The CHNS

CHNS is an ongoing longitudinal survey to collect health and nutrition data at the individual, household and community levels. Information collected for each family individual included clinical and sociodemographic characteristics, lifestyle factors, anthropometrics, dietary habits and indicators of health status. The study was conducted from 1989 to 2011 with nine rounds of surveys beginning in eight provinces. A ninth province (Heilongjiang) was added in 1997, and three autonomous cities, Beijing, Shanghai and Chongqing, were added in 2011. The survey used a multistage, random cluster progress to draw samples in the provinces. Counties in the nine provinces were stratified based on income levels (low, middle and high) and a weighted sampling scheme was used to randomly select four counties in each province. Additionally, two cities including the provincial capital and a lower income city were selected. By 2011, the survey provinces covered 47% of the Chinese population (according to the 2010 census). The details of the design and data collection for the CHNS have been described elsewhere.22 The survey was approved by the Institutional Review Committees of the University of North Carolina at Chapel Hill and the National Institute of Nutrition and Food Safety, China Centres for Disease Control and Prevention. All participants provided written informed consent.

Study participants

Only the 2009 CHNS data were analysed for the purpose of this study because the 2009 survey collected fasting blood samples for the first time. The sample size of the study was estimated based on parameters, such as confidence level (1-α), power of test (β) and tolerance (δ). In the study, the confidence level was 0.95 (1-α), and the tolerance was 0.3. According to the literature,22 the response rate was 90%, and the pass rate of the questionnaire was predicted to be 70%. Participants were eligible for enrolment if they were 18 years old or older, not pregnant and overweight/obese and had available biomarker data (n=4372). Among them, participants were excluded for the following reasons: missing data on sleep duration (n=47) or metabolic components (n=62) and a history of myocardial infarction and stroke (n=114). A total of 4149 participants (1952 men and 2197 women) between the ages of 18 and 94 years were included in the final analysis.

Data collection and clinical assessments

A self-administered questionnaire was completed for each individual to collect sociodemographic information including age, smoking status, alcohol consumption, medical history of diabetes, hypertension, myocardial infarction or stroke and use of medications for diabetes, hypertension and dyslipidaemia. Smoking status was categorised as never, former or current smoker. Alcohol drinking status in the previous year was evaluated and categorised as either ‘ever’ or ‘never.’

The body weight of the participants was measured to the nearest 0.1 kg, and height was measured to the nearest 0.1 cm in subjects in light clothing and without shoes. Body mass index (BMI) was calculated as body weight divided by the square of the height (kg/m²). Overweight/obesity was defined as a BMI≥23 kg/m² based on the WHO criteria for Asia.23 The waist circumference (WC) of each subject was measured at the midpoint between the lowest rib and the top of the iliac crest. Trained staff obtained blood pressures for each individual after 5 min of seated rest using a mercury sphygmomanometer. The right arm was selected. Patients were required not to wear tight cloth; avoid the following activities 1 hour before blood pressure: intense sports or exercise, eating, drinking (except water), especially drinks with caffeine; expose under very high/very low temperature; take medicine which may affect blood pressure; etc. Measures were collected in triplicate, and the average of the three readings was used in the analyses.

Sleep duration assessment

Sleep duration of all adults was assessed from self-reported questionnaires that asked about the number of hours of sleep obtained in a 24-hour period. For sleep assessment, individuals were asked, ‘How many hours each day do you usually sleep, including daytime and nighttime?’. Short sleep duration was defined as ≤6 hours/day, optimal sleep as 7 and 8 hours/day, and long sleep as ≥9 hours/day. Self-reported has been shown to correlated highly with objective measures (Pearson’s correlation coefficients, range 0.84–0.95).24

Biochemical assessment

Fasting (8–12 hours overnight) blood samples were obtained from each participant and were separated immediately and frozen at −86°C for subsequent laboratory analysis. Serum glucose was measured by the GOD-PAP method using a Hitachi 7600 analyzer (Randox Laboratories, UK). Serum insulin was measured using radioimmunoassay (North Institute of Bio-Tech, China). Insulin resistance was assessed using the homoeostasis model assessment of insulin resistance (HOMA-IR) defined as follows: fasting plasma glucose (FPG, mmol/L)×fasting blood insulin (μU/mL)/22.5. Serum apoA1 and apoB levels were determined by the immunoturbidimetric method (Randox Laboratories, UK). Serum apoA1 and apoB levels were determined by the immunoturbidimetric method (Randox Laboratories, UK). Decreased apoA1 levels were defined as <15th percentile (0.85 g/L), elevated apoB were defined as ≥85th percentile (1.3 g/L) and elevated apoB/apoA1 ratio was defined as ≥85th percentile.25 Low-density and high-density lipoprotein cholesterol (LDL-C and HDL-C, respectively) were measured by the enzymatic method (Kyowa, Japan). Serum triglyceride (TG) concentration and total cholesterol (TC) were measured by the GPO-PAP method and CHOD-PAP method (Kyowa, Japan), respectively. All lipid measurements were performed on the Hitachi 7600 automated analyzer (Hitachi, Tokyo, Japan). Serum high-sensitivity C-reactive protein (hs-CRP) concentrations were measured using the immunoturbidimetric assay (Denka Seiken, Japan).

Definitions of body size phenotypes

Individuals were classified as metabolically at-risk if they met two or more metabolic abnormalities according to the criteria for the diagnosis of metabolic syndrome recommended by the Adult Treatment Panel (ATP) III of the National Cholesterol Education Programme.26 These metabolic components were defined as follows: (1) TG concentration ≥150 mg/dL; (2) HDL-C levels <40 mg/dL for men and <50 mg/dL for women; (3) systolic/diastolic blood pressure (SBP/DBP)≥130/85 mm Hg or antihypertensive medication use; and (4) FPG≥100 mg/dL or antidiabetes medication use. Individuals with one or none of the abnormal components were classified as metabolically healthy. The following two body size phenotypes were classified based on consideration of both BMI values and the metabolic status: metabolically healthy overweight/obesity (MHOO) (BMI ≥23 kg/m² and <2 metabolic syndrome components) and metabolically unhealthy overweight/obesity (MUOO) (BMI ≥23 kg/m² and ≥2 metabolic syndrome components).

Patients involvement

No patients were involved in the development of research question, the outcome measures, the design or implementation of the study. There are no plans about dissemination of the results.

Statistical analysis

Statistical analyses were performed using SPSS software (V.19; SPSS, Chicago, Illinois, USA). P<0.05 was considered statistically significant. Continuous variables were expressed as the means and SEs, and categorical data were expressed as percentages. Differences in continuous variables between groups were analysed using general linear models, while the differences in categorical variables were analysed using the χ² test or Fisher’s exact test. A Bonferroni post hoc analysis was used for multiple comparisons when necessary. Univariate and multiple logistic regression models were used to determine the independent associations between sleep duration (≤6 hours/day, 7–8 hours/day, ≥9 hours/day) and the variables of apoA1, apoB and the apoB/A1 ratio using the optimal sleep duration of 7–8 hours/day as the reference category. Multiple logistic regression analysis was performed with low apoA1 levels, high apoB levels and high apoB/apoA1 ratios as the dependent variable. The potential confounders entered into the model was based on its significance in univariate analysis (p<0.15) and clinical implication. Models were initially adjusted for age and sex, and were additionally adjusted for potential confounding factors, including smoking habits, alcohol consumption and BMI.

To evaluate potential differential associations of sleep duration by age (<45 vs ≥45 years), sex (male vs female), smoking status (current vs ever/never smoker), alcohol consumption (yes vs no) and medical history of chronic disease (presence vs absence), including diabetes mellitus and hypertension, stratified analyses and interaction tests were performed using logistic regression analysis with elevated apoB as the outcome of interest. ORs with 95% CIs from logistic regression models were used to assess the associations between sleep duration and apoB level and were shown by forest plots. In sensitivity analyses, we examined the relationship between sleep duration and elevated apoB levels using alternative definitions for metabolic health status used in prior literature.27 Metabolically healthy was defined as having one or none of the following metabolic parameters (SBP/DBP≥130/85 mm Hg or antihypertensive medication use; TG levels ≥150 mg/L; FPG levels ≥100 mg/dL or antidiabetes medication use; HDL-C levels <40 mg/dL in men or <50 mg/dL in women; HOMA-IR >90th percentile; and hs-CRP >90th percentile).