Binge eating prevalence according to obesity degrees and associated factors in women

Objective: Investigate binge eating (BE) prevalence in women according to the obesity degree and assess the associated factors. Methods: Cross-sectional study with female adults presenting body mass index (BMI) ≥ 35 kg/m2. The analyzed variables were: sociodemographics, health status, obesity history, lifestyle, eating behavior and obesity degree. In order to analyse BE it was used the Binge Eating Scale (BES), which is considered positive when BES ≥ 18 points. Prevalence and prevalence ratios (PR) were calculated with confidence intervals (CI) of 95%. Multivariate analysis was carried out using Poisson regression. Results: BE prevalence was 53.2%, and the prevalence in super superobese women (BMI ≥ 60 kg/m2) was 75%. After multivariate analysis, associations were observed between the age group 40-49 years old (PR = 2.0; 95% CI = 1.2-3.4) and the “snacking habit” (PR = 1.9; 95% CI = 1.2-2.9). Conclusion: The prevalence of BE in severe obese women was high. Association with the “snacking habit” can be a BE marker that should be monitored in the severely obese individuals that fit this profile.


INTRODUCTION
Obesity is considered a serious public health problem that affects approximately 500 million adults worldwide 1 .This illness afflicts different age groups, all social levels, both genders and its prevalence has been increasing significantly in recent decades 2 .In Brazil, 49% of the adult population are overweight and 14.8% are obese, according to the Family Budget Survey (FBS) 2008-2009 3 .Besides, severe obesity incidence has increased and it represents between 0.5 and 1% of the adult Brazilian population 4 .
Due the fact that obesity is a chronic disease with multifactorial etiology 2 ; obese individuals present emotional alterations that can be classified either as determining factors or as a consequence of obesity 5 .Obese individuals can show psychological suffering related to social discrimination, self-image depreciation, insecurity, and a general feeling of failure related to weight loss.Women are more exposed to media and western society demands for a beauty ideal, related to slenderness, which can generate emotional stress 6,7 .
Bariatric surgery is a treatment option, widely used in severe obesity cases 8,9 .Psychological disorders presence is commonly an exclusion criteria for bariatric surgery candidates; however, due to lack of consensus in the scientific literature, these are not considered as a formal counter indication for this procedure 10 .Nevertheless, psychological and psychiatric monitoring are recommended in the pre-and post-surgery periods, as bariatric surgery candidates usually demonstrate association with psychiatric disorders and the proper treatment will promote better prognostic after the surgery 9,10 .
Due to the close relationship between binge eating (BE) and severe obesity, it is important to know the magnitude of this problem and its associated factors in obese patients that seek treatment, mainly because BE is a risk factor for comorbidities, such as: psychiatric disorders, and also hinders adherence to dietotherapic treatments 11,12 .This study aimed to investigate BE prevalence in women according to the obesity degree as well as to evaluate the associated factors.

METHODS
It was used a cross-sectional study carried out at the Severe Obesity Nutrition Ambulatory, at the Federal University of Goiás School Hospital (in Portuguese: Ambulatório de Nutrição em Obesidade Grave -ANOG).The invitation to participate of the research occurred sequentially to all patients that had appointments at ANOG; however inclusion in the research was not sequential.Some patients did not fulfill the eligibility criteria and some did not sign the informed consent form (ICF).
The inclusion criteria considered adult women (20-59 years old), who presented obesity degree II or superior, i.e., body mass index (BMI) ≥ 35 kg/m 2 , and sought nutritional treatment at ANOG and/or were bariatric surgery candidates.It was included in the research the women who fit the criteria above and accepted to be part of the study as well as signed the ICF.The exclusion criteria were: pregnant women; women who were breastfeeding; women with special needs; and post-bariatric surgery patients.
Data were collected between February, 2008 and June, 2012, during the first consultation.It was done through standardized pre-tested anamnesis application and anthropometric data collection by nutritionists or licensed nutrition academics.During consultation, the Binge Eating Scale (BES) 13 was applied.At this moment patients chose the affirmatives that best described their feelings regarding eating behavior.In order to avoid comprehension difficulties, as most patients verified low schooling level, it was done a simple instrument reading for all patients, following the literature recommendation 13 .
BES is a translated and validated Likert scale for the Brazilian population, considered an investigative instrument that allows the evaluation of symptoms severity during binge eating episodes as well as tracking binge eating disorders (BED) occurrence in obese individuals during weight loss treatments 13,14 .BES classifies BE severity degrees, categorized by the following scores: ≤ 17 points no BE, 18 to 26 points moderate BE, and score ≥ 27 severe BE 13 .BE was the outcome variable in this study, defined as a BES score ≥ 18 points.
A multidisciplinary team accompanied the patients in this research, but it did not include a psychiatrist.Therefore, a BE-tracking instrument was selected in order to achieve the research goals and the individuals monitored with BE were forwarded to psychiatric follow-ups.The term BE is used to describe the feeding behaviour characterized by moments of extreme food ingestion in short periods of time, followed by a loss of control feeling by the individual, not followed by compensatory methods 14,15 .
The studied variables included: sociodemographical data (age, marital status, schooling, and family income); health conditions (actual comorbidities and weight loss medication use); obesity clinical history (obesity onset, obesity in the family, and previous weight loss treatments); lifestyle (physical activities practice, use of tobacco or alcohol); eating behavior (number of meals per day, "snacking habit", Binge Eating -BE), and anthropometric data (weight, height, and BMI).
Patients were questioned about: age in full years, which was then categorized in 20-29 years old, 30-39 years old, 40-49 years old, 50-59 years old; marital status (with partner/without partner), schooling years, which was then categorized in less than 9 years and ≥ 9 years; and family income calculated by minimum wage (MW) * .It was categorized as < 1 MW/ 1 to 2 MW/ > 2 MW.Current comorbidities presence (yes/no) and weight loss medication use were confirmed by medical prescriptions for * In 2012 the minimum wage in Brazil was R$ 622.00.
Binge eating and associated factors continuous or mandatory medication and/or medical records.Regarding weight loss medication, anorexigen and psychiatric medications were considered, and to this second variable were included anxyolitics and antidepressives (yes/no).
Other variables studied included: obesity onset (childhood/adolescence/post-gestation/adulthood); obesity in family (yes/no); previous weight loss treatment (self-initiated or with a health professional help); physical activity practice (regular, irregular or none).In this category were considered as regular at least 150 minutes per week of moderate-intense aerobic activity 16 ; use of tobacco (yes/no) for which ex-smokers were counted as smokers; alcoholic beverages consumption (yes/no); number of meals per day (< 4/≥ 4 meals/ day) and if "snacking" was a habit.For the "snacking habit", the question asked was: "Do you have the habit of eating any food or drinking beverages, with the exception of water, in an unplanned manner and outside established periods for main meals and snacks?If yes, do you consider this a regular habit, as part of your routine at least once a day?" (yes/no) 17 .
The anthropometric variables evaluated were weight and height and later it was calculated the BMI.Platform-type mechanical scales, with a maximum capacity of 300 kg and 100 g precision, were used to determine weight.Height measurement was carried by using a stadiometer, coupled to the scales, with a 0.1 cm precision.
Before weight measurements, the scales were calibrated, and then the surveyee was positioned in the center of the scale, in a firm position, barefoot and with the minimum amount of clothes possible, with arms parallel to the body.Weight measurements were verified in the front of the scale 17 .It is important to highlight that the examiners participated of preliminary training.They were qualified to correctly carry out the anthropometric evaluation procedure in order to guarantee higher data reliability.
To classify obesity it was used the criteria from the World Health Organization (WHO) as well as from the Brazilian Association of Obesity Study and Metabolic Syndrome (in Portuguese: Associação Brasileira para o Estudo da Obesidade e da Síndrome Metabólica -Abeso).It was considered a BMI between 35 and 39.9 kg/m 2 obesity degree II and ≥ 40.0 kg/m 2 as degree III 9,18 .It was also considered the criteria defined by the American Society for Metabolic and Bariatric Surgery (ASMBS), which considers superobesity if the BMI is between 50 and 59.9 kg/m 2 and super superobesity if the BMI is ≥ 60 kg/m 2 19 .
Database structure was built using the Epi-info software, version 3.5.3.Data were typed in double entries and only after data validation entry the statistical analysis was carried out using the Stata software, version 8.0.
Because this research is part of a wider study 21 , the number of patients for the study was calculated a posteriori through the Epi-Info 6.0 software.It was selected the crosssectional studies option and the Statcalc tool.The parameters utilized were: 80% power, confidence level 95%.Ex-position variables considered were "snacking habit" with a 40% prevalence in non-exposed and 1:1 ratio (one exposed to one non-exposed) and weight loss medication use with a 50% prevalence among non-exposed and 1:5 ratio (1 exposed to 5 non-exposed).As a result, it would be necessary 90 patients to analyze the association between medicine and 92 patients to analyse "snacking habit".Thereby the sample size considered was enough to carry out the research.
Statistical analysis included BE prevalence and prevalence ratios measurement (PR), with corresponding 95% confidence intervals (CI 95%).Pearsons's Chi-Square Test or Fisher's Exact Test were used to verify the association between BE and the studied variables.Variables that presented p < 0.20 in bivariate analysis were selected for multivariate analysis through a Poisson regression, with 5% significance level.
This study was approved by the Federal University of Goiás School Hospital Ethics Committee (in Portuguese: Comitê de Ética do Hospital das Clínicas da Universidade Federal de Goiás -CEP/HC/UFG).
Degree II obese Degree III obese Superobese Super superobese

Binge eating and associated factors
The majority of patients (69.9%) were either married or in a consensual relationship, 51.6% had at least 9 years of schooling, and 43.9% achieved a family income between 1 and 2 minimum wages (Table 1).
The highest BE prevalence was verified in women between 40 and 49 years old (76.2%), with less than 9 years of schooling (60.0%), family income between one and two minimum wages per month (60.0%), obesity onset in their adolescence (60.0%), that had undergone through pre vious weight loss self-treatment (62.5%), who eat four or more meals per day (62.0%), and that present "snacking habit" (64.4%).This last variable was the only one that showed significant association (p = 0.020) (Table 2).
The current comorbidity that presented highest BE prevalence was dyslipidemia (69.2%).Regarding weight loss medication use, BE frequency in those making use of psychiatric medication was higher in relation to the use of anorexigens.Only one patient reported the use of anxiolytics, but did not present BE.BE was more frequent in sedentary women or in those practicing irregular physical activity (55.8%), in non- smokers (54.4%), and those who did not consume alcohol (55.0%), but no significant association was observed (Table 2).Current comorbidity that showed more BE prevalence was dyslipidemia (69.2%).About weight loss medication, BE frequency in those women who used psychiatric medication was bigger when compared to women who used anorexigen medication.Only one patient mentioned using anxiolytics, but did not present BE.BE was more frequent in women that practiced physical activity in an irregular way or were sedentary (55.8%), in those who were non-smokers (54.5%) and those who did not use alcoholic beverages (55.0%).However, it was not noticed significant association (Table 2).
The variables included in multivariate analysis were: age (p = 0.082), number of meals per day (p = 0.018), "snacking habit" (p = 0.020), and dyslipidemia (p = 0.185).Association with BE was maintained only in the age group of 40-49 years old, with BE prevalence twice as higher in relation to the age group of 30-39 years old, as well as with the "snacking habit" with a BE prevalence 1.9 times higher regarding those that did not present this behavior (Table 3).

DISCUSSION
This study detected a high BE prevalence (53.2%) in women suffering from severe obesity, who sought nutritional treatment at ANOG and part of them were bariatric surgery candidates.High BE prevalence in severe obese individuals was also observed in a study encompassing patients of both genders, in pre-operatory bariatric surgery conditions, with 44% prevalence 22 and in a research including women in pre-operatory bariatric surgery treatment, with 77.3% prevalence 23 .
A research that used the same instrument to track BE in obese patients waiting for bariatric surgery observed a 56.7% BE prevalence.From those, 25.4% were considered moderate BE and 31.3% were considered severe BE 24 .Such a high occurrence of BE was found as well in obese individuals who did not seek weight loss treatment 6 and among college female students, who did not seek weight loss programs, with 54.5% BE prevalence in those overweight and obese 25 .
Although the initial hypothesis that BE prevalence would increase along with the obesity degree, the increase observed was not statistically significant, even though it reached 75% among the super superobeses.Several studies have found elevated BE prevalence in obese individuals, although the association with obesity degrees was not assessed 6,25,26 in the same manner as this study.Only one prospective study in Brazil, evaluated 28 adults who suffered from overweight or obesity, found a higher BE occurrence associated with a higher degree of obesity 27 .In the current study, there was no association between BE and obesity degrees, which could be due to the patients profile.All patients were severely obese with an average BMI of 47.4 ± 6.6 kg/m 2 , 58.5% presented obesity degree III and 26.6% presented superobesity.Also, the women showed very specific characteristics, such as low education levels (48.4% attended school for less than 9 years), and low family income (43.9% reported income between 1 and 2 minimum wages).
In a population-based transversal-type study carried out with 2097 adults, a BE prevalence of 7.9% was verified, being more prevalent in women (9.6%) and among younger individuals, between 20 and 29 years old (11.0%), with a decreasing trend as age advances 26 .Another study evaluated 2855 adults and elderlies who often visited shopping malls in five Brazilian cities observed a higher BE prevalence in the female gender (13.0%), and in younger women, between 18 and 40 years old (17.5%) 28 .Regarding age, the findings mentioned above do not match with this study findings, as the age group presenting higher BE prevalence was 40 to 49 years of age (76.2%), with a BE risk twice as higher regarding women in the age group 30-39 years old.The age group 20-29 years old was not associated with BE.These findings might have occurred due to different characteristics of the study sample, revealing that in women suffering from severe obesity the association with age does not follow the same trend of gen-eral female population.The association verified between BE and the age group 40-49 years old raises the hypothesis of a relationship with menopause, as the age group is concomitant to this process.Women in menopause show a higher prevalence of anxiety 29 , which could cause the beginning of eating disorders.
The "snacking habit" is a frequent eating pattern in obese patients submitted to surgical techniques with gastric restrictions 12 .Especially after the first six months post-operatory, this behaviour is observed as it is impossible for patients to eat excessive amounts of food, and then they usually develop a "snacking behavior" 12 .This habit is a factor that prevents adherence to dietotherapic treatment and presents as consequences unsatisfactory weight loss and difficulty in maintaining weight loss 12 .Research carried out with postoperatory bariatric surgery patients have found the prevalence of "snacking habit" between 37.0% 30 and 44.0% 31 .However, the few existing research that assessed this behaviour in severely obese individuals within the pre-operatory period found prevalence between 26.4% 32 and 19.5% 33 .
Results presented by this research are worrying, as 50% of severe obese patients presented "snacking habit" before undergoing bariatric surgery, which is a percentage higher than observed in other studies.This "snacking habit" could compromise weight loss both while preparing for the surgery as well as after the surgery.This study is, to the best of the authors' knowledge and after systematic reviews, the only study that analyzes the association between BE and "snacking habit" in Brazilian women with severe obesity, candidates to bariatric surgery.The "snacking habit" affected 64.4% of those women with BE, and was significantly associated even after multivariate analysis.Thereby these findings are relevant and contribute to scientific knowledge in the field.
The primary hypothesis is that eating impulsivity can occur because of very restrictive diets and/or frustrations caused by unsuccessful and slow treatments.Satisfactory weight loss and its maintenance require a longer time than what is expected by the patient, which are factors that could also contribute for the occurrence of BE and "snacking habit".This highlights the relevance of a multidisciplinary team in severe obesity treatment, with the participation of psychologists, psychiatrists and nutritionists, reinforcing the several questions involved in "snacking habit" and binge eating.
The fact that BES is a tracking instrument and not a diagnosis instrument could be cited as this research limitation; however this point was discussed in the methodology section.BES is an easy analysis and low cost questionnaire; it is also useful to analyze BE occurrence but not to diagnose BED.However, BES, as BE a tracking instrument in severely obese individuals is useful to clinical support as well as to nutritional treatment.It can also increase obese patients comprehension of their own situation and, therefore, optimize intervention strategies.