; risks for self-harm, violence, and criminality; optimum treatment intensity; and prognosis.

The personality field is moving (as is most of psychopathology) inexorably toward more dimensional conceptualizations of personality pathology, but clinicians need to be on board for this important shift to finally take hold. ■