Article Text

Download PDFPDF

9% of a US community sample had a DSM-IV personality disorder
  1. Tara E Galovski, PhD, Associate Professor
  1. Center for Trauma Recovery
 University of Missouri, St. Louis, USA

    Statistics from

 QUESTION: What is the prevalence and correlates of personality disorders in the community? Is there a higher prevalence in some demographic subgroups?


    Cross sectional survey of participants followed up from a previous study.


    Baltimore, USA; 1997–1999.


    In 1981, 3481 randomly sampled adults were assessed using the Diagnostic Interview Schedule and a proportion were examined by psychiatrists. Approximately 15 years later, surviving participants were re-interviewed. A sample completed personality examinations for this study (n=742); age range 34–94 years; 63% women; 60% white.


    Personality disorders were assessed using a combined DSM-IV and ICD-10 version of the International Personality Disorder Examination. Information was collected about family background, childhood behaviours, general health, academic performance, social activities and impairments. Disorders were clustered using DSM-IV categories (A odd / eccentric, B dramatic / emotional / erratic, C anxious / fearful). Data were weighted to account for unequal selection probabilities.

    Main outcome measures

    DSM-IV and ICD-10 personality disorders.

    Main results

    The estimated prevalence of DSM-IV personality disorders was 9%. The prevalence of disorders is associated with demographic characteristics such as age and gender (see tables on website).


    About one in eleven people in the US community may suffer from a personality disorder.


    We must be cautious when interpreting the findings given the low prevalence of individual disorders. The authors ‘clustered&’ the disorders into DSM-IV categories. Demographic features are correlated with these clusters to provide an objective measure of psychosocial functioning as well as descriptive data. We must consider how well individual measures apply to each cluster when assessing the meaningfulness of this data. For instance, having never been married or ‘divorced or separated’ may be an adequate measure of poor psychosocial functioning for a Cluster B or C individual (in that the lack of a partner causes distress), whereas a Cluster A individual may be more distressed if married and more content if single. Measures can only indicate psychosocial impairment if the individual's status on that particular measure causes significant distress or is a result of functional impairment.

    Personality disorders are generally thought to develop in late adolescence and early adulthood, remaining relatively stable across the lifespan.2 There is no way to ascertain the directionality of the relationships reported in this study. Are there developmental precursors to personality disorders that contribute to high school drop-out rates? Could personality disorder risk factors be identified earlier? Could intervention change the course of these disorders and reduce rates of failed marriages and high school drop-outs? Or, contrarily, does dropping out of high school, getting divorced and so on contribute to the expression of the personality disorder? Finally, are these relationships mediated by the type of disorder present? The presence of multiple personality disorders has the highest risk for poor long-term and therapeutic outcome.3 Do people with multiple personality disorders account for the rate of high school drop-outs and failed marriages? Demographic associations differed across clusters, however, there is also considerable variability within each cluster. Do patterns of association between clusters hold true for each disorder within the clusters? This study provides the groundwork for further exploration of these questions in a patient population.

    Histrionic, borderline and dependent personality disorders are more frequently diagnosed in women, while antisocial personality disorder is more frequently diagnosed in men, even when symptom descriptions are identical.4 Sex differences may be an artefact of stereotypes of femininity and masculinity in the DSM-IV / ICD-10 criteria or clinician bias in applying criteria.5 In this study there was a gender differential, particularly in Cluster A and B. The odds of having a Cluster A or B disorder were 4 or 5 times greater for men than women. Unfortunately, the gender differential per disorder was not reported. It would be interesting to see if one or two disorders within the cluster account for the disparity. Although the study exercised methodological rigor in applying ICD-10 and DSM-IV criteria and had a thorough series of inter-rater checks, it would be interesting to note any gender differences in diagnosis using a dimensional approach.6

    There are several clinical applications of this data. The gender differentials reinforce the need to ‘dig a bit deeper’ and be aware of one‘s own clinical bias when assessing women for disorders in which men historically predominate and vice versa. The prevalence of personality disorders (especially Cluster B disorders) was inversely related to age. Although these results could be explained through attrition, cohort effect or a decline in expression of the disorder, there may also be a decrease in the severity of these disorders with age. Individuals may benefit from treatment or may self-correct maladaptive behaviours over time. This is a spark of hope. One often hears of therapists loathe to take on people with personality disorders, yet even the most challenging of these patients can achieve meaningful change.7

    View Abstract

    Supplementary materials



      9% of a US community sample had a DSM-IV personality disorder
      Tara E Galovski (commentator)

      Samuels J, Eaton W, Bienvenu J et al. Prevalence and correlates of personality disorders in a community sample. British Journal of Psychiatry 2002 June; 180: 536�42.

      Web-only Supplementary Material
      Adjusted odds ratios for DSM-IV personality disorders by demographic characteristics

      [View PDF]


    • For correspondence: J Samuels, Department of Psychiatry and Behavioral Sciences, Johns Hopkins Hospital, Baltimore, USA.

    • Sources of funding: National Institute of Health Grants; Johns Hopkins University School of Medicine.

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.