Classification & Assessment in Psychiatry · Paper A
ICD-11 personality disorder: dimensional trait-domain model
Think of the old system as a set of labelled boxes on a shelf, one box per personality disorder type, and you had to decide which single box a patient belonged in. ICD-11 threw the boxes away and replaced them with two dials instead. The first dial is a volume dial: how loud is the impairment, from none, to mild, to moderate, to severe? The second is a set of five separate sliders, one for each trait domain. You can push several of them up at once for the same person, because real people are rarely just one thing. The borderline pattern qualifier is the one extra sticker you can still stick on top, kept from the old system because clinicians found it too clinically useful to throw away. The exam trap is reaching for an old box name, obsessive-compulsive personality disorder, for example, when the question is set in ICD-11 terms. The same trap applies to reaching for a DSM-5 label, such as antagonism or psychoticism, when ICD-11 uses its own separate vocabulary that only partly overlaps.
ICD-11 (the World Health Organization's current diagnostic manual) rebuilt how personality disorder is diagnosed. The old approach used named boxes: borderline personality disorder, antisocial personality disorder, obsessive-compulsive personality disorder, and so on. ICD-11 dropped almost all of those named boxes. In their place it uses two separate ratings that get combined. The first rating is severity: how much difficulty does this person's personality cause them, day to day, in relationships and at work? This is scored as personality difficulty, or mild, moderate, or severe personality disorder. Severity is what actually decides whether someone has a diagnosis at all. The second rating is a set of up to five trait domain qualifiers, added on top of the severity rating to describe the flavour of the difficulty. The five domains are negative affectivity (a tendency toward intense negative emotion), detachment (a pull toward social and emotional distance), and dissociality (disregard for other people's rights and feelings). The remaining two are disinhibition (acting on impulse without weighing consequences) and anankastia (rigid perfectionism and excessive control). A person can carry one domain, several domains, or in some presentations none clearly enough to name. On top of both of these, ICD-11 kept exactly one named pattern from the old system: the borderline pattern qualifier, for people with the classic mix of unstable relationships, unstable self-image, and impulsivity. Everything else that used to be a named box, antisocial, obsessive-compulsive, avoidant, and the rest, is gone as a category. It is instead described using severity plus whichever trait domains fit. In practice, this means an exam question that asks you to name a specific old-style personality disorder type under ICD-11 is testing something specific. It is checking whether the reader recognises that type no longer exists as a category; the correct move is to describe it as a severity level plus one or more trait domains instead.
A trainee diagnoses a patient with obsessive-compulsive personality disorder, the same label their consultant would have written ten years ago. Under the diagnostic manual now in use, ICD-11, that category does not exist any more. This is not a minor renaming exercise. It is a genuine rebuild of how personality disorder is diagnosed, and MRCPsych exams test the new system directly. This page builds up, from nothing, what actually replaced the old named categories, and why.
ICD-11 diagnoses personality disorder using two combined ratings instead of one label. The first is severity: personality difficulty, or mild, moderate, or severe personality disorder, based on how much the pattern disrupts a person's self-functioning and relationships. The second is up to five optional trait domain qualifiers, added on top of severity to describe the flavour of the difficulty: negative affectivity (a tendency toward frequent, intense negative emotions such as anxiety or low mood), detachment (a lasting pull away from social and emotional closeness), dissociality (disregard for other people's feelings and rights), disinhibition (poor impulse control and low conscientiousness), and anankastia (a rigid, perfectionistic, controlling style). One further qualifier, the borderline pattern, is kept from the old system and can be added on top of both when it fits.
Here is the trap almost every exam candidate falls into: treating a trait domain as if naming it were the diagnosis. It is not. Severity is what actually decides whether someone has a personality disorder at all. A person can show an extreme-sounding trait, rigid perfectionism, for example, with no distress and no dysfunction anywhere in their life. That person has normal personality variation, not anankastia, however unusual the trait looks on the surface. You should always check impairment before reaching for a domain name, not after.
The old ICD-10 system used named categorical types: borderline, antisocial, histrionic, obsessive-compulsive, and several more. In practice, most real patients met criteria for several of these categories at once, because the categories overlapped heavily rather than describing clean, separate conditions. A dimensional model, rating severity and then describing which trait domains contribute, captures that overlap honestly instead of forcing a single best-fit label onto a messy clinical picture. That is the first principle behind the whole rebuild: severity plus domains describes real patients better than one box ever did.
Work through an assessment in order. First, ask whether there is any functional impairment or distress at all; without it, there is no personality disorder to rate. Second, rate severity: personality difficulty, or mild, moderate, or severe, based on how much dysfunction the pattern causes. Third, identify which trait domain qualifiers fit the pattern, up to five, added on top of the severity rating. Fourth, check whether the classic unstable-relationships-and-self-image picture is present; if so, add the borderline pattern qualifier on top of everything else. A complete formulation always has at least two parts, severity plus domain, never a single word.
Several genuine look-alikes are not personality disorder trait domains at all. A discrete major depressive episode can mimic negative affectivity, but it has a clear onset and course that a lifelong trait does not share. Autism spectrum condition can mimic detachment, but it begins in early childhood with its own specific communication and interest features, not a personality pattern consolidating later in life. DSM-5's Alternative Model uses a similar but separately-labelled system: it calls the disregard-for-others domain antagonism, not dissociality, and it includes a psychoticism domain that ICD-11 simply has no match for.
Diagnosis under ICD-11 starts with confirming functional impairment or distress, then rating its severity, then layering on whichever trait domain qualifiers the presentation actually shows. A clinician does not diagnose from a checklist of named symptoms belonging to one category; instead, they build the diagnosis from these combined ratings, the same way you would work through a real case. Old-style categorical labels, if a patient or an old letter still uses one, are translated into this severity-plus-domain language rather than carried forward unchanged.
For presentations that fit the borderline pattern qualifier specifically, NICE guidance on borderline personality disorder continues to underpin management: structured psychological therapies, such as dialectical behaviour therapy, are first-line, with medication reserved for specific comorbid symptoms rather than treating the personality disorder itself. Across personality disorder more broadly, management follows from the severity-and-domain formulation itself, since the domains identify which specific difficulties, emotional instability, impulsivity, rigid perfectionism, need addressing in therapy, rather than treatment being dictated by an old category name.
Whenever a question names an old categorical personality disorder type and asks for the ICD-11 answer, the safe response is always two-part: a severity level plus the trait domain(s) that fit, never a single renamed box. If the question says DSM-5 or Alternative Model, switch vocabulary to antagonism and remember psychoticism has no ICD-11 counterpart.
ICD-11 replaced named personality disorder categories with a severity rating, up to five trait domain qualifiers, and one retained borderline pattern qualifier, combined together rather than chosen between.