Night eating syndrome (NES) is characterized by a time-delayed pattern of eating relative to sleep, where most food is consumed in the evening and night. This study aimed to investigate the clinical significance of NES and nocturnal snacking by exploring the relationship between NES and (1) obesity, (2) binge eating disorder (BED) and (3) psychological distress.
One hundred and eighty bariatric surgery candidates, 93 members of a non-surgical weight loss support group and 158 general community respondents (81 males/350 females, mean age: 45.8±13.3 years, mean body mass index (BMI): 34.8±10.8 and BMI range: 17.7–66.7).
NES diagnosis required within the previous 3 months: (1) no appetite for breakfast, (2) consumption of 50% of daily energy after 1900 hours and (3) sleep difficulties 3 nights/week. Nocturnal snacking (awakening to eat) was recorded. Validated questionnaires assessed BED, symptoms of depression, appearance dissatisfaction (AD) and mental health-related quality of life (MHQoL). NES and binge eating (BE) (1 episode/week) were confirmed by interview.
NES criteria were met by 11.1% of the total cohort. Across all groups, BE (P=0.001), BMI (P=0.003) and male gender (P=0.013) explained 10% of NES variance. Individuals with co-morbid NES and BE reported similarly elevated psychological distress as other binge eaters. NES alone was not associated with psychological distress. Those with NES who consumed nocturnal snacks reported poorer MHQoL (P=0.007) and greater depressive symptoms (P=0.039) and hunger (P=0.013) than others with NES. Low MHQoL (P=0.007) and male gender (P=0.022) explained 27% of the variance in the nocturnal snacking group.
In this study, NES was positively associated with BMI, BE and male gender. Elevated psychological distress was only apparent in those who consumed nocturnal snacks. Further characterization and understanding of the clinical significance of NES and nocturnal snacking is required.
The escalating worldwide prevalence of obesity has drawn attention to the association between ‘non-normative’ eating patterns, weight gain and obesity. Night eating syndrome (NES) is such a condition, observed most frequently among groups of overweight and obese individuals. First characterized by Stunkard et al.1 in 1955, awareness of NES as a behavioral entity has only emerged in the last decade. Individuals with NES are characterized by a time-delayed pattern of eating relative to sleep, where most food is consumed late in the day and into the evening and night.2 Features of the syndrome have generally included morning anorexia, evening hyperphagia and insomnia,3, 4, 5, 6, 7, 8, 9, 10 and more recent research criteria have stipulated the consumption of nocturnal snacks, where individuals wake from sleep to eat.11, 12, 13, 14, 15 Individuals with NES, who report nocturnal snacking may constitute a group with more severe symptoms than those who do not wake to eat.16 At present, NES is not formally listed as an eating or sleep disorder, and no agreed diagnostic criteria exist.
NES prevalence estimates are low in community samples (range: 0.4–1.6%),5, 17, 18 but markedly higher among groups of obese persons seeking medical or surgical weight loss (range: 6–64%).19 Despite this, only one cross-sectional study has shown a positive relationship between NES and body mass index (BMI).20 Symptoms of depression are more consistently associated with NES,1, 4, 7 but may be influenced by the concurrence of binge eating disorder (BED). BED involves regular episodes of excessive, uncontrolled overeating, and is strongly associated with psychological distress.21 In obese populations, BED has been linked with NES 6, 9, 10, 22 and nocturnal snacking;23, 24 however, the nature of the association is uncertain. It is currently unclear whether NES, as a discrete condition, is associated with emotional distress, impairment or disability, and thereby represents an eating disorder of clinical significance.21, 25, 26 NES could simply constitute a variant of normal eating behavior that may be linked to weight gain and obesity.
The aim of this study was to investigate the clinical significance of NES by exploring the relationship between NES and (1) BMI, (2) BED and (3) psychological distress. Three groups differing in body weight and treatment-seeking status (bariatric surgery candidates, weight loss support group members and general community respondents) were recruited. Associations between NES and binge eating, symptoms of depression, body image/appearance dissatisfaction (AD), mental health-related quality of life (MHQoL) and eating behavior were investigated. It was hypothesized that:
NES prevalence would increase with increasing BMI;
NES would be associated with higher psychological distress than non-NES (and non-BED) after controlling for any distress related to age, gender and BMI;
Individuals reporting co-morbid NES and BED would display higher levels of psychological distress than those with either NES or BED; and
The subgroup of NES who woke to consume nocturnal snacks would show greater overlap with BED and higher psychological distress than NES who did not wake to eat.
All participants were recruited between August 2004 and January 2006. Inclusion required an age between 18–70 years. Subjects were excluded if they had undergone previous bariatric surgery. Six individuals were also excluded due to night-shift work,27 as was one student with a pattern of late-night studying and eating. The primary data were obtained via self-report surveys. Of 648 distributed, 431 eligible surveys were returned, representing an overall response rate of 66.5%.
The study was approved by the Monash University Standing Committee on Ethics in Research involving humans, and was conducted in accordance with the Helsinki Declaration of 1975 as revised in 1983. All subjects were informed regarding the nature of the questionnaires and consented to study involvement.
The majority of all respondents were Caucasians. Ninety three percent were residents of the state of Victoria, Australia, and the remainder resided in the neighboring states of South Australia and New South Wales.
Bariatric surgery candidates
Consecutive, eligible persons accepted into the bariatric surgery program at The Centre for Bariatric Surgery, The Avenue Hospital, Melbourne, Australia. Two hundred and thirty of 240 subjects agreed to participate and were provided with a survey pack and consent form. Of these, 180 completed surveys and consent forms were returned, a response rate of 78%.
Weight loss support group respondents
This sample provided data from obese persons who were not seeking bariatric surgery. Subjects were recruited from ‘Take Off Weight Naturally’ (TOWN), a weight loss company that consists of over 130 support groups within Victoria. Research flyers were disseminated throughout the groups and interested individuals were asked to contact the research center to arrange for the survey pack to be sent out. In total, 158 survey packs were distributed, and 93 completed surveys were returned, which represents a response rate of 59%. Names and contact phone numbers were volunteered by 71% of respondents.
General community respondents
Subjects were randomly recruited through flyers placed on notice boards in the general community, flyers on notice boards in two large metropolitan hospitals, and through survey distribution at a large Australian university. Flyers invited any interested individuals, who were not actively seeking weight loss to contact the research centre to arrange for the survey pack to be sent out. Of 260 distributed packs, 158 completed surveys were received; a response rate of 61%. Names and contact phone numbers were volunteered by 72% of respondents.
A cover sheet on the weight loss support group and general community surveys requested respondent's age, home post code and contact phone number (both optional), height, weight and date this weight was last checked. Although based solely on self-report, 87% of subjects from the support group and general community stated that they had weighed themselves within the previous month. The surgical group consented to have demographic and anthropometric information obtained from clinic notes. All survey packs were otherwise identical and consisted of a questionnaire on NES, and five validated surveys listed below.
Eating disorder diagnoses and eating behavior
A self-report survey screened for NES diagnostic criteria based on those of Stunkard et al.10 in 1996. The survey informed subjects that the questions related to past 3 months only. Six questions requiring a yes/no response were listed:
Do you usually have no appetite for breakfast?
Do you skip breakfast on 3 or more days of the week?
Do you usually eat the majority of your food intake, that is greater than half of the calories that you would eat over a 24 h period, after 1900 hours?
Do you have trouble getting to sleep and/or staying asleep on 3 or more days of the week?
Have you experienced awakenings during the night on at least 3 nights of the week over the last 3 months?
When you awaken during the night, do you find yourself frequently consuming snacks?
NES diagnosis required that persons within the previous 3-month period usually: (1) had no appetite for breakfast, (2) consumed 50% or more of total energy intake after 1900 hours and (3) had trouble getting to sleep or staying asleep on three or more nights of the week.
The Questionnaire on Eating and Weight Patterns – Revised (QEWP-R),28, 29 a 28-iem instrument to assess BED criteria as outlined in the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV),21 was used to screen for characteristics of binge eating. The QEWP-R also collects data on weight and shape concerns and symptoms of bulimia nervosa.
Following completion of the NES questionnaire and the QEWP-R, all surgical candidates participated in a short semistructured interview, and community respondents and support group members who reported characteristics of binge eating or night eating underwent a semistructured phone interview. The purpose of the semistructured clinical and phone interviews was to verify survey responses. The interview moved systematically through both surveys, repeating all questions. Subjects were provided with fuller descriptions of difficult concepts such as the experience of loss of control. During the interview, a brief diet history by an experienced dietitian (SLC) determined whether subjects usually ate greater than half their dietary energy after 1900 hours.
Eating behaviors and cognitions were further assessed using The Three Factor Eating Questionnaire (TFEQ).30 This widely used tool contains 51 items that measure three dimensions of human eating behavior: (1) cognitive dietary restraint (deliberate restriction of food intake; intent to diet); (2) disinhibition of eating (the inability to resist social, emotional or external eating cues); (3) subjective feelings of hunger. Scores increase with increasing eating pathology.
Psychological health and quality of life
Symptoms of depressive illness were assessed using the revised Beck Depression Inventory (BDI),31 a 21-item self-report instrument that assesses traits of a major depressive episode. The BDI has been used widely in subjects ranging from normal weight to obese. The validity and internal consistency of the BDI are well documented,31, 32 although weight ranges in study populations were not specified. Within a possible range of 0 and 63, a score of 0–9 was considered ‘normal’; 10–16 ‘mild’; 17–29 ‘moderate’; 30–63 ‘severe depression’.33
The Multidimensional Body Self Relations Questionnaire (MBSRQ)34 was used to calculate an AD score, as a measure of body image distress. The appearance orientation (AO) subscale assesses the importance an individual places on physical appearance and presentation. Appearance evaluation (AE) provides a measure of how an individual self-assesses their own appearance and attractiveness. The level of AD is determined by calculating the difference between the AO and AE.35
The Medical Outcomes Trust Short Form-36 (SF-36),36, 37 a widely used and validated survey of general health and outcomes, was used to assess health-related QoL. The SF-36 consists of eight domains individually weighted into physical and mental components, which can be combined to calculate physical component summary (PCS) and mental component summary (MCS) scores. A lower MCS indicates poorer self-rated psychological health and more social disability due to emotional problems. The SF-36 MCS was used as a measure of psychological distress.
Descriptive statistics were used to calculate the mean±s.d. for continuous variables when the total study group was divided according to recruitment origin, and to define ‘NES only,’ ‘BE only’ and ‘Co-morbid NES and BE.’ Two control groups matched for age, gender, BMI and ‘recruitment origin’ to ‘NES only’ and ‘BE only’ were derived from persons in the ‘no NES or BE’ group. These matched control groups were also presented as mean±s.d. Binary logistic regression identified factors independently predictive of NES. Factors entered into the model included ‘recruitment origin,’ gender, BMI, age, depression score, SF-36 MCS and PCS, AD score and BE. The difference in mean values for ‘NES only,’ ‘BE only’ and their matched controls, and ‘NES only,’ ‘BE only’ and ‘Co-morbid NES and BE,’ was assessed using independent t-tests, and χ2 analysis for gender. Characteristics of the subgroups of NES who did and did not consume nocturnal snacks were also compared using independent t-tests, and χ2 analyses for gender and BE. Finally, within all NES and within the total cohort, binary logistic regression and linear regression identified factors predictive of nocturnal snacking. Factors entered into the models included ‘recruitment origin,’ gender, BMI, age, depression score, SF-36 MCS and AD score. SPSS version 12.0.1 was used for statistical analysis.
There was no difference in gender, age or BMI between participants and non-responders in the surgical group. Owing to the anonymity of the surveys disseminated to the community groups, differences between responders and non-responders could not be assessed.
Data were obtained from persons of a wide BMI range. The final groups comprised the surgery candidates (n=180, BMI range: 31.9–66.7), weight loss support group members (n=93, BMI range: 21.3–60.2) and community respondents (n=158, BMI range: 17.7–45.5). Table 1 lists descriptive features of each group.
Prevalence of eating pathology
Based on responses to the self-report questionnaire, 62 persons met NES criteria; however, following the confirmatory clinical or phone interview, this number reduced to a final group of 48. Rates of NES were significantly different between groups at 19.4% in the surgery candidates, 4.3% in the support group and 5.7% in the community, χ2(2, n=428)=11.33, P<0.001 (Table 1). A subgroup of NES also reported nocturnal snacking (NES+snacks). This occurred in 7.2, 1.1 and 1.3% of each recruitment group, respectively, and was also statistically different between groups, χ2(2, n=428)=5.43, P=0.005. In particular, rates of NES and nocturnal snacking were higher among the surgical candidates than the support and community groups.
When binge eaters who reported one binge episode per week (n=14) were compared to those reporting two or more binges per week (n=38); both groups showed similar demographic, psychological and behavioral characteristics (data not shown). These two groups were therefore combined and collectively termed binge eaters (BE). Prevalence of BE varied between recruitment groups, at 24.4% in the surgery candidates, 5.4% in the support group and 1.9% among the community respondents, χ2(2, n=428)=25.14, P<0.001. Co-existing NES and binge eating was present in 4.4% of the total cohort. No subject met criteria for bulimia nervosa.
Correlates and characteristics of NES
The effect of age, gender, BMI and ‘recruitment group’ on the presence of NES was assessed in a binary logistic regression model. In the analysis, BMI (P<0.001) and male gender (P=0.034) explained a significant proportion of the variance in NES diagnosis (r2=0.049). Figure 1 illustrates NES prevalence according to BMI category. χ2 Analysis showed there was a significant difference in NES prevalence across five BMI categories, χ2(4, n=48)=22.71, P<0.001.
A second binary logistic regression model explored the association between NES and psychological distress while controlling for BMI and gender. Factors entered included BDI depression score, SF-36 MCS, AD score and binge eating. BE were almost seven times more likely to manifest NES (odds ratio: 6.9; 95% confidence interval (CI): 3.5–13.7). Overall, BE status (P=0.001), BMI (P=0.003) and male gender (P=0.013) explained 10% of the variance in NES diagnosis. χ2 Analysis confirmed the strong association between NES and BE, χ2(1, n=431)=38.56, P<0.001, and showed a positive trend toward NES and male gender, χ2(1, n=431)=3.81, P=0.051.
The group containing ‘NES only’ (n=29) was matched for age, gender, BMI and ‘recruitment origin’ to a comparison group derived from persons in the ‘No NES or BE’ category. The ‘BE only’ (n=33) were also compared to a matched group of subjects without NES or BE. Table 2 demonstrates there was no statistical difference between the ‘NES only’ group and matched controls for all psychological or behavioral variables. In contrast, BE showed significantly higher scores for symptoms of depression, SF-36 MCS, weight and shape concern, dietary disinhibition and hunger, compared to matched controls without binge eating.
Characteristics of subjects with ‘Co-morbid NES and BE’ (n=19) were also assessed (Table 2). In total, 40% of those with NES manifested binge eating, and a similar proportion of BE also manifested NES (37%). Weight-related variables, measures of psychological distress and eating behaviors of ‘Co-morbid NES and BE’ were compared to ‘NES only’ (n=29) and ‘BE only’ (n=33) using independent t-tests. Table 2 illustrates that weight-related variables were similar between the three eating disordered groups. Yet importantly, the ‘Co-morbid NES and BE’ and ‘BE only’ groups scored similarly high on all psychological variables. Comparison of the ‘Co-morbid NES and BE’ and ‘NES only’ groups revealed significantly lower psychological distress among those with ‘NES only’. Symptoms of depression measured by the revised BDI, AD score, dietary disinhibition and hunger, and importance of weight and shape were all significantly lower in the ‘NES only’ group.
Comparison of groups who did and did not consume nocturnal snacks
Characteristics of the subgroup of NES who consumed nocturnal snacks (NES+snack; n=16) were compared with NES who did not report this behavior (NES-no snack; n=32). Mean BMI was similar in both groups, t(46)=−0.92, P=0.362. Gender distribution was not statistically different, χ2(1, n=48)=2.47, P=0.116, however; males comprised 44% of ‘NES+snacks’ compared to 22% of the ‘NES-no snack’ group. Binge eating was not more prevalent in the ‘NES+snacks’ (43.8%) than ‘NES-no snack’ group (37.5%), χ2(1, n=48)=0.17, P=0.676, therefore BE were not excluded from subsequent analysis of psychological traits. The ‘NES+snack’ group showed significantly higher symptoms of depression on the revised BDI, t(43)=−2.13, P=0.039, a lower SF-36 MCS score, t(43)=2.83, P=0.007 and greater TFEQ hunger score, t(46)=−2.59, P=0.013. In a binary logistic regression model, low SF-36 MCS (P=0.007) and male gender (P=0.022) explained 27% of the variance in the ‘NES+snacks’ group.
Finally, within the total cohort (n=431), the characteristics of 30 individuals who reported the consumption of nocturnal snacks were explored. Of this group, n=16/30 also reported full NES criteria, that is the ‘NES+snacks’ group, and n=14/30 were not diagnosed with NES. Significantly more males comprised the small group of nocturnal snackers, χ2(1, n=431)=4.467, P=0.035 with 12.3% of all males reporting nocturnal snacking compared to 5.7% of females. In the binary logistic regression model, gender, (P=0.002), the revised BDI score, (P=0.019) and SF-36 MCS score, (P=0.038) explained 7.3% of the variance in nocturnal snacking behavior. The average score on the revised BDI for the nocturnal snackers was 19.3±11.4 compared to 10.3±8.5 in the remainder of the cohort. Table 3 presents the results of a linear regression analysis to assess factors driving the three measures of psychological distress. Nocturnal snacking within the total cohort was positively associated with BDI score and negatively associated with MHQoL. Again, NES was not associated with any psychological measure.
This study compared NES in a large cohort of persons ranging widely in BMI and treatment-seeking status. Characteristics of NES were contrasted with non-NES and with persons manifesting BE and co-morbid NES and BE. The clinical significance of nocturnal snacking was also explored. Of primary interest were differences in BMI and markers of psychological distress. Importantly, regardless of weight control endeavors, NES prevalence was positively associated with BMI. Until now, this association has been generally accepted due to consistently higher NES prevalence estimates in cohorts of overweight and obese when compared to the general community. Only one cross-sectional study has directly supported the positive relationship between NES and BMI,20 while the majority have shown no connection.3, 5, 6, 7, 8 The lack of association may be the result of a small BMI range within homogenous populations and inadequate power to detect group differences. The current data, collected specifically to provide BMI values across a broad spectrum, clearly show a strong, independent association between NES and BMI.
Binge eating behavior was also closely linked with NES. Most strikingly, BE were almost seven times more likely to manifest NES than non-BE. Co-morbid NES and BE was 4% in the total cohort. Yet in the NES group, 40% reported binge eating, and among the BE group 37% also reported NES. These findings represent similar rates of co-morbid NES and BE to other obese study groups;6, 10, 22 however, higher8 and lower9, 10 rates have been reported. Overlap between nocturnal snacking and BED has also been reported within obese populations.23, 38, 39, 40 While BED and NES have been studied and described as separate entities,13, 41 these data highlight that co-occurrence of the two conditions is common. Similarities may exist in the consumption of an objectively large amount of food for the circumstances, and binge eating, which can extend over several hours, is common in the late afternoon and evening. While a perceived lack of control in binge eating is essential, night eating behavior may also be under poor self-control.42
Yet despite some behavioral similarities between binge eating and night eating, associated levels of psychological distress appear markedly different. The group with NES scored low on all psychological measures, and was comparable to persons without NES. This was in sharp contrast to the BE group who yielded significantly higher symptoms of depression and AD, more weight and shape concern, dietary disinhibition and hunger, and lower MHQoL than matched non-bingeing controls. Furthermore, when NES and binge eating co-occurred, binge eating was the factor clearly associated with elevated psychological distress.
A number of studies have examined the link between NES (diagnosed according to various criteria) and associated psychological disturbance. However, few studies have controlled for the influence of binge eating. Those that have controlled for binge eating have reported similar levels of psychological functioning between NES and non-NES in a normal weight sample of Black females18 and an obese sample.8 Another study, which notably diagnosed NES by consumption of 25% total energy after the evening meal and/or awakenings to eat 3 times in a week (nocturnal snacking), did report a positive association between NES and symptoms of depression in obese NES compared to weight-matched controls.13 A high level of psychological disturbance (and concurrent binge eating) have also been reported among nocturnal eaters referred to a sleep clinic for polysomnography.40
The present study was the first to examine differences between subgroups who did and did not consume nocturnal snacks. Frequent nocturnal snackers reported higher symptoms of depression and hunger, and lower MHQoL compared to the NES who did not wake to eat. This association between nocturnal snacking and psychological distress provides clinical significance to the behavioral feature of waking to eat, and supports the proposal that nocturnal snackers are a group with more severe impairment.16 This also highlights the hitherto lack of distinction regarding the status of NES as an eating disorder or variant of normal eating behavior. NES as defined in the present study appeared to be an extension of normal eating behavior linked with weight gain and obesity. Further examination is required and should also consider the level of control over nocturnal eating and the time-delayed pattern of eating, and the link between NES and stress.11, 43, 44
Nocturnal snackers also tended to be male. Although gender differences in nocturnal snackers have not previously been assessed, one other study of morbidly obese treatment seekers has observed an increased risk of NES in males.20 In contrast, similar gender distributions between NES and non-NES have been found in surgical populations,6, 45 and obese3, 7 and general samples.3, 7, 17 Although limited by a relatively small male sample, our findings suggest that depressed men are more likely to engage in nocturnal snacking behavior. While there are no clear explanations, work-related stress,46 obstructive sleep apnea40, 47 or another sleep-related disorder, such as sleepwalking or restless legs syndrome,40 are possible correlates. Waking to use the bathroom,42 or insomnia, where eating acts as a ‘time killer’40 are other factors that may lead to nocturnal snacking. The identification and treatment of low mood could assist to alleviate nocturnal snacking, and reduce the associated risk of weight gain.
Finally, lower evening leptin levels may contribute to nighttime hunger and stimulate nocturnal snacking. Low circulating leptin has been observed in normal weight and obese NES, compared to weight-matched non-NES,11 although similar leptin levels have also been reported.12 The present study did not specifically measure evening hunger, but found elevated hunger ratings in nocturnal snackers using a general measure of self-reported hunger.
A strength of the current study is the inclusion of a large cohort of subjects comprising a broad BMI range, recruited from geographically similar locations within the same timeframe. BMI was derived from clinic measurements of weight and height in the obese surgical candidates, and self-report in the community and weight loss support groups. Although self-report weights tend toward underestimation, particularly as body weight increases,48 the majority of our overweight and obese respondents were weighed manually. Furthermore, self-reported weights in a general population have shown adequate sensitivity and good specificity when compared to actual weight.49 Although the recruitment methods differed slightly and the three original groups possessed distinct characteristics, our methods of statistical analyses controlled for possible confounders, in particular differences in BMI and the presence of binge eating.
Another potential limitation of this study was the collection of data by self-report questionnaires. To minimize this weakness, the surveys selected had been validated previously within a range of population groups. Furthermore, the QEWP-R and NES survey were used as a screening tool,50 and research criteria for BED and NES were verified by either clinical or phone interview. An on-going limitation of research involving NES is the lack of formally validated diagnostic criteria and assessment methods. This study employed the most commonly applied criteria10 and validated self-reported behavior in an interview. Nocturnal snacking, which is emerging as an important component of NES, was also considered.
NES as defined in this study showed a strong positive association with obesity, while frequent nocturnal snacking conferred an elevated risk of psychological disturbance. These findings highlight two clinically significant relationships, and importantly, provide a step toward differentiating a variant of normal eating behavior from disordered eating associated with emotional distress or impairment. The high degree of overlap between NES and binge eating is also noteworthy and merits additional study. Finally, we suggest that male gender may be a risk factor for NES and nocturnal snacking. Awareness of NES and nocturnal snacking and the risks they impose on weight gain and psychological distress are still little known in general practice. Future work should further define NES features of clinical importance to guide the development of agreed diagnostic criteria, and develop targeted intervention strategies.
We thank all of the study participants for their time and involvement, and the staff at the Centre for Obesity Research and Education (CORE) and The Centre for Bariatric Surgery in Windsor, Victoria for their on-going support and assistance.
About this article
Hospital Pediatrics (2019)