Borderline Personality Disorder and Health Care Utilization
in a Primary Care Setting


ABSTRACT: We examined health care utilization by subjects with symptoms of borderline personality disorder. The study included 194 female subjects between the ages of 17 and 52 who were consecutively seen for nonemergency medical care in a health maintenance organization. Each subject completed the borderline personality scale of the Personality Diagnostic Questionnaire Revised (PDQ-R), as well as the Self-Harm Inventory, which correlates with the diagnosis of borderline personality. For each instrument, subjects with scores suggestive of borderline personality were compared with subjects without this disorder in regard to their utilization of health care resources (eg, number of combined telephone contacts and physician visits, number of physicians seen, number of prescriptions obtained) during the preceding 12 months. Subjects with more severe borderline personality disorder (according to the PDQ-R) made significantly more contacts with the health care facility (i.e., telephone calls and physician visits) and received more prescriptions.

STUDIES INDICATE that a significant minority of patients seen in primary care settings (15% to 30%) have psychiatric disorders.1-6 The research literature consistently suggests that those with psychiatric disorders use primary care services more than nonimpaired individuals.6-10 Studies of several specific Axis I psychiatric disorders, including dysthymia,11 depression,12 somatization,13-15 panic disorder,16,17 and factitious disorder,18 also support this finding. One study reports a significant association between pathologic personality traits and medical utilization,19 but we found no study that examines the utilization of health care services by patients with specific personality disorders.

The current study was done to explore the impact of a particular personality disorder, borderline personality, on health care utilization in the primary care setting of a health maintenance organization (HMO). Borderline personality disorder is characterized by a veneer of normalcy in brief social encounters as well as (1) fleeting and long-standing quasipsychotic episodes (eg, transient depersonalization, derealization, rage reactions); (2) chronic impulsivity manifested as self-regulatory deficits (eg, substance abuse/dependence, eating disorders, promiscuity, gambling) and overt self-destructive behavior (eg, suicide attempts, self-mutilation, high-risk hobbies or behavior, sadomasochistic or abusive interpersonal relationships); (3) chaotic and chronically unsatisfying interpersonal relationships; and (4) chronic dysphoria (ie, persistent depression, anxiety, emptiness, anger).20,21 Again, the unique feature of borderline personality is the superficial appearance of normalcy in transient social encounters coupled with significant underlying psychologic dysfunction. Borderline personality was chosen for study because of its high prevalence in psychiatric settings (15% to 25% in inpatient and outpatient populations22,23) as well as in primary care settings, the difficulty in treating/caring for these individuals in primary care settings, and the potential for self-destructive behavior and medical noncompliance in these patients.


The study population consisted of 194 female subjects between the ages of 17 and 52 (mean age, 33.6; SD, 9.2) who were consecutively seen for nonemergency medical care by a female family physician (L.A.S.) in an HMO. The response rate was 97%.

Two instruments were used. The first was the borderline personality scale of the Personality Diagnostic Questionnaireñ Revised (PDQ-R),24 an 18-item self-report inventory that screens subjects for symptoms compatible with borderline personality. The borderline personality scale of the PDQ-R is based on the criteria for borderline personality noted in the Diagnostic and Statistical Manual of Mental Disorders, 3rd Edition, Revised (DSM-III-R).25 The PDQ-R has been reported to be a useful screening tool in both clinical26,27 and nonclinical28 populations.

The second measure, the Self-Harm Inventory (SHI),* is a 22-item self-report questionnaire that explores subjects self-destructive behavior. Each item is preceded by the statement, Have you ever intentionally, or on purpose. . . . Items include overdosed, burned yourself, attempted suicide, cut yourself, and engaged in sexually abusive relationships. A score on the SHI is the total number of endorsed items, each of which represents a pathologic response (i.e., there are no nonpathologic items in the inventory). Scores on the SHI have been shown to highly correlate with borderline personality* as measured by both the borderline personality scale of the PDQ-R (r = .67) and the Diagnostic Interview for Borderlines29 (r = .76). A cutoff score of 5 represents an overall accuracy in diagnosis of 83.7% (with diagnosis based on the Diagnostic Interview for Borderlines).*

All subjects participated in the project after their nonemergency visit. In a quiet room on site, each subject completed a testing booklet that included a demographic inquiry, the borderline personality scale of the PDQ-R, and the SHI. The medical records of all participants were reviewed after testing. For the preceding 12 months, the following measures of health care utilization were noted from the medical record: (1) the total number of contacts by the subject to the facility, which consisted of the combination of telephone calls and physician visits; (2) the number of different physicians seen by the subject; and (3) the number of prescriptions provided to the subject. All subjects had consistently obtained care at the HMO during the preceding 12-month period. The medical records were reviewed blind to subjects psychologic testing results.


According to the recommended cutoff score of 5 for the borderline personality scale of the PDQ-R,30 38 subjects (20%) reported symptoms suggestive of borderline personality disorder. According to the recommended cutoff score of 5 on the Self-Harm Inventory, 30 subjects (16%) met the criteria for borderline personality disorder. The results of one-way analyses of variance comparing these groups with regard to use of medical services are presented in the Table. Subjects with borderline personality symptoms showed significantly greater utilization of resources (eg, telephone and physician contacts) than subjects who did not evidence borderline personality disorder on the PDQ-R.

The data were analyzed as continuous variables to determine whether an increasing score on the measures of borderline personality might relate to increasing use of medical services. To investigate this possibility, Pearsonís correlation coefficients were computed between both PDQ-R and SHI scores and the degree of medical care utilization by subjects. Scores on the PDQ-R were correlated to the number of facility contacts (telephone calls and physician visits) by subjects (r = .21,
P < .05) and the number of prescriptions
(r = .16, P < .05) but not to the number of different physicians who had seen the subject. Scores on the SHI were unrelated to any of the measures of medical care utilization.


In this study, participants with borderline personality symptoms (according to the PDQ-R) evidenced significantly higher utilization of primary care resources than those without borderline personality symptoms. In addition, increasing PDQ-R scores for borderline personality (i.e., severity) predicted increased use of medical resources. These findings are consistent with the majority of studies indicating that individuals with psychiatric disorders evidence higher rates of health care utilization. These studies also indicate that individuals in primary care settings with borderline personality present an additional risk of cost through higher utilization. There may be an underlying cost for mental health services, which was not explored in this study.

The second measure of borderline personality, the SHI, showed no significant correlation with medical care utilization. This finding is difficult to explain but may have to do with the nature of the measure. The SHI surveys overt self-destructive behavior, whereas the PDQ-R may be a measure of more generalized psychiatric disturbance. Conceivably, at least some individuals with self-destructive tendencies avoid medical treatment, even when needed, as one more form of self-harm behavior. Further research on the relationship between self-harm behavior and the nature of health care utilization is needed.

There were no significant differences in the number of physicians that subjects saw for treatment as a function of borderline personality disorder. This is surprising, since borderline individuals are characterized by their impulsivity and low frustration tolerance. We would have predicted a significant between-group difference, with borderline subjects seeing a greater number of different physicians. However, the population demographics may explain this in that the borderline subjects in this study appear to be fairly high functioning.

Somewhat surprisingly, of those individuals whose PDQ-R scores indicated borderline personality, 63% were married, 26% were single, and only 11% were divorced. For those identified as borderline by the SHI, 57% were married, 19% were single, and only 7% were divorced. Regarding educational levels, only one individual (3%) in both cohorts did not complete high school. Fifty percent of those with borderline personality as shown by the PDQ-R had attended college, and 16% had a bachelorís degree or higher. A similar educational profile was obtained when examining those with borderline personality diagnosed by the SHI. Again, this sample of borderline individuals appears to be much higher functioning than individuals from other borderline personality samples. Therefore, we suggest that in a lower functioning population, these findings might be even more pronounced.

Taking into account the previous studies on psychiatric disorders and medical utilization, these findings suggest that psychiatric disorders including personality disorders generally predict higher rates of health care utilization. Therefore, in predicting health care utilization patterns in populations with psychiatric disorders, and, specifically, personality disorders, greater costs can be expected.

We believe this is the first study to investigate potential differences in health care utilization by patients with, versus without, borderline personality disorder in a primary care setting. A potential limitation in this study is the use of self-report instruments as measures of personality disorder. The strengths of this study include the use of two measures of borderline personality (ie, borderline personality scale of the PDQ-R and the SHI), the use of multiple measures of health care utilization (ie, contacts to the facility, which included both telephone contacts and physician visits, as well as number of prescriptions), and the recruitment of subjects from an enclosed health care delivery system (ie, a health maintenance organization) where utilization patterns could be accurately assessed. Given the prevalence of borderline personality disorder in all clinical settings, these findings are noteworthy.


1. Joukamaa M, Lehtinen V, Karlsson H, et al: SCL-25 and recognition of mental disorders reported by primary health care physicians. Acta Psychiatr Scand 1994; 90:320-323

2. Li PL, Jones I, Richards J: The collection of general practice data for psychiatric service contracts. J Public Health Med 1994; 16:87-92

3. Maniam T: Psychiatric morbidity in an urban general practice. Med J Malaysia 1994; 49:242-246

4. Sato T, Takeichi M: Lifetime prevalence of specific psychiatric disorders in a general medicine clinic. Gen Hosp Psychiatry 1993; 15:224-233

5. van Hemert AM, Hengeveld MW, Bolk JH, et al: Psychiatric disorders in relation to medical illness among patients of a general medical out-patient clinic. Psychol Med 1993; 23:167-173

6. Hansson L, Nettelbladt P, Borquist L, et al: Screening for psychiatric illness in primary care. a cross-sectional study in a Swedish health district. Soc Psychiatry Psychiatr Epidemiol 1994; 29:83-87

7. Kendrick T, Burns T, Freeling P, et al: Provision of care to general practice patients with disabling long-term mental illness: a survey in 16 practices. Br J Gen Pract 1994; 44:301-305

8. Hansson L, Borquist L, Nettelbladt P, et al: The course of psychiatric illness in primary care patients: a 1-year follow-up. Soc Psychiatry Psychiatr Epidemiol 1994; 29:1-7

9. Stefansson CG, Svensson C: Identified and unidentified mental illness in primary health careósocial characteristics, medical measures and total care utilization during one year. Scand J Prim Health Care 1994; 12:24-31

10. Williams P, Tarnopolsky A, Hand D, et al: Minor psychiatric morbidity and general practice consultation: the West London Survey. Psychol Med 1986; 16(suppl):1-37

11. Howland RH: General health, health care utilization, and medical comorbidity in dysthymia. Int J Psychiatry Med 1993; 23:211-238

12. Wells KB, Stewart A, Hays RD, et al: The function and well-being of depressed patients. results from the Medical Outcomes Study. JAMA 1989; 262:914-919

13. Miranda J, Perez-Stable EJ, Munoz RF, et al: Somatization, psychiatric disorder, and stress in utilization of ambulatory medical services. Health Psychol 1991; 10:46-51

14. Escobar JI, Burnham MA, Karno M, et al: Somatization in the community. Arch Gen Psychiatry 1987; 44:713-718

15. Quill TE: Somatization disorders: one of medicineís blind spots. JAMA 1985; 254:3075-3079

16. Katon WJ, Von Korff M, Lin E: Panic disorder: relationship to high medical utilization. Am J Med 1992; 92:7S-11S

17. Klerman GL, Weissman MM, Quellette R, et al: Panic attacks in the community. social morbidity and health care utilization. JAMA 1991; 265:742-746

18. Feldman MD: The costs of factitious disorders. Psychosomatics 1994; 35:506-507

19. Reich J, Boerstler H, Yates W, et al: Utilization of medical resources in persons with DSM-III personality disorders in a community sample. Int J Psychiatry Med 1989; 19:1-9

20. Sansone RA, Sansone LA: Borderline personality disorder: office diagnosis and management. Am Fam Physician 1991; 44:194-198

21. Sansone RA, Sansone LA: Borderline personality disorder. interpersonal and behavioral problems that sabotage treatment success. Postgrad Med 1995; 97:169-179

22. Widiger TA, Rogers JH: Prevalence and comorbidity of personality disorders. Psychiatr Ann 1989; 19:132-136

23. Gunderson JG, Zanarini MC: Current overview of the borderline diagnosis. J Clin Psychiatry 1987; 48:5-11

24. Hyler SE, Rieder RO: Personality Diagnostic QuestionnaireñRevised (PDQ-R). New York, New York State Psychiatric Institute, 1987

25. Diagnostic and Statistical Manual of Mental Disorders, 3rd Ed, Revised. Washington, DC, American Psychiatric Association, 1987

26. Dubro AF, Wetzler S, Kahn MW: A comparison of three self-report questionnaires for the diagnosis of DSM-III personality disorders. J Pers Disord 1988; 2:256-266

27. Hyler SE, Lyons M, Rieder RO, et al: The factor structure of self-report DSM-III Axis II symptoms and their relationship to cliniciansí ratings. Am J Psychiatry 1990; 147:751-757

28. Johnson JG, Bornstein RF: Utility of the Personality Diagnostic Questionnaire-Revised in a nonclinical population. J Pers Disord 1992; 6:450-457

29. Kolb JE, Gunderson JG: Diagnosing borderline patients with a semi-structured interview. Arch Gen Psychiatry 1980; 37:37-41

30. Hyler SE, Rieder RO, Williams JBW, et al: A comparison of clinical and self-report diagnoses of DSM-III personality disorders in 552 patients. Compr Psychiatry 1989; 30:170-178