The empirical literature points to the association of severe childhood trauma with both borderline personality disorder and posttraumatic stress disorder (PTSD) and has highlighted the overlapping nature of both disorders (1). In adhering to the DSM, clinicians have been placed in the contradictory position of having to affirm that individuals manifesting characteristic symptoms of both borderline personality disorder and PTSD have two separate illnesses. This approach condemns individuals to potentially numerous diagnostic categories and has contributed to cross purposes in treatment (1, 2).

An attempt to collapse the conceptual space between the DSM-IV axis I and II diagnoses of PTSD and borderline personality disorder, respectively, is represented by the elaboration of the disorder of extreme stress or complex PTSD (1, 3–5) as a new nosological entity. Complex PTSD is currently described with the associated features and disorders of PTSD in association with an interpersonal stressor (e.g., childhood sexual and/or physical abuse) and integrates the disorders of affect regulation, dissociation, somatization, and altered perceptions of the self and others (5). The construct of complex PTSD may be a useful addition to the DSM nomenclature (1, 3–5).

The high rates of childhood sexual abuse in subjects with both borderline personality disorder and complex PTSD (3, 4, 6) led to the current empirical evaluation of a group of women with such histories.

The goal of this study was to determine in a subset of women who met diagnostic criteria for both borderline personality disorder and complex PTSD if early-onset (i.e., ≤12 years of age) versus late-onset (i.e., ≥13 years of age) sexual abuse (3) was the primary predictor of a diagnosis of borderline personality disorder.

Method

A convenience group of 65 women (≥18 years of age), drawn from their telephone response to flyers posted in three Canadian urban mental health centers, was composed of 38 people who had a history of early-onset sexual abuse (age: mean=4.63 years, SD=2.66) and 27 who had a history of late-onset sexual abuse (age: mean=15.56, SD=2.59) (3, 4). There were no significant demographic differences between the groups. A substantial proportion of the participants were Caucasian (92.3%) outpatients (98.5%) who were an average age of 38.32 years (SD=11.26, range=19–64). After complete description of the study to the participants, written informed consent was obtained. An investigator (L.M.M.) then administered measures in a one-time interview. A psychology doctoral student provided a second set of diagnostic scores, and interrater reliabilities of 96% were achieved.

A diagnosis of borderline personality disorder was determined through administration of the Revised Diagnostic Interview for Borderlines (7), while a diagnosis of current and/or lifetime complex PTSD was assessed by use of the Structured Interview for Disorders of Extreme Stress (5). Psychometric properties are reported in the Trauma Center Assessment Package (8).

The Traumatic Antecedents Questionnaire, included in the Trauma Center Assessment Package (8), assessed lifetime experiences in 10 domains (e.g., neglect, physical trauma, sexual trauma) at four developmental periods (i.e., ≤6, 7–12, 13–18, and ≥19 years of age). A study reported in the Trauma Center Assessment Package found Traumatic Antecedents Questionnaire scores to be significantly related to symptoms of PTSD and symptoms of complex PTSD.

The women with early-onset sexual abuse and those with late-onset sexual abuse were compared by using Pearson’s chi-square tests with continuity adjustment for categorical data (e.g., intrafamilial sexual abuse [yes/no] and borderline personality disorder/complex PTSD [yes/no]). Logistic regression was used to determine predictors of diagnostic outcomes.

Results

The diagnoses of both borderline personality disorder and complex PTSD were significantly higher (Pearson’s χ²=57.33, df=1, N=65, p<0.0001, two-tailed) in the participants reporting early onset of sexual abuse (94.7%, N=36) than in the subjects with late-onset abuse (0%), suggesting a relationship between early onset of sexual abuse and the diagnoses of both borderline personality disorder and complex PTSD (Cramer’s V=0.94, p<0.0001). The two groups differed significantly (t=5.23, df=63, p<0.0001) on descriptors of sexual abuse: The early-onset group reported higher intrafamilial rates (72% versus 44%, respectively) (Pearson’s χ²=17.90, df=1, p<0.0001, N=65; Cramer’s V=0.53, p<0.0001, two-tailed), no single incidents of abuse (0% versus 41%) (Pearson’s χ²=24.42, df=2, p<0.0001, N=65; Cramer’s V=0.61, p<0.0001, two-tailed; Holm’s sequential Bonferroni: χ²=19.23^a, df=1, p<0.0001; Cramer’s V=0.58, p<0.0001, two-tailed, Fisher’s exact test=^a, respectively), higher lifetime revictimization (97% versus 51%) (Pearson’s χ²=76.35, df=1, p<0.0001, N=65; Cramer’s V=0.50, p<0.0001, two-tailed), and higher biparental neglect (71% versus 22%) (Pearson’s χ²=15.06, df=1, p<0.0001, N=65; Cramer’s V=0.48, p<0.0001, two-tailed) than the late-onset group.

Logistic regression analysis showed that intrafamilial (paternal) sexual abuse and sexual abuse were significant predictors of meeting the criteria for both borderline personality disorder and complex PTSD (Wald’s χ²=5.11, df=7, p<0.05 [beta=2.08, p=0.02], and Wald’s χ²=4.18, df=7, p<0.05 [beta=0.36, p=0.04], respectively). Intrafamilial (paternal) sexual abuse increased the odds of meeting the criteria for both diagnoses by 26%, while sexual abuse increased the odds of meeting the criteria for both borderline personality disorder and complex PTSD by 25%.

Multicollinearity was addressed by removing the predictor variable for intrafamilial (paternal) sexual abuse from the equation in a repeated regression. The new analysis echoed the first and found sexual abuse to be the most significant predictor (Wald’s χ²=4.18, df=7, p<0.05 [beta=0.36, p=0.04]) of the diagnoses of both borderline personality disorder and complex PTSD.

Using two-way and three-way contingency table analyses, respectively, we examined the predictor variable for intrafamilial sexual abuse, as opposed to the variable for extrafamilial abuse, and the variable for chronicity (i.e., 10 or more incidents of sexual abuse), as opposed to the variable for acute (i.e., more than one and fewer than 10 incidents of sexual abuse) or the variable for a single acute of sexual abuse in regard to the diagnoses of both borderline personality disorder and complex PTSD. Both predictor variables were shown to be statistically significant (Pearson’s χ²=18.31, df=1, p<0.0001, two-tailed; Cramer’s V=0.53; p<0.0001) (single acute versus chronic sexual abuse, respectively: Pearson’s χ²=21.16, df=2, p<0.0001; Cramer’s V=0.57, p<0.0001; Holm’s sequential Bonferroni: χ²=19.23, df=1, p<0.0001^a, Fisher’s exact test=^a, two-tailed; Cramer’s V=0.58, p<0.0001).

Discussion

To our knowledge, this study represents the first attempt to empirically integrate and make sense of the degree of overlap between symptoms of borderline personality disorder and complex PTSD in a clinical group of sexually abused women. Since virtually all of the women with a history of childhood sexual abuse met the diagnostic criteria for both borderline personality disorder and complex PTSD, the findings offer robust support that this group can be separated from the axis II diagnosis of borderline personality disorder and subsumed under the construct of complex PTSD.

Complex PTSD, as a diagnosis, is reflective of an admixture of axis I (state) and axis II (trait) symptoms and thus offers an expanded way of thinking about a single diagnosis for this group of women. In contrast to the dichotomous position of comorbid diagnoses, complex PTSD embodies the constellation of symptoms resulting from such early and adverse events (4).

Consideration of disorders of extreme stress (complex PTSD) as a nosological classification in DSM for a subset of borderline patients broadens our way of thinking about diagnoses and allows more room to consider the hierarchy of problems that the initial trauma sets into motion. A shared understanding of the survivor’s characteristic disturbances of relationship directs clinicians in using the various tools needed to help reduce the suffering so often observed in this population (1, 4).

Two limitations of this study—namely, the use of an all-female group and the use of a convenience group—indicate that our findings may not be generalizable to women with a history of sexual abuse. In regard to the first point, a substantial proportion of those diagnosed with borderline personality disorder are women (1).

This study highlights the need for replication in order to clarify diagnostic dilemmas and the integrity of disorders of extreme stress or complex PTSD in contrast to comorbid diagnoses. Furthermore, the significance of future empirical, methodically rigorous, correlational, and multivariate research in illuminating the ways in which childhood sexual abuse may contribute to adult psychopathology is underscored.

Presented in part at Harvey Stancer Research Day, University of Toronto, Department of Psychiatry, Toronto, June 21, 2001. Received Jan. 30, 2002; revision received June 18, 2002; accepted July 25, 2002. From the Faculty of Nursing and the Women’s Mental Health & Addiction Research Program, Center for Addiction & Mental Health, Department of Psychiatry, Clarke Division, University of Toronto. Address reprint requests to Dr. McLean, Women’s Mental Health & Addiction Research Program, Center for Addiction & Mental Health, Department of Psychiatry, Clarke Division, University of Toronto, 250 College St., Rm. 610B, Toronto, Ont., M5T 1R8 Canada; linda_mclean@camh.net (e-mail). The authors thank Anne Alonso, Ph.D., Kjell E. Rudestam, Ph.D., and Charles. E. Elliott, Ph.D., of the Fielding Graduate Institute, Santa Barbara, Calif.; Nira Kolers, Ph.D., the Coach House Clinic, Toronto; and Noreen Stuckless, Ph.D., Sunnybrook and Women’s Health Sciences Center, Women’s College Campus, Toronto, for their expertise.