Professional Documents
Culture Documents
Elise M. Rossiter
W. Stewart Agras
Christy F. Telch
John A. Schneider
(Accepted 25 April 1992)
The Personality Disorders Examination (PDE) was administered to 7 1 bulimia nervosa pa-
tients at baseline assessment in a study comparing the effectiveness of cognitive-behavioral
treatment with desipramine or the combination of both treatments. Personality disorder sub-
scales were combined into single DSM-Ill-R cluster scores. A high cluster B score (consist-
ing of antisocial, borderline, histrionic, and narcissistic features) significantly predicted poor
outcome at 16 weeks and was a better predictor of outcome than borderline personality
characteristics alone or any other DSM-Ill-R cluster score or Combination of cluster scores.
In contrast pretreatment depression level, self-esteem, degree of dietary restraint, frequency
of purging, and history of anorexia nervosa were not significantly related to outcome. At
1-year follow-up there was still a trend toward high cluster B scores predicting poor treat-
ment outcome. Cluster B score was not significantly correlated with percentage of sessions
attended nor did subjects with higher cluster B scores have a better outcome with either
specific treatment. These results suggest that further investigation of alternative treatments is
warranted with high cluster B individuals to determine if treatment effectiveness can be im-
proved. 0 1993 by john Wiley & Sons, lnc.
Relatively few studies have evaluated predictors of outcome in the treatment of bulimia
nervosa. Wilson, Rossiter, Kleifield, and Lindholm (1986) reported that nonresponders
to cognitive-behavioral treatment had significantly lower self-reported past weights than
treatment responders. Similarly, Agras, Dorian, Kirkley, Arnow, and Bachman (1987)
found that in bulimics treated with imipramine, those with lower reported body mass
indices (BMIs) did less well than those with higher BMIs. Fairburn, Kirk, OConnor,
Anastasiades, and Cooper (1987), found that self-esteem, as measured by the Rosen-
berg Self-Esteem Scale (Rosenberg, 1965), was a significant predictor of treatment out-
The authors are affiliated with The Department of Psychiatry, and Laboratory for The Study of Behavioral Medi-
cine, Stanford University School of Medicine. Requests for reprints should be addressed to W Stewart Agros, Stan-
ford University School of Medicine, TD-209, Laboratory for the Study of Behavioral Medicine, Stanford, CA 94305
Table 1. Prevalence rates of BPD a n d other personality disorders (I’D) i n published studies
Authors Instrument BPD Any PD
disorder is frequently difficult because people exhibit traits that are not limited to a sin-
gle personality disorder. We grouped together the features of cluster B into a single
composite score, and tested this as a predictor of treatment outcome. We hypothesized
that if personality characteristics affected outcome then the cluster B score would be a
better predictor of outcome than borderline characteristics alone.
METHOD
Participants
Participants in this investigation were 71 females between the ages of 18 and 65 years
who met DSM-111-R (APA, 1987) criteria for bulimia nervosa and were the participants
in a randomized study comparing the efficacy of desipramine, cognitive-behavioral ther-
apy, or their combination (Agras et al., 1992). Exclusion criteria included the presence
of: (1)a concurrent medical condition that would preclude the use of antidepressants
including evidence of conduction disturbance on electrocardiography; (2) current ano-
rexia nervosa; (3) current drug or alcohol abuse; (4)current psychosis; and (5) current
depression with suicidal risk of sufficient severity to preclude the use of antidepres-
sants on an outpatient basis. In the above study both the cognitive-behavioral and com-
bined treatments were significantly superior to medication alone and not different from
each other at the posttreatment assessment. The combined group had reduced purging
by 80.1% with 64% of subjects abstinent, the cognitive-behavioral group had reduced it
352 Rossiter et al.
by 82.6% with 48% abstinent, and the medication group had reduced it by 42.7% with
33% of subjects abstinent from purging. At 1-year follow-up the abstinence rates were:
53%, 38%, and 27% for the combined, cognitive-behavioral, and medication treatments,
respectively.
Sample Description
*
The participants were aged 29.6 8.9 years. Thirty-two percent were married, 56.3%
were single, and the remainder were either divorced (8.5%) or separated (2.8%). Fifty-
six percent had completed college, 36.7% had completed some college, and the remain-
der had either completed high school (5.6%) or some high school (1.4%);83% were
employed at the time of the study. Binge eating had begun at 19.9 5 5.7 years of age
*
and purging a few months later at 20.7 5.9 years of age. The participants reported an
ideal weight of 53.7 2 5.8 kg and their weight as measured at baseline was 59.9 -+ 9.1
*
kg. Frequency of purging at pretreatment was 9.2 6.9 times per week and binge eat-
ing 7.5 & 5.7 times per week. Twenty-two percent had been diagnosed as having an-
orexia nervosa in the past.
Eating Questionnaire, cognitive restraint factor; (4) purge frequency at baseline; (5) his-
tory of anorexia nervosa; and (6) the PDE cluster B score. The PDE cluster B score was
derived by converting the individual raw scores on the borderline, narcissistic, histri-
onic, and antisocial subscales of the PDE to standard z scores. This was done to com-
pensate for the unequal number of items on the four factors. These four standard scores
were then summed to arrive at the PDE cluster B score. The Bonferroni correction was
applied to interpret significance levels on the multiple logistic regression to compen-
sate for the large number of independent variables. Thus, results were considered sta-
tistically significant at the .008 level or beyond.
The multiple logistic regression was then repeated using the standard score of the bor-
derline scale of the PDE as opposed to the entire cluster B score to see if combining these
factors improved the predictive ability over and above that of the borderline factor alone.
Finally, the prevalence of each type of personality disorder diagnosis identified using
the PDE was calculated as well as the prevalence of at least one Axis I1 diagnosis among
subjects.
RESULTS
The prevalence of each personality disorder diagnosis is presented in Table 2. As seen
in Table 2 nearly 43% of subjects met DSM-111-R (APA, 1987) criteria for at least one
personality disorder. The most prevalent diagnosis was borderline personality, with
21.4% of subjects meeting criteria for this diagnosis. In addition, 28.5% of participants
met criteria for histrionic, narcissistic, and antisocial personality disorders.
The multiple logistic regression revealed that the only significant predictor of out-
come at 16 weeks was the cluster B score of the PDE, x2 = 9.51, p < .002. The odds
ratio was .67 which means that for the first unit increase on the PDE cluster B score
(indicating increased pathology) the odds of successful outcome (i.e., abstinence from
purging) decreased by .67. For the second unit of increase the odds of successful out-
come decreased by an additional .67 squared or .45. The mean PDE cluster B score was
0 (these were summed standard scores) with a standard deviation of 3.2. The minimum
score was -5.86 and the maximum was 8.04. None of the other independent variables,
that is, pretreatment frequency of purging, history of anorexia nervosa, Stunkard and
Messick Dietary Restraint, BDI, or Rosenberg Self-Esteem scales significantly predicted
treatment outcome.
A Paranoid 2 2.9
Schizoid 0 0.00
Schizotypal 1 1.4
B Borderline 15 21.4
Histrionic 5 7.1
Narcissistic 7 10.0
Antisocial 8 11.4
C Compulsive 7 10.0
Dependent 10 14.3
Avoidant 10 14.3
Passive-aggressive 1 1.4
At least one Axis I1 diagnosis 30 42.9
354 Rossiter et al.
Post-hoc Analyses
A post-hoc analysis was performed to determine whether any of the other personal-
ity clusters, that is, cluster A (dependent, avoidant, passive-aggressive, and obsessive
compulsive) and cluster C (paranoid, schizoid, and schizotypal) either alone or in com-
bination with each other or cluster B, significantly enhanced the prediction of treatment
outcome using the same logistic regression strategy. None of the other clusters or com-
binations of clusters significantly enhanced prediction of abstinence at 16 weeks over
and above the predictive ability of cluster B alone.
A second post-hoc analysis of covariance was performed to determine whether sub-
jects with higher cluster B scores tended to have differential outcomes in either the
cognitive-behavioral or medication treatment conditions. This revealed no significant
advantage for high cluster B subjects in either treatment.
To examine whether cooperation with treatment may have affected the outcome dif-
ferentially for those with high cluster B scores, a third post-hoc analysis was performed.
This analysis revealed that there was no significant correlation between cluster B score
and percentage of sessions attended.
DISCUSSION
The prevalence rates of BPD and any Axis I1 diagnoses found in this study were com-
parable to those of other studies reported in the literature. Our prevalence rates most
closely resembled those of Levin and Hyler (1986) with some 22% of subjects meeting
criteria for BPD and 40% meeting criteria for at least one Axis I1 diagnosis.
The results of this study support clinical observations suggesting that patients with a
higher level of personality disturbance have a poorer outcome in the treatment of bu-
limia nervosa. Of note was the superiority in predicting outcome using a cluster score
rather than relying on a single diagnostic category, that is, BPD per se. When we used
only the BPD score, our results replicated those of Garner et al. (1990) in that they only
approached significant predictive ability. However, the cluster B score was highly sig-
nificant in predicting outcome at 16 weeks. Of particular interest was the specific pre-
dictive ability of cluster B relative to the other personality disorder clusters. Although a
large percentage of subjects received high scores on cluster C (avoidant, dependent,
compulsive, and passive-aggressive), this did not appear to be related to poor treat-
ment outcome. Thus, there appears to be something unique about those characteristics
in the impulse controlldramatic spectrum that lead to poorer outcome in the cognitive-
Cluster B Personality 355
ment of BPD and by Johnson, Connors, and Tobin (1987) in the treatment of bulimia
nervosa. This broader and more intensive focus of treatment may lead to improvement
in bulimic symptoms by helping such patients control the intense affective instability
and longstanding interpersonal difficulties that are so prevalent in these patients.
This research was supported by NIMH grant 38637 to W. Stewart Agras.
Agras, W.S., Dorian, B., Kirkley, B.G., Arnow, B., & Bachman, J. (1987). Imipramine in the treatment of
bulimia: A double-blind controlled study. International Journal of Eating Disorders, 6 , 29-38.
Agras, W.S., Rossiter, E.M., Arnow, B., Telch, C.F., Raeburn, S.D., Schneider, J., Bruce, B., Perl, M., &
Koran, L. (1992). Pharmacologic and cognitive-behavioral treatment for bulimia nervosa: A controlled com-
parison. American Journal of Psychiatry, 149, 82-87.
American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., revised)
Washington D.C.: Author.
Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depres-
sion. Archives of General Psychiatry, 4, 561-571.
Cooper, J.L., Morrison, T.L., Bigman, O.L., Abramowitz, S.I., Blunden, D., Nassi, A,, & Krener, P.C. (1988).
Bulimia and borderline personality disorder. International Journal of Eating Disorders, 7, 43-49.
Fairburn, C.G., Kirk, J., OConnor, M., Anastasiades, P., & Cooper, P.J. (1987). Prognostic factors in bulimia
nervosa. British Journal of Clinical Psychology, 26, 223-224.
Garner, D.M., Olmsted, M.P., Davis, R., Rockert, W., Goldbloom, D., & Eagle, M. (1990). The association
between bulimic symptoms and reported psychopathology. International Journal of Eating Disorders, 9, 1-15.
Gartner, A.F., Marcus, R.N., Halmi, K., & Loranger, A.W. (1989). DSM IIIR personality disorders in patients
with eating disorders. American Journal of Psychiatry, 146, 1585- 1591.
Gwirtsman, H.E., Roy-Byrne, P., Yager, I., & Gerner, R.H. (1983). Neuroendocrine abnormalities in bulimia.
American Journal of Psychiatry, 240, 559-563.
Hyler, S.E., Lyons, M., heder, R.O., Young, L., Williams, J.B.W., & Spitzer, R.L. (1990). The factor struc-
ture of self-report DSM I11 Axis I1 symptoms and their relationship to clinician’s ratings. American Journal
of Psychiatry, 147, 751-757.
Johnson, C., Connors, M., & Tobin, D. (1987). Symptom management of bulimia. Journal of Consulting and
Clinical Psychology, 55, 668-676.
Johnson, C., Tobin, D., & Dennis, A. (1990). Differences in treatment outcome between borderline and non-
borderline bulimic subjects at one year follow-up. International Journal of Eating Disorders, 9, 617-627.
Johnson, C., Tobin, D., & Enright, A. (1989). Prevalence and characteristics of borderline patients in an eating-
disordered population. Journal of Clinical Psychiatry, 50, 9- 15.
Levin, A.P., & Hyler, S.E. (1986). DSM-IIIR personality diagnosis in bulimia. Comprehensive Psychiatry, 27,
47-53.
Linehan, M.M. (1987a). Dialectical behavior therapy for borderline personality disorder. Theory and method.
Bulletin of the Menninger Clinic, 51, 261-276.
Linehan, M.M. (1987b). Dialectical behavior therapy: A cognitive behavioral approach to parasuicide. Journal
of Personality Disorders, 1, 328-333.
Linehan, M.M. (1987~).Dialectical behavior therapy in groups: Treating borderline personality disorders and
suicidal behavior. In C.M Brady (Ed.), Womens therapy in groups: Paradigms of feminist treatment (pp.
145-162). New York: Springer.
Loranger, A.W., Susman, V.L. Oldham, J.M., & Russakoff, L.M. (1987). The personality disorder examina-
tion: A preliminary report. Journal of Personality Disorders, 1, 1-13.
Pope, H.G., Frankenburg, F.R., Hudson, J.I., Jonas, J., & Yergelun-Todd, D. (1987). Is bulimia associated
with borderline personality disorder? A controlled study. Journal of Clinical Psychiatry, 48, 181- 184.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.
Rossiter, E.M., Wilson, G.T., & Goldstein, L. (1989). Bulimia nervosa and dietary restraint. Behaviour Research
and Therapy, 27, 465-468.
Sansone, R.A., Seuferer, S., Fine, M.A., & Bovenzi, J. (1989). Prevalence of borderline personality symptom-
atology among women with eating disorders. Journal of Clinical Psychology, 45, 603-610.
Schmidt, N.B., & Telch, M. J. (1989, November). Prevalence of personality disorders among bulimics, non-bulimic
binge eaters and normal controls. Paper presented at the annual meeting of AABT, Washington, DC.
Stunkard, A.J., & Messick, S. (1985). The three-factor eating questionnaire for measure dietary restraint, dis-
inhibition and hunger. Journal of Psychosomatic Research, 29, 71-83.
Turner, R.M. (1989). Case study evaluations of a bio-cognitive-behavioral approach for the treatment of bor-
derline personality disorder. Behavior Therapy, 20, 477-489.
Cluster B Personality 357
Wilson, G.T., Rossiter, E.M., Kleifield, E., & Lindholm, L. (1986). Cognitive-behavioral treatment of bulimia
nervosa: A controlled evaluation. Behaviour Research and Therapy, 24, 277-288.
Wonderlich, S.A., Swift, W.J., Slotnick, H.B., & Goodman, S. (1990). DSM IIIR personality disorders in eat-
ing disorder subtypes. Znternational Journal of Eating Disorders, 9, 607-616.
Yager, J., Landsverk, J., Edelstein, C.K., & Hyler, S.E. (1989). Screening for Axis I1 personality disorders in
women with bulimic eating disorders. Psychosomatics, 30, 255-262.
Yates, W.R., Sieleni, B., & Bowers, W.A., (1989). Clinical corrlates of personality disorder in bulimia ner-
vosa. lnternational journal of Eating Disorders, 8, 473-477.
Zanarini, M.C., Frankenburg, F.R., Pope, H.G., Hudson, J.I., Yergelun-Todd, D., & Cicchetti, J. (1990). Axis
I1 comorbidity of normal weight bulimia. Comprehensive Psychiatry, 30, 20-24.