点击显示 收起
1 From the Department of Psychiatry, Weight and Eating Disorders Program, University of Pennsylvania School of Medicine, Philadelphia
2 Supported by grants from the National Institute of Diabetes Digestive and Kidney Disease (DK-50058-04 and K24-DK065018) to TAW. Novartis Nutrition is acknowledged for providing the OPTIFAST product used in this study. 3 Address reprint requests to TA Wadden, Department of Psychiatry, Weight and Eating Disorders Program, University of Pennsylvania School of Medicine, 3535 Market Street, Suite 3029, Philadelphia, PA 19104. E-mail: wadden{at}mail.med.upenn.edu.
ABSTRACT
Background: Some investigators fear that dieting may precipitate binge eating and other adverse behavioral consequences.
Objective: The objective of the study was to examine whether dieting would elicit binge eating and mood disturbance in individuals free of these complications before treatment.
Design: A total of 123 obese women were randomly assigned to 1) a 1000 kcal/d diet that included 4 servings/d of a liquid meal replacement (MR); 2) a 1200–1500 kcal/d balanced deficit diet (BDD) of conventional foods; or 3) a nondieting (ND) approach that discouraged energy restriction. All women attended weekly group sessions for 20 wk and biweekly sessions from week 20 to week 40.
Results: At week 20, participants in the MR, BDD, and ND groups lost 12.1 ± 6.7%, 7.8 ± 6.0%, and 0.1 ± 2.4% of initial weight, respectively (P < 0.001). During the first 20 wk, there were no significant differences among groups in the number of persons who had objective binge episodes or in reports of hunger or dietary disinhibition. Symptoms of depression decreased significantly more (P < 0.001) in the MR and BDD groups than in ND participants. At week 28, significantly more (P < 0.003) cases of binge eating were observed in MR participants than in the 2 other groups. No differences, however, were observed between groups at weeks 40 or 65 (a follow-up visit). At no time did any participant meet criteria for binge-eating disorder.
Conclusion: Concerns about possible adverse behavioral consequences of dieting should not dissuade primary care providers from recommending modest energy restriction to obese individuals.
Key Words: Obesity • dieting • weight loss • binge eating • depression
INTRODUCTION
America is experiencing an epidemic of obesity. Fully 64% of adults are either overweight or obese, and the rates are only expected to rise in the next decade (1, 2). As a result, the US Surgeon General and several government agencies have called for increased efforts to treat obesity, as well as to prevent its occurrence (3-5). The National Heart, Lung, and Blood Institute (4) has recommended that individuals with a body mass index (BMI; in kg/m2) 30, which is the criterion for obesity, exercise 30 min/d and decrease their calorie intake by 500 or more kcal/d, with the goal of losing 10% of initial body weight. Diet and exercise may be supplemented by other therapies if needed.
Some investigators believe that dieting, defined as intentional caloric restriction to reduce body weight (6), is associated with the development of eating disorders (7-10). Young women, in particular, often diet aggressively in pursuit of an ever-thinner ideal (11). A small but significant minority develops anorexia nervosa or bulimia nervosa (12). With the latter condition, dieting is believed to cause both physiologic and psychological deprivation that leads to binge eating (9). This behavior is characterized by the consumption of a large amount of food in a brief period and is accompanied by a subjective loss of control and subsequent remorse (13, 14). Dieting has also been reported to be associated with depression, low self-esteem, and negative body image (8-10, 15). Results of a classic study of average-weight male volunteers, who reduced their body weight 24% below normal limits, revealed marked symptoms of depression, apathy, and food preoccupation, in addition to binge eating (16).
Whereas aggressive dieting clearly may be associated with ill effects in persons of average weight, the relevance of these findings to obese individuals who wish to lose weight through modest caloric restriction is not clear. The National Task Force on the Prevention and Treatment of Obesity recently reviewed the literature on this topic and concluded that, in obese adults, modest caloric restriction does not appear to be associated with the development of binge eating or adverse psychological effects (6). This conclusion was based on the strongest available evidence (17-23). None of the studies reviewed, however, was designed specifically to determine whether dieting would precipitate binge eating or other behavioral complications in obese individuals. The present study was designed to this end. It assessed whether caloric restriction would elicit binge eating in obese individuals who were free of this complication before treatment. We assessed the effects of a traditional 1200–1500 kcal/d reducing diet comprised of conventional foods, as well as of a 1000 kcal/d diet that included 4 servings daily of a liquid meal replacement. The latter diet was included because of concerns that low-calorie and very-low-calorie liquid diets, in particular, precipitate binge eating (24). Changes in these 2 groups in eating behavior, as well as mood, self-esteem, and body image, were compared with those in a third condition in which participants were instructed not to diet.
SUBJECTS AND METHODS
Participants
Participants were 123 women with a mean (± SD) age of 44.2 ± 10.0 y, weight of 97.3 ± 13.0 kg, height of 164.3 ± 6.4 cm, and BMI of 35.9 ± 4.5. Seventy-nine were European American, 43 were African American, and 1 was Hispanic American. Participants were selected from respondents to advertisements in local newspapers that described a weight management program. Enrollment was limited to women because they are far more likely than men to diet and to develop eating disorders. Participation was further limited to those with a BMI of 30–43 and who were free of physical contraindications including: types 1 or 2 diabetes; uncontrolled hypertension (>140/90 mm Hg); a history of cerebrovascular, cardiovascular, kidney, or liver disease; the use of medications known to affect body weight (eg steroids); pregnancy or lactation; a weight loss 5 kg and/or the use of anorectic agents in the prior 6 mo; and the use of selective-serotonin re-uptake inhibitors (SSRI), monoamine oxidase (MAO) inhibitors. Additional exclusions included binge eating, major depression, and other psychiatric conditions that significantly affected daily function. Characteristics of the participants are shown in Table 1.
View this table:
TABLE 1. Baseline characteristics of participants randomly assigned to meal replacement (MR), balanced-deficit diet (BDD), or nondieting (ND) conditions1
Respondents completed a telephone screening. Those who met the above criteria were scheduled for a 1-h interview with a psychologist who described the nature and requirements of the study and obtained applicants’ informed consent. The psychologist reviewed candidates’ responses to the Beck Depression Inventory (25) and the Questionnaire on Weight and Eating Patterns (26, 27), the latter of which assesses binge eating. Women with significant symptoms of depression, as defined by a BDI score >25, were excluded from the study, as were those who had engaged in any binge eating episodes in the previous 3 mo. The latter diagnosis required that the applicant consume both an objectively large amount of food, as judged by the psychologist, and report loss of control (13). Women who remained eligible and interested in participating were referred to their primary care physicians who conducted a history and physical examination to determine that participants were free of the physical complications noted above. A total of 55 respondents were excluded from the study—18 because of binge eating, 31 for depression/psychiatric illness, and 6 because of medical complications. All were provided an appropriate referral. The study protocol was approved by the University of Pennsylvania’s Committee on Subjects Involving Human Beings. (We note that an additional 21 women were recruited into the study and randomly assigned to a pilot intervention. It examined the efficacy in inducing weight loss of individual brief behavioral treatment [ie 20–25 min sessions] provided once a month. Results of this group [which was included to assess a new method of managing obesity in primary care practice] will be described in a separate report.)
Procedures
The 123 participants were randomly assigned to one of 3 conditions: 1) balanced-deficit diet (BDD; n = 43); 2) meal replacement plan (MR; n = 41); or 3) nondieting approach (ND; n = 39). The specifics of each intervention are described in the next section; however, the 3 groups shared several commonalities. All participants attended weekly group treatment sessions during the first 20 wk and every-other week sessions during weeks 22–40. Group sessions lasted 90 min, included 7 to 10 participants, and were led by a clinical psychologist. A registered dietitian co-led 6 sessions in each of the 3 treatment conditions. After week 40, participants attended follow-up group sessions at week 52 and week 65.
Treatment conditions
Balanced-deficit diet
Participants in this group were instructed to consume a self-selected BDD of 1200–1500 kcal/d, with 15% of calories from protein, 30% or fewer from fat, and the remainder from carbohydrate. This dietary regimen, which is based on the Food Guide Pyramid, is recommended by the LEARN Program for Weight Control (28). Participants received a copy of the LEARN manual, and treatment sessions reviewed dietary adherence, as recorded in daily food records. Women could eat any foods they wished, provided they adhered to their dietary goals. They began dieting at week 2, after a 1-wk run-in, during which they recorded their food intake but were asked to maintain their usual eating habits. Treatment sessions also reviewed other behavioral-weight-control topics covered in the LEARN manual including stimulus control, slowing eating, social support, cognitive restructuring, and relapse prevention (29). Activity goals included walking (or engaging in other aerobic activity) for 150 min/wk by the end of week 20, with an increase to 180 min/wk by week 40. Participants kept records of their physical activity that they reviewed in weekly group sessions from week 1 to week 20 and biweekly meetings from week 22 to week 40. Figure 1 shows the phases of treatment for participants in this group and the 2 other treatment conditions.
FIGURE 1.. The figure shows the weeks at which the Eating Disorder Examination (EDE) and other measures were administered, as well as the schedule on which different treatment components were delivered to participants in the 3 groups.
Meal replacement
Women in this condition received the same treatment as those in the BDD group with one major difference. From week 2 to week 13, they were prescribed a 1000 kcal/d MR plan that consisted of 4 servings/d of a liquid diet (OPTIFAST 800; Novartis Nutrition Co, Minneapolis), combined with an evening meal of a frozen food entree, a serving of fruit, and a green salad. Each serving of the liquid diet provided 160 kcal, with 14 g of protein, 20 g of carbohydrate, and 3 g of fat. Beginning at week 14, participants gradually decreased their consumption of the liquid diet, so that by week 17 they were prescribed a 1200–1500 kcal/d diet of conventional foods, the same as women in the BDD. The MR plan was included to ensure that at least one group of participants achieved marked caloric restriction. Previous studies have shown that MR plans, whether comprised of conventional foods or liquid diets, facilitate obese individuals’ adherence to their calorie goals and produce significantly larger weight losses than self-selected diets (of conventional foods) that have the same calorie goal (30, 31).
Nondieting approach
Participants in this group were explicitly instructed not to reduce their calorie intake. This recommendation was provided in the context of a program developed by Polivy and Herman (32) and by Polivy (J Polivy, unpublished observations, 1991) that discussed the ineffectiveness of dieting for achieving long-term weight control. To encourage them to give up dieting, participants began the program by making a list of all the diets they had tried, determining the number of pounds they had lost and regained, and calculating what they had spent on these efforts. At week 6, women were encouraged to adopt a new eating plan. It prescribed that they: 1) eat at least every 4 h to avoid becoming hungry; 2) consume whatever foods they desired, thus, eliminating the notion of bad foods that should be avoided; and 3) stop eating when they felt full. Registered dietitians provided women in this group the same 6 lectures on healthy eating that were presented to participants in the 2 other conditions (but no calorie recommendations were provided). These women also were instructed to increase their physical activity by the same number of minutes/wk as those in the BDD and MR groups. ND participants also received instruction in improving self-esteem and body image, as well as in living more fulfilling lives, regardless of body weight (33-35).
Dependent measures
Weight
Body weight was measured using a balance-beam scale or an electronic digital scale (model 6800; Detecto). Participants were weighed in light clothing, without shoes, and were measured on the same scale throughout the study. Consistent with behavioral-weight-loss therapy, participants in the BDD and MR conditions were weighed at all treatment visits, while those in the ND group were weighed only at monthly intervals. Change in body weight was used as an estimate of participants’ adherence to their prescribed calorie intake. We predicted that participants in the MR group would lose significantly more weight at week 20 than would women in the BDD group who, in turn, would lose significantly more than the ND participants. We did not assess self-reported caloric intake because of its substantial underestimation of energy intake, as determined by doubly labeled water, the gold standard in this area (36). (Cost prevented us from using this latter method.)
Binge eating
The Eating Disorder Examination (EDE) was used to assess the occurrence of binge eating and related behaviors. The EDE is a structured clinical interview that is administered by a trained examiner (13). It assesses participants’ eating-related attitudes and behaviors and yields diagnoses for the major eating disorders. For the purposes of the present study, we were interested principally in the frequency with which participants engaged in 3 behaviors assessed by the EDE. 1) An objective binge episode refers to eating a large amount of food (as judged by the examiner), during a discrete time (ie any 2-h period), with the patients’ report of loss of control during the episode. (The amount of food eaten must be definitely larger than most people would eat during a similar period of time under similar circumstances.) 2) With subjective binge episodes, patients report loss of control but do not consume an objectively large amount, as judged by the interviewer. (An example would be eating 2 pieces of pizza and feeling "out of control" because only one piece was planned.) 3) An objective overeating episode refers to the consumption of a large amount of food but without the experience of loss of control. (An example would be eating a whole pizza, a large bag of chips, and a pint of potato salad and reporting enjoyment of the eating episode, without loss of control.) Only objective binge episodes count toward the diagnosis of binge-eating disorder (BED). Criteria for BED include at least 2 binge episodes a week for 6 mo, marked distress about this eating, and the absence of purging or other behaviors to compensate for the binge episodes (14). We wished to test the hypothesis that the MR and BDD conditions would be associated with significantly more objective binge episodes and diagnosed cases of BED than would the ND condition. We also examined subjective binge episodes and objective overeating episodes because they may be associated with emotional distress and weight gain, respectively.
The EDE was administered to participants at baseline and weeks 9, 20, 28, 40, and 65 by masters- or doctoral-level practitioners. (This is the schedule on which all measures were administered unless otherwise noted.) Examiners were blind to participants’ treatment conditions. At each assessment, participants described their eating for the previous 28 days, and the examiner queried them to determine the number of objective binge episodes, subjective binges, and objective overeating episodes. Before the study began, all examiners submitted at least 3 audiotapes of practice assessments that were reviewed by expert raters (ie, Christopher Fairburn or Elaine Wilson) to determine that the assessors adhered to the structured protocol and correctly classified the amount of food eaten (ie, objectively large amount) and the occurrence of loss of control. Examiners met regularly throughout the study to review ambiguous diagnostic cases and to maintain calibration. The EDE has been found to have excellent reliability and validity when administered in this manner (37).
Eating-related behaviors
Behavioral and cognitive aspects of eating behavior were measured by the Eating Inventory (38). It assesses 1) cognitive restraint (ie, intentional effort to restrict overall energy intake or consumption of certain foods); 2) disinhibition (ie, loss of control over food intake; and 3) hunger (ie, physical and psychological symptoms of hunger). The study tested the hypothesis that the 2 dieting conditions would be associated with significant increases in disinhibition and hunger compared with the ND group. (The Eating Inventory was not administered at week 65.)
Mood and self-esteem
Mood was assessed by the BDI (25) and self-esteem by the Rosenberg Self-Esteem scale (39). Higher scores on the 2 scales are associated with greater symptoms of depression and poor self-esteem, respectively. We wished to test the hypothesis that the 2 dieting groups would be associated with worsening of mood and self-esteem compared with the ND group.
Body image
Body image dissatisfaction was assessed by the Body Shape Questionnaire (40) and the Body Dysmorphic Disorder Examination-Self-Report (BDDE-SR) (41). Both instruments have excellent reliability and validity; higher scores on each indicate greater dissatisfaction with appearance. (The BBDE-SR was not administered at week 65.)
Statistical analyses
Differences between the 3 treatment conditions in changes in weight at weeks 9, 20, 28, 40, and 65 were assessed by a repeated measures analysis of variance (ANOVA). In cases in which the treatment by time interaction was significant, follow-up comparisons were conducted using Tukey’s HSD, which sets the experiment-wise error rate at P < 0.05, thus, controlling for multiple comparisons. Weight data were analyzed using 3 methods. A completers analysis examined only participants who were present at a given assessment, with no adjustment for missing values. In a last-observation-carried-forward analysis (LOCF), missing data were replaced with the participant’s last measured value. A more conservative intention-to-treat analysis (ITT), described previously (42), also was used in which participants who discontinued treatment were assumed to gain 0.3 kg/mo after leaving the study. This value was selected based on the 1-y weight regain of 3.0 kg observed in trials of group behavior modification (29). The 3 sets of analyses yielded similar statistical conclusions. Thus, only the results of the ITT analysis are reported here.
Changes in the occurrence of objective binge episodes were analyzed by examining the number of days, during each 28-d assessment, that participants were judged to have engaged in objective binge eating. Differences among groups in the number of days were compared using a repeated-measures ANOVA. In addition, the number of participants (ie, cases) in each of the 3 groups who engaged in any objective binge episodes, during each 28-d assessment period, was analyzed using both general estimating equations (GEE) and conditional logit regression. The results, however, for the overall tests would not converge, presumably because of the large number of zero values (ie indicating no cases). Thus, differences among groups were compared using the Fisher’s exact test. Using the Bonferroni adjustment, differences among groups at each time period had to reach P < 0.01 to be considered statistically significant. Subjective binge episodes and objective overeating episodes were analyzed in the same manner. Differences among the 3 treatment groups in changes in appetite, mood, self-esteem and body image were examined using repeated-measures ANOVAs. With all data, except body weight, only treatment completers were included in the repeated-measures ANOVAs. There were not adequate guidelines with these variables for imputing missing values (as there were with data for body weight). ANOVAs were conducted using SPSS (version 11.5; Chicago) and Fisher’s exact test using SAS (version 8.2; Cary, NC). GEE and conditional logit regression were conducted using both SPSS (version 11.5) and STATA (version 8; College Station, TX).
Of the 43 women who were randomized to the BDD group, 37 (ie, 86%) remained in the study at week 20, 30 (70%) at week 40, and 26 (60%) at week 65. Of the 41 MR participants, 37 (90%), 31 (76%), and 28 (68%) remained at weeks 20, 40, and 65, respectively. Corresponding numbers for the 39 participants assigned to ND were 38 (97%), 34 (89%), and 28 (74%), respectively. Chi-square analyses revealed no significant differences among the 3 groups in attrition from therapy. The great majority of dropouts were attributable to scheduling conflicts, dissatisfaction with treatment, or lack of time to meet the study requirements (including the frequent and lengthy assessments). There were no significant differences among groups in the factors associated with participants’ attrition.
RESULTS
Weight
Figure 2 shows that weight changes of participants in the 3 groups were generally consistent with their prescribed caloric intakes. At week 20, women in the MR group lost an average (± SD) of 12.1 ± 6.7% of initial weight, compared with losses of 7.8 ± 6.0% and 0.1 ± 2.4% for the BDD and ND participants, respectively. Weight losses of all 3 groups differed significantly (all: P < 0.001) from each other at this time. At week 40, the MR and BDD participants lost 11.5 ± 8.9% and 8.4 ± 8.7%, respectively, compared with 0.8 ± 3.2% for women in the ND group. Losses of the first 2 groups differed significantly (P < 0.001) from those of the third but not from each other. Women in the MR and BDD groups gained weight from weeks 40 to 65 but concluded the study with mean losses of 8.6 ± 10.0% and 6.3 ± 8.3% of initial weight, respectively, which did not differ significantly from each other. Both groups lost significantly (P < 0.001) more weight than the ND participants who, at week 65, had gained 0.8 ± 3.4% from baseline. (All weight losses reported are based on the intention-to-treat analysis, described in Statistical analyses.)
FIGURE 2.. The figure shows the percentage reduction in initial weight for participants in the meal replacement (MR; x), balanced-deficit diet (BDD; ), and nondieting (ND; ) groups. Data are based on an intention-to-treat analysis described in the Methods section. A repeated-measures ANOVA revealed a significant treatment-by-time interaction (P < 0.001). For any given assessment period, groups with different letters (ie, a, b, and c) differed significantly from each other (P < 0.015), as determined by Tukey’s HSD.
Binge eating
As shown in Figure 3 A, no objective binge episodes were observed in any participants in any of the groups at week 9, the time when women in the MR and BDD conditions dieted the most aggressively (as inferred from their weight losses). Similarly, at week 20, no objective binge episodes were observed in any of the MR participants. A single episode was observed in 2 BDD participants but there were no significant differences among groups in the number of cases at this time. A repeated-measures ANOVA revealed a significant (P < 0.04) treatment by time interaction for differences in the number of days on which patients reported objective binge episodes. Follow-up comparisons, however, revealed no significant differences between groups at week 20. (For the MR, BDD, and ND participants, the mean number of days of binge eating at week 20 was 0.0 ± 0.0, 0.1 ± 0.2, and 0.0 ± 0.0, respectively.)
FIGURE 3.. The percentages of participants in the meal replacement (MR;
Differences, however, among conditions were statistically significant (both: P < 0.003) at week 28, at which time each of 4 participants in the MR group was judged to have had one objective binge episode during the prior 28 d, and a fifth woman had 2 episodes (see Figure 3A ). No episodes were observed in the 2 other groups. MR participants engaged in objective binge episodes an average of 0.2 ± 0.5 d (out of 28 d), compared with means of 0.0 for the 2 other conditions (both: P < 0.015).
At week 40, there were no significant differences among groups in the number of cases of binge eating (see Figure 3A ) or the mean number of days (ie 0.1 ± 0.7, 0.0 ± 0.0, and 0.0 ± 0.0. for the MR, BDD, and ND groups, respectively). Objective binge episodes had remitted at this time in the 4 MR participants who had experienced one episode at week 28. The fifth participant was determined to have had 4 episodes at week 40 (during the previous 28 d). At the follow-up assessment at week 65, there were no significant differences among conditions in either measure of binge-eating status. At no time during treatment or follow-up did any participant in any of the treatment groups meet the diagnostic criteria for BED.
Subjective binge episodes
Before treatment, 67% to 77% of women in the 3 groups were judged to have had no episodes of subjective binge eating in the prior 28 d (see Figure 3B ). On average, participants (across the 3 groups) had engaged in this behavior 1.3 ± 4.2 d of the previous 28 d. At no time were there significant differences among groups in the number of days on which subjective binge episodes were reported. The number of participants in all conditions who experienced subjective binge episodes declined during the first 9 wk of treatment (as did the number of days to a mean of 0.2 ± 0.6 d) but then returned toward baseline levels at week 20. There were no significant differences among groups at anytime during the study in the number of cases of subjective binge episodes. At week 40, 7 participants in the BDD group, compared with only 1 in the ND condition, reported subjective binge episodes but the difference (P < 0.03) did not reach the level required for statistical significance (ie, P < 0.01). The number of cases in the BDD group (at week 40) remained below baseline levels.
Objective overeating episodes
Objective overeating episodes were relatively infrequent in all participants. At baseline they were observed (across the 3 groups) an average of only 0.2 ± 1.0 d (during the prior 28 d). At no time were there significant differences among groups in the number of days on which objective overeating episodes were reported. Cases of these episodes did not increase in any of the groups over the course of the study, and there were no statistically significant differences among conditions, at any time, in the number of individuals who reported objective overeating episodes.
Eating-related behaviors
Contrary to our hypotheses, neither disinhibition nor hunger increased significantly in the MR or BDD conditions during caloric restriction. At no time were there significant differences among the 3 groups on changes in these 2 measures. There was, however, a main effect of time. Collapsing across the 3 conditions, mean scores on disinhibition fell significantly (P < 0.001) from a baseline value of 9.8 ± 3.3 to 6.6 ± 3.2 at week 40. Hunger similarly declined significantly (P < 0.001) from 6.5 ± 3.2 at baseline to 4.2 ± 2.8 at week 40. Figure 4 shows that women in the MR and BDD groups reported increases in dietary restraint at all periods, compared with small decreases in ND participants. An ANOVA revealed that the ND group differed significantly (both: P < 0.001) from both the MR and BDD conditions at all periods. There were, however, no significant differences between these latter 2 groups at any time.
FIGURE 4.. The figure shows changes in restraint, as measured by the Eating Inventory (38) in participants in the meal-replacement (MR; ), balanced-deficit diet (BDD; ), and nondieting (ND; ) groups. A repeated-measures ANOVA revealed a significant treatment-by-time interaction (P < 0.001). Groups with different letters differed significantly from each other (all: P < 0.001), as determined by follow-up comparisons using Tukey’s HSD.
Mood and self-esteem
Symptoms of depression declined in the MR and BDD groups during the 40 wk of treatment (see Figure 5 ). Mean reductions in each of these groups were significantly greater than those in the ND participants at week 20 (both: P < 0.03) and at week 40 (both: P < 0.02), with no differences between the MR and BDD participants. At no time were there significant differences among the 3 groups in changes in self-esteem. There was, however, a main effect of time. Collapsing across the 3 groups, symptoms of poor self-esteem declined significantly (P < 0.02) from a mean baseline value of 17.5 ± 4.7 to 15.6 ± 4.9 at week 40.
FIGURE 5.. The figure presents changes in participants’ scores on the Beck Depression Inventory (25). A repeated-measures ANOVA revealed a significant treatment-by-time interaction (P < 0.001). At weeks 20 and 40, women in the meal replacement (MR;
Body image
Figure 6 shows that participants in all 3 groups reported large reductions in negative body image, as assessed by the Body Shape Questionnaire. At no time were there significant differences in changes among the 3 groups. There was, however, a main effect of time. Collapsing across groups, reductions at all assessment periods were significantly (P < 0.001) different from baseline. The same pattern of results was obtained with the BDDE-SR. There were no statistically significant differences among the 3 groups in changes on this measure. Collapsing across the 3 conditions, however, mean scores on the BDDE-SR fell from a baseline value of 53.2 ± 24.6 to 41.3 ± 25.4 at week 20 (P < 0.001) and remained at 33.2 ± 21.2 at week 40 (P < 0.001).
FIGURE 6.. The figure presents changes in symptoms of body image dissatisfaction as measured by the Body Shape Questionnaire (40). A repeated-measures ANOVA revealed a significant effect of time (P < 0.001) but no treatment-by-time interaction. Thus, body image dissatisfaction declined significantly over time in all groups, but there were no significant differences in changes among women in the meal replacement (MR;
DISCUSSION
This is the first prospective, randomized trial that was designed specifically to assess whether dieting is associated in obese individuals with the precipitation of binge eating and related behavioral complications. We found no evidence during the first 20 wk of treatment that either a 1000 kcal/d diet, which included a liquid MR, or a 1200–1500 kcal/d diet, comprised of conventional foods, was associated with binge eating or other disordered eating. Moreover, neither of these diets was associated, during this time, with increases in hunger, dietary disinhibition, or symptoms of depression. On the contrary, women who consumed either the MR or BDD reported significantly greater reductions in symptoms of depression than did participants in the ND condition. Women in all 3 treatment groups experienced decreases in hunger and disinhibition. The strength of the present findings resides in the selection of a well-defined sample (known before treatment to be free of objective binge episodes and depression), in the randomization of participants to dietary conditions, and in the frequent assessment of eating behavior by blinded examiners who used consistent criteria to judge binge episodes.
Anecdotal observations (43), as well as one research report (24), have suggested that liquid formula, very-low-calorie diets (providing <800 kcal/d) may be associated with binge eating, particularly after patients have terminated severe caloric restriction and resumed consumption of conventional foods. Consistent with this hypothesis, 5 participants in our MR condition (who had been prescribed a 1000 kcal/d diet that included 4 servings/d of a liquid diet) were judged to have experienced 1 or 2 episodes of binge eating in the month before their week 28 assessment. This behavior appeared of little clinical significance in 4 participants, in whom no objective binge episodes were observed at week 40. A fifth participant, however, was found at week 40 to have had 4 such episodes, and her BDI score at this time had increased to 17 (indicative of symptoms of mild depression) from a baseline score of 12. This individual, who had lost 27% of initial weight and achieved a BMI of 23.5, moved overseas and, regrettably, did not complete the assessment at week 65. Neither this nor any other individual reported purging or similar compensatory behavior, as found in persons with bulimia nervosa. It is impossible in the present study to determine whether the isolated cases of binge eating observed at week 28 were associated with the severity of caloric restriction (from week 2 to week 13), the limitation of food choices, the substantial weight loss achieved, or some other factors.
No participant in the MR condition met the criteria for BED at anytime during the study, even if the duration of binge eating required to meet the diagnosis was reduced from 6 mo to 1 mo. This finding, combined with the overall rarity of objective binge episodes (and the general lack of increase in subjective binges or objective overeating episodes), suggests that a 1000 kcal/d MR plan is unlikely to precipitate clinically significant eating disorders in persons who are free of such complications before treatment. This possibility cannot be ruled out, given our small sample size, but seems even more unlikely in persons who consume a 1200–1500 kcal/d diet comprised of conventional foods. At no time did our BDD participants experience significantly more objective binge episodes than did women in the ND group. In the ND group, one participant was judged to have had 2 objective binge episodes at the week 65 assessment, despite her apparently not having dieted at this time. This occurrence is consistent with previous studies of obese individuals with BED that found that 35–50% of subjects reported that they were not dieting at the time of their first binge eating episode (6).
Our conclusions about the generally benign effects of dieting are consistent with those of 3 less intensive studies (17, 20, 44) in which participants lost modest amounts of weight (0.5–5.0 kg). Our conclusions, however, must be qualified by several observations. First, we studied only 123 individuals, 23% of whom did not complete the week 40 assessment and 33% the week 65 follow-up visit. The vast majority of dropouts were free of objective binge eating (and other complications) at the time of their last assessment but we cannot be certain of their ultimate outcome. Second, our participants were selected to be free of objective binge episodes and significant symptoms of depression. We wished to determine whether dieting would precipitate adverse behavioral effects in persons who were free of them before treatment. Additional studies, however, are needed to assess the effects of dieting in persons (including men), who before treatment have a history of binge eating, depression, personality disorders, substance abuse, or self-injurious behavior (45). These may be the very individuals who are vulnerable to experiencing adverse effects of dieting (although at least 2 studies found no worsening of binge eating in persons diagnosed with this condition who were treated with a very-low-calorie diet) (22, 46). Another qualification is that dieters in this study were provided a highly structured and supportive group program that included an exercise intervention. This program undoubtedly facilitated participants’ successful weight reduction. Obese individuals who diet on their own may try to follow sensible guidelines to reduce their calorie intake (by 500 kcal/d) and to increase their physical activity. However, they may not be successful in losing weight and, thus, experience frustration, dejection, and self-blame (47; also J Polivy, unpublished observations, 1991). Such individuals might experience the behavioral and emotional complications that have been reported in restrained eaters, persons who are chronically preoccupied with eating fewer calories (or avoiding certain foods) but are not necessarily successful in these efforts or in losing weight (9, 15, 48). Moreover, as they feel more frustrated or desperate, do-it-yourself dieters could become vulnerable to a host of unhealthy weight loss strategies, including skipping meals or fasting for prolonged periods (44). Thus, large community samples must be examined to obtain the best estimate of the incidence of binge eating in response to self-directed weight loss efforts.
The nondieting approach used in this study served as a control condition against which to judge the effects of caloric restriction. This was an "active" control that received a fully developed therapy (32; also J Polivy, unpublished observations, 1991) that is advocated by many critics of traditional dieting approaches (10, 49-51). We chose this intervention, rather than an inactive control (ie, measurement only) to increase the likelihood of participants remaining in the study and completing the multiple, lengthy assessments. While participants did not lose weight, they did experience improvements in body image and self-esteem, comparable to those of participants who dieted and lost weight. This finding confirms previous reports of the benefits of cognitive therapy for improving these aspects of psychological status in obese individuals, independent of weight loss (32, 34, 51).
In summary, we believe that findings of the adverse effects of dieting in persons of average weight (16) or in those with bulimia nervosa or anorexia nervosa have, in some cases, been inappropriately generalized to overweight and obese individuals who seek to lose weight by caloric restriction. By the same measure, the present findings of the generally benign consequences of dieting in obese individuals should not be misconstrued as a denial of the potential perils of aggressive weight loss efforts in adolescent girls and young women (12). Our findings pertain specifically to overweight and obese adults who seek to lose weight by following an appropriate program of diet, physical activity, and behavior therapy, as recommended by the National Heart, Lung, and Blood Institute expert panel (4). The present findings indicate that concerns about the possible adverse effects of dieting should not dissuade overweight and obese individuals from pursuing weight loss. The health benefits of modest weight loss (and increased physical activity) are compelling, as demonstrated most recently by results of the Diabetes Prevention Program (52).
ACKNOWLEDGMENTS
We thank Christopher Fairburn for consulting on the assessment of binge eating and Christopher Fairburn and Elaine Wilson for training our examiners in the use of the Eating Disorder Examination. Janet Polivy is acknowledged for consulting on the development of the nondieting approach, which borrowed heavily from her manual, Stop Dieting: A Program to Enhance Self-Acceptance. We thank James Rosen for his consultation on improving body image in obese individuals. Myles Faith, Jesse Berlin, and Mark Cary are acknowledged for their statistical consultation and Kirstin Job and Vicki Clark for their editorial assistance. We also thank an anonymous reviewer whose statistical expertise contributed substantially to the final version of the manuscript.
TAW and GDF participated in all aspects of the research which included designing the study, providing treatment, analyzing the data, and writing the manuscript. DBS and RIB also provided treatment and consulted on manuscript preparation. DAA and MG administered the Eating Disorder Examination (EDE), and MG supervised additional EDE assessors. RSS and RVL coordinated the recruitment, assessment, and treatment of participants and were responsible for data collection and management. SP contributed to data management and analysis. TAW, GDF, and RIB have served as consultants to Novartis Nutrition, which contributed the OPTIFAST products used in the study. None of the other authors had a relation with any of the study’s sponsors or contributors.
REFERENCES