Setting
The China Health and Nutrition Survey (CHNS) started in 1989. It was intended to represent a range of economic and demographic variation in China. Data of participants of CHNS came from 12 provinces (from north to south, including Heilongjiang, Liaoning, Shandong, Henan, Hubei, Hunan, Jiangsu, Guangxi, Guizhou, Ningxia, Shaanxi and Yunnan) and 3 municipal cities (Beijing, Shanghai and Chongqing). Blood biomarkers were tested for the first time in 2009. Details of the study design and sampling strategies are available at the World Wide Web site (https://www.cpc.unc.edu/projects/china) and elsewhere(16, 17).
All the documentation and procedures comply with Good Clinical Practice (GCP), Human Ethics Protocol Rules and related Chinese laws. The CHNS project was approved by the office of human research ethics of the University of North Carolina at Chapel Hill and the Human & Clinical Research Ethics Committee of China-Japan Friendship Hospital (Study ID: 07-1963). Approved consent forms and other documents are available online at the World Wide Web site (http://apps.research.unc.edu/irb/index.cfm?event=home.dashboard.irbStudyManagement&irb_id=07-1963).
Data collection methods
The data of demographic, anthropometric, lifestyle, memory status and perceived stress index were collected by trained interviewers used a questionnaire. Height and weight were measured based on a standard protocol. Height was measured to the nearest 0.1 cm, and weight in lightweight clothing was measured to the nearest 0.1 kg. BMI was calculated as weight in kg divided by height in square metres.
Trained nurses drew fasting blood from participants’ antecubital vein in the morning. Blood samples were treatment (centrifuged at 3000g for 10 mins at room temperature and separated into 9 aliquots) within 2 hours of collection in local hospitals. Aliquots were storage in -80 degree freezers.
Serum TG (Lot number: 192AIF), total cholesterol (TC) (Lot number: 203AIG), high-density lipoprotein cholesterol (HDL-C) (Lot number: 548AIE) and low-density lipoprotein cholesterol (LDL-C) (Lot number: 362AIG) were detected using the enzymatic colorimetric method (Kyowa Medex Co., Ltd, Takatsuki-shi, Osaka, Japan). The calibrators and control serums were provided by the department of laboratory medicine of China-Japan Friendship Hospital and had the same lot number.
Definition of body size phenotypes and BMI levels
Participants were classified as follows (18):
Underweight: BMI <18.5 kg/m2;
Normal weight: BMI 18.5-23.9 kg/m2;
Overweight: BMI 24.0-27.9 kg/ m2;
Obese: BMI of 28-31.9 kg/m2;
Severely obese: BMI ≥32.0 kg/m2.
Definition of age levels
Participants were classified as follows (19):
Mid-life adults: <65 years;
Older adults: ≥65 years.
Assessment of memory status and cognitive function
The global cognitive score was calculated using composite scores of memory, counting back and subtraction scores. The cognitive screening items of questionnaire used in CHNS included a subset of items from the telephone interview for cognitive status-modified (20, 21). The questionnaire included two questions for assessing self-reported memory status and four tests for testing memory performance. The first question asked about memory status: “How is your memory?” The response categories were “very good”, “good”, “OK”, “bad” and ”very bad”. Those who reported “bad” or “very bad” were defined as having a poor memory. The second question asked about changes in memory status: “In the past twelve months, how has your memory changed?” The response categories were “improved”, “stayed the same” and “deteriorated”. Those who reported “deteriorated” were defined as self-reported memory decline.
The following four tests were related to cognitive function on specific memory tasks. Four tests were administered in the following order: 1) the first was a word list memory test for immediate memory, in which an examiner read a list of 10 unrelated words at 2-second intervals and immediately asked the participant to repeat to them as many words as possible in any order (score 10); 2) the following two tests were mind control ability tests in which an examiner counted backward from 20 to 1 (score 2) and calculated 100 minus 7 and subtracted 7 again and again (score 5); and 3) the last test was a test for delayed memory, in which a list of words were repeated to an examiner after a period of time (score 10). An orientation test was not included in the analysis as it was only assessed in 2015 wave. The cognitive function score was used to assess memory performance, which was the sum of the scores of the four tests and could range from 0-27 points. The Cronbach alpha internal consistency coefficient of this scale was 0.73, which is above the acceptable cut-off value of 0.70.
Study population
In total, 15143 CHNS participants were included, and 5256 participants had complete memory status data. After excluding the participants with use lipid lowering agent, 4574 participants remained. After further excluding the participants with missing or incomplete gender, age and education level data, 4538 participants remained. Subsequently, 18 participants with missing BMI data and 2274 participants with missing biomarker data were excluded, 2246 participants (1120 men and 1126 women) remained. (Fig. 1)
Statistical methods
The current study was restricted to 2246 participants to examine the influence of BMI on the association between serum lipids and cognitive function among Chinese population. For the baseline characteristics of participants, all data are shown as the means±standard deviations (SDs) for normal variables and as medians (interquartile ranges) for skewed variables. Differences in characteristics between gender and age subgroups were tested for significance. The unpaired t-test or Mann-Whitney U test was used to compare the differences between continuous variables, and the chi-square test was used for categorical variables.
Multivariable linear regression analyses examined serum lipids level as predictors of gender- and age-specific measure of cognitive function in different BMI levels, which were adjusted for confounding factors, including gender, age, nationality, BMI, systolic blood pressure (SBP), diastolic blood pressure (DBP), smoking status, alcohol consumption and education level. The statistical analysis was performed with program R 3.4.3.