OBJECTIVE: To examine the association between the habit of eating at night, and the 5-y preceding and 6-y subsequent weight changes in a middle-aged population, with particular focus on the obese.
DESIGN: Prospective study with initial examination of the cohort in 1982–83, re-examination in 1987–88 and a third examination in 1992–93.
SUBJECTS: The Danish MONICA cohort includes an age- and sex-stratified random sample of the population from the Western part of the Copenhagen County. Out of 2987 subjects participating in 1987–88, a total of 1050 women and 1061 men had been examined in 1982–83, and 1993–94 too. Subjects working night shifts were excluded.
MEASUREMENTS: Night eating in 1987–88, 5-y preceding and 6-y subsequent weight change.
RESULTS: In total, 9.0% women and 7.4% men reported ‘getting up at night to eat’. Obese women with night eating experienced an average 6-y weight gain of 5.2 kg (P=0.004), whereas only 0.9 kg average weight gain was seen among obese women who did not get up at night to eat. No significant associations were found among all women, or between night eating and the 5-y preceding weight change for women. Night eating and weight change were not associated among men.
CONCLUSION: Night eating was not associated with later weight gain, except among already obese women, suggesting that getting up at night to eat may be a contributor to further weight gain among the obese.
Several environmental factors, presumably in interaction with genetic factors, seem to exert a constant pressure on the human body to increase energy intake and decrease energy expenditure, and the sensitivity to these factors may vary between subgroups of the population.1, 2 One group of behavioural traits that may generate imbalance in the energy regulation by increasing energy intake, and hence possibly relate to obesity and weight gain, are abnormal eating patterns such as restrained eating, binge eating and night eating.3, 4, 5, 6, 7, 8, 9, 10
The night-eating syndrome was described in 1955 as a condition characterised by morning anorexia (minimal or no calorie intake at breakfast), evening hyperphagia (at least 50% of daily calorie intake after the evening meal) and insomnia, and it has later been broadened to include night time awakenings with food intake.11, 12, 13, 14 Several studies have used this definition and found frequencies of the night-eating syndrome ranging from 8 to 27% among obese patients.3, 13, 14, 15, 16, 17 Although the symptom of night eating is a main component in the night-eating syndrome, systematic studies of the relation of this symptom to overweight, independent of the syndrome, are warranted.18
We therefore hypothesised that night eating, that is, getting up at night to eat, as reported in the general population, could be related to later weight gain, and we investigated whether night eating per se predicts weight gain, or weight gain predicts night eating.
In 1982, a total of 4807 Danish citizens born in 1922, 1932, 1942 or 1952 were invited to participate in the Danish MONICA project, which was carried out as part of the Glostrup Population studies in the greater Copenhagen County. The MONICA cohort includes an age- and sex-stratified random sample of the population from the Western part of the Copenhagen County, selected from the Central Person Register. In all, 226 subjects of non-Danish origin were excluded, thereby reducing the sample to 4581 subjects. A total of 3608 (79%) accepted the invitation, forming the original baseline examination from 1982 to 83 (MONICA 1 (M-82)). This group was invited to a re-examination 5 y later in 1987–88 (Gen-MONICA (M-87)), and 2987 (83%) subjects participated. Finally, in 1993–94 (MONICA 10 (M-93)), the cohort was invited to a third examination, and 2436 (68%) of the initial 3608 individuals participated in all the three examinations.
The non-participants included 324 subjects who died in the follow-up period between M-82 and M-93, but characteristics of the non-participants have been described in detail elsewhere.19 Out of 2839 subjects who gave answers to the night-eating question in 1987–88, 518 were excluded due to non-participation in either M-82 or M-93. Furthermore, 210 night and shift workers were excluded from further analysis, since night and shift workers may have misinterpreted the questionnaire and hence reported night eating despite actually working at night and therefore not getting up at night to eat.
Height was measured to the nearest 0.5 cm with subjects standing without shoes, heels together and head in a horizontal Frankfurter plane. Body weight was measured to the nearest 0.1 kg using a SECA scale, with subjects wearing only light clothes.
Follow-up and historical information on weight change
Preceding weight change was calculated by subtracting M-82 measurements of body weight from M-87 measurements of body weight. Subsequent weight change was calculated by subtracting M-87 measurements of body weight from M-93 measurements.
An extensive questionnaire on sociodemographic variables and lifestyle was filled in at all the three examinations. In M-87, subjects were specifically asked: ‘Do you get up at night to eat?’ to signify night eating. A total of 2839 out of the 2987 subjects gave answers to this question. Subjects recorded their present smoking habits, and were subsequently classified as current, former, occasional or never-smokers. Leisure-time activity was recorded on a four-point scale ranging from almost completely inactive, some physical activity, regular activity and regular hard physical training for competition. Education was recorded as the number of years in school, and the subjects’ education was classified as ≤7, 8–11 and ≥12 y. Restrained eating at meals was recorded by asking: ‘Do you refrain from eating your fill at meals?’ and subjects classified on a four-point scale ranging from never, occasionally, often and frequently. Subjects with a body mass index (BMI) ≥30 were classified as obese.
The difference in proportions between night eaters and others was tested using a χ2 test. Association between night eating and the preceding weight change was analysed with multiple logistic regression models, with night eating as the response variable. Change in body weight from M-82 to M-87 (kg) was included as the main explanatory continuous variable. A first series of models included only night eating and weight change. In a second series of models, age, smoking habits (recoded as current smokers and others), physical activity (recoded as inactive, some activity and regular/hard activity), years in school, restrained eating at meals and BMI in M-82 were included as explanatory variables. To examine whether the association between preceding weight change and night eating differed according to obesity in M-82, a third series included the product of night eating and obesity in the equation.
Multiple linear regression was used to estimate the effect of night eating on subsequent weight change from M-87 to M-93 as response variable, and night eating was included as the main explanatory variable. In a first series of crude models, only the response variable and the main explanatory variable were included. A second series included the explanatory variables age, smoking habits, years in school, leisure time physical activity, restrained eating at meals, baseline BMI and weight change from M-82 to M-87. To examine whether the association between night eating and subsequent weight change differed according to obesity in M-87, a third series included the product of night eating and obesity in the equation.
All data were analysed using the SAS 8.02 statistical software package.
The 518 subjects excluded due to non-participation in M-82 or M-93 were found to differ somewhat from the participants, in that they included slightly more older people, and slightly more people with a BMI ≥30 when examined at M-87.
In total, 95/1050 (9.1%) women and 78/1061 (7.4%) men reported night eating. Table 1 gives the characteristics among night eaters compared to others on selected variables. Among the women, a significantly greater fraction of night eaters were older than 54 y of age, had ≤7 y of school, were daily smokers and unemployed.
Night eating and preceding weight change
Night eating was not associated with weight changes between M-82 and M-87 in the crude and adjusted analyses. Obesity in M-82 did not modify the association between preceding weight change and night eating (Table 2).
Night eating and subsequent weight change
For men, night eating was not associated with subsequent weight change (M-87 to M-93). For women, obesity was found to modify the association between night eating and subsequent weight gain (P=0.01). Analyses revealed that obese women with night eating experienced a greater average 6-y weight gain (4.2 kg, 95% confidence interval 1.3–7.0) than the obese women without night eating (Table 3). The total average 6-y weight gain for obese night-eating women was 5.2 kg, whereas obese non-night-eating women experienced only 0.9 kg average weight gain (Figure 1).
In the present study, we used a simple yes/no question to assess the lifestyle of getting up at night to eat, and examined its relation to weight change. Our findings indicate that, when using this assessment of the night-eating phenomenon, obesity and night eating had a joint effect on 6-y weight change for women but not for men, suggesting that night eating may be a significant contributor to further weight gain among already obese women, but not for others.
A newly published study on meal patterns and obesity in Swedish women confirms that obese women have a meal pattern that is distinct from normal weight women. They found that obese women consumed more meals in the evening and night and more between-meal snacking than normal weight women.20 It is possible that a feedback regulation in the non-obese night eaters avoids excessive energy intake by means of a lowered energy intake, or a rise in physical activity the following day, whereas such a feedback mechanism may not be operative in the obese or the obesity-prone. Obstructive sleep apnoea may be perpetuating night eating due to frequent awakenings, but since sleep apnoea is twice as common among men as women it is a less likely explanation.
The present study did not find differences in weight gain among women with and without night eating in general. However, the insignificant tendency to a greater weight gain among the night-eating women, as presented in Figure 1, could also be a simple consequence of lack of statistical power.21 Hence, a weak association between a relatively seldom symptom of night eating and subsequent weight gain among women may be difficult to detect if several pathways to weight gain supervene. Indeed, when analyses were focusing on the obesity-susceptible subgroups, associations became significant.
Additionally, it may be argued that the measurement of night eating with a simple yes/no question with no specified time frame may have reduced the sensitivity of the analyses. However, the fact that we were indeed able to detect interactions using this simple question suggests that a more sophisticated instrument, specifying the time frame for this phenomenon, should be able to discriminate even better, since a random measurement error likely to be introduced by the simple question would tend to attenuate our results.
As regards the reported gender differences, it could be argued that both eating at night and weight change are related to childbearing and may potentially have influenced the findings of differences between men and women, but in our study the exclusion of fertile women (<45) (n=302) did not alter the results (data not shown), making this possibility less likely. Also, the restrained eating may relate to both night eating and weight change specifically among women, and may explain part of the found gender differences. However, in the present study, restrained eating among women did not confound the associations between night eating on subsequent weight change. On the other hand, sex-specific genetic effects on BMI and weight change have been reported previously.1, 22, 23 Studies of the night-eating syndrome and related conditions have found these conditions to be associated with psychopathology,13, 24, 25, 26 and the difference in associations between men and women in the present study may reflect sex-specific differences in the psychopathology of the obese. It could be that night eating is promoting weight gain because of a positive imbalance in the energy regulation caused by excessive energy intake during night. However, the gender and obesity differences in the association suggest that night eating is not a uniform phenomenon, but a condition that operates in combination with obesity, and only among women. Another interpretation may be that some other mechanism than night eating was operating and caused the weight change. If this is the case, this simple assessment of the phenomenon may serve as an indicator or an intermediate factor for an underlying determinant. Finally, energy intake at night is likely to constitute a lower fraction of daily caloric need for obese men than for obese women, which would make it easier to detect an association in women.
The lack of an association between preceding weight gain and night eating is another finding that needs an explanation. In view of the adverse consequences of night eating, the women suffering from night eating may tend to cease night eating when a certain level of weight gain is evident. Since we do not have information on the night-eating status from M-92, we are unable to perform a further examination of this issue. On the other hand, this explanation would be in congruence with the notion of night eating as a psychopathological condition,1 which may introduce irregular eating patterns at certain periods of life, especially among women.
The nature of the night-eating phenomenon is yet to be established, but abnormal nocturnal levels of the hormones leptin and ghrelin may provide a biologic mechanism involved in night eating. For instance, Birketvedt et al13 found reduced nocturnal levels of leptin among normal- and overweight women with the night-eating syndrome compared with controls, and hence indicated that attenuated nocturnal levels of leptin may promote night eating, by inhibiting sufficient suppression of appetite. Ghrelin is believed to operate in the opposite direction of leptin by stimulation of feeding and wakefulness.27 Augmented levels of plasma ghrelin were found both immediately before meals and also at 0100 h, followed by an acute post-prandial decline in ghrelin levels.28, 29, 30 English et al31 reported a lack of this post-prandial decline specifically among the obese. We speculate that insufficient post-prandial decrease in plasma ghrelin levels could promote getting up at night to eat especially in the obese. Blundell et al1 offer an alternative view, as they argue that particularly obese women may try to exercise cognitive control over food intake which is largely under physiological control, and hence this attempted self-control may lead to disordered eating patterns.
In populations followed over a long time period, there is a strong tendency for weight gain to be followed by weight loss and vice versa,33 and weight gain subsequent to night eating in the present study could be a result of this tendency. Also, night eating could be a consequence of previous weight change and inadequate attempts to control weight gain. However, analyses of night eating and subsequent weight gain were adjusted for preceding weight gain, and preceding weight gain was not related to night eating when analysed separately. Therefore, it is not likely that weight change prior to night eating can explain the subsequent weight gain in the present study.
Possible differences of non-participation and dropout from follow-up may also bias the findings. However, when comparing the BMI for those who attended only the baseline examination with those attending all three examinations, such a bias was not apparent. Still, those who never turn up for examination have been found to be more obese than participants.32 Such a bias may, however, have led to an underestimation of BMI, and, hence attenuated, not strengthened the associations between night eating and weight change among the obese.
Other variables than those included may confound the present findings. For instance, binge-eating disorders have previously been associated both with weight gain and various definitions of night eating, and inclusion of binge-eating disorder could influence the findings.18, 34 Future studies with information on diet and dieting pattern may provide valuable information in the assessment of the relation between night eating and weight gain.
In conclusion, this simple lifestyle phenomenon of night eating defined as ‘getting up at night to eat’ did not seem to be associated with subsequent weight gain, except in a subgroup of obese women.
This study was supported by grants from The Danish Medical Research Council (the FREJA programme), The Danish National Research Foundation, The Augustinus Foundation, and The Danish Ministry of Health.