Is It Time to Rethink Borderline Personality Disorder?

New research shows overlap between borderline and other personality disorders.

Posted Dec 18, 2018

A running debate in abnormal psychology and psychiatry is whether there really is a diagnosable entity of borderline personality disorder. From the first proposal of this term to reflect, literally, the “border” between neurosis and psychosis, shifts in diagnostic thinking have focused on the deficits this disorder involves in emotion regulation, sense of self, and ability to negotiate boundaries with others. The American Psychiatric Association’s Diagnostic and Statistical Manual, 5th Edition (DSM-5), published in 2013, retained the personality disorders as categorical types instead of, as was expected, using an empirically-based dimensional system. Borderline personality disorder would have remained an entity in this new system, but mental health professionals making diagnoses would also specify an individual's ratings on a set of associated personality traits. Based on empirical approaches that regard personality as reflecting individual variations along central dimensions, or facets, critics of the categorical system had argued that this change was long overdue. The compromise reached with the DSM-5 was that a “beta” version of a dimensional rating system would be tested until it was clear that the diagnostic shift was indeed justified.

If you there are people in your life who have been diagnosed with this disorder, or if you yourself have received such a diagnosis, the idea that borderline isn’t a discrete entity but instead can range in magnitude and expression might resonate well with you. By its very nature, personality doesn’t easily fit into pigeonholed categories. Even the most respected framework for understanding personality, the Five Factor Model, doesn’t place people into groups based on a single trait. Although there’s a temptation to say that someone is “neurotic,” or “conscientious,” the proper application of the Five Factor Model defines an individual’s personality in terms of high and low scores on all five of its ratings scales. The even more refined version of the Five Factor Model uses a total of 30 scores (six “facets” of the five traits) to capture an individual’s personality with all of its nuances and subtleties. How, then, can it be realistic to group people with the maladaptive personalities associated with a personality disorder into clear-cut groups? It’s handy to have those terms, and clinicians have found the types to provide them with a lexicon they can use in their practice, but the question remains as to whether those terms have legitimacy.

University of Toronto’s Carolyn Watters and colleagues (2018) proposed that the entire system of diagnosing personality disorders needs a data-based reboot that determines how much overlap there is among them, and whether they are actually distinct entities. To test out their proposition, the Canadian team completed the ambitious task of scouring the available literature for evidence on the validity of personality disorders as categories, even obtaining unpublished findings from the prominent researchers in the field.

The compromise solution in the DSM-V, called the Alternative Model for Personality Disorders (AMPD), rests on the assumption that there is an overall level of personality functioning reflected in the categorical diagnosis but that there are 25 separate facets which present unique configurations of what would be six major personality disorders. The personality disorders in the AMPD include antisocial personality disorder (ASPD), avoidant personality disorder (AVPD), borderline personality disorder (BPD), narcissistic personality disorder (NPD), obsessive-compulsive personality disorder (OCPD), and schizoid personality disorder (SPD). Watters et al. used the data from as many studies as they could obtain to determine whether the patterns of scores shown by participants across a variety of samples would support the existence of the six original personality disorders remaining in the AMPD. In other words, would a pattern of scores emerge that would suggest those six categories are distinct enough, and distinct enough from each other, to remain in the revised DSM formulation? After all, the purpose of having the six types would be to reduce diagnostic confusion and tighten the entire system. However, if those six types aren’t “real,” then there should be no discernible patterns in those 25 facets.

From the 25 independent samples the authors obtained reflecting from 7,000 to 10,000 participants per personality disorder category, University of Toronto researchers were able to determine just how well the proposed facets led to distinctions among the six major personality disorder entities. Just as importantly, they were able to look for personality traits that showed significant relationships with more than one personality disorder. Historically, the problem with categorical diagnoses has been this matter of overlap. If two or three disorders all share one or more definitional qualities, are they really justifiable as distinct categories?

Looking at the personality traits that defined each of the personality disorders in the AMPD, the 25 qualities clustered into the five sets (with sample items included here) of negative affectivity (NA; anxiousness, depression, insecurity), detachment (DET; intimacy avoidance, suspiciousness, withdrawal), antagonism (ANT; deceitfulness, hostility, grandiosity), disinhibition (DIS; impulsivity, distractibility, risk-taking), and psychoticism (PSY; eccentricity, unusual beliefs). For a personality disorder category to “hang together” means that the traits it’s supposed to reflect also cluster together and don’t show up as correlated with another category. The “average proposed,” then, should be higher than the “average nonproposed.” 

BPD, which has often taken the brunt of criticism with regard to categorical rating systems, proved in this analysis to be, indeed, a weak basis for categorization. The only diagnostic category to receive both significant “average proposed” and “average nonproposed” correlations across facets was BPD. Because it received significant correlations with all of the facet scale scores, BPD could not be uniquely identified by any of the personality dimensions. In the words of the authors, “It cannot be ignored that the overall correlation magnitudes highlight the well-established problem of excessive co-occurrence between BPD and other forms of personality pathology.” Furthermore, the pattern of results suggests that formal diagnosis of BPD could benefit from adding a symptom that reflects “cognitive and perceptual dysregulation and suspiciousness.”

One of the limitations of their analysis, as noted by Watters et al., was their self-report measure of the personality facets might not have been ideal. However, balancing this off were the strengths of the studies included in the analysis not being limited to published findings and to having used a broad range of samples, including those from the community and those from clinical populations. The charges that BPD was a particularly weak player in the entire system seem justified by its very poor, in technical terms, “discriminant validity.”

To sum up, none of the personality disorders proposed to remain as categorical types came out as completely healthy. However, BPD seemed particularly weak as a justifiable category. All of this begs the question of whether it’s worth having personality disorders at all in a diagnostic rating system. If personality trait scores do the job, why bother with these potential sources of error? Furthermore, if people can have more than one PD diagnosis at a time, how can treatment be targeted toward the “correct” diagnostic category? If you are someone looking for help for yourself or for someone close to you, the Toronto study suggests it may be worth digging below the surface of a “type” and looking more realistically at the qualities that comprise an individual’s entire personality.


Watters, C. A., Bagby, R. M., & Sellbom, M. (2018). Meta-analysis to derive an empirically based set of personality facet criteria for the alternative DSM-5 model for personality disorders. Personality Disorders: Theory, Research, and Treatment. doi 10.1037/per0000307