Compulsive exercise to control shape or weight in eating disorders: prevalence, associated features, and treatment outcome

Transcription

1 Available online at Comprehensive Psychiatry 49 (2008) Compulsive exercise to control shape or weight in eating disorders: prevalence, associated features, and treatment outcome Riccardo Dalle Grave a, Simona Calugi a, Giulio Marchesini b, a Department of Eating Disorder and Obesity, Villa Garda Hospital, Garda (Vr), Italy b Unit of Clinical Dietetics, Alma Mater Studiorum University of Bologna, Policlinico S. Orsola, I Bologna, Italy Abstract Objective: The study was aimed at assessing the prevalence of compulsive exercising to control shape and weight in eating disorders (EDs) and its relationship with treatment outcome. Method: Compulsive exercising to control shape and weight, defined according to a modified version of the Intense Exercising to Control Shape or Weight section of the Eating Disorder Examination (EDE), was assessed in 165 consecutive ED inpatients entering a protocol based on the transdiagnostic cognitive behavior theory and treatment of EDs. Baseline assessment also included anthropometry, the global EDE interview, the Beck Depression Inventory, the State-Trait Anxiety Inventory (STAI), the Eating Disorders Inventory Perfectionism Scale, and the Temperament and Character Inventory. Results: Of the patients, 45.5% were classified as compulsive exercisers, the prevalence being highest (80%) in restricting-type anorexia nervosa (AN), lowest in EDs not otherwise specified (31.9%), and intermediate in binge/purging AN (43.3%) and in purging-type bulimia nervosa (39.3%). Compulsive exercising to control shape and weight was independently predicted by the EDE restraint score (odds ratio, 1.32; 95% confidence interval, ; P =.014) after adjustment for ED; the total amount of exercise was associated with EDE restraint, as well as with the Temperament and Character Inventory reward dependence. At follow-up, an improved EDE global score was predicted by lower baseline values, higher baseline STAI and STAI improvement, and lower amount of exercise in the last 4 weeks. Voluntary treatment discontinuation was not predicted by baseline exercise. Discussion: Compulsive exercising to control shape and weight is a behavioral feature of restricting-type AN, associated with restraint and temperament dimensions, with influence on treatment outcome Elsevier Inc. All rights reserved. 1. Introduction Excessive exercising refers to an unhealthy behavior adopted by a large subgroup of patients with eating disorders [1]. The term is used by the Manual of Mental Disorders (DSM IV) [2] to characterize exercising as a symptom of bulimia nervosa (BN) and emphasizes the quantitative dimension of exercise. However, some authors have Abbreviations: AN, anorexia nervosa; BDI, Beck Depression Inventory; BMI, body mass index; BN, bulimia nervosa (purging type); BPAN, bingeing with/without purging type anorexia nervosa; CBT, cognitive-behavior therapy; EDE, Eating Disorder Examination; EDNOS, eating disorder not otherwise specified; RAN, restricting type anorexia nervosa; STAI, State-Trait Anxiety Inventory; TCI, Temperament and Character Inventory. Corresponding author. Tel.: ; fax: addresses: rdalleg@tin.it (R. Dalle Grave), sim-cal@libero.it (S. Calugi), giulio.marchesini@unibo.it (G. Marchesini). suggested a different term to take into account the compulsive feature of exercising because this dimension seems to be the most important predictor of disordered eating attitudes and behaviors [3]. Clinical indices of the compulsive dimension are the maintenance of a rigid exercise schedule, priority of exercising over other activities, detailed record keeping, and feelings of distress if unable to exercise [1,3]; but in most cases, the quantitative and the qualitative dimensions of exercising are not disentangled. The extent to which exercise is used to change shape or weight and the degree of guilt experienced whenever exercise must be postponed are the dimensions most strongly associated with eating disorder psychopathology and reduced quality of life, tested by self-reported questionnaires [4]. Only a few data are available on the prevalence of excessive exercise across eating disorder diagnostic categories. The multisite international Price Foundation Genetic Study found a prevalence of excessive exercising in 44.4% X/$ see front matter 2008 Elsevier Inc. All rights reserved. doi: /j.comppsych

2 R. Dalle Grave et al. / Comprehensive Psychiatry 49 (2008) of participants with anorexia nervosa (AN) (54.5% among the purging subtype of AN), in 20.6% of those with BN, in 20.8% of those with eating disorder not otherwise specified (EDNOS), and in 43.5% of those with a lifetime diagnosis of AN and BN [1]. A retrospective study of hospital records failed to observe significant differences in the prevalence of excessive exercising between patients with AN and those with BN or EDNOS [5]. On the contrary, another study in an inpatient setting found a significantly higher prevalence of excessive exercising among AN patients compared with BN during the acute phase of their disorder [6]. Compared with subjects with no/leisure time exercise, the individuals with eating disorders who report excessive exercising also have lower minimum body mass index (BMI), younger age at interview, higher levels of perfectionism, eating disorder symptoms, obsessions and compulsions, persistence [1], anxiety [1,7], lower scores of novelty seeking [1], and higher obsessive-compulsive disorder symptoms and traits [8]. By contrast, subjects without compulsive exercising were also reported to have a higher prevalence of self-induced vomiting, laxative misuse, and binge eating [9]. No agreement exists on the criteria to assess excessive/ compulsive exercising in individuals with eating disorders. The different definitions used to determine both the frequency and the duration of exercise (eg, more than 3 hours a day [1], at least 5 times a week for at least 1 hour without stopping [7], at least 5 days a week over the past 3 months [10]) are open to criticism, whereas its compulsive origin may be more clearly defined. However, the influence of compulsive exercising on treatment outcome has been rarely studied. In patients with AN, exercise is associated with longer inpatient treatment [5] and a shorter time to relapse [11]; but no data are available in BN and EDNOS patients. We investigated the prevalence and the associated features of compulsive exercising to control shape and weight evaluated by a direct interview in a large group of inpatients with eating disorders and assessed its role on treatment outcome. 2. Methods 2.1. Participants One hundred sixty-five female patients (age, 26.0 ± 7.8 years) with an eating disorder of clinical severity participated in the study. All patients were voluntarily and consecutively admitted to the eating disorder inpatient unit of Villa Garda Hospital between November 2003 and October The patients were referred from all over Italy by general practitioners or by outpatients' eating disorder specialists. Indications for admission were the failure of less intensive treatments (eg, outpatient treatment) or the presence of an eating disorder of clinical severity not manageable in an outpatient setting. Patients with active substance abuse, schizophrenia, and other psychotic disorders were not included in the study. The research was reviewed and approved by the Institutional Review Board of Villa Garda Hospital, Verona; and all participants (or the legal guardian for the 17 patients younger than 18 years) gave written informed consent Inpatient treatment protocol The inpatient treatment has been described elsewhere [12]. The treatment is derived from the transdiagnostic cognitive behavior theory and treatment of eating disorders [10], but has been adapted to make it suitable for an inpatient setting. The treatment is manual based [13], lasts 20 weeks (13 for inpatients followed by 7 weeks of residential dayhospital), and is divided into 3 stages. In stage 1 (weeks 1 to 4), the focus is on engaging and educating the patient, obtaining maximum early behavior change (including the interruption of compulsive exercising to control shape and weight), and creating a personalized formulation of the disorder. In stage 2 (weeks 5 to 17), the content is dictated by the extended formulation developed in stage 1. It always addresses the patient's eating disorder psychopathology (the overvaluation of eating, shape and weight and their control, and its various expressions); but in a subgroup of patients, it also addresses one or more of the following maintaining mechanisms in additional modules: clinical perfectionism, core low self-esteem, mood intolerance, interpersonal difficulties. In stage 3 (weeks 18 to 20), the focus is on maintaining progress after treatment end and on organizing the outpatient follow-up. The treatment is provided by a multidisciplinary, noneclectic team composed of physicians, psychologists, dieticians, and nurses, all trained in the transdiagnostic cognitive behavior theory and therapy for eating disorders Assessment and measures Data collection was made on the first day of hospital admission and on the last day of day-hospital treatment Demographic and clinical variables Demographic and clinical variables, including family data and a detailed medical and eating disorder history, were obtained by physicians in the course of a direct interview. Weight was measured on a medical balance and height by a stadiometer. Patients were weighted with underwear and without shoes Eating disorder diagnosis and psychopathology The Eating Disorder Examination (EDE) 12.0D [10] was carried out for diagnostic purposes and to evaluate the specific eating disorder psychopathology by a senior specialist in the field (RDG). The EDE is an investigatorbased interview that assesses the frequency of key behavioral and attitudinal aspects of eating disorders during the preceding 4 weeks (28 days). It evaluates the major areas of eating disorder psychopathology in 4 subscales (restraint,

3 348 R. Dalle Grave et al. / Comprehensive Psychiatry 49 (2008) eating concern, shape concern, and weight concern), with good discriminant validity in distinguishing individuals with eating disorders from controls [14,15] and from restrained eaters [16]. Interrater reliability of the EDE global score has been estimated to be 0.97 to 0.99 [16]. The EDE, in its validated EDE Italian translation [17], was used to generate the operational definitions of AN, BN, and binge eating disorder according to DSM-IV. The AN cases were further classified as restricting-type AN (AN with restrictive eating, no purging, and no bingeing behavior [RAN]), AN with bingeing and/or purging behavior (bingeing with or without purging type AN [BPAN]), or purging only type AN (purging and no bingeing AN [PAN]). Eating disorders that did not meet the above operational definitions were classified as cases of EDNOS. All EDNOS patients had an eating disorder of clinical severity, as defined by Fairburn and Walsh [18] Personality and general psychopathology The Temperament and Character Inventory (TCI) [19] was used to evaluate personality characteristics based on a psychobiologic model of personality. This model includes 4 temperament (novelty seeking, harm avoidance, reward dependence, and persistence) and 3 character dimensions (self-directedness, cooperativeness, and self-transcendence) [19,20]. The inventory has good internal consistency [19,21] and intertester and test-retest reliability [19], and has been validated in its Italian version [22]. The Beck Depression Inventory (BDI) [23] and the State- Trait Anxiety Inventory (STAI Form Y-1) [24] were used to assess the presence and severity of depression and trait levels of anxiety, respectively. Both have excellent internal reliability, reasonably good test-retest reliability, and good criterion validity [23,25,26], and have been validated in their Italian versions [27,28]. Finally, the Eating Disorders Inventory Perfectionism Scale (EDI-P) was used to measure perfectionism. The EDI- P is the most frequently used index of perfectionism in participants with eating disorders [29] and correlates well with other measures of perfectionism, such as the Frost Multidimensional Perfectionism Scale [30]. The EDI-P measures 2 principal constructs: self-oriented perfectionism (ie, requiring perfection of oneself) and socially prescribed perfectionism (ie, perceiving that others are demanding perfection of oneself) [31,32]. The EDI has been validated in its Italian version [33] Compulsive exercising to control shape or weight Participants were divided into compulsive exercisers and noncompulsive exercisers on the basis of a few welldefined questions defining the intention of exercise and its compulsive nature, derived from the Intense Exercising to Control Shape or Weight section of the EDE, as follows: (1) Over the past 4 weeks, have you exercised with the aim of burning up calories to control your shape or weight? (2) Have you felt compelled or obliged to exercise? (3) Have you exercised even when it caused severe interference with important activities? (4) Have you exercised to a level that might be harmful for you? (5) Have you felt distressed if you were unable to exercise? The patient was classified as compulsive exerciser in the presence of a positive answer to the first question and to anyone of the remaining. The interviewer also rated the number of days and the average amount of time (in minutes) per day spent on exercising, and calculated the total amount of exercise over the last 4 weeks. In case of doubt, the exercise was classified as noncompulsive Statistical analyses Statistical analyses were carried out by SPSS Version 11.0 (SPSS, Chicago, IL) and StatView 5.0 (SAS, Cary, NC). Continuous variables were categorized as mean ± SD or as median (interquartile range [IQR]) and categorical variables as frequency and percentage. Weight data (in kilograms) were transformed into BMI units to facilitate comparison between sexes. Analysis of variance, Mann-Whitney test, and χ 2 test were used to test the significance of difference between compulsive exercisers and noncompulsive exercisers. First, logistic regression analysis was used to identify the association of compulsive exercising with specific eating disorder categories. Correlation analysis was performed to establish links between the amount of exercise over the last 4 weeks and the other examined variables. All variables that proved to be significantly correlated were included in a linear regression analysis to identify the determinants of amount of exercise over the last 4 weeks. Changes in clinical parameters at the end of treatment were tested for significance by means of t test for paired data or Wilcoxon signed rank test for nonnormally distributed variables. Correlation analysis was performed to establish links between the change in EDE global score at the end of treatment and other examined variables. Variables that proved significant at univariate analysis were entered after a block procedure. In the first step, we entered the baseline EDE global score. In the second step, we entered the baseline measures of psychopathology and personality. In the last step, we entered the changes in psychopathology and personality measures. The significance limit was set at P b Results 3.1. Participant characteristics Sixty-five participants (39.4%) met the diagnostic criteria for AN, 28 (17.0%) met the criteria for BN, and 72 (43.6%) were classified as EDNOS. The 3 groups did not differ significantly in age and age of onset of the eating disorder; but differences in BMI, previous minimum and maximum BMI, and premorbid BMI were reported across groups. Patients with AN also had a higher number of previous inpatient treatments than EDNOS (Table 1). Among AN patients, 35 women had RAN and 30 had BPAN (9 with PAN). All patients with BN had purging-type BN.

4 R. Dalle Grave et al. / Comprehensive Psychiatry 49 (2008) Table 1 Clinical data in relation to eating disorder diagnosis (mean ± SD) AN (n = 65) BN (n = 28) EDNOS (n = 72) Age (y) 25.6 ± ± ± Current BMI 14.3 ± ± ± 3.8, b.001 (kg/m 2 ) Maximum BMI 21.2 ± ± ± b.001 (kg/m 2 ) a Minimum BMI 13.3 ± ± ± 2.3, b.001 (kg/m 2 ) a Premorbid BMI 20.5 ± ± ± (kg/m 2 ) Age of eating 16.5 ± ± ± disorder onset (ys) Previous inpatient treatments b 1 (4) 0 (1) 0.5 (2) a Since menarche occurred. b Median (IQR) (Kruskall-Wallis test). P b.05 vs AN. P b.05 vs BN Compulsive exercisers vs noncompulsive exercisers Of the patients, 45.5% were classified as compulsive exercisers. Their prevalence was higher in RAN (80%) than in BPAN (43.3%), BN (39.3%), and EDNOS (31.9%) (χ 2 = 22.63, P b.001). Among BPAN patients, 5 of 9 patients with PAN were classified as compulsive exercisers (55.5%). Patients with BPAN and RAN reported a significantly higher number of days of exercise than EDNOS patients (Table 2). Patients with RAN also reported a significantly higher total amount of exercise over the last 4 weeks than EDNOS and BN patients (Table 2). Limiting the analysis to compulsive exercisers, again the number of days of exercise, not the total amount of exercise over the last 4 weeks, differed across the ED categories. Compulsive exercisers had significantly higher scores in the EDE restraint, shape concern, and weight concern subscales and in the EDE global score, as well as lower scores in the novelty seeking subscale of TCI and a lower number of episodes of self-induced vomiting (Table 3). Logistic regression analysis only identified the EDE F P Table 3 Psychometric testing in eating disorder patients, according to compulsive exercise Noncompulsive exercisers Compulsive exercisers restraint score (odds ratio, 1.32; 95% confidence interval, ; P =.014) as independent predictor of baseline compulsive exercising after adjustment for the type of eating disorder. In the whole population, the amount of exercise over the last 4 weeks was positively associated with EDE restraint (r = 0.270, P b.01), EDE weight concern (r = 0.164, P =.05), F P value Current BMI (kg/m 2 ) 17.3 (3.9) 16.4 (3.5) Maximum BMI 22.7 (4.9) 22.0 (4.8) (kg/m 2 ) Minimum BMI 14.6 (2.5) 14.3 (2.2) (kg/m 2 ) Age (y) 27.0 (7.6) 24.7 (8.5) ED duration (mo) (92.3) 94.3 (92.6) EDE Objective bulimic 5 (29.5) 0 (20) episodes a,b Subjective bulimic 5 (28) 5 (36.5) episodes a,b Self-induced 10 (56) 0 (28) vomiting a,b Laxative misuse a,b 0 (3) 0 (0) Diuretic misuse a,b 0 (0) 0 (0) Restraint 3.2 (1.7) 4.0 (1.4) Eating concern 3.4 (1.7) 3.6 (1.3) Weight concern 3.4 (1.7) 4.0 (1.6) Shape concern 3.5 (1.4) 4.0 (1.1) Global score 3.3 (1.3) 3.9 (1.1) STAI (Form Y) 57.6 (13.8) 57.7 (14.1) BDI 29.0 (14.6) 29.9 (12.4) TCI Novelty seeking 19.2 (6.2) 17.4 (5.0) Harm avoidance 22.5 (6.8) 21.3 (6.2) Reward dependence 15.5 (3.7) 14.7 (3.8) Persistence 4.8 (1.8) 5.3 (1.8) Self-directedness 20.3 (8.5) 19.8 (7.0) Cooperativeness 31.0 (5.5) 29.9 (6.6) Self-transcendence 12.2 (6.2) 14.0 (5.8) a Over the last 28 days before the examination. b Analysis of variance or Mann-Whitney test, as appropriate. Table 2 Prevalence, number of days, and total amount of exercise over the last 4 weeks spent on exercise across diagnostic groups RAN (n = 35) BPAN (n = 30) BN (n = 28) EDNOS (n = 72) P value All cases Compulsive exercisers 28 (80.0%) 13 (43.3%) 11 (39.3%) 23 (31.9%) b.001 No. of days of exercise a 28 (22) 0 (28) 0 (11) 0 (6) b.001 Amount of exercise (min) a 1800 (13440) 0 (13440) 0 (8400) 0 (13440) b.001 Analysis limited to compulsive exercisers No. of days of exercise a 28 (4) 28 (0) 28 (22) 14 (22).002 Amount of exercise (min) a 3780 (13230) 3360 (13296) 1500 (8340) 1680 (13386).034 a In the last 28 days. P b.05 vs EDNOS.

5 350 R. Dalle Grave et al. / Comprehensive Psychiatry 49 (2008) EDE shape concern (r = 0.193, P =.05), EDE global score (r = 0.211, P =.01), and TCI reward dependence (r = 0.184, P =.05). Stepwise regression analysis revealed that only the EDE restraint score (β =.218, t = 2.359, P b.05) and TCI reward dependence (β = 1.69, t = 2.238, P b.05) were independent predictors of the amount of exercise over the last 4 weeks Compulsive exercise and treatment outcome One hundred twenty-six patients (76.4%) completed treatment (continuers), whereas 39 (23.6%) were classified as dropouts because of voluntary treatment discontinuation before the planned 20 weeks. The dropout rate was not different between compulsive exercisers and noncompulsive exercisers (21.1% vs 26.7%, χ 2 = 0.700, P = not significant). The median number of days of exercise in the last 4 weeks before entering the treatment program was 3 (IQR, 28) in dropouts and 0 (28) in continuers (Z = 0.909, P =.315); and the median amount of exercising time was 54 (13 440) minutes over the last 4 weeks vs 0 (13 440), respectively (Z = 1.100, P =.271). The mean BMI of completers increased from 17.0 kg/m 2 (SD, 3.8) on admission to 20.3 kg/m 2 (SD, 2.0) at discharge (P b.001). There was a trend toward a larger increase in BMI in compulsive exercisers (+3.7 ± 2.2 kg/m 2 vs 3.0 ± 2.6 kg/ m 2, P =.080), largely explained by the larger-than-average increase observed in AN patients (+4.9 ± 1.6 kg/m 2, P vs baseline b.001, n = 48) compared with BN (+1.0 ± 1.5) and EDNOS (+2.9 ± 2.4). In the whole data set, BMI increase was associated with a significant reduction in the number of objective and subjective bulimic episodes, episodes of selfinduced vomiting and laxative misuse, and days and amount of exercise over the last 4 weeks, and with a significant improvement of the 4 EDE subscales, of EDE global score, and of BDI and STAI scores (Table 4). The changes in EDE global score were significantly associated with its baseline values (r = 0.447, P b.001) and other baseline parameters (TCI self-directness [r = 0.204, P =.024], the number of days of exercise [r = 0.281, P =.002], and the amount of exercise over the last 4 weeks [r = 0.283, P =.001]), as well as with the changes in harm avoidance (r = 0.285, P =.003) and selfdirectedness (r = 0.299, P =.001) at TCI, with the changes in BDI (r = 0.387, P b.001) and STAI (r = 0.364, P b.001), and with the changes in the number of days (r = 0.311, P b.001) and the amount of exercise over the last 4 weeks (r = 0.280, P =.002). Linear regression analysis carried out using a block procedure revealed that only the baseline global score of EDE (β =.784, t = 12.86, P b.001), STAI total score (β =.428, t = 4.19, P b.001), STAI improvement (β =.338, t = 3.83, P b.001), and the amount of exercise over the last 4 weeks (β =.158, t = 2.87, P =.005) were independent predictors of changes in EDE global score (r 2 = 0.707). The results were largely confirmed in a separate analysis of AN cases, where baseline EDE was confirmed as the most Table 4 Psychosocial measures at admission and at the end of treatment (mean ± SD or median [IQR]) Admission Discharge Effect size t/z b P value BMI (kg/m 2 ) 17.0 ± ± b.001 EDE Objective bulimic 1 (28) 0 (0) 6.78 b.001 episodes a,b Subjective bulimic 6 (30) 0 (2) 5.86 b.001 episodes a,b Self-induced 1.5 (50) 0 (0) 6.91 b.001 vomiting a,b Laxative misuse a,b 0 (1) 0 (0) 4.94 b.001 Diuretic misuse a,b 0 (0) 0 (0) Days of exercise a,b 0 (28) 0 (0) 5.41 b.001 Total amount of 0 (13440) 0 (8400) 5.21 b.001 exercise (min) a,b Restraint 3.5 ± ± b.001 Eating concern 3.5 ± ± b.001 Weight concern 3.6 ± ± b.001 Shape concern 3.7 ± ± b.001 Global score 3.6 ± ± b.001 STAI (Form Y) 57.3 ± ± b.001 BDI 29.0 ± ± b.001 Paired t test and Z values of differences are also reported. a Over the last 28 days before the examination. b Paired t test or Wilcoxon signed rank test, as appropriate. significant predictor of changes in EDE global score (β =.717, t = 5.83, P b.001). 4. Discussion The principal finding of the study is that almost half of the individuals admitted to an inpatient cognitive behavior treatment of eating disorders reports compulsive exercising to control weight or shape. This type of exercising is more common in RAN than in any other eating disorder subtype, and it is associated with the EDE restraint score. Similarly, the amount of exercising over the last 4 weeks is predicted at baseline by 2 psychological evaluations: the EDE restraint score and TCI reward dependence. The study has 2 main strengths. Firstly, it presents data from a large sample of individuals with clinically severe eating disorder that includes several EDNOS patients, a group scarcely evaluated in previous research. Secondly, it evaluates both the presence of compulsive exercising and the total amount of exercise in the last 4 weeks using a predefined interview, partly derived from the DSM-IV definition of excessive exercise and very similar to the more recent definition of driven exercise proposed by EDE 15. At variance with the EDE 12 definition, where the diagnosis of intense exercisers is limited to subjects exercising 5 times or more per week to control shape and weight [15], in our sample, the focus was given on its compulsive nature.

6 R. Dalle Grave et al. / Comprehensive Psychiatry 49 (2008) The prevalence of compulsive exercising in our sample (45.5%) is moderately higher than the 39% prevalence of intense exercise observed in the only other study that included subjects with all the eating disorder diagnostic categories (AN, BN, and EDNOS) [1]. We confirmed that exercise, either excessive or compulsive, is more commonly associated with AN than with BN [1,6] or EDNOS [1]. We found a similar prevalence of compulsive exercising in patients with PAN (55.5%) and BAN (38.1%), but a higher prevalence in those with RAN (80%). Our compulsive exercisers also reported a significantly lower number of episodes of self-induced vomiting. These results, different from those reported in the Price Foundation Genetic Study [1], could be the consequence either of the different assessment of exercise or of a different severity of eating disorders. Our patients belonged to a distinctive subgroup of RAN with severe dietary restriction and very low weight, hospitalized in a specialist inpatient unit, whereas the RAN individuals studied by Shroff et al belonged to a very heterogeneous sample [1]. Compulsive exercising, when associated with rigid dietary restriction, leads to severe malnutrition and medical instability, a cause of in-hospital treatment [34]. Accordingly, compulsive exercising might be overrepresented in our in-patient RAN sample. The EDE restraint scores were predictors of both compulsive exercising and of the amount of exercise over the last 4 weeks, in keeping with the association between excessive exercising and food restriction in AN [35]. Animal studies found that rats increase running as a consequence of food deprivation [36-38]. However, the cognitive processes involved in the control of eating, shape, and weight might be more important than the biological effect of food restriction. This is consistent with the notion that the EDE restraint subscale measures the cognitive, not the caloric, dimension of food restriction. The correlation of exercise with other EDE measures of psychopathology (the EDE shape concern and weight concern subscales and the EDE global score) further supports this hypothesis. Future studies should investigate the relationship between cognitive eating restraint and the biological effect of food restriction on physical activity in AN individuals. A low score of TCI reward dependence was an additional predictor of the amount of exercise over the last 4 weeks. Low scores in this temperamental dimension identify individuals described as practical, tough minded, cold, and socially insensitive. These subjects rarely initiate open communication and typically have difficulties in sharing something with other people [19]. A second temperamental dimension characterizing compulsive exercisers was a low score of novelty seeking, identifying subjects described as slow tempered, indifferent, uninquisitive, unenthusiastic, unemotional, reflective, thrifty, reserved, tolerant of monotony, systematic, and orderly [19]. Both temperamental traits could explain the tendency of AN individuals to exercise in a very systematic way and alone most of the time, and their capacity to tolerate monotonous and repetitive exercises. We did not confirm the direct association between exercising and depression or anxiety, measured by the STAI and BDI questionnaires [1,7,35]. This is not in keeping with the hypothesis that anxiety in combination with food restriction may lead to high levels of physical activity, and with the alleged anxiolytic function of physical activity [35]. The improvement of EDE score was predicted by lower baseline EDE global score, higher STAI scores, lower amount of exercise in the last 4 weeks, and changes in STAI scores, whereas TCI scales had no effects, in contrast with a previous report [39]. The final outcome may be influenced by the interaction of eating disorder psychopathology with basal anxiety severity, anxiety improvement, and the amount of exercise in the 4 weeks before treatment start. These data indicate the importance of including additional strategies in the inpatient cognitive behavior therapy to help the subgroup of patients with higher level of anxiety and amount of exercise in the past 4 weeks. The study has some limitations. Firstly, data were derived from a single inpatient unit, mainly treating adult patients; and external validation is needed. Secondly, our patients belonged to a well-characterized subgroup of patients with severe eating disorder; and the results might not apply to subjects with less severe disease. Finally, the assessment was based on direct interview and on self-reported questionnaires. Because denial is a common process in AN [40,41],it is possible that a subgroup of participants denied their symptoms or gave unreliable answers. Nonetheless, the prevalence of compulsive exercise observed in our sample is similar to that reported in previous studies [1]. Compulsive exercising at baseline had very few effects on treatment outcome in our setting. Although the study was carried out during an intensive inpatient treatment, patients were exposed to environmental stressors (particularly to family stressors) during the last 7 weeks in a residential dayhospital; and there was no systematic supervision of their physical activity. A recent study showed that excessive exercisers have a shorter time to relapse [11], and relapse is a crucial issue in eating disorder treatment; our experience of intensive treatment raises hopes that the effects of treatment may be maintained. References [1] Shroff H, Reba L, Thornton LM, Tozzi F, Klump KL, Berrettini WH, et al. Features associated with excessive exercise in women with eating disorders. Int J Eat Disord 2006;39: [2] American Psychiatry Association. Diagnostic and statistical manual of mental disorders. 4th, text revision, ed. Washington (DC): American Psychiatry Association; [3] Adkins EC, Keel PK. Does excessive or compulsive best describe exercise as a symptom of bulimia nervosa? Int J Eat Disord 2005;38: [4] Mond JM, Hay PJ, Rodgers B, Owen C. An update on the definition of excessive exercise in eating disorders research. Int J Eat Disord 2006;39: [5] Solenberger SE. Exercise and eating disorders: a 3-year inpatient hospital record analysis. Eat Behav 2001;2:

7 352 R. Dalle Grave et al. / Comprehensive Psychiatry 49 (2008) [6] Davis C, Katzman DK, Kaptein S, Kirsh C, Brewer H, Kalmbach K, et al. The prevalence of high-level exercise in the eating disorders: etiological implications. Compr Psychiatry 1997;38: [7] Penas-Lledo E, Vaz Leal FJ, Waller G. Excessive exercise in anorexia nervosa and bulimia nervosa: relation to eating characteristics and general psychopathology. Int J Eat Disord 2002;31: [8] Davis C, Kaptein S. Anorexia nervosa with excessive exercise: a phenotype with close links to obsessive-compulsive disorder. Psychiatry Res 2006;142: [9] Brewerton TD, Stellefson EJ, Hibbs N, Hodges EL, Cochrane CE. Comparison of eating disorder patients with and without compulsive exercising. Int J Eat Disord 1995;17: [10] Fairburn CG, Marcus MD, Wilson GT. Cognitive-behavioral therapy for binge eating and bulimia nervosa. A comprehensive treatment manual. In: Fairburn CG, Wilson GT, editors. Binge eating: nature, assessment and treatment. New York: The Guildford Press; p [11] Strober M, Freeman R, Morrell W. The long-term course of severe anorexia nervosa in adolescents: survival analysis of recovery, relapse, and outcome predictors over years in a prospective study. Int J Eat Disord 1997;22: [12] Dalle Grave R. A multi-step cognitive behaviour therapy for eating disorders. Eur Eat Dis Rev 2005;13: [13] Dalle Grave R. Terapia Cognitivo Comportamentale dei Disturbi dell'alimentazione Durante il Ricovero. Verona: Positive Press; [14] Cooper Z, Cooper PJ, Fairburn CG. The validity of the eating disorder examination and its subscales. Br J Psychiatry 1989;154: [15] Fairburn CG, Cooper Z. The eating disorder examination. In: Fairburn CG, Wilson GT, editors. Binge eating. 12th ed. New York: The Guilford Press; p [16] Wilson GT, Smith D. Assessment of bulimia nervosa: an evaluation of the eating disorder examination. Int J Eat Disord 1989;8: [17] Mannucci E, Ricca V, Di Bernardo M, Rotella CM. Studio del comportamento alimentare con una intervista strutturata: la eating disorder examination. Il Diabete 1996;8: [18] Fairburn CG, Walsh T. Atypical eating disorders (eating disorders not otherwise specified). In: Fairburn CG, Brownell KD, editors. Eating disorders and obesity: a comprehensive handbook. 2nd ed. New York: Guilford Press; p [19] Cloninger CR, Svrakic DM, Przybeck TR. The Temperament and Character Inventory (TCI): a guide to its development and use. St Louis (Mo): Center for Psychobiology of Personality; [20] Cloninger CR, Svrakic DM, Przybeck TR. A psychobiological model of temperament and character. Arch Gen Psychiatry 1993;50: [21] Sato T, Hirano S, Narita T, Kusunoki K, Kato J, Goto M, et al. Temperament and character inventory dimensions as a predictor of response to antidepressant treatment in major depression. J Affect Disord 1999;56: [22] Fassino S, Abbate-Daga G, Amianto F, Leombruni P, Boggio S, Rovera GG. Temperament and character profile of eating disorders: a controlled study with the Temperament and Character Inventory. Int J Eat Disord 2002;32: [23] Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry 1961;4: [24] Spielberger CD, Gorsuch RL, Lushene RE. Manual for the State-Trait Anxiety Inventory (self evaluation questionnaire). Palo Alto: Consulting Psychologists Press; [25] Ramanaiah NV, Franzen M, Schill T. A psychometric study of the State-Trait Anxiety Inventory. J Pers Assess 1983;47: [26] Spielberger CD, Vagg PR. Psychometric properties of the STAI: a reply to Ramanaiah, Franzen, and Schill. J Pers Assess 1984;48:95-7. [27] Baggio A, Ferrari R, Partinico M, Vidotto G, Visentin M. Il Beck Depression Inventory per la valutazione della depressione nel dolore cronico. Il contributo degli item somatici. Int J Pain Ther 1997;7: [28] Lazzari R, Pancheri P. Questionario di Valutazione dell'ansia di Stato e di Tratto (State-Trait Anxiety Inventory). Firenze: Organizzazioni Speciali; [29] Shafran R, Cooper Z, Fairburn CG. Clinical perfectionism: a cognitivebehavioural analysis. Behav Res Ther 2002;40: [30] Frost RO, Marten P, Lahart C, Rosenblate R. The dimensions of perfectionism. Cogn Ther Res 1990;14: [31] Joiner T, Schmidt N. Dimensions of perfectionism, life stress, and depressed and anxious symptoms. J Soc Clin Psychol 1995;14: [32] Sherry SB, Hewitt PL, Besser A, McGee BJ, Flett GL. Self-oriented and socially prescribed perfectionism in the Eating Disorder Inventory Perfectionism subscale. Int J Eat Disord 2004;35: [33] Rizzardi M, Trombini G, Trombini Corazza E. Adattamento Italiano dell'eating Disorder Inventory-2. Firenze: Organizzazioni Speciali; [34] Yager J, Andersen AE. Clinical practice. Anorexia nervosa. N Engl J Med 2005;353: [35] Holtkamp K, Herpertz-Dahlmann B, Mika C, Heer M, Heussen N, Fichter M, et al. Elevated physical activity and low leptin levels cooccur in patients with anorexia nervosa. J Clin Endocrinol Metab 2003; 88: [36] Exner C, Hebebrand J, Remschmidt H, Wewetzer C, Ziegler A, Herpertz S, et al. Leptin suppresses semistarvation induced hyperactivity in rats: implications for anorexia nervosa. Mol Psychiatry 2000;5: [37] Finger FW. The effect of food deprivation and subsequent satiation upon general activity in the rat. J Comp Physiol Psychol 1951;44: [38] Pirke KM, Broocks A, Wilckens T, Marquard R, Schweiger U. Starvation-induced hyperactivity in the rat: the role of endocrine and neurotransmitter changes. Neurosci Biobehav Rev 1993;17: [39] Anderson CB, Joyce PR, Carter FA, McIntosh VV, Bulik CM. The effect of cognitive-behavioral therapy for bulimia nervosa on temperament and character as measured by the temperament and character inventory. Compr Psychiatry 2002;43: [40] Vandereycken W, Vanderlinden J. Denial of illness and the use of selfreporting measures in anorexia nervosa patients. Int J Eat Disord 1983; 2: [41] Couturier JL, Lock J. Denial and minimization in adolescents with anorexia nervosa. Int J Eat Disord 2006;39:212-6.

8 ID Title Pages Compulsive exercise to control shape or weight in eating disorders: prevalence, associated features, and treatment outcome 7