STRENGTHS AND LIMITATIONS OF THIS STUDY

The study was conducted with a sizeable, representative sample of the US adult population.

This study used the Visceral Adiposity Index (VAI) to assess adipose dysfunction, potentially providing a more accurate evaluation than the traditional body mass index.

This study incorporated a larger number of confounding elements.

The cross-sectional design of the study precludes any inference of causality between sleep duration and VAI.

The study recognises the potential for a reciprocal connection between VAI and sleep duration, as well as the chance of recall bias when assessing self-reported sleep duration.

Introduction

In recent decades, obesity has become a worldwide pandemic and a major public health concern.1 Epidemiological evidence suggests that obesity is connected a greater likelihood of developing diabetes, cardiovascular disease (CVD) and cancer, leading to a substantial financial burden.2 Therefore, prevention and treatment of obesity should be carried out appropriately. It has been established that behavioural factors, in particular diet and exercise, contribute significantly to the emergence of obesity.3–6 Obesity is characterised by a prolonged mismatch between energy intake and energy expenditure. Adopting a healthy lifestyle that involves eating a nutritious diet and engaging in regular exercise could potentially reduce the energy intake and enhance calorie expenditure, thus preventing obesity.

In recent years, in addition to other health behaviours, sleep has been identified as a potential factor in obesity. Sufficient sleep and regular sleeping patterns are essential components of a healthy lifestyle. According to previous studies, sleep deprivation has been linked to metabolic diseases such as CVD and obesity. For example, Kim et al concluded from the Korean Health and Nutrition Survey that sleep deprivation had a significant association with both general obesity and abdominal obesity.7 In addition, Jaiswal et al found that a shorter sleeping period was significantly linked to higher body mass index (BMI).8 Recently, an updated meta-analysis by Bacaro et al confirmed that short sleeping period was related to obesity, yet there is no connection between long sleep duration and obesity.9

Despite the consensus of various studies that there is an association between sleep duration and obesity, there is some discrepancy in the data. Vgontzas et al revealed that there was no notable connection between short sleep duration and the development of obesity.10 Nagai et al did not observe a notable association between sleep duration and the likelihood of obesity among Japanese citizens living in the community.11 The discrepancies in the relationship between sleep and obesity may be attributable to the methods used to measure sleep and obesity, adjustments made to account for potential confounding factors and differences in the characteristics of study populations. Most of the studies discussed above relied on BMI to measure obesity. However, this method was not able to differentiate between subcutaneous adiposity and visceral adiposity, thus making it difficult to understand the impact of sleep on the development of adiposity.12 An increasing amount of evidence has indicated that visceral fat, rather than subcutaneous fat, is correlated to a range of adverse outcomes of obesity.13 Despite the fact that various imaging techniques, like CT and MRI, could accurately determine the distribution of adipose tissue,14 accurately evaluating visceral body fat remains a major challenge.

Given the high cost and difficulty of implementation, these technologies are not suitable for the majority of epidemiological or population-based studies. A mathematical index, known as the Visceral Adiposity Index (VAI), has been developed to measure the amount of visceral body fat. The calculation of VAI is primarily derived from simple characteristics, such as BMI, waist circumference (WC), high-density lipoprotein cholesterol (HDL-C) and triglycerides (TGs).14 It has been established through multiple studies that VAI is a reliable measure of the presence and performance of visceral adipose tissue, which is specific to gender.14–16 Besides, VAI has been frequently employed to estimate the dangers of illnesses such as diabetes, hyperuricaemia and CVD.17 18 To date, no investigations have clarified the connection between sleep and VAI in a substantial-scale cohort study. To bridge this research gap, we conducted this cross-sectional study to examine the relationship between sleep and VAI based on data from The National Health and Nutrition Examination Survey (NHANES).

MethodPatient and public involvement

No patient was involved in the design and conduct of this study.

Study population

The NHANES uses a complex, multistage probability design to collect and evaluate data that accurately reflect the health and nutritional condition of individuals, both adults and children, who are part of the non-institutionalised civilian population of the USA (https://www.cdc.gov/nchs/nhanes/index.htm). Data were extracted from six cycles of the NHANES database, which were 2007–2008, 2009–2010, 2011–2012, 2013–2014, 2015–2016 and 2017–2018. In the present study, participants ≥18 years of age who were included in the study had to provide complete information regarding sleep duration, VAI, age, gender, race, marital status, education status, poverty, smoking status, drinking status, physical activity, energy intake and self-reported chronic diseases including hypertension, stroke, heart attack, congestive heart failure, coronary heart disease, diabetes and cancer. Participants who had missing data for VAI, sleep duration or other important covariates at the baseline assessment were excluded. Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Assessment of VAI

The NHANES collected anthropometric data (height, weight and WC) and lipid parameters (TG, HDL-C) to calculate VAI according to previous studies.14 BMI was calculated by dividing a person’s weight in kilograms by the square of their height in metres. The formula for men was VAI=(WC/(39.68+1.88×BMI)))×(TG/1.03)×(1.31/HDL-C) and for women it was VAI=(WC/(36.58+1.89× BMI)))×(TG/0.81)×(1.52/HDL-C). The TG and HDL-C were expressed in mmol/L while the WC was expressed in cm. It was found that a higher VAI correlated with a greater estimated visceral adiposity, which was a risk factor for CVD.19

Determination of sleep duration

The amount of sleep was determined in the NHANES through the following question: ‘How much sleep do you usually get at night on weekdays or workdays?’. According to the American Academy of Sleep Medicine and the Sleep Research Society, self-reported sleep duration can be divided into three categories: short (less than 7 hours each day), middle (7–9 hours every day) and long (over 9 hours each day).20 21 Despite being a retrospective self-report and not a concrete measure of typical sleep, the sleep duration item from the NHANES has been found to be reliable in multiple studies.22 23

Covariates

In the current study, we included the following potential confounding factors for adjustment: basic demographics including age, gender (Female/Male), race (white, Mexican, black other), marital status (unmarried/married), education (grade or less/high school/some college/college or more), family poverty income ratio, smoking status (never/former/now), alcohol drinking status (never/former/mild/moderate/heavy), BMI, physical activity (inactive/insufficient/sufficient), healthy diet was measured by Healthy Eating Index-2015 energy intake,24 and self-reported chronic diseases, including hypertension, diabetes, heart attack, coronary heart disease, stroke, congestive heart failure and cancer. According to the WHO recommendations, BMI less than 18.5 kg/m2 is classified as underweight, 18.5 to <25 kg/m2 is deemed as a healthy weight, 25.0 to <30 kg/m2 is classified as overweight and any BMI of 30.0 kg/m2 or higher is considered to indicate obesity (https://www.who.int/europe/news-room/fact-sheets/item/a-healthy-lifestyle---who-recommendations).

Statistical analysis

In compliance with NHANES recommendations, a weighted analysis was performed to address the intricate sampling design in the study. The sampling weight was calculated by multiplying the 2-year weights of the Mobile Exam Centre by one-sixth. Continuous variables are presented as weighted mean, and categorical variables are presented as weighted percentage. Individuals were divided into three categories based on duration of sleep, namely those with short, medium and long sleep patterns. To analyse the differences in baseline characteristics among the three sleep duration groups, one-way analysis of variance and χ2 test were employed. The weighted multiple generalised linear regression was used to assess the association between sleep duration (both as a categorical and continuous variable) and VAI. The results were expressed as β and 95% CIs. We also assessed the variance inflation factors of each covariate. These analyses were performed in three models. Model I was the crude model without any adjustment. Model II was adjusted for age and model III was adjusted for model II, plus gender, race, education, marital status, poverty, smoking status, alcohol status, physical activity, energy intake and self-reported chronic diseases. Since the relationships between sleep duration and VAI may differ by other potential variables, we conducted subgroup analyses stratified by age, gender, BMI status and physical activity status. Furthermore, we used restricted cubic spline (RCS) analysis to examine the nonlinear association between sleep duration and VAI while accounting for confounding variables.25 To investigate the threshold impact of sleep duration on the VAI and determine the inflection point, we employed smooth curve fitting and generalised additive models. Statistical analyses were conducted using R statistical software (V.4.2.2, http://www.R-project.org) and EmpowerStats (http://www.empowerstats.com). All results were deemed significant at p<0.05.

Results

A total of 11 252 participants were included in our study based on our inclusion and exclusion criteria. Figure 1 summarises the detailed process of cohort selection. The average age of the participants was 49 years, ranging from 20 to 80 years, with a majority being female. The average sleep duration and VAI were 7.05 hours/day and 2.03. There were significant differences among age, gender, race, marital status, education status, poverty, smoking, drinking status, energy intake, healthy diet, physical activity, VAI, history of hypertension, diabetes, stroke and congestive heart failure among different sleep duration groups (table 1). In terms of sleep duration as a continuous variable, there was a reduction of 0.05 in the VAI with each additional hour of sleep in both the crude and adjusted models which accounted for multiple confounders (online supplemental table S1). When the sleep duration was analysed as a categorical variable, compared with the short sleep group, the middle sleep group had a lower VAI in the crude (β=−0.20, 95% CI −0.33 to −0.06; p<0.01) and the multiple confounders adjusted model (β=−0.15, 95% CI −0.28 to −0.01, p=0.04) (online supplemental table S1). There was no significant difference in VAI among the long sleep group in both the crude and multiple confounders adjusted models (p>0.05) (online supplemental table S1). As shown in online supplemental figure S1, the middle sleep duration group had a significant lower VAI compared with the short sleep duration group. No significant difference was observed between the middle and long sleep duration groups. Thus, we selected the middle sleep duration as our reference.

Table 1

Characteristics of included population based on sleep duration (N=11 252)

Figure 1

The detailed flow chart of included population in the current study. NHANES, National Health and Nutrition Examination Survey; VAI, Visceral Adiposity Index.

As shown in table 2, compared with individuals with a middle sleep duration, those with a short sleep duration had a significantly higher VAI. No significant difference was found in individuals with a long sleep duration (table 2). To estimate the impact of other variables on the association between sleep duration and VAI, we performed subgroup analysis stratified by age, gender, BMI status and physical activity. In subgroup analysis, significant associations between short sleep duration and VAI were observed among participants of young age (≤65 years), males and participants with normal weight and sufficient physical activity (table 3). A significant association was observed between long sleep duration and VAI among participants with normal weight (table 3). Through the utilisation of RCS, a non-linear association between sleep duration and VAI was identified postadjustment for various covariates (p<0.05). An L-shaped relationship between the duration of sleep and VAI was revealed by the smooth curve fitting, which was displayed in figure 2. Before the inflection point (sleep duration=7.5 hours/day), longer sleep duration was associated with lower VAI. However, VAI gradually increased with the sleep duration thereafter, although it was not significant (figure 2).

Table 2

Multivariate analysis of association between sleep duration and Visceral Adiposity Index

Table 3

Subgroup analysis of multivariate analysis of association between sleep duration and Visceral Adiposity Index

Figure 2

Non-linear relationship between sleep duration and Visceral Adiposity Index.

Discussion

In the current study, our findings revealed that individuals who experienced short sleep duration had a significantly higher VAI compared with those in the middle sleep group. However, this significant association was not seen in the long sleep group. An L-shaped relationship was observed between sleep duration and VAI. It was observed that a sleep duration of less than 7.5 hours had a significantly negative association with VAI. When sleep duration was more than 7.5 hours, VAI gradually increased with the sleep duration, although not significantly. It was observed that young male participants with normal weight and sufficient physical activity had a noteworthy connection between sleep duration and VAI. It is widely acknowledged that the relationship between sleep duration and VAI is very significant and continues to draw much consideration. In 2000, a comprehensive epidemiological study was undertaken to explore the association between obesity and sleep duration, and the results indicated that obesity was associated with shorter sleeping hours.26 Over the years, multiple investigations have been conducted to determine the link between obesity and the amount of sleep. Early in 2008, Cappuccio et al conducted a meta-analysis based on cross-sectional studies, and it indicated that short sleepers, both children and adults, had a greater risk of obesity.27 Later, an updated meta-analysis by Guimarães et al further confirmed the notable association between short sleep and the onset of obesity.28 No link has been established between BMI and sleep duration in certain studies.29 However, most of the aforementioned studies opted for BMI, as it is a widely used and easily obtained measure of adiposity, which makes it convenient to analyse and interpret. Prior research has indicated that BMI may not be the most reliable measure of adiposity since it is influenced by water retention and muscle mass. It is suggested that the obesity paradox is the result of flaws in the BMI.30 31 VAI is a marker of visceral adiposity that reflects dysfunction in adipose tissue and is associated with an unhealthy metabolic phenotype, regardless of nutritional status.32 A high VAI was found to be significantly correlated with an augmented risk of developing diabetes and metabolic syndrome over the long term, regardless of the individual’s age, race or BMI.33

To the best of our knowledge, this is the first time a non-linear relationship between sleep duration and VAI has been observed; no study has previously explored this connection. Short sleep duration was significantly associated with a higher VAI, which was partly consistent with previous studies. It is thought that the association between sleep and weight status is due to several factors, although the precise cause is yet to be determined. It is believed that inadequate sleep can cause a decrease in leptin secretion and insulin sensitivity, causing an imbalance between energy intake and expenditure, which can lead to fat accumulation. Additionally, long-term lack of appetite-suppressing hormones can cause an increase in caloric intake and weight gain.34 35

The subgroup analysis revealed that a noteworthy association between sleep duration and VAI was only observed in young, male participants with normal weight and sufficient physical activity. Grandner et al noted a stronger association between sleep duration and BMI among younger individuals, which is in agreement with the findings of this study.23 Saito et al conducted a comprehensive cross-sectional study in East Asia and identified an association between gender and sleep-related obesity.36 The study demonstrated that men who slept for longer periods of time had higher levels of visceral obesity, whereas the opposite was true for women.36 This divergence could be partially due to the varying demographics. Cho et al revealed a variety of associations between sleep duration and obesity that depended on gender and age.37 The current study showed that participants with normal weight and sufficient physical activity had an increased risk of VAI with short sleep duration. Therefore, it is important to maintain a regular sleep schedule for the normal weight population with sufficient physical activity, so as to reduce the occurrence of overweight or obesity.

The current study adopted VAI instead of BMI as the measurement of adipose dysfunction, which may offer a more valuable evaluation of adiposity.14 Additionally, a subgroup analysis was conducted and it uncovered some noteworthy results. However, there are some limitations. First, it is important to note that this was a cross-sectional analysis, thus precluding any inferences of causality. Additionally, the potential bidirectional relationship between VAI and sleep duration could not be excluded. A prior study has proposed that obesity may be a precursor to inadequate sleep, suggesting a potential reciprocal relationship between sleep duration and body weight.38 Further research through multiple longitudinal cohorts with different age groups would help to validate and expand on these findings. Mechanistic studies are also essential to draw definitive conclusions. Second, it is important to consider the potential for recall bias, as the sleep duration was self-reported. Moreover, the accuracy of dietary evaluation may have been overestimated or misclassified. Owing to recall bias, cross-sectional nature, and a dearth of information concerning sleep habits over the medium to long term, most population-level sleep data are restricted. The utilisation of gold-standard techniques, such as actigraphy, to quantify sleep duration may be employed for future investigations. Third, we cannot be certain whether our conclusion can be applied to children; thus, further research is necessary to determine whether the same benefits can be observed in the other populations.

Short sleep duration was found to be independently linked to increased VAI, particularly in men. To gain a better understanding of this association, longitudinal studies that measure sleep and adiposity over an extended period of time are necessary. Such findings could be beneficial in providing clinical advice and exploring the relationship between sleep and obesity.

Data availability statement

Data are available on reasonable request. Data are accessible on reasonable request. The datasets used and analysed in this study can be obtained from the corresponding author on request.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants and all NHANES protocols have been approved by the National Center for Health Statistics Research Ethics Review Board (NCHS IRB/ERB Protocol Number: #2005-06, #2011-17,#2018-01). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

We are thankful to the personnel and members of the National Center for Health Statistics of the Centers for Disease Control and Prevention (CDC) and the individuals who participated in the National Health and Nutrition Examination Survey. Grammar correction and language polishing are performed using the Stork software (https://www.storkapp.me).