Selection of study site and participant eligibility

The Hainan Cohort study was designed as a longitudinal study that recruited community dwellers in five main areas of Hainan (figure 1): Haikou (the capital city located on the northern coast of Hainan), Sanya (central city in the south of Hainan), Qionghai (a city located in the east of Hainan), Dongfang (a city located on the western coast of Hainan) and Qiongzhong (Li and Miao Autonomous County in central Hainan). These areas cover both urban and rural locations, as well as different ethnic groups who speak different dialects, and are representative of the diverse sociodemographic characteristics of Hainan. We determined the sample size of each area based on the population distribution of Hainan and other objective factors, such as biological sample preservation, transportation and local coordination. Then, a multistage cluster random-sampling method was used to obtain samples from the general population. In the first stage, 18 districts were selected from five major areas through simple cluster sampling. Second, communities or units in each district were randomly selected considering population stability, local medical conditions and adherence to future follow-up assessments. The inclusion criteria were as follows: (1) age between 35 years and 74 years old; (2) residence in Hainan for at least 5 years; (3) no severe physical disability, mental illness or severe medical condition; and (4) ability to complete questionnaire interviews, physical examinations and follow-up assessments. All the participants enrolled in this study voluntarily and signed an informed consent form on enrolment. A unique identification was created for each participant using their national identification card.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Locations of the five survey sites in the Hainan Cohort.

Patient and public involvement

Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Study plan

The Hainan Cohort is a dynamic cohort with no end date, and we will collect follow-up data for each participant periodically until death. Follow-up will be conducted using two methods: monitoring and long-term follow-up surveys. For basic monitoring of the participants, we are cooperating with the Social Insurance Service Centre and Healthcare Security Administration of Hainan. Using the independent medical insurance number of each urban and rural participant who signed the informed consent form, information about medical treatment, we will collect hospitalisation and incidence of diseases from the department of medical insurance data management in January and July of each year. The Mortality Surveillance System of the Hainan Centre for Disease Control and Prevention allows us to track cause-specific mortality in all the participants in the cohort. Causes of death will be classified according to the 10th version of the International Statistical Classification of Diseases and Related Health Problems. The National Basic Public Health Service Platform (http://www.nbphsp.org.cn/) has achieved full coverage in Hainan, enabling us to ascertain the onset time of common chronic diseases, such as hypertension and diabetes, in the cohort.

Long-term follow-up will be conducted every 5 years. All surviving participants will be visited and will complete an in-person assessment, which is the same as that conducted at baseline, that includes a questionnaire interview, physical examination, and blood and urine sample collection. The onset of endpoint events and the loss of cohort participants will be investigated and compared with the monitoring data to reduce possible omissions and mis-statements caused by conventional monitoring. Endpoint events for major chronic diseases will be confirmed by experienced physicians based on well-accepted international standards. The household registration system of the public security department has the relocation records of all permanent residents, which can be combined with the national network medical insurance system to follow up any study participants who can still be contacted after relocation and finally control the rate of lost to less than an estimated 8%. Figure 2 shows the overall study plan.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Study plan of the Hainan Cohort study. NCD, non-communicable disease.

Baseline assessments

In December 2017, a pilot study was carried out before the formal investigation. The aim of the pilot study was to evaluate whether each item on the questionnaire was easy to understand by persons in different age groups and from different cultural backgrounds and whether the options were logical. Additionally, the pilot study allowed us to find the most suitable method for asking difficult questions after discussion to facilitate the training of investigators in the future. The baseline assessments were performed from June 2018 to October 2020 and included a questionnaire survey, physical examination, blood and urine sample collection, laboratory measurements and outdoor environmental data collection.

After overnight fasting, all participants were required to bring their ID cards or household registration books to the local health examination centre at the specified time. Following the obtainment of written informed consent, a validated face-to-face questionnaire interview was conducted by well-trained interviewers from the School of Public Health, Hainan Medical University. The questionnaire collected general demographic (name, sex, age, ethnicity, education level, marital status, occupation and annual household income), medical history (diabetes mellitus, coronary heart disease, stroke, hypertension, hyperuricaemia, cancer, etc), family disease history (coronary heart disease, hypertension, stroke, diabetes mellitus, cancer and chronic obstructive pulmonary disease), lifestyle habit (smoking, alcohol consumption, tea consumption, dietary pattern, sleeping status, physical activity, passive smoking and indoor pollution), female reproductive history and mental status data. Table 1 shows detailed information of the baseline measures.

Physical examinations were conducted by trained doctors or nurses from the First Affiliated Hospital of Hainan Medical University; collected measurements included height, waist circumference, hip circumference, blood pressure, heart rate, lung function, body composition (weight, body mass index (BMI), body fat percentage, skeletal muscle mass, bone mass, daily energy requirement, etc), as well as 12-lead electrocardiography. All measurements were collected by trained investigators in a large, bright room with a suitable temperature using standard procedures. Height was measured without shoes and hats using an ultrasonic height measuring instrument (SK-X80, China). Waist and hip circumferences were measured while the participants were wearing thin clothing. The investigator stood on the right side of the participant and measured them with a tape measure (Hoechstmass, China). The above three measurements were accurate to the nearest 0.1 cm. Blood pressure and heart rate were measured twice consecutively using a validated electronic sphygmomanometer (HEM-7071; OMRON, Japan) on the right arm after the participants had rested for at least 5 min in a sitting position. Forced expiratory volume in 1 s and forced vital capacity were measured using a spirometer (SP10; Contec, China); the participants took a deep breath and blew at the fastest speed and with maximum strength until all the gas in their lungs was exhaled. Body composition was measured using a body fat detector (BC-6017611; Tanita, Japan). Participants removed their coats, shoes and socks and checked that the soles of their feet were clean; when necessary, they used a wet towel to wipe their feet, and they removed all metal objects (such as mobile phones and keys) to avoid affecting the result. Weight was measured to the nearest 0.1 kg. Electrocardiography was performed (ECG-1350C; NEC Lighting, China), and the experienced physician reviewed each ECG. BMI was calculated as weight (kg) divided by height (m) squared. Table 2 shows the details of the equipment for the physical examination and clinical measurements.

Before blood collection, all participants fasted for at least 10 hours as required. To prevent the occurrence of hypoglycaemic events, all blood samples were collected between 07:00 and 10:00. Three tubes of peripheral venous blood from each participant were extracted for a total of 15 mL; two tubes were EDTA anticoagulation tubes for the separation of plasma and blood cells, and one tube was a non-anticoagulation tube for the separation of serum and clotted blood. The collected blood samples were sent to the laboratory for testing within 2 hours. The serum samples were used to detect biochemical indicators with automatic biochemical analyser (7600–110; Hitachi, Japan), including alanine aminotransferase, aspartate aminotransferase, albumin, alkaline phosphatase, gamma-glutamyl transpeptidase, triglycerides (TGs), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), urea, creatinine, uric acid (UA), fasting plasma glucose (FPG) and glycated haemoglobin. Routine blood parameters, including white blood cell count, red blood cell count, red cell distribution width, platelet count, etc, were measured using an automatic blood cell analyser (BC-6900; Mindray, China). The remaining blood samples were stored in a 4°C refrigerator and centrifuged within 1 day after collection. Plasma, serum, and white and red cells were collected after 15 min of centrifugation at 3000 rpm. Aliquots of 0.2 mL of centrifuged blood samples were retained in each tube and stored in a −80°C refrigerator to prevent DNA breakage or degradation. At least 50 mL of midstream morning urine on an empty stomach was collected using a urine cup and immediately placed into a refrigerator at −4°C after collection. After testing, the remaining urine samples were divided into two 15 mL centrifuge tubes using disposable plastic straws (10–15 mL each) and sent back to the laboratory for preservation within 24 hours. The routine urine test indicators included urine glucose, urine protein, urobilirubin, etc. All these test items were measured by the general standardised methods of the Clinical Laboratory of the Affiliated Hospital of Hainan Medical University.

Outdoor environmental data, including daily maximum temperature, daily minimum temperature and average temperature, were collected from the National Meteorological Bureau and local meteorological departments. Air quality monitoring data at the survey site, including the daily air quality index, air quality levels and primary pollutant (SO₂, NO₂, CO, O₃, PM₁₀ and PM_2.5) concentrations, were obtained through the air quality platform of the National and Hainan Environmental Protection Departments.

Hypertension was defined as a sitting systolic blood pressure (average of 2 readings with a difference of <5 mm Hg) of 140 mm Hg or diastolic blood pressure of 90 mm

Hg and/or the use of antihypertensive drugs.25 Diabetes mellitus was defined as a fasting blood glucose value of ≥7.0 mmol/L, prior diagnosed diabetes mellitus or the use of antidiabetic agents.26 Metabolic syndrome (MS) was defined according to the criteria of the International Diabetes Federation (IDF) for Chinese adults. MS was defined according to the IDF criteria and the Chinese specific abdominal obesity standard in adults (waist circumference in Chinese men ≥90 cm, waist circumference in Chinese women ≥80 cm), along with the presence of two or more of the following: (1) TG level of >150 mg/dL (1.7 mmol/L) or drug treatment for elevated TGs, (2) HDL level of <40 mg/dL (0.9 mmol/L) in men, < 50 mg/dL (1.1 mmol/L) in women or drug treatment for low HDL; (3) systolic blood pressure of ≥130 mm Hg or diastolic blood pressure of ≥85 mm Hg or drug treatment for hypertension, and (4) FPG of ≥100 mg/dL (5.6 mmol/L) during fasting status or diagnosed diabetes.27

In addition, a resurvey was conducted in a random subsample of the Hainan Cohort during the time of the baseline survey. Approximately 5% of participants were randomly chosen and completed the resurvey. The procedures and content of the resurvey were consistent with those of the baseline survey.

Strengths and limitations

This cohort study has several clear strengths. First, as far as we know, the Hainan Cohort study is the first independent biospecimen-based prospective study on NCDs in tropical China to analyse detailed information including sociodemographic characteristics, lifestyle habits, personal and family histories of diseases, dietary pattern, sleeping status, physical activity levels, female reproductive history, and anthropometric and clinical measurements. Second, the sample size was relatively large, which gives us sufficient power to explore the relationship between risk factors and common chronic NCDs, such as diabetes mellitus, chronic obstructive pulmonary disease, myocardial infarction and stroke (see sample size estimation in online supplemental materials 1). Third, the sample biospecimens collected at baseline will allow us to conduct nested case–control studies on genetic variations and metabolomic biomarkers for NCDs. Fourth, due to rigorous data quality control, the age distribution and sex ratio of the baseline data can generally be representative of those in the Chinese population (based on the seventh Chinese National Census). Finally, we linked the cohort database with the medical insurance system, the mortality surveillance system and the national basic public health service platform and will combine these data with those obtained from long-term follow-up surveys to ensure that we obtain accurate endpoint information.

This study also has limitations. First, we did not collect the basic data of eligible people who did not participate in the survey according to the sampling plan, which may lead to selection bias. Second, for some NCDs with a low incidence rate (such as lung cancer), the sample size of the study is insufficient. We will consider expanding the number of participants in the next wave of follow-up. Third, although we carefully checked the collected questionnaires and the resurvey showed good concordance with the baseline survey, we could not rule out possible recall and measurement bias during data collection. Finally, more than 98% of the participants in the cohort were Han Chinese; thus, caution should be taken when generalising the results of this work to other ethnicities.