psychopathology in the eating disorders: the influence of physical activity

18
Psychopathology in the Eating Disorders: The Influence of Physical Activity' CAROLINE D A V I S ~ The Department of Psychiatry, The Toronto Hospital, and the Faculw of Medicine University of Toronto, Canada and Kinesiology and Health Science. York University. Toronto, Canada D. BLAKE WOODSIDE AND MARION P. OLMSTED The Department of Psychiatry The Toronto Hospital, and the Faculty of Medicine Universiw of Toronto, Canada SIMONE KAPTEIN Kinesiology and Health Science York Universiw Toronto. Canada This study was designed to examine whether high-level exercise (HE) and starvation are associated with greater psychopathology in patients with eating disorders than either low weight or HE on its own. Patients with either anorexia nervosa or bulimia nervosa were dichotomized according to their exercise status prior to hospital admission. High-level exercisers reported greater psychopathology than moderatehonexercisers, and these effects were not confounded by differences in body weight. Conclusions are inevitably speculative because of the cross-sectional nature of the data. It is probable, however, that certain psychological characteristicsmay foster high-level exercising among patients who develop eating disorders, whereas other characteristicsmay be exacerbated by exercise in combination with starvation. Etiology and Psychopathology Several decades of research have demonstrated that no single theory can ade- quately explain the development of eating-related disorders; instead, the eating disorders are the final common pathway of a heterogeneous set of predisposing 'This research was funded by a grant (410-94-1510) from the Social Science and Humanities Research Council, Ottawa, Canada. *Correspondence concerning this article should be addressed to Caroline Davis, York University, 343 Bethune College, 4700 Keele Street, Toronto, Ontario M3J 1P3, Canada. e-mail: [email protected]. 139 Journal of Applied Biobehavioral Research, 1999, 4, 2, pp. 139-1 56. Copyright Q 1999 by Bellwether Publishing, Ltd. All rights reserved.

Upload: caroline-davis

Post on 23-Jul-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Psychopathology in the Eating Disorders: The Influence of Physical Activity

Psychopathology in the Eating Disorders: The Influence of Physical Activity'

CAROLINE D A V I S ~ The Department of Psychiatry, The Toronto Hospital, and the Faculw of Medicine

University of Toronto, Canada and

Kinesiology and Health Science. York University. Toronto, Canada

D. BLAKE WOODSIDE AND MARION P. OLMSTED The Department of Psychiatry

The Toronto Hospital, and the Faculty of Medicine Universiw of Toronto, Canada

SIMONE KAPTEIN Kinesiology and Health Science York Universiw Toronto. Canada

This study was designed to examine whether high-level exercise (HE) and starvation are associated with greater psychopathology in patients with eating disorders than either low weight or HE on its own. Patients with either anorexia nervosa or bulimia nervosa were dichotomized according to their exercise status prior to hospital admission. High-level exercisers reported greater psychopathology than moderatehonexercisers, and these effects were not confounded by differences in body weight. Conclusions are inevitably speculative because of the cross-sectional nature of the data. It is probable, however, that certain psychological characteristics may foster high-level exercising among patients who develop eating disorders, whereas other characteristics may be exacerbated by exercise in combination with starvation.

Etiology and Psychopathology

Several decades of research have demonstrated that no single theory can ade- quately explain the development of eating-related disorders; instead, the eating disorders are the final common pathway of a heterogeneous set of predisposing

'This research was funded by a grant (410-94-1510) from the Social Science and Humanities Research Council, Ottawa, Canada.

*Correspondence concerning this article should be addressed to Caroline Davis, York University, 343 Bethune College, 4700 Keele Street, Toronto, Ontario M3J 1P3, Canada. e-mail: [email protected].

139

Journal of Applied Biobehavioral Research, 1999, 4, 2, pp. 139-1 56. Copyright Q 1999 by Bellwether Publishing, Ltd. All rights reserved.

Page 2: Psychopathology in the Eating Disorders: The Influence of Physical Activity

140 DAVIS ET AL.

individual and environmental factors (McClure, Timimi, & Westman, 1995). His- torically, however, certain causal perspectives have gained greater prominence than others in the scientific study of anorexia nervosa (AN) and bulimia nervosa (BN). For example, early theorists tended to emphasize the importance of socio- cultural influences-most importantly, the pressure on women in Western cul- tures to conform to unrealistically thin standards of female sexual attractiveness (e.g., Bruch, 1973; Garner & Garfinkel, 1978, 1980). Considerable importance has also been given to the notion that eating disturbances are largely the result of dysfunctional parent-child relationships, in particular those characterized by enmeshment and overprotectiveness so that the child’s strong need for autonomy during adolescence may take the form of control over her own body (Minuchin, Roseman, & Baker, 1978).

In the past few years, there has been increased emphasis on the role that per- sonality-and its biological underpinnings-play in the pathogenesis of the eat- ing disorders. Support for this position has come from at least three types of research: (a) the prospective analysis of personality in the later development of disordered eating; (b) comparisons among symptomatic patients, their weight- restored asymptomatic counterparts, and healthy control subjects; and (c) comor- bidity studies of eating disorders and personality disorders.

When all these data are considered together, a relatively consistent picture emerges. For example, studies have found that depression, anxiety, and obses- sionality are strongly and reliably correlated with dieting and weight preoccupa- tion in nonclinical studies (Davis & Fox, 1993; Davis, Shapiro, Elliott, & Dionne, 1993; Rogers & Petrie, 1996; Wichstrom, 1995), that these characteris- tics predict eating disturbances in longitudinal research (Leon, Fulkerson, Perry, & Early-Zald, 1995), and that they are also reported to a greater degree among long-term weight-recovered AN patients than among healthy control subjects (O’Dwyer, Lucey, & Russell, 1996; Pollice, Kaye, Greeno, & Weltzin, 1997; Strober, 1980). In addition, major depressive episodes and anxiety disorders have been found to predate the onset of AN and BN in a substantial number of cases, suggesting that they may represent one pathway into the eating disorders (Bulik, Sullivan, Carter, & Joyce, 1996; Deep, Nagy, Weltzin, Rao, & Kaye, 1995). In turn, these same characteristics tend to be exacerbated by weight loss and malnu- trition (Keys, 1950).

Physical Activity and Psychopathology

The course of an eating disorder is often one of escalating and increasingly intractable symptomatology-a slippery downward slope of illness. Some have suggested that a psychological and behavioral synergy, in the form of a feedfor- ward loop, is intrinsic in this process (Davis, 1997b; Davis et al., 1995; Pollice et a]., 1997). Psychological vulnerabilities (in conjunction with individually

Page 3: Psychopathology in the Eating Disorders: The Influence of Physical Activity

PSYCHOPATHOLOGY, ACTIVITY, EATING DISORDERS 141

relevant precipitating factors) can contribute to the onset of dieting, which, in turn, makes the psychopathology worse, leading to further food restriction and weight loss. A set of maladaptive symptoms and behaviors can quite quickly become self-perpetuating and resistant to change.

There is now evidence, both from animal experimentation and from clinical research, that physical activity plays a central role in this cycle of pathology (Davis, 1997b; Davis, Kennedy, Ralevski, & Dionne, 1994; Epling & Pierce, 1992, 1996; Katz, 1986). Although the facts seem counterintuitive, when caloric restriction and strenuous physical activity occur together, they tend to potentiate one another. Animal studies over the past 30 years have demonstrated reliably that when experimental rodents are severely food restricted (i.e., limited to one 90 min meal/day) and they have free access to a running wheel, they will increase their activity and decrease their food intake, which soon leads to pronounced weight loss and further increases in activity (Epling & Pierce, 1992; Routtenberg & Kuznesof, 1967). Over the space of a week or so, these animals show a break- down of the typical diurnal pattern of activity; they run continuously throughout the light and dark periods, and they soon die (Russell & Morse, 1996).

There are startling parallels to the exercise-induced weight-loss syndrome described above in the behavioral profile of many patients with eating disorders. For example, about 80% of patients with AN, and 50% with BN, exercise exces- sively during an acute phase of their disorder (Davis, Katzman, et al., 1997). Many also report that their physical activity levels increase steadily in almost direct proportion to their decrease in food intake and body weight and that their need to be physically active becomes obsessional and irrational (Davis et al., 1994).

Explanations for this behavioral phenomenon have been largely biological and have been based almost entirely on data obtained from animal studies. Evi- dence indicates that when starvation and strenuous physical activity occur together-and over a sustained period of time-significant alterations of central nervous system functioning begin to occur. For example, a greater increase in the hypothalamic metabolism of serotonin (5-HT) has been found in the exercising semistarved rats than in the sedentary semistarved control animals (Broocks, Schweiger, & Pirke, 1991). Others have similarly found evidence of 5-HT dys- function that is specific to the combination of strenuous exercise and food restric- tion and is not simply the consequence of weight loss (Aravich, Doerries, & Rieg, 1994; Aravich, Rieg, Ahmed, & Lauterio, 1993). These findings have important implications for understanding both the etiology and the progression of the eating disorders. Dysregulation of 5-HT is of particular interest because it has strong connections to Obsessive-Compulsive Disorder (OCD) and is strongly implicated in the regulation of hunger, mood, and activity/arousal (e.g., Barr, Goodman, Price, McDougle, & Charney, 1992; Solanto, Urrutia, & Morales, 1995). This allows for speculation that obsessionality is likely to be exacerbated

Page 4: Psychopathology in the Eating Disorders: The Influence of Physical Activity

142 DAVIS ET AL.

among patients who are exercising excessively in the face of starvation and dietary chaos.

Clinical Research Findings

To date, few attempts have been made to find psychological and behavioral markers, in the human condition, that map onto the biological evidence from animal research. In a preliminary examination of these issues, we found a posi- tive relationship between obsessive-compulsive (OC) symptomatology and exer- cise frequency in a group of high-level exercising women and a group of women diagnosed with AN (Davis et al., 1995). We also found that weight preoccupation was positively associated with frequency of exercise and with pathological atti- tudes toward exercise (e.g., exercising in the face of fatigue and illness) among the AN patients. Our next study was undertaken to broaden the scope of this first investigation. Using a quasi-experimental design, we compared a group of exces- sively exercising AN patients with those who were moderate/nonexercisers. We found, as predicted, that both OC symptomatology and OC personality character- istics were greater among the excessively exercising AN patients than among their less active counterparts (Davis, Kaptein, Kaplan, Olmsted, & Woodside, 1998). This raised the possibility that OC traits may function as an antecedent factor in the pathogenesis of AN and that extensive exercising may also exacer- bate symptomatology once the eating disorder has developed. The only other study that compared exercising with nonexercising eating-disordered patients examined only a limited number of psychological factors and did not include any measure of obsessionality (Brewerton, Stellefson, Hibbs, Hodges, & Cochrane, 1995). Although these investigators found greater body dissatisfaction among the exercising patients, the significance of that finding is compromised because their sample included both AN and BN patients and because there was no attempt to control for body size in the statistical analysis.

The present study was designed to expand the breadth of previous research. Its purpose was essentially twofold. First, we intended to test the hypothesis that exercise and very low body weight (starvation) are associated with greater obses- sionality than either low weight (AN patients who were not high-level exercisers) or exercise in the absence of low body weight (BN patients who were high-level exercisers). Employing a human analogue of the animal model described earlier, we compared emaciated AN patients (with no prior history of BN or bulimic symptomatology) and nonemaciated BN patients (with no prior history of AN). Patients in both groups were further classified as “high-level exercisers” or as “moderate/nonexercisers.”

The second purpose of the study was to establish, in the event that the predicted differences in obsessionality were found, whether these were unique differences between the exercise groups or whether they were part of a more

Page 5: Psychopathology in the Eating Disorders: The Influence of Physical Activity

PSYCHOPATHOLOGY, ACTIVITY, EATING DISORDERS 143

global psychopathology that differentiated the excessive exercisers from the nonexercisers. To this end, we included, in our assessment protocol, a broad range of symptoms and personality characteristics associated with poor psychological functioning, in addition to the measures of state and trait obses- sionality.

Method

Subjects

Female patients enrolled in one of the eating disorder programs at The Tor- onto Hospital between 1994 and 1997 took part in this study (mean age = 27.4 years, SD = 7.9 years). Patients were diagnosed according to the revised third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III- R)/DSM-IV (American Psychiatric Association, 1987, 1994) criteria. From a larger sample of patients, only those classified as AN (restrictor subtype) with no history of BN ( n = 38), as well as BN patients with no prior diagnosis of AN (n = 54), were included in the analyses. There was no significant age difference between the two groups.

According to a criterion used in previous clinical and noncliiical research (Brewerton et al., 1995; Davis & Fox, 1993; Davis et al., 1998; De Coverley Veale, 1987; Nudelman, Rosen, & Leitenberg, 1988), both patient groups were further classified either as excessive exercisers (AN = 18, BN = 21) or as moderate/nonexercisers (AN = 20, BN = 33). Inclusion in the former category required participation in some form of leisure-time exercise activity, for a mini- mum of 6 hr a week averaged over the 12 months prior to asse~sment.~ Physical activity data were obtained during a detailed, structured interview. The patient was asked to indicate the physical activities in which she had participated regu- larly over the past year. Sport/exercise activities like jogging, swimming, and biking were included, as well as the less conventional activities found among eat- ing disordered patients like pacing and repetitive stair climbing. For every activ- ity that was identified, she was then asked to specify the number of weeks of participation during that year, the average number of sessions of that activity per week, and the average duration of each session in minutes ( 1 to 30, 3 I to 60, 6 1

3Clearly, any criterion for dichotomizing a variable that is inherently continuous is somewhat arbitrary. Our decision to treat exercise level as a categorical variable was based on two factors. First of all. the distribution of the exercise data was non-normal. Essentially, it was bimodal with a large number of patients reporting little or no exercise activity in the I2 months prior to assessment and another group at the other end of the continuum reporting that they had done a great deal of activity. Second, in a number of other studies (referenced above), the 6 hr/week minimum has been employed as the criterion for a classification of "high-level exercise," and it happened that there was a natural break in the distribution of our scores at approximately this point.

Page 6: Psychopathology in the Eating Disorders: The Influence of Physical Activity

144 DAVIS ET AL.

to 90,91+). Physical activity was quantified by multiplying Weeks Per Year x Sessions Per Week x Duration Per Session in half-hour units (1,2,3,4, consecu- tively) for each activity and then summing across all activities. We have used this method of quantifying physical activity in a number of previously published clin- ical studies (e.g., Davis et al., 1994, 1995; Davis, Katzman, et al., 1997).

Measurements

Obsessive-compulsive symptoms. The Maudsle y 0 bsessional -C om pu 1s i ve Inventory (Hodgson & Rachman, 1977) is a 30-item self-report scale that assesses the various symptoms typically associated with OCD, in particular, intrusive thoughts, slowness and doubting, frequent checking, and worries about cleanliness. There is good evidence that the scale has adequate internal consis- tency and test-retest reliability (Richter, Cox, & Direnfeld, 1994).

Obsessive-compulsivepersonality. This was assessed by the total of the six subscales (viz., Emotional Constriction, Orderliness, Parsimony, Perseverance, Rigidity, and High Superego) of an inventory whose items reflect the “obses- sional” or “anal” personality type derived from psychoanalytic theory (Lazare, Klerman, & Armour, 1966, 1970). A number of reviewers have concluded that this scale is one of the best measures of obsessive-compulsive personality organi- zation (Garamoni & Schwartz, 1986; Pollak, 1979).

Narcissism. The O’Brien Multiphasic Narcissism Inventory (OMNI; O’Brien, 1987, 1988) is a measure of narcissistic pathology as it is currently, and was his- torically, understood! A factor analysis of the items has identified three moder- ately correlated factors: the first reflects tendencies to have exploitative interpersonal relations, to display a sense of entitlement, and to be blatantly exhibitionistic; the second reflects the belief that one can, and should, control others; and the third reflects self-deprecation to the point of martyrdom. This scale has demonstrated respectable reliability and has been validated with clinical groups diagnosed with Narcissistic Personality Disorder (O’Brien, 1987, 1988).

Neurotic perfectionism. The Neurotic Perfectionism Scale (NPQ; Mitzman, Slade, & Dewey, 1994) was designed to indicate the degree to which indi- viduals set unrealistically high targets, are unduly concerned about making mistakes, and are driven by a fear of failure. Test items reflect the specific atti- tudes and experiences of neurotic perfectionism that are thought to be linked to

4The most commonly used measure of narcissism has been the Narcissistic Personality Inventory (NPI; Raskin & Hall, 1979). However, despite its putative clinical orientation, the NPI has consis- tently been related to aspects of psychological adjustment such as positive self-esteem, good body image, and low levels of anxiety and depression (e.g., Emmons, 1984; Jackson. Ervin, & Hodge, 1992; Raskin, Novacek, & Hogan, 1991; Raskin & Terry, 1988). It has, therefore, been viewed as a measure of the more adaptive aspects of narcissism. The OMNI, on the other hand, has been associ- ated with poorer psychological adjustment (e.g., Davis, Claridge, & Cerullo, 1997).

Page 7: Psychopathology in the Eating Disorders: The Influence of Physical Activity

PSYCHOPATHOLOGY, ACTIVITY, EATING DISORDERS 145

eating disorders. Recent evidence suggests that a heightened risk for eating dis- orders occurs only when general perfectionism exists in the presence of a ten- dency to be anxious, hypercritical, and dissatisfied (Davis, 1997a; Slade & Dewey, 1986).

Selfesteem. The Janis-Field Feelings of Inadequacy Scale (Janis & Field, 1959) consists of 23 questions for which the respondent is asked to give self- ratings on anxiety in social situations, self-consciousness, and feelings of per- sonal self-worth. This scale has been used extensively as a measure of self- esteem (e.g., O’Neil & White, 1987; Stoltz & Galassi, 1989) and, in a compre- hensive review of self-esteem scales, it was considered to be among the most worthwhile measures (Blascovich & Tomaka, 1991). The authors of the scale have reported split-half reliabilities of .83 and an alpha coefficient of 91. Because of the wording of the items, a low score indicates good self-esteem.

Body imuge. The Body Esteem Scale (Franzoi & Shields, 1984) lists 35 body parts and body functions and asks respondents to assess each item on a 5-point Likert scale. Since its development, this scale has been used extensively in body- image research (e.g., Beren, Hayden, Wilfley, & Grilo, 1996; Henriques, Calhoun, & Cann, 1996; Montepare, 1996; Rierdan & Koff, 1997; Tiggemann, 1996). Factor-analytic procedures have identified three separate but correlated subscales for males and three for females. The female subscales are Sexual Attractiveness (assessing aspects of the body that are associated with physical attractiveness), Weight Concern (assessing aspects of the body that may be phys- ically altered through dietary restraint or exercise), and Physical Condition (per- taining to strength and agility). The alpha coefficients for these scales range from .81 to .86. In this study, the sum of the three subscales is used as a measure of global body esteem.

GZobuZpsychoputhoZ~~. The SCL-90 (Derogatis, Rickels, & Rock, 1976) is a multidimensional symptom self-report inventory with each item rated on a 5- point scale of distress. In addition to the nine primary symptom dimensions (e.g., anxiety, depression, interpersonal sensitivity), there are three global indices of pathology. The most comprehensive of the global measures is the Global Sever- ity Index, which combines information on numbers of symptoms and intensity of distress. The SCL-90 has been widely used and validated over the past two decades.

Body muss index. Weight/height (kg/m2) was calculated from height and weight measurements taken at the time of assessment. In the case of women in the inpatient program, these data were obtained from chart records.

Procedure

After receiving informed consent, basic demographic data were obtained by the interviewer. Following completion of the self-report questionnaire package,

Page 8: Psychopathology in the Eating Disorders: The Influence of Physical Activity

146 DAVIS ET AL.

the physical activity interview took place, and height and weight were measured. All patients were assessed individually by the author or her research assistant.

Results

Table 1 presents means, standard deviations, and minimum and maximum values for all variables used in the analyses, which are listed separately by patient group and exercise status.

Multivariate analysis of variance (MANOVA) procedures were carried out with patient group (AN vs. BN) and exercise status (excessive exercisers vs. moderatehonexercisers) as independent variables in the model and with the seven psychometric variables and body mass index (BMI) as the dependent vari- ables. The MANOVA tests for the hypotheses of no Overall Group, Exercise, or Group x Exercise effects were all statistically significant @ < .0328, .OOO 1, and .028 1, respectively). With respect to obsessionality, the univariate results indi- cated that, for OC symptoms, there was a significant interaction. Post hoc com- parisons using Tukey’s Honestly Significant Difference Test indicated that excessively exercising AN patients had significantly higher scores than the other three groups, which did not differ from each other. For OC personality, there was a significant group and exercise effect, indicating that scores were higher for AN patients (collapsing across exercise) and for those who were excessive exercisers (collapsing across patient group). For the remaining five measures of psychopa- thology, there was a statistically significant effect of exercise status but no differ- ences among patient groups and no significant interactions. In all cases, the excessive exercisers reported greater psychopathology. Finally, and not surpris- ingly, there was a highly significant difference between the patient groups on BMI, but there was neither a main effect for exercise status nor was there a sig- nificant interaction.

Table 2 presents a summary (ANOVA) table for each independent variable, excluding the interaction term except when it reached statistical significance. The table also includes means and standard deviations for the significant main effects.

Discussion

Previous research has demonstrated striking similarities between the exercise and feeding behavior of a great many AN patients and that of food-deprived experimental animals who develop the exercise-induced weight-loss syndrome described earlier (Davis et al., 1994). In both situations, calorie restriction and physical activity tend to function synergistically in promoting a potentially fatal drop in body weight. This animal model has enhanced our understanding of the biological factors that may function in the development and progression of some eating disorders: (a) aspects of 5-HT dysknction are specific to the combination of exercise and weight loss rather than to starvation on its own (Aravich et al.,

Page 9: Psychopathology in the Eating Disorders: The Influence of Physical Activity

PSYCHOPATHOLOGY, ACTIVITY, EATING DISORDERS 147

Table 1

Means, Standard Deviations, Minima, and Maxima for AN Variables Listed by Patient Group and by Exercise Status

Variable M SD Minimum Maximum

Excessively exercising AN patients ( n = 18)

OC symptoms OC personality Narcissism Perfectionism Self-esteem Body-esteem Global psychopathology BMI

17.6 4.7 66.5 12.7 19.5 4.4

166.6 25.6 99.0 15.5 70.0 17.0 2.3 0.8

15.2 1.8

9.0 38.5 10.3

124.0 72.0 44.0 0.9

12.3

Moderate/nonexercising AN patients (n = 20)

OC symptoms 10.0 7.3 1 .o OC personality 57.8 11.4 32.0 Narcissism 17.8 4.9 7.0 Perfectionism 148.7 39.5 79.0 Self-esteem 84.2 20.3 42.0 Body-esteem 86.4 24.7 36.0 Global psychopathology 1.9 1 .o 0.4 BMI 15.7 2.5 10.9

Excessively exercising BN patients (n = 2 1)

OC symptoms 12.2 7.3 1 .o OC personality 51.8 12.0 31.8 Narcissism 22.5 5.3 11.0 Perfectionism 170.2 20.6 120.0 Self-esteem 95.8 10.2 80.0 Body-esteem 74.1 13.1 40.0 Global psychopathology 2.1 1 .o 0.9 BMI 25.2 4.3 18.6

26.0 83.0 26.0

210.0 115.0 102.0

3.5 17.7

27.0 80.9 24.0

197.0 115.0 133.0

3.6 18.7

26.0 71.8 29.0

200.0 109.0 89.0 4.0

35.4

(table continues)

Page 10: Psychopathology in the Eating Disorders: The Influence of Physical Activity

148 DAVIS ET AL.

Table I (Continued)

Variable M SD Minimum Maximum

Moderate/nonexercising BN patient (n = 33)

OC symptoms 10.2 6.1 1 .o 25.0 OC personality Narcissism Perfectionism

49.1 12.6 12.6 72.0 18.6 6.1 7.0 30.0

147.6 33.4 87.0 203 .O Self-esteem 86.5 17.8 53.0 115.0 Body-esteem 85.1 20.9 51.0 131.0 Global psychopathology 1.6 0.9 0.1 3.3 BMI 28.4 6.9 18.6 48.7

Note. AN = anorexia nervosa; OC = obsessive-compulsive; BMI = body mass index; BN = bulimia nervosa.

1993, 1994); (b) dysfunction in the 5-HT regulatory system is associated with anxiety disorders such as OCD (Petty, Davis, Kabel, & Kramer, 1996; Pigott, 1996; Solanto et al., 1995); and (c) at high levels, opiates (such as those produced by food restriction and exercise) tend to suppress appetite and increase hyper- activity in certain animals (Marrazzi et al., 1990).

The present study sought to broaden our understanding of the exercise/ starvation dyad in the case of human eating disorders by examining psychopatho- logical differences between excessively exercising patients and those who were less active. The first prediction was that high-level exercise and low body weight would be associated with greater OC traits and symptoms than either high-level exercising or starvation on its own. This hypothesis was largely supported. With respect to OC personality traits, patients in the AN group were more obsessive- compulsive than those with BN, as others have found (e.g., Vitousek & Manke, 1994). In addition, those who exercised a great deal were more obsessive- compulsive than those who did not. As expected, these effects were additive. On the other hand, the two main effects interacted in the case of OC symptomatol- ogy, implying a synergistic influence when high-level exercise is combined with dieting. Scores were significantly higher among AN exercisers compared with the other three groups, which did not differ from each other. Moreover, scores for the AN exercisers were higher than the mean score reported on this measure for a group of patients with OCD (Richter et al., 1994).

We are not able to conclude, however, that over-exercising is uniquely associ- ated with greater obsessionality. On the broad range of other measures of psycho- pathology included in the analyses, there was a consistent and statistically

Page 11: Psychopathology in the Eating Disorders: The Influence of Physical Activity

PSYCHOPATHOLOGY, ACTIVITY, EATING DISORDERS 149

Table 2

2 x 2 (AN vs. BN x Exercisers vs. Nonexercisers) Anahses of Variance

Source MS F P <

Exercise status Patient group Exercise x Group

Exercise status

Patient group

Exercise status Patient group

Exercise status Patient group

Exercise status Patient group

Exercise status Patient group

Exercise status Patient group

Obsessive-compulsive symptoms

479.3 11.89 ,0009 160.7 3.99 .0490 184.8 4.58 .0350

Obsessive-compulsive personality

7 1 1.2 4.68 .0332 ExEx: x = 58.6, SD = 12.3 Nonex: x = 52.4, SD = 12.1

2,724.8 17.92 .0001 AN: x= 61.0, SD= 12.0 BN: x = 50.2, SD = 12.4

Narcissism

159.2 5.53 .0212 ExEx: x= 21.1, SD = 4.9 56.6 1.97 .1647 Nonex: x= 18.3, SD = 5.6

Neurotic perfectionism

8,148.2 8.30 .005 1 ExEx: x = 168.5, SD = 22.9 13.2 0.01 .9079 Nonex: x= 148.0, SD = 35.7

Self-esteem

2,697.4 9.65 .0026 ExEx: x = 97.3, SD = 12.6 13.2 0.00 .9897 Nonex: x= 85.6, SD = 18.7

Body-esteem

3,488.8 8.93 ,0037 ExEx: x = 72.2, SD= 14.9 16.3 0.04 $386 Nonex: x = 86.5, SD = 20.4

Global psychopathology

5.1 6.34 .0138 ExEx: x = 2.2, SD= 0.9 1.0 1.21 .2741 Nonex: w = 1.7, SD = 0.9

(table continues)

Page 12: Psychopathology in the Eating Disorders: The Influence of Physical Activity

150 DAVIS ET AL.

Table 2 (Continued)

Source MS F P <

Body mass index

Exercise status 72.3 3.04 .0851 AN: x= 15.5, SD = 2.2 Patient group 2,613.2 109.91 .0001 BN: x= 27.2, SD = 5.9

~ ~~~~

Note. ExEx = excessive exercisers; Nonex = nonexercisers; AN = anorexia nervosa; BN =

bulimia nervosa.

significant effect of exercise status, although there were no differences between AN and BN patients. In every case, the high-level exercisers reported greater levels of psychopathology. They had greater body disparagement, higher levels of narcissism and neurotic perfectionism, and lower self-esteem. They also had a higher severity index on the measure of global psychopathology. These dissimi- larities were not confounded by body weight; there was no difference in BMI between exercisers and nonexercisers collapsing across patient g r ~ u p . ~

The absence of psychological differences between AN and BN patients in this study-except for the measure of obsessive-compulsive personality-is perhaps not surprising, given the strong overlap between the two conditions and the fact that a high proportion of AN patients (of the restrictor subtype) eventually develop BN or bulimic symptomatology if the disorder persists (Eckert, Halmi, Marchi, Grove, & Crosby, 1995). In past research, the most consistently found differences between the two patient groups have been increased obsessionality, conformity, and constriction of affect in AN (e.g., Casper, 1990), and anxiety and depressive disorders in BN (Bulik, 1997; Bulik et al., 1996; Walters et al., 1992). It seems that the psychopathological similarities between the two conditions may be greater than their differences.

Given that these data are cross-sectional, it is impossible to determine whether the exercise differences are a consequence of prolonged hyperactivity or whether those who combine strenuous exercise with severe food restriction-a

'Although the Exercise Status main effect did not reach statistical significance in the present study and there was no significant interaction of Group x Exercise Status, it should be noted that among AN patients, the BMI between the two exercise groups was virtually identical (Table I ) . On the other hand, among BN patients, the difference between the two exercise groups was substantial (25.2 vs. 28.4, respectively). In fact, if the BN data are analyzed separately, this difference is statisti- cally significant (p = .04). Together, these results suggest two things. First, when the AN and BN data are combined, the variability in BMI is much greater than when either patient group is considered separately. As a consequence, the power of the statistical test is lower in the former case than in the latter. Second, the very similar, and very low, BMI in the AN groups may indicate a floor effect-that is, that there is a limit below which BMI is physiologically unlikely to fall whether patients are starv- ing and exercising, or just starving.

Page 13: Psychopathology in the Eating Disorders: The Influence of Physical Activity

PSYCHOPATHOLOGY, ACTIVITY, EATING DISORDERS 151

seemingly more aberrant set of behaviors than either on its own-have a more virulent form of the eating disorders. Indeed, both may occur. Some of the neuro- chemical alterations caused by exercise and dieting are likely to exacerbate aspects of psychopathology such as obsessionality. Indeed, they may actually sustain, and even worsen, the very behaviors from which they initially arose.

On the other hand, it may also be that some psychological characteristics are more likely to foster over-exercising among eating-disordered patients. For example, previous research has indicated that high-level exercising women have more obligatory and perfectionistic attitudes toward exercise, are more weight preoccupied, and have a greater drive for thinness than women who exercise less assiduously, suggesting that over-exercisers are at increased risk for developing an eating disorder and that this profile of characteristics has etiological significance in the disorder (Davis et al., 1995; Iannos & Tiggemann, 1997; Yates, 1991). Narcissistic personality traits are also relevant in this context. In our need to regulate and protect self-esteem, we have multiple sources of self- affirmation available. In a culture that admires fitness and thinness, there may be an added incentive for the narcissist to engage in behaviors like dieting and exer- cise that are highly regarded and rewarded.

Scores on the Global Severity Index of the SCL-90 have indicated a higher degree of mood-related distress among the exercisers than the nonexercisers. It could be that strenuous exercising, independent of the physiological conse- quences of dieting, exacerbates depression and anxiety because it contributes to energy depletion. On the other hand, over-exercisers may have had greater affec- tive instability premorbidly. Support for the latter point of view stems from evi- dence of the antidepressant effects of physical activity (Martinsen & Morgan, 1 996) and our clinical observations that many eating disordered patients report- edly use exercise-at least initially-as a mood-regulation strategy.

Although the present study identified important associations between high- level exercising and psychopathology in eating disordered patients, it is also important to address its limitations. Causal links between the maladaptive behav- iors discussed in this paper and psychological functioning are not clear; by neces- sity, our conclusions have, therefore, been speculative, although for the most part they have been grounded in well-supported theory and are in accordance with previous research. Although prospective population studies are the ideal next step in untangling issues of causality, the very low incidence and prevalence of eating disorders in the population make this approach difficult to execute. Another tactic, and one that is strongly recommended, is to study the long-term outcome of weight-restored patients who are no longer dieting or exercising to excess. Such a strategy would allow us to make psychological comparisons between the healthy state and the ill state; it would also allow for outcome com- parisons among those who were exercising excessively during their disorder compared with those who were less active.

Page 14: Psychopathology in the Eating Disorders: The Influence of Physical Activity

152 DAVIS ET AL.

References

American Psychiatric Association. ( 1 987). Diagnostic and statistical manual, for mental disorders (3rd ed., rev.). Washington, DC: Author.

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

Aravich, P. F., Doerries, L. E., & Rieg, T. S. (1994). Exercise-induced weight loss in the rat and anorexia nervosa. Appetite, 23, 196.

Aravich, P. F., Rieg, T. S., Ahmed, I., & Lauteno, T. J. (1 993). Fluoxetine induces vasopressin and oxytocin abnormalities in food-restricted rats given voluntary exercise: Relationship to anorexia nervosa. Brain Research, 612, 180- 189.

Barr, L. C., Goodman, W. K., Price, L. H., McDougle, C. J., & Charney, D. S. ( 1992). The serotonin hypothesis of obsessive compulsive disorder: Implica- tions of pharmacologic challenge studies. Journal of Clinical Psychiatry, 53,

Beren, S. E., Hayden, H. A., Wilfley, D. E., & Grilo, C. M. (1996). The influence of sexual orientation on body dissatisfaction in adult men and women. Inter- national Journal ofEating Disorders, 20, 135- 14 1.

Blascovich, J., & Tomaka, J. (1991). Measures of self-esteem. In J. P. Robinson & P. R. Shaver (Eds.), Measures ofpersonality and social psychological atti- tudes (Vol. 1, pp. 1 15-160). San Diego, CA: Academic.

Brewerton, T. D., Stellefson, E. J., Hibbs, N., Hodges, E. L., & Cochrane, C. E. (1 995). Comparison of eating disorder patients with and without compulsive exercising. International Journal of Eating Disorders, 17,4 13-4 16.

Broocks, A., Schweiger, U., & Pirke, K. M. (1991). The influence of semistarva- tion-induced hyperactivity on hypothalamic serotonin metabolism. Physiol- ogy and Behavior, 50,385-388.

17-28.

Bruch, H. (1973). Eating disorders. New York, NY: Basic Books. Bulik, C. M. (1997). Predictors of the development of bulimia nervosa in women

with anorexia nervosa. The Journal of Nervous and Mental Disease, 185,

Bulik, C. M., Sullivan, P. F., Carter, F. A., & Joyce, P. R. (1 996). Lifetime anxiety disorders in women with bulimia nervosa. Comprehensive Psychiatry, 37,

Casper, R. (1990). Personality features of women with good outcome from restricting anorexia nervosa. Psychosomatic Medicine, 52, 156- 170.

Davis, C. (1997a). Normal and neurotic perfectionism in eating disorders: An interactive model. International Journal of Eating Disorders, 22,42 1-426.

Davis, C. (1 997b). Eating disorders and hyperactivity: A psychobiological per- spective. Canadian Journal of Psychiatry, 42, 168- 175.

Davis, C., Claridge, G., & Cerullo, D. (1997). Personality factors predisposing to weight preoccupation: A continuum approach to the association between

704-707.

368-374.

Page 15: Psychopathology in the Eating Disorders: The Influence of Physical Activity

PSYCHOPATHOLOGY, ACTIVITY, EATING DISORDERS 153

eating disorders and personality disorders. Journal of Psychiatric Research,

Davis, C., & Fox, J. (1993). Excessive exercise and weight preoccupation in women. Addictive Behaviors, 18,20 1-2 1 1.

Davis, C., Kaptein, S., Kaplan, A. S., Olmsted, M. P., & Woodside, D. B. (1 998). Obsessionality in anorexia nervosa: The moderating influence of exercise. Psychosomatic Medicine, 60, 192- 197.

Davis, C., Katzman, D. K., Kaptein, S., Kirsh, C., Brewer, H., Kalmbach, K., Olmsted, M. P., Woodside, D. B., & Kaplan, A. S. (1997). The prevalence of high-level exercise in the eating disorders: Etiological implications. Compre- hensive Psychiatry, 38,32 1-326.

Davis, C., Kennedy, S. H., Ralevski, E., & Dionne, M. (1994). The role of physi- cal activity in the development and maintenance of eating disorders. Psycho- logical Medicine, 24,957-967.

Davis, C., Kennedy, S. H., Ralevski, E., Dionne, M., Brewer, H., Neitzert, C., & Ratusny, D. (1995). Obsessive compulsiveness and physical activity in anorexia nervosa and high-level exercising. Journal of Psychosomatic Research, 39,967-976.

Davis, C., Shapiro, C. M., Elliott, S., & Dionne, M. (1993). Personality and other correlates of dietary restraint: An age by sex comparison. Personality and Individual Diflerences, 14,297-305.

De Coverley Veale, D. ( 1987). Exercise dependence. British Journal qf Addic- tion, 82, 735-740.

Deep, A. L., Nagy, L. M., Weltzin, T. E., Rao, R., & Kaye, W. H. (1995). Pre- morbid onset of psychopathology in long-term recovered anorexia nervosa. International Journal of Eating Disorders, 17,29 1-297.

Derogatis, L. R., Rickels, K., & Rock, A. F. (1976). The SCL-90 and the MMPI: A step in the validation ofa new self-report scale. British Journal of Psychia-

Eckert, E., Halmi, K., Marchi, P., Grove, W., & Crosby, R. (1 995). Ten-year fol- low-up of anorexia nervosa: Clinical course and outcome. Psychological Medicine, 25, 143- 156.

Emmons, R. A. (1984). Factor analysis and construct validity of the Narcissistic Personality Inventory. Journal of Personality Assessment, 48,29 1 -300.

Epling, W. F., & Pierce, W. D. (1992). Solving the anorexia puzzle. Toronto, Canada: Hogrefe & Huber.

Epling, W. F., & Pierce, W. D. (1996). An overview of activity anorexia. In W. F. Epling & W. D. Pierce (Eds.), Activity anorexia theory, research, and treat- ment (pp. 3-12). Mahwah, NJ: Lawrence Erlbaum.

Franzoi, S. L., & Shields, S. A. (1984). The Body Esteem Scale: Multidimen- sional structure and sex differences in a college population. Journal qf Per- sonality Assessment, 48, 173- 178.

31,467-480.

try, 128,280-289.

Page 16: Psychopathology in the Eating Disorders: The Influence of Physical Activity

154 DAVIS ET AL.

Garamoni, G. L., & Schwa-, R. M. (1986). Type A behavior pattern and com- pulsive personality: Toward a psychodynamic-behavioral integration. Clini- cal Psychology Review, 6,311-336.

Garner, D. M., & Garfinkel, P. E. (1978). Sociocultural factors in anorexia ner- vosa. Lancet, ii, 674.

Gamer, D. M., & Garfinkel, P. E. (1980). Sociocultural factors in the develop- ment of anorexia nervosa. Psychological Medicine, 10,647-656.

Henriques, G. R., Calhoun, L. G., & Cann, A. (1996). Ethnic differences in women’s body satisfaction: An experimental investigation. The Journal qf Social Psychology, 136,689-697.

Hodgson, R. J., & Rachman, S. (1977). Obsessional-compulsive complaints. Behavior Research and Therapy, 15,389-395.

lannos, M., & Tiggemann, M. (1997). Personality of the excessive exerciser. Per- sonality and Individual Differences, 22,775-778.

Jackson, L. A., Ervin, K. S., & Hodge, C. N. (1992). Narcissism and body image. Journal of Research in Personality, 26,357-370.

Janis, 1. L., & Field, P. B. (1959). A behavioral assessment of persuasibility: Con- sistency of individual differences. In C. I. Jovland & I. L. Janis (Eds.), Person- ality andpersuasibility (pp. 29-54). New Haven, CT: Yale University Press.

Katz, J. L. (1986). Long-distance running, anorexia nervosa, and bulimia: A report of two cases. Comprehensive Psychiatry, 21,74-78.

Keys, A. (1950). The biology qfhuman starvation. Minneapolis, MN: University of Minnesota Press.

Lazare, A., Klerman, G. L., & Armour, D. J. (1966). Oral obsessive and hysteri- cal personality patterns: An investigation of psychoanalytic concepts by means of factor analysis. Archives of General Psychiatry, 14,624.

Lazare, A., Klerman, G. L., & Armour, D. J. (1970). Oral, obsessive and hysteri- cal personality patterns. Journal of Psychiatric Research, 7,275-290.

Leon, G. R., Fulkerson, J. A., Perry, C. L., & Early-Zald, M. B. (1995). Prospec- tive analysis of personality and behavioral vulnerabilities and gender influ- ences in the later development of disordered eating. Journal of Abnormal

Marrazzi, M. A., Mullings-Britton, J., Stack, L., Powers, R. J., Lawhorn, J., Graham, V., Eccles, T., & Gunter, S. (1 990). Atypical endogenous opioid sys- tems in mice in relation to an auto-addiction opioid model of anorexia ner- vosa. Life Sciences, 47, 1427-1435.

Martinsen, E. W., & Morgan, W. P. (1996). Antidepressant effects of physical activity. In W. P. Morgan (Ed.), Physical activity and mental health (pp. 93- 106). London, UK: Taylor & Francis.

McClure, G. M., Timimi, S., & Westman, A. (1995). Anorexia nervosa in early adolescence following illness-the importance of the sick role. Journal of Adolescence, 18,359-369.

psycho lo^, 104, 140-149.

Page 17: Psychopathology in the Eating Disorders: The Influence of Physical Activity

PSYCHOPATHOLOGY, ACTIVITY, EATING DISORDERS 155

Minuchin, S., Roseman, B., & Baker, I. (1978). Psychosomatic families: Anor- exia nervosa in context. Cambridge, MA: Harvard University Press.

Mitzman, S. F., Slade, P., & Dewey, M. E. (1994). Preliminary development of a questionnaire designed to measure neurotic perfectionism in the eating disor- ders. Journal of Clinical Psychology, 50,5 16-522.

Montepare, J. M. (1996). Actual and subjective age-related differences in women’s attitudes toward their bodies across the life span. Journal of Adult Development, 3, I 7 1 - 182.

Nudelman, S., Rosen, J. C., & Leitenberg, H. (1988). Dissimilarities in eating attitudes, body image distortion, depression, and self-esteem between high- intensity male runners and women with bulimia nervosa. international Jour- nal of Eating Disorders, 7,625-634.

O’Brien, M. L. (1987). Examining the dimensionality of pathological narcissism: Factor analysis and construct validity of the O’Brien Multiphasic Narcissism Inventory. Psychological Reports, 61,499-510.

O’Brien, M. L. (1988). Further evidence of the validity of the O’Brien Multi- phasic Narcissism Inventory. Psychological Reports, 62,879-882.

O’Dwyer, A. M., Lucey, J. V., & Russell, G. F. M. (1996). Serotonin activity in anorexia nervosa after long-term weight restoration: Response to D-fenflu- ramine challenge. Psychological Medicine, 26,353-359.

O’Neil, M . K., & White, P. (1987). Psychodynamic group treatment of young adult bulimic women: Preliminary positive results. Canadian Journal of Psy- chiatty, 32, 153-155.

Petty, F., Davis, L. L., Kabel, D., & Kramer, G. L. (1996). Serotonin dyshnction disorders: A behavioral neurochemistry perspective. Journal of Clinical Psy- chiatry, 57, 1-6.

Pigott, T. A. (1996). OCD: Where the serotonin selectivity story begins. Journal of Clinical Psychiatry, 57, 11-20.

Pollak, J. M. ( 1979). Obsessive-compulsive personality: A review. Psychological Bulletin, 36,225-241.

Pollice, C., Kaye, W. H., Greeno, C. G., & Welkin, T. E. (1997). Relationship of depression, anxiety, and obsessionality to state of illness in anorexia nervosa. International Journal of Eating Disorders, 21,367-376.

Raskin, R., & Hall, C. S. (1979). A narcissistic personality inventory. Psychulog- ical Reports, 45,590.

Raskin, R., Novacek, J., & Hogan, R. (1991). Narcissistic self-esteem manage- ment. Journal of Personality and Social Psychology, 60,911-91 8.

Raskin, R., & Terry, H. (1988). A principal-components analysis of the Narcis- sistic Personality Inventory and hrther evidence of its construct validity. Journal of Personality and Social Psychology, 54,890-902.

Richter, M. A., Cox, B. J., & Direnfeld, D. M. (1994). A comparison of three assessment instruments for obsessive-compulsive symptoms. Journal of Behavioral Therapy and Experimental Psychiatty, 25,143- 147.

Page 18: Psychopathology in the Eating Disorders: The Influence of Physical Activity

156 DAVIS ET AL.

Rierdan, J., & Koff, E. (1997). Weight, weight-related aspects of body image, and depression in early adolescent girls. Adolescence, 32,6 15-624.

Rogers, R. L., & Petrie, T. A. (1996). Personality correlates of anorexic symp- tomatology in female undergraduates. Journal of Counseling and Develop- ment, 75, 138-144.

Routtenberg, A., & Kuznesof, A. W. (1967). Self-starvation of rats living in activity wheels on a restricted feeding schedule. Journal of Comparative Physiology and Psychology, 64,4 14-42 1.

Russell, J. C., & Morse, A. D. (1 996). The induction and maintenance of hyper- activity during food restriction in the rat. In W. F. Epling & W. D. Pierce (Eds.), Activity anorexia theory, research. and treatment (pp. 1 13- 12 1 ). Mahwah, NJ: Lawrence Erlbaum.

Slade, P. D., & Dewey, M. J. (1986). Development and preliminary validation of SCANS: A screening for identifying individuals at risk of developing anorexia and bulimia nervosa. International Journal of Eating Disorders, 5,

Solanto, M. V., Urrutia, V., & Morales, A. (1 995). Serotonin dysregulation and psychopathology in anorexia nervosa and obsessive-compulsive disorder. Eating Disorders, 3,56-73.

Stoltz, R. F., & Galassi, J. P. (1989). Internal attributions and types of depression in college students: The learned helplessness model revisited. Journal of Counseling Psychology, 36,3 16-32 1.

Strober, M. ( 1980). Personality and symptomatological features in young, non- chronic anorexia nervosa patients. Journal of Ps,vcho.somatic Research, 24, 353-359.

Tiggemann, M. (1 996). “Thinking” versus “feeling” fat: Correlates of two indices of body image dissatisfaction. Australian Journal of Psychologv, 48,21-25.

Vitousek, K., & Manke, F. (1994). Personality variables and disorders in anor- exia nervosa and bulimia nervosa. Journal of Abnormal Psychology, 103,

Walters, E. E., Neale, M. C., Eaves, L. J., Heath, A. C., Kessler, R. C., & Ken- dler, K. S. (1992). Bulimia nervosa and major depression: A study of com- mon genetic and environmental factors. Psychological Medicine, 22, 6 17- 622,

Wichstrom, L. (1995). Social, psychological and physical correlates of eating problems. A study of the general adolescent population in Norway. Psycho- logical Medicine, 25,561-519.

Yates, A. (1991). Compulsive exercise and the eating disorders. New York, NY: BrunnerMazel.

5 17-538.

137-147.