Criterion and Clinician Bias Against Sexual- and Gender-Minoritized Individuals in the Diagnosis of Borderline Personality Disorder (BPD): Vignette Experiment
Marley Warren, University of Michigan
Abstract
Sexual- and gender-minoritized (SGM) individuals experience elevated diagnosis of borderline personality disorder (BPD) regardless of underlying maladaptive personality traits – suggesting clinician bias. Data from a randomized vignette experiment was used to investigate clinician and criterion bias against SGM individuals in the diagnosis of BPD. Participants (N = 426) – consisting of Ph.D. and master’s students in psychology and social work, current clinical social workers, psychologists, psychiatrists, and medical doctors – were randomly assigned to one of three vignette conditions (a cisgender heterosexual man, a cisgender gay man, and a heterosexual transgender woman) describing a fictional presenting client. Participants selected from a list of primary and secondary diagnoses, indicated their agreement with each BPD criterion, and parts of the Personality Disorder Inventory for DSM-5 (PID-5). Participants assigned the gender minoritized vignette were significantly more likely to diagnose (and agree with the diagnosis of) BPD relative to participants assigned the cisgender heterosexual vignette. No significant differences were found in BPD criterion agreement or PID-5 domains between the three vignettes, suggesting clinician bias. Clinicians should explore connections between patients’ symptoms and minority stress processes, weigh the degree of impairment caused by symptoms, and use more rigorous measures of BPD to reduce stigmatization.
Introduction
Borderline Personality Disorder (BPD) is characterized by three central components: unstable interpersonal relationships, impulsivity, and identity disturbance (APA, 2013). Individuals living with BPD – about 2% of the general population (Ellison et al., 2018) – experience elevated levels of externalizing and internalizing problems including mood, anxiety, and substance use disorders (Ha et al., 2014). Across both nationally representative and clinical samples, sexual and gender minoritized (SGM) individuals are diagnosed with BPD at significantly higher rates (i.e., ratios as high as 2:1) than cisgender heterosexual individuals – even after controlling for the maladaptive personality domains underlying BPD (via the Personality Inventory for DSM-5; Grant et al., 2011; Krueger et al., 2012; Reuter et al., 2016; Rodriguez-Seijas et al., 2021a). The extent to which the inequality in BPD diagnostic rates corresponds with differences in presenting psychosocial distress between SGM and cisgender heterosexual patients is unclear. Holding psychosocial distress and BPD symptomology constant in a vignette experiment, we assessed BPD diagnosis as a function of patient sexual orientation and gender identity.
Several of the DSM-5 diagnostic criteria for BPD appear conflated with minority stress processes among SGM individuals. The criteria efforts to avoid abandonment and impaired interpersonal functioning appear conflated with experiences of SGM-specific rejection sensitivity (i.e., the anxious expectation of future social rejection on the basis of one’s SGM identity; Carlton, 2021; Pachankis et al., 2008). Identity disturbance can similarly be seen as a response to living in a systematically identity-repudiating society. Indeed, the DSM-5 continues to refer to confusion about sexual identity as a diagnostic feature of BPD (APA, 2013). SGM individuals are also more likely than cisgender heterosexual individuals to engage in some impulsive behaviors, such as sexual behaviors that confer risk for negative health outcomes (e.g., unprotected anal intercourse, low frequency of condom use, having more than four sexual partners) and substance use, often concurrently (Cochran et al., 2004; Drabble & Trocki, 2005; Pachankis et al., 2015; Taggart et al., 2019). These behaviors, while potentially self-damaging, may be normative within SGM individuals’ communities (Kashubeck-West & Szymanski, 2008; Pachankis, 2014; Pollard Nadarzynski, & Llewellyn, 2018; Taggart et al., 2019). The suicidal ideation and self-injurious behaviors criterion may also be strongly connected with structural minority stressors (e.g., Rodriguez-Seijas et al., 2022), meaning that SGM individuals may be diagnosed with BPD at higher rates than cisgender heterosexual regardless of underlying personality pathology.
Overall, the DSM-5 diagnostic criteria for BPD may not accurately reflect BPD, maladaptive personality, or even distress among SGM individuals. Practitioners unaware of SGM group norms risk interpreting potentially normative and expectable behaviors as signs of BPD.
Methods
This study utilized data from a randomized vignette-based experiment with a large, diverse sample (N = 426) of mental health providers to evaluate BPD diagnosis as a function of SGM status. Participants – consisting of current psychology doctoral students, doctoral-level licensed clinical and counseling psychologists, licensed psychiatry attendings and residents, licensed clinical social workers, and licensed masters-level mental health counselors – were randomly assigned to read one of three vignettes (a cisgender heterosexual male, cisgender gay male, and a heterosexual transgender female) designed to reflect common forms of psychosocial dysfunction experienced by SGM individuals (e.g., substance use, sexual risk behavior, and identity concealment).
Vignette exposure was the primary independent variable of this study. The primary dependent variable of the study, BPD diagnosis, was assessed four ways; (1) participants selected a single diagnosis from a list of thirteen plausible options, (2) participants optionally selected comorbid diagnoses from the same list, and (3) participants indicated their agreement with the diagnosis of BPD on a Likert- style scale ranging from 0 (strongly disagree) to 3 (strongly agree). (4) Vignettes were also diagnosed with BPD if their Personality Inventory for DSM-5 (PID-5; Krueger et al., 2012) negative affect and disinhibition and/or antagonism latent domain scores were elevated. For each PID-5 item, participants could select from four options on a Likert-style scale ranging from 0 (very false or often false) to 3 (very true or often true). Participants were additionally asked to state their agreement with the nine DSM-5 criterion for BPD on a four-point Likert-style scale.
The final two measures included were the Sexual Orientation Counselor Competency Scale (SOCCS; Bidell, 2005) and the Gender Identity Counselor Competency Scale (GICCS; O’Hara et al., 2013), which measure the professional skills and knowledge participants have about providing SGM individuals with mental health care as well as the personal attitudes they hold about SGM individuals.
Analytic Plan
Binary logistic regressions were conducted in SPSS (version 28, IBM, 2022) to analyze the relationship between vignette group and binary BPD diagnosis. Four contrasts were specified: cisgender heterosexual versus SM treatment, cisgender heterosexual versus GM treatment, SM versus GM treatment, and cisgender heterosexual versus SGM. For post-hoc investigations of any significant results, we performed independent sample t-tests of means between vignette groups.
A one-way MANOVA was used to assess the relationship between vignette group and agreement with each individual BPD criterion (dependent variables). A one-way MANOVA was also utilized to determine if the rate of endorsement of PID-5 latent domains (dependent variables) varied significantly by vignette group.
SOCCS and GICCS subscales were compared within vignette groups via independent sample t-tests. Specifically, we compared clinician competence means between those that received the SM vignette and did not diagnose BPD (group one) against the means of those that received the SM vignette and did diagnose BPD (group two). Finally, binary logistic regression was used to determine if BPD diagnosis varied as a function of SOCCS and GICCS scales and vignette treatment group (i.e., the interactive effects of both).
Results
Regarding BPD diagnosis, participants that received the GM vignette were significantly more likely to select BPD as the primary diagnosis relative to participants that received the cisgender heterosexual vignette (Beta = 0.272, OR = 1.312, p = 0.322), but not relative to the SM vignette. No other significant differences emerged between vignettes in terms of BPD primary, combined, or PID-5-based diagnosis.
Regarding Likert-style BPD agreement, t-tests revealed significantly different means for three of the four contracts specified. No significant differences in mean agreement between participants assigned the cisgender heterosexual vignette (M = 1.53, SD = 1.070) and those assigned the SM vignette (M = 1.63, SD = 1.077) were found, t(266) = -0.730, p = 0.466. Overall, results suggest a GM-specific rather than generalized SGM main effect for BPD
diagnosis and agreement.
Participants agreed with all nine BPD criteria at similar rates, regardless of which vignette they were assigned. Specifically, MANOVA results failed to reject the null hypothesis of homogeneity between all vignettes on all criteria, WIlks’ Lambda = 0.954, F (16, 786) = 1.165, p = 0.290, partial η2 = 0.023, observed power = 0.774. MANOVA results similarly revealed that regardless of the vignette group contrast specified, none of the PID-5 latent domains or trait facets varied significantly in endorsement rate by vignette group.
Regarding clinicians’ skills, knowledge, and attitudes, binary logistic regressions revealed that greater clinician skill was associated with higher rates of BPD diagnosis (primary and combined) whereas greater clinician knowledge predicted lower rates of BPD diagnosis (primary and combined) across most contrasts. The lack of a systematic pattern in which contrasts were significant suggests that particular clinician competency traits did not significantly contribute to disparities in the diagnosis of BPD between vignettes.
In terms of intra-group (i.e., intra-vignette) differences in SOCCS and GICCS scores, only knowledge differed significantly between those that selected BPD as the primary diagnosis (M = 0.7326, SD = 0.2030) and those that did not (M = 0.8228, SD = 0.1709). Both skills and knowledge scores differed significantly between those that selected BPD as a primary and/or comorbid diagnosis.
Discussion
This study utilized a randomized vignette experiment among a large diverse sample of mental health care providers and trainees to investigate clinician and criterion bias against SGM individuals in the diagnosis of BPD. Participants assigned a clinical vignette of a GM patient were significantly more likely to select BPD as a primary diagnosis and agree with the diagnosis of BPD relative to participants assigned a cisgender heterosexual vignette. We found no significant differences in BPD diagnosis or agreement between participants assigned the SM vignette and those assigned the cisgender heterosexual vignette, suggesting a GM-specific rather than a generalized SGM main effect.
Differences in BPD diagnosis and agreement between vignettes (when present) were not explained by particular DSM-5 criterion; we found no significant differences in the rate of endorsement of any BPD criterion between treatment groups. Nor can differences in BPD diagnosis and agreement be attributed to perceived differences in underlying maladaptive personality; there were no significant differences in rate of PID-5-based BPD diagnosis or agreement with PID-5 latent domains/trait facets between vignettes.
Clinical and Research Implications
Results suggest that when clinicians are given the freedom to provide their own diagnosis based on clinical first impressions, they are significantly more likely to diagnose GM individuals with BPD compared to when they are restricted to diagnosing BPD based on structured assessments (e.g., PID-5) or individual criteria. Differences in participants’ SGM treatment skills, knowledge, and attitudes failed to account for the observed disparities in the diagnosis of BPD, pointing to a widespread and generalized bias against GM individuals.
To avoid the diagnostic bias we observed and to better understand the connections that behaviors associated with BPD (e.g., rejection sensitivity, identity disturbance, interpersonal difficulties, impulsive behaviors, and suicidal behaviors) have with minority stress processes for patients, clinicians should measure the level of impairment associated with each behavior in addition to item endorsement. A practical way to do so would be to rely more heavily on structured assessments (such as the PID-5) to determine potential diagnoses, then explore the possible connections that patients’ symptoms have with minority stress processes in unstructured interviews (c.f. Rodriguez-Seijas, 2021a). Such exploration may reduce the stigma associated with BPD among nurses, psychiatrists, and other healthcare professionals (Aviram, Brodsky, & Stanley, 2006; Young, 2021; Chartonas et al., 2017). Such exploration may also facilitate other, potentially more interculturally informed diagnoses, such as posttraumatic stress disorder, major depressive disorder, and generalized anxiety disorder.
Additional consideration among clinicians should be given to the fact that access to common treatments for BPD, such as dialectical behavioral therapy and cognitive behavioral therapy, have transdiagnostic effects that do not require a personality disorder diagnosis (Rodriguez-Seijas et al., 2022).
Future vignette research should control for the gender of the individual featured in the vignette and include gender non-binary conditions. Given that real-world clinical assessments differ significantly from vignettes (e.g., providing information about body language, the sound of patients’ voices, and patients’ interpretations of their behavior), researchers should utilize recorded vignettes or simulated therapy sessions.
Impact Statement
Borderline personality disorder (BPD) is one of the most stigmatized and stigmatizing psychiatric diagnoses among clinicians and the general public; this research contextualizes the symptoms of BPD within the psychosocial context of sexual- and gender-minoritized individuals, the better to reduce misconceptions and stigma among the public. The (over)diagnosis of BPD among SGM individuals is not without policy consequence; it may present an additional legal barrier for individuals seeking gender-affirming care or disability services. The diagnosis itself may provide a treatment-access barrier, as many clinicians refuse to work with clients that have BPD. The potential impact on the field of clinical psychology is to encourage increased scrutiny of the conceptual validity of BPD among SGM individuals.
