The evolving nosology of personality disorder and its clinical utility

There has been increasing consensus that the classification of personality disorder in the DSM‐IV and ICD‐10 was no longer fit for purpose. There was no good evidence that there are nine to eleven discrete personality disorder categories, the system was too complex, and most categories were not used. The evidence pointed toward the dimensional nature of personality disturbance, with severity being the strongest determinant of disability and prognosis¹.

It was therefore not surprising that the American Psychiatric Association in the DSM‐5 and the World Health Organization in the ICD‐11 moved toward dimensional models of personality disorder classification. The DSM‐5 Work Group proposed a model that included an evaluation of severity (Criterion A) and a description of 25 traits (Criterion B) which were organized into five domains, as well as six individual personality disorders based on DSM‐IV categories. The proposal was rejected, but published in the DSM‐5 Section III and labelled the Alternative Model of Personality Disorders. Despite not being part of the official classification, the model has acquired an acronym – AMPD – and has received multiple studies evaluating its utility and validity.

The ICD‐11 model also involves a dimensional measure of severity (mild, moderate and severe personality disorder) and a subsyndromal condition called “personality difficulty”. Once severity has been determined, the personality dysfunction can be further delineated using one or more of the five trait domains labelled negative affectivity, detachment, disinhibition, dissociality and anankastia. The model does not retain traditional personality types, with the exception of a borderline specifier².

Research on the AMPD model progressed rapidly once a self‐report instrument, the Personality Inventory for DSM‐5 (PID‐5), was developed. This instrument demonstrated adequate psychometric properties, including a replicable factor structure, convergence with existing personality instruments, and expected associations with clinical constructs³. Contradicting the beliefs of the DSM‐5 Committee that the AMPD model lacked clinical utility, clinicians reported that the model demonstrated stronger relationships to ten of eleven clinical judgments than the DSM‐5 categories⁴.

Due to its more recent development, the ICD‐11 model has received less clinical scrutiny. However, studies generally report good construct validity and test/retest reliability⁵. Five domains also appear to be the best fitting model for traditional personality disorder symptoms, although the anankastia, detached and dissocial domains may be more clearly delineated than the negative affective and disinhibition domains⁶.

It has been documented that the AMPD traits (measured using the PID‐5) can describe the ICD‐11 trait domains⁷. Despite being derived independently, the AMPD and ICD‐11 share four of the five domains; the exceptions are anankastia in the ICD‐11 and psychoticism in the AMPD. Both models show relative continuity with traditional personality disorder categories and capture most of their information. The ICD‐11 model is superior in capturing obsessive‐compulsive personality disorder, whereas the DSM‐5 model is superior in capturing schizotypal personality disorder⁸.

In addition, both models show some continuity with dimensions of personality in the general population, measured using the Five Factor Model. Negative affectivity is linked with neuroticism, detachment with low extraversion, disinhibition with low conscientiousness, and dissociality with low agreeableness. The ICD‐11 anankastia is linked with high conscientiousness, while AMPD psychoticism does not particularly align with any of the five factors⁸.

On the face of it, both new models seem more “true” to the existing evidence about personality pathology than the DSM‐5 official classification. Yet, the most important rationale for making such a paradigm shift – the development and evaluation of treatments – has not yet been subjected to significant study. It should be noted that there is little justification for retaining the old model of personality disorder classification regardless of how the new model performs. Only borderline personality disorder has an evidence base, and this essentially tells us that a host of treatments are similarly effective and none have shown specific efficacy for this disorder as opposed to general psychological distress and dys­function⁹.

Nevertheless, treatment studies using the new classification are urgently needed. A number of frameworks have been put forward which, on the basis of a careful assessment of severity and trait domains, lead to a coherent and holistic formulation which is usually shared with the patient and results in the adoption of a consensual approach to treatment⁹.

A potential problem is the retention of traditional personality disorder categories in both models. In the AMPD model, six individual personality disorders are retained. Since non‐personality disorder specialist clinicians generally only use three diagnoses (borderline personality disorder, antisocial personality disorder, and personality disorder not otherwise specified), a danger is that they will simply continue with their current practice. The ICD‐11 model only retains one personality disorder – the borderline personality disorder specifier – but its inclusion may also compromise the change to more evidence‐based practice. While the old categories have no scientific underpinnings, their familiarity may hinder clinicians embracing the new classifications.

In summary, the changes in the classification of personality dis­order represent the beginning of a paradigm shift in diagnosis. The ICD‐11 and AMPD are reasonably consistent with each other. Both place severity of personality disorder at the centre of diagnosis, as the evidence suggests. Both have dimensional trait domains consistent with models of personality such as the Five Factor Model. Both seem to be understood and preferred by clinicians. It is unfortunate that in both models the need has been felt to cling on to traditional categories. The complexity that this created in the AMPD model may be a part of the reason why it was rejected by the DSM‐5 Committee. The ICD‐11 Committee felt the need to compromise with a borderline specifier in order not to suffer a similar fate².

The ICD‐11 personality disorder classification is now official and will be required to be used in many countries from January 2022. Whether and when the AMPD, or some form of it, becomes official is unclear. It is hoped that clinicians will see the new classifications as useful and that their use will lead to greater understanding of the concept of personality disorder, resulting in better clinical care.

The importance of personality in the treatment of psychiatric disorders (and physical disorders for that matter) is obvious in most studies which have measured it. Yet, personality is often an afterthought in clinical practice, given to patients when things go awry. If personality pathology can be recorded with relative ease (through brief questionnaires and interviews) and we can let go of traditional categories, then it is my view that its utility in planning and predicting the outcome of treatment will become self‐evident.