Relationship between hedonic hunger and subjectively assessed sleep quality and perceived stress among university students: A cross-sectional study

Abstract

Purpose

This study examined the relationship between hedonic hunger (HH), sleep quality, and stress levels among university students in the United Arab Emirates and the Kingdom of Bahrain.

Methods

We used a cross-sectional design with participants (N = 565) recruited via convenience sampling. Data were collected with a self-administered, standardized, and validated online questionnaire. HH was assessed with the Palatable Eating Motives Scale (PEMS) and Power of Food Scale (PFS), sleep quality and sleep components were assessed with the Pittsburgh Sleep Quality Index (PSQI), stress was evaluated with the Perceived Stress Scale (PSS), and physical activity was examined with the International Physical Activity Questionnaire. Descriptive and analytical statistics were used to assess the relationship between HH and sleep quality and perceived stress.

Results

There were positive associations between total PSQI scores and total PEMS (β = 0.14, 95% confidence interval [CI]: 0.06–0.25, P = 0.001) and PFS (β = 0.21, 95% CI: 0.45–1.04, P < 0.001). The likelihood of poor sleep quality increased by 8% (odds ratio [OR] = 1.08, P = 0.020) and 43% (OR = 1.43, P < 0.001) for each one-unit increase in PEMS and PFS scores, respectively. We also found positive associations between PSS scores and total PEMS (β = 0.19, 95% CI: 0.26–0.63, P < 0.001) and PFS (β = 0.23, 95% CI: 1.04–2.22, P < 0.001) scores.

Conclusion

Reducing HH and stress levels may help to enhance sleep quality among university students. Conversely, improving sleep quality and reducing stress levels could improve HH in this population.

Abbreviations

BMI

Body mass index

HH

Hedonic Hunger

IPAQ

International Physical Activity Questionnaire

PAQ-SF

International Physical Activity Questionnaire-Short Form

MET

Metabolic Equivalent of Task

OR

Odds ratio

PEMS

Palatable Eating Motive Scale

PFS

Power of Food Scale

PSS

Perceived Stress Scale

PSQI

Pittsburgh Sleep Quality Index

CI

Confidence interval

1. Introduction

Hedonic hunger (HH) refers to an imbalance of food-regulating hormones resulting in the persistent desire to consume highly palatable foods (i.e., foods high in sugar, salt, and fat) in the absence of physiological hunger. The level of HH varies among individuals, and those who score the highest on the HH scale may have problematic behavioral and physiological characteristics [1]. True homeostatic hunger only occurs a few times a day as it is primarily concerned with the energy balance regulation of peripheral hormones (leptin and ghrelin) [2]. Research suggests HH may be correlated with BMI [3], and it has been shown that individuals with higher HH tend to lose control when eating; this may result in the development of binge eating disorder, which may lead to obesity in the long term if left untreated [4]. Emotional eating, with its driving factors such as HH, mental stress, and other negative emotions (such as depression and anxiety), and sleep have been reported to interplay with each other in shaping body weight, especially among those people with obesity [5].

Several studies have shown that short sleep duration alters appetite-regulating hormones. For example, sleep deprivation lowers leptin and increases ghrelin levels, which contribute to increased appetite and weight gain over time [[6], [7], [8]]. A previous study reported that short sleep time was related to a higher body mass index (BMI), and those with short sleep duration were more likely to have obesity [9]. In addition, cross-sectional data suggest that energy consumption later in the day and during the night is associated with weight gain [10]. Several studies have indicated that diet impacts sleep and food choices are associated with sleep duration and quality [11,12]. Unhealthy eating habits are prevalent among university students, including night eating [13], frequent snacking, skipping meals [14], and high caffeine intake [15]. Markwald et al. reported that increased food intake after a short sleep duration was a physiological adaptation to provide the required energy to sustain additional wakefulness; however, when food was readily available, consumption exceeded this requirement [16]. Transitioning to adequate sleep duration has been associated with decreased energy consumption and weight loss [16]. Research conducted among university students in the United Arab Emirates (UAE) indicated that individuals with irregular mealtimes were likely to experience poor overall sleep quality and shorter sleep duration [17].

Chronic stress is linked to the etiology of obesity by interfering with both energy intake and expenditure mechanisms, which directly or indirectly contribute to increased appetite and caloric intake, and decreased physical activity [18]. The association between stress and food intake appears to be highly individualized [19]. Stress impairs cognitive functions such as executive control and self-regulation by increasing the intake of foods high in calories, fat, and sugar [18]. Existing studies have shown that medical students are at high risk for developing mental health problems, including eating disorders, because of academic stress, high workloads, and exposure to illnesses and death during their clinical training [20]. Associations between psychosocial stress and increased consumption of palatable foods (high in fat, sodium, and refined carbohydrates) and weight gain have been observed in overweight and obese individuals [19].

Previous studies attempted to investigate the association between sleep and HH, but no studies have considered all three indicators (HH, sleep, and stress). Research suggests that a range of factors may mediate the relationship between stress and sleep; one such factor is HH. HH may play a role in the relationship between stress and sleep, as stress can lead to an increased desire for highly palatable foods, and together those factors may lead to disruptions in sleep. Previous research has suggested a relationship between HH, stress, and sleep [21,22], however, HH as a mediator between stress and sleep is yet to be investigated. Among the scarce studies is the one that investigated sleep quality and HH among university students in Turkey. That study reported that poor sleep quality and shorter sleep duration are associated with high HH, which increased the tendency to overeat unhealthy food and played a role in weight gain [3]. Furthermore, another study from the UAE that examined food consumption frequency and perceived stress scale (PSS) scores among female university students reported a relationship between moderate stress and high consumption of fast foods and soft drinks [23]. Therefore, to add to the existing literature, we examined the relationship between sleep quality and perceived stress levels in relation to HH among university students in the Gulf Cooperation Council (GCC) region. Our results will help tailor suitable interventions for this critical age population. We hypothesized that students who sleep for a shorter time or have poor sleep quality and high-stress levels would be more likely to develop HH than those who have adequate sleep duration and quality and low-stress levels.

2. Materials and methods

2.1. Study population and design

This study used a cross-sectional design and involved university students in the UAE and Bahrain. Participants were recruited using convenience sampling and data were collected between March and April 2021 using a self-developed online questionnaire. Links to the questionnaire were shared through the University of Sharjah's email list and social media student groups for Bahrain universities. The questionnaire was available in both Arabic and English. The English questionnaires were translated into Arabic by a qualified bilingual translator whose native language was Arabic and then back-translated to English by another qualified translator. The inclusion criteria were: students studying in universities in either BH or UAE, aged ≥18 years, and who provided consent to participate in this study. This research was approved by the Research Ethics Committee of the University of Sharjah (REC-21-03-03-08-S). A pilot study was conducted with 10 students to assess the fluidity and comprehension of the questionnaire and identify and correct any errors that may have occurred during questionnaire preparation. The pilot test participants were not included in the final sample.

2.2. Sociodemographic data

Sociodemographic data included students’ sex, age, nationality, marital status, financial status, disease status, smoking status, university, college, academic year, cumulative grade point average, and registered semester credit hours.

2.2.1. Anthropometric assessment

Anthropometric data (e.g., height and weight) were obtained from participants’ self-report. The BMI for each participant was calculated using these data.

2.2.2. Sleep quality assessment

Sleep quality was assessed using the PSQI, which is a self-reported questionnaire that focuses on sleep-related habits and experiences over the previous month [24]. Participants were given the choice to complete the PSQI in English or Arabic. The Arabic version of the PSQI used in this study was previously validated and found to be reliable [25]. In a study by Raniti et al. the PSQI had a Cronbach's alpha of 0.73 [26]. The PSQI has 19 items that cover seven subjective sleep quality components: sleep latency (the time it takes to fall asleep), sleep period (actual time spent sleeping), sleep efficiency (percentage of time spent sleeping in bed), sleep disturbances, use of sleep drugs, and daytime dysfunction [ [24,27]]. Participants rated each object on a scale from 0 (best) to 3 (worst) [24]. PSQI global scores range from 0 to 21, with a score ≥6 suggesting clinically relevant poor sleep quality and a score ≤5 indicating good sleep quality [24].

2.2.3. HH assessment

The PFS is a brief and effective scale that assesses the psychological impact of living in food-abundant climates [28,29]. It comprises 15 statements about the availability, presence, and taste of food, and tests appetite rather than the intake of palatable foods [28]. The three domains covered are: “food available” (when food is available but not physically present); “food present” (when food is physically present but has not been tasted); and “food tasted” (when food has been tasted but not yet consumed) [29]. Responses are on a 5-point Likert scale from 1 (I don't agree) to 5 (I strongly agree). For each participant, we calculated three domain scores and an aggregate score [29]. Domain scores were calculated as the mean of the question items representing the corresponding domains, and the aggregate score was the mean of the three domain scores [30]. The PFS was previously found to have sufficient internal consistency and test-retest reliability, and the scale's real-world predictive validity was assessed in a recent study [28,31]. In a study by Cappelleri et al. the PFS has a Cronbach's alpha ranging from 0.81 to 0.91 [29].

The PEMS revised scale is a 20-item self-reported questionnaire [32,33]. The scale assesses the frequency and reason behind individuals' motives for consuming tasty foods and drinks during the past year. PEMS items are categorized into four motive domains: coping motive (e.g., “To forget my worries”), reward enhancement motive (e.g., “Because I like the feeling”), social motive (e.g., “Because it helps me to enjoy a party”) and conformity motive (e.g., “To fit in with a group I like”) [34]. Participants are given examples of tasty foods and drinks (e.g., sweets, fast foods, snacks, non-alcoholic sugary drinks) and asked to choose the most appropriate response using a 5-point Likert-like scale from 1 (Never/Almost never) to 5 (Almost always/Always). Their responses were determined by how frequently these motives triggered the consumption of such foods; for example, “to deal with negative feelings” (coping motive), “to enhance positive emotions unrelated to social situations” (reward enhancement motive), “for social reasons” (social motive), and “because of external sources” (conformity motive). Furthermore, each motive domain was scored by calculating the mean score for each item. The total PEMS score is the sum of the mean scores for all motive domains and reflects the general intake of tasty foods for non-metabolic reasons [32]. PEMS motives have high internal consistency and test-retest reliability [34], and the validity of this scale was previously demonstrated [33,34]. A study by Bilici et al. determined that the PEMS had a Cronbach's alpha of 0.94 [35]. Because of poor loading on coping or any other motive, the wording of item 15 was changed from that in the original version (“Because you feel more self-confident and sure of yourself”) to “Because it helps to lower my stress” in the latest PEMS version [32], which was used in this study.

2.2.4. Stress assessment

The PSS is a self-report measure that assesses the degree to which stressful conditions in a person's life in the previous month are appraised [36]. It comprises 10 items rated on a 5-point scale (0 = never, 1 = almost never, 2 = sometimes, 3 = fairly often, 4 = very often) [37]. PSS scores are calculated by reversing the responses of four specified items (4, 5, 7, and 8) and then summing all scale items (e.g., 0 = 4, 1 = 3, 2 = 2, 3 = 1, and 4 = 0) [38]. Total PSS scores range from 0 to 40, with higher scores indicating higher perceived stress. In this study, participants were classified into three groups based on their total PSS score: low stress (scores 0–13), moderate stress (scores 14–26), and high stress (scores 27–40) [37]. The reliability and validity of the Arabic version of the PSS were previously evaluated [39] along with the PSS-10 [40]. The PSS had a Cronbach's alpha of >0.70 in a systematic review by Lee et al. [41].

2.2.5. Physical activity assessment

We used the International Physical Activity Questionnaire-Short Form (IPAQ-SF) to assess participants' physical activity levels [42]. The questionnaire is available in Arabic [43] and English [42] versions and its validity and reliability have previously been reported [44]. In a study by Moghaddam et al. the IPAQ-SF had a Cronbach's alpha of 0.7 [45]. The IPAQ-SF evaluates participants' activity levels over the last 7 days and in terms of vigorous, moderate, walking, or sitting activities [42]. Participants had to perform these physical activities for at least 10 min to be considered completed [44]. The metabolic equivalent task (MET) has been reported as a continuous variable; for scoring, the MET value for a given activity was multiplied by the minutes and the number of days per week the activity was performed [44]. Based on the category criteria, this was further reported as a categorical variable and classified as low, moderate, or high physical activity [44].

2.2.6. Sample size

Using Raosoft software, estimating a prevalence of 50% at 95% confidence with an error margin of 5% required a minimum of 377 participants. Given the lack of existing data about the relationship between HH, sleep quality, and stress among university students, a conservative 50% proportion was selected as an estimate that gives the maximum sample size.

2.3. Statistical analysis

All variables were analyzed as continuous (mean ± standard deviation [SD]), categorical variables (frequency and percentages), or both. Independent t-tests were used to analyze continuous variables, and crosstabs were used to analyze categorical variables and compare data by sex and location (UAE or Bahrain). P-values were obtained from independent t-tests (sig. 2-tailed) and Pearson's chi-square tests (sig. 2-sided).

Based on the analysis of standard residuals, the data contained no outliers (Standard Residual Min and Standard Residual Max lying between −3.29 and 3.29). A scatter plot of the standardized residuals showed that the data met the assumptions of linearity and homogeneity of variance. Additionally, the histogram and the P–P plot of standardized residuals indicated that the data were normally distributed.

Sociodemographic variables were analyzed as categorical variables (frequency and percentages) with crosstabs conducted to determine the statistical significance (p-value from Pearson's chi-square) for each variable by sex. For questionnaires measuring stress (PSS), sleep quality (PSQI), Mean Palatable Eating Motives Scale (PEMS), Power of Food Scale (PFS), and physical activity levels, responses were given a numerical value (score). Scores were thereafter analyzed as continuous variables (mean ± SD). Furthermore, two separate multiple regression tables illustrate the correlation of hedonic hunger (PEMS and PFS scores) with sleep quality (PSQI score) and with perceived stress (PSS). A third table presents the results of a logistic linear regression analysis of sleep quality and its components (PSQI and PSQI components) in relation to hedonic hunger (PFS and PEMS scores). The models in both of the multiple regression tables were adjusted for age, BMI, physical activity, and smoking. Likewise, the logistic linear regression analysis table presents two models, the first is adjusted for age and BMI alone, whereas the second adjusts for age, BMI, physical activity, and smoking. The selection of statistical methods was based on an earlier study investigating the quality and duration of sleep in relation to hedonic hunger [3].

Analyses were performed using data from both countries. The PSQI global score (0–21) and seven sleep component scores (0–3) were used to express sleep behavior (continuous variable). In addition, HH was analyzed as a continuous variable using both subscale scores (1–5) and total PEMS scores (4–20). PSS was analyzed as both a continuous variable (0–40) and a categorical variable (low, moderate, and high perceived stress) to classify students’ stress levels. Multiple linear regression was used to calculate the standardized beta coefficient, standard error, 95% confidence intervals (CI), and P-values.

First, we tested if there was a correlation between HH and sleep quality, according to sex, and then examined the correlation between HH and PSS by sex. All variables were analyzed as continuous variables. In addition, the values that expressed these correlations were calculated as crude and adjusted values. All models were adjusted for confounding factors (age, BMI, physical activity, and smoking status). Multiple logistic regression was used to calculate the odds ratio (OR), 95%CI, and P-value to assess the risk for poor sleep quality. Model 1 was adjusted for age and BMI, and Model 2 was further adjusted for physical activity level and smoking status. HH and PSS were analyzed as continuous variables, whereas the overall sleep quality and sleep components were analyzed as categorical variables. A moderated mediation pathway between stress, HH, and sleep quality was analyzed to explore the possible connection. Specifically, the direct and indirect effects of stress on sleep quality, with HH as a potential mediator. A moderation analysis using the Hayes Process macro was conducted to explore whether sex moderated the relationship between HH and the sleep quality or perceived stress. Missing data, either list- or pair-wise, were excluded from all analyses. A P-value <0.05 was considered statistically significant. Statistical analysis was conducted using IBM Corp. Released in 2020. IBM SPSS Statistics for Windows, Version 27.0. Armonk, NY: IBM Corp.

3. Results

Hedonic hunger, sleep quality, and stress were analyzed using the combined data from both countries. Data combination was justified because an independent t-test and cross-tab analyses highlighted a non-significant difference between the main components in students from the UAE and Bahrain (Supplementary Table 1). Supplementary Tables 2 and 3 show the relationships between HH, sleep, and stress by sex. Supplementary Table 4 presents the correlation matrix between all major variables. A moderation analysis highlighted a significant positive correlation between HH, sleep quality, and perceived stress for both males and females. However, the moderation analysis showed that sex did not significantly moderate the relationship between HH and the sleep quality or perceived stress. These findings suggest that the relationship between HH and sleep quality and perceived stress is not influenced by sex.

3.1. Participants characteristics

Participants’ sociodemographic characteristics by sex are presented in Table 1. The vast majority of the participants were females, with a mean age was 21.19 ± 3.8 years. Most participants were from GCC followed by Arab non-GCC countries. There was a significant (P < 0.001) difference in smoking status between females and males. The mean BMI was within the normal range (24.5 ± 6.0 kg/m2) with males having significantly (P < 0.001) greater BMI than females. The majority of participants reported moderate physical activity.

Table 1.

Sociodemographic characteristics of participating university students by sex.

Variables	All students
	Female (n = 453)	Male (n = 112)	Total (n = 565)	P-value
Age (years)	21.03 ± 3.63	21.84 ± 4.30	21.19 ± 3.78	0.040
Nationality, n (%)
UAE local	129 (28.5)	26 (23.3)	155 (27.4)	0.650
GCC	136 (30.0)	39 (34.8)	175 (31.0)
Arab non-GCC	132 (29.1)	32 (28.6)	164 (29.0)
Non-Arab	56 (12.4)	15 (13.4)	71 (12.6)
Marital status, n (%)
Single	431 (95.1)	100 (89.3)	531 (94.0)	0.040
Married	21 (4.6)	12 (10.7)	33 (5.8)
Divorced	1 (0.2)	0 (0.0)	1 (0.2)
Smoking status, n (%)
Non-smoker	424 (93.6)	78 (69.6)	502 (88.8)	<0.001
Smoker	29 (6.4)	34 (30.4)	63 (11.2)
Disease status, n (%)
No	373 (82.3)	93 (83.0)	466 (82.5)	0.850
Respiratory diseases	15 (3.3)	6 (5.4)	21 (3.7)
Cardiovascular diseases	10 (2.2)	3 (2.7)	13 (2.3)
Digestive diseases	15 (3.3)	3 (2.7)	18 (3.2)
Endocrine diseases	13 (2.9)	2 (1.8)	15 (2.7)
Other	27 (6.0)	5 (4.5)	32 (5.7)
Financial status, n (%)
Low	389 (85.9)	92 (82.1)	481 (85.1)	0.450
Medium	44 (9.7)	12 (10.7)	56 (9.9)
High	20 (4.4)	8 (7.1)	28 (5.0)
College, n (%)
Medicine and health sciences	197 (43.6)	20 (17.9)	217 (38.5)	<0.001
Humanities	105 (23.2)	41 (36.6)	146 (25.9)
Applied sciences	138 (30.5)	49 (43.8)	187 (33.2)
Graduate studies	11 (2.4)	1 (0.9)	12 (2.1)
Dual program	1 (0.2)	1 (0.9)	2 (0.4)
Education level, n (%)
Foundation year	24 (5.3)	6 (5.4)	30 (5.3)	0.080
Year 1	120 (26.5)	21 (18.8)	141 (25.0)
Year 2	75 (16.6)	25 (22.3)	100 (17.7)
Year 3	57 (12.6)	24 (21.4)	81 (14.3)
Year 4	124 (27.4)	22 (19.6)	146 (25.8)
Year 5	18 (4.0)	6 (5.4)	24 (4.2)
Year 6	1 (0.2)	1 (0.9)	2 (0.4)
Post-graduate	34 (7.5)	7 (6.3)	41 (7.3)
CGPA, n (%)
<2.0	22 (4.9)	1 (0.9)	23 (4.1)	0.001
2.0–2.4	31 (6.8)	15 (13.4)	46 (8.1)
2.5–2.9	85 (18.8)	35 (31.3)	120 (21.2)
3.0–3.5	161 (35.5)	35 (31.3)	196 (34.7)
3.6–4.0	154 (34.0)	26 (23.2)	180 (31.9)
Semester registered credit hours	(n = 412)	(n = 99)	(n = 511)	0.040
	21.05 ± 28.15	28.04 ± 41.88	22.41 ± 31.36
BMI (kg/m2)	23.93 ± 5.75	27.14 ± 6.45	24.57 ± 6.00	<0.001
BMI classification, n (%)	(n = 447)	(n = 111)	(n = 558)	<0.001
Underweight	60 (13.4)	4 (3.6)	64 (11.5)
Normal	231 (51.7)	47 (42.3)	278 (49.8)
Overweight	101 (22.6)	29 (26.1)	130 (23.3)
Obese	55 (12.3)	31 (27.9)	86 (15.4)
Physical activity, n (%)
No physical activity	134 (29.6)	32 (28.6)	166 (29.4)	0.180
Low physical activity	99 (21.9)	16 (14.3)	115 (20.4)
Moderate physical activity	158 (34.9)	42 (37.5)	200 (35.4)
High physical activity	62 (13.7)	22 (19.6)	84 (14.9)

BMI, body mass index; CGPA, cumulative grade point average; GCC, Gulf Cooperation Council; UAE, United Arab Emirates. Continuous (numeric) variables presented as mean ± standard deviation obtained from an independent t-test. Categorical (nominal and ordinal) variables are shown as frequency (percentage) obtained from a crosstabs test; P-values were obtained from independent t-tests (sig. 2-tailed) or Pearson's chi-square tests (sig. 2-sided).

BMI classification values were underweight (<18.5 kg/m²), normal weight (18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²), and obese (≥30.0 kg/m²).

Financial status currency was in Bahraini Dinar (BD) and United Arab Emirates Dirham (AED), with values categorized as low: <500 BD/<5000 AED (<1361 USD), Medium: 500–1000 BD/5000–10,000 AED (1361–2722 USD), and high: >1000 BD/>10,000 AED (>2722 USD).

3.2. Hedonic hunger, sleep quality, and stress

Descriptive statistics for HH, sleep quality, and stress levels for all participating students are presented in Table 2. The mean total PEMS score was 10.95 ± 3.00, and mean subscale scores were 2.88 ± 1.15 for coping, 3.14 ± 1.02 for reward enhancement, 3.15 ± 1.01 for social, and 1.77 ± 0.86 for conformity. Mean scores for the PFS domains were: 2.99 ± 1.69 for food available, 3.40 ± 1.14 for food present, and 3.30 ± 0.97 for food tasted, with a mean aggregated score of 3.21 ± 0.96. The PSQI global mean score was 7.78 ± 3.38 points, and the majority of students reported poor sleep quality. The mean PSS-10 score was 23.09 ± 6.98, and most students experienced moderate or high stress.

Table 2.

Hedonic hunger, sleep quality, and stress level scores for participating university students.

Characteristics	Total Students
	(n = 565)
Hedonic hunger (mean ± SD)
PEMS (mean ± SD)
Coping motive	2.88 ± 1.15
Reward enhancement motive	3.14 ± 1.02
Social motive	3.15 ± 1.01
Conformity motive	1.77 ± 0.86
Total PEMS score	10.95 ± 3.00
PFS (mean ± SD)
Food available	2.99 ± 1.69
Food present	3.40 ± 1.14
Food tasted	3.30 ± 0.97
Aggregated score	3.21 ± 0.96
Sleep components (PSQI)
Subjective sleep quality (mean ± SD)	1.44 ± 0.91
Adequate subjective sleep quality, n (%)a	345 (61.1)
Inadequate subjective sleep quality, n (%)a	220 (38.9)
Sleep latency (mean ± SD)	1.62 ± 1.03
Adequate sleep latency, n (%)a	257 (45.5)
Inadequate sleep latency, n (%)a	308 (54.5)
Sleep duration (mean ± SD)	1.03 ± 0.98
Adequate sleep duration, n (%)a	422 (74.7)
Inadequate sleep duration, n (%)a	143 (25.3)
Sleep efficiency (mean ± SD)	0.66 ± 1.01
Adequate sleep efficiency, n (%)a	464 (82.1)
Inadequate sleep efficiency, n (%)a	101 (17.9)
Sleep disturbance (mean ± SD)	1.32 ± 0.57
Little/No sleep disturbance, n (%)a	379 (67.1)
Sleep disturbance, n (%)a	186 (32.9)
Use of sleep medication (mean ± SD)	0.23 ± 0.64
Lack of need for sleep medication, n (%)a	529 (93.6)
Need for sleep medication, n (%)a	36 (6.4)
Daytime dysfunction (mean ± SD)	1.48 ± 0.91
Less daytime dysfunction, n (%)a	295 (52.2)
More daytime dysfunction, n (%)a	270 (47.8)
Global PSQI (mean ± SD)	7.78 ± 3.38
Sleep quality ≤5, n (%)a	164 (29.0)
Sleep quality >5, n (%)a	401 (71.0)
Perceived Stress Scale (PSS)
Total stress level (mean ± SD)	23.09 ± 6.98
Low stress, n (%)	45 (8.0)
Moderate stress, n (%)	342 (60.5)
High perceived stress, n (%)	178 (31.5)

PEMS, Palatable Eating Motive Scale; PFS, Power of Food Scale; PSQI, Pittsburgh Sleep Quality Index; SD, standard deviation.

Continuous (numeric) variables are presented as mean ± standard deviation obtained from independent t-test. Categorical (nominal and ordinal) variables shown as frequency (percentage) obtained from the crosstabs test; P-values obtained from independent t-tests (sig. 2-tailed) or Pearson's chi-square tests (sig. 2-sided).

^aTotal PSQI score and sleep behavior components were analyzed as categorical variables; PSQI global score (0–21) and seven sleep components scores (0–3) were used to express sleeping behavior as continuous variables. In addition, the total PSQI score and sleep behavior components were analyzed as categorical variables. These scores were categorized into two outcomes: total sleep quality was categorized as good sleep quality (≤5) or poor sleep quality (>5). Sleep components were categorized as either adequate or inadequate; subjective sleep quality as adequate (high and medium sleep quality, score of 0 or 1) or inadequate (low and very low sleep quality, score of 2 or 3). Sleep latency was categorized as adequate (very short and short sleep latency, score of 0 or 1) or inadequate (medium and long sleep latency, score of 2 or 3). Sleep duration was categorized as adequate (6–7 h and >7 h, score of 0 or 1) or inadequate (5–6 h and <5 h, score of 2 or 3). Sleep efficiency was categorized as adequate (high and medium sleep efficiency, score of 0 or 1) or inadequate (low and very low sleep efficiency, score of 2 or 3). Sleep disturbance was categorized as adequate (no or low sleep disturbances, score of 0 or 1) or inadequate (medium and high sleep disturbances, score of 2 or 3). Need for medication was categorized as adequate (no and low use, score of 0 or 1) or inadequate (medium and high use, score of 2 or 3). Daytime dysfunction was categorized as less (none and low day dysfunction, score of 0 or 1) or more (medium and high day dysfunction, score of 2 or 3).

3.3. Relationship between HH and sleep

The relationships between HH and PSQI scores using multiple linear regression analysis are shown in Table 3. PEMS scores showed a significant (P < 0.001) positive association with PSQI scores after full adjustment, and the coefficient of the determinant was 2.7%. In both the crude and adjusted models “reward,” “coping,” and “conformity” motives were positively associated with PSQI scores. Total PFS scores were positively correlated with PSQI scores after controlling for confounding factors (P < 0.001), and the coefficient of determination was 0.50%. In addition, the three domain scores (food available, food present, and food tasted) showed positive associations with the total PSQI score after adjustment for confounding factors (food available (P<0.001); food present (P < 0.001); food tasted (P<0.001).

Table 3.

Multiple linear regression analysis of the relationship between hedonic hunger (PEMS and PFS subscale scores) and total PSQI score.

Hedonic hunger		Total (N = 565)
		β	SE	95%CI	P-value
PEMS
Coping motive	Crude	0.17	0.12	0.25–0.73	<0.001
	Adjusteda	0.17	0.13	0.24–0.74	<0.001
Reward enhancement motive	Crude	0.10	0.14	0.06–0.61	0.020
	Adjusted	0.10	0.14	0.04–0.60	0.020
Social motive	Crude	0.06	0.11	−0.09–0.46	0.180
	Adjusted	0.06	0.14	−0.09–0.46	0.190
Conformity motive	Crude	0.10	0.17	0.06–0.71	0.020
	Adjusted	0.10	0.17	0.06–0.72	0.020
Total PEMS	Crude	0.14	0.05	0.07–0.25	<0.001
	Adjusted	0.14	0.05	0.06–0.25	<0.001
PFS
Food available	Crude	0.17	0.08	0.17–0.50	<0.001
	Adjusted	0.16	0.09	0.16–0.50	<0.001
Food present	Crude	0.17	0.12	0.27–0.76	<0.001
	Adjusted	0.17	0.13	0.27–0.75	<0.001
Food tasted	Crude	0.12	0.15	0.14–0.72	<0.001
	Adjusted	0.12	0.15	0.13–0.71	<0.001
Aggregated factor	Crude	0.21	0.15	0.45–1.02	<0.001
	Adjusted	0.21	0.15	0.45–1.04	<0.001

PEMS, Palatable Eating Motive Scale; PFS, Power of Food Scale; PSQI, Pittsburgh Sleep Quality Index; β, standardized beta coefficient; SE, standard error; CI, confidence interval.

Standardized beta coefficients, standard errors, 95% confidence intervals, and P-values were obtained from a multiple linear regression analysis.

^aAll the models adjusted for age, body mass index, physical activity, and smoking.

Table 4 shows the results of the multiple logistic regression that was performed to determine the ability of HH to predict the outcome of overall poor sleep quality (PSQI) and poor sleep components among participating students. Model 1 was adjusted for age and BMI, and Model 2 was additionally adjusted for physical activity and smoking status. Model 2 showed that poor sleep quality was positively associated with PEMS scores; the estimated OR favored an 8% increase (P = 0.020) in poor sleep quality for each one-unit increase in HH. There was also a 7% increase (P = 0.030) in having short sleep duration (<5 h or 5–6 h) among students with high PEMS scores compared with those who slept for 6–7 h or >7 h (reference group). Students with increased HH were 12% more likely to have medium/high sleep disturbances (OR = 1.12, P < 0.001) than those with no/low sleep disturbances (reference group). There was an increased likelihood (14%, P = 0.020) of students with high PEMS scores using medications to sleep compared with the reference group. Students with high PEMS scores had 8% higher odds (OR = 1.08, P = 0.010) of increased medium/high daytime dysfunction.

Table 4.

Multiple logistic regression models for the associations between total PSQI and PSQI components with hedonic hunger (PFS and PEMS) among participating students (n = 558).

PSQI and sleep components		Total sleep quality		Subjective sleep quality		Sleep latency		Sleep duration		Sleep efficiency		Sleep disturbances		Use sleep medications		Daytime dysfunction
		OR (95%CI)	P-value	OR (95%CI)	P-value	OR (95%CI)	P-value	OR (95%CI)	P-value	OR (95%CI)	P-value	OR (95%CI)	P-value	OR (95%CI)	P-value	OR (95%CI)	P-value
Hedonic hunger	Total PEMS
	Crude
	Good quality	1 [Reference]	0.010	1 [Reference]	0.260	1 [Reference]	0.420	1 [Reference]	0.040	1 [Reference]	0.470	1 [Reference]	<0.001	1 [Reference]	0.010	1 [Reference]	<0.001
	Poor quality	1.08 (1.02–1.15)		1.03 (0.98–1.09)		1.02 (0.97–1.08)		1.07 (1.00–1.14)		0.97 (0.91–1.05)		1.13 (1.06–1.20)		1.16 (1.04–1.30)		1.08 (1.03–1.15)
	Model 1
	Good quality	1 [Reference]	0.020	1 [Reference]	0.370	1 [Reference]	0.450	1 [Reference]	0.030	1 [Reference]	0.763	1 [Reference]	<0.001	1 [Reference]	0.010	1 [Reference]	0.010
	Poor quality	1.08 (1.01–1.15)		1.03 (0.97–1.09)		1.02 (0.97–1.08)		1.07 (1.01–1.15)		0.99 (0.92–1.06)		1.13 (1.06–1.20)		1.15 (1.03–1.29)		1.08 (1.02–1.14)
	Model 2
	Good quality	1 [Reference]	0.020	1 [Reference]	0.390	1 [Reference]	0.460	1 [Reference]	0.030	1 [Reference]	0.760	1 [Reference]	<0.001	1 [Reference]	0.020	1 [Reference]	0.010
	Poor quality	1.08 (1.01–1.15)		1.03 (0.97–1.09)		1.02 (0.96–1.08)		1.07 (1.00–1.15)		0.99 (0.92–1.06)		1.12 (1.06–1.20)		1.14 (1.02–1.28)		1.08 (1.02–1.14)
	Total PFS – Aggregated score
	Crude
	Good quality	1 [Reference]	<0.001	1 [Reference]	0.030	1 [Reference]	0.220	1 [Reference]	0.090	1 [Reference]	0.870	1 [Reference]	<0.001	1 [Reference]	0.030	1 [Reference]	<0.001
	Poor quality	1.44 (1.19–1.75)		1.22 (1.02–1.47)		1.12 (0.94–1.33)		1.19 (0.97–1.46)		1.02 (0.81–1.28)		1.62 (1.33–1.98)		1.52 (1.04–2.21)		1.55 (1.29–1.86)
	Model 1
	Good quality	1 [Reference]	<0.001	1 [Reference]	0.050	1 [Reference]	0.230	1 [Reference]	0.070	1 [Reference]	0.430	1 [Reference]	<0.001	1 [Reference]	0.050	1 [Reference]	<0.001
	Poor quality	1.43 (1.17–1.75)		1.20 (1.00–1.45)		1.12 (0.93–1.33)		1.21 (0.98–1.49)		1.10 (0.87–1.39)		1.60 (1.31–1.95)		1.45 (1.00–2.12)		1.51 (1.26–1.82)
	Model 2
	Good quality	1 [Reference]	<0.001	1 [Reference]	0.060	1 [Reference]	0.230	1 [Reference]	0.080	1 [Reference]	0.450	1 [Reference]	<0.001	1 [Reference]	0.060	1 [Reference]	<0.001
	Poor quality	1.43 (1.17–1.74)		1.20 (0.99–1.44)		1.12 (0.93–1.34)		1.21 (0.98–1.49)		1.10 (0.87–1.38)		1.60 (1.31–1.95)		1.44 (0.99–2.01)		1.51 (1.26–1.82)

PSQI, Pittsburgh Sleep Quality Index; PEMS, Palatable Eating Motive Scale; PFS, Power of Food Scale; OR, odds ratio; CI, confidence interval.

ORs, 95%CIs, and P-values were obtained from multiple logistic regression.

Model 2 also showed that poor sleep quality was positively associated with PFS score, with a 43% increased likelihood (OR = 1.43, P < 0.001) of having poor sleep quality as HH increased. Participants with high PFS scores had a 60% increased likelihood of having medium/high sleep disturbances (OR = 1.60, P < 0.001) compared with the reference group. High HH was associated with a 51% increased likelihood (OR = 1.15, P < 0.001) of having medium/high daytime dysfunction.

3.4. Relationship between HH and stress

Table 5 shows the multiple linear regression analysis of the relationship between HH and PSS scores. There was a positive association between total PEMS and PSS scores (P < 0.001), with those with high PSS scores being more likely to consume palatable foods for coping motives (P < 0.001). The relationship between PFS aggregated scores and PSS scores was positive (P < 0.001). Students with high PSS scores were more likely to consume palatable foods when food was present.

Table 5.

Multiple linear regression analysis of the relationship between hedonic hunger (PEMS and PFS subscale scores) and total Perceived Stress Scale scores.

Hedonic hunger		Total (n = 565)
		β	SE	95%CI	P-value
PEMS
Coping motive	Crude	0.27	0.25	1.12 to 2.10	<0.001
	Adjusteda	0.26	0.25	1.07 to 2.07	<0.001
Reward enhancement motive	Crude	0.14	0.29	0.37 to 1.50	<0.001
	Adjusted	0.13	0.29	0.31 to 1.44	<0.001
Social motive	Crude	0.13	0.29	0.31 to 1.45	<0.001
	Adjusted	0.13	0.29	0.30 to 1.42	<0.001
Conformity motive	Crude	0.04	0.34	−0.33 to 1.03	0.310
	Adjusted	0.04	0.34	−0.34 to 0.99	0.340
Total PEMS	Crude	0.20	0.10	0.28 to 0.66	<0.001
	Adjusted	0.19	0.10	0.26 to 0.63	<0.001
PFS
Food available	Crude	0.20	0.17	0.49 to 1.16	<0.001
	Adjusted	0.19	0.17	0.45 to 1.13	<0.001
Food present	Crude	0.22	0.25	0.85 to 1.84	<0.001
	Adjusted	0.22	0.25	0.83 to 1.81	<0.001
Food tasted	Crude	0.13	0.30	0.31 to 1.51	<0.001
	Adjusted	0.12	0.30	0.25 to 1.44	<0.001
Aggregated factor	Crude	0.23	0.30	1.08 to 2.25	<0.001
	Adjusted	0.23	0.30	1.04 to 2.22	<0.001

PEMS, Palatable Eating Motive Scale; PFS, Power of Food Scale; β, standardized beta coefficient; SE, standard error; CI, confidence interval.

Standardized beta coefficients, standard errors, 95% confidence intervals, and P-values were obtained from a multiple linear regression analysis.

^aAll the models were adjusted for age, body mass index, physical activity, and smoking.

3.5. Relationship between stress and sleep mediated by HH

Fig. 1 shown below demonstrates a mediation pathway between stress, HH, and sleep quality. It is indicated that the total effect size of stress (PSS) on sleep quality (PSQI score) was statistically significant (β = 0.35, P < 0.001). Performing the Sobel test to explore the indirect effect of stress on sleep with PEMS as a mediator did not produce a statistically significant result, however; based on multiple regression conducted using stress and PEMS as predictors of sleep quality, stress was found to have a statistically significant direct effect on sleep quality (β = 0.34, P < 0.001). The same tests were conducted to measure the effect of PFS as a mediator between stress and sleep quality. Similarly, multiple regression using PFS and stress as predictors indicated that stress has a statistically significant direct effect on sleep quality (β = 0.32, P < 0.001). However, Sobel test results did indicate a significant indirect effect of PFS as a mediator in the relationship (z = 2.86, P = 0.050). Overall, these findings suggest that hedonic hunger may play a role in the relationship between stress and sleep, with the desire for palatable foods potentially mediating this relationship.

Fig. 1.

Standardized beta coefficients for the relationship between perceived stress and sleep quality as mediated by hedonic hunger.

*p < 0.001

^aon PEMS,

^bon PFS

^cPEMS with stress,

^dPFS with stress

^ewith PEMS,

^fwith PFS.

The total PSQI score and sleep behavior components were analyzed as categorical variables, PSQI global score (0–21) and seven sleep components scores (0–3) were used to express sleeping behavior as continuous variables. In addition, the total PSQI score and sleep behavior components were analyzed as categorical variables. They were categorized into two outcomes: total sleep quality categorized as good sleep quality (≤5) and poor sleep quality (>5). Sleep components were categorized as either adequate or inadequate; subjective sleep quality as adequate (high and medium sleep quality, score of 0 or 1) or inadequate (low and very low sleep quality, score of 2 or 3). Sleep latency was categorized as adequate (very short and short sleep latency, score of 0 or 1) or inadequate (medium and long sleep latency, score of 2 or 3). Sleep duration was categorized as adequate (6–7 h and >7 h, score of 0 or 1) or inadequate (5–6 h and <5 h, score of 2 or 3). Sleep efficiency was categorized as adequate (high and medium sleep efficiency, score of 0 or 1) or inadequate (low and very low sleep efficiency, score of 2 or 3). Sleep disturbance was categorized as adequate (none or low sleep disturbances, score of 0 or 1) or inadequate (medium and high sleep disturbances, score of 2 or 3). Need for medication was categorized as adequate (none and low use, score of 0 or 1) or inadequate (medium and high use, score of 2 or 3). Daytime dysfunction was categorized as less (none and low day dysfunction, score of 0 or 1) or more (medium and high day dysfunction, score of 2 or 3).

Model 1: adjusted for age and body mass index.

Model 2: additionally, adjusted for physical activity and smoking status.

4. Discussion

The present study was the first to investigate the relationship between sleep quality and stress in relation to HH among university students in the UAE and BH, with consideration of possible confounding factors. We found strong positive associations between high HH, poor sleep quality, and high-stress levels. Food consumption is regulated by two different pathways: homeostatic and hedonic. Hedonic pathways indicate that there is an increased desire to consume highly palatable foods during periods of energy abundance [2]. Lutter and Nestler noted that highly palatable foods elicit responses in the mesolimbic dopamine pathway; these foods induce the release of dopamine, which is thought to coordinate food reward processes [2].

The findings of the current work are consistent with those found in Turkey among university students, where Açik and colleagues (2021) reported a positive relationship between increased hedonic hunger and poor sleep quality, while an inverse relationship was observed between ideal sleep duration and hedonic hunger [3]. Other two studies conducted among college students in the US showed that students consumed palatable foods for social motives more than for other PEMS motives [ [32,34]]. Similarly, we found that social motive scores were significantly higher than those of the other motives. A previous study found a positive relationship between HH and neural responsivity in brain regions associated with oral somatosensory activity during the intake of highly palatable foods and increased motivation for the consumption of these foods [46]. The mean total PEMS score in our study was higher than that of known studies combined. This difference may be explained by the fact that the number of restaurants and fast-food chains has expanded dramatically in all Middle Eastern countries, which has led to an increased number of people eating out. Therefore, individuals from these countries may be more inclined to consume fat-and carbohydrate-rich foods [47].

Furthermore, our results were consistent with a study conducted among college students in the US that showed a positive association between PFS subscales scores and motivation to consume highly palatable foods [28]. Previous studies that evaluated neural activity in food-seeking and reward-related brain regions found hedonic hunger was associated with heightened urges to eat, regardless of hunger conditions [1]. A literature review highlighted that higher PFS scores were related to greater activity in the postcentral gyrus areas in the brain, which is linked to both somatosensory processing of food cues and obesity [1].

Decreased quality and quantity of sleep are thought to increase HH by increasing ghrelin and decreasing leptin, thereby increasing the prevalence of obesity [48]. A previous study reported that PEMS scores were higher in those with both short and long sleep durations compared with ideal sleep duration; long periods of sleep may increase the bodily inflammatory response, which may lower satiety in adipocytes and hunger hormones in the brain [49]. This mechanism may help to explain how long sleep duration contributes to obesity and HH [50]. Another study among university students in Turkey concluded that improving sleep quality and duration may help lower HH [3]. The decline in systemic insulin sensitivity and glucose use has been associated with low satiety hormone levels [50,51]. Consistent with previous outcomes, we found a positive association between HH and poor sleep quality; a higher PSQI score (poor sleep quality) was associated with a higher coping motive score.

Physical or emotional discomfort has been linked to increased consumption of highly palatable foods [19]. This may be explained by high cortisol levels in conjunction with high insulin levels [52]. Ghrelin, a hunger hormone, may also play a role, as foods high in fat and sugar appear to contribute to feelings of comfort, which impacts individuals to consume food when feeling stressed [19]. Based on a UAE study involving female university students, moderate stress was associated with high consumption of fast food and soft drinks and low consumption of fresh fruits and vegetables [23]. The eating habits of individuals are thought to change when they are stressed [53]. Our study showed a significant positive association between high PSS and HH (PEMS and PFS) scores. A comparison of the motive domains showed that most students consumed highly palatable foods to cope with negative emotions when they felt highly stressed. A study from Saudi Arabia revealed that PSS was associated with unhealthy changes in eating patterns; female students reported an increased preference for sweets under stress, whereas males preferred consuming fast food [54]. However, to our knowledge, no study has directly examined the relationship between HH (PEMS and PFS) and PSS, meaning we cannot directly compare our findings with other studies.

This study had several strengths. To our knowledge, this was the first research conducted in the GCC region that analyzed the relationship between HH, sleep, and stress among university students. Our sample was also sufficient for a cross-sectional study (565 participants) and included students from 33 different universities from two countries in the GCC region (UAE and BH), whereas many previous studies focused on one university in one country. However, some inherent limitations should be considered when interpreting our results. As the data were self-reported, it is possible that participants misreported or misclassified their height, weight, sleep, stress, and HH because of recall bias. Furthermore, students were self-determined to participate in the analysis, as in other online survey studies. Due to the convenience sampling technique followed, the findings obtained may therefore not be generalizable to all university students, and the validity of applying the conclusions of our study outside the context of that study is not applicable. Another possible limitation was the difference in the number of female and male participants; however, this ratio mirrored the actual ratio of females to males in UAE and Bahrain universities, which is approximately 2/3:1/3. In addition, this study was conducted during the COVID-19 pandemic, which may have contributed to errors. Finally, the cross-sectional design of this study means that causality cannot be deduced. Thus, only correlation/association, non-causality, and relationships could be drawn from the current work.

5. Conclusion and recommendations

Due to the bidirectional relationship of the examined outcomes, it can be concluded that reducing HH and stress levels may help to enhance sleep quality, conversely, improving sleep quality and reducing stress levels may help alleviate HH among university students.

It is recommended to conduct well-controlled clinical trials investigating the relationship between hormones and related genes that control the circadian rhythm are warranted to better understand this relationship. For example, biochemical tests (blood or urine tests) may be conducted to objectively assess levels of sleep and appetite-regulating hormones, such as leptin, ghrelin, cortisol, and melatonin. Additional experimental intervention research that controls for various confounding and interfering factors should be conducted to further explore the direct effect of poor sleep quality and high-stress levels on HH. Altering dietary and lifestyle behaviors may be essential for improving sleep quality and relieving stress among university students.

Author contribution statement

Narmin Khaled Abdulla, B.Sc.: Performed the experiments; Analyzed and interpreted the data; Wrote the paper.

Reyad R. Obaid, Ph.D.: Contributed reagents, materials, analysis tools or data.

Mariam N. Qureshi, B.Sc.; Amna A. Asraiti, B.Sc.; Maryam A. Janahi, B.Sc.: Performed the experiments; Wrote the paper.

Salma J. Abu Qiyas, B.Sc.: Analyzed and interpreted the data.

MoezAlIslam E. Faris, Ph.D.: Conceived and designed the experiments; Performed the experiments; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement

Data will be made available on request.

Declaration of interest’s statement

I, the undersigned author of the above-mentioned case report, hereby declare the following:

• 1.I have obtained written informed consent from the patient for the publication of this case report, any accompanying data and images. Where consent was obtained from someone other than the patient, I confirm that this proxy was authorised to provide consent on the patient's behalf.
• 2.Where the patient is a minor, we followed local laws on the age and circumstances under which they may consent for themselves. If they were not of legal age to consent, consent was obtained from an authorised proxy i.e. the parents or legal guardian. If the minor has reasonable understanding of the informed consent and implications, signature was also obtained from the minor.
• 3.Where the patient provided consent themselves, I confirm that they had capacity to do so, and any mental or physical disabilities were taken into consideration in the process of informing and obtaining written consent.
• 4.Where the patient has died, I confirm that the consent given still allows for publication.
• 5.I confirm that all content presented in this case report, associated data and images have been anonymized to the best possible extent.
• 6.The original signed and dated consent form is held by the treating institution and will be retained according to institutional policies and procedures.
• 7.The informed consent form (please do not include with your submission) includes the patient's name, age, medical history, diagnosis, treatment, and any other relevant information.
• 8.The patient/authorized proxy have been fully informed of the purpose of this case report, the potential risks and benefits of publication, and the consequence of disclosing their personal information.
• 9.The patient or authorized proxy has been given the opportunity to ask questions regarding publication of the case report, has had their questions answered fully and has approved the final version of the manuscript, all associated data and images prior to publication.
• 10.The patient or legal guardian(s) has been informed that their consent and participation in the publication of this case report is entirely voluntary. They have been informed that they have the right to withdraw their consent at any time.
• 11.I confirm that at least one of the authors of this paper was involved in the care of the patient.
• 12.I confirm that my article complies with the local law on consent and privacy.

By signing this declaration form, I acknowledge that I have read and understood the information provided above, and I attest to the accuracy of this declaration. I understand that any false or misleading information may result in the rejection of the case report or other disciplinary actions.

As corresponding author, I hereby declare that I sign this document on behalf of all the authors of the above-mentioned case report.

Appendix A. Supplementary data

The following is the Supplementary data to this article: