Ethnic variations in the structure of borderline personality disorder symptomatology

Abstract

The goal of this study was to examine differences in the factor structure of borderline personality disorder symptoms among different ethnic groups. The authors obtained information regarding ethnic identity and endorsement of borderline personality disorder criteria for an ethnically diverse community sample of 1140 young adult subjects from south Florida. Using this information the authors conducted an exploratory factor analysis examining differences between Caucasian, Hispanic and African American groups. A principal-components factor analysis (PCA) with Varimax rotation for each ethnic group revealed a reasonably generalizable four-factor structure: affective dysregulation, cognitive disturbance, disturbed relatedness and behavioral dysregulation. The emergence of a four-factor structure across three separate, relatively large samples suggests that the factors obtained have merit. However, the loadings of some BPD symptoms, such as impulsivity, varied for each ethnic group. The results of this study indicate that ethnic variations in borderline personality disorder should be considered during assessment and treatment of this disorder. Also, future research should examine if this same factor structure holds for ethnic minorities with BPD diagnoses, examine ethnic differences in the etiology and maintenance of BPD symptomatology, and explore the effects that these differences might have in treatment settings.

Introduction

Due to the polythetic nature of borderline personality disorder (BPD), the symptom presentations of patients can appear quite different. In addition to the various combinations of BPD symptoms, high rates of comorbidity with axis I disorders provide even more variation of BPD presentation (Oldham et al., 1995). Understanding the causes of variation in BPD symptom presentation is important not only for research purposes, but also for the diagnosis and treatment of this disorder in clinical settings. Although recent factor analytic studies have examined the factor structure of BPD criteria in both inpatient (Sanislow et al., 2000) and adolescent (Becker et al., 2006) samples, little to no research has been conducted on ethnic differences in the presentation of the disorder. An examination of the structure of BPD in varying ethnic groups would determine if there is a universal structure regardless of ethnicity, or if there are ethnic variations in diagnostic structure.

Although there have been few studies on ethnicity and BPD, some studies have examined the disorder in different ethnic groups. For example, Grilo et al. (2004) found that DSM-IV criteria for BPD were diagnostically efficient with a group of Hispanic subjects, and they also found that suicidal behavior was the best inclusion criterion and affective instability the best exclusionary criterion for a diagnosis. In another study Becker et al. (2005) examined the discriminant efficiency of BPD and antisocial personality disorder (ASPD) diagnostic criteria in a sample of substance abusing Hispanic men. They found that BPD criteria do not differentiate between BPD and ASPD, but ASPD criteria can differentiate the two disorders. Although these studies contribute important information to the BPD diagnosis as it relates to Hispanics, they do not address ethnic differences in the structure of BPD.

Previous factor analytic studies have made some progress understanding the structure of the BPD diagnosis. Furthermore, these studies have resulted in a variety of solutions. Fossatti et al. (1999), for example, conducted a confirmatory factor analysis on DSM-IV criteria in 564 inpatients and outpatients and concluded that borderline personality disorder is a unitary construct. Alternatively, Rosenberger and Miller (1983) conducted a factor analysis of DSM-III criteria for BPD in a sample of 106 college students and found a two factor solution: one for interpersonal and identity criteria, and one for the dysregulation of behavior and affective criteria. In another study Clarkin et al. (1993) conducted a factor analytic study on 75 inpatient women diagnosed with BPD according to DSM-III-R criteria and found a three factor solution with the first factor consisting of problematic interpersonal relationships and identity, the second factor consisting of affective qualities and suicidality, and the final factor consisting of impulsivity. Finally, Becker and colleagues (2006) conducted a factor analysis on 123 inpatient adolescents with DSM-III-R criteria. They arrived at a four-factor solution with the first factor consisting of suicidal behavior and emptiness/boredom, the second factor consisting of affective dysregulation and identity disturbance, the third factor consisting of unstable relationships and abandonment fears, and the fourth factor consisting of impulsiveness and identity disturbance. Each of these previous factor analytic studies has not only resulted in a different number of factors, but a symptom may load onto one factor in one study and a different factor in another study. These findings suggest that the underlying structure of the disorder remains at issue.

One structure that has shown particular merit comes from a study of DSM-III-R BPD criteria in 141 acutely ill inpatients; in this study Sanislow et al. (2000) suggested three factors: disturbed relatedness (“unstable relationships,” “identity disturbance,” and “emptiness and boredom”), behavioral dysregulation (“impulsiveness and “suicidal threats or gestures”), and affective dysregulation (“affective instability,” “inappropriate anger,” and “abandonment fears”). These factors were replicated in a second study with confirmatory factor analysis on 668 treatment-seeking subjects using DSM-IV-TR criteria (Sanislow et al., 2002). Yet, although this factor structure has been replicated, it may still be uniquely affected by treatment-seeking effects and comorbidity status.

There are a variety of issues that could result in different factor structure of BPD symptomatology, including inpatient versus outpatient samples, college samples, the developmental status of the sample (adolescent versus adult), gender, small sample size, and comorbidity status. Each of the previous factor analytic studies appears to have been influenced by one or more of these issues, with inpatient status, treatment-seeking, and comorbidity being among the most frequent. Furthermore, although many of the previous factor analytic studies had a certain percentage of minorities in their sample, none of these studies examined the effects of ethnicity on BPD symptom structure.

The examination of ethnic differences in BPD structure may provide important information as to why factor analytic studies have not revealed a common structure, as well as provide important information about ethnic differences in the disorder. Because of the important psychological and socio-cultural differences that are associated with different ethnic groups, some symptoms may be more tolerated in one ethnic group, whereas other symptoms may be viewed as more aberrant. For example, it has been suggested that cultural values may moderate the way that individuals regulate their emotions (Butler et al., 2007) and furthermore, cultures may encourage or discourage different emotional responses according to differing circumstances (Kitayama et al., 2000, Mequita, 2001). Additionally, given the prevalence of emotional symptoms in BPD and the important differences in ethnic/cultural experience and expression of emotion, the factor structure of BPD symptoms may be greatly influenced by ethnicity.

In the current study we examined borderline personality disorder symptomatology structure in a relatively large, ethnically diverse representative community sample to determine if self-reported ethnic identity contributed any heterogeneity to the underlying factor structure of the disorder symptomatology. Furthermore, the sample used in the current study was not affected by some of the issues present in previous studies such as small sample size, inpatient status, gender invariance, and college student sample. Also, because it is a community sample, comorbidity and treatment-seeking effects may not have been as prevalent.