Cohort creation

Liyang, referred to as Liyang city (county-level city), is actually a county-level administrative unit in China’s administrative division, with the urban and rural areas representing 20.28% and 79.72% of the population, respectively, based on the census results of Jiangsu Province in 2020 (online supplemental table 1). In 1986, The State Council of China decreed that ‘counties’ that met all of the following criteria could be called ‘cities’. First, the total population reaches more than 500 000. In addition, the county government is located in towns with a non-agricultural population of more than 120 000. Second, the annual gross national product reaches more than ¥400 million. On 8 December 1990, Liyang City met all the above criteria and was officially called Liyang ‘City’. In China, such county-level administrative units are often called ‘county-level cities’. As an economically developed county-level city located at the junction of Jiangsu, Zhejiang and Anhui, and in the Yangtze River Delta, Liyang City was ranked among the top 10 counties and districts in Jiangsu Province, and its comprehensive strength, including economy, environment, and science and technology, ranked 25th among the top 100 counties in China in 2020. In the same year, the Gross Domestic Product (GDP) of Liyang had reached $17.73 billion, which is a leading level in China. Also, residents’ per capita disposable income had reached $7090, far higher than the national average ($5056). Despite this, the resource imbalance in Liyang still exists, mainly from a medical, education and income standpoint (online supplemental table 2). The per capita disposable income of urban residents ($8784) is 1.84 times that of rural residents ($4763). Besides, the GDP gap among the towns is also large, with the highest GDP 7.28 times the lowest.

To address the urgent need for understanding the prevalence and risk factors of NCDs in resource-imbalanced regions, the baseline survey of the Liyang Study was launched from March 2019 to June 2020 in Liyang city, Jiangsu Province. Samples were randomly selected from the general population using multistage stratified cluster sampling. Considering the consistency of residents, population stability and local medical conditions, all the towns and streets in Liyang city were selected as recruitment points, and the health centres (17 in total) responsible for daily healthcare in each town or street were selected as survey points (figure 1). Note: administrative divisions in China include provinces, cities, counties(districts) and towns(streets). Towns are the same level as streets and they are the most basic administrative units in China. Towns belong to rural areas, and residents living in them are mostly residents with rural household registrations, while streets belong to urban areas, and residents living in them have both urban and rural household registrations. The detailed population information of each town or street in Liyang city is shown in online supplemental table 1. We preliminarily planned to conduct a large-scale survey covering 10 000 people among the general population of Liyang city. First, to ensure a sufficient sample size and facilitate implementation and calculation, a sample of approximately 600 cases was planned for the survey at each survey site (health centre). However, due to the loss of address, refusing investigation of some urban residents, and some residents with rural household registrations living in the urban district, the sample size collected in several survey sites did not reach the planned 600 cases, so additional investigation on 261 alternative subjects was carried out at other survey sites to fill the gap. Specifically, because the sample size collected by several survey sites did not reach the requirements of the original plan, however, to ensure that the total number of people reached the expected target (a large cohort of more than 10 000 people), we randomly selected alternative participants (261 cases in total) from villages or communities belonging to the remaining survey sites (a total of 11 health centres recruited with a sample size of more than 600 shown in online supplemental table 1) whose participants are actively cooperating with this survey in the quality control stage. Next, two administrative villages (belonging to the town) or communities (belonging to the street) in the vicinity of each health centre were randomly selected, and approximately 300 cases were planned to be randomly selected from each village or community as the respondents. Thus, 10 200 respondents were recruited for the survey before the questionnaire was issued, with an average sample size of 600 per survey site (ranging from 443 to 633) (online supplemental table 1). A total of 144 individuals were excluded from the study because they refused to sign informed consent (n=82), failed a medical examination (n=26), were physically unable to complete the study (n=4), were not permanent residents (n=23) or were younger than 18 years (n=9). Overall, 10 056 permanent residents (living for more than 6 months before the survey) from different households, aged ≥18 years, who had signed written informed consent, completed the baseline survey with a response rate of 98.6%. The sampling process and specific inclusion and exclusion criteria are illustrated in figure 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Locations of the 12 towns (streets) with 17 health centres in the Liyang Study. Note: The Licheng town has three health centers, Tianmuhu, Shangxing and Shezhu town have two heath centers,respectively, and the other 8 towns have only one health centers, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Flow diagram of baseline sampling in the Liyang Study. *Reasons for refusing to participate in the baseline survey included refusing to sign informed consent, failing a medical examination, and being physically unable to participate.

Data collection

The Liyang Study collected a large amount of information through questionnaire surveys, including potential confounding variables, mainly sociodemographic characteristics (such as age, sex, nationality, residence status, marital status, education level, living status, and family annual income and expenditure). The primary exposure variables were as follows: health-related behaviours (smoking, drinking, physical activity level, drinking tea and occupational exposure history), activities of daily living, instrumental activities of daily living to assess physical function status, abbreviated mental test to evaluate cognitive function, and Food Frequency Questionnaire (FFQ) to measure dietary habits. Moreover, the possible outcome variables included physician-diagnosed NCDs (ie, the symptoms persisted or related medical treatment continued over the past year), awareness of NCD-related knowledge, self-reported medical history and HRQoL as measured on the EuroQol 5-Dimensions 5-Levels validated by the Chinese population.13 Detailed information on the questionnaire is presented in table 1.

Table 2 presents the details of the physical examinations and laboratory tests. The physical examination consisted of self-reported height, weight and waist circumference, and double measurements of height, weight, waist circumference and blood pressure using an international anthropometric instrument. While measuring height, the subjects removed their shoes and leaned against a calibrated wall upright. Waist circumference was measured at a level of 1.0 cm above the navel. An electronic weight scale (G&G TC-200K, USA) was used to measure subjects’ weight according to the instructions. After an overnight fast, intravenous catheters were inserted into the antecubital vein for blood sampling by medical professionals. Blood samples were drawn into a 5 mL serum separator tube (non-anticoagulant blood, yellow cap) and a 5 mL EDTA-K2 anticoagulation tube (purple cap). Blood samples were kept at room temperature for 1 hour before centrifugation at 3000 rpm for 15 min (TM48). The serum supernatant was used to measure blood glucose, total cholesterol, triglycerides, high-density lipoprotein and low-density lipoprotein (MINDRAY BS-820). After centrifugation, the anticoagulated whole blood was separated into plasma, white blood cells and red blood cells. The plasma and white blood cells were stored at −80°C and forwarded to measure gene expression. In addition, the white blood cells were centrifuged at 2000 rpm for 2 min before storage. All separated blood samples were sent to the Center for Disease Control and Prevention (CDC) within 24 hours and stored at −80°C (Haier DW-86L338J). The details of the flow chart for blood separation and detection of biochemical indices are shown in figure 3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Flow chart of blood separation and detection of biochemical indexes.

Data collection methods

Trained investigators conducted the baseline face-to-face interview. The baseline data in the Liyang Study were collected in two stages. In the first stage, we issued informed consent forms for the investigation in advance and set up centralised survey sites in local health centres. Subsequently, we required the respondents to carry their unique identification (ID) cards to the corresponding survey sites for inperson questionnaire surveys, which were conducted by trained and qualified interviewers. Professional physicians were responsible for physical measurements and blood sample collection. In the second stage, if residents could not participate in the centralised investigation, we arranged for interviewers to visit the participants’ households for supplementary investigations, after obtaining informed consent from the respondents. Additionally, the results of physical measurements and laboratory tests were reported to the respondents within 2 weeks of the survey.

Quality control

At the beginning of the investigation, strict quality control procedures ensured that the quality of the study was in accordance with the unified standard. Prior to the baseline interview, a quality control committee was formed, comprising staff from Soochow University and Liyang CDC, and a quality control manual was developed by the committee. According to the manual, specific quality control measures were divided into two sections: primary and secondary. The primary quality control implemented by Soochow University involved checking the completeness and accuracy of the questionnaire and establishing the quality control level of the laboratory. Secondary quality control implemented by the Liyang CDC was to recheck all questionnaires in each town (street). For any errors detected, such as missing items, logical errors or incomplete filling, the respondents will be surveyed again: the investigator will communicate with the respondents by phone or door to door to ask them to fill the problematic questionnaire again. And then we will revise the wrong items according to the results from the repeat survey. In addition, a pilot study was conducted in three health centres (Daibu, Tianmuhu and Pingqiao) in December 2017 to assess aspects such as whether the questionnaire was easy to understand and if each question was clear. After the pilot survey, we modified the language and items of the questionnaire according to feedback from the investigators.

Cohort follow-up

A follow-up survey will be conducted every 3 years, beginning in 2022. Like the baseline survey, all surviving study participants will be invited for repeated interviews, including questionnaire interviews and physical examinations. Although the Liyang Study collected blood samples from all participants at baseline, further blood samples, such as the genetics of patients with stroke, will be collected during follow-up only for specific substudies. Follow-up will be conducted both actively and passively. For active follow-up, a structured questionnaire using a computer-assisted personal interview system will be used to collect data. In addition, we will communicate with respondents not living (probably a small number) anymore at their original residence during the follow-up period, by phone in a timely manner. Although differences in data collection methods between baseline and follow-up may be a source of bias, a series of quality control measures implemented in the baseline stage effectively ensured the quality of data collected from paper questionnaires. First, investigators received uniform interviewing training before collecting data. Second, after data collection, we implemented strict quality control measures, including re-investigating missing, incomplete and logical errors items. Finally, at the data entry stage, we made a double entry and re-investigated the inconsistencies by phone. In addition to the steps involved in active follow-up, passive follow-up includes connecting to the primary Health Information System of local hospitals to collect basic healthcare information. Physician-diagnosed NCDs, medication history and family history of hypertension, diabetes, stroke, CHD, dyslipidaemia, COPD, asthma and cancer will be further identified by linking to the Chronic Disease Network Direct Report Management System of Changzhou city. All deaths during the follow-up period will be checked according to cause-of-death classification through linkage with data from the Local Cause of Death Registration and Reporting System. Combined with passive follow-up, these existing systems will allow us to accurately gather basic sociodemographic information and track the incidence and prognosis of disease and death during follow-up. In addition, the under-representation of urban residents in this study may lead to selection bias. Hence, two measures should be taken into account to address this problem when analysing follow-up data. First, statistical methods such as standardised weighting or inverse probability weighting should be used to adjust the combined sample data according to the sampling ratio to lower selection bias.14 15 Second, the rural and urban populations will be analysed separately so that the research findings can better reflect the different health status of the general population in Liyang city.

Findings to date

The baseline characteristics of the study participants were summarised as numbers with percentages and means±SD for categorical and continuous variables, respectively. Pearson’s χ² analysis for categorical variables and Wilcoxon’s rank-sum test for continuous variables were used to determine differences in baseline characteristics and the prevalence of NCDs according to sex. All data analyses were performed using SAS V.9.4 (SAS Institute, Cary, North Carolina, USA). A value of p<0.05 was considered statistically significant using two-sided tests.

Of the 10 056 individuals who participated in the Liyang Study, 47.08% (n=4734) were men, with a mean age of 49.9 years, and the vast majority (92.26%, n=9278) lived in rural areas. Approximately 31.51% (n=3172) had obtained high school or higher education. Most participants were married and employed (84.19%, n=8464 and 77.19%, n=7762, respectively). The proportion of participants with less than ¥50 000 per capita family annual income was 26.73% (n=2688). The prevalence of smoking and drinking was significantly higher in men than in women (50.91% for men, n=2410 vs 0.79%, n=42, for women and 41.38%, n=1959 for men vs 2.61%, n=139 for women, respectively). Overall, 52.83% (n=5313) of all participants had a normal BMI; the proportion of overweight and obesity was significantly higher in men than in women. Mean height, weight, waist circumference, systolic blood pressure, diastolic blood pressure, fasting blood glucose and triglycerides were significantly higher in men than in women (all p<0.05) (table 3).

The prevalence of the most frequent NCDs, based on self-reported and doctor-diagnosed data, in the Liyang Study is shown in table 4. The top three overall most prevalent chronic diseases in the study were hypertension (18.06%, n=1815), digestive diseases (7.88%, n=791), and arthritis or rheumatism (5.28%, n=530). Women had a significantly higher prevalence of diabetes and arthritis than men, while the opposite was true for chronic lung diseases such as COPD and heart diseases (all p<0.05).

Patient and public involvement

No participants were involved in the design and implementation of the study. In addition, all participants or their relatives were informed of the use of data in this study and the results of all examinations and tests were simply reported to each participant.

Strengths and limitations

The Liyang Study has several strengths. First, the most significant advantage of this study is that it provides referential data, based on a large sample size (n=10 056) and a wide range of variables, providing significant evidence for identifying and measuring the prevalence and risk factors of common NCDs. Another key advantage is that the field investigators and research team have significant fieldwork experience; moreover, professional medical institutions participated in the project, aided by the Liyang CDC and Soochow University, to provide project review and verification, as well as technical guidance. This multicollaborative team ensured the quality of our research. Second, the collection of a large number of biological samples at baseline, including whole blood, plasma, serum and white blood cells, will help effectively capture and determine relevant information about the genetics and variation of NCDs. The finding will have implications for disease prediction, prevention and awareness in China and in other countries. Finally, disease diagnosis and follow-up will be performed by linking various health record systems to realise real-time confirmation of patients' health status and medical history tracking.

It is also essential to highlight the limitations of the study. First, data collection on NCDs at baseline, especially mental and memory-related diseases, was self-reported, which may result in recall bias. However, effective researcher training, good field implementation, scientific and technical guidance, and good linkages with local health record systems will ensure the accuracy and reliability of the information. Moreover, studies have proven that self-reported information on NCDs possesses relatively good reliability.16 Second, the study population was exclusively from Liyang city, Jiangsu Province, which may affect the generalisability of our results. However, the results of this large cohort study can, to some extent, indicate the main risk factors and prevalence of common NCDs in most developed areas of China. Third, due to the loss of address, refusal of investigation by some urban residents, and some residents with rural household registrations living in urban districts, the proportion of urban residents who completed the baseline survey was lower than that of urban residents in Liyang city, which may lead to bias in estimating the association of NCDs among the general population in Liyang city. Nevertheless, when we compare the age distribution and sex ratio of different age groups in the Liyang Study with those of the Liyang census (2020), we found that our cohort is quite similar to the entire population of Liyang city, supporting the representativeness of our study (online supplemental table 3). Furthermore, we will account for rural and urban imbalances in analyses examining differences in NCD outcomes. Finally, in a cohort with an average age of 50 years, early life data for most participants will not be available. There is growing evidence that early life exposure may have long-term effects on risk of chronic diseases in adulthood.4 17 However, the high prevalence and incidence of common NCDs among middle-aged and elderly people will enable our study to provide sufficient information for future analysis.

Collaboration

All data collected are stored at the Soochow University, Soochow Medical College, School of Public Health. The Liyang Study encourages collaboration to maximise the use of data and sample information. Although the research database is currently not freely available in the public domain, as it contains sensitive information, any ideas and suggestions on potential collaborations are welcome. Researchers interested in collaborating can contact the authors via shenyueping@suda.edu.cn.