Pascale M. Lehoux, *, Howard Steiger, and Sheila Jabalpurlawa

Eating Disorders Unit, Douglas Hospital, Montreal, (Verdun), Quebec, Canada

Objective: This study compared 55 women with active bulimic symptoms, 18 in remission from a bulimic eating disorder, and 31 who showed no evidence of a past or present eating disorder, on selected personality and psychiatric features. Method: Discriminant function analyses were used to isolate dimensions that differentiated active patients from patients in remission, and controls (i.e. that would logically constitute “state”-related disturbances, and then dimensions that differentiated clinical cases (whether active or in remission) from non-eating-disordered controls (i.e., that might reflect stable trait pathology associated with bulimic syndromes, whether active or not). Results: Measures of depression, suicidality, and anxiety loaded significantly on the first function (differentiating from active bingers from all other cases), whereas narcissism differentiated both clinical groups from non-eating-disordered controls. Discussion: In light of theoretical and empirical evidence stressing the etiological role of narcissistic disturbances in bulimic syndromes, we interpret our findings as suggesting that narcissism may be a common trait (persisting even after remission of bulimic symptoms) in those who develop bulimic eating syndromes. Alternatively, depression, suicidality, and anxiety appear to be state-dependent features that resolve in many cases, along with remission of bulimic symptoms. We discuss various clinical and theoretical implications of our findings. ©2000 by John Wiley & Sons, Inc. In J Eat Disord 27: 36-52, 2000

Key words: bulimic syndromes; remission; trait pathology

INTRODUCTION

Studies on bulimic patients in treatment often report that improvements in bulimic symptoms coincide with substantial reductions in depression, mood lability, anxiety, parasuicidality, and other symptoms (Steiger,Leung, Thibodeau,Houle, & Ghadirian, 1993). Such observations suggest that at least some psychiatric manifestations in bulimic

syndromes need to be construed as “state”disturbances associated with eating disorder (ED) sequelae, not “trait” features that exist independently of the ED. Many findings have, however, associated bulimic syndromes with stable psychopathological traits (Steiger & Stotland, 1996; Wonderlich, Fullerton, Swift, & Klein, 1994). All of the preceding could be interpreted as suggesting an underlying trait pathology that preexists bulimic symptom onset and persists following its remission.
The question therefore arises: In bulimic syndromes, what psychopathological characteristics constitute state-related disturbances (associated only with the actively bulimic state),and what constitute trait disturbances, present whether the sufferer is in an active phase of ED or not?
We explored such state/trait distinctions, comparing loadings on selected psychopathological indices across active bingers, bingers in remission, and non-eating disordered controls. Among the psychological dimensions explored, pathological narcissism was of particular interest here, given (a) the theoretical view that bulimic features may be a symptom, in narcissistically compromised individuals, of overinvestment in body and self-image (Johnson, 1991), and (b) recent findings indicating that narcissism loads more heavily, on average, in binge eating sufferers than in women with other psychiatric disturbances such as affective and anxiety disorders (Steiger, Jabalpurlawa, Champagne, & Stotland,1997).

METHOD

Participants

Active bingers and bingers in remission were recruited through outpatient services at a specialized eating disorders unit (EDU). ED diagnoses were confirmed using the eating Disorders Examination (EDE) interview (Fairburn & Cooper, 1993),described fully below.
According to interviews, of 55 active bingers, 35 met criteria outlined in the 4th ed. Of the Diagnostic and Statistical Manual for Mental Disorders (DSM-IV; American Association, 1994) for bulimia nervosa (BN) purging subtype, 8 for BN nonpurging subtype, and 8 a bulimia-spectrum eating disorder not otherwise specified in (ED-NO5; 4 showing a subthreshold BN purging subtype [binging once vs. twice weekly] and 4 a syndrome consistent with binge eating disorder [binge eating without compensatory behaviors]). Four more cases displayed concurrent anorexia nervosa (AN), mild enough to be consistent with low-intensity outpatient treatment, but otherwise fulfilled BN criteria. According to interviews, these patients had suffered their ED for an average of 115.92 (±86.92) months. They also binged an average 17.51 (±8.49) days monthly. Those who purged by vomiting reported doing so an average of 19.56 (±9.14) days monthly.
Bingers in remission (n =18) consisted of former bingers who denied binging more than twice a month on EBB interview. These cases were former patients or patients in final stages of treatment at the same EDU. There were originally 24 bingers in remission as diagnosed by the EDE. However, 6 of these cases were ultimately dropped from the study because they reported binge or purge episodes during a subsequent 3-week experience sampling period (in a second phase of this study, not addressed in this report). They did not seem in remission as far as bulimic symptoms were concerned. Distribution of past ED diagnoses in the remaining sample of 18 bingers in remission as follows: BN purging subtype (n = 14), BN nonpurging subtype (n = 1), ED-NOS (n = 2), and mild AN with binge-purge features (n = 1). These cases had suffered their ED for an average of 88.33 (±55.52) months.
Normal control women (n = 31) were recruited who showed no past or present ED upon EDE interview or overt psychiatric problem upon informal clinical assessment. All denied binging or purging behaviors upon interview (and throughout the 3-week experience sampling). One additional case who exhibited bulimic symptoms upon interview was dropped from the study.
Across active bingers, bingers in remission, and normal control groups, mean (±standard error), age was 27.45 ± 1.00, 27.39 ± 1.38, and 25.84 ± 1.12, respectively, mean body mass index (BMI: Beumont, Al-Alami, & Touyz, 1988) was 21.64 ± .50, 22.16 ± .63, and 21.81 ± .56, respectively, and mean Hollingshead two-factor index of social standing (Myers & Bean, 1964) was 38.52 ± 1.98, 39.64 ± 2.87, and 35.94 ± 2.87, and 35.54 ± 16.32, respectively. Analyses of variance (ANOVAs) detected no between-group differences.

Measures

ED diagnoses were established using items relevant to diagnosis from the EDE (Fairburn & Cooper, 1993). The EDE is a carefully structured interview designed to estimate the severity of ED symptoms (e.g., binge/purge behaviors, body image disturbances, pursuit of thinness). On this interview, interrater reliability exceeds .90 on all but 3 of 62 items and internal consistency of subscales is excellent. The interview has proven validity for discriminating clinical and nonclinical populations.
Depressive symptomatology was assessed using the Beck Depression Inventory (BDI; Beck & Beck , 1972) which comprises 13 items assessing signs of depression. Correlations of BDI-13 scores to those of the full-length BDI reportedly range form .89 to .97 and alpha consistently exceeds .85.
Dimensions of personality pathology were measured using selected subscales from the Dimensional Assessment for Personality Pathology (DAPP; Livesley, Jackson, & Schroeder, 1989, 1992). The DAPP is a self-report measure of personality pathology, developed using a carefully structured approach. The full DAPP includes 18 factor-based dimensions, all achieving a coefficient alpha of .90 or better. For this study, we applied only selected subscales: (1) narcissism (inflated self-importance, need for attention, and admiration); (2) stimulus-seeking (reckless sensation seeking); (3) anxiousness (longstanding trait anxiety, ruminativeness, indecisiveness, and guilt proneness); (4) compulsivity (concern with order, cleanliness, and precision); (5) restricted expression (of feelings); (6) self-harm (recurrent thoughts of self-harm and suicide). We analyzed total scores on each subscale (adjusted for missing answers in isolated cases). On all scales, higher scores imply greater endorsement of the relevant trait.
Data from and experiencing-sample record (ESR) component of this study (not addressed here) were used to verify that the bingers in remission were indeed free of symptoms over a follow-up interval (3 weeks in most cases). Our ESR was designed to enable convenient, ongoing (in a pocket-sized booklet) of perceptions of daily social interactions, concurrent self-perceptions, moods, and eating behaviors. Recording of each record was cued to daily social interactions lasting more than 10 min. Along with each record, subjects were asked to report eating behaviors and attitudes for the period between the current and previous social interactions. Subjects noted meals, snacks, eating binges, vomiting, laxative abuse, and exercising in the interest of weight loss, according to objective definitions (provided in the booklet).

Procedure

Following informed consent, participants were interviewed, and completed self-report questionnaires.

RESULTS

A direct discriminant function analysis was performed to assess loadings of the seven psychopathological characteristics in active bingers in remission, and in normal control subjects. The classification variables were depression (measured with the BDI), narcissism, self-harm, compulsivity, anxiousness, restricted-expression, and stimulus-seeking (all measured using the DAPP). Of the total 104 cases, 3 were excluded from the analysis because they had missing data on one of the discriminating variables. No cases were identified as multivariate outliers. For the 101 cases retained (53 active bingers, 17 bingers in remission, and 3 controls), assumptions of linearity, normality, multicollinearity, and homogeneity of variance-covariance matrices were met. Pooled within-group correlations among the predictors are shown in Table 1.
Two discriminant functions were calculated, with a combined x²(14) = 107.49, p <.001. After removal of the first function, there were still strong associations between groups and predictors, x²(6) = 27.53, p < .001. The two discriminant functions accounted for 70.7% and 20.3% respectively, of between-group variability. The first discriminant function maximally separated active bingers from bingers in remission and control subjects, whereas the second discriminant function discriminated active bingers and bingers in remission from control subjects.
The loadings of predictors into discriminant functions (Table 2) suggested that the best predictors for distinguishing between active bingers and the other two groups (first function) were depression (from the BDI), anxiety, self-harm, restricted-expression, stimulus-seeking, and compulsivity (all from the DAPP). Loadings less than .50 will not be interpreted. Active bingers exhibited greater levels of depression than did bingers in remission and controls, the means and

standard deviations for the three groups, respectively, being 18.06 ± 8.11, 4.76 ± 5.08, and 3.61 ± 3.93, F(2,98) = 56.75, p < .001; higher levels of anxiousness than did bingers in remission and controls, the means and standard deviations for the three groups, respectively, being 66.19 ± 12.09, 55.88 ± 14.06, and 43.56 ± 9.98, F (2,98) 37.37, p < .001; and higher levels of self-harm, 30.07 ± 13.26, 17.35 ± 8.02, and 14.64 ± 4.36, F (2,98) = 24.42, p < .001. Post-hoc analyses using Bonferroni-adjusted multiple comparisons procedures revealed significant differences between active bingers, on the one hand, and bingers in remission and controls, on the other, on depression self-harm levels. Significant differences among all three groups were obtained on anxiousness levels.

Table 1. Pooled within-group correlations among predictors

Self-Harm

Compulsivity

Depression Anxiousness Restricted- Expression Sensation-Seeking

Narcissim .10 ?.05 .23 .43 ?.07 .23
Self-Harm .05 .60 .37 .26 .27
Compulsivity .07 .18 .24 ?.19
Depression .63 .37 .20
Anxiousness .31 .08
Restricted Expression ?.09
Sensation-seeking

Table 2. Pooled within-groups correlation between discriminating variables and canonical discriminant functions.

Function 1

Function 2

Depression .92 ?.32
Anxiousness .74 .36
Self-Harm .61 ?.13
Restricted-expression .37 ?.05
Stimulus-seeking .33 ?.00
Compulsivity .23 .04
Narcissism .50 .72

Significant pooled within groups correlation between discriminating variables and
canonical discriminent function.

Only one predictor, narcissism, loaded above .50 onto the second discriminant function, separating active bingers and bingers in remission from controls. Active bingers and bingers in remission both exhibited higher levels of narcissism than did controls [means and standard deviations for the three groups, respectively, being 59.50 ± 11.62, 61.06 ± 10.16, and 43.56 ± 9.98, F (2.98) = 24.17, p < .001]. Post-hoc analyses using Bonferroni-adjusted multiple-comparison procedures revealed significant differences between the two binge eating groups and the controls. The two binge eating groups did not differ.
Finally, the classification results for the total usable sample of 101 women were computed. The number of cases correctly classified were the following: 80.31 (77.23%), 39 (73.60%), 12 (70.60%), and 27 (87.10%), respectively, for the full sample, active bingers, bingers in remission, and controls. By chance alone, these values compare to 40.06 (39.66%), 27.56, (52.00%), 2.89 (17.00%), and 9.61 (31.00%), respectively, for the full sample, the active bingers, bingers in remission, and the controls, who would have been expected to be correctly classified.
DISCUSSION

Our results suggest that psychopathological features characteristic of active bingers differ from features that are characteristic of such cases once in remission. The latter features we interpret as trait tendencies. From among a range of psychopathological dimensions, depression, anxiety, and suicidality emerged as reliable predictors of membership in our active binger group and differentiated such cases from bingers in remission and controls. An obvious interpretation is that depression , anxiety, and suicidal tendencies often constitute state characteristics , pronounced during actively bulimic phases, but not traits that persist following abstinence from bulimic symptoms. This conceptualization is compatible with observations that have shown depression, anxiety, and self-harming behaviors to be prominent in bulimic patients but to resolve rapidly along with remission of bulimic symptoms (Wonderlich, et al., 1994).
Conversely, narcissism differentiated bingers, whether active or in remission, from controls. As this dimension was elevated whether the sufferer was binging actively at the time of assessment or not, it could represent an enduring trait that may be associated with the propensity to develop bulimic symptoms–an interpretation compatible with many formulations that link binge eating causally to self-disturbances of a narcissistic type (Johnston, 1991; Strober, 1991). Such views assume binge/purge syndromes to be symptomatic of an underlying developmental pathology, which impacts on the sufferer’s capacity to maintain a stable sense of self-worth, and to self-regulate without excessive reliance on positive regard from the external social environment. Such sensitivities are presumed to underlie the bulimic’s proclivities toward preoccupation with physical perfection and need of mastery (Johnson & Wonderlich, 1991).
Our cross-sectional design is insufficient to rule out an alternative interpretation of our results–that narcissim simply reflects a state problem that is relatively refractory to change and persists individuals who binge eat even after remission of bulimic symptoms. We find this a less plausible interpretation, however, given the loadings of narcissism in our group in remission.
The present findings have several clinical implications: (1) If depression and anxiety are often state disturbance in bulimic syndromes, it may often be the case that such symptoms will resolve with the resolution of ED. Therefore, it may be indicated to treat bulimic symptoms first, on the assumption that other symptoms will resolve with remission of eating symptoms. If so, clinical judgment may need to guide decisions concerning the need to continue psychotherapy aimed at self disturbances (when these seriously impede the sufferer’s adaptation) or to leave well-enough alone (when these features are assumed to be refractory to treatment efforts or are of lesser adverse impact on the individual’s functioning and prognosis).

Sheila Jabalpurlawa was a research coordinator at EDU. The authors are grateful to Dr. Lise Gauvin for her advice on statistical analyses.

REFERENCES

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC,: Author

Beck,A.T., & Beck,.W.(1972). Screening depressed patients practice: A rapid technique. Postgraduate Medicine, 81-85

Beaumont, P., Al-Alami,M. & Tocyz.S. (1988). Relevance of a standard measurement of undernutrition to the diagnosis of anorexia nervosa. Use of Quetelet’s body mass index. International Journal of Eating Disorders, 7, 399-405

Fairburn, C.G., & Cooper, Z. (1993), The Eating disorder Examination. In C. Fairburn & G.T. Wilson (Ed.s) Binge eating: Nature, assessment and treatment (pp. 317-332), New York: Guilford Press

Johnson,C. (1991). Treatment of eating-disordered patients with borderline and false self/narcissistic disorders. In C. Johnson (Ed.) , Psycho dynamic treatment of anorexia nervosa and bulimia. (Pp. 165-193), New York: Guilford Press

Johnson, C. & Wonderlich,S. (1992). Personality characteristics as a risk factor in the development of eating disorders. In J.H. Crowther, D.L. Tennenbaum, S.E. Hobfell, & M.A.P. Stephens (Eds.). The etiology of bulimic nervosa: The individual and familial context. Washington, D.C.: Hemisphere Publishing.

Livesly, W.J. Jackson, D.N. & Schroeder, M.L. (1989). A study of the factorial structure of personality pathology. Journal of Personality disorders, 3. 292-306.

Livesly, W.J., Jackson, D.N. & Schroeder, M.L. (1992) Factorial structure of traits delineating personality disorders in clinical and general population saples.
Journal of Abnormal Psychology, 101, 432-440.

Myers,J.K., & Bean, L.L. (1964). A decade later: A follow-up of social class and mental illness. New York: John Wiley & Sons.

Steiger, H., Jabalpurwala, S. Champagne, J. & Scotland, S. (1997) A controlled study of trait narcissism in anorexia and bulimic nervosa. International Journal
of Eating disorder, 22, 173-276.

Steiger, H., Leung, F. , Thibaudeau, J. Houle, & Ghadirian, A.M. (1993). Comorbid features in bulimics before.