Ttulo: Eating disorder risk and the role of clothing in collegiate cheerleaders' body images Autores: Toni M.

Torres-McGehee , Eva V. Monsma , Thomas P. Dompier and Stefanie A. Washburn Ttulo: Journal of Athletic Training. 47.5 (September-October 2012): p541. Tipo de documento: Report DOI: http://dx.doi.org/10.4085/1062-6050-47.5.03 Texto completo: COPYRIGHT 2012 National Athletic Trainers' Association, Inc. http://www.nata.org/journal-of-athletic-training Resumen: Context: With increased media coverage and competitive opportunities, cheerleaders may be facing an increase in eating disorder (ED) prevalence linked to clothing-related body image (BI). Objective: To examine ED risk prevalence, pathogenic weight control behaviors, and variation in clothing-specific BI across position and academic status among collegiate cheerleaders. Design: Cross-sectional study. Setting: National Collegiate Athletic Association Division I and II institutions. Patients or Other Participants: Female collegiate cheerleaders (n = 136, age = 20.4 [+ or -] 1.3 years, height = 160.2 [+ or -] 8.1 cm, weight = 57.2 [+ or -] 8.3 kg). Main Outcome Measure(s): Participants self-reported height, weight, and desired weight and completed the Eating Attitudes Test. Body image perceptions in 3 clothing types (daily clothing, midriff uniform, full uniform) were assessed using sex-based silhouettes (body mass index = 18.3 kg/[m.sup.2] for silhouette 1, 23.1 kg/[m.sup.2] for silhouette 4). Results: The ED risk for cheerleaders was estimated at 33.1%. However, when body mass index was controlled using backward stepwise logistic regression, flyers had greater odds (odds ratio = 4.4, 95% confidence interval = 1.5, 13.2, P = .008) of being at risk compared with bases, but no difference was noted between the base and back-spot positions (odds ratio = 1.9, 95% confidence interval = 0.5, 6.6, P = .333). A main effect of BI perceptions was seen (P < .001), with a significant interaction by clothing type ([F.sub.2,133] = 22.5, P < .001, [[eta].sup.2] = 0.14). Cheerleaders desired to be smaller than their perceived BIs for each clothing type, with the largest difference for midriff uniform (2.6 [+ or -] 0.8 versus 3.7 [+ or -] 0.9), followed by full uniform (2.7 [+ or -] 0.8 versus 3.5 [+ or -] 0.9) and daily clothing (2.8 [+ or -] 0.8 versus 3.5 [+ or -] 0.9). Conclusions: Cheerleaders, especially flyers, appear to be at risk for EDs, with greatest BI dissatisfaction when wearing their most revealing uniforms (ie, midriffs). Universities, colleges, and the national governing bodies of these squads need to focus on preventing eating disorders and BI dissatisfaction and promoting self-esteem.

Key Words: athletes, body image dissatisfaction, self-esteem Texto completo: Although the prevalence of eating disorders among athletes varies, (1-4) collegiate cheerleaders remain an understudied population in the literature on eating disorder risk. However, prevalence rates may be among the highest in athletes given the appearance demands of this aesthetic sport. Previous researchers estimated that 35% of female athletes were at risk for anorexia nervosa and 38% for bulimia nervosa (3) and that 31% of elite females in thin-build sports compared with 5.5% of the control population (5) and 25% of elite female athletes in endurance sports, aesthetic sports, and weight-class sports had clinical eating disorders compared with 9% of the general population. (6) Recently, TorresMcGehee et al (7) estimated that 29.7% of collegiate auxiliary units (dancers, color guard, majorettes) were at risk for eating disorders and reported elevated body image dissatisfaction, suggesting that eating disorder prevalence may be a function of body image. Body image dissatisfaction may be an even greater concern among cheerleaders because of frequent media coverage, regular evaluations of their aesthetic qualities (ie, physiques), and physically demanding training regimes. During the last decade, cheerleading squads have become more dynamic, competitive, and athletic and required greater physical demands and advanced skills (eg, tumbling, building pyramids, tossing). Media attention in the form of television coverage, close-ups, and associated preferences for scanty uniforms may be related to mechanisms underlying body image and eating disorder relationships in cheerleaders. Not only are cheerleaders expected to represent their institutions at athletic events, they are required to coordinate spirit-raising events and maintain academic success. As in most aesthetic sports, the pressures to be thin and to look physically fit (5,7) are prominent in cheerleading because of the subjective evaluation embedded in selection and competition success. Cheerleaders are commonly judged not only on performance but also on overall appearance. Moreover, weight expectations (personal or others') may vary with squad position. For example, a base or a back spot may need to be stronger and, thus, heavier to adequately toss, catch, and hold a flyer, who is likely selected because of small size and may be expected to maintain the lightest weight possible to prevent injury to the base. Bases and back spots (who weigh more) may be constantly compared with flyers (who weigh less) and therefore feel more self-conscious about their bodies. Body image dissatisfaction and eating disorder prevalence likely vary by position, but we know of no researchers who have considered such position-specific differences. Most studies of eating disorders and body image in cheerleaders are limited to adolescent participants (8-11) or contained very small sample sizes of collegiate cheerleaders (eg, Black et al [n = 9], (1) Greenleaf et al In = 1], (2) Reel and Gill [n = 76] (10)) or heterogeneous samples of lean-sport athletes, including cheerleaders], (12) that did not include estimates of eating disorder prevalence. A previous investigation (11) showed that adolescent cheerleaders did not appear to be at higher risk for eating disorders compared with girls in general; however, collegiate cheerleaders were not studied. Although research on collegiate cheerleaders is limited, these females may carry forward the same risk factors for

eating disorder risk as do adolescent cheerleaders or girls in general. This notion was supported by Reel and Gill (10) when they demonstrated that body dissatisfaction predicted eating disorders in a sample of both adolescent and collegiate cheerleaders. Interestingly, 53% of the collegiate sample indicated that revealing team uniforms contributed to weight pressures, (10) suggesting that body image may depend on clothing type. It is reasonable to assume that body image dissatisfaction might be higher among cheerleaders when they are wearing open midriff uniforms compared with daily clothing or full uniforms. Cheerleaders who may be increasingly dissatisfied with their body image in revealing uniforms may be at greater risk for disordered eating and the female athletic triad of low energy availability, menstrual cycle dysfunction, and compromised bone health. (13) Thus, the primary purposes of our study were to estimate eating disorder risk and examine pathogenic weight control behaviors in a sample of collegiate cheerleaders, comparing relative risk by position (ie, bases, flyers, back spots) and academic status. A second purpose was to determine the magnitude of body image dissatisfaction for clothing type (daily clothing, midriff uniform, or full uniform) across position and academic status. Based on the findings of Reel and Gill, (10) we expected body image dissatisfaction to be greatest with wearing of the midriff uniform. METHODS Participants Thirty National Collegiate Athletic Association (NCAA) Division I and 10 Division II cheerleading coaches were contacted for access to female cheerleaders; 24 and 2 coaches, respectively, responded (65% team response rate). This yielded 340 possible cheerleading participants. Of those, 136 cheerleaders (40%) participated in the study (age range = 18 to 23 years, age = 20.4 [+ or -] 1.3 years, height = 160.2 [+ or -] 8.1 cm, weight = 57.2 [+ or -] 8.3 kg). They were categorized by position (54 bases, 61 flyers, 21 back spots) and academic status (48 freshmen, 42 sophomores, 21 juniors, 25 seniors). Instrumentation Personal Demographic and Anthropometric Data. Basic demographic data were collected through a questionnaire that included cheerleading position (ie, base, flyer, back spot) and academic status (ie, freshman, sophomore, junior, senior). Participants self-reported height, current weight, highest weight, lowest weight, and ideal weight. Although the potential for reporting bias exists and females are more likely to underreport their weight, previous authors (14) found that selfreported height and weight are generally valid in younger adults. Eating Attitudes Test. The Eating Attitudes Test (EAT26) was administered to screen for eating disorder characteristics and behaviors. This is a well-validated instrument, with reliability (internal consistency) of [alpha] = .90. (15) The [alpha] coefficient of the current study was .88. Although not diagnostic, the EAT-26 is commonly used as a screening tool to identify early characteristics and behaviors indicating the potential presence of an eating disorder. (15) It includes 3 subscales:

dieting, bulimia, and food preoccupation/oral control. Five supplemental questions identify risky behaviors, such as binge eating; vomiting to control weight or shape; use of laxatives, diet pills, or diuretics to lose or to control weight; exercising more than 60 minutes per day to lose or control weight; and loss of 20 pounds or more in the past 6 months. The first 4 supplemental questions are evaluated on a Likert scale (l = never, 2 = once a month or less, 3 = 2-3 times per month, 4 = once per week, 5 = 2-6 times per week, or 6 = once a day or more); the question about weight loss of 20 pounds or more was answered with yes or no. An individual is categorized as at risk for eating disorder attitudes and behaviors if the associated EAT-26 score is greater than 20 or if she meets the risk criteria for 1 supplemental question. If the EAT-26 score is lower than 20 and the individual does not meet the risk for behavioral criteria on the supplemental questions, then she is considered not at risk for eating disorder characteristics and behaviors. Gender-Specific BMI Figural Stimuli Silhouette (SIL). The Figural Stimuli Survey was used to assess body disturbance based on perceived and desired body images (Figure). (16) In a sample of 16728 white females ranging in age from 18-100 years, Bulik et al (17) extended the work of Stunkard et al (16) by associating specific BMI anchors with each image, thereby enhancing the practical use of the scale. The Figural Stimuli Survey scale (Figure) consists of sex-specific BMI figural stimuli SILs associated with Likert-type ratings of oneself against 9 SILs; each SIL is associated with a number that represents a specific BMI ranging from 18.3-45.4 kg/[m.sup.2] (ie, SIL 1 = 18.3, SIL 2 = 19.3, SIL 3 = 20.9, SIL 4 = 23.1). (17) In previous research, (18) the sex-specific BMI figural stimuli SIL test-retest analysis for females' current body image was r = 0.85 (P < .0001) and for ideal body image was r = 0.82 (P < .0001). Validity coefficients using Pearson r correlations ranged from 0.690.84 for comparisons between perceived BMI values and actual BMI measures. (19) [FIGURE OMITTED] Consistent with earlier investigations, (7,20) we used SIL surveys as a basis of comparison for questions about perceived and desired body image in daily clothing versus uniform (full or midriff). Participants were asked to select an SIL (numbered 1-9) that best represented (a) "how you appear in normal daily clothing (eg, what you wear to school)," (b) "how you would prefer to appear in your normal daily clothing," (c) "how you appear in a midriff uniform," d) "how you would prefer to appear in a midriff uniform," (e) "how you appear in a fulllength uniform," and (f) "how you would prefer to appear in a full-length uniform." Scores were recoded based on the BMI value associated with each score. (17) Self-reported BMI was used to compare each person's actual body size with the associated SIL for the 6 questions. Procedures After we acquired approval from the University of South Carolina Institutional Review Board, participants completed the personal information survey, cheerleading background, Eating Attitudes Test-26, and sex-specific BMI base silhouette questionnaire. (15,16) All surveys were distributed via e-mail through SurveyMonkey.com (Palo Alto, CA). At 10 and 20 days after the initial notification

e-mail was sent, a follow-up reminder e-mail was sent to nonrespondents. The survey was open for a total of 30 days. Data Analysis We used SPSS (version XVII; SPSS Inc, Chicago, IL) for all analyses. The sample size was determined using an [alpha] of .05 and a moderate effect size using the Cohen method. (21) Our a priori sample size estimate indicated that we needed 125 participants. Both univariate and multivariate (logistic regression) analyses were conducted to compare eating disorder risk (at risk or not at risk) as the dichotomous dependent variable. The independent variables were position, educational status, and BMI; BMI was included in the multivariate model because analysis of variance revealed differences in physical measurements across position (Table 1). Height and weight were not included in the multivariate model because of their covariance with BMI. Overall risk and risk by position and educational status were calculated by dividing the number of cheerleaders at risk by the number of cheerleaders in each group. Relative risks with 95% confidence intervals (CIs) were calculated with base position and freshman serving as the reference groups. (22) Backward stepwise logistic regression was used to compare eating disorder risk by position while controlling for educational status and BMI and is reported as odds ratios with 95% CIs. Body image dissatisfaction was examined with the Likert SIL anchor data and a 3 (position: base, flyer, back spot) x 3 (clothing type: SIL daily clothing, SIL midriff uniform, SIL full uniform) x 2 (perceived body image, desired body image) analysis of variance with repeated measures on the last 2 factors. Tukey post hoc analyses were used to examine pairwise comparisons across cheerleading positions. The Mauchly test of sphericity was performed to determine whether a correction factor should be applied. An a priori [alpha] level was set at .05. We provide the BMI based SIL means established by Bulik et al (17) for comparative purposes, but we did not use these in statistical analyses examining body image variation across groups because the uneven intervals between BMI values associated with incremental Likert anchors would inherently inflate the type I error rate. RESULTS Physical measurements (height, weight, BMI, etc) for all cheerleaders and by cheerleading position are reported in Table 1. Differences were evident across position (bases, flyers, back spots) for all weight variables, with Tukey post hoc analyses indicating that flyers were smaller than both back spots and bases for all variables. Height differences were noted among groups; flyers were the shortest, followed by bases, who had higher BMIs than both flyers and back spots (P < .01). Comparison of Eating Disorder Risk and Pathogenic Weight Control Behaviors The overall likelihood of being at risk for eating disorders was 33.1% (95% CI = 25.2%, 41.0%) for all cheerleaders. Flyers were at greatest risk for eating disorders (36.1%, 95% CI = 24.0%, 48.1%), and back spots were at lowest risk (28.6%, 95% CI = 9.25%, 47.9%; Table 2). College seniors had the highest risk (48.0%, 95% CI = 28.4%, 67.6%), and juniors had the lowest risk (19.1%, 95% CI

= 2.3%, 35.84%); however, the cell frequency for juniors was less than 5, making it an unreliable estimate. Comparison across groups revealed no significant relative risks on univariate analysis of categorical variables (Table 2). Yet when controlling for BMI using backward stepwise logistic regression, flyers had greater odds (odds ratio = 4.4, 95% CI = 1.5, 13.2, P = .008) of being at risk than bases, but no difference was seen between the base and back-spot positions (odds ratio = 1.9, 95% CI = 0.5, 6.6, P = .333). Body mass index remained significant (odds ratio = 1.5, 1.2, 1.8, P < .001), indicating a 0.368 increase in the log odds of being at risk for each unit increase in BMI. Academic status was eliminated from the model. Fortyfive participants (33.1%) were classified as at risk for eating disorders based on behaviors (n = 40), EAT-26 subscales alone (n = 0), or both behaviors and EAT-26 subscales (n = 5). With regard to pathogenic weight control behaviors overall, 11.8% (n = 16) reported binge eating at least 2-3 times a month; 9.6% (n = 13) vomited to control weight or shape at least once a month; 19.9% (n = 27) used laxatives, diet pills, or diuretics to control weight at least once a month; 1.5% (n = 2) exercised for more than 60 minutes at least once a day to control weight or shape; and 2.2% (n = 3) reported losing 20 pounds or more in the past 6 months (Table 3). Body Image Repeated-measures analysis of variance indicated a clothing type-by-cheerleading position interaction ([F.sub.2,133] = 13.8, P < .001, [[eta].sup.2] = 0.17). Tukey post hoc tests revealed differences between flyers and bases (P < .001), with the greatest discrepancy between perceived and ideal body images for each clothing type. The body image perception-by-clothing type interaction was significant ([F.sub.2,133] = 22.5, P < .001, [[eta].sup.2] = 0.14): cheerleaders desired to be smaller than their perceived body image for each of the clothing types, with the largest difference for midriff uniform (2.6 [+ or -] 0.8 versus 3.7 [+ or -] 0.9) compared with daily clothing (2.8 [+ or -] 0.8 versus 3.5 [+ or -] 0.9) and full uniform (2.7 [+ or -] 0.8 versus 3.5 [+ or -] 0.9; Table 4). DISCUSSION This study is unique because we estimated the prevalence of eating disorder risk behaviors and examined risk within individual cheerleading positions (base, flyer, back spot) in the largest known sample of collegiate cheerleaders. The prevalence estimate for eating disorder risk for all cheerleaders was 33%, which is consistent with other studies examining athletes in aesthetic sports (1,2,5,6,23) and performance squads. (7,10,11,25) More specifically, Black et al (1) estimated their highest eating disorder prevalence to be among cheerleaders (33%); however, only 9 cheerleaders were included. Eating disorder risk in other aesthetic-sport groups (aerobics, auxiliary performers, cross-country, diving, figure skating, gymnastics, modern dance) ranged from 24% to 50%, (1,2,6) with gymnastics highest at 50% and cheerleading most similar to auxiliary performers (ie, dancers, color guard, and majorettes) at 29.7%. (7) In contrast, a recent study (26) examining dieting attitudes using the EAT26 in female college students (n = 299) revealed a lower percentage (12.9%) of these students at risk for disturbed eating tendencies or behaviors.

After we controlled for BMI and academic status, logistic regression showed that flyers were at higher risk than bases and back spots, indicating that the cheerleaders in this sample were not at equal risk for eating disorders across positions. Flyers were also smaller (weighed less) than back spots and bases, suggesting possible selection bias. Flyers may be either selected or self-selected into these positions based on size and may, therefore, be more at risk for eating disorders, independent of position. Although no differences were evident for the total at-risk sample by academic status, freshmen and seniors had the highest prevalence for eating disorder risk (Table 2). In the multivariate model, flyers and those with higher BMIs were more likely to be at risk for eating disorders; however, the overall incidence was greater than 10%, so the odds ratios may be inflated. (27) Subsequent authors should track eating disorder risk across years in college to better understand contributing mechanisms and pinpoint contextsensitive intervention targets. Pathogenic Behaviors The pathogenic behaviors of the NCAA cheerleaders in our study included misuse of laxatives, diuretics, or other medications and were consistent with recent findings on auxiliary performers (ie, dancers, color guard, majorettes) (7) and equestrian athletes. (20) However, the prevalence in this sample was significantly higher than in other studies of NCAA varsity sport athletes (Table 3), particularly for the use of diet pills, laxatives, or diuretics to lose or control weight. This is especially concerning because cheerleading is not considered an NCAA varsity sport, and so cheerleaders are not protected by NCAA rules regarding the use of dietary supplements and weight loss agents. Varsity Brands, Inc is the parent company of the largest national cheer organization and has partnered with the NCAA to establish risk management guidelines for cheer squads. However, eating disorder risk in these collegiate women may require additional vigilance until the NCAA or other governing bodies implement balanced policies to address this issue. Body Image The trend for body image dissatisfaction in cheerleaders is similar to that for the general female college student population: in general, females possess less body image satisfaction. (28) However, body image in cheerleading may have ties to both social and culturally driven pressures to achieve a certain body shape and contextual demands for thinness to maximize performance. (29) Accordingly, we examined the role of body image dissatisfaction from the perspective of clothing type (daily clothing, midriff uniform, full uniform). Although no differences in body image dissatisfaction were observed across cheerleading position or academic status, our findings were consistent with recent studies (7,20) on collegiate auxiliary dancers and equestrians that used SILs to examine body image dissatisfaction across clothing type. In all samples, desired SILs were significantly smaller than perceived SILs, implying that, like equestrians and auxiliary performers, cheerleaders were dissatisfied with their bodies. Not surprisingly, when clothing type was considered, cheerleaders had the greatest degree of body image dissatisfaction with the midriff uniform (Table 3), confirming Reel and Gill's (10) earlier findings that revealing team uniforms contribute to weight pressures among cheerleaders. The role of revealing uniforms in body image

dissatisfaction is important because uniforms have become increasingly revealing during the past 15 years, likely as a function of media coverage. This added pressure may cultivate cheerleaders' mindsets for unhealthy body comparisons, competitive thinness, and pressures to look "good" (thin), especially in those who appear on national television (and hear that "TV adds 10 pounds"). (30) Subsequent investigators should examine the role of national television appearances more carefully to understand the possible effects of media-related pressures. With the popularity and competitiveness of collegiate cheerleading continually increasing, the prevalence of eating disorder risk and body image dissatisfaction within the sport has the potential to increase as well. Our findings show that collegiate cheerleaders are at risk for eating disorders and exhibit body image dissatisfaction in patterns similar to those of other collegiate nonvarsity sport performers. (7) Understanding how cheerleaders perceive their bodies can have practical implications for their weight loss behaviors and mental status. Our study confirms the need to examine the high percentage of pathogenic behaviors to control or lose weight, independent of eating disorder status, especially in understudied aesthetic populations. In addition, the external pressures on body image dissatisfaction indicate an increased risk for developing eating disordered thoughts and behaviors. An external factor contributing to this increase in body image dissatisfaction was uniform type (eg, midriff or full uniform); however, previous researchers (10,31) have implicated cheerleading coaches as influential in weight loss pressures. Identifying perceived body images from social agents associated with the home (eg, parents and peers) and athletic environment (eg, coaches) should be considered for future research. Limitations Although this study revealed several body image characteristics and underlying mechanisms of eating disorders in cheerleaders, the following limitations should be recognized. First, only female cheerleaders were investigated; with the growth of mixed-sex squads, future researchers need to examine eating disorders and body images in male cheerleaders. Our 40% response rate for an Internet-based survey is good, but it is low for estimating prevalence rates in a population and may result in a biased estimate. Second, the EAT-26 was used to screen for eating attitudes and behaviors. This is a widely used and psychometrically sound instrument, yet it is not an accurate diagnostic instrument by itself. Because we screened for and did not diagnose eating disorder characteristics and behaviors, we cannot definitely conclude that the collegiate cheerleaders classified as at risk actually had eating disorders. Obsessive dieters without morbid concerns and generally disturbed individuals who respond positively on surveys without having significant eating concerns could have also inflated the EAT-26 scores in the absence of a diagnosable eating disorder. (32,33) Given the scoring of the EAT-26, it is also possible to have similar EAT-26 total score mean values for those athletes classified as at risk and not at risk (eg, not at risk with a total EAT-26 score less than 20 but reported as at risk due to answers on the behavioral questions). Finally, silhouettes were used to determine B! dissatisfaction. Although this is only a small snapshot of the construct, some of the longer and more common instruments that include body-related perceptions are fee based or time consuming

(eg, Eating Disorder Inventory-3 and Eating Disorder Evaluation). In addition, these longer instruments can be impractical, especially in nonclinical settings. Thus, these findings should be interpreted with caution. CONCLUSIONS Initiating more preventive actions will help decrease the risk to young female cheerleaders. Currently, cheerleaders are bound only by the college or university rules regarding drug testing, but monitoring them (like other varsity studentathletes) for the use of weight loss supplements permitted by the NCAA may help to decrease the risk of unhealthy weight loss behaviors. The NCAA does acknowledge the importance of early recognition of the female athlete triad (34) and supports integrating screening for indicators. However, as in auxiliary performers, (7) the NCAA does not require a preparticipation physical examination for collegiate cheerleaders. Interestingly, in 2006, NCAA and Varsity Brands, Inc, (35) collaborated to undertake a risk management initiative that includes cheerleading in the NCAA's Catastrophic Injury Insurance Program, which requires cheerleading squads to be supervised by a safety-certified coach or advisor, but they do not require any additional medical personnel (eg, athletic trainer, team physician, dietitian, counselor). Providing collegiate cheerleading squads with medical personnel may help in the early identification of signs and symptoms of eating disorders (including body image dissatisfaction), so that prompt, appropriate referrals can be initiated. In summary, coaches, universities and colleges, and national governing bodies of these squads need to focus on programs for preventing body image dissatisfaction and disordered eating and promoting self-esteem. Targeting the governing bodies for cheerleading (eg, American Association of Cheerleading Coaches & Administrators) when advocating to include education about preventing and recognizing eating disorders in cheerleading and spirit squad members could help to decrease the risk. Implementing policies to address the possible use of banned substances or weight loss agents by the NCAA would protect cheerleaders by holding them to the same standards as athletes competing in other varsity sports and may help limit unhealthy behaviors (eg, taking banned dietary supplements) to control weight or lose weight. Until the NCAA recognizes cheerleading as a varsity sport, cheerleading coaches and the current governing bodies of these squads need to focus on programs for preventing body image dissatisfaction and disordered eating and promoting self-esteem. They should also require medical personnel to oversee screening for eating disorders.

Ttulo: Association of eating behaviors and obesity with psychosocial and familial influences Autores: Stephen L. Brown , Glenn R. Schiraldi and Peggy P. Wrobleski Ttulo: American Journal of Health Education. 40.2 (March-April 2009): p80. Tipo de documento: Report Texto completo: COPYRIGHT 2009 Taylor & Francis Ltd. http://www.tandf.co.uk/journals/ Texto completo:

ABSTRACT Background: Overeating is often attributed to emotions and has been linked to psychological challenges and obesity. Purpose: This study investigated the effect of emotional and external cue eating on obesity and the correlation of emotional and external cue eating with positive and negative psychological factors, as well as early familial eating context. Methods: 483 young adults attending two universities completed instruments measuring obesity, emotional and external cue eating, familial eating patterns, depression, anxiety, stress behaviors and somaticism, optimism, self-esteem, resilience, gratitude, humility, happiness, religiosity, and disordered eating. Results: Disordered eaters (with anorexia, bulimia, purging signs) reported worse mental health and more emotional eating. Gender was the only consistent predictor of obesity and external cue eating. In addition to gender, being offered food for comfort as a child was an important predictor of emotional cue eating. Discussion: More emphasis should be given to familial eating context, particularly the practice of offering children food for comfort, as a potential precursor to young adult emotional eating behavior. Translation to Health Education Practice: Findings point to a potential need to monitor and to train primary caregivers and those supervising young children in other settings regarding the use of food for non-nutritional purposes, and to provide training to children on more constructive methods of coping with strong emotions. BACKGROUND Most of the research on food behaviors in psychology has focused on obesity or eating disorders, with relatively few studies targeting behaviors such as emotional or external cue overeating and their correlates to psychological health and familial influence. (1) Understanding these correlates is important because emotional overeating behavior increases risk for obesity and may potentially be a precursor to more serious disordered eating. Distinguishing emotional or external cues to eat from physiological hunger requires an awareness of the signals that differentiate desire to eat based on emotions or outside stimulus from appetite based on true physical need. The emotional stressors that trigger bouts of emotional eating differ among individuals, although most episodes occur when people are at home by themselves. (2) Common emotional cues to eating include feelings of anger, hopelessness, lack of control, and boredom as well as positive emotions, such as celebration. (3) Additionally, using food to cope with emotional

problems can become a habitual response, which not only may lead to weight-related problems, but also may prevent individuals from learning constructive coping skills for effectively resolving emotional distress. Depression, anxiety, and poor self-esteem are often related to emotional eating behavior. (4) There is some evidence that disordered eating practices might mediate the relationship between depression and obesity, with more severe disordered eating practices like binge eating strongly associated with depression. (5,6) Moreover, dieting frequency has been found to be positively associated with depression and inversely associated with self-esteem. (7) Fisher et al., who studied eating patterns of female adolescents, found that even sub-clinical abnormal eating attitudes were strongly correlated with both low self-esteem and high anxiety. (8) With the exception of self-esteem, there is scant research on the association between emotional eating and positive psychological traits. A study among more than 80,000 adolescents found that in addition to selfesteem, general emotional well being was protective against disordered eating behaviors in both males and females. (9) Although there is no research on the influence of childhood family context on adult eating behavior, some have studied its influence on adolescent eating behavior. For example, among adolescents, high levels of family connectiveness have been associated with lower rates of extreme dieting and other disordered eating behaviors among both males and females, (9,10) and with more frequent breakfast eating among overweight males. (10) Laliberte and his colleagues suggest that eating-disordered individuals are more likely to perceive their families as prioritizing appearance and achievement over other family characteristics. (11) They conclude that an individual engaged in disordered eating practices "reflects that family's attempt to conform to the standards of "self-restraint, success, and physical appearance." Further research into the influence familial factors may have in the etiology of clinical and even sub-clinical disordered eating behavior may prove valuable in future efforts to prevent the development of unhealthy eating practices. PURPOSE Only in recent years, researchers have begun to investigate the etiology of overeating behavior in people with sub-clinical disordered eating practices (i.e., emotional and external cue eating). Perhaps with a better understanding of the etiology of sub-clinical disordered eating practices, we will be better equipped to design prevention programs that identify and alter unhealthy eating habits before they progress into more serious eating disorders. This exploratory study had two purposes. The first was to measure the association of emotional and external cue eating with obesity. We hypothesized that emotional and external cue eating would correlate with and predict Body Mass Index (BMI). The other purpose was to investigate the correlation of emotional and external cue eating with depression, anxiety, self-esteem, other psychological factors not previously studied, and aspects of early familial eating context. We hypothesized that disordered eaters would have higher scores for negative and lower scores for positive psychological traits, that negative psychological factors would correlate positively and positive psychological

factors would correlate negatively with emotional and external cue eating, and that familial emotional eating patterns would correlate with individual eating patterns. METHODS Sample Following human subjects approval, participants were sampled from two large public universities (one in the Midwestern and one in the Mid-Atlantic United States). The researchers solicited participants from students in lower-level (mostly freshmen and sophomores) personal health classes in early Fall 2003. After reading the consent form, students were allowed to complete the instruments in class. Those who did not want to participate were excused from class early without penalty; very few students declined to participate. Of 507 students who began the survey, 483 completed all the self-report instruments. Two-thirds (67%) of participants were Caucasian, 19% were AfricanAmerican, 7% were Asian American, 5% were Hispanic, and 2% were other. Fifty-five percent were female. Instrumentation The entire questionnaire of 200 items took approximately 30-40 minutes to complete and included 13 variables. Obesity. In this study, obesity was approximated by the BMI which is calculated as BMI= [(weight in pounds / (height in inches).sup.2]) X 703. A BMI of 18.5 to 25 is considered normal; a BMI of 25-29 is considered overweight; a BMI of 30 or greater is considered obese; a BMI less than 18.5 is considered underweight. (12) Participants self-reported their heights and weights to facilitate this calculation. Emotional and external cue eating. Two subscales of the Dutch Eating Behavior Questionnaire (DEBQ) measuring overeating triggered by negative emotions (psychosomatic theory) and eating in response to visual and environmental cues to eat were used. (3,13) The DEBQ is widely used in the field of dietary behavior because it has consistently shown high internal consistency, factorial validity, and dimensional stability. (14) Five response choices include: never, seldom, sometimes, often, and very often. In this sample, Cronbach's alphas for the Emotional and External cue eating were .95 and .83 respectively. Familial eating patterns. Four additional items were created to measure family patterns potentially associated with emotional eating: (I) While growing up, did your family show they love or care by offering food?; (2) While growing up, how often were you offered food to comfort you when you were emotionally upset?; (3) Who usually offered you this food, while you were growing up?; and (4) While growing up, how often did this person (or persons) eat for comfort when he or she was emotionally upset? Questions 1, 2, and 4 were patterned after the DEBQ and allowed the same response choices; (15) question 3 offered the response options father, mother, both parents, someone else, or not applicable. Depression. The 20-item Zung Self-Rated Depression Scale (SDS) addresses each of the four most commonly found characteristics of depression: the pervasive effect, the

physiological equivalents, other disturbances, and psychomotor activities. (15) Responses range from 1 (little of the time) to 4 (most of the time). A score of 25 to 49 is considered Normal; 50-59, Mildly Depressed; 60-69, Moderately Depressed, and above 69; Severely Depressed. The scale has been able to discriminate between clinically depressed patients and normal controls. (15) Cronbach's alpha for this sample was .82. Stress behaviors and somaticism. The 49-item Strain Questionnaire was designed to measure frequency of behavioral, cognitive, and somatic stress complaints. (16) The scale has shown high test-retest and internal consistency reliability and correlates with the Beck Depression Inventory (.71). The somatic and behavioral subscales were used in this study. Cronbach's alphas for this sample were .88 and .71 respectively. Anxiety. According to the manual, the Speilberger State/Trait Anxiety Inventory (STAI) is the most widely used measure of anxiety in the world. The Trait subscale consists of 20 items designed to measure general anxiety proneness. The test-retest reliability for the Trait scale for male and female college students over a six month period are .73 and .77 respectively. (17) In a sample of 126 college women, the Trait scale correlated highly with the IPAT Anxiety Scale (.75), and the Manifest Anxiety Scale (.80). (17) Cronbach's alpha for this sample was .91.

Optimism. Optimism, defined as the tendency to believe that one will generally experience good outcomes in life, (18,19) was measured by the revised Life Orientation Test (LOT), a 10-item measure focusing on the assessment of generalized outcome expectancies. (20) The mean for an undergraduate population was 14.3, with a testretest reliability of .79 and a Cronbach's alpha of .76. (20) Cronbach's alpha for this sample was also .76. The scale has shown convergent validity with the Self-Mastery Scale (r=.55) and the Rosenberg Self-Esteem Inventory (r=.54), and discriminant validity with the State-Trait Anxiety Inventory (r=-.59) and the Guilford-Zimmerman Temperament Survey (r=-.50).

Self-Esteem. The Rosenberg Self-Esteem Inventory (RSE) has been used to measure self-esteem in diverse groups of adults and adolescents since 1962. (21) This 10-item, Likert-type scale has been validated in many groups including college students and adults. The RSE has a two-week, test-retest reliability in college students of .85-.88. It also correlates significantly with other selfesteem measures such as the Coopersmith Self-Esteem Inventory and divergently with measures of depression and anxiety. (22) A study among a large group of college students found a mean of 32.3 and standard deviation of 4.8. (23) Cronbach's alpha for this sample was .89. Resilience. The 5-point Resilience Scale has eight items representing acceptance of self and life by indicating adaptability, balance, flexibility, and a balanced perspective on life. (24) The remaining 17 items represent personal competence by measuring selfreliance, independence, determination, invincibility, mastery, resourcefulness, and perseverance. (24) The scale had a Crohbach alpha of .91 among a random sample of older adults. Cronbach's alpha for this sample was also .91. Gratitude. The Gratitude Questionnaire-6 (GQ-6) includes six self-report items designed to assess experiences and expressions of gratefulness and appreciation in daily life, as well as feelings about receiving from others. Items reflect the intensity, frequency, span, and density facets of gratitude. Responses are given on a 5-point Likert-type scale. Cronbach's alpha estimates have ranged from .76 to .84. Cronbach's alpha for this sample was .81. The GQ-6 has shown moderate (.30 to .50) convergent correlations with life satisfaction, vitality, optimism, and hope; and modest (less than .4) discriminant correlations with depression, and anxiety. (25) Humility. Humility was measured using a new scale developed by two of the authors. The instrument contains 16 items measured on a 5-point Likert-type scale from strongly agree to strongly disagree. The measure includes four factor subscales: esteem for others, acceptance of own fallibility, recognition of need for external support, and low demand for recognition. In a large sample of college undergraduates, Cronbach's alpha reliability was .73 and two-week test-retest reliability was .73. It has shown positive (.55) correlation with the GQ-6, the Resilience Scale (.49), the revised LOT (.43) and the RSE (.40); and moderate discriminant correlations with the SDS (-.30) and the STAI (-.40). Happiness. The Happiness Measure (HM) is thought to be the most widely used instrument for measuring happiness in general adult samples. (26) It consists of two, self-reporting items measuring emotional well-being: an 11-point, happiness/ unhappiness scale, and a question asking for the time spent in happy, unhappy, and neutral moods. Religiosity. a revised version of the Duke Religion Index (DRI) included two items assessing organizational religious behavior (e.g. church attendance) using a 5-point scale ranging from "once a year or less" to "2-3 times per week" and three items concerning non-organized religion (e.g. prayer or meditation) and spiritual beliefs (e.g. involvement of religion in all aspects of life) using a 5-point scale ranging from "Definitely Not True" to "Definitely True". (27) The five-item scale has shown a Cronbach's alpha of .75 and moderate correlations with other measures of religiousness such as the Age Universal Religious Scale and the Santa Clara Strength of Religious Faith. Cronbach's alpha for this sample was .87.

Disordered Eating. Three questions adapted from Yanovski (28) were used to screen participants with clinical or sub-clinical eating disorders: (1) During the past 6 months, did you often eat within any two-hour period what most people would regard as an unusually large amount of food, PLUS have the feeling that your eating was out of control?; (2) If yes, did you do anything to counteract the effect of eating a large amount of food, like making yourself vomit, take laxatives, strict dieting, fasting or exercising a lot?; and (3) I am tormented by the idea that I am fat or might gain weight AND/OR if I don't have a specific routine for my daily eating, I'll lose control and I'll gain weight. (28) The first two questions screen for bulimia-like behavior and the third question is used to screen for anorexia-like thoughts or behavior. Analysis Three questions were used to screen participants for disordered eating (i.e., anorexia, bulimia, purging behaviors). Those with any "yes" response were compared to those with all "no" responses using independent t-tests for interval variables and chi-squared tests for categorical variables. To address the study hypotheses regarding sub-clinical emotional and external cue eating and their correlates, those screened as disordered eaters were not included in any additional analyses. Prior to completing analyses of nondisordered cases, all variables were checked for linearity, homoscedasticity, and normality. Two variables, Resilience and BMI had kurtosis greater than 2.0. To preserve the original scale, BMI was not transformed; one outlier, more than four standard deviations from the mean, was removed to correct kurtosis. The kurtosis in Resilience could not be corrected by any transformations and two outliers, more than four standard deviations from the mean, were removed. No other response patterns were apparent for the outliers. Spearman correlations were run among all the variables for participants with no apparent disordered eating; coefficients were corrected for attenuation due to measurement reliability. Next, after confirming low multicollinearity, multivariate linear regression (with variables entered simultaneously) was used to predict Emotional Cue Eating and External Cue Eating from the positive psychological variables, the negative psychological variables, and the interval familial eating variables. Finally, to include the categorical independent variables of gender, race, and the familial eating variable (who offered comfort food), a multivariate logistic regression was conducted. For this model, the dependent variables were dummy coded. For BMI, the NIH cutoff of 25 and above was used. For Emotional Cue Eating and External Cue Eating, the respondents were divided into equal halves (high versus low). RESULTS Two hundred seventy-five participants (57%) answered "yes" to one or more of the questions used to screen for disordered eating. Disordered eaters were more likely to be female and to express depressive, somatic, stress behavior and anxiety symptoms, and to be emotional eaters (Table 1). Additionally, disordered eaters were less likely to have high optimism, self-esteem, resilience, gratitude or humility (Table 1). They were also more likely to say a family member offered food to comfort them while growing up (Table 1). There were no significant differences in BMI or race/ethnicity (Table 1).

Table 2 shows correlations among variables for non-disordered eaters. BMI showed no significant correlation with any other variables. In addition to Emotional Cue Eating, External Cue Eating was moderately correlated with the familial eating patterns of showing love through food (.36) and offering food to comfort (.30). It was slightly correlated with parental emotional eating (.17), anxiety (.25), and low optimism (-.21). Emotional Cue Eating was strongly correlated with having a family member offer food for comfort (.55); moderately correlated with parental Emotional Cue Eating (.41), all the negative psychological measures (.30 to .46), and three positive psychological measures: optimism (-.21), self-esteem (-.34), and resilience (-.40). It was slightly correlated with family showing love through food (.29) and gratitude (.17). Because many of the measures used typically share variance, regression analyses were conducted to isolate unique contribution. Table 3 shows results of simultaneous regressions for Emotional Cue Eating, and External Cue Eating. Because no variables were significantly correlated with BMI, no regression was conducted for BMI. Three variables predicted External Cue Eating in a simultaneous model: anxiety (B=.48), depression (negatively, B=.31), and family showing love through food (B=.27). Two variables predicted Emotional Cue Eating: family offered food to comfort (B=.54) and happiness (positively, B=.21). Because demographic variables, including gender and race/ethnicity, are sometimes associated with eating behaviors and obesity, a logistic regression model was built to include these variables (Table 4). To do this, the outcome variables were dummy coded. Women were much less likely than men in this sample to report a high BMI (AOR=0.1). External Cue Eating was also a significant, but minor predictor of BMI (AOR=0.8). Although women had lower means for External Cue Eating, the differences were not statistically significant in this model. In addition to gender, the primary predictor of External Cue Eating was who offered comfort food. According to the logistic regression, those who were offered food by "both parents" or "someone else" were much more likely to express External Cue Eating than those who responded "not applicable." Anxiety was also a significant, but minor predictor of External Cue Eating (AOR=1.1). There were two important predictors of Emotional Cue Eating in this study: women were four times as likely to be emotional eaters, and those whose families offered food to comfort were nearly three times as likely (AOR=2.6) to be emotional eaters. DISCUSSION In a model including gender and race/ ethnicity, the only significant predictors of emotional eating were gender and being raised in a family that offered food for comfort. Earlier studies have established that women have a greater tendency to eat for emotional reasons. (29) The unique contribution of this study, is the finding that those who report being frequently offered food to comfort them when they were emotionally upset as children were over two-and-a-half times more likely to admit to frequent emotional eating as young adults. However, these young adults were not significantly more likely to report having observed a parent participate in emotional eating. What is unclear from this finding is the nature of these food offerings. For example, did the individual, as a child, initiate these exchanges by explicitly asking for or hinting for comfort food when emotionally upset? It is also unclear whether the parent(s) offered

every person or child in the family food for comfort or whether the interaction was unique to this individual. Further, whether a child transitioned from being offered comfort foods to seeking them autonomously, and the age at which this transition occurred, is also unknown from these preliminary findings. These, and other issues regarding the relationships and environment surrounding the tender of comfort foods to children, need further investigation. In addition to men, in this study, those who were more anxious were also slightly more likely to admit to external cue eating. This study contributes to the understanding of external cue eating by showing a pattern among those who offered food for comfort. Those who were offered food by both parents, or someone other than a parent, were more likely to admit external cue eating as young adults than those who said they were not offered food for comfort as a child. The only consistent predictor of BMI was gender. In this sample of young adults, women were much less likely to be overweight then men. The men in the sample had BMI proportions similar to national averages. (13) Specifically, 55% had BMI scores less than 25, 40% had BMI scores between 25 and 29, and 5% had scores 30 or greater. However, the women in this sample reported BMI rates much lower than national averages. (13) Eighty-two percent reported weights and heights for BMI scores less than 25, 14% had BMI scores between 25 and 29, and 4% had BMI scores 30 or greater. Limitations The finding of no association between the emotional eating patterns, or any of the familial or psychological precursors with BMI may mean that these factors have limited influence on the BMI of nondisordered eaters in this sample of young adults, which is similar to results seen in other samples. (30) An alternative explanation is the possibility that the weight and/or height data, which were volunteered by participants, were misreported. It may also mean that the effects of emotional eating patterns are not yet apparent in samples of young adults, most of who were in their twenties. In addition to self-report data, other limitations include fairly narrow, nonrandom sampling; which limits generalizability and recall from childhood. Based on these findings, we cannot judge the actual frequency of participants' being offered comfort food as children. This level of understanding could be achieved by designing a prospective study with observations of food offerings during childhood. Such a longitudinal approach may also help determine whether emotional eating patterns eventually lead to increases in BMI in later adulthood. However, it is apparent that those with salient memories of these exchanges, or perceptions that these exchanges took place frequently, are much more likely to admit to emotional eating patterns as young adults. Other Findings Similar to earlier studies, (5-8) those screened for disordered eating in this study generally had poorer mental health. This is evidenced by higher scores for depression, anxiety, and stress symptoms, as well as lower scores for optimism, self-esteem, resilience, gratitude, and humility.

One striking finding was not part of the original hypotheses. Forty-four percent of the men and more than two-thirds of the women answered "yes" to at least one of the questions used to screen for disordered eating. This is higher than estimates from the recent replication of the National Comorbidity Study, in which rates for anorexia nervosa, bulimia nervosa, and binge eating disorder were estimated at 0.3%, 0.5%, and 2.0% among all men in the U.S. and 0.9%, 1.5%, and, 2.5% among all women. (31) This is also higher than the DSM estimates that 1-4% of all young women in the U.S. will have one or more of these. (32) Part of the explanation for higher proportions is that the abbreviated method used to assess disordered eating in this study, an affirmative response to any of the three screening questions, likely led to an inflated prevalence compared to more comprehensive measures of disordered eating. However, there may also be other reasons for proportions higher than clinical estimates. For example, more recent research by the American Psychiatric Association estimate that 8% of all women have clinical eating disorders and this estimate does not include binge eating, which is thought to be the most common of the three eating disorders. Another explanation is that rates calculated for all ages are not representative of rates in younger populations, particularly adolescent and young adult rates. Additionally, rates may be even higher among college students. For example, a national poll on college campuses in 2006 found that nearly 20% of respondents admitted to currently suffering with or to having suffered from an eating disorder, although nearly threequarters had never received treatment. (33) In this study, 20% of the men and 29% of the women responded affirmatively to the binge eating question: "During the past 6 months, did you often eat within any two-hour period what most people would regard as an unusually large amount of food, PLUS have the feeling that your eating was out of control? "As expected, a lesser number, 13% of the men and 15% of the women, gave affirmative responses to the purging question: "If yes, did you do anything to counteract the effect of eating a large amount of food, like making yourself vomit, take laxatives, strict dieting, fasting or exercising a lot?" The most dramatic responses were to the question that was intended to screen for anorexia-like thinking: "I am tormented by the idea that I am fat or might gain weight AND/OR if I don't have a specific routine for my daily eating, I'll lose control and I'll gain weight." Twenty-nine percent of the men and 60% of the women marked that they agreed with this statement. An alternative explanation for this finding is that this item is not only measuring anorexia-like thinking, but also excessive concerns about weight gain and/or body image among college students. A large study of adolescent and young adult women found that 48% were dissatisfied with their bodies and that 49% had preoccupations with their weight. (34) In a review of the literature, Bergstorm and Neighbors concluded that heterosexual college students, which account for the majority of this sample, appear to be most susceptible to these body image disturbances, which often stem from incorrect assumptions about the normative beliefs of those of the opposite sex regarding preferred body image. (35) For example, college women perceive that men their age prefer much thinner women and college men perceive that women their age prefer more muscular men than is the case. TRANSLATION TO HEALTH EDUCATION PRACTICE

These preliminary findings point to a need for further research on the influence of childhood family context on adult eating patterns. Specifically, it is important to investigate the possibility that emotional eating may, at least in part, be a behavior learned from parents or other caregivers. If supported, this may point to a greater need for health educators to train caregivers regarding the use of food for non-nutritional purposes among young children. Three types of familial influence were measured in this study: (1) Is food being used as a symbol of love, care or celebration?; (2) Is food being used as means of emotional comfort or coping during difficult times?; and (3) Are caregivers showing a personal example of using food for comfort? Other familial food patterns not measured in this study may also be helpful to address including the practice of forcing children to empty their plates; even when this act causes them emotional distress and the practice of using food as a punishment or reward. These findings may also point to the need for health educators to monitor the use of food as part of a reward or punishment system for children in other settings, such as is common in many elementary schools and youth groups. Many teachers and youth leaders feel that this is their only remaining means for classroom or group discipline and may need to be trained in alternative disciplinary methods. In summary, these preliminary findings may point to a need for health educators to provide training to children and adolescents, directly or through training care givers and teachers, on more constructive methods of coping with strong emotions or stress. These could include guidance in youth development and resilience skills such as stress management, social support, and communication. This paper was submitted to the Journal on June 12, 2008, revised and accepted for publication on September 30, 2008.

Ttulo: Never Enough Ttulo: Publishers Weekly. 259.22 (May 28, 2012): p99. Tipo de documento: Book review, Brief article, Young adult review Texto completo: COPYRIGHT 2012 PWxyz, LLC http://www.publishersweekly.com/ Texto completo:

Never Enough Denise Jaden. Simon Pulse, $9.99 trade paper (40p) ISBN 978-1-4424-2907-9 In her sophomore novel, Jaden (Losing Faith) offers an intimate and enlightened rendering of anorexia and bulimia; The story is told from the perspective of high school junior Loann, who lives in the shadow of her more beautiful and gifted older sister, Claire. Loann's feelings of inadequacy and resentment are overtaken by fear when she discovers that her sister is going to extreme measures to remain thin ("It wasn't just the skinny arms and legs .... Deeper, in her eyes, it was as though she'd lost something of who she used to be"). Loann's horror intensifies when she realizes her parents are incapable of handling Claire's eating disorder and finding a solution. While struggling to understand her sister, Loann is drawn to a boy at school with serious problems of his own. Loann's fight against forces that might be beyond her control is both harrowing and inspiring. While Jaden does not provide simple answers for the problems presented, she dramatically illustrates the importance of speaking out and reaching out. Ages 14up. Agent: Michelle Humphrey, Martha Kaplan Agency. (July)