Personality disorder is common and shortens life — and is absent from the GBD

The Global Burden of Disease study is the reference framework for how much illness the world carries and where. Personality disorder is not in it. A systematic review and meta-regression published in The Lancet Psychiatry on 3 November assembles the epidemiology that would be needed to put it there [s1].

What was pooled

The review followed PRISMA and GATHER guidelines and included cross-sectional or longitudinal studies reporting prevalence or mortality for any personality disorder using DSM or ICD diagnostic criteria in representative samples [s1]. Personality disorder was defined to encompass paranoid, schizoid, schizotypal, antisocial, borderline, histrionic, narcissistic, avoidant, dependent, obsessive-compulsive, and other specified and unspecified types [s1].

The exclusions matter as much as the inclusions. Studies using self-report instruments were excluded, as were non-representative samples — clinical, incarcerated, or university populations [s1]. Those are the settings where personality disorder prevalence is highest and most often quoted, so excluding them lowers the resulting estimates substantially and makes them population figures rather than clinic figures.

Searches ran in PubMed, Embase, and PsycINFO from 1 January 1980 to 13 February 2024 [s1]. Study quality was assessed using Joanna Briggs Institute critical appraisal tools; 59 of the studies were rated high quality and one moderate [s1]. The study is registered with PROSPERO as CRD42021269169 [s1].

Sixty studies across 28 countries were identified [s1]. Prevalence analyses included 139,373 individuals; mortality analyses included 392,420 individuals with a personality disorder [s1].

The estimates

Pooled prevalence of any personality disorder ranged from 4.1% (95% uncertainty interval 2.8–6.0) in low-income and middle-income countries to 5.2% (3.7–7.3) in high-income countries, at a mid-age of 43 years [s1]. Prevalence was higher among male than female individuals, and decreased with age [s1].

Personality disorder was associated with elevated mortality. Pooled standardised mortality ratios were 4.7 (95% UI 3.7–5.8) for inpatients, 1.8 (1.4–2.2) for outpatients, and 2.2 (1.8–2.9) for inpatients and outpatients combined [s1]. The data indicated moderate diagnostic stability over time [s1].

The heterogeneity is the caveat

Between-study heterogeneity was high for both prevalence (I²=98.8%) and mortality (I²=99.4%) [s1]. Those values mean almost all the variation between studies comes from real differences between them rather than sampling error — and the meta-regression identified where: diagnostic criteria, interviewer type, and region [s1]. Publication bias was identified in the prevalence analyses [s1].

A pooled estimate drawn from studies that disagree that much is a summary of a scattered literature, not a measurement. The uncertainty intervals reported alongside each figure are the honest part of the presentation, and they are wide.

The mortality figures carry an additional structural caveat. Standardised mortality ratios of 4.7 for inpatients versus 1.8 for outpatients [s1] partly reflect who becomes an inpatient, not only what the disorder does — the sickest people are the ones admitted.

The measurement problem, illustrated

A meta-analysis published on 17 November in Personality Disorders makes the diagnostic-method problem concrete for one diagnosis [s2]. It searched MEDLINE, PubMed, PsycINFO, PsycARTICLES, Embase, and Web of Science for English-language publications between 1980 and 2023 reporting original prevalence data for borderline personality disorder in men in the general population, and found 11 eligible studies covering 2,027,764 people [s2].

The results split cleanly by method. Seven studies using clinical interviews reported a pooled prevalence of 0.9% in men, with a men-to-women ratio of 43:100 [s2]. Four studies using self-report measures reported a pooled prevalence of 8.1%, with a men-to-women ratio of 73:100 [s2].

That is a ninefold difference in prevalence and a substantially different sex ratio, produced by instrument choice alone [s2]. Prevalence also differed according to sampling method, participant age, and cultural and economic factors [s2]. The authors note that the variation may be influenced by gender bias in diagnostic criteria and clinical practice as well as by methodological and environmental factors, and call for refined measurement tools [s2].

It also explains one of the Lancet Psychiatry review's design choices. Excluding self-report instruments [s1] is what keeps a nine-fold measurement artefact out of a global prevalence estimate.

What follows

The case the review makes is procedural rather than clinical: personality disorder is common, associated with premature mortality, and diagnostically stable, yet remains excluded from the Global Burden of Disease study, and the synthesis is offered as the foundation for future global estimates and policy responses [s1].

Whether the GBD adopts it is the thing to watch. A condition that is not counted does not appear in the burden rankings that drive funding allocation — which is a separate problem from whether it is treated well.

This article is informational and does not constitute medical advice.

Sources

• [s1] The global epidemiology of personality disorder: a systematic review and meta-regression. The Lancet Psychiatry, 3 November 2025. https://doi.org/10.1016/S2215-0366(25)00299-8
• [s2] A systematic review and meta-analysis of the prevalence of borderline personality disorder among men in the general population. Personality Disorders: Theory, Research, and Treatment, 17 November 2025. https://doi.org/10.1037/per0000756