Zara J. Damania
Abstract
This study aims to investigate whether there is a bidirectional relationship between poor sleep quality, high body mass index (BMI) and disordered eating (binge-eating and night-time eating). Participants were a community-derived sample (N= 330) of people recruited through advertisements placed at the Australian National University (ANU) campus and a number of online platforms. An online questionnaire asked participants for their height, weight and recent experiences of sleep and eating. Multiple regression analyses found that: (a) worse overall sleep quality and binge-eating (but not night-time eating) were positively associated with high BMI accounting for a significant 8% of the variability in BMI; and (b) high BMI and night-time eating (but not binge-eating) were positively associated with worse overall sleep quality accounting for 35.6% of the variability in worse overall sleep quality. These results indicate that disordered eating (binge- and/or night-time eating) partially describe the association between poor sleep quality and high BMI. Future research could be conducted using objective- rather than self-reported- measures of sleep quality, BMI and eating behaviour to control for inaccuracies that self-reported measures might pose.
Best services for writing your paper according to Trustpilot
The Bidirectional Association between High Body Mass Index, Poor Sleep Quality and Disordered Eating
This paper investigates whether there is a bidirectional relationship between poor sleep quality, high body mass index and disordered eating. Current research corroborates the association between poor sleep quality and being overweight or obese. Sleep quality is a broad concept that includes: sleep duration, difficulty falling and/or staying asleep and the use of sleep medications (Buysse, Reynolds, Monk, Berman & Kupfer, 1989; Krystal& Edinger, 2008). This study makes use of the Pittsburgh Sleep Quality Index (PSQI) to measure sleep quality. PSQI is an effective and widely used self-reported instrument that is high in reliability and validity, consisting of questions that are easy to understand and answer (Buysse et al., 1989; Smyth, 1999). PSQI measures subjective sleep quality in seven distinct areas, including: sleep latency and sleep duration (Krystal& Edinger, 2008; Smyth, 1999).
One of the key interests in this study is the association between PSQ and having a high body mass index (BMI); i.e., being overweight or obese. According to the World Health Organisation’s classification, a BMI of ?25 indicates that a person is overweight and ?30 indicates that a person is obese (World Health Organization, 2000). Empirical evidence corroborates an association between PSQ and having a high BMI (hBMI). For instance, longitudinal studies and studies on large mixed-race and socioeconomically diverse samples found that sleeping less than seven hours and having trouble falling and/or staying asleep was positively associated with hBMI (Gangwisch, Malaspina, Boden-Albala& Heymsfield, 2005; Meyer, Wall, Larson, Laska & Neumark-sztainer, 2012). Furthermore, empirical findings from cross-sectional studies with mixed-race samples indicate that: obese individuals experience shorter sleep durations compared to non-obese individuals; for every hour of sleep lost the risk of obesity increased by 80%; and PSQ leads to decreases in physical activity which is consequently associated with hBMI (Cappuccio, et al., 2008; Gupta, Mueller, Chan & Meininger, 2002)
Very few studies that attempted to explain how PSQ is associated with hBMI found that sleep apnea might mediate this relationship (Yeh& Brown, 2014). Sleep apnea refers to sleep disturbance due to continual interruptions to airflow through the nose and mouth on at least 30 occasions during a seven-hour sleep period (Guilleminault, Tilkian & Dement, 1976). However, sleep apnea is relatively uncommon in the population (Tishler, Larkin, Schulchter &Redline, 2003) while PSQ is more common (Buysse, Reynolds, Monk, Berman & Kupfer, 1989). Therefore, it can be inferred that only a small proportion of hBMI individuals with PSQ suffer from sleep apnea and there might be other potential explanations for the association between PSQ and hBMI (Yeh& Brown, 2014).
Given that no other studies have attempted to further investigate factors that mediate the relationship between PSQ and hBMI, this study attempts to do so by investigating whether disordered eating mediates this relationship. Disordered eating includes both: binge eating and night-time eating. Binge eating (BE) refers to consuming unusually large amounts of food in a relatively short time-span and perceived lack of control over one’s eating behaviour (American Psychiatric Association, as cited in Johnson, Carr-Nangle, Nangle, Antony & Zayfert, 1997). This study uses the Binge Eating Scale (BES) – a questionnaire that measures whether and to what extent individuals binge eat by inquiring about their eating behaviours and tendencies (Gormally, Black, Daston& Rardin, 1982). Whereas, night-time eating (NTE) refers to consuming >25% of one’s caloric intake after dinner and/or after waking up at night, at least twice a week (Allison et al., 2010; Stunkard, Grace&Wolff). This study uses the Night-time Eating Questionnaire (NEQ) to identify whether and the frequency of which participants engage in NTE behaviour (Striegel-Moore, Franko & Garcia, 2009).
Yeh and Brown (2014) suggest that difficulty falling asleep and shorter sleep durations provide hBMI individuals with more time to eat, consequently leading to weight gain over time. This is in accordance with Andersen, Stunkard, Sorenson, Peterson and Heitmann (2004) and Crispim, Zimberg, dos Reis, Tufik and de Mello (2011) who respectively found that NTE was associated with both PSQ and weight gain in hBMI individuals. Similarly, empirical research has indicated that BE is associated with PSQ and hBMI in obese individuals (Yeh& Brown, 2014).
In light of both: the lack of research investigating potential mediators of the association between PSQ and hBMI and research indicating that disordered eating is associated with PSQ and hBMI, the present study aimed to determine whether disordered eating (NTE and BE) mediates the relationship between PSQ and hBMI. The hypotheses of this study were: (1) Poor sleep quality and disordered eating will be associated with high BMI; and, (2) High BMI and disordered eating will be associated with higher scores of poor sleep quality.
Method
Participants
Participants were recruited through advertisements placed at the Australian National University (ANU) campus and a number of online platforms. Study inclusion criteria were: being ? 18 years old and a BMI of 18.5 (normal weight) or more. 678 participants opted to participate in this study; however data from only 330 participants were used because the remaining 348 did not meet the study inclusion criteria or did not complete the study. Of the 330 participants, 107 (32.4%) were males, 223 (67.6%) were females, the ages ranged from 18-87 years and the mean age was 27.42 years (SD=10.36).
Procedure
Participants accessed the study by clicking on an embedded URL in the advertisement. If they met the study criteria and consented to participate, they responded to an online questionnaire inquiring about: their recent experiences of sleep and eating and height and weight, to calculate their BMI. SPSS statistical software (version 22) was used to perform all statistical analyses. Two standard multiple regression analyses were performed to test the two hypotheses.
Materials
Demographics including education level were collected. BMI was calculated by computing participants’ weight (in kilograms) over their height (in meters); with a BMI of ?25 indicating overweightness and ?30 indicating obesity. Next, the PSQI assessed seven subjective domains of sleep. An overall PSQI score (ranging from 0 to 21) of >5 indicated moderate to severe sleep difficulties. Overall sleep score has high internal consistency reliability with a Cronbach’s ? of .83 (Smith & Wegener, 2003). Thirdly, BE was measured using the BES; which consists of 16-items reflecting behaviours and feelings related to eating. An overall BES score (ranging from 0 to 46) of >27 indicated binge-eating and a higher overall score indicated worse binge eating. In this study, BES had high internal consistency with a Cronbach’s ? of .92. Finally, NTE was measured using NEQ which consists of 15 questions. An overall NEQ score (ranging from 0 to 52) of >25 indicated NTE behaviour. In this study, the NEQ showed sufficient internal consistency with a Cronbach’s alpha of .73.
Results
A number of outliers were detected for each of the variables; however, none of these were excluded because they represented clinically relevant cases. Kolmogorov-Smirnov statistics of overall sleep quality, BMI, BE and NTE were found to be non-significant (i.e., p<.05); which means that these key variables were normally distributed.
Two multiple regression analyses (MRA) were conducted to investigate whether: (a) PSQ and disordered eating (BE and NTE) were associated with high BMI; and (b) whether high BMI and disordered eating (BE and NTE) were associated with higher scores of PSQ. Means and standard deviations of the key variables are shown in Table 1.
Table 1
Means and Standard Deviations of Key Study Variables
Variable Mean Standard Deviation
BMI 26.08 8.55
PSQI 6.37 3.66
BES 13.78 10.25
NEQ 14.60 6.85
Note. N=330
PSQI= PSQI global score= total sleep quality score
BES= binge eating scale score
NEQ= night eating questionnaire score
BMI= body mass index
The first MRA found that overall PSQ and BE (but not NTE) were positively associated with high BMI (the dependent variable) accounting for a significant 8% of the variability in BMI, R2 =.080, adjusted R2=.071, F (3,326) = 9.40, p=.000. Examination of the beta weights suggested that BE was the strongest unique contributor to high BMI (see Table 2).Unstandardized (B) and standardized (?) regression coefficients for each predictor in the regression model are provided in Table 2.
Table 2
Unstandardized (B) and Standardized (?) Regression Coefficients for each Predictor in a Regression Model Predicting high BMI
Variable
B [95% CI]
?
Total sleep quality
.399
.171
Binge-eating
Night-eating
.118
.033
.142*
.026
Note. N=330.
C.I.= Confidence level
*p<.05
The second MRA found that high BMI and NTE (but not BE) were positively associated with overall PSQ (the dependent variable) accounting for 35.6% of the variability in overall PSQ, R2 =.356, adjusted R2=.350, F (3,326) = 59.99, p=.000. Unstandardized (B) and standardized (?) regression coefficients for each predictor in the regression model are provided (see Table 3).
Table 3
Unstandardized (B) and Standardized (?) Regression Coefficients for
each Predictor in a Regression Model Predicting Worse Overall Sleep Quality
Variable B[95% CI] ?
BMI .051 .120*
Binge-eating .267 .500
Night-eating .032 .090*
Note. N=330
C.I.= Confidence level
*p<.05
Discussion
This study aimed to determine whether disordered eating (NTE and BE) explains the relationship between PSQ and hBMI. The first hypothesis was mostly supported by the results which indicated that PSQ and BE (but not NTE) were positively associated with hBMI. The second hypothesis was also mostly supported by the results which indicated that hBMI and NTE (but not BE) were positively associated with PSQ.
First, the finding that: there is a bidirectional relationship between hBMI and PSQ is in accordance with empirical research findings and meta-analyses which found that: (a) sleeping less than seven hours, having trouble falling and/or staying asleep was positively associated with hBMI; (b) obese individuals experience shorter sleep durations compared to non-obese individuals; and (c) for every hour of sleep lost the risk of obesity increased by 80%; (Cappuccio, et al., 2008; Gangwisch, Malaspina, Boden-Albala& Heymsfield, 2005; Gupta, Mueller, Chan & Meininger, 2002). Second, the finding that: BE was associated with hBMI was in accordance with Yeh and Brown (2014) whose meta-analyses also found that BE was positively associated with hBMI. Third, the finding that NTE was associated with PSQ is consistent with those of Crispim, Zimberg, dos Reis, Tufik and de Mello (2011) who found that NTE was associated with PSQ. The findings that did not support the hypotheses of the study included: NTE is associated with high BMI and BE is associated with PSQ. These findings are inconsistent with those of Andersen, Stunkard, Sorenson, Peterson and Heitmann (2004) who found that NTE was associated with weight gain in individuals with hBMI and Yeh and Brown (2014) whose meta-analyses found that BE is associated with PSQ.
Overall, these findings mostly corroborate empirical findings and theories which claim that: there is a bidirectional positive association between hBMI and PSQ, BE is positively associated with hBMI and NTE is positively associated with PSQ. However, it is possible that there are alternative explanations for the results of this study and /or potential confounds that might have influenced the results. The first limitation of the study is that the majority of the participants were females, well-educated and of a normal weight. Therefore, it is potentially difficult to generalize these findings to the mixed gendered populations, individuals with and less well-educated populations. Second, these findings are based on subjective self-reported measures of sleep quality, eating behaviour and BMI. This is potentially problematic because individuals might be inaccurate- due to not knowing the responses to particular questions or deliberately lying- in providing response. Lastly, internet access was required to participate in this study; this is a limitation because it excludes individuals who cannot gain internet access easily and/or individuals who are not technologically knowledgeable.
Overall, the results of this study implicate that individuals should strive to improve their eating and sleeping habits since they seem to affect each other and that NTE should be controlled because it is associated with PSQ. In light of the limitations of this study, future research could include: mixed-gender samples (with equal amounts of male and female patticipants), using paper-based questionnaires and using objective measures of BMI, sleep quality and eating behaviour rather than subjective/self-report methods. Incorporating these suggestions could provide findings that are: more easily generalizable to the general population and more accurate measures of the key variables in the study.
In conclusion, this report investigated whether poor sleep quality and disordered eating (binge-eating and night-time eating) were associated with high BMI and whether high BMI and disordered eating were associated with higher scores of poor sleep quality. The findings indicate that poor sleep quality and binge-eating (but not night-time eating) were positively associated with high BMI and high BMI and night-time eating (but not binge-eating) were positively associated with overall poor sleep quality. These findings provide useful implications for future research and for individuals’ health behaviours.
