Association between ultra-processed food intake, night eating behavior, and sleep quality: a cross-sectional study from Türkiye

Abstract

Background

Ultra-processed food (UPF) consumption and night eating behavior are increasingly common in modern dietary patterns and may negatively affect sleep quality. While each factor has been independently associated with adverse health outcomes, their combined influence on sleep-related well-being is less understood.

Methods

This cross-sectional study included 1111 adults aged 18–65 years from Türkiye. Data were collected using a structured online questionnaire, including the Screening Questionnaire for Highly Processed Food Consumption (sQ-HPF), Night Eating Questionnaire (NEQ), and Pittsburgh Sleep Quality Index (PSQI). Pearson correlation, multiple linear regression, and binary logistic regression analyses were conducted to explore the associations between UPF intake, night eating behavior, and sleep quality.

Results

UPF intake was positively correlated with both night eating (r = 0.288, P < 0.001) and PSQI scores (r = 0.216, P < 0.001). Night eating was also positively associated with poorer sleep quality (r = 0.285, p < 0.001). Regression analyses showed that age, marital status, number of meals and snacks, NEQ scores, and PSQI scores were independently associated with UPF intake. NEQ and UPF scores were also independently associated with poor sleep quality.

Conclusion

UPF consumption and night eating behavior appear to be important behavioral factors associated with reduced sleep quality. Interventions targeting these dietary behaviors may contribute not only to improved sleep health but also to enhanced overall quality of life in adults.

Introduction

Ultra-processed food (UPF) consumption has skyrocketed globally in recent years, shaping modern eating patterns and raising significant public health concerns [1, 2]. Chen et al. [3] define UPFs as manufactured products usually high in calories, low in nutrients, and heavy in salt, added sugars, saturated fats, and other food additives [3]. High UPF consumption has been linked to many negative health effects including depression, obesity, cardiometabolic diseases, and higher risk of dying from any cause [4, 5].

Though much has been studied about the effects on physical health [6, 7], more recent studies have started to investigate how UPF consumption influences fundamental behaviors like eating patterns [8] and sleep [9]. Sleep controls metabolic, immunological, and psychological health. Diet composition, especially the kind of carbohydrates and the glycemic index, is more and more acknowledged to affect sleep quality [10]). Diets low in fiber and high in refined carbohydrates have been found to disturb sleep architecture and circadian cycle [11].

According to their nutritional makeup, UPFs may contribute to sleep problems through a higher glycemic load, micronutrient deficiencies, and systemic inflammation [3, 12]). Numerous studies have connected UPF use to shorter sleep duration, decreased sleep efficiency, and an increased incidence of symptoms of insomnia [13, 14].

Meal timing has emerged as a critical factor in sleep health, in addition to food composition. Eating a significant amount of daily calories in the evening or at night is known as “night eating,” and it is associated with increased psychological discomfort, decreased melatonin production, and circadian misalignment [15–17].

Although research has examined UPF intake and sleep or night eating patterns independently, very few have considered these elements jointly. Thus, the relationship between sleep control, eating patterns, and dietary quality is still not well known [18, 19]. Night eating itself might be linked to bad metabolic and psychological consequences, perhaps by means of disrupted circadian rhythms and compromised appetite control [15, 20]. Likewise, UPF consumption could exacerbate these effects by means of higher calorie consumption and metabolic stress [13].

While various studies on UPF consumption, nighttime eating, and sleep quality are available in the literature [3, 8, 14, 18, 21]; Eroğlu, Ekici, & Açıkalın Göktürk [22]), , there is a need for studies with a larger number of participants that address the topic from a broader perspective. Sleep and dietary behaviors likely share a complex bidirectional relationship. Evidence suggests that poor sleep quality can increase the consumption of ultra-processed foods due to hormonal dysregulation and impaired cognitive control, while diets rich in ultra-processed foods may, in turn, disrupt sleep through inflammatory and metabolic pathways [23–25].

Emerging evidence highlights several potential pathophysiological pathways linking ultra-processed food (UPF) consumption, night eating behavior, and sleep disruption. Diets high in UPFs—particularly those rich in sugar, fat, and additives—can disrupt metabolic and hormonal regulation. These disruptions may influence circadian rhythms and impair the balance of appetite-related hormones such as leptin and ghrelin [23]. Moreover, St-Onge et al. [25] emphasized a bidirectional cycle, where diet affects sleep via the tryptophan–melatonin pathway, and in return, insufficient sleep enhances UPF consumption by altering hormonal responses. Night eating syndrome has also been linked to irregularities in cortisol and melatonin secretion, further contributing to metabolic imbalance and poor sleep quality [26].

This study was meant to look at the links between UPF consumption, night eating patterns, and sleep quality in a broad population of people to fill in these gaps. Based on previous literature, we hypothesized that higher ultra-processed food intake and night eating behavior would be associated with lower sleep quality among Turkish adults. However, given the cross-sectional nature of the study, these relationships should be interpreted as associations rather than causal inferences. Knowing these linked lifestyle factors could help to direct thorough strategies to improve both nutritional and sleep health.

Materials and methods

This descriptive cross-sectional study was carried out between September 2024 and January 2025 and included 1,111 adults (797 females and 314 males) aged 18–65 years. Data collection was conducted online using a structured Google Forms survey created by the researchers. Participants were recruited through a snowball sampling strategy via social media platforms, including Twitter, Facebook, WhatsApp, and Instagram, within the Ankara province of Türkiye.

Individuals were eligible to participate if they were aged 18–65, had internet access, voluntarily provided consent by selecting the agreement checkbox at the beginning of the survey, and completed the questionnaire in full. Participants under the age of 18, those with incomplete survey responses, individuals who self-reported a diagnosis of severe psychiatric (e.g., psychotic spectrum disorders) or neurological conditions (e.g., epilepsy, multiple sclerosis) that could interfere with dietary behavior or sleep quality, and individuals working night shifts were excluded from the study. These conditions were identified based on participants’ self-declaration. Pregnant and lactating individuals were also excluded due to potential alterations in eating and sleep patterns caused by physiological and hormonal changes during this period. On average, participants completed the survey in approximately 15 min.

Ethical clearance for this research was granted by the University of Health Sciences Gülhane Scientific Research Ethics Committee (Decision No: 2024/570). All procedures involving human participants adhered to the ethical principles outlined in the Declaration of Helsinki.

The online survey collected data on participants’ sociodemographic background (including age, gender, marital status, educational attainment, income level, and employment status), eating behaviors (such as frequency of main meals and snacks), self-reported anthropometric details (height and weight), as well as ultra-processed food intake, night eating patterns, and sleep quality. All scales, including the sQ-HPF, NEQ, and PSQI, were administered simultaneously within a single online questionnaire, reflecting participants’ habitual behaviors and perceptions over a similar retrospective period. A visual summary of the participant recruitment process, including inclusion and exclusion criteria, is presented in Fig. 1.

Fig. 1.

Participant recruitment and assessment flowchart

Flowchart illustrating the participant recruitment and selection process. The figure summarizes the recruitment strategy, eligibility criteria, reasons for exclusion, final sample size (n = 1111), evaluated variables, and measurement instruments (including sQ-HPF, NEQ, PSQI, and self-reported anthropometrics) used in the study.

Eating behaviors

At the beginning of the questionnaire, participants were presented with brief definitions and examples distinguishing main meals and snacks. Main meals were described as structured eating occasions typically consumed at breakfast, lunch, or dinner times and generally consisting of a variety of food groups. In contrast, snacks were defined as smaller, unstructured eating episodes occurring between meals, often including items such as biscuits, chips, fruits, or beverages. Participants were instructed to classify their eating occasions accordingly based on these definitions. Specific examples of time intervals (e.g., morning, mid-afternoon, late evening) and food types were also provided to facilitate accurate self-reporting. Importantly, no minimum caloric threshold (e.g., 50 kcal) was applied to differentiate between meals and snacks. This strategy allowed for capturing participants’ habitual eating patterns within a culturally contextualized and flexible reporting framework.

Anthropometric measurements

Anthropometric indicators, including height and weight, were based on self-reported data provided by participants. The online form included guidance to help individuals accurately report their body measurements. To improve accuracy in self-reported anthropometric data, the Google Form included specific instructions such as using a wall and flat surface to measure height, and a calibrated scale for weight. Participants were encouraged to report recent values, and reminders were provided to avoid estimation. Nonetheless, self-reported anthropometric data may still be prone to reporting bias due to potential under- or overestimation. Therefore, these values should be interpreted with caution in the absence of direct clinical assessment. However, several studies support the validity of self-reported anthropometric measurements. For instance, Lassale et al. [27], using data from the Nutrinet-Santé study, found high validity of web-based self-reported height and weight, with intraclass correlation coefficients (ICCs) exceeding 0.9. Similarly, Fayyaz et al. [28] conducted a systematic review demonstrating good agreement between measured and self-reported anthropometric data, reporting ICC values greater than 0.9 and acceptable limits of agreement.

Body mass index (BMI) was computed using the standard formula: weight in kilograms divided by height in meters squared (kg/m²). Following the classification criteria by Madden and Smith [29], participants were categorized as underweight (< 18.50 kg/m²), normal weight (18.50–24.99 kg/m²), overweight (25.0–29.99 kg/m²), or obese (≥ 30.0 kg/m²).

Ultra-processed foods consumption

The level of ultra-processed food (UPF) intake was assessed using the Screening Questionnaire for Highly Processed Food Consumption (sQ-HPF), developed by Martínez-Pérez et al. [30]. The instrument was adapted and validated for the Turkish population by Erdoğan Gövez et al. [31], reporting a Cronbach’s alpha coefficient of 0.65. This 11-item scale requires participants to respond to whether they consume specific food products, assigning 1 point for each affirmative answer and 0 for negative responses. The total score ranges from 0 to 11, with a threshold of 6 or above indicating a high level of UPF intake. A higher score reflects more frequent consumption of such foods.

Night eating behavior

The Night Eating Questionnaire (NEQ), developed by Allison et al. [32], was used to evaluate patterns characteristic of night eating behavior. In this study, ‘night eating behavior’ refers to evening and nocturnal eating patterns and does not imply a clinical diagnosis of Night Eating Syndrome (NES). Its Turkish adaptation and psychometric evaluation were conducted by Atasoy et al. [33], demonstrating construct validity and acceptable internal consistency (Cronbach’s alpha = 0.69) in Turkish adult populations. The instrument consists of 14 items, most of which are scored on a five-point Likert scale ranging from 0 to 4. Scoring excludes item 13, and item 7 is not rated using the Likert format. The total possible score spans from 0 to 52, with elevated scores indicating a higher tendency toward night eating behaviors. In the present study, NEQ scores were analyzed as a continuous variable to assess night eating tendencies rather than to diagnose Night Eating Syndrome (NES). This approach aligns with previous non-clinical applications of the NEQ in Turkish adult populations [33].

Sleep quality

Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI), a widely utilized tool introduced by Buysse et al. [34] that evaluates sleep patterns over the previous month. The scale includes 19 self-assessment items, which are organized into seven key domains: perceived sleep quality, time to fall asleep, total sleep duration, sleep efficiency, frequency of sleep disturbances, use of sleep aids, and daytime performance impairment. Each domain yields a score between 0 and 3, with the overall score ranging from 0 to 21. A global score exceeding 5 is typically interpreted as indicating poor sleep quality. This threshold is based on the original validation by Buysse et al. [34] and has been supported by the Turkish adaptation study by Ağargün et al. [35], which confirmed its applicability in Turkish adult populations. The Turkish adaptation and validation of the PSQI were performed by Ağargün et al. [35], with internal consistency values of 0.83 in the original version and 0.80 in the Turkish version.

Statistical analysis

Statistical analyses were carried out using IBM SPSS Statistics version 27.0. Descriptive statistics provided an overview of participants’ demographic, lifestyle, and behavioral profiles. Continuous variables were summarized using means and standard deviations, while categorical variables were described in terms of frequencies and percentages.

The normality of the data distributions was assessed through the Kolmogorov–Smirnov test in conjunction with visual inspections via histograms and Q-Q plots. Depending on distribution characteristics, Spearman correlation coefficients were calculated to evaluate associations among ultra-processed food intake, night eating tendencies, and sleep quality indicators. To identify determinants of night eating behavior, a multiple linear regression model was constructed with NEQ score as the outcome variable. Multivariable linear regression analyses were performed using a backward elimination approach, in which nonsignificant variables (p > 0.05) were removed stepwise to obtain the most parsimonious model. All independent variables were entered simultaneously at the beginning of the model to control for potential confounding effects. Sociodemographic factors such as BMI, income level, and marital status were initially included but excluded from the final models due to lack of statistical significance. Independent variables included gender, age, meal and snack frequency, UPF intake score, and PSQI score. Additionally, binary logistic regression models were applied to examine factors associated with high UPF intake and impaired sleep quality. A global PSQI score greater than 5 was used to define poor sleep quality, in accordance with the original validation study by Buysse et al. [34]. These models included socio-demographic and behavioral covariates. Results were reported as odds ratios (ORs) along with 95% confidence intervals (CIs), and significance was determined at a p-value threshold of < 0.05. All assumptions related to regression analyses including linearity, independence of residuals, multicollinearity, and homoscedasticity were evaluated and met. Model fit was assessed using the Hosmer–Lemeshow goodness-of-fit test for both logistic regression models. The test indicated acceptable fit for both models (P = 0.230 and p = 0.110, respectively). A post-hoc power analysis was performed using G*Power (version 3.1.9.7) to evaluate the adequacy of the achieved sample size (n = 1111). Assuming an alpha of 0.05, a power of 0.95, and a small to medium effect size (f² = 0.02–0.03), the sample was found sufficient for both linear and logistic regression analyses. All variables were collected via a mandatory-response Google Form, and no missing data were encountered during the data collection process.

Results

Table 1 presents a descriptive overview of the study participants. The average age of the sample was 28.7 years (SD = 12.46), and the mean body mass index (BMI) was calculated as 23.9 kg/m² (SD = 4.61). Respondents reported consuming approximately 2.4 main meals (SD = 0.54) and 1.6 snacks (SD = 0.78) daily.

Table 1.

Descriptive characteristics of the study sample

Variables	Total (n=1111) X̅±SD	Female(n=797) X®±SD	Male (n=314) X®±SD	P-value
Age (years)	28.7±12.46	28.9±11.80	29.8±13.79	0.126
Body Mass Index (kg/m2)	23.9±4.61	23.3±4.43	24.1±4.49	0.113
Number of daily main meals	2.4±0.54	2.4±0.53	2.4±0.55	0.130
Number of daily snacks	1.6±0.78	1.6±0.78	1.6±0.78	0.216
sQ-HPF Score	4.8±2.68	4.9±2.65	4.8±2.76	0.366
NEQ Score	13.7±5.24	12.6±5.20	13.9±5.34	0.003*
PSQI Score	6.7±2.90	6.9±2.87	6.4±2.93	0.002*
	Number (%)	Number (%)	Number (%)
Education level				0.062
Primary school	30 (2.7)	22 (2.8)	8 (2.5)
Middle school	21 (1.9)	15 (1.9)	6 (1.9)
High school	182 (16.4)	106 (13.3)	76 (24.2)
University	842 (75.8)	629 (78.9)	213 (67.8)
Master's degree/ Doctorate	36 (3.2)	25 (3.1)	11 (3.5)
Marital status				0.111
Single	754 (67.9)	520 (65.2)	234 (74.5)
Married	357 (32.1)	277 (34.8)	80 (25.5)
Income status				0.076
Income < Expenses	279 (25.1)	218 (27.4)	61 (19.4)
Income = Expenses	597 (53.7)	440 (55.2)	157 (50.0)
Income >Expenses	235 (21.2)	139 (17.4)	96 (30.6)
BMI Classification				0.301
Underweight	91 (8.2)	70 (8.8)	21 (6.7)
Normal	638 (57.4)	456 (57.2)	182 (58.0)
Overweight	268 (24.1)	86 (23.3)	82 (26.1)
Obese	114(10.3)	85 (10.7)	29 (9.2)
UPF Consumption - High	465 (41.9)	334 (41.9)	131 (41.7)	0.955
UPF Consumption - Low	646 (58.1)	463(58.1)	183 (58.3)
Sleep Quality				0.003*
Good	402 (36.2)	267 (33.5)	135 (43.0)
Poor	709 (63.8)	530 (66.5)	179 (57.0)

sQ-HPF Screening Questionnaire for Highly Processed Food Consumption, NEQ Night Eating Questionnaire, PSQI Pittsburgh Sleep Quality Index, BMI Body Mass Index, UPF Ultra-Processed Food, SD Standard Deviation, * Statistically significant at P < 0.05 level

The mean scores for the assessed instruments were as follows: 4.8 (SD = 2.68) for the Screening Questionnaire of Highly Processed Food Consumption (sQ-HPF), 13.7 (SD = 5.24) for the Night Eating Questionnaire (NEQ), and 6.7 (SD = 2.90) for the Pittsburgh Sleep Quality Index (PSQI).

In terms of gender, the majority were female (71.7%), while 28.3% were male. Educational attainment was highest among university graduates (75.8%), with others having completed high school (16.4%), primary school (2.7%), middle school (1.9%), or holding postgraduate degrees (3.2%). Regarding marital status, 67.9% were single and 32.1% were married. Income status distribution showed that 53.7% reported income equal to expenses, 25.1% reported income below expenses, and 21.2% indicated income above their expenditures.

According to BMI classifications, the majority (57.4%) were of normal weight, while 24.1% were overweight, 10.3% were obese, and 8.2% were underweight. In terms of ultra-processed food consumption, 41.9% of individuals were classified as high consumers. Regarding sleep quality, 63.8% of participants were identified as poor sleepers, whereas 36.2% had good sleep quality. Although the mean PSQI score was 6.7, this does not imply that all participants experienced poor sleep quality. The prevalence figure (63.8%) represents those who exceeded the established cutoff of 5, while others with lower scores contributed to the overall mean.

Table 2 outlines the correlation coefficients among the key variables. A moderate and statistically significant positive association was identified between sQ-HPF and NEQ scores (r = 0.288, P < 0.001). Similarly, sQ-HPF and PSQI scores were weakly but significantly correlated (r = 0.216, P < 0.001). NEQ and PSQI scores also showed a comparable moderate correlation (r = 0.285, P < 0.001).

Table 2.

Correlational findings among ultra-processed food intake, night eating behavior, and sleep quality indicators

	sQ-HPF score	NEQ score
sQ-HPF score	-
NEQ score	r = 0.288 P < 0.001*	-
PSQI score	r = 0.216 P < 0.001*	r = 0.285 P < 0.001*

Spearman correlation coefficients are presented *P<0.001 sQ-HPF Screening Questionnaire for Highly Processed Food Consumption, NEQ Night Eating Questionnaire, PSQI Pittsburgh Sleep Quality Index

Table 3 presents findings from the multiple linear regression analysis examining factors independently associated with night eating behavior. Gender (β = 0.084, P = 0.003), age (β = -0.084, P = 0.007), number of main meals (β = -0.088, P = 0.002), snack frequency (β = 0.107, P < 0.001), UPF intake score (β = 0.185, P < 0.001), and sleep quality score (β = 0.254, P < 0.001) were all found to significantly predict NEQ scores. This model accounted for 41.7% of the variance in night eating behavior (R² = 0.417, P < 0.001).

Table 3.

Linear regression model for night eating prediction

	Night Eating Questionnaire score
Model	Beta (β)	t	VIF	95% CI Lower	95% CI Upper	P-value
Gender	0.084	3.026	1.031	0.344	1.613	0.003*
Age (years)	−0.084	−2.697	1.287	−0.061	−0.010	0.007*
Number of main meals	−0.088	−3.124	1.052	−1.383	−0.316	0.002*
Number of snacks	0.107	3.795	1.069	0.348	1.092	< 0.001*
Screening Questionnaire of Highly Processed Food Consumption score	0.185	5.908	1.310	0.241	0.481	< 0.001*
The Pittsburgh Sleep Quality Index score	0.254	8.958	1.078	0.359	0.560	< 0.001*
		R2 = 0.417; p < 0.001*

Variable values: Gender (Male = 1, Female = 0), *P<0.05, Note: VIF = Variance Inflation Factor; all VIF values are below 5, indicating no multicollinearity

Table 4 displays the binary logistic regression results examining factors independently associated with high UPF consumption. Age, marital status, meal and snack frequency, NEQ scores, and PSQI scores were independently associated with elevated UPF intake. Older age was associated with lower odds of high UPF intake (OR = 0.951, 95% CI: 0.933–0.969, P < 0.001), whereas being single increased the likelihood (OR = 1.895, 95% CI: 1.205–2.980, P = 0.006). Higher numbers of meals (OR = 1.382, 95% CI: 1.073–1.779, P = 0.012) and snacks (OR = 1.380, 95% CI: 1.158–1.644, P < 0.001) were also linked to elevated UPF consumption. Additionally, higher NEQ scores (OR = 1.089, 95% CI: 1.061–1.117, P < 0.001) and poorer sleep quality (OR = 1.081, 95% CI: 1.033–1.130, P = 0.001) were positively associated. The model explained 25.3% of the variance (R² = 0.253), with visual representations provided in Fig. 2.

Table 4.

Logistic regression model for prediction of highly processed food consumption

	Highly Processed Food Consumption
Model	OR	%95 CI	Beta	Standard error	P-value
Age (years)	0.951	0.933–0.969	−0.051	0.010	< 0.001*
Marital status	1.895	1.205–2.980	0.639	0.231	0.006*
Number of main meals	1.382	1.073–1.779	0.323	0.129	0.012*
Number of snacks	1.380	1.158–1.644	0.322	0.089	< 0.001*
Night Eating Questionnaire score	1.089	1.061–1.117	0.085	0.013	< 0.001*
The Pittsburgh Sleep Quality Index score	1.081	1.033–1.130	0.077	0.023	0.001*
	R2 = 0.253; p < 0.001*

Variable values: Marital status (Single = 1, Married = 0), OR Odds ratio, CI Confidence interval, *P<0.05, For categorical variables, reference groups were as follows: gender – female (0); marital status – married (0)

Fig. 2.

Factors associated with ultra-processed food consumption

Table 5 details the outcomes of a logistic regression model evaluating the factors independently associated with poor sleep quality. The analysis found that gender, marital status, night eating behavior, and ultra-processed food consumption were independently associated with poor sleep quality. Compared to male participants, females demonstrated greater odds of experiencing poor sleep (OR = 0.671, 95% CI: 0.505–0.891, P = 0.006). Individuals who were single had significantly higher odds of poor sleep than their married counterparts (OR = 2.297, 95% CI: 1.721–3.066, P < 0.001).

Table 5.

Logistic regression model for prediction of sleep quality

	Sleep Quality
Model	OR	%95 CI	Beta	Standard error	P-value
Gender	0.671	0.505–0.891	−0.399	0.145	0.006*
Marital status	2.297	1.721–3.066	0.832	0.147	< 0.001*
Night Eating Questionnaire score	1.079	1.050–1.109	0.076	0.014	< 0.001*
Screening Questionnaire of Highly Processed Food Consumption score	1.130	1.075–1.188	0.122	0.026	< 0.001*
	R2 = 0.239; p < 0.001*

Variable values: Gender (Male = 1, Female = 0), Marital status (Single = 1, Married = 0), OROdds ratio, CIConfidence interval, *P < 0.05

Furthermore, elevated NEQ scores (OR = 1.079, 95% CI: 1.050–1.109, P < 0.001) and greater UPF intake (OR = 1.130, 95% CI: 1.075–1.188, P < 0.001) were associated with an increased likelihood of poor sleep outcomes. The regression model explained approximately 23.9% of the variance in sleep quality (R² = 0.239), and findings are visually represented in Fig. 3.

Fig. 3.

Factors associated with sleep quality

Forest plot displaying odds ratios (ORs) and 95% confidence intervals for factors independently associated with ultra-processed food consumption. Variables included in the logistic regression model were age (in years), marital status (0 = married, 1 = single), number of main meals, number of snacks, Night Eating Questionnaire (NEQ) score, and Pittsburgh Sleep Quality Index (PSQI) score. The vertical dashed line represents the null value (OR = 1.0), indicating no association.

Forest plot displaying odds ratios (ORs) and 95% confidence intervals for factors independently associated with poor sleep quality. Variables included in the logistic regression model were gender (male = 1, female = 0), marital status (0 = married, 1 = single), Night Eating Questionnaire (NEQ) score, and Screening Questionnaire for Highly Processed Food Consumption (sQ-HPF) score. The vertical dashed line represents the null value (OR = 1.0), indicating no association.

Discussion

The diagram visualizes the bidirectional associations between ultra-processed food intake, night eating behavior, and sleep quality. Arrows indicate the tested associations based on multivariable regression models and correlation analyses. Other covariates included in the models (e.g., gender, age, number of meals/snacks, marital status) are not displayed to enhance visual clarity, as the primary focus is on the interrelationships among the three main behavioral variables.

This study is among the few that have simultaneously evaluated the associations between ultra-processed food (UPF) consumption, night eating behavior, and sleep quality within a single analytic framework. Figure 4 visually summarizes the hypothesized relationships among the three primary behavioral variables. The figure reflects the bidirectional associations tested through multivariable regression and correlation analyses and supports the theoretical structure of the study. The results indicate a positive and statistically significant association between poorer sleep quality and both high UPF consumption and more pronounced night eating habits. Given the cross-sectional design, these associations do not imply temporal or causal directionality. Our findings are consistent with previous literature and contribute to a more nuanced understanding of the complex dietary and behavioral pathways that influence sleep health.

Fig. 4.

Conceptual framework illustrating the hypothesized bidirectional relationships among ultra-processed food consumption, night eating behavior, and sleep quality

Recent systematic reviews and meta-analyses published in the past few years further support these associations, highlighting consistent links between ultra-processed food intake, disrupted eating behaviors, and impaired sleep parameters such as sleep quality, insomnia symptoms, and circadian rhythm misalignment [13, 14, 36, 37]. These findings reinforce the associations we observed in the current study and strengthen the biological plausibility of the proposed mechanisms.

To our knowledge, this is the first study in Türkiye to assess UPF consumption, night eating behavior, and sleep quality concurrently using the sQ-HPF, NEQ, and PSQI within a unified analytic approach. However, several cross-sectional studies among Turkish populations have explored one or two of these factors in combination: Eroğlu et al. [22] reported a positive association between night eating syndrome and UPF consumption; Ekici et al. [38] observed similar levels of night eating behavior among Turkish adults; Mengi Çelik et al. [39] found that eating behaviors significantly influence UPF intake; and Çakir et al. [40] demonstrated correlations between poor sleep quality and unhealthy dietary patterns. By integrating all three components, the present study aims to address this gap in the existing Turkish literature.

By integrating all three measures, the present study addresses this gap in the Turkish literature. Previous systematic reviews and meta-analyses have demonstrated associations between high UPF consumption and poorer sleep outcomes, which corroborate the patterns observed in the present study [4, 13, 14].

Ultra-processed foods may impair sleep quality through multiple physiological pathways. These include neuroendocrine disruptions—such as increased ghrelin and cortisol levels and decreased leptin concentrations—that promote hunger and impair melatonin secretion [41–44]. Diets high in added sugars and fats may stimulate the hypothalamic–pituitary–adrenal (HPA) axis, leading to circadian misalignment and poor sleep architecture (St-Onge, Mikic, & Pietrolungo [25]; Gangwisch et al., [42]. Additionally, recent research has highlighted the role of the gut-brain axis, suggesting that poor dietary patterns may alter the gut microbiota and affect serotonin and melatonin synthesis, further disturbing sleep–wake cycles [11, 25].

Chronic consumption of ultra-processed foods also appears to reduce dietary intake of key micronutrients, such as magnesium, zinc, and B vitamins. These nutrients play essential roles in melatonin synthesis and circadian rhythm regulation. Their deficiency may contribute to sleep disorders [43].

Several studies have shown that consuming a large portion of daily caloric intake during the evening or night hours may suppress nocturnal melatonin secretion by delaying the onset of pineal activity, thereby disrupting the natural sleep–wake cycle and misaligning circadian rhythms [15, 20]. Night eating has also been linked to increased cortisol levels and dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis, contributing to impaired sleep continuity [45]; Rogers, Banks, & Jenkins [46]; Sakthivel, Hay, & Mannan [47]. Furthermore, frequent nighttime eating may lead to desynchronization of peripheral clocks, including those regulating glucose metabolism and appetite signaling, further disrupting sleep architecture (Sakthivel, Hay, & Mannan [47].

Particularly, we discovered a distinct correlation between night eating behavior and UPF consumption. This may be partly explained by the high palatability and convenience of UPFs, which can promote late-night snacking and disrupt normal metabolic and circadian processes [2, 30]. In addition, their low fiber and protein content may lead to inadequate satiety responses, further encouraging hedonic eating in the evening hours [48, 49].

The observation that these associations remained statistically significant after controlling for key covariates represents a notable contribution of this work. Although descriptive statistics revealed no major gender-based differences in most sociodemographic and behavioral variables, males in the current study reported significantly higher night eating scores and lower sleep quality. These patterns were further supported by the regression analyses, where gender was found to be independently associated with both night eating behavior and sleep quality. This finding aligns with prior studies in adult populations indicating that men may experience poorer sleep and more dysregulated eating behaviors compared to women, possibly due to sex-specific hormonal regulation (e.g., melatonin, cortisol), behavioral tendencies, and psychosocial influences [50, 51]. Additionally, a previous study conducted in Türkiye also showed gender-related differences in the association between sleep and dietary patterns.

Cross-national comparisons reveal both consistencies and contextual differences. Several international studies have demonstrated associations between UPF intake, night eating behaviors, and poor sleep quality [25, 36, 37], in line with the findings of the present study. However, cultural and dietary habits may shape the expression of these behaviors across populations. The elevated night eating scores observed in our sample may be partially attributed to culturally embedded eating patterns in Türkiye, such as the frequent consumption of refined carbohydrates at dinner (e.g., bread, pastries) and traditional night-time snacks consumed with tea [52, 53]. In addition, the relatively young and highly educated profile of our participants may influence dietary timing and sleep behaviors, which may differ from patterns observed in Western populations. These contextual and demographic factors should be considered when interpreting cross-national comparisons.

We caution against inferring causality from these findings due to the cross-sectional nature of the study. Previous research has frequently cited obesity, depression, or physical inactivity as mediating variables between diet and sleep disturbances [54–57]; Shen, Zhu, Liu, & Jia [58]). Although modest in strength, the associations identified between UPF intake and sleep quality are in line with earlier research and may hold clinical significance given the cumulative impact of dietary patterns on sleep health.

From a wider perspective, psychosocial elements like stress, erratic work hours, and socioeconomic limitations could influence both sleep results and dietary habits. UPFs may be more important for night shift employees or those with limited access to fresh food alternatives; stress-related cortisol dysregulation could promote both night eating and bad sleep. Poor sleep may still reasonably cause more UPF intake and night eating as a coping mechanism, therefore generating a bidirectional cycle [13, 59]. This supports the idea that dietary and sleep patterns should be seen in the larger ecological framework of contemporary lifestyles and environmental stressors (Aseem, Chaudhry, & Hussain [60]; Hall et al., [61, 62].

Although previous studies have reported associations between ultra-processed food consumption, night eating behavior, and sleep disturbances, many of these investigations suffer from limitations such as small sample sizes, lack of validated measurement instruments, and failure to adjust for key sociodemographic variables [13, 18]. Additionally, some studies focus solely on sleep duration, rather than assessing sleep quality in a comprehensive manner. The simultaneous evaluation of these three variables is also rare in the current literature, limiting the ability to understand their potential interactive effects [14]. The present study addresses these gaps by utilizing a large adult sample, employing validated instruments across all domains (sQ-HPF, NEQ, PSQI), and applying multivariable regression models to adjust for confounders. This approach allows us to gain a more in-depth understanding of the intersecting influences of dietary and behavioral factors on sleep health.

This study has several strengths, including a large sample size and the use of validated instruments to assess ultra-processed food consumption, night eating behavior, and sleep quality. The use of multivariable regression models also enhanced the analytical rigor by adjusting for key sociodemographic factors. This integrative approach not only strengthens the analytical rigor but also contributes to a more comprehensive understanding of the behavioral pathways linking diet and sleep health. This integrative approach not only fills a notable gap in the literature but also enables a more comprehensive understanding of the behavioral pathways linking diet and sleep health.

However, the study’s cross-sectional design precludes causal inference. In addition, reliance on self-reported data, particularly for anthropometric measurements and dietary intake, may introduce reporting bias. Moreover, the use of an online self-administered questionnaire (Google Form) may carry limitations related to measurement accuracy, which should be taken into account when interpreting the findings.

Additionally, the use of a snowball sampling method through social media may have introduced selection bias, thereby limiting the generalizability of our findings to broader populations. Additionally, the relatively low internal consistency (Cronbach’s alpha = 0.65) of the Turkish version of the sQ-HPF may be considered a limitation in interpreting the findings related to UPF consumption; however, this level is generally acceptable for newly adapted tools in non-interventional, descriptive studies. Although the tools used in this study (sQ-HPF, NEQ, and PSQI) have been previously validated in Turkish populations through rigorous adaptation procedures, subtle cultural nuances in item interpretation may still exist and should be considered when interpreting the findings.

Socioeconomic status and work-related factors were not within the scope of the present study and thus not assessed; however, we recognize their potential importance in influencing dietary and sleep behaviors. In addition, educational level was not assessed in the current study, although it may also play a moderating role in shaping dietary habits, health literacy, and sleep behaviors. Therefore, future studies should incorporate these variables to provide a more comprehensive understanding of these complex relationships.

Although physical activity and screen time are important behavioral factors that may influence both sleep and dietary behaviors, we did not assess these variables in the present study to reduce participant burden. Additionally, overall diet quality was not evaluated. Although UPF intake is often used as a proxy for poor dietary habits, a comprehensive dietary quality index (e.g., Healthy Eating Index or Mediterranean Diet Score) was not included. This limits the ability to determine whether the observed associations are independent of broader dietary patterns. In addition, no mediation or moderation analyses were conducted.

Additionally, individual chronotype and meal timing (chrononutrition) were not assessed in the present study, although both may influence night eating patterns and sleep quality. Individuals with an evening chronotype may be particularly susceptible to circadian misalignment and late-night eating behaviors. Future studies should consider incorporating these chronobiological variables to gain deeper insights into the temporal dynamics of eating and sleep behaviors.

Although BMI was included in preliminary analyses, it did not demonstrate significant associations with key outcomes and was not included in the final models. Additionally, potential moderators such as perceived stress were not assessed due to survey constraints. These factors should be considered in future studies to explore more complex interrelationships.

While not directly examined in the present study, psychological variables such as depressive symptoms, emotional eating, and perceived stress have been shown to mediate the relationship between diet and sleep quality. Emotional dysregulation may contribute to both unhealthy eating behaviors and poor sleep patterns, forming a potential biopsychosocial loop [63, 64]. Future research should incorporate these factors to provide a more comprehensive understanding of the mechanisms linking UPF intake, night eating, and sleep disturbances. Moreover, future studies may consider using advanced statistical approaches such as path analysis or structural equation modeling (SEM) to evaluate whether night eating behavior acts as a mediator in the relationship between ultra-processed food consumption and sleep quality. Such techniques could offer deeper insight into the indirect pathways and complex interactions among these variables.

Furthermore, BMI was derived from self-reported height and weight, and this anthropometric indicator alone may not capture the underlying metabolic health status or systemic inflammation. As the current study did not aim to assess clinical parameters related to metabolic stress or inflammatory markers, caution is warranted in interpreting BMI-related inferences.

Additionally, the explanatory power of the regression models was limited (R² = 0.24–0.41), indicating that the included variables only partly account for the variance in night eating behavior and sleep quality. This suggests the potential influence of additional unmeasured psychological, environmental, or behavioral factors that were beyond the scope of this study.

Moreover, although the study targeted adults, the relatively young mean age of the sample (~ 28.7 years) may limit the generalizability of findings to older adult populations. Therefore, future studies including broader and more balanced age groups may help to identify potential age-related patterns in the associations among dietary behaviors, night eating, and sleep quality. In addition, although subgroup analyses by gender were presented, further exploration based on age and BMI did not yield statistically significant results and were therefore not included. Although descriptive comparisons revealed some differences in NEQ and PSQI scores by gender, the regression models identified gender as a statistically significant factor in both outcomes. However, the study was not designed to explore sex-specific pathways in detail. Future studies could benefit from incorporating gender-stratified models to better capture potential differences in behavioral patterns. In addition, future studies could explore other potential interactions in greater depth to uncover subgroup-specific patterns beyond gender.

Building upon these observations, future studies may test whether reducing ultra-processed food consumption or modifying night eating behavior improves sleep quality through interventional approaches. Likewise, improving sleep hygiene could potentially reduce UPF intake and disordered eating tendencies. Longitudinal studies may also help clarify whether poor sleep precedes increased UPF consumption via hormonal dysregulation, or vice versa. These hypotheses could guide the development of integrated dietary and behavioral interventions.

Finally, these results suggest that effective interventions to improve sleep quality should address not only what people eat, but also when they eat. Programs that integrate nutritional education with chronobiological and behavioral strategies -such as reducing UPF exposure, promoting earlier meal timing, and supporting sleep hygiene- may yield synergistic benefits for both dietary and sleep health. Addressing ultra-processed food consumption and night eating behavior may not only improve sleep quality but also contribute to enhanced health-related quality of life, particularly among young adult populations—such as the sample in the present study—who may be especially vulnerable to poor dietary and sleep patterns.

Authors' contributions

EM.E. Ö.MÇ. P.G and AH.G. Conceptualization, Data Curation, Writing –Original Draft, EM.E. Formal Analysis, Supervision, Prepared Figs. 1, 2, 3 and 4.

Funding

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

Data availability

The datasets generated and analyzed during the current study are not publicly available due to institutional and ethical regulations. However, they are available from the corresponding author upon justified and documented request, subject to approval by the relevant ethics committee and institutional data protection policies.

Declarations

Ethics approval and consent to participate

This study was approved by the University of Health Sciences Gülhane Scientific Research Ethics Committee (Approval No: 2024/570), and conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants prior to participation. All collected data were kept confidential, anonymized prior to analysis, and stored securely to protect participant privacy.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.