Therefore, it is not surprising that when the ICD-11 personality disorder classification committee met in 2010, they felt that a paradigm shift was necessary. Changing the symptoms within categories or changing the number of categories would be insufficient. Tinkering around the margins would not address the fundamental problems with the ICD-10 classification system. Led by Peter Tyrer, the initial proposal for the ICD-11 classification set out to abolish all categories of personality disorder except for the general description of a personality disorder. Personality disorder was conceptualized along a dimension of severity. So, to qualify for a diagnosis of personality disorder, general diagnostic features such as problems in interpersonal relationships and impaired functioning must be present. The diagnosis could then be further specified as “mild,” “moderate,” or “severe” based upon descriptions of degrees of severity. Assessment of severity was based on the prominence of the abnormal traits and their impact on the individual's social and occupational functioning, as well as the risk to themselves or others. Although some description details have been changed since the initial formulation, particularly an increased emphasis on aspects of self-functioning, the fundamental concept has remained and is now embedded in ICD-11 [the final definitions of severity are outlined in the World Health Organization's ICD-11 website (2018)].

What is new, however, is the concept of personality difficulty which, while not a disorder, is described as “problems associated with interpersonal interactions.” More specifically, the definition refers to “pronounced personality characteristics that may affect treatment or health services but do not rise to the level of severity to merit a diagnosis of personality disorder.” In contrast to personality disorders, personality difficulty is manifested in “cognitive and emotional experience and expression only intermittently or at low intensity” (14). There is some existing data suggesting that the concept of personality difficulty may be clinically useful. The UK National Morbidity Survey assessed 8,400 individuals for personality status and mental health. When personality difficulty was defined as a score of one operational criterion less than personality disorder on a SCID-II nearly half the respondents (48.3%) fulfilled criteria. Having this diagnosis was not trivial; they were significantly more likely to consult their general practitioners, be admitted to a mental hospital, attend a community medical center or see a mental health worker (15). Some are concerned that the term will lead to over-medicalisation of behavior and become referred to as a diagnosis even though it is explicitly stated that it is not one. However, the concept has potential use in the general medical and general practice sphere where clinicians often note that patients with the same physical diagnoses may require quite different treatment approaches due to personality quirks.

With regard to the descriptions of personality pathology, the ICD-11 proposal was quite radical. Rather than compromising by retaining some diagnostic categories, as the DSM-5 proposal did, the model included replacement of all ICD-10 diagnostic categories. This significant shift toward a dimensional descriptive framework was guided by literature review (3) and subsequently informed by field trails (16–18). The field trails produced mixed findings; while the anankastia and detached domains appeared reasonably robust, the negative affectivity and dissocial were less so. Two of the analyses suggested that five trait domains were a better fit for the data (17). While the initial model had four traits which were (a) negative affectivity, (b) dissociality, (c) anankastia, and (d) detachment, after considerable debate, disinhibition was later added. As we shall see, its inclusion may have led to as many problems as it seemed to solve.

Despite being derived independently, the five personality trait domains are similar to the DSM-5 Alternative Model for Personality Disorders (AMPD); the major difference being the anankastia domain in ICD-11 vs. the DSM-5 psychotic domain. This similarity is reassuring as well as allowing DSM AMPD domains to be translated to ICD-11 domains, notably by Bach et al. (19, 20).

The proposal received mixed feedback; most clinicians appeared to like it, but some personality disorder researchers considered it too minimalist and feared dropping established categories (21–23). While most experts believed that personality pathology was dimensional in nature, a majority preferred a hybrid model (a mixture of dimension and categories) (4). Just before the WHO final approval of the ICD-11 representatives from the European, International, and North American Societies for the Study of Personality Disorders expressed their strong concerns (24). The central issue was the loss of Borderline Personality Disorder. Further discussion between some ICD-11 working group members and representatives from the Societies reached a compromise (25, 26). This included specifying a borderline pattern qualifier (which is very similar to the DSM-5 Borderline Personality Disorder diagnosis) and the general definition having more focus on self as well as interpersonal functioning. For a more detailed overview of the process see Tyrer et al. (27).

One problem with such a radical change in the classification system is that the findings of previous studies cannot always be directly translated into the new system. There is, therefore, limited specific evidence of the utility and validity of the ICD-11 classification at this time. Fortunately, the DSM-5 AMPD, which despite not being the official DSM-5 classification has been the most used in recent personality disorder research studies (28), and can be reasonably translated into the ICD-11 system. The similarities are important in allowing the large body of work supporting the DSM-5 AMPD model to be generalized to the ICD-11 personality disorder model. Table 1 shows the alignment between the two models.

While the ICD-11 levels of severity compliment the DSM-5 impairment levels of personality functioning, they also add to them. ICD-11 provides a separate list of explicit emotional, cognitive, and behavioral manifestations to help determine the severity of an individual personality disorder (29). Therefore, severe personality disorder may include stress-related distortions in an individual's situational and interpersonal beliefs which could lead to psychotic-like perceptions. The system also explicitly includes risk of harm to self and others as part of determining personality disorder severity. Finally, the model includes the complexity and pervasiveness of the disturbance as a determinant of severity. Severe personality disorder by definition, affects most, if not all, areas of personality functioning, while mild personality disorder will only involve some areas.

The ICD-11 trait domains also have the advantage of being conceptually aligned with the AMPD trait domains. There is now consistent evidence that both systems converge and largely capture the same descriptive features (20, 30, 31) so that the extensive research on AMPD traits (with the obvious exception of psychoticism) may be generalized to ICD-11 trait domains when it comes to clinical information and guidelines. Bach et al. (19) developed an algorithm to delineate the five ICD-11 trait domains (including anankastia) using the well-established Personality Inventory for DSM-5 (PiD) (32). The ICD-11 five factor structure was supported across US and Danish data. Other populations, including Iranian and Brazilian samples, have replicated the five factor structure (33, 34). Sellbom et al. (35) reported that the ICD-11 traits demonstrated expected associations with categorical personality disorder symptom scores and five factor traits in a sample of Canadian psychiatric patients.

Oltmanns and Widiger (30) produced a scale specifically developed for the ICD-11 trait domains—the Personality Inventory for ICD-11 (PiCD). Evaluation in several studies generally supports the validity of the ICD-11 trait domains. Importantly, the domains converged predictably with external criteria, notably the five factor model. As predicted, negative affectivity is associated with neuroticism, detachment with low extraversion, dissociality with low agreeableness, anankastia with high orderliness, and disinhibition with low orderliness (36). Crego and Widiger (37) used a sample of 323 patients in mental health treatment and reported meaningful and expected convergence of the ICD-11 domains with SNAP and DAPP-BQ. The anankastia domain was noted to have strong predictable relationships with compulsivity measures on both scales, which were not present in the PID-5.

There is limited evidence on the utility of the trait domains. A study by Hansen et al. (38) reported that clinicians rated the ICD-11 slightly more useful than the ICD-10 classification. The final definitions of the ICD trait domains were produced in 2018 so specific evidence of their utility is not available yet. However, the predicted relationships with other measures of personality suggest they may prove useful in clinical practice.

This trait domain has been discussed more than the other domains for two reasons. The first is that it is not a domain in DSM-5 AMPD model. The second is that some studies suggest it is not a separate domain but part of a bipolar disinhibition-anankastia factor. Regarding the first reason; compulsivity, which is roughly equal to anankastia, was originally proposed as a distinct domain in the DSM-5 trait model (39). It was abandoned for reasons which are not clear. The generated list of 37 facets for the DSM-5 AMPD included (lack of) orderliness, (lack of) perfectionism, and (lack of) rigidity, which are aligned with an anankastia or compulsivity domain. However, these facets were subsumed into the disinhibited domain (32). Our review (3) found that anankastia or something similar was reported in most analyses which looked for it, while a disinhibition dimension was less consistently found.

That said, there are legitimate concerns about the relationship of the disinhibited and anankastia domains in the ICD-11 classification. Two studies have reported a four factor solution including a bipolar disinhibition-anankastia domain (30, 31). However, other studies support a five factor solution in which anankastia and disinhibition are two distinct domains (17, 19, 35). The clinical reality may be that complex personality disorder patterns can be characterized by both disinhibition and anankastia (40).

Given the consistency of an anankastia domain, or something like it, in the literature, it may be that the disinhibited domain deserves more scrutiny. Disinhibition may be a more general trait which renders personality pathology, and possibly other psychopathology, more obvious across a variety of disorders. The behaviors associated with rigid perfectionism exist and it is not intuitive for clinicians to diagnose “low disinhibition” or “lack of disinhibition” as a personality description. In the end, clinical utility will probably determine whether both the anankastia and disinhibited domains are helpful, or whether one should be dropped.

The ICD-11 classification allows clinicians to specify a borderline pattern qualifier, which essentially consists of the nine DSM-5 diagnostic features. This coding was a pragmatic solution to objections from many senior researchers around the loss of this historically important construct (24). Nevertheless, the ICD-11 working group felt that the severity and trait model could fully account for the borderline pattern (27). The borderline pattern qualifier appears closely linked to the overall severity of the personality pathology and may be a redundant addition to the classification (41). It is hoped that research will reveal whether the borderline pattern qualifier provides useful or distinct information beyond that provided by the trait domains (26).