Estimating the prevalence of borderline personality disorder in psychiatric outpatients using a two-phase procedure

Abstract

The prevalence of borderline personality disorder (BPD) in outpatient clinics varies greatly (7%-27%) depending on the setting and methodology. We examined the cross-sectional rate of BPD in a general adult outpatient university clinic using a 2-phase procedure: (1) we screened all registered patients with the self-report SCID-II-PQ and (2) we administered the Revised Diagnostic Interview for Borderlines (DIB-R). Sixty-six percent (239/360) of the clinic patients completed the screening: About 72.4% (173/239) (95% confidence interval [CI] = 66.7%, 78.1%) were positive for BPD on the Structured Clinical Interview for DSM-IV Personality Disorders–Patient Questionnaire (SCID-II-PQ), and 22.6% (54/239) (95% CI = 17.3%, 27.9%) were positive for BPD on the DIB-R. Our BPD rate was somewhat higher than recent semistructured interview studies (9%-18%). We believe this is due, in part, to our cross-sectional design and our decision not to exclude acute Axis I disorders. Mostly, however, we believe that our 22.6% incidence of BPD arises from the high morbidity of our sample. Demographic data from 130 of 131 DIB-R completers reveal the following: mean age was 40.2 years, 75.4% were female, most patients were unable to work, and they averaged 3.8 lifetime hospitalizations.

Introduction

Borderline personality disorder (BPD) is a pervasive pattern of impulsiveness, self-destructive threats or behaviors, instability of moods, chaotic interpersonal relationships, and unstable sense of identity [1]. Patients with BPD have disproportionately high rates of health care resource utilization, high rates of self-injurious behavior, and a suicide rate of about 10% [2]. Since the advent of effective psychotherapies [3], it is imperative that outpatient clinics have an accurate estimate of the number of BPD patients in their care to plan resources and services.

The literature on the epidemiology of personality disorders is plagued by numerous methodological problems. First, the diagnostic criteria for BPD changed from the 1980 to the 1994 editions of the Diagnostic and Statistical Manual of Mental Disorders (DSM), making early studies less relevant. Second, different diagnostic tools produce very different estimates of the prevalence of BPD. That is, compared with semistructured interviews, unstructured clinical assessments are less reliable, with more false negatives [4]. Conversely, self-report questionnaires produce too many false positives [5]. Third, research samples can be biased by clinical setting, access to treatment, help seeking, severity of illness, and comorbidity issues [6].

In the general population, looking at the studies with the broadest community random sample selection, the prevalence of BPD varied from 0.4% to 1.8%, with a pooled rate of 1.1% [7], [8], [9], [10]. In clinical samples, BPD is usually the most common personality disorder [11]. The rates of BPD in pre-1989 clinical interview psychiatric inpatients studies are 15% [11]. More recent inpatient studies with semistructured interviews have reported rates of 40% to 44% [12], [13].

The prevalence of BPD in outpatient samples varies depending on whether the clinic serves insured or uninsured patients or specialized populations. In outpatient samples, the rates of BPD have varied from 8% to 27%. The average prevalence across 8 pre-1989 clinical interview outpatient studies is 8% [11]. The average prevalence of BPD in outpatient studies that use self-report questionnaires is 27.7% to 30.3% [7], [14]. Recent semistructured interview studies have reported rates of 9.3% to 18% in consecutively enrolled subjects, with a pooled rate of 11.9% (n = 1657) [4], [15], [16], [17]. These 4 studies were conducted in a catchmented clinic serving a more affluent and educated part of the city, psychoanalysis applicants, and university clinics where patients were insured.

The 2-phase procedure (ie, screening with a self-report questionnaire and then interviewing those who screened positive) is an efficient methodology to identify cases of personality disorder. Nussbaum and Rogers [18] screened an inpatient forensic population with the Structured Clinical Interview for DSM-IV Personality Disorders–Patient Questionnaire (SCID-II-PQ) [19], followed by the Structured Clinical Interview for DSM-IV Personality Disorders (SCID--II) interview. They obtained a 2.4% false-negative rate of BPD and a 30.5% false-positive rate of BPD. Lenzenweger et al [5] screened university students with the Personality Disorder Examination (PDE) self-report screen, followed by the PDE interview. They obtained a 0% false-negative rate and an 84.3% false-positive rate for all the personality disorders. Thus, self-report questionnaires perform remarkably well in screening for personality disorders, as very few cases are missed [18].

The goal of the present study was to estimate the point prevalence of BPD in a general adult outpatient university clinic by screening with the BPD section of the SCID-II-PQ and then assessing the positive cases with the Revised Diagnostic Interview for Borderlines (DIB-R) [20]. The DIB-R has been used extensively to diagnose BPD, but never in prevalence studies. It is time consuming (1 hour) and requires substantial clinical experience to administer and score [20].