Changes in Patient Measures as Predictors of Therapists’ Ratings of Treatment Collaboration and Change in Eating Disorder Subgroups

Abstract

This study examined how implicit and explicit changes following integrative inpatient treatment of adolescents with eating disorder (ED) may predict the posttreatment ratings of psychodynamic therapists of their patients’ openness to therapeutic processes and their change (Therapist Evaluation Inventory). The relative contribution of inpatients’ ego functions was compared with that of their mental distress and ED symptoms in two subgroups: restricting type anorexia (AN-R) and binging/purging type EDs (B/P). Data indicated that the implicit personality variable of elevated ability to modulate affects was the best predictor of therapist-rated global outcome among patients with B/P symptoms, whereas in patients with AN-R, evolving openness to implicit negative affects and a reduction in reported distress were best predictors. In patients with AN-R, attenuated affect control was also significantly correlated with therapist posttreatment ratings. These data point that in addition to addressing behavioral/symptomatic aspects, personality variables should be addressed in the psychological treatment of EDs.

Keywords

therapists’ ratings of change therapeutic change eating disorders implicit and explicit measures

Recent research has increasingly focused on the assessment of change processes and outcomes of psychotherapeutic interventions from both patient and therapist perspectives. This shift toward multiple perspectives is important for two reasons: the first is providing insight into what constitutes change from different perspectives and the second is that patient–therapist discrepancies may influence patients’ motivation, attendance, and alliance—all considered significant, relevant parameters for treatment outcomes (Barber, Connolly, Crits-Cristoph, Gladis, & Siqueland, 2000). Several studies have shown that comparisons of patient and therapist viewpoints on psychotherapy outcomes yield only small to moderate levels of agreement (Farrell, 1999; Pekarik & Wolff, 1996). In psychodynamically oriented therapies, a source of discrepancies between patients and therapists may relate to the optimal focus of psychotherapy: Should it focus on symptomatic relief and conscious personality changes, or should it be geared toward changes in the patient’s inner world and deep-rooted personality organization? According to psychodynamic formulations, change in the inner world is the most important treatment aim and the underlying basis for symptomatic and behavioral improvement (Lush, Boston, & Grainger, 1991). Thus, in analyzing how patients’ perspectives or patients’ measures relate to psychodynamic therapists’ perspectives, assessment tools ought to detect changes in patients’ inner world as well as symptomatic and behavioral changes.

Assessment of patients’ and therapists’ estimates of collaboration and outcome is of particular relevance in the treatment of eating disorders (EDs), which elicits concern among treatment providers. The challenges inherent to treating ED relate to the complex pathogenesis in patients with ED (Herzog et al., 2000). In addition, young patients with ED, particularly those diagnosed with anorexia (AN), are often considered highly difficult to treat because of their inclination to resist change in their symptoms (Couturier & Lock, 2006). Thus, understanding patients’ and therapists’ estimates of change may promote better alliance and enhance treatment efficiency.

Assessment of Therapists’ and Patients’ Perspectives on Therapeutic Change in EDs

Empirical data suggest that underlying personality constructs and psychiatric comorbidity patterns may be of considerable importance both in the predisposition to an ED (Fairburn & Harrison, 2003) and in influencing the process and outcome of the disorder and the response to treatment (Thompson-Brenner & Westen, 2005a; Vitousek & Manke, 1994). Along these lines, several studies have examined the influence of personality aspects versus symptomatic parameters on treatment outcomes in EDs. For example, Thompson-Brenner and Westen (2005a, 2005b) conducted a naturalistic project assessing the therapist’s perspective of factors influencing the outcome of different types of treatment, mainly in patients in the binging/purging (B/P) spectrum. They found that the use of psychodynamic interventions focusing on personality dynamics (e.g., emotional and interpersonal functioning) was associated with changes in these target domains and was positively associated with global outcomes, whereas cognitive behavioral treatment (CBT) interventions did not cause such changes. Nevertheless, CBT interventions targeting ED symptoms addressed them more effectively than psychodynamic interventions. In a study examining the perspectives of patients with ED and their therapists’ perspectives (mostly CBT experts), it was found that patients differed from therapists with respect to the definition of a good therapeutic outcome (De la Rie, Noordenbos, Donker, & Furth, 2008). Whereas the therapists emphasized the importance of a reduction in ED symptoms, the patients underscored the importance of the relationship with the therapist and of addressing underlying personality issues (such as self-esteem).

The studies presented were based on therapists’ and patients’ reports on treatment process and outcome. In the present study, we aimed to examine the relations between the changes in patients’ measures of personality and symptomatic variables and therapists’ ratings of patient outcome, to uncover what predicts the therapist’s perspective on outcome. Based on the extensive literature conceptualizing EDs in terms of deficient ego functions for coping with stress (e.g., Bruch, 1973), we examined the personality domain of ego functions.

Personality Dynamics: Ego Functions in ED Patients

Ego functions is a construct that refers to psychological functioning and adaptation. Several formulations of ego functions have been proposed, such as the formulations of Beres (1956). Beres presents a global formulation of ego functions comprising several specific arenas, such as relation to reality, ability to regulate impulses, level of object relations, organized versus deviant thinking, and maturity of defensive operations. Higher levels of ego deviations in these areas are related to greater pathology.

Bruch (1973) led to the conceptualization of EDs in terms of severe ego pathology, manifested, for example, by faulty reality testing, severe disturbances in body image, inaccuracy in identifying bodily and emotional states, and the presence of deficient regulatory mechanisms.

Several aspects of ego functions have been found to differ among patients diagnosed with restricting type anorexia nervosa (AN-R), versus patients diagnosed with bulimia nervosa (BN), and with anorexia nervosa of the binging/purging type (AN-B/P). Accordingly, patients with AN-R are described more often as constricted and controlling their emotions than patients with AN-B/P and BN (Fassino et al., 2001; Vitousek & Manke, 1994), whereas patients with BN and AN-B/P are more likely to manifest deficient impulse and affect regulation (Vitosek & Manke, 1994). In a previous study of our group, significant between-group differences were revealed (Rothschild, Lacoua, & Stein, 2009), showing that inpatients with AN-R presented with higher levels of ego functions in comparison with patients with AN-B/P and BN, specifically on measures of impulse control and thinking disturbances.

Aims and Hypotheses of the Present Study

In a previous study, we found that subsequent to integrative multimodal treatment, stabilization of ED symptoms was associated with a significant improvement in the implicit Rorschach measure of ego functions adequacy and reported distress (Rothschild et al., 2009). The data further showed a different change trend in ego functions between patients with AN-R and patients with B/P symptoms, indicating that significant between-group differences were found at admission but not at discharge. In the present study, we explored whether the psychotherapists’ post discharge ratings of inpatient collaboration and treatment outcome would be differently influenced by the extent to which the patients changed over the course of treatment in symptoms compared with the personality construct of ego functions. We defined the symptomatic domain as a combination of mental distress and core ED-related symptoms, in line with data pinpointing high comorbidity of EDs and emotional distress (e.g., Stice, Burton, & Shaw, 2004). We hypothesized that in both ED subtypes, changes in patient’s ego functions would better predict therapists’ posttreatment outcome ratings than changes in ED symptoms and mental distress (as the latter are conceived to represent transient changes). We further hypothesized that different change patterns in deficient domains conceived as characteristic of each group would predict therapists’ outcome ratings in the two ED subgroups. Specifically, elevation in affect regulation capacity was expected to predict higher therapist ratings in patients with B/P symptoms (AN-B/P and BN), whereas attenuation in affect constrictedness was expected to predict higher therapist ratings in patients with AN-R.

Most of the research on personality characteristics and psychological dynamics in patients with ED is based on self-report measures, notwithstanding extensive data suggesting that patients with ED may misperceive their state because of denial or deficient awareness (Vitousek & Manke, 1994). Therefore, in assessing therapeutic change in patients with ED, the use of assessment tools aiming to detect both the patients’ inner world and symptomatic changes becomes crucial. In this study, we used the Rorschach as an implicit measure to investigate underlying personality processes and standardized self-report scales to examine explicit processes (for a review, see Bornstein, 2002; Ganellen, 2007; McClelland, Koestner, & Weinberger, 1989).

The term implicit–explicit used in cognitive literature and in social cognition domain captures parallel distinctions such as unaware–aware, unconscious–conscious, and automatic–controlled (Greenwald & Banaji, 1995). In the cognitive psychology domain, the implicit cognitions are viewed as traces of past experience that affect performance, although the earlier experience is not recalled and is unavailable to self-report (Greenwald & Banaji, 1995; Jacoby, Lindsay, & Toth, 1992). Based on this distinction, it was suggested that investigation of implicit cognitive processes require indirect measures, which neither inform the subject of what is being assessed nor request self-report regarding it (Greenwald & Banaji, 1995).

McClelland et al. (1989) implemented the implicit–explicit distinction to personality assessment, suggesting that most self-report measures assess explicit or self-attributed motives that a person acknowledges. In contrast, projective tests assess implicit motives that influence an individual’s behavior automatically, often without any awareness on the individual’s part. Ganellen (2007) went on suggesting that the Rorschach Comprehensive System (CS; Exner, 2003) as a relatively unstructured performance-based or implicit measure of personality elicits information about personality characteristics that the person does not acknowledge. For example, high values in the Schizophrenia Index (SCZI) examining thought organization and reality testing have been found to be related to overt psychotic behavior, whereas patients with psychoses often do not recognize and report these symptoms (Meyer & Archer, 2001).

Several research studies on interpersonal dependency further provide examples of the distinction between the Rorschach as an implicit method and the explicit self-report method for predicting different aspects of an individual’s behavior within this domain (Bornstein, 2002; Ganellen, 2007). For example, Bornstein (1998) examined help-seeking behavior during an experimental session and found that implicit dependency was more predictive of spontaneous help seeking when participants were not informed about the nature of the study, whereas self-reported dependency was more related to controlled help seeking, when participants were informed about the purpose of the study.

Following this line of thinking and research data, implementing both implicit (Rorschach) and explicit (self-report scales) measures is aimed not only to overcome the deficiency in self-perception characteristics found in patients suffering from ED (Vitousek & Manke, 1994) but also to assess the influence of different facets of personality dynamics as predictors of therapists’ outcome ratings.

To our knowledge, there are no studies that examined the predictive relationship between patients’ measures and therapists’ ratings. Moreover, our study is the first to examine the potential of implicit and explicit measures of treatment outcome to predict the therapists’ ratings.

Method

Participants

The study was conducted in an adolescent ED inpatient department located in a general hospital in central Israel. Data were derived from a longitudinal study described in previous studies (Rothschild et al., 2009; Rothschild, Lacoua, Eshel, & Stein, 2008), following female adolescent inpatients from the acute phase of their illness at admission to the phase of achieving stabilization in ED symptoms at discharge. The current participants were 62 female adolescent inpatients aged 13 to 18 years who (a) met Diagnostic and Statistical Manual of Mental Disorders, 4th edition (American Psychiatric Association, 1994) criteria for full-blown AN or BN at admission; (b) were never diagnosed with a bipolar disorder, schizophrenic spectrum disorder, substance use, organic brain disorder, or any medical disorder potentially affecting food consumption and weight (e.g., diabetes mellitus or thyroid disorders); and (c) continued with inpatient treatment until achieving and maintaining their required weight (for all patients with AN) and demonstrating no restricting, binging, and purging behaviors (for all patients) for at least a 2-week period.

The original cohort included 98 patients. Thirty-six of them were excluded from this study because of the following: dropout from the inpatient department and refusal to participate in a second evaluation at that time (n = 19), not meeting with the aforementioned diagnostic criteria for inclusion in the study (n = 4), refusal to participate in the assessments (n = 4), fewer than 14 responses in their Rorschach protocol (n = 5), or a change in their primary therapists during hospitalization (n = 4).

Of the 62 inpatients included in the study, 33 were diagnosed with AN-R and 29 with B/P symptoms: either AN-B/P (n = 10) or BN of the purging subtype (n = 19). Patients with AN-B/P were included along with patients with BN, following studies suggesting that AN-B/P is more akin in its core ED and personality-related attributes to BN than to AN-R (e.g., Wonderlich, Crosby, Mitchell, & Engel, 2007).

Comparing the demographic and clinical characteristics between the two study groups revealed no significant differences in age at admission (AN-R: M = 15.54 ± 1.67, B/P: M = 16.13 ± 1.18), duration of inpatient treatment (AN-R: M = 6.54 ± 2.27 months, B/P: M = 5.96 ± 2.30 months), history of loss and divorce (AN-R: 27.27%, B/P: 27.59%), and comorbidity with obsessive compulsive disorder (AN-R: 39.39%, B/P: 20.69%). Nonetheless, patients with B/P symptoms exhibited significantly greater rates of past sexual abuse (3.3 ± 4.5 years before ED onset; AN-R: 6.06%, B/P = 27.59%, χ² = 4.76, p < .05) and comorbid depressive disorders compared with patients with AN-R (AN-R: 36.36%, B/P: 79.31%, χ² = 11.58, p < .001).

Measures

Inpatient Baseline and Outcome Variables

Inpatient variables were assessed at admission and after integrative treatment (at discharge). The 10 measures included the following: an objective body mass index (BMI), 6 explicit measures (self-reports) and 3 implicit measures derived from the Rorschach.

Bodily and Behavioral Symptoms

Body mass index

The BMI was used to measure the normalcy of body weight by dividing weight by height squared (Bray, 1992).

Self-reported concerns and behaviors related to ED

The short version of the Eating Attitude Test (EAT-26; Garner, Olmsted, Bohr, & Garfinkel, 1982) measured self-reported pathological eating- and weight-related preoccupations and behaviors (Cronbach α = .93 in this study).

Mental Distress

Rorschach measure of mental distress: Experience stimulation (es)

The CS marker of mental distress (Exner, 2003) is a summary index measuring current stress levels elicited by emotional felt distress (SumShd) and intrusive ideation (FM + m).

Self-reported measures of mental distress

We assessed self-reported mental distress in terms of depression and anxiety:

Self-reported depressive symptoms

The 21-item Beck Depression Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) measured self-reported depressive symptoms (α = .92 in this study).

Self-reported state anxiety

The 20-item State Anxiety subscale of the State-Trait Anxiety Inventory (Spielberger, Gorsuch, & Lushene, 1970) measured the participants’ self-reported anxiety at the time of evaluation (α = .94).

A self-report distress variable (SR-distress) was constructed, comprising the standardized scores of the Beck Depression Inventory and the State-Trait Anxiety Inventory –State Anxiety subscale. The correlation between the scales was .78.

Ego Functions

Rorschach measure of ego function adequacy: Ego impairment index (EII-2)

The EII-2, as reformulated by Viglione, Perry, and Meyer (2003), comprised the weighted sum of five components controlled for the number of responses in each protocol, drawn from Beres’s (1956) ego assessment model: (a) Sum FQ-, indicating poor reality testing; (b) WSum6, indicating thought disturbance; (c) M-, indicating distortion in interpersonal perceptions; (d) “critical contents” of An, Bl, Ex, Fi, Fd, Sx, Xy, AG, and MOR, indicating poor ability to inhibit needs and urges; and (e) PHR and GHR, indicating poor and good human representations, respectively. The EII-2 was found to differentiate between clinical groups presenting different levels of psychological impairment. High EII-2 values were associated, for example, with schizophrenic spectrum disorders and psychoses (Viglione et al., 2003).

Rorschach measure of affect modulation: Color-Form Level equation

Because affect modulation is a central ego function deficit in patients with ED but not included in the EII-2, we used the Color-Form Level equation (Mihura, Nathan-Montano & Alperin, 2003) to assess the balance between well-modulated (FC) and relatively unmodulated (CF + C + Cn) emotional experiences and expressions. The Color-Form Level formula uses the weights for the CS chromatic color scores as follows: (FC × 0.5) − (CF × 1.0) − (C × 1.5) − (Cn × 1.5). Lower Color-Form Level scores were significantly correlated with elevated self-reported measures indicating higher emotional impulsivity (Mihura et al., 2003).

Self-reported measures of ego functions

Three personality-related subscales of the Eating Disorders Inventory-2 (EDI-2; Garner, 1991) have been used to assess ego functions. The ability to inhibit needs and urges was assessed by the 11-item EDI-2 Impulse Regulation (IR) subscale (α =.82). The ability to identify and convey affects and emotional states was assessed by the 10-item Interoceptive Awareness (IA) subscale (α =.83). The capacity to form adequate, secure, reciprocal social relationships was assessed by the seven-item Social Insecurity (SI) subscale (α =.83).

Interrater Reliability for the Rorschach

All Rorschach protocols were scored by the first author, who was blind to the patients’ specific ED diagnosis and who is highly experienced in the Rorschach administration and scoring according to CS. To evaluate interrater agreement, 34 randomly selected protocols were scored independently by a senior intern who had 4 years extensive training in the CS and who was also blind to the patients’ subgroup and to the first author’s scoring. Intraclass correlation coefficients (ICCs) for all Rorschach scores used in this study were computed at the protocol level for absolute agreement. The ICC indicated excellent interrater reliability (Cicchetti, 1994): .82 for FC, .85 for CF, .88 for C, .91 for EII-2, and .95 for es.

Therapists’ Perspectives on Inpatients’ Process and Outcomes

We used a therapist’s rating scale modified for patients with ED, from the 15-item Therapist Evaluation Inventory (TEI; Kazdin, Esveldt-Dawson, French, & Unis, 1987). The TEI assessed therapist ratings of individual psychotherapy processes and outcomes for children with psychiatric disturbance, on a 5-point Likert-type response scale from Not at all (1) to Very much (5).¹ We chose this scale due to its relevance in assessing symptomatic change as well as the concept of ego functions (e.g., changes in patients’ ability to cope with frustration, and awareness to the inner world).

The TEI comprises two a priori subscales: (a) Receptiveness and Openness to Treatment (six items), assessing collaboration with the therapeutic process; and (b) Progress and Future Projection (nine items, hereafter the Change subscale), assessing symptomatic changes and the ability to generalize treatment outcomes in the future. In previous studies, a summary global outcome score yielded high internal consistency (α = .97; Thurber & Snow, 1991). Internal reliability in the present study was .95 for the global TEI. A maximal likelihood factor analysis confirmed the a priori structure.²

Procedure

Recruitment and testing procedures

The study was approved by the Review Board of the Sheba Medical Center. All participants, and their parents or other legal guardians in the case of minors aged 18 years and younger, signed a written informed consent to participate in the study. Two experienced child and adolescent psychiatrists independently conducted standardized intake interviews of the inpatients within 14 days of admission, using the Eating Disorders Family History Interview (Strober, 1987) for the ED diagnosis and the Structured Clinical Interview for DSM-IV Axis I Disorders–Patient edition (Version 2.0) for comorbid psychiatric diagnoses (First, Spitzer, Gibbon, & Williams, 1995).

The self-report questionnaires and the Rorschach test were administered individually twice in the inpatient department: first within 14 days of admission and second prior to discharge, upon achieving stabilization of the ED symptoms and the patients’ physical condition as determined by relevant physical examinations and laboratory testing. The Rorschach tests were administered by trained licensed clinical psychologists and senior interns in clinical psychology, who were blind to patients’ subgroup affiliation and self-report outcomes. Administration and scoring of the Rorschach protocols followed Exner’s (2001) CS guidelines. Following discharge, the patient’s individual psychodynamically oriented psychotherapist (see below) completed the TEI questionnaire. The therapists neither had any access to the pretest nor to the posttest assessment material of the patients.

All 62 participants completed the Rorschach testing at both phases, but only 48 agreed to complete all the self-report measures at both phases (B/P: n = 24, AN-R: n = 24). The therapists rated the participants who completed the self-report measures as more open to treatment (t = 2.19, p < .05) and as presenting a better global outcome (t = 2.12, p < .05) than those who did not complete the self-report measures.

Integrative, multimodal inpatient treatment procedure

The integrative treatment protocol in this department corresponds with other structured inpatient programs for adolescents with EDs (American Psychiatric Association, 2006; Anzai, Lindsay-Dudley, & Bidwell, 2003). The protocol includes the following: a behaviorally oriented nutritional rehabilitation program, twice-weekly sessions of individual psychodynamically oriented psychotherapy, one weekly family therapy session, twice-weekly sessions of dynamic group therapy, twice-weekly CBT group sessions, and one weekly group expressive movement therapy. The inclusion of psychodynamic psychotherapy in the treatment regimen is designed to address interpersonal aspects and contact with inner personality dynamics referring to affects, needs, and conflicts. The psychotherapy in this department is conducted by senior interns in clinical psychology, licensed clinical psychologists, and licensed child and adolescent psychiatrists, all with clinical experience of 5 to 25 years.

Results

Before conducting the statistical analysis, the number of Rorschach responses (R) in each protocol was controlled for its linear effect on Color-Form Level and es as a covariance to control for between-subjects differences in productivity. In all analyses, the variables were also controlled for age, sexual abuse, and comorbidity with depression, which were found to differentiate between the study groups. All statistical analyses used the SAS V9.1.2 (SAS, 2004) software for Windows XP. We further compared between the two B/P subgroups (patients with AN-B/P and BN) to examine whether they present with similar configurations, using the t test. The data showed that at baseline the two groups were similar in all the study variables (EAT, EDI-2 [IR, SI], SR-distress, es, EII-2, and Color-Form Level), with the exception of BMI (t = 3.91, p < .001) and EDI-2 IR (t = 2.45, p < .05). In the acute state, patients with AN-B/P presented lower BMI and reported better IA. The data further showed that the two groups were similar in the change scores (discharge minus acute state scores) in all variables, with the exception of BMI (t = 4.37, p < .001). These results allowed us to combine the two groups in our further analyses.

Changes in Patients’ Measures

Before examining the study hypothesis, we examined the extent to which the patients’ variables changed from admission to discharge. Two multivariate analyses of variance (MANOVAs) were conducted: (a) for the Rorschach measures of ego functions (EII-2 and Color-Form Level variable) and (b) for the self-reported ego functions (EDI-2 IA, IR, and SI). Four univariate ANOVAs were conducted: (a) for ED symptoms (BMI and EAT-26) and (b) for mental distress variables (es, SR-distress). Diagnosis and phase were the independent variables, with diagnosis as the between-subject variable and phase as the within-subject variable (see Table 1).

Table 1.

Two-Way Analyses of Variance of Study Variables for Patients With Restricting Anorexia Nervosa (n = 33) and Binging/Purging Spectrum (n = 29) at the Acute Phase and the Stabilization of ED Phase.

	AN-R^a		B/P Spectrum^b
Variable	Acute Phase, M (SD)	Stabilization of ED Phase, M (SD)	Acute Phase, M (SD)	Stabilization of ED Phase, M (SD)	Diagnosis, F (ES d)^c	Phase, F (ES d)^c	Diagnosis × Phase, F (ES d)^c
ED symptoms
Body mass index	14.94 (1.93)	19.24 (1.45)	18.80 (3.99)	20.15 (2.29)	16.94** (0.96)	110.96*** (1.14)	30.19*** (0.59)
Eating Attitudes Test^d	41.80 (18.56)	32.17 (22.20)	46.67 (15.00)	28.83 (14.74)	0.03 (0.04)	28.75*** (0.78)	2.57 (0.23)
Mental distress
SR-distress^e (state anxiety + depression)	0.02 (0.83)	−0.19 (0.94)	0.67 (0.76)	−0.21 (0.74)	2.43 (0.40)	21.19*** (0.69)	7.89** (0.42)
es (Experience stimulation; stress levels)	6.83 (3.14)	6.99 (4.39)	7.76 (4.23)	7.87 (4.31)	1.09 (0.23)	0.06 (0.04)	.00 (0.00)
Ego functions
EII-2 (Ego Impairment Index)	0.74 (1.30)	0.62 (1.21)	1.77 (1.80)	1.02 (1.25)	5.51* (0.52)	5.38* (0.31)	2.93 (0.23)
Color-Form Level (affect regulation)	−1.74 (1.69)	−1.33 (1.90)	−2.56 (2.23)	−1.98 (2.18)	3.36 (0.37)	2.42 (0.25)	.08 (0.04)
EDI-2 IA^d (Interoceptive Awareness)	7.30 (4.62)	6.09 (3.46)	15.35 (8.04)	9.06 (5.44)	15.90*** (0.99)	18.77*** (0.68)	8.59*** (0.46)
EDI-2 IR^d (Impulse Regulation)	5.50 (5.04)	5.91 (5.97)	9.95 (5.90)	6.50 (5.11)	3.38 (0.46)	3.48 (0.28)	5.57* (0.36)
EDI-2 SI^d (Social Insecurity)	6.09 (4.09)	7.88 (5.02)	8.94 (5.22)	5.21 (5.38)	0.01 (0.02)	1.54 (0.20)	12.49*** (0.57)

Note. ED = eating disorder; SR-distress = self-report distress. The means presented in the table are estimate means after controlling for the covariates.

AN-R = Anorexia nervosa, restricting type.

Binging/purging spectrum = anorexia nervosa − binging/purging type + bulimia nervosa.

Cohen’s d scores were computed using the pulled residual error term, including between- and within-subject variance.

AN-R: n = 24, B/P: n = 24, df = 1, 46.

SR-distress = standardized scores of the two self-report scales: the Beck Depression Inventory and State-Trait Anxiety Inventory–State Anxiety subscale.

p < .05. **p < .01. ***p < .001.

Changes in ED symptoms and mental distress

BMI approached the normal range at discharge. The patients with AN-R showed greater weight gain than the patients with B/P symptoms (including patients with AN-B/P and normal weight patients with BN), as indicated by the significant interaction effect. Both diagnostic subgroups reported a significant decline in disordered eating-related preoccupations and behaviors (EAT-26) at discharge.

A significant phase effect and a significant interaction effect were revealed for the self-reported distress variable (SR-distress), showing that an attenuation of explicit distress emerged in both diagnostic subgroups, with a greater improvement among patients with B/P symptoms than among patients with AN-R. No significant main effects emerged for the univariate ANOVA for the Rorschach es marker.

Changes in ego functions

Two MANOVAs, examining the patients’ implicit and explicit ego functions, revealed a significant improvement both for the Rorschach variables (EII-2 and Color-Form Level), F(2, 59) = 3.82, p = .03, η² = .11, and for the self-report measures, F(3, 44) = 6.19, p = .001, η² = .30. This was shown specifically for the general Rorschach measure EII-2 and the EDI-2 IA subscale. A significant diagnosis effect also emerged for both the Rorschach variables, F(2, 59) = 4.01, p = .02, η² = .12, and for the self-report measures, F(3, 44) = 5.43, p = .003, η² = .27, indicating that the inpatients with AN-R presented with higher levels of implicit and explicit ego functions in comparison with inpatients with B/P symptom. Interaction effects were revealed for all EDI-2 ego deficit subscales (IA, IR, SI), F(3, 44) = 4.73, p = .006, η² = .24, but not for the Rorschach variables. These data showed that following treatment and attainment of stabilization in ED symptoms, inpatients with AN-R reported greater IR and SI, whereas patients with B/P symptoms reported an improvement in all reported ego functions.

Predictors of the Therapist’s Perspectives on Inpatient Outcome

We used t tests to examine the differences between the diagnostic subgroups in therapists’ TEI ratings of the patients’ overall outcome and the two specific TEI subscales. Therapists described the inpatients with B/P symptoms as showing a significantly better overall outcome score compared with the inpatients with AN-R (AN-R: 42.35 ± 12.69, B/P: 49.04 ± 10.25, t = 2.26, p < .05), with a medium effect size (Cohen’s, 1977, d = .57). Specifically, the therapists described the inpatients with B/P symptoms as more open and cooperative in the therapeutic process than inpatients with AN-R (AN-R: 18.09 ± 5.91, B/P: 22.38 ± 4.67, t = 3.13, p < .05), with a large effect size (d = .79). No between-group difference was detected in the Change subscale (AN-R: 21.69 ± 7.26, B/P: 24.13 ± 6.63, t = 1.34, ns).

We hypothesized that changes in patients’ personality variables (ego functions) would predict the therapists’ ratings to a greater extent than the changes in ED symptomatology and mental distress. We first computed Pearson correlations between the residual gain scores of the study variables and the TEI subscales (see Table 2) separately for each subgroup. The relative contribution of residual gain scores in symptom measures (ED symptomatology and distress measures) and personality measures (ego functions) to the global TEI score was then tested by hierarchical regression analyses conducted separately for each subgroup. Symptom measures entered in the first and second steps and personality measures in the third step. BMI as a symptom measure entered only for the patients with AN-R (see Table 3).

Table 2.

Pearson Correlation Coefficients Between the TEI and Inpatients’ Implicit and Explicit Residual Gain Scores for the Two Eating Disorder Subtypes.

Group	Variable	TEI Global	TEI Openness	TEI Change
Binging/purging spectrum (n = 29)^a	Eating Attitudes Test^b	−.09	−.07	−.09
	SR-distress^c	−.04	.02	−.09
	es (stress levels)	.17	.21	.11
	EII-2: Ego Impairment Index	.24	.20	.23
	Color-Form Level	.42*	.26	.47*
	EDI-2 IA: Interoceptive Awareness^b	.05	.13	−.01
	EDI-2 IR: Impulse Regulation^b	.13	.11	.12
	EDI-2 SI: Social Insecurity^b	.03	.06	−.004
Restricting anorexia nervosa (n = 33)	Body mass index	−.001	−.03	.03
	Eating Attitudes Test^b	−.30	−.29	−.29
	SR-distress	−.37	−.38*	−.35
	es (stress levels)	.50**	.56**	.42*
	EII-2: Ego Impairment Index	.32	.32	.30
	Color-Form Level	−.40*	−.40*	−.36*
	EDI-2 IA: Interoceptive Awareness^b	.07	−.05	.16
	EDI-2 IR: Impulse Regulation^b	−.06	−.16	.02
	EDI-2 SI: Social Insecurity^b	−.15	−.14	−.14

Note. TEI = Therapist Evaluation Inventory; SR-distress = self-report distress; EDI-2 = Eating Disorders Inventory-2.

Binging/purging spectrum = anorexia nervosa − binging/purging type + bulimia nervosa.

AN-R: n = 24, B/P: n = 24, df = 1, 46.

SR-distress = standardized scores of the two self-report scales: the Beck Depression Inventory and State-Trait Anxiety Inventory–State Anxiety subscale, AN-R: n = 24, B/P: n = 24.

p < .05. **p < .01.

Table 3.

Estimates of Predictors for Global TEI, Ordered by Blocks.

Group	Predictors^a	Global TEI
Group	Predictors^a	B	SE	β	ΔR² ^b
Binging/purging spectrum (n = 24)	Step 1—Symptoms				.01
	Eating Attitudes Test	−6.49	3.39	−.56
	Step 2—Distress				.04
	SR-distress	−3.12	3.51	−.29
	es	6.16	3.12	.59
	Step 3—Personality measures				.35
	EII-2	−0.75	2.86	−.06
	Color-Form Level	6.92	2.73	.58*
	EDI-2 IA	−2.60	3.82	−.21
	EDI-2 IR	3.39	3.95	.30
	EDI-2 SI	6.80	4.23	.61
Restricting anorexia nervosa(n = 24)	Step 1—Symptoms				.12
	Eating Attitudes Test	−0.47	3.47	−.03
	BMI	2.05	2.46	.15
	Step 2—Distress				.47
	SR-distress	−13.89	4.01	−.92**
	es	6.95	2.54	.56*
	Step 3—Personality measures				.15
	EII-2	2.05	1.80	.18
	Color-Form Level	−2.04	1.87	−.20
	EDI-2 IA	1.18	2.10	.10
	EDI-2 IR	0.24	3.57	.02
	EDI-2 SI	6.52	3.37	.48

Note. TEI = Therapist Evaluation Inventory; SR-distress = self-report distress; es = experience stimulation; IA = Interoceptive Awareness; IR = Impulse Regulation; SI = Social Insecurity; BMI = body mass index; EII-2 = Ego Impairment Index; EDI-2 = Eating Disorders Inventory-2.

Predictors were calculated as residual gain scores.

Partial R² is calculated for each step’s contribution to its preceding steps.

p < .05. **p < .01.

As seen in Table 2, for the patients with B/P symptoms, elevated ability to modulate affect (Color-Form Level variable) was associated with therapists’ higher global outcome ratings, because of higher ratings in the Change subscale. As for the patients with AN-R, an opposite finding was attained for the affect modulation variable. Higher ratings in all TEI scales were related to a reduction in affect control. In addition, whereas greater openness to treatment was correlated with a reduction in self-reported distress (SR-distress), higher scores on all TEI scales were associated with an elevation in the implicit measure of distress (es).

The regression analysis for the patients with B/P symptoms explained 40% of the variance in global TEI. This regression analysis showed that the symptom measures explained only 5% of the variance in global TEI, whereas the personality block explained 35% of the variance in the global TEI. Specifically, the only significant variable predicting therapist-rated global outcome was the personality implicit measure of affect modulation as a positive predictor. Greater capacity to modulate affects predicted higher therapists’ global outcome ratings.

In contrast to our hypothesis, a different configuration was found among the patients with AN-R, showing that the symptom measures were found as significantly predicting the global TEI score, whereas the personality variables did not. The first step with ED symptoms explained 12% of the variance in the global TEI. The mental distress variables entering in the second step (SR-distress, es) were found as significant predictors, explaining 47% of the variance in global TEI. The Rorschach measure of distress (es) entered as a positive predictor, whereas self-reported distress entered as a negative predictor, showing that therapists’ higher outcome ratings were related to an elevation in implicit distress alongside a reduction in the reported distress experience. The personality measures entering in the third step raised the explained variance in TEI to 74%.

Discussion

This study examined the hypothesis that improvement in the personality variable of ego functions would better predict therapists’ ratings of openness and collaboration in treatment as well as therapeutic change than would changes in ED and distress symptoms. This examination was done following the application of an integrative treatment approach in inpatients diagnosed with AN-R and B/P symptoms.

The regression analysis examining the study hypothesis was partially confirmed. It showed that the best predictor of higher outcome ratings by the therapists for inpatients with B/P symptoms was, as expected, their elevated ability to regulate their affects, as indicated by the implicit Rorschach measure (Color-Form Level variable). As for the patients with AN-R, the Pearson correlations indicated significant positive relations between a reduction in affect control (Color-Form Level variable) as well as an elevation in implicit mental distress (es) and elevated global TEI ratings. However, in the hierarchical regression, changes in symptomatic measures indicating elevation in implicit mental distress (es) and reduction in reported distress (SR-distress), but not changes in personality variables, were found to predict the therapists’ global ratings in patients diagnosed with AN-R. These findings suggest, nevertheless, that the therapists identified in both ED subtypes enduring characteristics requiring modification, namely, constricted affective life in patients with AN-R and affective dysregulation in patients with B/P symptoms, alongside changes in distress symptoms, to enable a greater collaboration with treatment and a more favorable outcome.

A question may be raised as to the generalization of our findings in adolescents to adult patients with ED. Nevertheless, patients with ED of all ages often have considerable difficulties in identifying and expressing their emotions, and affective dysregulation is considered an enduring characteristic of paramount importance in predicting the outcome of patients with B/P symptoms (Kaye, 2008; Vitousek & Manke, 1994). Similar conclusions can be reached with respect to adult patients diagnosed with AN-R, which are characterized by elevated inhibition, rigidity, and harm avoidance (e.g., avoidance of change; Kaye, 2008; Vitousek & Manke, 1994). These characteristics can be considered akin the constricted emotionality found in the present study as correlated with therapists’ ratings of change.

The different association found with respect to personality configuration in the two ED subgroups and the TEI suggest that the therapists regard the two ED subgroups as presenting different personality dynamics, thus calling for different therapeutic approaches. With respect to the patients with B/P symptoms, our results may support the notion that specific treatment interventions geared toward reducing affective dysregulation should be included in their treatment (Thompson-Brenner & Westen, 2005b). Following this suggestion, a few studies have adapted the dialectic behavioral therapy (Linehan, 1993), used in borderline personality patients to EDs, in patients with B/P symptoms (e.g., Kröger et al., 2010). This model focusing on mindfulness, emotion regulation, and distress tolerance skills was found to be associated with a reduction in B/P behaviors an improvement in general psychopathology. In the case of AN-R, unfortunately, there are as yet no sufficient data with respect to the effectiveness of strategies geared toward greater openness to one’s negative affects. In the only study evaluating modified dialectic behavioral therapy in AN, no significant improvement has been shown in weight increase, or overall well-being (Kröger et al., 2010).

Clinical and Empirical Implications

The mostly insignificant association between the therapists’ ratings and the patients’ self-reports in comparison with the significant associations found between the TEI and several Rorschach measures deserve further consideration. One explanation is that it may stem from the low IA found in patients with ED in general (Vitousek & Manke, 1994), expressed in a deficiency in identifying and reporting internal processes as required by self-report measures. The current findings may also point out that psychodynamic therapists are more attuned to the patients’ implicit processes and dynamics, captured by the Rorschach test, and less to their conscious experiences. This supposition may suggest that therapists ought to be more aware of and relate to the patients’ perspective as to what constitutes a genuine change during treatment. Such an approach may enhance the therapeutic alliance, facilitate the process of treatment, and improve its outcome (Barber et al., 2000).

The gap between the patients’ implicit and explicit distress experiences as predicting the therapist ratings may further portray the different aspects detected by them. Therapists’ evaluations of better treatment outcomes in patients with AN-R have been predicted by a reduction in the patients’ report of depressive and anxious symptoms (SR-distress) and also by higher implicit distress (the Rorschach es measure). The last result may point out that from the therapists’ perspective improvement is predicted by elevation in implicit distress, suggesting openness to experiencing painful emotions (es measure). These multisource data appear to further emphasize the merit of implementing, for both clinical and research considerations, assessment tools to detect implicit phenomena in combination with self-report measures, as they assess different facets of experience and processes in therapy.

The data concerning the relevance of the Rorschach Color-Form Level equation as detecting the different configurations of affect regulation in ED subgroups, may indicate the need to use this measure in the assessment of ego function deficits in addition to the EII-2. Further research is required to examine the validity and clinical utility of this measure in other clinical groups indicating deficient ego functions.

Limitations and Implications for Future Research

Although our findings are promising they are still preliminary and should be regarded with caution due to the relatively small number of patients involved in the study. The current patient numbers reflect, in part, the reluctance of adolescent inpatients with ED to comply with treatment and assessment procedures. Second, as our sample consists only of inpatients, our findings cannot be generalized to populations with less severe forms of ED. Future research should replicate and substantiate our findings in larger cohorts of ambulatory adolescent and adult patients.

Another limitation of our study is that we assessed the therapist’s perspective only among the individual psychodynamically-oriented psychotherapists, focusing primarily on personality dynamics. Further research should address what constitutes therapeutic change as perceived by all different multidisciplinary team personnel, who are often more concerned with the physiological and ED symptoms. This is particularly important considering that the posttreatment scores on the EAT-26 were still pathological (>20; see Garner et al., 1982), indicating that although presenting with adequate weight and no overt ED behaviors, these patients were still preoccupied with their eating and weight. Future research should also aim to explore the therapists’ perspectives with respect to patients dropping out of treatment compared with those who comply with their treatment. Last, future studies should employ a scale equivalent to the TEI in order to address the patient’s perspective on his or her openness to treatment and change.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

References

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

American Psychiatric Association. (2006). Practice guideline for the treatment of patients with eating disorders (3rd ed.). Washington, DC: Author.

Anzai

Lindsay-Dudley

Bidwell

R. J.

(2003). Inpatient and partial hospital treatment for adolescent eating disorders. Child and Adolescent Clinics of North America, 11, 279-309.

Barber

J. P.

Connolly

M. B.

Crits-Christoph

Gladis

Siqueland

(2000). Alliance predicts patients’ outcome beyond in-treatment change in symptoms. Journal of Consulting and Clinical Psychology, 68, 1027-1032.

Beck

A. T.

Ward

C. H.

Mendelson

Mock

J. E.

Erbaugh

J. K.

(1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.

Beres

(1956). Ego deviation and the concept of schizophrenia. Psychoanalytic Study of the Child, 11, 164-235.

Bornstein

R. F.

(1998). Implicit and self-attributed dependency strivings: Differential relationships to laboratory and field measures of help-seeking. Journal of Personality and Social Psychology, 75, 779-787.

Bornstein

R. F.

(2002). A process dissociation approach to objective-projective test score interrelationships. Journal of Personality Assessment, 78, 47-68.

Bray

G. A.

(1992). An approach to the classification and evaluation of obesity. In Bjorntorp

Brodoff

B. N.

(Eds.), Obesity (pp. 294-308). Philadelphia, PA: Lippincott.

10.

Bruch

(1973). Eating disorders: Obesity, anorexia nervosa and the person within. New York, NY: Basic Books.

11.

Cicchetti

D. V.

(1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6, 284-290.

12.

Cohen

(1977). Statistical power analysis for the behavioral sciences (2nd ed.). New York, NY: Academic Press.

13.

Couturier

J. L.

Lock

(2006). Denial and minimization in adolescents with anorexia nervosa. International Journal of Eating Disorders, 39, 212-216.

14.

De la Rie

Noordenbos

Donker

Furth

E. V.

(2008). The quality of treatment of eating disorders: A comparison of the therapists’ and the patients’ perspective. International Journal of Eating Disorders, 41, 307-317.

15.

Exner

J. E.

(2001). Rorschach workbook for the comprehensive system. Asheville, NC: Rorschach Workshops.

16.

Exner

J. E.

(2003). The Rorschach: A comprehensive system: Vol. 1. Basic foundations (4th ed.). New York, NY: Wiley.

17.

Fairburn

C. G.

Harrison

P. J.

(2003). Eating disorders. Lancet, 361, 407-416.

18.

Farrell

A. D.

(1999). Evaluation of the Computerized Assessment System for Psychotherapy Evaluation and Research (CASPER) as a measure of treatment effectiveness in an outpatient training clinic. Personality Assessment, 11, 345-358.

19.

Fassino

Abbate-Daga

Amianto

Leombruni

Fornas

Garzaro

. . . Rovera

G. G.

(2001). Outcome predictors in anorexic patients after 6 months of multimodal treatment. Psychotherapy and Psychosomatics, 70, 201-208.

20.

First

M. B.

Spitzer

R. L.

Gibbon

Williams

J. B. W.

(1995). Structured clinical interview for DSM-IV axis I disorders: Patient edition (SCID-I/P, Version 2.0). New York: New York State Psychiatric Institute.

21.

Ganellen

R. J.

(2007). Assessing normal and abnormal personality functioning: Strengths and weaknesses of self-report, observer, and performance-based methods. Journal of Personality Assessment, 89, 30-40.

22.

Garner

D. M.

(1991). Eating Disorders Inventory-2: Professional manual. Odessa, FL: Psychological Assessment Resources.

23.

Garner

D. M.

Olmsted

M. P.

Bohr

Garfinkel

P. E.

(1982). The eating attitude test: Psychometric features and clinical correlates. Psychological Medicine, 112, 871-878.

24.

Greenwald

A. G.

Banaji

M. R.

(1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27.

25.

Herzog

D. B.

Greenwood

D. N.

Dorer

D. J.

Flores

A. T.

Ekeblad

E. R.

Richards

. . . Keller

M. B.

(2000). Mortality in eating disorders: A descriptive study. International Journal of Eating Disorders, 28, 20-26.

26.

Jacoby

L. L.

Lindsay

D. S.

Toth

J. P.

(1992). Unconscious influences revealed: Attention, awareness, and control. American Psychologist, 47, 802-809.

27.

Kazdin

A. E.

Esveldt-Dawson

French

N. H.

Unis

A. S.

(1987). Problem-solving skills training and relationship therapy in the treatment of antisocial child behavior. Journal of Consulting and Clinical Psychology, 53, 76-85.

28.

Kaye

W. H.

(2008). Neurobiology of anorexia and bulimia nervosa. Physiology & Behavior, 94, 121-135.

29.

Kröger

Schweiger

Sipos

Kliem

Arnold

Schunert

Reinecker

(2010). Dialectical behavior therapy and an added cognitive behavioral treatment module for eating disorders in woman with borderline personality disorder and anorexia nervosa or bulimia nervosa who failed to respond to previous treatment. An open trial with a 15-month follow-up. Journal of Behavior Therapy and Experimental Psychiatry, 41, 381-388.

30.

Linehan

M. M.

(1993). Dialectical-behavioral treatment of borderline personality disorder. New York, NY: Guilford Press.

31.

Lush

Boston

Grainger

(1991). Evaluation of psychoanalytic psychotherapy with children. Psychoanalytic Psychotherapy, l5, 191-234.

32.

Meyer

G. J.

Archer

R. P.

(2001). The hard science of Rorschach research: What do we know and where do we go? Psychological Assessment, 13, 486-502.

33.

McClelland

D. C.

Koestner

Weinberger

(1989). How do self-attributed and implicit motives differ? Psychological Review, 96, 690-702.

34.

Mihura

J. L.

Nathan-Montano

Alperin

R. J.

(2003). Rorschach measures of aggressive drive derivatives: A college student sample. Journal of Personality Assessment, 80, 41-49.

35.

Pekarik

Wolff

C. B.

(1996). Relationship of satisfaction to symptom change, follow-up adjustment, and clinical significance. Professional Psychology: Research and Practice, 27, 202-208.

36.

Rothschild

Lacoua

Eshel

Stein

(2008). Changes in defensiveness and in affective distress following inpatient treatment of eating disorders: Rorschach comprehensive system and self-report measures. Journal of Personality Assessment, 90, 356-367.

37.

Rothschild

Lacoua

Stein

(2009). Changes in ego functions and distress among two ED subgroups following integrative treatment. Eating Disorders: The Journal of Treatment & Prevention, 17, 242-260.

38.

Spielberger

C. D.

Gorsuch

R. L.

Lushene

R. E.

(1970). STAI manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Clinical Press.

39.

Stice

Burton

E. M.

Shaw

(2004). Prospective relations between bulimic pathology, depression, and substance abuse: Unpacking comorbidity in adolescent girls. Journal of Consulting and Clinical Psychology, 72, 62-71.

40.

Strober

(1987). The eating disorders family history interview. Los Angeles: University of California Press.

41.

Thompson-Brenner

Westen

(2005a). A naturalistic study of psychotherapy for bulimia nervosa, Part 1: Comorbidity and therapeutic outcome. Journal of Nervous and Mental Disease, 193, 573-584.

42.

Thompson-Brenner

Westen

(2005b). A naturalistic study of psychotherapy for bulimia nervosa, Part 2: Therapeutic intervention in the community. Journal of Nervous and Mental Disease, 193, 585-595.

43.

Thurber

Snow

(1991). The therapist evaluation inventory: Evidence for an unidimensional scale. Journal of Clinical Psychology, 47, 113-115.

44.

Viglione

D. J.

Perry

Meyer

(2003). Refinements in the Rorschach ego impairment index incorporating the human representational variable. Journal of Personality Assessment, 81, 149-156.

45.

Vitousek

Manke

(1994). Personality variables and disorders in anorexia nervosa and bulimia nervosa. Journal of Abnormal Psychology, 103, 137-147.

46.

Wonderlich

S. A.

Crosby

R. D.

Mitchell

J. E.

Engel

S. G.

(2007). Testing the validity of eating disorder diagnoses. International Journal of Eating Disorders, 40(Suppl.), S40-S45.