Sleep quality is a complex, multidimensional construct that encompasses both quantitative aspects – such as sleep duration, efficiency and latency—and subjective perceptions of how restorative and satisfying the sleep experience is. It serves as a fundamental pillar of biological function, with significant associations with various dimensions of human health(Reference Buysse, Reynolds and Monk1). Maintaining high sleep quality is essential for metabolic regulation, immune system integrity and cognitive performance. Conversely, chronic poor sleep quality is a major modifiable risk factor for several debilitating health problems, including obesity, diabetes, hypertension, CVD, stroke and mental health issues such as anxiety and depression(Reference Grandner2,Reference Medic, Wille and Hemels3) . Around 70 % of American adults experience poor-quality sleep monthly, and sleep-related issues affect approximately 50–70 million people(4). Just 35 % of adults achieve the recommended 7–9 h of sleep per night(Reference Knutson, Phelan and Paskow5) and mental health issues(Reference Ohayon6). There are several modifiable behavioural risk factors for poor sleep quality, such as cigarette smoking, alcohol consumption(Reference Bartel, Gradisar and Williamson7), caffeine intake(Reference Clark and Landolt8) and poor diet(Reference St-Onge, Roberts and Shechter9). While many studies support the impact of sleep on diet, the influence of eating behaviours on sleep quality has been less studied(Reference Duan, Gu and Polotsky10).
Dietary modifications may offer a promising new approach to improving sleep quality(Reference Wilson, St-Onge and Tasali11). A cross-sectional study found that eating habits and meal timing were strongly linked to sleep quality(Reference Faris, Vitiello and Abdelrahim12). Another study found that eating closer to bedtime negatively affects sleep quality in healthy individuals, particularly women(Reference Crispim, Zimberg and dos Reis13). A study on night-shift workers indicates that the timing and composition of the last meal before bedtime could be crucial key factors for good diurnal (daytime) and nocturnal (nighttime) sleep for those who work night shifts(Reference Nogueira, Pellegrino and Cipolla-Neto14). A narrative review also reported that individuals with nocturnal lifestyles often experience insufficient sleep, particularly those who replace regular meals with snacks and consume most of their food late in the evening or at night(Reference Peuhkuri, Sihvola and Korpela15). In a systematic review, it was concluded that consuming healthy foods was linked to improved sleep quality, while consuming more processed and high-sugar foods was linked to poorer sleep quality(Reference Godos, Grosso and Castellano16). A study on sleep apnoea patients concluded that increased food consumption in the evening may reduce the sleep quality in patients with moderate and severe sleep apnoea(Reference de Melo, del Re and dos Santos Quaresma17). Conversely, several studies have reported no significant association between meal timing and sleep quality(Reference Duan, Gu and Polotsky10,Reference Luz, da Fonseca and Santos18,Reference van Egmond, Moulin and Schiöth19) . Emerging evidence reveals key physiological mechanisms linking eating behaviours to sleep quality. Central to this connection are hormones such as insulin, leptin and ghrelin, which regulate nutrient metabolism in alignment with circadian rhythms(Reference Bae, Fang and Rustgi20). Nighttime eating may disrupt these metabolic processes, impairing sleep quality due to their intrinsic relationship with the sleep-wake cycle(Reference Amaral, Andrade-Silva and Kuwabara21). Furthermore, research demonstrates that specific nutrients – including glucose, amino acids, Na ethanol and caffeine – along with meal timing can directly reset circadian rhythms in rodent models(Reference Froy22).
Meal timing patterns may also reflect cultural norms. In Iran, meal schedules differ from those in many Western countries. Lunch is typically consumed between approximately 12.00 and 15.00, while dinner is often eaten relatively late in the evening, commonly between about 20.00 and 22.00, and in some households even later(Reference Zeraattalab-Motlagh, Lesani and Janbozorgi23). Despite the accumulating evidence on diet–sleep interactions, a significant knowledge gap persists regarding how cultural dietary architectures and meal timing influence sleep quality – particularly within non-Western populations. While previous research has predominantly focused on Western cohorts(Reference Peuhkuri, Sihvola and Korpela15,Reference St-Onge, Grandner and Brown24–Reference St-Onge26) , the impact of traditional eating behaviours and regional meal schedules remains largely underexplored. This study addresses this gap by examining the interplay between the timing and nutritional composition of the last meal within the Iranian context, where late-night dining and specific culinary traditions are normative(Reference St-Onge, Ard and Baskin27–Reference Almoosawi, Prynne and Hardy29). By elucidating these relationships in a population with distinct dietary customs, our findings aim to facilitate the development of culturally tailored nutritional strategies to enhance sleep quality.
Subjects and methods
Study design and participants
This cross-sectional study was conducted from October 2021 to February 2023 among 450 apparently healthy adults (aged 20–59 years) recruited from Tehran University’s Jahad Daneshgahi center. Participants were selected through convenience sampling due to COVID-19 restrictions. All participants provided self-reported health status confirmation and completed the study requirements. For calculating a sample size of 450 participants, we used the following formula: n = z2 p (1–p)/d2(Reference Daniel30); where z is the statistic corresponding to the level of confidence (95 %), p is the expected prevalence (0·25) and d is the precision (0·04)(Reference Mirghaed, Gorji and Panahi31). The inclusion criteria were individuals aged 20–59 years and residents of Tehran. Exclusion criteria included pregnancy or lactation and adherence to any specific dietary plan that could independently influence sleep quality or metabolic markers. Specifically, individuals following therapeutic diets (e.g. for managing diabetes or hypertension), restrictive weight-loss diets (e.g. ketogenic and very-low-calorie diets), or specialised regimens (e.g. intermittent fasting, veganism or high-protein diets) were excluded. The rationale for this exclusion was to eliminate potential confounding effects of extreme macronutrient distributions or altered eating windows on sleep quality and meal timing assessments, thereby ensuring that the observed associations reflect habitual dietary patterns rather than clinical or structured interventions. This study was approved by the Ethics Committee of Tehran University of Medical Sciences (IR.TUMS.Medicine.REC.1401.424). All study participants received written informed consent before entering the study. The study methods followed the ethical principles outlined in the Declaration of Helsinki.
Dietary intake assessment
Three non-consecutive 24-h dietary recalls (including two weekdays and one weekend day) were collected by certified dietitians following standard multiple-pass methodology. After an initial in-person interview at the clinic, two additional recalls were conducted by phone on randomly assigned days. For each recall, participants reported all food/beverage consumption with exact clock times, enabling calculation of the last meal-to-sleep interval (time between final energy consumption and self-reported bedtime)(Reference Lesani, Jayedi and Karimi32). Daily food intakes from three 24-h dietary recalls were converted into grams per day using household measures(Reference Ghafarpour, Houshiar-Rad and Kianfar33). To achieve this, the Nutritionist IV software (First Databank) was used, which had been adapted for Iranian foods, to calculate the energy and nutrient intake for each day of the study.
Data collection
Data collection was performed through structured face-to-face interviews conducted by trained researchers using a validated lifestyle and demographic questionnaire. The comprehensive assessment encompassed three primary domains:
-
(1) Socio-demographic factors: including age, sex, marital status, educational attainment, occupational status and living arrangements.
-
(2) Lifestyle behaviours: including tobacco use, alcohol consumption and physical activity levels. Special attention was given to dietary supplement intake, which was defined as the habitual use of any supplements over the past month. This included general health-promoting supplements (e.g. multivitamins, n-3 fatty acids and minerals) as well as those potentially influencing sleep–wake cycles or metabolic health (e.g. Mg, melatonin or herbal extracts). Supplement use was subsequently categorised as a binary variable (user v. non-user) for inclusion in the multivariable models to control for potential confounding effects.
-
(3) Medical history: including a detailed record of chronic disease status and medication use to ensure adherence to the study’s inclusion and exclusion criteria.
Physical activity
We used the validated International Physical Activity Questionnaire to measure physical activity(Reference Vasheghani-Farahani, Tahmasbi and Asheri34). Participants recalled time spent in walking, moderate-intensity and vigorous-intensity activities during the previous week. Total physical activity was calculated in metabolic equivalent of task-minutes/week and categorised into three levels: low (<600), moderate (600–3000) and high (>3000).
Anthropometric measurements
All measurements were obtained under standardised conditions following an overnight fast. Weight was measured to the nearest 0·1 kg using a calibrated digital scale (Seca) with participants wearing minimal clothing and no footwear. Height was recorded to the nearest 0·5 cm using a wall-mounted stadiometer while participants stood barefoot with their back straight and heels together. BMI was calculated as weight in kilograms divided by height in metres squared (kg/m2) and classified according to WHO criteria:
Underweight: <18·5 kg/m2, normal weight: 18·5–24·9 kg/m2, overweight: 25·0–29·9 kg/m2 and obesity: ≥30·0 kg/m2. Body composition was assessed using the Tanita BC-418 bioelectrical impedance analyser (Tanita Corporation). Participants were instructed to: Maintain an overnight fast, avoid strenuous physical activity for 24 h prior and abstain from water intake for 2 h before measurement.
Sleep quality
We assessed sleep quality using the validated Pittsburgh Sleep Quality Index (PSQI), a 19-item questionnaire evaluating seven components over the past month: sleep quality, latency, duration, efficiency, disturbances, sleep medication use and daytime dysfunction. Each item is scored 0–3 (higher scores = worse sleep), with a global score range of 0–21 (scores >5 indicate poor sleep). The Persian version demonstrated good reliability (Cronbach’s α = 0·83) and validity in Iranian populations, aligning with the original validation by Buysse et al. (1989). Participants completed the supervised questionnaire, and our data showed strong internal consistency (α = 0·79)(Reference Buysse, Reynolds and Monk1).
Assessment of circadian rhythm
The morningness-eveningness questionnaire was used(Reference Rahafar, Meysam and Sadeghpour35). It is a self-assessment scale designed to measure an individual’s circadian rhythm preference. It consists of nineteen questions that assess lifestyle habits, sleep patterns and wakefulness. Participants rate their preferences, and the scores range from 16 to 86. A higher score indicates greater morningness, while a lower score suggests eveningness.
Meal quality
In this study, we used a nutrient-rich food index to assess the nutrient density of meals as a measure of dinner quality. The nutrient-rich food index was developed and validated by Fulgoni and colleagues (2010), known as NRF9·3, utilising algorithms that have the strongest predictive relationship with diet quality. The NRF9·3 index is based on nine positive nutrients (protein, fibre, vitamins A, C and D, Ca, Fe, potassium and Mg) and three negative nutrients (saturated fat, added sugars or total sugars and sodium). The index is calculated by summing the percentage of daily values for the nine positive nutrients and subtracting the percentage of maximum recommended values for the three negative nutrients – all measured per 100 kcal or a standard reference amount, with values capped at 100 %(Reference Fulgoni, Keast and Drewnowski36).
Definition of meal timing and time interval from the last meal to bed
In accordance with prior studies, breakfast was defined as a substantial eating event occurring between 05.00 and 11.00. Lunch was defined as the primary meal consumed between 11.00 and 16.00, and dinner as the primary meal consumed between 16.00 and 23.00. These specific time intervals were adopted to facilitate direct comparisons with findings from published research. The time interval from the last meal to bedtime was defined as the duration (in hours) between the cessation dinner and the self-reported time of going to sleep. This variable was calculated based on the timings of the last eating event and the habitual bedtime.
Data analysis
The interval between the last meal and bedtime was calculated for each participant. To evaluate the association between meal-to-bedtime intervals and sleep quality, we categorised the time intervals into equal tertiles (T1, T2 and T3). This approach was preferred over using arbitrary clinical cutpoints to ensure equal distribution of participants across categories, thereby maximising statistical power. The first tertile (T1) represented the shortest interval (≤1·83 h), the second (T2) an intermediate interval (1·89–2·64 h) and the third (T3) the longest interval (≥2·67 h). The first tertile (T1) represented the shortest interval, while the third tertile (T3) represented the longest interval before sleep. Similarly, the nutritional quality of the dinner was assessed and participants were classified into tertiles ranging from the lowest quality (T1 ≤ 53) to the highest quality (T3 ≥ 112). This categorisation allows for a more robust comparison between different levels of meal timing and dietary quality in relation to sleep outcomes.
The distribution of continuous variables was evaluated for normality with the Kolmogorov–Smirnov test. Non-normally distributed data underwent logarithmic transformation; persistent non-normality led to the use of non-parametric methods. Data are summarised as mean (standard deviation) for normal variables or as counts and percentages for categorical variables. Inter-group comparisons for continuous and categorical variables were made using the Independent Samples t test and the χ 2 test, respectively. To examine the associations between exposures and outcomes, multivariable logistic and ordinal regression models were employed. Logistic regression models were used to examine the association between meal timing and overall sleep quality because the PSQI global score was dichotomized according to the conventional cut-off (>5 v. ≤5), indicating poor v. good sleep quality. For analyses of individual PSQI components, ordinal regression models were applied because each component is measured on a 0–3 ordered scale representing increasing severity of sleep disturbances. These models were adjusted for the following confounders: age, educational level, occupation, marital status, living situation, smoking, physical activity, BMI, chronotype, sleep quality, history of chronic disease, supplement use, total energy intake and menopausal status. Covariates were selected a priori based on evidence from previous studies examining determinants of sleep quality and potential confounders in the relationship between dietary behaviours and sleep outcomes. Additionally, sensitivity analyses were performed by stratifying the study population based on sex (men v. women) and BMI categories (normal weight, overweight and obese). This approach was employed to evaluate whether the association of the dinner-to-bedtime interval and dinner quality with sleep quality remained consistent across different physiological and demographic subgroups. All statistical analyses were performed with SPSS Statistics, Version 26.
Results
The characteristics of the study participants are presented in Table 1. The study included 450 participants (208 men and 242 women) with a mean age of 38·12 (sd 9·71) years. Overall, 251 participants (55·8 %) were categorised as having good sleep quality. The proportion of men with good sleep quality was significantly higher than that of women (P <0·001). A total of ninety-six participants (21·3 %) reported a time interval of less than 3 h between their last meal and sleep. The mean interval between the last meal and sleep was 2·38 (sd 1·08) h (range: 0·50–7·33). The mean dinner time was 20.47 (sd 0·57) h, and the average sleep duration was 07.11 (sd 01.14) h. In terms of chronotype, based on the Morningness-Eveningness Questionnaire classification, 190 participants (35 %) were categorised as morning type, 225 (50 %) as intermediate type and 35 (7·8 %) as evening type. The distribution of chronotype did not differ significantly between men and women (P = 0·70). Regarding physical activity, 208 participants (46·2 %) had low physical activity levels, 151 (33·6 %) had moderate activity and 91 (20·2 %) had high activity levels. A significant sex difference was observed in physical activity distribution, with women more frequently classified in the low physical activity category compared with men (P = 0·027).
General characteristics of the participants stratified by sex
Table 1. Long description
The table presents the characteristics of 450 study participants, including 208 men and 242 women, with a mean age of 38.12 years. It includes various characteristics such as marital status, smoking status, education level, chronic diseases, occupation status, supplement intake, living status, MEQ, physical activity level, time interval between last meal and sleep, and sleep quality. The table is divided into columns for overall characteristics, women, and men, with corresponding percentages and p-values indicating statistical significance. Notable trends include a higher proportion of men with good sleep quality compared to women, and a significant sex difference in physical activity levels, with women more frequently classified in the low physical activity category.
Values are presented as number (percentage) or mean sd (standard deviation) in total population and gender categories, calculated using χ 2 test and independent t test. The Pittsburgh Sleep Quality Index (PSQI): Good/Poor, Nutrient Rich Foods Index (NRF9_3 score).
The mean NRF9·3 score for the dinner meal was 130·26 (sd 127·34). According to this index, a score above 100 indicates high nutritional density, reflecting good-to-excellent nutritional quality.
Relationships between time interval from the last meal to bed and Pittsburgh Sleep Quality Index scores
The OR and 95 % CI for poor sleep quality (PSQI global score) across tertiles of the time interval between the last meal and bedtime are presented in Table 2. After adjustment for sex, age, occupation, marital status, smoking, energy intake, physical activity and education, a longer time interval was associated with higher odds of poorer sleep quality (OR for highest v. lowest tertile = 2·03, 95 % CI: 1·24, 3·32; P-trend = 0·005.
Associations between tertiles of time interval from the last meal to bed and sleep quality
Table 2. Long description
The table presents data on the associations between tertiles of the time interval from the last meal to bedtime and sleep quality, measured by the Pittsburgh Sleep Quality Index (PSQI). The table is divided into rows for the total population, normal weight, overweight, obese, men, and women, with columns for different time intervals (T1, T2, T3) and their respective odds ratios (OR) and 95% confidence intervals (CI). The data is presented for both unadjusted and adjusted models. Notable trends include higher odds of poorer sleep quality with longer time intervals, particularly in the total population and among overweight and obese individuals. The P-values and P-trends indicate the statistical significance of these associations.
Logistic regression was used to analyse data, adjusted for: Age, sex, education, occupation, marriage status, life status, smoking, menopause, Physical Activity Questionnaire, BMI, Morningness-Eveningness Questionnaire, chronic disease, supplement usage. PSQI: The Pittsburgh Sleep Quality Index.
* Logistic regression was used to analyze data, adjusted for: Age, sex, education, occupation, marital status, life status, smoking, menopause, physical activity questionnaire, and body mass index.
** Adjusted for all confounders except for BMI.
*** Adjusted for all confounders except for sex.
In the sensitivity analysis stratified by BMI and sex, distinct patterns emerged (Table 2). Among participants with overweight (adjusted OR = 2·50, 95 % CI: 1·21, 5·23; P-trend = 0·029) and obesity (unadjusted OR = 3·01, 95 % CI: 1·45, 6·25; P-trend = 0·007), longer meal-to-sleep intervals (T3 ≥ 2·67 h) were significantly associated with higher odds of poor sleep quality compared with the shortest interval (T1). Furthermore, when stratified by sex, a longer interval was significantly linked to poor sleep quality, specifically in men (adjusted OR = 2·38, 95 % CI: 1·09, 5·20; P-trend <0·001). In contrast, no significant associations were observed between the meal timing interval and sleep quality among individuals with normal weight or within the female subgroup (P-trend >0·05).
In contrast, analyses of individual PSQI components showed that longer meal-to-sleep intervals (T3 ≥ 2·67 h) were associated with significantly better subjective sleep quality (OR = 0·49, 95 % CI: 0·31, 0·77; P-trend = 0·002), shorter sleep latency (OR = 0·64, 95 % CI: 0·41, 0·98; P-trend = 0·041) and longer sleep duration (OR = 0·59, 95 % CI: 0·38, 0·91; P-trend = 0·017), while a borderline improvement was observed for habitual sleep efficiency (OR = 0·59, 95 % CI: 0·35, 0·99; P-trend = 0·056) (Table 3).
Associations between tertiles of time interval from the last meal to bed and aspects of sleep quality
Table 3. Long description
The table presents data on the associations between different time intervals from the last meal to bedtime and various aspects of sleep quality. It includes three tertiles of time intervals: T1 (less than or equal to 1.83 hours), T2 (1.84 to 2.64 hours), and T3 (2.67 hours or more). The table has 8 rows and 7 columns, with columns for unadjusted and adjusted models, odds ratios (OR), and 95% confidence intervals (CI) for each tertile. The rows represent different sleep quality aspects: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medication, and daytime dysfunction. Notable trends include significant improvements in subjective sleep quality, sleep latency, and sleep duration for longer meal-to-sleep intervals (T3), with odds ratios and confidence intervals provided for each aspect.
Ordinal regression analyses were conducted using an adjusted model: Age, sex, education, occupation, marriage status, life status, smoking, menopause, Physical Activity Questionnaire, BMI, Morningness-Eveningness Questionnaire, chronic disease, and supplement usage.
Relationships between dinner quality and Pittsburgh Sleep Quality Index scores
The association between dinner quality and sleep quality did not follow a uniform pattern across tertiles. Compared with the lowest tertile of dinner quality, participants in the second tertile had higher odds of poor sleep quality (OR = 1·66, 95 % CI: 1·01, 2·73), whereas those in the highest tertile showed a weaker and non-significant association (OR = 1·24, 95 % CI: 0·75, 2·04).
Sensitivity analysis revealed significant subgroup variations in the association between dinner quality and sleep (Table 3). Among women, those in the lowest tertile of dinner quality (T1) had significantly higher odds of poor sleep quality compared with those in the highest tertile (T3) in the adjusted model (OR = 2·45, 95 % CI: 1·10, 5·43, P-trend = 0·024). A similar significant trend was observed in individuals with normal weight (OR = 2·43, 95 % CI: 1·05, 5·64, P-trend = 0·040). However, no significant associations were found between dinner quality and sleep outcomes among men or individuals with overweight and obesity (P >0·05).
Analyses of individual PSQI components did not reveal statistically significant associations with dinner quality; however, sleep duration showed a modest but non-significant tendency toward improvement at higher dinner quality levels (T3 v. T1: OR = 0·88, 95 % CI: 0·57, 1·36) (Tables 4 and 5).
Associations between tertiles of the quality of the dinner and sleep quality
Table 4. Long description
The table presents data on the associations between tertiles of dinner quality and sleep quality, specifically focusing on the Pittsburgh Sleep Quality Index (PSQI). It includes unadjusted and adjusted models for the total population, as well as subgroups categorized by normal weight, overweight, obese, men, and women. The table is divided into columns for different tertiles of dinner nutrient richness factor (NRF) scores (T1, T2, T3) and includes odds ratios (OR) with 95 percentage confidence intervals (CI). Notable trends include a modest but non-significant tendency toward improved sleep duration at higher dinner quality levels. The table also provides P-values and P-trends for statistical significance. Row-wise data includes unadjusted and adjusted models for each subgroup, with specific odds ratios and confidence intervals listed for each tertile.
Logistic regression was used to adjust for age, sex, education, occupation, marriage status, life status, smoking, menopause, Physical Activity Questionnaire, BMI, Morningness-Eveningness Questionnaire, chronic disease, and supplement usage.
** Adjusted for all confounders except for BMI.
*** Adjusted for all confounders except for sex.
Associations between tertiles of the quality of the dinner and aspects of sleep quality
Table 5. Long description
The table presents data on the associations between tertiles of dinner quality and various aspects of sleep quality. It includes three tertiles (T1, T2, T3) based on dinner nutritional repeat frequency (NRF) scores, with T1 being the lowest (less than or equal to 53) and T3 being the highest (greater than or equal to 112). The table has eight rows representing different sleep quality aspects: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medication, and daytime dysfunction. Each row contains unadjusted and adjusted models with odds ratios (OR) and 95 percentage confidence intervals (CI) for T2 and T3 compared to T1. Notable trends include a modest but non-significant tendency toward improved sleep duration at higher dinner quality levels (T3 vs. T1: OR 0.88, 95 percentage CI: 0.57, 1.36). The table also shows that sleep disturbances have a significant association with dinner quality in the unadjusted model (T3 vs. T1: OR 1.77, 95 percentage CI: 1.09, 2.89).
Ordinary regression analyses were used, adjusted for age, sex, education, occupation, marriage status, life status, smoking, menopause, Physical Activity Questionnaire, BMI, Morningness-Eveningness Questionnaire, chronic disease and supplement usage.
Discussion
This cross-sectional study examined the association between the timing of the last meal before sleep and the nutritional quality of dinner with sleep quality, assessed using the PSQI, in a sample of Iranian adults. After adjusting for demographic and lifestyle variables, we observed that a longer interval between the last meal and bedtime was associated with poorer overall sleep quality. Interestingly, analyses of specific sleep components revealed mixed effects, with some improvements in subjective sleep quality, sleep latency and efficiency when the interval was longer.
The interpretation of our results requires careful consideration of the distinct cultural meal timing norms in Iran. In the present study, the mean dinner time was 20.47 (sd 0.57) h, with a mean interval between the last meal and sleep of 2·38 (sd 1·08) h, ranging as close as 30 min before bedtime. These descriptive data align with observational findings in Iranian populations where eating dinner after 21.00 is a normative behaviour rather than an outlier pattern(Reference Zeraattalab-Motlagh, Lesani and Janbozorgi23). Our distribution showed that even in the longest interval tertile (T3), the starting point was only ≥ 2·67 h, which in many Western guidelines would still be categorised as late eating(Reference Irish, Kline and Gunn37,Reference Qaseem, Kansagara and Forciea38) .
This distribution suggests that shorter intervals between dinner and sleep are culturally embedded in this population. Consequently, the physiological disruption typically associated with late-night eating in Western cohorts – where dinner is usually consumed much earlier – might not be as pronounced here due to habitual adaptation(Reference Saidi, Rochette and Dambel39). The fact that our participants, on average, slept for 07.11 (sd 01.14) h despite these late meal timings underscores the necessity of culturally sensitive frameworks. It questions the universal applicability of strict pre-sleep fasting recommendations and suggests that in the Iranian context, a 2-h window might be the practical ‘norm’ rather than a metabolic risk factor.
Association between last meal timing and sleep quality
Our results indicate that shorter intervals between the last meal and sleep were associated with better overall sleep quality. In our adjusted models, participants with shorter meal-to-sleep intervals demonstrated significantly lower odds of poor global sleep quality, suggesting that eating closer to bedtime may not necessarily impair perceived sleep in this population. This finding contrasts with conventional sleep hygiene recommendations but aligns with emerging evidence suggesting that the relationship between evening food intake and sleep may depend on contextual and behavioural factors(Reference Dashti, Scheer and Jacques40,Reference St-Onge, Pizinger and Kovtun41) .
This observation diverges from conventional sleep hygiene recommendations, which generally advise allowing at least 2–3 h between eating and bedtime to prevent sleep disturbances related to digestion, reflux or metabolic activity(Reference Keser and Yüksel42–Reference BaHammam and Pirzada44). Several studies have documented poorer sleep outcomes with later eating patterns, including reduced sleep duration and increased risk of insomnia(Reference Chung, Bin and Cistulli45–Reference Yan, Li and Fan47). For example, Yasuda et al. (Reference Yasuda, Kishi and Fujita46) reported in young Japanese adults that a longer gap between dinner and sleep was associated with better subjective sleep indices, while Chung et al. (Reference Chung, Bin and Cistulli45) found no significant relationship in university students. Conversely, Lehmann et al. (Reference Lehmann, Saidi and Giacomoni48) reported that eating closer to bedtime (1·5 h) was linked with less fragmented sleep in young athletes compared with a longer gap (3·5 h). Compared with these studies, our findings suggest that the effect of meal timing on sleep may vary substantially across populations. In our sample, where the mean dinner time was approximately 20.47, eating closer to bedtime was associated with better perceived sleep quality, highlighting the potential role of habitual meal timing patterns(Reference Katsi, Papakonstantinou and Soulaidopoulos49).
These discrepancies also suggest that the relationship between meal timing and sleep is not linear and may be moderated by factors such as age, sex, activity level, cultural eating habits and meal composition. Meal timing may influence sleep through several interconnected biological and behavioural mechanisms. From a chronobiological perspective, food intake acts as a secondary zeitgeber for peripheral circadian clocks involved in metabolic regulation(Reference Katsi, Papakonstantinou and Soulaidopoulos49). Consuming meals late in the evening may therefore shift metabolic rhythms and interact with the circadian secretion of hormones such as insulin, cortisol and melatonin, potentially influencing sleep initiation and maintenance(Reference St-Onge, Pizinger and Kovtun41). In addition, behavioural factors such as satiety, evening routines and psychological relaxation associated with customary dinner practices may also shape subjective sleep experiences(Reference Dashti, Scheer and Jacques40). Together, these pathways suggest that the impact of dinner timing on sleep may vary across different dimensions of sleep quality, including sleep latency, efficiency and subjective sleep perception(Reference Pot50). In our study population, late dinners are culturally common, and eating close to bedtime may have been perceived as routine, potentially explaining why participants reported better sleep quality under these conditions. Cultural normalisation of late meals may attenuate the metabolic or psychological stress that such behaviour induces in populations with earlier eating traditions.
Given that shorter meal-to-sleep intervals in our study were associated with better global sleep ratings, several physiological mechanisms may help explain this observation(Reference Peuhkuri, Sihvola and Korpela15). From a physiological perspective, several mechanisms could explain the seemingly counterintuitive benefits of shorter meal-to-sleep intervals. Consuming a balanced meal before bedtime may increase availability of tryptophan and carbohydrates, which promote serotonin and melatonin synthesis, facilitating sleep initiation(Reference Lieberman, Agarwal and Fulgoni51). Late eating may also help stabilise blood glucose levels overnight, preventing awakenings due to nocturnal hypoglycaemia(Reference Pigeon, Carr and Gorman52). Moreover, eating may serve as a calming evening routine that reduces psychological stress and facilitates sleep onset, particularly in cultures where family dinners represent a regular social ritual occurring later in the evening(Reference Grandner2,Reference Irish, Kline and Gunn37) .
Our sensitivity analyses further revealed that the association between a longer dinner-to-bedtime interval and poor sleep quality is particularly pronounced in men and individuals with excess body weight (overweight/obesity). These findings suggest that the metabolic response to meal timing may be moderated by adiposity and biological sex. Individuals with higher BMI often exhibit altered circadian regulation of appetite-related hormones and increased insulin resistance, which may exacerbate the impact of meal timing on sleep-wake cycles(Reference St-Onge, Pizinger and Kovtun41,Reference Stenvers, Scheer and Schrauwen53) .
The sex-specific findings in men align with some prior studies suggesting that men may be more susceptible to the metabolic disruptions caused by altered eating windows, possibly due to differences in hormonal profiles or lifestyle factors. The lack of a significant association in women and normal-weight individuals highlights the importance of personalised nutritional recommendations that consider body composition and sex-specific physiological needs(Reference Saidi, Rochette and Dambel39,Reference Katsi, Papakonstantinou and Soulaidopoulos49) .
At the same time, our findings of mixed effects across PSQI subcomponents suggest complex underlying interactions. In our analyses, shorter meal-to-sleep intervals were associated with better global PSQI scores, whereas longer intervals were related to improvements in certain individual components such as sleep efficiency. This apparent discrepancy highlights the multidimensional nature of sleep quality and suggests that meal timing may influence different aspects of sleep in distinct ways(Reference Duan, Gu and Polotsky10,Reference Luz, da Fonseca and Santos18,Reference Yan, Li and Fan47,Reference Mollayeva, Thurairajah and Burton54) . The global score is calculated as the sum of seven distinct domains, including subjective sleep quality, sleep latency, duration, efficiency, disturbances, medication use and daytime dysfunction. These components do not necessarily represent a single underlying construct but rather capture different aspects of sleep perception and behaviour. Consequently, improvements in certain domains (e.g. shorter sleep latency or higher efficiency) may coexist with neutral or adverse changes in other components, resulting in a global score that does not fully reflect domain-specific improvements. Methodological studies have highlighted that the PSQI global score may obscure heterogeneous patterns across individual sleep dimensions, particularly in epidemiological studies using self-reported sleep measures(Reference Buysse, Reynolds and Monk1,Reference Mollayeva, Thurairajah and Burton54) .
Furthermore, subjective sleep measures such as the PSQI may capture perceived sleep satisfaction or habitual sleep experiences rather than objective sleep physiology. Behavioural and cultural factors – such as customary meal timing or evening routines – may influence how individuals evaluate their overall sleep quality, even when specific sleep parameters change in different directions(Reference St-Onge, Grandner and Brown24). This may partly explain why shorter meal-to-sleep intervals were associated with better global sleep ratings in our population, while longer intervals appeared beneficial for certain physiological components of sleep. Future studies using objective sleep measures such as actigraphy or polysomnography may help clarify whether these divergent patterns reflect true physiological differences or limitations inherent to subjective sleep assessment tools.
Dinner quality and sleep outcomes
In the present study, dinner quality assessed using the NRF9·3 index was not linearly associated with global sleep quality. However, exploratory analyses suggested a nonlinear pattern across tertiles of dinner quality(Reference Drewnowski55).
Previous studies have primarily examined overall diet quality or long-term dietary patterns rather than focusing on dinner specifically(Reference Crispim, Zimberg and dos Reis13,Reference Godos, Grosso and Castellano16,Reference Kim, DeRoo and Sandler56) . Higher-quality diets, rich in fruits, vegetables, fibre and micronutrients, have consistently been associated with better sleep outcomes, whereas diets high in saturated fats, sugar and processed foods are linked with sleep disturbances(Reference Wilson, St-Onge and Tasali11,Reference Godos, Grosso and Castellano16,Reference Chaput57) . For example, St-Onge et al. (Reference St-Onge, Roberts and Shechter9,Reference St-Onge26) found that diets high in fibre and low in saturated fat improved sleep efficiency, while Chaput(Reference Chaput57) noted that poor dietary quality contributes to systemic inflammation, which disrupts sleep regulation.
Our stratified analysis showed that the negative impact of poor dinner quality on sleep was more evident in women and normal-weight individuals. This suggests a potential sex-specific sensitivity to nutrient density before sleep. Previous research indicates that women may report higher subjective sleep disturbances in response to dietary deficiencies(Reference St-Onge, Mikic and Pietrolungo58). Furthermore, the significant association in the normal-weight group – but not in those with higher BMI – might suggest that in individuals with overweight or obesity, other factors such as metabolic inflammation or obstructive sleep apnoea may play a more dominant role in sleep disruption, potentially masking the specific effect of dinner’s nutritional quality(Reference Grandner, Schopfer and Sands-Lincoln59).
These findings partially contrast with previous studies that reported associations between overall diet quality and sleep outcomes, suggesting that meal-specific dietary quality may exert weaker or more context-dependent effects than habitual dietary patterns(Reference St-Onge, Pizinger and Kovtun41). One reason for the lack of association in our study may be the compensatory effect of other meals. Participants with poor-quality dinners may have consumed more balanced breakfasts or lunches, thereby attenuating the impact of dinner on nightly sleep. Nutrients involved in sleep regulation, including tryptophan and carbohydrates, influence serotonin and melatonin pathways across the entire day; therefore, adequate nutrient intake during earlier meals may partially compensate for lower nutrient density at dinner, potentially attenuating associations between dinner quality and sleep outcomes(Reference Peuhkuri, Sihvola and Korpela15). Additionally, dinner in Iranian culture is often later, lighter and highly variable in composition, potentially diluting associations detectable in more homogeneous populations. Measurement limitations – particularly reliance on self-reported food records and subjective sleep assessments – may have also introduced error, reducing our ability to detect modest but meaningful associations(Reference St-Onge, Mikic and Pietrolungo58).
Another possibility is that single-meal quality may be less influential than overall dietary patterns. While dinner is important due to its proximity to sleep, long-term dietary habits likely play a stronger role in shaping circadian regulation, hormonal rhythms and inflammatory responses(Reference Godos, Grosso and Castellano16,Reference Bae, Fang and Rustgi20,Reference Froy22) . Future research should integrate assessments of daily dietary patterns with meal-specific quality to clarify how both interact with sleep physiology.
Biological mechanisms linking food timing, diet and sleep
Meal timing is a key modulator of circadian rhythms, influencing sleep–wake cycles through hormonal and metabolic pathways(Reference BaHammam and Pirzada44,Reference Challet60) . Eating late may alter secretion patterns of insulin, leptin and ghrelin, which are tightly coupled with circadian rhythms and influence appetite regulation, metabolism and sleep drive(Reference Stenvers, Scheer and Schrauwen53,Reference Kluge, Schüssler and Bleninger61) . Ghrelin, for instance, promotes non-REM sleep, while disruptions in leptin signalling have been associated with insomnia(Reference Kluge, Schüssler and Bleninger61). Thus, the timing of the last meal may serve as a ‘zeitgeber’ (time cue) that interacts with the central circadian clock in the suprachiasmatic nucleus, reshaping sleep architecture.
Regarding diet composition, micronutrients such as Mg, vitamin B6 and tryptophan are critical for the synthesis of neurotransmitters that regulate sleep(Reference Peuhkuri, Sihvola and Korpela15,Reference Lieberman, Agarwal and Fulgoni51) . Diets high in processed foods and saturated fat may impair sleep quality by increasing systemic inflammation through cytokines such as IL-6 and TNF-α (Reference Irwin, Olmstead and Carroll62). In contrast, foods rich in complex carbohydrates and protein may support sleep by promoting serotonin and melatonin pathways(Reference Peuhkuri, Sihvola and Korpela15). These mechanisms may explain why high-quality diets tend to be associated with better sleep, though the effect of a single meal (like dinner) may be smaller compared with long-term dietary patterns.
Strengths and limitations
This study has several strengths. It is among the first to investigate dinner timing and quality in relation to sleep quality in an Iranian adult population, providing insights into a cultural context where late-night eating is common. The relatively large sample size and adjustment for multiple confounders increase confidence in the robustness of the associations observed.
However, several limitations must be considered. First, the cross-sectional design precludes causal inference, and reverse causality cannot be ruled out (e.g. poor sleep leading to late-night eating).
It is also possible that the observed associations reflect reverse causality. Individuals experiencing poor sleep may modify their eating behaviours, such as delaying meals due to insomnia or consuming food closer to bedtime during nocturnal wakefulness. Conversely, some individuals may intentionally eat earlier in the evening in an effort to improve sleep quality. Another possible explanation is a form of behavioural or healthy user bias. Individuals who generally experience better sleep may feel more comfortable eating closer to bedtime and may not perceive late eating as disruptive. Conversely, individuals with sleep difficulties may intentionally avoid late meals in an effort to improve their sleep. Because both meal timing and sleep quality were assessed simultaneously in this cross-sectional study, it is not possible to determine the temporal direction of these relationships(Reference St-Onge, Grandner and Brown24,Reference Grandner, Jackson and Gerstner25) . Prospective cohort studies and randomised interventions are therefore needed to clarify whether meal timing influences sleep outcomes or whether sleep patterns drive changes in eating behaviour. Second, both dietary intake and sleep quality were self-reported, introducing recall bias and subjective error. Additionally, the use of three 24-h dietary recalls may not fully capture habitual dietary intake or usual meal timing patterns, as both food consumption and eating schedules can vary from day to day. Furthermore, the timing of the last meal and bedtime was self-reported, which may introduce measurement error in estimating the interval between eating and sleep. Participants may have difficulty accurately recalling small snacks or beverages consumed close to bedtime, potentially leading to misclassification of the exposure variable(Reference Freedman, Schatzkin and Midthune63). Such reporting inaccuracies are common in dietary assessment studies and may attenuate observed associations. Third, although multiple covariates were included in the models, some potentially important factors such as caffeine intake, alcohol consumption, light exposure, bedroom environment and work schedules were not available in the dataset. Residual confounding due to these unmeasured factors cannot be excluded. Fourth, analyses of individual PSQI components involved multiple statistical tests, which may increase the risk of Type I error. Therefore, these findings should be interpreted with caution. In addition, some PSQI components – particularly sleep medication use – had a relatively low prevalence in the study population. As a result, the corresponding regression estimates showed wider confidence intervals, reflecting limited statistical precision for these specific subcomponent analyses. Therefore, findings related to these individual domains should be interpreted with caution. Fifth, the NRF9·3 index may not fully capture the cultural nuances of Iranian meals, and future work should validate culturally adapted tools for meal quality assessment. Finally, cultural eating practices should be considered when interpreting our findings. In Iran, dinner is often consumed relatively late in the evening, and eating close to bedtime may represent a habitual and socially normative behaviour. Therefore, the associations observed in this study may not be directly generalisable to populations where dinner is typically consumed much earlier. Future studies in diverse cultural contexts are needed to determine whether these relationships hold across different meal timing traditions.
Implications and future research
Our findings suggest that in Iranian adults, eating closer to bedtime may not be as detrimental to sleep as previously assumed and, in some cases, may even be beneficial. These challenges generalised sleep hygiene advice that discourages late-night eating and highlights the importance of considering cultural eating practices when making recommendations.
Nevertheless, the mixed results across sleep components underscore the need for more nuanced guidelines that account for not only the timing but also the composition of meals. Randomised controlled trials using objective sleep measures (e.g. actigraphy or polysomnography) are needed to disentangle the complex interplay between meal timing, meal composition and sleep.
Future studies should also explore whether individual factors such as chronotype, metabolic health status or genetic predispositions moderate the meal–sleep relationship. Such personalised approaches may ultimately inform tailored interventions for improving sleep through dietary strategies.
Conclusions
In conclusion, the association between the timing of the last meal and sleep quality appears to be complex and potentially bidirectional. Although shorter meal-to-sleep intervals were associated with better overall sleep quality in this population, the findings across individual sleep domains were heterogeneous. These results suggest that the relationship between meal timing and sleep may depend on cultural eating patterns, behavioural factors and individual characteristics. Longitudinal and experimental studies are needed to clarify the direction and underlying mechanisms of these associations.
Acknowledgements
The authors would like to thank the participants for their sincere cooperation in this study. This study was approved and funded by a grant from Tehran University of Medical Sciences (TUMS), Tehran, Iran.
The authors declare that no funds, grants or other support were received during the preparation of this manuscript.
F. R. and S. S-B. contributed to the conception/design of the research; B. J. G., R. N., S. G. and N. B. contributed carrying out the study; F. R. contributed to the analysis; F. R. and B. J. G. contributed to the interpretation of the data and drafting the first version of the manuscript; S. S-B. and K. Dj. critically revised the manuscript and S. S-B. agrees to be fully accountable for ensuring the integrity and accuracy of the work.
The authors report there are no competing interests to declare.
