Journal of Personality Disorders, 26, 2012, 551-567 © 2012 The Guilford Press ANALOGOUS DIAGNOSTIC CATEGORIES Lai ET AL. 020 ARE DSM-IV-TR BORDERLINE PERSONALITY DISORDER, ICD-10 EMOTIONALLY UNSTABLE PERSONALITY DISORDER, AND CCMDIII IMPULSIVE PERSONALITY DISORDER ANALOGOUS DIAGNOSTIC CATEGORIES ACROSS PSYCHIATRIC NOMENCLATURES? Ching Man Lai, MPhil, Freedom Leung, PhD, Jianing You, MPhil, and Fanny Cheung, PhD This study examined the validity of the borderline construct which encompasses diagnostic criteria from the DSM-IV-TR Borderline Personality Disorder (BPD), ICD-10 Emotionally Unstable Personality Disorder (EUPD), and CCMD-III Impulsive Personality Disorder (IPD) in a sample of 1,419 Chinese psychiatric patients. Participants completed the Chinese Personality Disorder Inventory and the Chinese Personality Assessment Inventory-2 assessing various disordered personality features. Adequate internal consistency was found for the borderline construct (α = .83). Exploratory factor analysis revealed two components: (1) affective and cognitive disturbances, and (2) impulse dysregulation, which were replicated by confirmatory factor analysis. Item analysis indicated that the various borderline criteria displayed similar levels of diagnostic efficiency, which does not support the elimination of fear of abandonment and transient psychotic features from the EUPD and IPD criteria set. Findings of this study suggest that BPD, EUPD, and IPD may represent analogous diagnostic categories across classification systems. Borderline personality disorder (BPD) was first introduced as an official diagnostic category in the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III; American Psychiatric Association, 1980). In the DSM-IV-TR (American Psychiatric Association, 2000), BPD was described as “a pervasive pattern of instability of interpersonal relationships, self-image and affects, and marked impulsivity beginning by early adulthood and presents in a variety of contexts” (American Psychiatric Association, This article was accepted under the editorship of Paul S. Links. From The Chinese University of Hong Kong This research was supported by funding from the Research Grants Council Earmarked Grant Projects CUHK4326/01H and UHK4333/00H to Prof. Fanny Cheung and Freedom Leung. Address correspondence to Freedom Leung, PhD, Department of Psychology, Chinese University of Hong Kong, Shatin, Hong Kong; E-mail:fykleung@psy.cuhk.edu.hk. 551 552 LAI ET AL. 2000, p. 706). Previous epidemiological studies have found that the prevalence of BPD is about 0.2% to 1.8% in the community sample, and about 10%–20% in psychiatric sample (Torgensen, Kringlen, & Cramer, 2001; Widiger & Weissman, 1991; Zimmerman, Rothschild, & Chelminski, 2005). CONSTRUCT VALIDITY OF BPD BPD is one of the most extensively studied personality disorders (PD) in the literature. BPD patients were found to have significant impairment in various areas of psychosocial functioning (Skodol et al., 2002; Skodol et al., 2005). Ample studies have reported high reliability of the BPD criteria (Fossati et al., 1999; Johansen, Karterud, Pedersen, Gude, & Falkum, 2004; Sanislow, Grilo, & McGlashan, 2000). Empirical efforts over the past decades in the West indicated that BPD is a valid psychiatric diagnosis with well-documented neurobiological and psychosocial correlates (Linehan, 1993; Paris, 1994a, b). However, skeptics are still present concerning the construct validity of BPD, in particular with its contextual validity in other cultures (Alarcon & Foulks, 1995; Luo & Huang, 2002; Zhong & Leung, 2009). To our knowledge, studies until now all supported the internal consistency (Leung, Chan, & Cheung, 2007; Leung, Cheung, & Cheung, 2004; Lu et al., 2001; Yang, Bagby, Costa, & Ryder, 2002; Yang, Rorbert, & Paul, 2000), test-retest reliability, and construct validity of the BPD construct in Chinese (Leung & Leung, 2009; Wang, Leung, & Zhong, 2008). Yang and colleagues (2000) reported sufficient internal consistency of the BPD construct in Chinese psychiatric patients using the Personality Diagnostic Questionnaire-4+ (Hyler, 1994; α = .71) and Personality Disorder Interview-IV (Widiger & Frances, 1985; α = .66). Leung and Leung (2009), in a sample of Chinese adolescents, demonstrated that BPD is a coherent personality construct and replicated Sanislow et al.’s (2000, 2002) three-factor model of BPD. Together, these findings suggest that the BPD construct represents a valid clinical syndrome among the Chinese. COMPARABLE CONSTRUCTS OF BPD IN THE ICD-10 AND THE CCMD-III There is no exact diagnostic label termed as BPD in the tenth edition of International Classification of Diseases (ICD-10; World Health Organization, 1992). After burgeoning evidence regarding the validity of BPD, the ICD has grudgingly introduced a derivative category coined as emotionally unstable personality disorder (EUPD) with the impulsive (EUPD-I) and borderline subtypes (EUPD-B). Similarly, the BPD construct has not been accepted in the Chinese Classification of Mental Disorders (CCMD; Chinese Psychiatric Association, 2001) in light of the strong resistance and objection among Chinese clinicians (Jia, 1998). Instead, CCMD adopted a new diagnostic category termed impulsive personality disorder (IPD) which is modeled closely after the impulsive subtype of EUPD. ANALOGOUS DIAGNOSTIC CATEGORIES 553 As shown in Table 1, a number of common diagnostic features can be observed among the three diagnostic categories. For instance, eight out of the ten EUPD criteria are similar to the BPD construct, while eight out of the ten diagnostic criteria of IPD can be found in the EUPD construct. For various reasons, either cultural inappropriateness or clinical irrelevance, the CCMD committee dropped the criteria, namely, frantic effort to avoid abandonment, chronic feeling of emptiness, and transient psychotic features, and added a new criterion, inability to plan ahead, to the diagnostic set of IPD. However, with a high number of common diagnostic criteria across IPD and BPD, it is unclear how IPD patients in China may be different from BPD patients in the West. Moreover, since the diagnostic criteria of BPD, IPD, and EUPD have never been subjected to analysis simultaneously in a cross-system manner, it is uncertain whether these three diagnostic categories reflect several distinct clinical entities or just analogous diagnostic categories across psychiatric nomenclatures. The objective of this study, therefore, is to examine the construct validity of the borderline syndrome which encompasses all the defining personality features of BPD, IPD, and EUPD in a Chinese psychiatric sample. METHODS PARTICIPANTS Participants of this study came from a larger study which aimed at establishing the clinical norms for the Chinese Personality Assessment Inventory-2 (CPAI-2; Cheung, Chueng, & Leung, 2008). The original study recruited 1,749 psychiatric patients from 10 psychiatric hospitals/clinics in China and two psychiatric hosptials/clinics in Hong Kong. Five categories of psychiatric disorders comprising the most common types of patients found in psychiatric hospitals and clinics in China and Hong Kong were included: schizophrenia, bipolar disorders, depressive disorders, anxiety and other neurotic disorders, and substance use disorders. The psychiatric hospitals in mainland China were selected to cover different geographical locations, including Beijing, Chengdu, Fuzhou, Guangxi, Henan, Jilin, and Nanjing to ensure the representativeness of the sample (more details of the original sample can be found in Cheung et al., 2008). For the current investigation, patients who were older than 45 and who did not function well enough to complete the whole set of personality assessment instruments were excluded. As a result, 1,419 patients (58.6% male) aged between 18 and 45 (M = 29.36; SD = 7.74) years were retained for this study. Among these patients, 2.5% of them completed primary school only, 27% completed junior high school, another 35.4% completed senior high school, and 35% had tertiary education. In terms of marital status, 56.3% of them were single, 37.3% were married, and 6.4% were divorced. The primary Axis I diagnoses of these participants included schizophrenia (n = 135), bipolar disorders (n = 202), depressive disorders (n = 284), anxiety and other neurotic disorders (n = 309), and substance use disorders (n = 135). There 554 LAI ET AL. TABLE 1. Summary of Diagnostic Criteria for DSM-IV-TR BPD, ICD-10 EUPD, and CCMD-III IPD CPAI-2 or CPDI items extracted for the simulated diagnostic assessment of DSM-IV-TR BPD, ICD-10 Diagnostic criterion EUPD-I, ICD-10 EUPD-B and CCMD-III IPD DSMIV-TR BPD ICD-10 EUPD-I CCMD-III EUPD-B IPD A. Common to all three diagnostic categories Unstable relationship I urgently seek another relationship as sources of care and support when an intimate relationship ends. My interpersonal relationship is often turbulent due to the rapid alternations between extremes of idealization and devaluation of my feelings towards others. Identity I often don’t know about myself (e.g., unstable sense disturbance of self; unstable or unclear sexual orientation, career prospects, or purposefulness). I am easily suggestible. My viewpoints can be changed remarkably within a day. Marked Sometimes I will behave impulsively regardless of the impulsivity consequences (e.g., restless spending to debt, substance abuse or alcoholic, reckless driving or speeding, having sex with people unfamiliar). I often like to do something risky or exciting, and sometimes even ignore the safety of others and myself. Suicidal behaviors I had recurrent self-injurious behaviors or ideation (e.g., self-mutilations, burning myself, etc.). Recently, I often think about suicide. Affective Sometimes I will cry for a while and then laugh for a instability while. I just cannot control myself. I am very sensitive and emotionally unstable. I am easy to get into a state of dysphoria, irritability, or anxiety, with little sense of affective stability. Anger dyscontrol I am impulsive and irritable, which makes me quarrel with others easily. I am easy to feel anger intensively, and get out of control. B. Common to DSM-IV BPD and ICD-10 EUPD Chronic emptiness I often have chronic feeling of emptiness, which makes me feel uncomfortable persistently. I frequently feel that I am at a loss for no reason. Abandonment fear I will go for drastic action to avoid abandonment of my beloved (e.g., making unceasing phone calls, tracking, write or beg people not to leave me, or even death threats). I will go to excessive lengths to obtain nurturance and support from others, to the point of volunteering to do things that are unpleasant C. Common to ICD-10 EUPD and CCMD-III IPD Quarrelsome When people talk about my weakness or mistakes, I behaviors always get mad. I get angry when my family tells me how I should live my life. Reward discounting People often say I am quick-tempered. Even if work is boring, I would still work hard on it.* D. Unique to DSM-IV BPD Transient psychotic Most of the time, I cannot make out or distinguish features whether my own experience is real or not. When I am faced with tremendous stress or emotionally pain, sometimes I will experience transient paranoid ideation, delusions, or dissociative symptoms (e.g., derealization, disengagement from the space and time of being, or even suspicion that others are harming me). E. Unique to CCMD-III IPD Inability to plan I seldom consider all angles before I make a decision. ahead I often act without thinking, and seldom look before I leap. √ — √ √ √ — √ √ √ √ — √ √ — √ √ √ √ — √ √ √ — √ √ — √ — √ √ — √ — — √ — √ — √ — √ √ — — — — — — √ Note. BPD = borderline personality disorder; IPD = impulsive personality disorder; EUPD-I = impulsive type of emotionally unstable personality disorder; EUPD-B = borderline type of emotionally unstable personality disorder. Negative-keyed item is asterisked. ANALOGOUS DIAGNOSTIC CATEGORIES 555 was 11.93% of the participants reported recurrent suicidal ideation, and 2.21% of them reported having attempted suicide in the past two weeks. The demographic and clinical characteristics of the participants by PD diagnosis are shown in Table 2. MEASURES Simulated Diagnostic Assessment for BPD, EUPD, and IPD. Since most of the clinicians in China were not familiar with making diagnoses of BPD and EUPD, a simulated diagnostic assessment procedure was developed in this study to assess the disordered personality features as specified in the DSMIV-TR BPD, ICD-10 EUPD, and CCMD-III IPD diagnostic criteria. Each PD feature was assessed by two highly relevant items extracting from the CPAI-2 and/or the Chinese Personality Disorder Inventory (CPDI; Leung, 2004). The CPAI-2, consisting of 541 dichotomous items, is an indigenously developed personality inventory, which covers personality characteristics for normal and clinical assessment (Cheung, Leung, Song, & Zhang, 2001). The CPDI, consisting of 96 dichotomous items, is also an indigenously developed instrument designed to assess disordered personality features as specified in the DSM-IV-TR. Accordingly, a total of 24 items were extracted to measure the 12 diagnostic criteria as specified in the DSM-IV-TR BPD, the ICD-10 EUPD, and the CCMD-III IPD. Each criterion may then receive three possible scores: 0 (absent or normal), 1 (probable or sub-threshold), and 2 (pathological). Triangulation sessions were conducted among the authors to check the appropriateness of the items being selected. The resulting 24 items obtained .85 on the Cronbach’s alpha estimate. Participants had to endorse both selected simulated items (i.e., score = 2) to be considered as fulfilling the specific diagnostic criterion. Diagnosis of the PD was determined according to the diagnostic rules specified in the respective diagnostic manual. To be diagnosed as a BPD case, a participant needs to fulfill five or more BPD criteria. To be diagnosed as an IPD case, a participant has to fulfill the criterion, affective instability, and marked impulsivity, plus at least three out of eight other IPD criteria. Regarding the diagnosis of EUPD, a participant diagnosed as the impulsive subtype must display at least three EUPD-I criteria, with one of which being quarrelsome behavior; while those diagnosed as the borderline subtype must display at least three EUPD-I criteria plus two or more EUPD-B criteria. The diagnostic criteria for BPD, IPD, and EUPD and items chosen for simulating the corresponding criteria are presented in Table 1. PROCEDURE Participants participated in the research project voluntarily with informed consent. They were reassured that any diagnosis or related information obtained from the study would not affect their treatment, and immediate withdrawal from the study was allowed at anytime. Clinicians participated in the project at the designated hospitals were responsible for the initial screening and conducting standard diagnostic interviews to ascertain psychiatric diag- 556 LAI ET AL. TABLE 2. Sample Characteristics (in Percentage) by Personality Disorder Diagnosis n Gender Male Female Age group 18–25 26–35 36–45 M (SD) Education Primary Secondary Tertiary Postgraduate Marital status Single Married Divorced Patient group Inpatient Outpatient Suicidality Presence of suicidal ideation Presence of suicide attempt Primary psychiatric disorders Anxiety disorders Depressive disorders Substance use disorders Bipolar disorders Schizophrenia DSM-IV-TR BPD CCMD-III IPD EUPD-I EUPD-B All 141 58 103 102 1419 57.04 42.96 63.79 36.21 54.37 45.63 60.78 39.22 58.63 41.37 43.66 30.28 26.06 28.82 (7.97) 39.66 36.21 24.14 29.36 (7.84) 45.63 31.07 23.3 28.68 (7.97) 43.14 25.49 31.37 29.42 (8.48) 39.04 34.67 26.29 29.36 (7.74) 4.23 65.49 26.76 3.52 1.72 67.24 22.41 8.62 4.85 65.05 24.27 5.83 3.92 66.67 24.51 4.9 2.54 62.44 31.71 3.31 57.04 33.8 9.15 56.9 34.48 8.62 57.28 32.04 10.68 53.92 35.29 10.78 56.31 37.28 6.41 78.87 21.13 74.14 25.86 66.02 33.98 73.53 26.47 76.11 23.89 14.49 5.22 12.28 3.51 14.71 2.94 11 4.08 11.93 2.21 19.01 14.08 16.2 18.31 32.39 18.97 8.62 25.86 20.69 25.86 24.27 9.71 20.39 20.39 25.24 26.47 6.86 18.63 18.63 29.41 21.78 20.01 9.51 14.24 34.46 Note. BPD = borderline personality disorder; IPD = impulsive personality disorder; EUPD-I = impulsive type of emotionally unstable personality disorder; EUPD-B = borderline type of emotionally unstable personality disorder. noses for the patients. Before data collection, clinicians attended a training session to standardize the diagnosis and research protocol. Participants were administered the CPAI-2, CPDI, and a questionnaire booklet asking about their basic sociodemographic information. The study was approved by the Social Science Faculty Ethic Committee of the Chinese University of Hong Kong. ANALYSES AND RESULTS INTERNAL CONSISTENCY AND CORRELATIONS BETWEEN DIAGNOSTIC CRITERIA OF BPD, IPD, AND EUPD Cronbach’s alpha of the borderline criteria set was .83, suggesting adequate internal consistency. The strong intercorrelations and item-total correlations of the “borderline” criteria provided further evidence to the internal consistency of the construct (see Table 3). Item-total correlation 1 2 3 4 5 6 7 8 9 10 11 12 - 0.35 0.39 0.29 0.34 0.23 0.32 0.27 0.27 0.25 0.39 0.38 - 0.34 0.27 0.41 0.33 0.25 0.25 0.38 0.25 0.25 0.30 - 0.36 0.30 0.37 0.48 0.39 0.30 0.19 0.29 0.34 - 0.32 0.19 0.42 0.37 0.23 0.20 0.23 0.26 - 0.23 0.29 0.26 0.36 0.22 0.22 0.29 - 0.31 0.24 0.24 0.13 0.21 0.28 - 0.43 0.24 0.19 0.26 0.30 - 0.26 0.23 0.15 0.24 - 0.14 0.14 0.24 - 0.14 0.21 - 0.29 1 Anger dyscontrol 0.55 2 Marked impulsivity 0.52 3 Affective instability 0.60 4 Identity disturbance 0.50 5 Unstable relationship 0.50 6 Suicidal behaviors 0.43 7 Chronic emptiness 0.55 8 Psychotic features 0.49 9 Abandonment fear 0.44 10 Inability to plan ahead 0.33 11 Reward discounting 0.40 12 Quarrelsome behaviors 0.48 ANALOGOUS DIAGNOSTIC CATEGORIES TABLE 3. Item-Total Correlations and Inter-Criterion Correlations of Defining Criteria of DSM-IV-TR BPD, ICD-10 EUPD, and CCMD-III IPD - Note. All correlation coefficients were significant at p < .001. Mean of inter-criterion correlations is .28 557 558 LAI ET AL. TABLE 4. Varimax Solution with Two Factors for the Borderline Construct with Defining Criteria of DSM-IV-TR BPD, ICD-10 EUPD, and CCMD-III IPD 1a 2b Chronic emptiness 0.80 0.13 Psychotic features 0.72 0.16 Affective instability 0.71 0.28 Identity disturbance 0.64 0.24 Suicidal behaviors 0.46 0.28 Marked impulsivity 0.16 0.74 Unstable relationship 0.23 0.66 Anger dyscontrol 0.33 0.63 Fear of abandonment 0.18 0.62 Inability to plan ahead 0.08 0.53 Quarrelsome behaviors 0.37 0.46 Reward discounting 0.24 0.39 Note. Extraction method is principal component analysis. Rotation method is varimax with Kaiser Normalization. Bolded factor loadings represent the criteria that correspond to each factor. aEigenvalue = 4.28; percent of variance = 22.65%. bEigenvalue = 1.09; percent of variance = 22.12%. EXPLORATORY FACTOR ANALYSIS (EFA) Prior to factor analyses, the sample was randomly split into two halves. Comparison between the two subsamples for age, gender, and Axis I disorder was performed to ensure independent grouping. Since no previous study has examined the factorial structure underlying the construct of the borderline syndrome, EFA (principal components analysis with varimax rotation) was conducted on all the 12 diagnostic criteria of BPD, IPD, and EUPD on a three-point level in the first half of the sample (S1; n = 710). Factorability of the data was revealed by the Bartlett’s test of sphericity, χ2 = 1846.29, df = 66, p < .001, and the Kaiser-Meyer-Olkin measure of sampling adequacy of 0.89. Results of EFA revealed two components with eigenvalue greater than 1.0 (factor 1 eigenvalue = 4.28; factor 2 eigenvalue = 1.09), which accounted for 44.77% of total variance. After rotation, the first factor, consisting of chronic emptiness, transient psychotic features, affective instability, identity disturbance, and suicidal behaviors, which we named affective and cognitive disturbances, explained 22.65% of the variance. The second factor, consisting of marked impulsivity, unstable relationship, anger dyscontrol, abandonment fears, inability to plan ahead, quarrelsome behaviors, and reward discounting, which we named impulse dysregulation, explained 22.12% of the variance. The standardized factor loadings are illustrated in Table 4. CONFIRMATORY FACTOR ANALYSIS (CFA) A major shortcoming of EFA is its capitalization on chance, hence different factorial structures could emerge for different samples (Kline, 2005). The two-step EFA-CFA approach allowed us to examine the replicability of the resulting factor structure from EFA in the second half of the sample (S2; n ANALOGOUS DIAGNOSTIC CATEGORIES 559 = 709). In addition, a series of other models identified in previous studies of BPD were tested for goodness of fit using EQS 6.0 (Bentler, 2005): 1. Unitary: one factor comprising of all 12 criteria from BPD, IPD, and EUPD. 2. EFA obtained solution in the present study: two correlated factor, namely (a) affective and cognitive disturbances; and (b) impulse dysregulation. 3. Three-dimensional model: derived from Sanislow et al.’s (2000, 2002) model: three correlated factors, namely (a) Disturbed relatedness, consisting of unstable relationships, identity disturbance, transient psychotic features, chronic emptiness, and quarrelsome behaviors; (b) Behavioral dysregulation, consisting of impulsivity, suicidal behaviors, inability to plan ahead and reward discounting; and (c) Affective dysregulation, consisting of affective instability, anger dyscontrol, and abandonment fears. 4. Four-dimensional model: derived from Leung and Leung’s (2009) model: four-correlated factors, namely (a) Affective dysregulation, consisting of affective instability and anger dyscontrol; (b) Impulsivity, consisting of marked impulsivity, suicidal behaviors, inability to plan ahead, and reward discounting; (c) Self disturbance, consisting of chronic emptiness and identity disturbance; and (d) Interpersonal disturbance, consisting of unstable relationship, abandonment fears, quarrelsome behaviors, and transient psychotic features. Results of CFA indicated that the unitary model only provided marginal fit to the data, χ2 (54) = 215.24, NFI = .87, NNFI = .88, CFI = .90, RMSEA = .07. Factor loadings ranging from 0.35 for inability to plan ahead to 0.68 for affective instability, were all significant. We then examined the various multidimensional models. Various fit indices indicated that the two-factor model obtained from our EFA result, χ2 (53) = 182.19, NFI = .89, NNFI = .90, CFI = .92, RMSEA = .06, provided the best fit to the data among other plausible models, three-factor model: χ2 (51) = 202.44, NFI = .88, NNFI = .88, CFI = .91, RMSEA = .07; four-factor model: χ2 (48) = 197.19, NFI = .88, NNFI = .87, CFI = .91, RMSEA = .07. Consistent with results of chi-square test for significant differences, the two-factor model fitted significantly better than the unitary model, ∆χ2 = 33.05, ∆df = 1, p < 0.001. Compared with the three-factor and four-factor model, the two-factor model reported a smaller χ2 value, even being more restricted in structure. Factor loadings and covariances of the two-factor model is presented in Figure 1. ESTIMATED PREVALENCE, ENDORSEMENT RATE AND DIAGNOSTIC EFFICIENCY According to the simulated diagnostic assessment, a total of 141(9.9%), 55(3.9%), 103(7.3%), and 102(7.2%) participants met the diagnosis of BPD, IPD, EUPD-I, and EUPD-B, respectively. Of all the participants, 1215(85.62%) were not given any of these PD diagnoses. Furthermore, a high percentage of the EUPD-I patients was further diagnosed as EUPD-B 560 LAI ET AL. 0.70 Affective instability 0.84 Identity disturbance 0.86 Self-mutilation/ Suicide 0.75 Chronic Emptiness 0.86 Psychotic Features 0.80 Anger dyscontrol 0.82 Marked impulsivity 0.70 0.54 0.51 Affective and cognitive disturbances 0.66 0.51 0.86 0.60 0.84 Unstable relationship 0.87 Fear of abandonment 0.93 Inability to plan ahead 0.85 Reward discounting 0.58 0.54 Impulse Dysregulation 0.49 0.36 0.52 0.84 0.54 Quarrelsome behaviors FIGURE 1. Final model of confirmatory factor analyses of DSM-IVTR BPD, ICD-10 EUPD and CCMD-III IPD. Standardized factor loadings are indicated by single-headed arrows; correlations between factors are indicated by double-headed arrows. All factor loadings and correlations significant at p < .05. (71.84%). Differences on demographic and clinical characteristics among the subjects with PD diagnoses to the non-PD subjects (see Table 2 for more details) were assessed by a series of contingency table analyses using the chisquare statistic. The results indicated that gender, age, education level, and marital status were not significantly dependent on the diagnoses of BPD, IPD, EUPD-I, and EUPD-B, which suggested that patients meeting these four groups of PD had similar demographic characteristics. Apart from that, the chi-square statistics indicated that no specific PD diagnosis was more significantly related to suicidal ideation, but the rate of self-reported suicide attempt only, χ2(1) = 6.15, p = .01, which was the highest among patients being diagnosed with BPD than those who were not. The odds of having attempted suicide was about three times for BPD to non-BPD patients. Table 5 shows the endorsement rate, sensitivity, specificity, positive predictive power (PPP), negative predictive power (NPP), and the overall efficiency for each of the borderline symptom criteria among different groups of PD patient. The overall diagnostic efficiency of each criterion is indicated by the Youden’s index, which is defined as J = sensitivity + specificity - 1 ANALOGOUS DIAGNOSTIC CATEGORIES 561 (Youden, 1950). Since marked impulsivity, affective instability, and quarrelsome behaviors were designated as the necessary criteria in the diagnostic set of IPD and EUPD-I, these three criteria were excluded from the analysis for the corresponding PD. Results indicated that the most prevalent criterion was chronic emptiness in BPD patients (80.71%), identity disturbance in IPD patients (74.55%), and anger dyscontrol in EUPD-I (83.5%) and EUPD-B patients (88.24%), whereas the least prevalent criterion was reward discounting in BPD patients (31.91%) and IPD patients (41.82%), and suicidal behaviors in EUPD-I (31.68%) and EUPD-B patients (42.00%). In general, the borderline symptom criteria showed high PPP (range for BPD = .92–.97; IPD = .97–.99; EUPD-I = .94–.98; EUPD-B = .94–.99), but relatively low NPP (range for BPD = .20–.54; IPD = .07–.30; EUPD-I = .14–.45; EUPD-B = .14–.47). For instance, a person who endorsed the criterion anger dyscontrol had a probability of 99% to be diagnosed as IPD, but a person who did not endorse that item still had a probability of 88% to be diagnosed as IPD. Taking both the measures of sensitivity and specificity together, the most efficient diagnostic criteria appeared to be chronic emptiness for BPD (J = .59), transient psychotic features for IPD (J = .51), anger dyscontrol for EUPD-I (J = .69), and EUPD-B (J = .73). On the other hand, inability to plan ahead was the worst performing criterion for discriminating IPD (J = .23), EUPD-I (J = .20), and EUPD-B (J = .21) patients. For BPD, the criterion of rewarding discounting, which is missing from the DSM-IV-TR BPD diagnostic set, was the one with the lowest overall efficiency (J = .23). DISCUSSION ARE BPD, IPD, AND EUPD THREE DISTINCT CLINICAL CONSTRUCTS? This study was the first to examine the validity of the BPD construct using the 12 diagnostic criteria from the DSM-IV-TR BPD, ICD-10 EUPD, and CCMD-III IPD. The high intercorrelations and Cronbach’s alpha value result, provide support for the internal consistency of the criteria set as a whole. Furthermore, CFA was also conducted to test a unitary factor model, the two-factor model obtained from EFA in this study, and the previously reported three-factor and four-factor models. Findings lent further support to the reliability of the construct in which the unitary model was found to have adequate goodness-of-fit. Paralleling with previous studies in psychiatric patients, the EFA-obtained two-factor model was revealed to provide the best fit to the data (Benazzi, 2006; Whewell, Ryman, Bonanno, & Heather, 2000). The two-factor solution reveals that affective and cognitive disturbances and impulse dysregulation may represent the two core components underlying the borderline construct. The result is in line with various previous factor analytic studies of BPD showing that the borderline syndrome is a unitary but multidimensional construct (Johansen et al., 2004; Leung & Leung, 2009; Sanislow et al., 2002). 562 TABLE 5. Item analysis for Defining Criteria of DSM-IV-TR BPD, ICD-10 EUPD, and CCMD-III IPD: Diagnostic Efficiency Statistics BPD Anger dyscontrol Marked impulsivitya IPD EUPD-I EUPD-B BR SEN SPE PPP NPP J SEN SPE PPP NPP J SEN SPE PPP NPP J SEN SPE PPP NPP J 23.01 0.85 0.71 0.56 0.96 0.36 0.79 0.71 0.50 0.99 0.12 0.85 0.83 0.69 0.98 0.33 0.85 0.88 0.73 0.99 0.34 13.40 0.94 0.61 0.55 0.95 0.54 0.90 1.00 — — 0.30 0.94 0.58 0.52 0.96 0.45 0.94 0.64 0.58 0.97 0.47 a 30.76 0.92 0.63 0.55 0.96 0.47 0.89 1.00 — — 0.26 0.92 0.62 0.54 0.97 0.39 0.92 0.72 0.64 0.98 0.42 Identity disturbance 20.60 0.75 0.79 0.55 0.97 0.28 0.71 0.75 0.46 0.99 0.10 0.75 0.58 0.34 0.95 0.17 0.75 0.70 0.45 0.97 0.19 Affective instability Unstable relationship 10.25 0.86 0.72 0.58 0.96 0.38 0.81 0.62 0.43 0.98 0.12 0.86 0.53 0.39 0.96 0.24 0.86 0.66 0.52 0.97 0.29 Suicidal behaviors 28.00 0.94 0.44 0.38 0.94 0.46 0.91 0.43 0.35 0.98 0.16 0.94 0.32 0.26 0.94 0.31 0.94 0.42 0.36 0.95 0.37 Chronic emptiness 24.05 0.79 0.81 0.59 0.97 0.31 0.74 0.73 0.47 0.99 0.10 0.79 0.59 0.38 0.96 0.19 0.79 0.71 0.50 0.97 0.22 Psychotic features 14.19 0.82 0.74 0.56 0.96 0.32 0.78 0.73 0.51 0.99 0.12 0.82 0.53 0.35 0.95 0.20 0.82 0.60 0.42 0.96 0.22 Fear of abandonment 25.71 0.90 0.54 0.44 0.94 0.39 0.87 0.49 0.36 0.98 0.14 0.90 0.33 0.23 0.94 0.22 0.90 0.43 0.33 0.95 0.27 Inability to plan ahead 13.48 0.78 0.49 0.26 0.93 0.20 0.75 0.47 0.23 0.97 0.07 0.78 0.43 0.20 0.94 0.14 0.78 0.43 0.21 0.94 0.14 Reward discounting 19.10 0.91 0.32 0.23 0.92 0.29 0.88 0.42 0.29 0.97 0.12 0.91 0.51 0.42 0.96 0.32 0.91 0.54 0.45 0.96 0.34 Quarrelsome behaviorsb 48.94 0.88 0.49 0.37 0.94 0.32 0.83 0.65 0.48 0.98 0.14 0.88 — — 1.00 0.41 0.88 0.73 0.60 0.97 0.33 Note. BPD = borderline personality disorder; IPD = impulsive personality disorder; EUPD-I = impulsive type of emotionally unstable personality disorder; EUPD-B = borderline type of emotionally unstable personality disorder; BR = base rate: number of participants with endorsement of the criterion divided by total number participants; SEN = sensitivity: percentage of participants who met the criterion given a positive PD diagnosis by the simulated diagnostic assessment; SPE = specificity: percentage of participants who did not met the criterion given a negative PD diagnosis by the simulated diagnostic assessment; PPP = positive predictive power: percentage of participants who received the corresponding PD diagnosis by the simulated diagnostic assessment given the presence of a criterion; NPP = negative predictive power: percentage of participants who did not receive the corresponding PD diagnosis by the simulated diagnostic procedure given the absence of a criterion; J = Youden’s index. aMarked impulsivity and affective instability are necessary criteria in CCMD-III IPD diagnostic set. bQuarrelsome behaviors are necessary criteria in ICD-10 EUPD-I diagnostic set. LAI ET AL. ANALOGOUS DIAGNOSTIC CATEGORIES 563 The 12 diagnostic criteria originated from BPD, EUPD, and IPD were found to cluster with each other. Along with the high correlations between the various PD features, it is plausible that BPD, EUPD, and IPD are not distinct constructs. Rather, the additional criteria included in EUPD and IPD diagnostic set, i.e., quarrelsome behaviors, reward discounting, and inability to plan ahead, may just be manifestations of the core disordered personality trait in a different context. This inference is consistent with what Zhong and Leung (2007) argued in a review article that these three PDs may represent “comparable clinical syndrome that is characterized by a pervasive pattern of mood and impulse control problems” (p. 80). The similar demographic profile displayed by the three groups of PD patients, on the other hand, lent support to the proposal that ICD-EUPD and CCMD-IPD might have described the same group of patients otherwise being diagnosed as BPD in terms of the DSM nomenclature, and vice versa. IS THE IMPULSIVE AND BORDERLINE SUBTYPE OF EUPD A VALID DIVISION? Findings of the present study also broach the discussion concerning the validity of subdividing the construct of EUPD into impulsive and borderline subtypes as described in the ICD-10 classification. Whewell and associates (2000) addressed this question by conducting EFA on the eight DSM-III-R BPD criteria in a sample of 288 BPD patients. Two factors, namely a calminternalizing factor and a mood-externalizing factor, have emerged. The mood-externalizing factor was argued to correspond to the impulsive subtype, while the hybrid of both factors was argued to correspond to the borderline subtype of EUPD. Partially convergent with Whewell et al.’s findings, we have also identified a two-factor solution. Yet, the pattern of factor loadings is dissimilar across studies. The pattern of factor loadings of our study instead is similar to that represented by the two subtypes of EUPD in the ICD-10. Apart from affective instability, all the other four criteria proposed to characterize the impulsive subtype have loaded onto the factor of impulse dysregulation. On the other hand, the borderline subtype and the factor of affective and cognitive disturbances have shared three criteria: chronic emptiness, identity disturbance, and suicidal behaviors. Intriguingly, the criterion of quarrelsome behaviors was found to have substantial loadings on both factors in EFA. This finding supports the diagnostic rule of designating quarrelsome behaviors as the necessary criterion for EUPD-I. In the literature, the problem of membership heterogeneity has been one of the frequent critiques with the construct of BPD in the DSM (Nurnberg et al., 1991; Zanarini et al., 1998a, b). Findings of our study are supportive to the binary subtype division of EUPD described by ICD-10. Therefore, the binary division of ICD-10 EUPD may represent one possible solution to address the issue of membership heterogeneity of the borderline construct. Questions remain, however, regarding the validity of this subtype division. First, according to the diagnostic criteria of EUPD, affective instability should 564 LAI ET AL. be restricted to the impulsive subtype. This line of thinking is not supported by our empirical findings which indicate that the affective instability criterion is associated with the borderline subtype. Second, as argued by Sanislow et al. (2002), identification of components in factor analysis does not necessarily imply subtypes of the disorder. High intercorrelations are revealed among the components and criteria, which indicate their interdependence in contributing to the psychopathology of the borderline construct. Apparently, personality features characterizing the borderline construct are not easy to be delineated from each other. IS IT VALID TO ADD OR DROP VARIOUS BPD, IPD, AND EUPD CRITERIA ON THE BASIS OF THEIR DIAGNOSTIC EFFICIENCY? To establish construct validity of the borderline construct, one cannot overlook its content validity, which refers to the extent to which a psychological instrument or diagnostic category reflects the specific intended domain of content. Despite the different diagnostic rules, BPD, IPD, and EUPD are all defined by a polythetic criterion set. This classification rule is inherently probabilistic in which each criterion was assumed to carry equal weight in defining the disorder. Elimination of relevant criteria with high diagnostic efficiency may diminish the content validity of the construct. As determined by the simulated diagnostic assessment, the criterion, reward discounting, which is absent from the diagnostic set of BPD, was found to have the lowest efficiency value. In addition, with its relatively low inter-criterion and item-total correlations, reward discounting was not as internally coherent as other existing BPD criteria. The relationships between reward discounting and other core BPD features need further investigation in future studies. On the contrary, the criterion transient psychotic features which is excluded from the diagnostic set of EUPD and IPD, is found to be a strong associated feature for the IPD and EUPD constructs. Of note, even being absent from the diagnostic set, this criterion has obtained a comparable level of efficiency with other defining criteria. The diagnostic and conceptual values of transient psychotic features for the EUPD and IPD constructs may need further examination in future studies. Similarly, findings of the present study indicated that the diagnostic criterion fear of abandonment outperformed inability to plan ahead and reward discounting within the IPD construct in terms of diagnostic efficiency. Further examination of this criterion is needed to explain why it has been omitted from the IPD construct. Finally, previous researchers often argued against the prototypic validity of the BPD construct on the grounds that differential diagnostic efficiency of the criteria were found (Johansen et al., 2004). Intriguingly, apart from suicidal behaviors and fear of abandonment, the remaining six BPD criteria were all found to have similar efficiency indexes with ranges from J = .55 to J = .59, which partially supported the prototype validity of the BPD construct specified in the DSM-IV-TR. ANALOGOUS DIAGNOSTIC CATEGORIES 565 LIMITATIONS AND STRENGTHS Findings of this study should be interpreted in light of certain methodological limitations. First, due to the fact that most of the clinicians in China were not familiar with making diagnoses of BPD and EUPD, and validated diagnostic instruments for assessing IPD and EUPD were not available, we were not able to conduct diagnostic interviews with our participants. Instead, we relied on a simulated diagnostic procedure to assess diagnostic criteria for BPD, IPD, and EUPD. Findings of this study, therefore, should be regarded as tentative at this stage and cross-validation of these findings with clinical diagnostic interview data is needed. Second, since all of our participants were psychiatric patients with at least one Axis I diagnosis, it is unclear to what extent active Axis I psychiatric symptoms might have confounded our assessment of personality traits. For instance, substance abuse may lead to various adjustment stresses to individuals, which later may manifest as emotional and behavioral disinhibition that resemble borderline personality features. Despite these limitations, this study has examined the construct validity of the borderline construct using all criteria from DSM-IV-TR BPD, CCMDIII IPD, and ICD-10 EUPD in Chinese psychiatric patients. Findings indicate that the borderline construct can be regarded as a coherent clinical entity with two interdependent components; i.e., affective and cognitive disturbances, and impulse dysregulation. Relationship of the various criteria and components supports the conceptualization of impulsive and borderline subtypes of EUPD in ICD-10. Finally, this study finds that the diagnostic criteria of fear of abandonment and transient psychotic features, which have been eliminated from the IPD and EUPD construct, exhibit sufficient diagnostic efficiency. Future research should further explore how these two diagnostic features are related to the other disordered personality features within the broad borderline construct. Clinicians in the U.S., Europe, and China seem to have encountered a group of psychiatric patients who display a rather similar personality and psychosocial functioning profile. However, clinical researchers on different continents seem to have placed emphasis on different symptoms in their understanding of the core pathology of these patients. As a result, three different diagnostic labels BPD, EUPD, and IPD have been put forward. The concepts of psychiatric nosology have a profound effect on our theory development and clinical practice. Further research to clarify the relationships among these constructs may expedite international communication and stimulate knowledge exchange between researchers across cultures. 566 LAI ET AL. REFERENCES American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author. American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text revision ed.). Washington, DC: Author. Alarcon, R., & Foulks, E. (1995). Personality disorders and culture: Contemporary clinical views (Part A). Cultural Diversity and Mental Health, 1, 3–17. Benazzi, F. (2006). Borderline personality–bipolar spectrum relationship. Progress in Neuropsychopharmacology & Biological Psychiatry, 30, 68–74. Bentler, P. (2005). EQS 6 structural equations program manual. Encino, Los Angeles, CA: Multivariate Software. Cheung, F. M., Cheung, S. F., & Leung, F. (2008). Clinical utility of the cross-cultural (Chinese) personality assessment inventory (CPAI-2) in the assessment of substance use disorders among Chinese men. Psychological Assessment, 20, 103–113. Cheung, F. M., Leung, K., Song, W. Z., & Zhang, J. X. (2001). The cross-cultural (Chinese) personality assessment inventory—2 (CPAI–2). Hong Kong: The Chinese University of Hong Kong. Chinese Psychiatric Association. (2001). The Chinese classification of mental disorders (CCMD-III). Jinan: Shandong Science Press. Fossati, A., Maffei, C., Bagnato, M., Donati, D., Namia, C., & Novella, L. (1999). Latent structure analysis of DSM-IV borderline personality disorder criteria. Comprehensive Psychiatry, 40, 72–79. Hyler, S. E. (1994). PDQ-4+ Personality questionnaire-4. New York: Author. Jia, Y. C. (1998). Some opinions to the “borderline personality disorder.” Journal of Clinical Psychological Medicine, 8, 248–249. Johansen, M., Karterud, S., Pedersen, G., Gude, T., & Falkum, E. (2004). An investigation of the prototype validity of the borderline DSM-IV construct. Acta Psychiatrica Scandinavica, 109, 289–298. Kline, R. B. (2005). Measurement models and confirmatory factor analysis. In R. B. Kline (Ed.), Principles and practice of structural equation modeling (pp. 165–208). New York: Guilford Press. Leung, F. (2004, December). Clinical application of the Chinese personality disorders inventory. Paper presented at the 3rd Annual Meeting of Taiwan Association of Clinical Psychology, Taiwan. Leung, F., Chan, K., & Cheung, F. (2007, July). Construct validity and prevalence of borderline personality disorder among Chinese psychiatric patients in China. Paper presented at the World Congress of Behavioral and Cognitive Therapies, Barcelona, Spain. Leung, F., Cheung, F., & Cheung, S. F. (2004, July). CPAI-2 profiles of Chinese female psychiatric patients displaying a significant level of borderline personality disorder features. Paper presented at the World Congress of Behavioral and Cognitive Therapies, Barcelona, Spain. Leung, S. W., & Leung, F. (2009). Construct validity and prevalence rate of borderline personality disorder among Chinese adolescents. Journal of Personality Disorders, 23, 494–513. Linehan, M. (1993). Cognitive-behavioral treatment of borderline personality disorder. New York: The Guilford Press. Lu, N., Liu, X., Zhu, C., Yang, Y., Xian, C., & Yu, Z. (2001). Four existing instruments for assessing personality disorders: I. Chinese Mental Health Journal, 15, 125–128. Luo, X. L., & Huang, Y., L. (2002). Discussion on the third edition of Chinese classification of mental disorders (CCMD-III). Journal of Clinical Psychological Medicine, 12, 303–304. Nurnberg, H. G., Raskin, M., Levine, P. E., Pollack, S., Siegel, O., & Price, R. (1991). The comorbidity of borderline personality disorder and other DSM-III-R axis II personality disorders. American Journal of Psychiatry, 148, 1371–1377. Paris, J. (1994a). Borderline personality disorder: A multidimensional approach. Washington, DC: American Psychiatric Press. Paris, J. (1994b). The etiology of borderline personality disorder: A biopsychosocial approach. Psychiatry, 57, 316–326. Sanislow, C. A., Grilo, C. M., & McGlashan, T. H. (2000). Factor analysis of the DSM-IIIR borderline personality disorder criteria in psychiatric inpatients. American Journal of Psychiatry, 157, 1629–1633. Sanislow, C. A., Grilo, C. M., Morey, L. C., Bender, D. S., Skodol, A. E., Gunderson, J. G., et al. (2002). Confirmatory factor analysis of DSM-IV criteria for borderline personality disorder: Findings from the collaborative longitudinal personality disorders study. American Journal of Psychiatry, 159, 284–290. Skodol, A., Gunderson, J., McGlashan, T., Dyck, I., Stout, R., Bender, D., et al. (2002). Functional impairment in patients with schizotypal, borderline, avoidant, or obsessivecompulsive personality disorder. American Journal of Psychiatry, 159, 276. Skodol, A., Oldham, J., Bender, D., Dyck, I., Stout, R., Morey, L., et al. (2005). Dimensional ANALOGOUS DIAGNOSTIC CATEGORIES representations of DSM-IV personality disorders: Relationships to functional impairment. American Journal of Psychiatry, 162, 1919. Torgensen, S., Kringlen, E., & Cramer, V. (2001). The prevalence of personality disorders in a community sample. Archives of general psychiatry, 58, 590–596. Wang, Y., Leung, F., & Zhong, J. (2008). The adaptation of McLean screening instrument for borderline personality disorder among Chinese college studies. Chinese Journal of Clinical Psychology, 16, 258–260. Whewell, P., Ryman, A., Bonanno, D., & Heather, N. (2000). Does the ICD 10 classification accurately describe subtypes of borderline personality disorder? British Journal of Medical Psychology, 73, 483–494. Widiger, T. A., & Frances, A. (1985). The DSMIII personality disorders. Perspectives from psychology. Archives of General Psychiatry, 42, 615–623. Widiger, T. A., & Weissman, M. M. (1991). Epidemiology of borderline personality disorder. Hospital Community Psychiatry, 42, 1015–1021. World Health Organization. (1992). The ICD-10 classification of mental and behavior disorders: Clinical descriptions and diagnostic guidelines. Geneva: World Health Organization. Yang, J., Bagby, R. M., Costa, P. T., & Ryder, A. G. (2002). Assessing the DSM-IV structure of personality disorder with a sample of Chinese psychiatric patirns. Journal of Personality Disorders, 16, 317–331. 567 Yang, J., Rorbert, R. M., & Paul, T. C. (2000). The cross-cultural generalization of axis-II constructs: A evaluation of two personality disorder assessment instrument in the people’republic of China. Journal of Personality Disorders, 14, 249–263. Youden, W. (1950). Index for rating diagnostic tests. Cancer, 3, 32–35. Zanarini, M. C., Frankenburg, F. R., Dubo, E. D., Sickel, A. E., Trikha, A., Levin, A., et al. (1998a). Axis I comorbidity of borderline personality disorder. American Journal of Psychiatry, 155, 1733–1739. Zanarini, M. C., Frankenburg, F. R., Dubo, E. D., Sickel, A. E., Trikha, A., Levin, A., et al. (1998b). Axis II comorbidity of borderline personality disorder. Comprehensive Psychiatry, 39, 296–302. Zhong, J., & Leung, F. (2007). Should borderline personality disorder be included in the fourth edition of the Chinese classification of mental disorders? Chinese Medical Journal, 120, 77-82. Zhong, J., & Leung, F. (2009). Diagnosis of borderline personality disorder in China: Current status and future directions. Current Psychiatry Reports, 11, 69–73. Zimmerman, M., Rothschild, L., & Chelminski, I. (2005). The prevalence of DSM-IV personality disorders in psychiatric outpatients. American Journal of Psychiatry, 162, 1911–1918.