Unfavorable Mealtime, Meal Skipping, and Shiftwork Are Associated with Circadian Syndrome in Adults Participating in NHANES 2005–2016

The study sample included NHANES data from the surveys conducted between 2005 and 2016. The NHANES is conducted annually in the US to assess the health of the non-institutionalized population, using a probability sampling design for participant recruitment. Various assessment methods are used, including surveys and physical, medical, dental, and biochemical tests.

In this study, six NHANES cycles were combined to generate an analytical sample of 10,486 adults who were 20 years and older (Figure 1). Exclusion criteria included the following: (1) implausible energy intake based on two 24 h food recalls (defined as an intake of <500 kcal and >6000 kcal in men, and <500 kcal and >5000 kcal intake in women); (2) participants with incomplete data on CircS components; and (3) pregnant women, due to irregular patterns of sleep and hormonal and metabolic changes during pregnancy [27].

Covariates included in the analysis were pre-selected based on current knowledge: age, sex, ethnicity (Non-Hispanic White, Non-Hispanic Black, Mexican American, and Other), energy intake measured via two 24 h recalls, physical activity (measured as Metabolic Equivalent of Task (MET) minutes per week using data from the validated Global Physical Activity Questionnaire and recoded as <600, 600–1200, and ≥1200 MET min/week), education (lower than 11th grade, high school, some college, and higher than college), smoking status (never, former, or current smoker), alcohol (categorized as “yes” or “no” through self-reported consumption over the previous 12 months), and Healthy Eating Index 2015 (HEI-2015). The HEI-2015 is a validated tool for examining how well an individual’s food intake aligns with the Dietary Guidelines for Americans [31]. It has 13 components, 9 of which highlight adequacy (foods to select more often for optimum health), and 4 components that focus on moderation (items to limit). The scores range from 0–100, and higher HEI-2015 scores reflect a better diet quality. The HEI-2015 scores were divided into quartiles in the analysis. The socioeconomic status (SES) was evaluated by dividing family income by the poverty threshold and classifying it into three categories: <1.3, 1.3–3.5, and >3.5 Poverty Income Ratio.

Sample characteristics were presented according to the midpoint of intake or meal skipping if meals were omitted. To test the differences between groups, Chi-square test and ANOVA were used for categorical and continuous variables, respectively. Due to the skewed distribution and wide dispersion of values in each quartile of meal timing, the interquartile range (IQR) was used to define the time ranges in each corresponding quartile. Multivariable logistic regression was performed to examine the associations of CircS with meal timing, meal skipping, and shiftwork. Three multivariable models were used: model 1, adjusted for age, sex, ethnicity, and energy intake; model 2, further adjusted for physical activity, education level, smoking status, and alcohol consumption; and model 3, further adjusted for quartiles of the HEI. Sampling weights were used to account for the complex survey design of NHANES data. In subgroup analyses, potential effect modifications of CircS with meal timing and shiftwork by various factors (age, sex, ethnicity, energy intake, educational level, physical activity, smoking status, and alcohol consumption) were explored by adding product terms into the logistic regression models. All the analyses were performed using STATA 18 (Stata Corporation, College Station, TX, USA). Statistical significance was considered when p-values were less than 0.05.

In total, 10,486 adults (50.9% women) with a mean (SD) age of 50.3 (17.6) years were included in the analysis (Table 1). The unweighted prevalence of meal skipping was high (n = 4372, 41.7%). Meal skippers tended to consume fewer calories than others, and consumed less protein, fat, and carbohydrate overall. Their HEI scores were also lower. Meal skippers were younger and tended to be men, smokers, and of Non-Hispanic Black or Mexican American descent. They were more likely to have lower education levels and income, elevated glucose and triglycerides, reduced HDL-C, and were more likely to report depression symptoms, short sleep, have Circadian Syndrome. Those with a favorable mid-eating time were more likely to be women, college educated, high-income earners, and of Non-Hispanic White descent.

The median times for breakfast, lunch, and dinner were 8:20, 12:30, and 18:30, respectively. Figure 2 shows the distribution of the midpoint of intake between breakfast and dinner. The median (IQR) was 13:15 (12:30–14:15). Among the participants, the unweighted prevalence of shiftwork was 18.2%.

Compared with a favorable mealtime (second quartile, midpoint of intake between 12:30 and 13:15), participants who skipped meals and those with an unfavorable mealtime had 39% (95%CI 1.16–1.67) and 24% (95%CI 1.07–1.44) higher odds of having CircS, respectively (Table 2). The positive association remained after adjusting for the HEI.

The association of CircS by quartiles of the mid-eating time was also investigated (). There was a non-linear association between the midpoint of meal intake and CircS. Both early and late midpoints of meal intake were associated with an increased odds of having CircS. Table S1

Among the components of CircS, meal skipping showed a positive association with all components except elevated triglycerides and blood pressure (Table 3). Those who skipped meals had ORs of 1.23 for central obesity (95%CI 1.03–1.47), 1.29 (95%CI 1.10–1.50) for elevated glucose, 1.22 (95%CI 1.04–1.44) for low HDL-C, 1.57 (95%CI 1.31–1.89) for depressive symptoms, and 1.37 for short sleep (95%CI 1.17–1.61).

Compared to favorable mealtimes, unfavorable mealtimes were associated with 20% higher odds of elevated glucose and short sleep, and 38% higher odds of depressive symptoms (Table 3). No associations were found between unfavorable mealtimes and other CircS components. Similar to meal skippers, participants with unfavorable mealtimes had the highest odds of depressive symptoms and short sleep among all the components of CircS.

Shiftwork was independently associated with CircS (Table 4). Compared to non-shift workers, a 40% higher likelihood of having CircS in shift workers was seen in model 2, and this significant association persisted in the fully adjusted model (OR = 1.37; 95%CI 1.01–1.87). The Healthy Eating Index did not significantly attenuate the relationship between shiftwork and CircS.

Subgroup analyses revealed no significant interactions between mealtimes, SES, and lifestyle in relation to CircS (Table 5). However, the association of CircS with unfavorable mealtimes and meal skippers was only observed in women, middle-aged adults, individuals with lower education levels and HEI scores, non-smokers, and those of Non-Hispanic White descent. Additionally, meal skippers between the ages of 20 and 39, those within the lowest and highest categories of income and physical activity level, high school and college graduates, former smokers, and individuals with the highest HEI scores were positively associated with CircS.

Similarly, although the interaction of shiftwork, SES, and lifestyle in relation to CircS did not reach statistical significance, the association of shiftwork and CircS was seen among men but not women (Table 6).

Our findings on meal timing are in line with emerging evidence identifying the time of eating as a risk factor for impaired metabolic outcomes. Late eating has been associated with CircS components such as elevated glucose [32], waist circumference [30], and triglycerides [30] and low HDL-C [32] and sleep duration [33]. The effect of late eating on cardiometabolic risk factors and weight loss success after a weight loss intervention was investigated in a cross-sectional analysis [30]. Compared to early eaters, late eaters had elevated triglycerides, greater BMIs, and a reduced insulin sensitivity at baseline and were more likely to have weight loss barriers [30]. In an RCT, a late dinner induced nocturnal glucose impairment and reduced the oxidation and mobilization of fatty acids, especially in habitual early sleepers, compared to an early isocaloric dinner [34]. In a Japanese cohort, late eating was associated with obesity, dyslipidemia, and MetS, the latter only in women [35]. In our study, we created a reference midpoint of meal intake to categorize favorable or unfavorable mealtimes in relation to CircS. When the mid-eating time between breakfast and dinner was divided into quartiles, the association with CircS was significantly U-shaped. Both early and late eaters had an increased likelihood of having CircS and were categorized under unfavorable meal timing accordingly. This quartile also included those with a much earlier mid-eating time between breakfast and dinner, ranging from 1:00 a.m. to 12:30 p.m. It reflected individuals with mistimed sleeping and eating habits that continued throughout the night and were not aligned with circadian rhythms.

Similarly, meal skipping is also associated with MetS [36], all-cause and CVD mortality [37], and lower HEI scores [38]. The findings on breakfast versus dinner skipping are conflicting. In a cross-sectional analysis among Korean adults, women who skipped breakfast had higher odds of glucose and triglyceride metabolism impairment, while skipping dinner was associated with a favorable fasting glucose [36]. Another study found no association between skipping breakfast and MetS [39]. In the current study, meal skippers had a higher likelihood of CircS. Compared to other main meals, skipping breakfast showed the strongest association with CircS (results not shown). While other skippers were not significantly associated with CircS, dinner skippers tended to have lower odds of CircS, as corroborated by the existing literature.

Finally, the findings on shiftwork and CircS in the present study are consistent with the previous literature. Circadian disruptions manifesting as MetS and short sleep have been widely reported in shift workers [40].

Over the past decade, Time-Restricted Eating (TRE)—limiting energy intake to a specific timeframe within a 24 h day—has been investigated as a potential approach to improve metabolic risk factors. Its effectiveness may depend on the length of the feeding window and the timing of food intake throughout the day. Confining meals to the morning, in line with the biological clock, as opposed to the evening has been associated with an improved insulin sensitivity, and lower inflammation, oxidative stress, and blood pressure [41,42,43,44]. A network meta-analysis of 12 RCTs on early versus later TRE revealed enhanced improvements in insulin resistance in early TRE (early vs. later TRE: −0.44; 95%CI −0.86 to −0.02; p < 0.05). While weight loss was similar in both interventions, early TRE was further associated with improved blood pressure and glucose metabolism [45].

As proposed by Zimmet et al. [8], metabolic alterations of glucose, blood pressure, and lipid profiles manifesting as cardiometabolic diseases are linked to disruptions in circadian rhythms. The internal circadian clock, consisting of central and peripheral clocks in the brain and peripheral tissues, respectively, generates these rhythms through transcription factors and gene expression, overseeing crucial physiological processes in the body, including metabolism [46]. The transcriptional pathways regulating the internal clock can be modulated and reset by a variety of signals, including light and feeding. For instance, breakfast consumption impacts the expression of clock genes and facilitates the typical patterns of rhythms [47]. In contrast, skipping breakfast impairs the expression of the internal clock and related genes and is associated with an elevated post-meal blood glucose response, irrespective of diabetes status [48]. The disruption of these pathways can lead to weight gain, an elevated blood pressure, impaired insulin sensitivity, and dyslipidemia. Modern lifestyle behaviors like shiftwork, prolonged exposure to artificial light, social jet lag (the misalignment between weekday and weekend sleep schedules), and nocturnal eating have been linked to circadian disruption [4].

Consistent and timely meals are important due to the various health risks associated with irregular eating patterns and meal skipping. Adopting a one-meal-per-day approach may contribute to weight loss, without any benefit to lipid profiles [49], and may be associated with elevated fasting blood glucose levels [50]. Evidence suggests that skipping breakfast leads to an increased body weight and insulin resistance [51] and a reduction in the overall diet quality and intake of the recommended level of nutrients [52]. Notably, a higher prevalence of MetS was observed in those with inconsistent mealtimes in a Swedish cohort of 60-year-old adults [51]. Irregular meals are also associated with an increased likelihood of MetS later in life [53].

Furthermore, prior evidence indicates that meal timing may be related to diet quality. In a Serbian study, earlier breakfast and dinner consumption were associated with an increased consumption of fruits and vegetables and higher Dietary Quality Scores (based on European Union Science Hub recommendations on fruits and vegetables, fiber, sodium, saturated fat, and sugar) [54]. Similarly, in a cross-sectional study of Brazilian women, those with an early midpoint eating time had healthier eating habits as measured by Brazilian HEI scores [55]. Nutrient-rich and diverse dietary patterns are crucial for disease prevention, and these findings emphasize the equally important role of consistent and appropriate meal timing.

This study has several noteworthy strengths. It is the first study to investigate CircS with variables associated to meal timing and shiftwork. Secondly, the use of a robust sampling design improves the generalizability of the findings to the general non-institutionalized US population. Furthermore, we controlled for multiple lifestyle and sociodemographic factors due to the extensive dataset. In addition, subgroup analyses allowed us to investigate interactions of meal timing, meal skipping, and shiftwork with sociodemographic and lifestyle factors in relation to CircS.

This study also has certain limitations, such as the lack of repeated measurements of CircS. Moreover, a self-reported sleep duration was used in the study rather than objective measures. The threshold for identifying symptoms of depression using the PHQ-9 was established at 5 instead of 10, specifically capturing mild depression, but the PHQ-9 is easy and efficient to use. While appropriate statistical models were used to account for confounding, the possibility of residual confounding remains, such as environmental factors including air pollution, noise, and the use of artificial lighting. Additionally, the cross-sectional design of this study limits the ability for causal inferences.