Binge eating disorder, weight control self-efficacy, and depression in overweight men and women

Transcription

1 (2004) 28, & 2004 Nature Publishing Group All rights reserved /04 $25.00 PAPER Binge eating disorder, weight control self-efficacy, and depression in overweight men and women JA Linde 1 *, RW Jeffery 1, RL Levy 2, NE Sherwood 3, J Utter 1, NP Pronk 3 and RG Boyle 3 1 Division of Epidemiology, University of Minnesota School of Public Health, Minneapolis, MN, USA; 2 University of Washington School of Social Work, Seattle, WA, USA; and 3 HealthPartners, Minneapolis, MN, USA OBJECTIVE: To examine binge eating, depression, weight self-efficacy, and weight control success among obese individuals seeking treatment in a managed care organization. DESIGN: Gender-stratified analyses of associations between binge eating, depression, weight self-efficacy, and weight change, using data from a randomized clinical trial that compared low-cost telephone-based, mail-based, and usual care interventions for weight loss. SUBJECTS: A total of 1632 overweight individuals (460 men, 1172 women; mean age: 50.7 y; mean body mass index: 34.2 kg/ m 2 ) were recruited from a large Midwestern US managed care organization. MEASUREMENTS: Height and weight were measured by study personnel at baseline, and self-reported weight was assessed at 6 and 12 months; self-reported depression status, binge eating, and self-efficacy for weight control were assessed at baseline. RESULTS: Lifetime prevalence rates for depression and probable binge eating disorder were high. Weight self-efficacy was inversely related to weight in both men and women. For women, depression was associated with lower weight self-efficacy and higher body weight. Women reporting depression or lower weight self-efficacy at baseline had less weight loss success at 6 and 12 months. Depression, binge eating disorder, and weight self-efficacy were not significantly associated with weight loss success in men. CONCLUSION: Negative emotional states are highly prevalent and predict poor treatment outcomes, particularly for obese women. As obese women with clinical depression typically are excluded from intervention studies, further research on how to address the intersection of obesity intervention and mood management may be warranted. (2004) 28, doi: /sj.ijo Published online 13 January 2004 Keywords: weight control; self-efficacy; binge eating; depression; treatment issues Introduction Obesity is a major problem in the US because it is highly prevalent and causally linked to a number of adverse health consequences. 1,2 Dramatic increases in the prevalence of obesity over the past 20 y have resulted in a discouraging health dilemma. Over half of US adults are overweight and more than one in four are clinically obese. 3,4 Moreover, although most of these individuals have tried to lose weight, there are no effective treatments for obesity and thus, most have failed to maintain any weight loss they may have achieved. 5 It is understandable, therefore, that increasing *Correspondence: JA Linde, Division of Epidemiology, School of Public Health, University of Minnesota, Twin Cities Campus, 1300 South Second Street, Suite 300, Minneapolis, MN , USA. linde@epi.umn.edu Received 22 April 2003; revised 28 October 2003; accepted 17 November 2003 concerns are being raised about possible psychological morbidity associated with obesity and unsuccessful weight control, such as binge eating and clinical depression. Scientific data on this issue are limited, however, especially with respect to the ability of psychological variables to predict weight loss. The current paper presents the results of analyses examining the relationships among binge eating, depression, self-efficacy, and weight control in a large sample of obese managed care organization members. Binge eating was first characterized as a pattern of overeating episodes followed by feelings of loss of control, culpability, and attempts to restrict eating to lose weight. 6 Estimates of the prevalence of these behaviors are as high as 23 55% in individuals seeking clinical treatment for obesity and 2 3% in community samples To guide further research in this domain, a set of criteria for binge eating disorder (BED) were proposed in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV). 13 Binge

2 eating has been linked with greater degree of overweight in community samples and in clinical samples. 14,15 It has also been associated with poorer weight loss treatment outcomes, 16,17 although small sample sizes and restricted samples have limited the generalizability of these findings. Data are particularly lacking on binge eating and weight loss outcomes in men, who are severely under-represented in most studies. 9,10 Lifetime risk estimates for major depression vary by gender, from 10 to 25% in population samples of women and from 5 to 12% in men. 13 Among obese individuals, symptom prevalence rates for depression approach 16%, with greater relative risk for women than for men. 18,19 Depressive symptoms have been positively associated with obesity and binge eating behaviors in treatment-seeking obese samples. 20,21 Little is known regarding the exact nature of the relationship between depression and weight loss success however, perhaps because clinical depression is a common exclusion criterion in weight loss intervention trials. 22,23 The concept of self-efficacy, or an individual s belief in his or her ability to perform and succeed in challenging situations, has been examined in the context of obesity and weight control in a limited number of prior studies. 24,25 Results of this work have demonstrated that weight-related self-efficacy usually improves following completion of a behavioral weight loss program Greater baseline selfefficacy also tends to predict greater weight loss success in such programs, although not always In addition, previous research has suggested that weight-related selfefficacy is lower in women than in men. 29 Most of these studies have been limited, however, by relatively small samples or restricted samples. 26,28,31 Further examination of these issues in a broader population thus is warranted. Relationships among binge eating and weight self-efficacy or binge eating and depression have been explicated in few studies. Researchers have found lower self-efficacy ratings among treatment-seeking overweight binge eaters as compared with nonbinge eaters, and binge eating severity has shown an inverse correlation with self-efficacy following treatment for BED. 33,34 Positive correlations between binge eating and depression have been found in obese community samples, as well as among those at risk for obesity. 35,36 Investigation of all three variables at once is rare, however. We are aware of only one study to examine all three of these variables in a clinically obese sample; these researchers found greater mean depression and lower mean weight control selfefficacy for binge eaters compared to nonbinge eaters. 26 The current study examined binge eating, depression, and weight control self-efficacy in obese volunteers for a weight loss trial, as well as the ability of these variables to predict weight loss over time. Distinguishing features of the study are the unusually large sample, broad inclusion criteria, and adequate numbers of both men and women to examine the gender differences in observed relationships. These characteristics make this study unique in its examination of the prognostic significance of these variables, especially depression, in relation to weight loss success in the clinically obese. Method Study description and eligibility procedures Data used in this paper were collected at baseline, 6, and 12 months of the Weigh-to-Be project. Weigh-to-Be is a study being conducted in collaboration with a large Minnesota managed care organization (MCO) to evaluate telephoneand mail-based weight loss interventions. 37 Participants in the study were recruited over a 12-month period by mail, flyers, postings to the MCO website, and by physician referral. Recruitment efforts were focused on four staffmodel clinics in the Minneapolis St Paul metropolitan area (two urban clinics, two suburban clinics). Recruitment materials encouraged those interested in the study to call a study telephone number to determine eligibility. Eligibility was evaluated by telephone interview. Criteria for study inclusion were: age 18 y or older, not currently pregnant, lactating, or planning a pregnancy within the next 6 months, and body mass index (BMI) greater than 27 kg/m 2 based on self-reported height and weight. Eligible, interested individuals were scheduled for an appointment at their MCO clinic, at which informed consent was obtained, height and weight were measured, and questionnaires were completed. The University of Minnesota and MCO IRB committees approved the study protocol. Of 2205 individuals scheduled for baseline visits, 1801 (82%) completed the visit, consented, and were enrolled in the study (508 men, 1293 women). Of the 404 individuals who scheduled but did not complete the baseline visit, 340 (84%) did not attend the visit, 60 (15%) attended but declined participation after orientation, and four (1%) were found ineligible upon arrival. 38 Based on available self-report data, baseline noncompleters were significantly younger (45.0 vs 50.7 y, t(2169) ¼ 8.01, Po0.0001) and had a higher mean BMI (33.5 vs 34.2 kg/m 2, t(2184) ¼ 2.11, Po0.05) than baseline completers. There were no significant differences in gender (72.4 vs 71.9% women, w2(1) ¼ 0.017, p ¼ 0.896] or mean weight in kilograms (95.0 vs 95.9 kg, t(2184) ¼ 0.881, P ¼ 0.379) between groups. Once recruited, participants were randomized to one of three conditions (a mail-based weight intervention, a telephone-based weight intervention, or a usual care group) by study coordinators, using identification numbers taken in order from the appointment book. Randomization was carried out in blocks of 15. After randomization, letters were sent to participants to inform them of group status. Weight loss protocols were derived from prior weight loss studies by members of the Weigh-to-Be research team and designed as potentially cost-effective delivery modes for weight loss within an MCO population. Treatment consisted of 10 selfpaced modules that promoted behavioral goal setting and 419

3 420 self-monitoring, decreases in caloric intake, and increases in physical activity. Interventions were designed for delivery by mail (written materials and regular mailings from trained staff to monitor treatment progress) or by telephone (same written materials as mail-based intervention, with the substitution of regular telephone contacts with trained staff to monitor treatment progress). Participants assigned to the usual care group were not offered weight intervention beyond that available through the MCO health plan. At the time of the study, the MCO health plan was offering telephone-based weight counseling, clinic-based advice on selecting a weight control program in the community, and clinic-based group programs provided by health plan staff, all for a $25 fee. Measures Demographic characteristics. Age, gender, educational attainment, ethnicity, and marital status were measured by self-report questionnaire administered at baseline. Height, weight, BMI, and weight change. During the baseline clinic visit, trained research staff measured height and weight. At this time, self-reported weights were also collected. Baseline BMI(kg/m 2 ) was computed from measured height and weight. The project design calls for direct measurement of weight at baseline and 24 months and collection of self-reported weight at all assessment points (baseline, 6, 12,18, and 24 months). The 6- and 12-month weight change were defined by subtracting measured weight (in kg) at baseline from selfreported values at follow-up (converted to kg). Since selfreported and measured weights correlate as highly as in weight treatment-seeking samples (with r ¼ 0.96 for men and r ¼ 0.98 for women in the current sample), and since self-reported weight is the strongest predictor of measured weight, we are confident that our weight change variables are a reasonably accurate assessment of relative weight change Indeed, some research has suggested that self-reports of weight, with coefficients as high as 0.97 in predicting actual weight, may be considered more reliable and valid than self-reports of less objective characteristics such as attitudes or health status. 45 Researchers agree that self-reported weights are extremely accurate indicators of actual weight that are particularly useful in cases when data are reported by telephone or by mail, as was the case at 6 and 43, 44, months in the current study. In a further effort to refine our data treatment relative to weight reporting, we performed all analyses on the full sample of 1801 participants and on a sample of 1632 participants (9.1% sample reduction) whose baseline selfreported vs measured weight discrepancies were less than 15 lb (6.82 kg). Results were highly similar. Next, based on the average discrepancy between self-reported and measured baseline weights, we applied a correction of þ 1.5 kg for men and þ 1.7 kg for women to weight data from the full sample of 1801, to adjust for under-reporting of self-reported weights at 6 and 12 months. This approach adjusted the weights and the magnitude of weight change, but did not affect the magnitude or significance of any of the findings for binge eating, depression status, or weight self-efficacy. In the interest of using the most weight-accurate sample, and given that we see no evidence in the literature to suggest that our self-report psychosocial measures are any less biased than self-reported weights, data from the reduced sample of 1632 are presented below. Binge eating status. BED status was assessed using three items from the Questionnaire on Eating and Weight Patterns, which has been used in community and weight treatment-seeking populations to validate the BED diagnostic criteria for DSM-IV and to determine the presence of BED. 9,10,13 Participants were asked to rate the presence and frequency of overeating and feelings of loss of control over their eating during the past 6 months. Participants were classified as having no bingeing (no periods of overeating or feelings of loss of control), some bingeing (binge episodes and feelings of loss of control, less than 2 days per week), or probable BED (binge episodes and feelings of loss of control at least 2 days per week for the past 6 months). During instrument development and validation procedures, this measure demonstrated a kappa of 0.60 for clinician vs selfreported symptom status, which is comparable with agreement rates for other psychiatric diagnoses. 10 Depression status. Depression status was assessed by two health history items asking participants whether a doctor or medical professional had ever told them that they had depression (history of depression) and whether they were currently taking prescribed medication for depression (depression medication status). Response options were Yes, No, or Don t Know ; responses were dichotomized by collapsing No and Don t Know into one No category. The correlation between the depression status items was high in this sample, f ¼ 0.59 for men and 0.64 for women. Weight control self-efficacy. Weight control self-efficacy was assessed using the 20-item Weight Efficacy Life-style Questionnaire (WEL), a measure judged reliable and valid in obese populations. 25 The WEL provides a total weight self-efficacy score based on the sum of the item scores (total range ¼ 0 180). Higher scores indicate greater self-efficacy. The WEL demonstrated good internal consistency reliability in this sample (Cronbach s coefficient a ¼ 0.95 for men, 0.93 for women). Data analysis. Statistical analyses were conducted using Version 8.2 of the Statistical Analysis System (SAS). 47 All analyses were stratified by gender. The means and frequencies were calculated for demographic, weight, and psychosocial variables (binge eating status, depression medication status, depression history, weight control self-efficacy). The

4 SAS General Linear Models procedure was used to analyze the ability of the psychosocial variables to predict weight change from baseline to 6 months and from baseline to 12 months, with baseline weight and treatment group entered as covariates. Education level and ethnicity were considered as covariates; as initial analyses showed both terms to be nonsignificant in all models (Ps ¼ ), these terms were excluded from the analyses presented here. Psychosocial variables were entered into separate models and leastsquares means were calculated for group weight change comparisons. Participants with missing baseline data were excluded from demographic and correlational analyses, and participants with missing follow-up data (weight outcomes and psychosocial variables) were excluded from general linear models analyses. Tables report effect sizes using the effect size correlation (r Yl ), with the exception of w 2 analyses, for which the contingency coefficient is reported. 48 Results Participant characteristics Table 1 presents demographic information and psychosocial variables, stratified by gender. Participants were primarily White, middle-aged, married, and college educated. BMI values were as high as 56.0 for men and 62.4 for women, and mean values indicate that the sample was quite obese. Men were older on average than women and more likely to be White, married, and college educated. Men reported significantly greater self-efficacy for weight control than women. Women were significantly more likely to engage in some binge eating or to have probable BED, to be on medication for depression at present, and to have a history of depression. Table 1 also presents weight change means at 6 and 12 months by gender and treatment group. From baseline to 6 months, weight changes ranged from to þ kg for men and from to þ kg for women. From baseline to 12 months, weight changes ranged from to þ kg for men and from to þ kg for women. Treatment group differences in weight outcomes are discussed in greater detail elsewhere. 37 Effect sizes for gender differences in demographic characteristics were mostly small to medium (r Yl in the range), although a large effect was observed for gender differences in baseline weight (r Yl approaching 0.50). 48 Correlations between baseline weight, psychiatric status, and weight self-efficacy variables. Table 2 presents correlations between baseline weight and predictor variables, stratified by gender. Owing to the relatively large sample sizes, significant correlations of Po0.01 only are highlighted in bold type. For women, all correlations between baseline weight, WEL score, BED status, depression medication status, and depression history were significant. For men, baseline weight and WEL scores were not significantly associated with depression status variables; nor was BED status significantly associated with depression medication status. Regardless of gender or significance level, the directions of association were the same for all variable sets: self-efficacy was negatively associated with baseline weight and with psychiatric status variables; psychiatric status variables were positively associated with each other and with baseline weight. 421 Table 1 Demographic information and descriptive statistics by gender Men Women Mean (SD) or % Mean (SD) or % w 2 or t P r Yl Age (y) (11.75) (12.37) 6.22 o Weight (kg) (15.40) (17.07) o Body mass index (kg/m 2 ) (4.50) (6.00) Ethnicity (% white) 95% 90% a Marital status (% married) 84% 66% o a Education (% attended college) 82% 76% a WEL Total score (20 items) (30.79) (28.82) 4.82 o Binge eating status a None 79% 73% Some bingeing 12% 14% Probable binge eating 9% 13% Depression medication (% yes) 8% 16% o a Depression history (% yes) 16% 30% o a Weight change to 6 months (kg) 3.13 (5.21) 2.03 (4.29) Control 2.25 (4.50) 1.56 (3.76) Mail 2.76 (4.20) 2.09 (4.79) Phone 4.27 (6.39) 2.50 (4.34) Weight change to 12 months (kg) 3.31 (5.84) 2.40 (5.58) Control 2.41 (4.64) 2.29 (5.30) Mail 3.95 (7.22) 2.40 (5.82) Phone 3.48 (5.11) 2.50 (5.66) WEL ¼ Weight Efficacy Life-style Questionnaire. Weight change values represent unadjusted means. a Contingency coefficient.

5 422 Weight change, psychiatric status, and weight self-efficacy. Tables 3 and 4 present results examining the associations between binge eating, depression (medication status or depression history), weight self-efficacy, and weight change from baseline to 6 months and from baseline to 12 months, stratified by gender. Psychosocial variables were entered into separate equations, with baseline weight and treatment group entered as covariates. For men, no variables were significantly associated with weight change at either 6 or 12 months beyond that predicted by treatment group status or baseline weight. For women, higher scores on the WEL, depression medication status (not on medication), and history of depression (no history) were significantly associated with greater weight loss at 6 months. Not taking depression medication was also significantly associated with greater weight loss at 12 months. For all variables, effect sizes were small, with r Yl approaching Table 2 gender Correlations between baseline weight and predictor variables by WEL BED MED HIST Baseline weight (WT; in kg) Men Women Weight self-efficacy (WEL) Men F Women F BED status Men F Women F Depression medication status (MED) Men F 0.59 Women F 0.64 N ¼ 438 men, 1118 women. HIST ¼ History of depression. Correlations in boldface (rx 0.13 for men; rx 0.08 for women) are significant, Po0.01. To examine further potential differences in the effects of psychosocial factors on weight change after the end of active treatment participation, General Linear Models equations were used to examine associations between binge eating, depression status (medication or history), weight selfefficacy, and weight change at 24-month follow-up. Results for all variables were nonsignificant (Ps ranging from for men, for women; data not shown). Results suggest that these psychosocial factors have more influence on the weight loss process during active treatment than follow-up, during which time participants demonstrated a pattern of weight regain: unadjusted mean weight losses at 24 months were 1.40 kg for men and 1.33 kg for women, as compared to 3.13 and 2.03 kg average losses at 6 months and 3.31 and 2.40 kg average losses at 12 months for men and women, respectively. Based on the statistically significant findings from the bivariate General Linear Models equations at 6 months for women, the WEL total score, depression medication status, and history of depression were entered into a multivariate model to predict weight change from baseline to 6 months. Baseline weight and treatment group were entered as covariates. Initially, a model containing all interaction terms for the psychosocial predictors was tested; as no interaction terms were significant (all Ps4.700), a follow-up model with main effects only was tested. Results of the follow-up model are presented in Table 5. The overall model was statistically significant, although small to moderate in effect (r Yl between 0.10 and 0.30). 48 Both depression medication status (not on medication) and self-efficacy for weight control were significantly associated with greater 6-month weight change. Discussion This study examined binge eating, depression, and weight control self-efficacy as predictors of weight change in Table 3 Associations of weight self-efficacy, binge eating status, and depression medication status with weight change at 6 and 12 months in men Weight change (kg) Baseline to 6 months Baseline to 12 months Scale or status Coefficient or adjusted mean (SE) P r Yl Coefficient or adjusted mean (SE) P r Yl WEL Total score 0.02(0.01) (0.01) Binge eating status No bingeing 3.32 a (0.30) 3.46(0.34) Some bingeing 2.48 a,b (0.76) 3.40(0.84) Probable binge eating 1.28 b (0.97) 1.13(1.18) Depression medication Not on medication 3.09(0.28) 3.45(0.32) On medication 2.92(0.90) 1.82(1.06) History of depression No 3.12(0.30) 3.40(0.33) Yes 2.86(0.65) 2.88(0.76) Ns range from 363 to 371 due to missing values. Unstandardized regression coefficients are presented for the WEL; adjusted means are presented for all other variables. Means are adjusted for baseline weight (kg) and treatment group status (control, mail, or phone). Adjusted means with different superscripts are significantly different, Po0.05.

6 Table 4 Associations of weight self-efficacy, binge eating status, and depression medication status with weight change at 6 and 12 months in women Weight change (kg) 423 Baseline to 6 months Baseline to 12 months Scale or status Coefficient or adjusted mean (SE) P r Yl Coefficient or adjusted mean (SE) P r Yl WEL Total score 0.02(0.01) (0.01) Binge eating status No bingeing 2.16(0.17) 2.48(0.22) Some bingeing 1.91(0.39) 2.84(0.53) Probable binge eating 1.60(0.41) 1.46(0.54) Depression medication Not on medication 2.25 a (0.16) 2.68 a (0.21) On medication 0.98 b (0.36) 0.91 b (0.48) History of depression No 2.26 a (0.17) 2.63(0.23) Yes 1.54 b (0.27) 1.84(0.35) Ns range from 847 to 880 due to missing values. Unstandardized regression coefficients are presented for the WEL; adjusted means are presented for all other variables. Means are adjusted for baseline weight (kg) and treatment group status (control, mail, or phone). Adjusted means with different superscripts are significantly different, Po0.01. Table 5 Multivariate model to predict weight change from baseline to six months in women Scale/status Coefficient or adjusted mean (SE) P r Yl Overall model o WEL Total score 0.02 (0.01) Depression medication Not on medication 2.27 a (0.17) On medication 1.06 b (0.47) History of depression No 2.04 (0.19) Yes 2.14 (0.34) N ¼ 845 due to missing values. Unstandardized regression coefficients are presented for the WEL; adjusted means are presented for all other variables. Means are adjusted for baseline weight (kg) and treatment group status (control, mail, or phone). Adjusted means with different superscripts are significantly different, Po0.05. participants recruited from a large MCO for weight loss intervention. Participants included a large number of men as well as women. Percentages of obese men and women with some bingeing or probable BED (21 and 27%, respectively) were consistent with the range of 23 55% seen in prior studies of obese samples seeking treatment. 7,8 The prevalence of depression (16% of men and 30% of women reporting history of depression, 8% of men and 16% of women on medication) was also similar to the 15.5% symptom prevalence rate among the obese coupled with an elevated risk for women (odds ratio ¼ 1.56, 95% CR ¼ ) found in a previous study of depression risk in obese populations. 19 The mean values for weight control self-efficacy were consistent with the range of means reported for multiple weight control samples in the WEL validation study as well. 25 With regard to gender differences, current results were comparable with previous results. Women reported greater binge eating, greater depression, and lower weight control self-efficacy than men. 9,13,18,29 The consistency of these results with previous findings suggests that this sample is comparable to other obese, treatmentseeking populations of men and women. The single most unique finding in this study was that depression predicted poorer weight loss in women. Women who reported taking medication at the time of the study (which may be considered as a proxy measure for current depression status) lost less than half as much weight as women not taking medication for depression, both from baseline to 6 months and from baseline to 12 months. Although the statistical effects were small, they were significant. Clinician wisdom and epidemiologic studies indicate that psychological comorbidities are very common in these women and it has been suggested that depression makes treatment more difficult. As weight loss trials typically have excluded depressed women, however, the extent of this effect has not been previously well documented. Our findings underscore the potentially important effects of concurrent depression on weight loss success in women and suggest that attention to depressive symptomatology may be needed for many obese women. Whether the observed results were due to negative effects of depression on weight loss success or to medication side effects is unclear, and our measure of depression was not as comprehensive as might have been obtained by a more extensive evaluation (eg by depression screening instruments such as the Center for Epidemiological Studies Depression questionnaire 49 or by interview). Nevertheless, the issue clearly seems to merit further study. Another unique aspect of this study is its ability to examine psychosocial correlates of obesity in a large sample of men as well as women. Interestingly, psychosocial measures used in this study (binge eating status, depression medication status, weight control self-efficacy) did not significantly predict weight loss success in men. These measures showed much greater predictive ability for women.

7 424 Depression (based on history as well as medication status) in women was associated with decreased self-efficacy in all weight control domains, with greater likelihood of BED status, and with greater baseline weight. In addition, from baseline to 6 months in the weight loss program, the WEL total self-efficacy score significantly predicted weight change for women. Although the directions of observation in men were similar to those seen in women, these findings add credibility to the speculations of some researchers that the links between obesity and psychosocial factors from female samples may not adequately model these links in men. 20 Fewer men than women seek treatment for obesity and it is reasonable to believe that their reasons for doing so, and their psychosocial reactions to the experience, may be different. Physical health status variables may provide one direction for investigation. Recent studies have suggested that men, particularly those of middle age (as sampled in the present study), cite the beneficial health effects of weight loss as a primary motivating factor when entering weight loss programs or when considering factors that might lead them to seek treatment. 50,51 Some apparent gender differences in the correlation and regression results in the study may have been driven by sample size, as the sample was comprised of approximately 2.5 times as many women as men. For example, the regression coefficient value for Total WEL score as a predictor of weight change from baseline to 6 months (B ¼ 0.02) was identical for men and women, although only statistically significant for women. In addition, the correlation between BED and depression medication status was statistically significant for women only, although the correlations for men and women were not significantly different when compared by Fisher s r to z transformation. Furthermore, effect sizes were small regardless of gender, with r Yl values clustering around 0.10 for all variables in regression analyses. 48 Future studies may benefit from attempts to equate sample sizes by gender in order to improve the literature base in this area as it pertains to men, as well as for statistical purposes. Some limitations in the present study have been described above. Another limitation of the present study is its restricted sampling range. Although the present sample was consistent with previous samples of obese individuals in terms of binge eating status, depression status, and weight self-efficacy, 7,8,18,19,25 current participants were mostly White, well educated and married members of a managed care organization in the Midwestern US. However, the prevalence of obesity in the US is greater among individuals of Black or Hispanic ethnicity and among individuals of lower socioeconomic status. 52 Consequently, future efforts should address this issue of great concern by sampling the broadest possible segment of the US population. This study also demonstrates that negative emotional states are highly prevalent among obese treatment seekers and predict poor treatment outcomes, particularly for obese women. As clinical depression is a common exclusion criterion in weight loss intervention trials, these findings raise the question of whether currently available obesity treatment protocols adequately address the needs of the subset of obese individuals who manifest symptoms of depression or who are currently taking medication for depression; further research in this arena may be warranted. Acknowledgements This research was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant 1R01-DK A1. We thank anonymous reviewers for their comments on an earlier draft of this manuscript. References 1 Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL. Overweight and obesity in the United States: prevalence and trends, Int J Obes Relat Metab Disord 1998; 22: Must A, Spadano J, Coakley EH, Field AE, Colditz G, Dietz WH. The disease burden associated with overweight and obesity. JAMA (1999); 282: Kuczmarski RJ, Flegal KM. Criteria for definition of overweight in transition: background and recommendations for the United States. Am J Clin Nutr 2000; 72: Mokdad AH, Serdula MK, Dietz WH, Bowman BA, Marks JS, Koplan JP. The spread of the obesity epidemic in the United States, JAMA 1999; 282: Jeffery RW, Drewnowski A, Epstein LH, Stunkard AJ, Wilson GT, Wing RR, Hill DR. Long-term maintenance of weight loss: current status. Health Psychol 2000; 19 (Suppl 1): Stunkard AJ. Eating patterns and obesity. Psychiatr Q 1959; 33: Gormally J, Black S, Daston S, Rardin D. The assessment of binge eating severity among obese persons. Addict Behav 1982; 7: Marcus MD, Wing RR, Lamparski DM. Binge eating and dietary restraint in obese patients. Addict Behav 1985; 10: Spitzer RL, Devlin MJ, Walsh BT, Hasin D, Wing R, Marcus M, Stunkard AJ, Wadden T, Yanowski S, Agras S, Mitchell J, Nonas C. Binge eating disorder: a multisite field trial of the diagnostic criteria. Int J Eat Disord 1992; 11: Spitzer RL, Yanovski S, Wadden T, Wing R, Marcus MD, Stunkard AJ, Devlin M, Mitchell J, Hasin D, Horne RL. Binge eating disorder: its further validation in a multisite study. Int J Eat Disord 1993; 13: Fairburn CG, Hay PJ, Welch SL. Binge eating and bulimia nervosa: distribution and determinants. In: Fairburn CG, Wilson GT (eds). Binge eating: nature, assessment, and treatment. Guilford Press: New York; pp Robertson DN, Palmer RL. The prevalence and correlates of binge eating in a British community sample of women with a history of obesity. Int J Eat Disord 1997; 22: American Psychiatric Association. Diagnostic and statistical manual of mental disorders, 4th edn. APA: Washington, DC; Smith DE, Marcus MD, Lewis CE, Fitzgibbon M, Schreiner P. Prevalence of binge eating disorder, obesity, and depression in a biracial cohort of young adults. Ann Behav Med 1998; 20: Telch CF, Agras WS, Rossiter EM. Binge eating increases with increasing adiposity. Int J Eat Disord 1988; 7: Agras WS, Telch CF, Arnow B, Eldredge K, Marnell M. One-year follow-up of cognitive-behavioral therapy for obese individuals with binge eating disorder. J Consult Clin Psychol 1997; 65:

8 17 Sherwood NE, Jeffery RW, Wing RR. Binge status as a predictor of weight loss treatment outcome. Int J Obes Relat Metab Disord 1999; 23: Carpenter KM, Hasin DS, Allison DB, Faith MS. Relationships between obesity and DSM-IV major depressive disorder, suicide ideation, and suicide attempts: results from a general population study. Am J Public Health 2000; 90: Roberts RE, Kaplan GA, Shema SJ, Strawbridge WJ. Are the obese at greater risk for depression? Am J Epidemiol 2000; 152: Musante GJ, Costanzo PR, Friedman KE. The comorbidity of depression and eating dysregulation processes in a diet-seeking obese population: a matter of gender specificity. Int J Eat Disord 1998; 23: Wadden TA, Stunkard AJ. Psychopathology and obesity. Ann NY Acad Sci 1987; 499: Andersen RE, Wadden TA, Bartlett SJ, Zemel B, Verde TJ, Franckowiak SC. Effects of lifestyle activity vs structured aerobic exercise in obese women: a randomized trial. JAMA 1999; 281: Tate DF, Wing RR, Winett RA. Using internet technology to deliver a behavioral weight loss program. JAMA 2001; 285: Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev 1977; 84: Clark MM, Abrams DB, Niaura RS, Eaton CA, Rossi JS. Self-efficacy in weight management. J Consult Clin Psychol 1991; 59: Cargill BR, Clark MM, Pera V, Niaura RS, Abrams DB. Binge eating, body image, depression, and self-efficacy in an obese clinical population. Obes Res 1999; 7: Dennis KE, Goldberg AP. Weight control self-efficacy types and transitions affect weight-loss outcomes in obese women. Addict Behav 1996; 21: Bernier M, Avard J. Self-efficacy, outcome, and attrition in a weight-reduction program. Cognit Ther Res 1986; 10: Forster JL, Jeffery RW. Gender differences related to weight history, eating patterns, efficacy expectations, self-esteem, and weight loss among participants in a weight reduction program. Addict Behav 1986; 11: Prochaska JO, Norcross JC, Fowler JL, Follick MJ, Abrams DB. Attendance and outcome in a worksite weight control program: processes and stages of change as process and predictor variables. Addict Behav 1992; 17: Stotland S, Zuroff DC. Relations between multiple measures of dieting self-efficacy and weight change in a behavioral weight control program. Behav Ther 1991; 22: Fontaine KR, Cheskin LJ. Self-efficacy, attendance, and weight loss in obesity treatment. Addict Behav 1997; 22: Miller PM, Watkins JA, Sargent RG, Rickert EJ. Self-efficacy in overweight individuals with binge eating disorder. Obes Res 1999; 7: Goodrick GK, Pendleton VR, Kimball KT, Poston WSC, Reeves RS, Foreyt JP. Binge eating severity, self-concept, dieting self-efficacy and social support during treatment of binge eating disorder. Int J Eat Disord 1999; 26: Fairburn CG, Doll HA, Welch SL, Hay PJ, Davies BA, O Connor ME. Risk factors for binge eating disorder: a community-based, case control study. Arch Gen Psychiatry 1998; 55: Marcus MD, Wing RR, Ewing L, Kern E, Gooding W, McDermott M. Psychiatric disorders among obese binge eaters. Int J Eat Disord 1990; 9: Jeffery RW, Sherwood NE, Brelje K, Pronk NP, Boyle RG, Boucher JL, Hase K. Mail and phone interventions for weight loss in a managed-care setting: Weigh-to-Be one-year outcomes. Int J Obes Relat Metab Disord 2003; 27: Pronk NP, Boucher JL, Gehling E, Boyle RG, Jeffery RW. A platform for population-based weight management: description of a health plan-based integrated systems approach. Am J Managed Care 2002; 8: Masheb RM, Grilo CM. Accuracy of self-reported weight in patients with binge eating disorder. Int J Eat Disord 2000; 29: Palta M, Prineas RJ, Berman R, Hannan P. Comparison of selfreported and measured height and weight. Am J Epidemiol 1982; 115: Tell GS, Jeffery RW, Kramer FM, Snell MK. Can self-reported body weight be used to evaluate long-term follow-up of a weight-loss program? J Am Diet Assoc 1987; 87: Raciti MA, Murphy J, Ward K, Leshine P, Fowler J, Abrams D, Follick M. Self-reported weight at the worksite: a validation study. Proceedings of the seventh annual meeting of the Society of Behavioral Medicine, March 1986, San Francisco, California. 43 Jeffery RW. Bias in reported body weight as a function of education, occupation, health and weight concern. Addict Behav 1996; 21: Stunkard AJ, Albaum JM. The accuracy of self-reported weights. Am J Clin Nutr 1981; 34: Stewart AL. The reliability and validity of self-reported weight and height. J Chron Dis 1982; 35: Pirie P, Jacobs D, Jeffery RW, Hannan P. Distortion in self-reported height and weight data. J Am Diet Assoc 1981; 78: SAS Institute. SAS OnlineDoc, version 8, SAS: Cary, NC Cohen J. Statistical power analysis for the behavioral sciences. 2nd edn. Lawrence Erlbaum Associates: Hillsdale, NJ; Santor DA, Coyne JC. Shortening the CESD to improve its ability to detect cases of depression. Psychol Assess 1997; 9: Hankey CR, Leslie WS, Lean MEJ. Why lose weight? Reasons for seeking weight loss by overweight but otherwise healthy men. Int J Obes Relat Metab Disord 2002; 26: Wolfe BL, Smith JE. Different strokes for different folks: why overweight men do not seek weight loss treatment. Eat Disord 2002; 10: Paeratakul S, Lovejoy JC, Ryan DH, Bray GA. The relation of gender, race and socioeconomic status to obesity and obesity comorbidities in a sample of U.S. adults. Int J Obes Relat Metab Disord 2002; 26: