East Asian Arch Psychiatry 2011;21:52-7

ORIGINAL ARTICLE

Dr Hiu-Mei Wong, MBBS, MRCPsych, FHKCPsych, FHKAM (Psychiatry), Department of Psychiatry, Shatin Hospital, New Territories, Hong Kong SAR, China.
Dr Lai-Yin Chow, MBChB, MRCPsych, FHKCPsych, FHKAM (Psychiatry), Department of Psychiatry, Tai Po Hospital, New Territories, Hong Kong SAR, China.

Address for correspondence: Dr HM Wong, Department of Psychiatry, Shatin Hospital, 33 A Kung Kok Street, Shatin, New Territories, Hong Kong SAR, China.
Tel: (852) 2636 7500; Fax: (852) 2647 5321; Email: whm254@ha.org.hk

Submitted: 24 November 2010; Accepted: 21 March 2011

pdf Full Paper in PDF

Introduction

Borderline personality disorder (BPD) is an important entity to recognise in clinical practice. Co-morbidity of this personality disorder with Axis I disorder may lead to a poor prognosis, poor treatment response, and increased risk of mortality. The prevalence of BPD is estimated to be about 2% in the general population,¹ about 10% among individuals seen in outpatient mental health clinics,² and about 20% among psychiatric inpatients.³ Despite its clinical importance and relatively high prevalence among psychiatric patients, BPD is an under-researched topic with very limited available data in the Chinese population. A survey conducted on 3,402 psychiatric outpatients in Shanghai in 2006 showed that 5.8% fulfilled the criteria for BPD described in the Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV), 46% of whom had co-morbid Axis I mood disorders.⁴ Using a structured clinical interview for the diagnosis of BPD may be useful for enhancing the validity and reliability of the diagnostic label in research settings. Moreover, because of the apparent overlap of BPD with many other Axis I mood disorders, an easily administered diagnostic instrument to better delineate this category was warranted.

Several diagnostic tools for personality disorders based on the DSM-IV⁵ or the International Classification of Diseases, 10th revision (ICD-10)⁶ had been translated into Chinese (Mandarin) and validated in Mainland China (Table 1^7-9). However, none has been validated in the Cantonese- speaking Chinese population in Hong Kong. Among the 3 existing instruments, the Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II) is the only one which allows the usage of a distinct portion to assess a particular subtype of personality disorder. It confers a distinctive advantage by enabling studies focused on a specific subtype of personality disorder (Table 1).

The original version of SCID-II for assessment of the 10 DSM-IV Axis II personality disorders entailed a semi-structured face-to-face interview developed by the Biometrics Research, New York State Psychiatric Institute, New York.¹⁰ The SCID-II interview schedule started with a brief overview that characterised the subject’s usual behaviours and relationships. The BPD subscale consisted of 9 items corresponding to the DSM-IV diagnostic criteria (Appendix). There was an initial interview question, followed by several pre-determined follow-up questions for each item. The interviewer coded each item with either ‘1’, ‘2’, or ‘3’, using clinical judgement based on information provided by the subjects (‘1’ = absent, ‘2’ = subthreshold, ‘3’= threshold or true). A rating of ‘3’ was warranted only if the subject had provided a convincing elaboration or example, or if there was clear evidence from the behaviours encountered during the interview. For instance, regarding the item 1 ‘frantic efforts to avoid real or imagined abandonment’ in the BPD subscale, if a person repeatedly took frantic actions to keep people that they were dependent on or attached to from leaving, such as pleading with them not to leave or physically restraining them, a rating of ‘3’ would be warranted. The interviewer also needed to confirm that the trait had been present prior to and independent of the Axis I condition and of long-standing, in order to tease out the trait and state the relationship. Subjects with 5 or more items scored as ‘3’ out of the 9 items were rated as having BPD. In its original design, the subject interviewed was usually the sole source of information. However, ancillary data from other sources, for example, a current or previous therapist or family member, were also permitted, especially in cases in which contradictory information was elicited from the subjects.

The Chinese version of SCID-II was developed by the Shanghai Mental Health Center in 2006.⁷ In its development, the original English version was translated into Mandarin and was back-translated into English for testing of its psychometric properties. Cultural factors were also taken into consideration in the translation and revision process. The BPD subscale demonstrated an internal consistency with a coefficient of 0.77, the kappa for inter-rater reliability was 0.72, and the kappa for test-retest reliability was 0.78.

In view of the cultural and dialectal differences between Mandarin-speaking Chinese and Hong Kongers, the applicability of this instrument was evaluated using the Cantonese spoken by Hong Kong Chinese in this study.

Methods

Objectives

The study aimed to assess the reliability and validity of the BPD subscale (Chinese version) of the SCID- II, by measuring the internal consistency, best-estimate clinical diagnosis–SCID diagnosis agreement kappa value, sensitivity, and specificity in a Cantonese-speaking Chinese sample in Hong Kong.

Ethical Considerations

Ethical approval was obtained from the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee. Written consent was obtained from all subjects. Subjects could withdraw from the study at any time. All subjects diagnosed to have psychiatric morbidity were offered referrals to a psychiatric service if they had not registered with one.

Sample Size Estimation

We assumed that the prevalence of BPD in psychiatric outpatients was 10 to 20%.² The number of subjects required to detect a statistically significant kappa (p ≤ 0.05) with the null-hypothesis value of kappa of 0.40 (the lowest value of kappa characterised by Landis and Koch¹¹ as representing clinically acceptable agreement), with 80% power, would be between 48 and 65,¹² thus 65 subjects were recruited.

Recruitment Sites and Procedure

Convenience sampling was used: patients attending the psychiatric outpatient clinic of the Prince of Wales Hospital in April 2007 were invited to participate in the study. In 2006, the Prince of Wales Hospital was an acute general hospital with a psychiatric outpatient service serving a general population of 607,544 in Hong Kong.¹³ Patient inclusion criteria were: (a) age between 18 and 64 years; (b) Chinese ethnicity; and (c) provision of written informed consent. Exclusion criteria included: (a) being physically unsuitable for interview; (b) inability to communicate in Cantonese; and (c) lack of mental capacity to be interviewed (e.g. moderate or severe mental retardation). Each participant was interviewed by a multidisciplinary clinical team formed by experienced psychiatrists, nursing staff, and social workers. Their aim was to provide the best- estimate diagnosis by determining whether the participant fulfilled the diagnostic criteria of BPD according to the DSM-IV classification system. Additional information was obtained from old case notes, if available. Other available informants including clinical psychologists, occupational therapists, family members and friends were interviewed for collateral information, as necessary.

The SCID-II rater (the first author) completed training by studying the user’s guide and watching an authorised training videotape issued by the authors of the SCID-II.¹⁴ The rater then carried out the SCID-II assessment within 1 week of the initial assessment by the multidisciplinary clinical team, while remaining blind to the best-estimate clinical diagnosis. In this assessment, the BPD subscale of the Chinese version of SCID-II was administered to each participant. Each assessment lasted 25 to 40 minutes. The SCID rater was not permitted to obtain other sources of information to ensure that the final diagnostic assessment would be based only on the result of the interview.

Data Analysis

Data were analysed using the Statistical Package for the Social Sciences, Windows version 15.0. A p value of less than 0.05 was considered statistically significant. The internal consistency of the BPD subscale of SCID-II expressed as Cronbach’s α coefficient was computed. The BPD status of each participant established by SCID assessment and clinical interview were compared. The gold-standard clinical diagnosis and SCID diagnosis agreement expressed as a kappa value was calculated. Moreover, the sensitivity, specificity, positive and negative predictive values for BPD diagnoses established by SCID-II assessment were calculated, using the clinician-established diagnosis as the gold standard.

Results

Demographic Characteristics

During the recruitment period, 87 patients attended the outpatient clinic and fulfilled the selection criteria. There were 53 females (61%) and 34 males (39%), with a mean (± standard deviation [SD]) age of 41 (± 10) years. Twenty two (25%) of the patients declined to participate in the study, whilst the remaining 65 gave informed consent and completed the study. The mean (SD) age of the latter was 41 (± 11) years, of whom 40 (62%) were women. Also, 39 (60%) of them were married, 13 (20%) were single, and the remainder were either divorced (19%) or widowed (2%). Table 2 shows the demographic characteristics of those who declined and those who completed the study, and reveals that the groups were comparable.

Principal Axis I Diagnosis

The principal Axis I diagnosis was obtained from the case notes. The majority of the participants (40%) suffered from mood disorders (major depressive disorder in 28% and bipolar disorder in 12%). Alcohol and substance use disorders accounted for 32%, whilst 15% suffered from anxiety disorders, 11% from schizophrenia and related psychotic disorders, and 2% from eating disorders.

Internal Consistency of the Borderline Personality Disorder Subscale

The Cronbach’s α coefficient of the BPD subscale of SCID-II was 0.82. Since this value was greater than 0.70, it was considered acceptable.¹⁵

Validity of the Borderline Personality Disorder Subscale

Of the 65 subjects, 13 (20%) were diagnosed to have BPD by clinician’s best-estimate diagnosis, whilst 15 (23%) fulfilled the SCID-II diagnostic criteria of BPD. Table 3 illustrates the distribution of diagnoses based these 2 methods of assessment. The BPD subscale of SCID-II demonstrated satisfactory sensitivity (0.92), specificity (0.94), as well as positive predictive (0.80) and negative predictive (0.98) values. The kappa for best-estimate¹¹ clinical diagnosis– SCID diagnosis agreement was 0.82.

Discussion

In this study, subjects were recruited from the clinical population with a higher base rate of BPD than in the community, because a sample with low base rates might lead to unstable kappa values.¹⁶ It was recommended that the kappa value be calculated only for diagnoses with a frequency of 5% or more.¹⁷ The base rate of in our sample (20%) was sufficient to estimate a kappa value. However, since a convenience sample was used in this study, the sample size was relatively small, and the BPD prevalence of (20%) in the current study was not necessarily representative of the psychiatric outpatient populations. This area could be explored in the future studies with larger sample sizes.

There are several limitations to this study. First, the results cannot be generalised to patients aged above 65 years and below 18 years. Second, due to manpower constraints, the inter-rater reliability, test-retest reliability, convergent and divergent validity were not measured. These measures should be included in future reliability studies to allow a more complete measure of external reliability. Third, the validity of personality questionnaire, which was available as a self-reported questionnaire / screening tool to reduce the time required for the clinician to administer the SCID-II interview schedule, had been omitted. These issues could be addressed in future research.

Our results demonstrate that the BPD subscale of the Chinese version of SCID-II has a reasonable validity in the Cantonese-speaking Chinese population of Hong Kong. Its internal consistency, best-estimate clinical diagnosis– SCID rater agreement, sensitivity, specificity, positive and negative predictive values measured against the clinician’s best-estimate diagnosis were also demonstrated to be satisfactory and comparable to those obtained in other studies.^7,18 Validation of the SCID-II BPD subscale in Cantonese Chinese is a step forward in the application of a structured instrument for diagnosing personality disorder in the course of psychiatric research in Hong Kong. It can be used to measure the prevalence of BPD in a clinical population, or in community epidemiological studies. Regarding clinical use, it can assist confirming and documenting suspected BPD. Finally, this instrument can help improve interview skills of students in the mental health professional, by providing a repertoire of useful questions to elicit information about the diagnosis of BPD.

References

1. Swartz M, Blazer D, George L. Estimating the prevalence of borderline personality disorder in the community. J Person Disord 1990;4:257-72.
2. Zimmerman M, Rothschild L, Chelminski I. The prevalence of DSM-IV personality disorders in psychiatric outpatients. Am J Psychiatry 2005;162:1911-8.
3. Widiger TA, Weissman MM. Epidemiology of borderline personality disorder. Hosp Community Psychiatry 1991;42:1015-21.
4. Wang LL, Zhang TH, Xiao ZP. A survey of the comorbidity of borderline personality disorder among psychiatric outpatients at Shanghai Mental Health Center. Shanghai Archives of Psychiatry 2007;19:136-9.
5. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, D.C.: American Psychiatric Association; 2000.
6. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders. Clinical descriptions and diagnostic guidelines / World Health Organization. Geneva: World Health Organization; 1990.
7. Dai Y, Xiao Z, Wang Z. The reliability and validity of a Chinese version of the Structured Clinical Interview for DSM-IV personality disorders. Shanghai Archives of Psychiatry 2006;18:1-4.
8. Han J, Xu Y, Cui Y. Evaluation on reliability and validity of the Chinese version of International Personality Disorder Examination. Chinese Journal of Psychiatry 1998;31:172-4.
9. Yang Y, Shen D, Wang J. The Reliability and Validity of PDQ-4+ in China. Chinese Journal of Clinical Psychology 2002;10:165-8.
10. First MB, Gibbon MG, Spitzer RL, Williams JB, Benjamin LS. Structured Clinical Interview For DSM-IV Axis II Personality Disorders (SCID-II). New York: American Psychiatric Press; 1997.
11. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74.
12. Sim J, Wright CC. The kappa statistic in reliability studies: use, interpretation, and sample size requirements. Phys Ther 2005;85:257- 68.
13. Census and Statistics Department, HKSAR. Interactive Data Dissemination System for the 2006 Population By-census 2006. http:// idds.censtatd.gov.hk/default.aspx. Accessed 1 Apr 2008.
14. First MB, Gibbon MG, Spitzer RL, Williams JB, Benjamin LS. User’s guide for the Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II). Washington, DC: American Psychiatric Press; 1997.
15. Nunnally JC. Psychometric theory. New York: McGraw-Hill; 1978.
16. Shrout PE, Spitzer RL, Fleiss JL. Quantification of agreement in psychiatric diagnosis revisited. Arch Gen Psychiatry 1987;44:172-7.
17. Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New York: John Wiley & Sons; 1981.
18. Skodol AE, Rosnick L, Kellman D, Oldham JM, Hyler SE. Validating structured DSM-III-R personality disorder assessments with longitudinal data. Am J Psychiatry 1988;145:1297-9.