Personality disorder: Good reasons to reclassify

A good classification is simple, principled and useful. ‘God created the world, I organised it’, said Carl Linnaeus in his library at Uppsala in 1741, in one of his many vain moments. But he was justified in being proud, for he had identified a common system that was applicable to all living organisms. We are light years away from this in psychiatry but there is no reason why we should not aspire. This even applies to personality disorder, one of the areas of psychiatry where nosology hardly rules. The current attitude of most clinicians towards personality disorder is one of embarrassed avoidance, a hold-your-nose condition that we hope will pass by quickly without needing to be engaged, except of course when we need to use the diagnosis as a reason for refusing to treat.

The current classification of personality disorder, unlike Linnaeus’ classification, is complicated, unprincipled and useless. It is disliked by almost everybody, even experts in the field, is not diagnosed clinically even when it is clearly present, and has an evidence base that is patently unacceptable, that of expert opinion only. It is therefore ripe for reform. One of the strongest arguments for a revised classification is the case for replacement of the current categorical system by a dimensional one. There are understandable objections to such a change as clinical decision-making has to be categorical and thresholds are uncertain, but in personality disorder, where decision-making is much less clear and we have an abundance of psychological research to help in identification, this is an excellent place to start.

The DSM-5 (Diagnostic and Statistical Manual of Mental Disorders, 5th Edition) Task Force made a valiant attempt to introduce a dimensional system of classification but, in trying to preserve most of the existing categories of personality disorder, made the classification so complex that it was rejected by the American Psychiatric Association as needing further study, and currently, the DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, 4th Edition) classification remains the active one for clinicians, except in some areas, including the United Kingdom, which are DSM-free zones and use ICD (International Classification of Diseases) only. The new ICD-11 classification is not yet published but has rejected categories in favour of a single dimensional spectrum extending from normal personality to severe personality disorder (Tyrer et al., 2015), and this seems to have clinical utility in that it helps in selection of treatment and prediction of outcome (Sanatinia et al., 2016). The more contentious issue is the replacement of categories by domain traits, which are secondary, not primary, diagnostic descriptors.

The domains have been derived from psychological studies carried out over many years and the findings are robust. The detached (schizoid), negative affective (formerly described as neurotic), antisocial (dissocial), obsessional (anankastic) and disinhibited domains have been delineated clearly and repeatedly. The additional one, schizotypy, is also well described but as it is regarded as part of the schizophrenia spectrum in ICD-10, it is not included among the personality disorders. The DSM-5 classification also included these domains (under the heading of facets), and these fit nicely with the Big Five domains in normal populations.

The ICD-11 and DSM-5 (now in alternative form) classifications appear on the surface to be a long way apart, but on closer examination they are not. Severity levels are also included in DSM-5, but in ICD-11, they become the main classification. Everybody is somewhere on a single dimension – no personality dysfunction, personality difficulty, mild personality disorder, moderate personality disorder and severe personality disorder, so there is no possibility of comorbidity within personality pathology (Tyrer et al., 2015). Personality difficulty is a sub-syndromal condition, not a formal disorder, but still associated with pathology. Once the level of severity is ascertained, further elaboration of the nature of the personality disturbance is provided by the domain descriptors; in more severe disorders, the domain traits tend to be more numerous.

In this issue (Bach et al., in press), further evidence is given to support a common domain structure in DSM-5 and ICD-11. The domain of anankastia is not in DSM-5 but there is clear evidence that it is a coherent and consistent personality domain in mental health populations (Kim et al., 2015; Mulder et al., 2016), and its absence from the DSM-5 domains here is apparent rather than real, as, not entirely satisfactorily, the opposite pole of disinhibition in DSM-5 is compulsivity, essentially another synonym for anankastia. The absence of schizotypy in ICD-11 is also entirely understandable as schizotypy continues to remain within the schizophrenia spectrum in ICD-11, so is not recognised as a personality disorder. The other three domains of negative affective, dissocial (antagonistic) and disinhibited traits are virtually identical.

But one group of conditions does not align itself with these domains – one very important to clinicians in the field – borderline personality disorder. Despite the evidence that these domains are robust, the research findings of people with borderline personality disorder do not pay any respect to this organised system but flagrantly misbehave, crossing over to all the domain traits, but particularly those of negative affectivity, disinhibition and antisociality (Mulder et al., 2016). This occurs at all levels of borderline personality disturbance and is not confined to the most severe. So, whereas the classification of someone with a mild degree of what used to be called Asperger’s syndrome could now be mild personality disorder in the detached domain, for most people with borderline personality disorder between two and four domain traits may be prominent at all levels of disorder.

How to deal with this in the ICD-11 classification or in a revised DSM-5 is currently a matter of debate. Some feel borderline personality disorder is best regarded as emotional dysregulation disorder (this is Marsha Linehan’s preference), and if the old DSM classification existed, it could be a simple task to move it from Axis II to the Axis I disorders. It is sad that no serious discussion took place with experts in personality disorder when Axis II was abolished, because it least had the value of drawing attention to a different area of psychopathology.

But the reality is that borderline personality disorder is a very heterogeneous condition that overlaps other disorders on every side; the many attempts at its subclassification show how unsatisfactory it is to have a single label. Where DSM-5, in its new form, and ICD-11 agree, is that most other personality disorders can be readily accommodated with both systems without the need to retain categorical labels, which have no evidence base and were committee-determined, not empirically derived. But many feel, with some justification, that the loss of borderline will be too much of a wrench for existing treatment and research programmes. We do not love this diagnosis but we feel we cannot let it go without a struggle.

Perhaps, the best solution is to be kind to the term and allow at least two forms of borderline personality disorder linked to the domain traits; borderline-1 as the core of negative affectivity and disinhibition, and borderline-2 as negative affectivity, dissociality and/or disinhibition, as there is increasing evidence of differences in presentation and outcome between forensic and non-forensic populations. Bach and colleagues have done a service in showing that the ICD-11 classification is not a leap into the unknown and the transition from ICD-10 to ICD-11 can be relatively smooth. Furthermore, they show that the loss of the categorical system of classification is a tremor that can readily be accommodated; it is not an earthquake.