Study population
The subjects were adults aged ≥ 18 years who lived in Tumushuke City, 51st Regiment, Xinjiang Production and Construction Corps above 6 months from September 2016 to August 2021, with a median follow-up time of 4.97 years. We started this study in September 2016. This study adopts the stratified random cluster sampling method. In the early stage, the Xinjiang Uygur Autonomous Region was stratified according to the southern Xinjiang/northern Xinjiang, the corps area/the non-corps area. Finally, the southern Xinjiang and corps areas were selected. We selected the third division after the first cluster sampling. After the second cluster sampling, we selected the 51st regiment as our research site. We conducted a census of permanent residents ≥ 18 in the 51st regiment and took hospitals and communities as our study sites for questionnaires, anthropometric measurements, and blood sample collection. The Uyghurs are the main permanent residents in the southern Xinjiang region, the area where this study is carried out is the Uyghur inhabited area. Considering that the living environment of the southern Xinjiang region is similar, the Uyghurs have the same dietary habits, genetic backgrounds, and living habits, and the random sampling method was strictly followed in the field, so it can be regarded as representative of the Uyghur population in southern Xinjiang.
The participants aged 30–74 years were selected. They had no history of cardiovascular disease (CVD) at baseline. They had complete baseline information and participated in at least one follow-up visit throughout the follow-up period. Floating population, population with mental illness or intellectual disability, pregnant women and people with chronic kidney disease were excluded from this study. According to the inclusion and exclusion criteria of this study (Fig. 1), 7705 subjects aged 30–74 years were included in the final analysis.
Questionnaire and follow-up
The epidemiological survey was carried out in the 51st Regiment of the Third Division of the Xinjiang Production and Construction Corps in 2016. The survey included a questionnaire, a collection of blood biochemical indicators, and a collection of physical indicators. And three follow-up surveys were conducted in 2019, 2020, and 2021 respectively. The follow-up survey content was consistent with the baseline survey content. The social security information, hospitalization information, and chronic disease information during the follow-up period were also collected.
Participants were interviewed face-to-face by standard questionnaires, which includes information on sociodemographic characteristics, medical history and lifestyle habits. All participants have lived in the rural areas of Southern Xinjiang for more than 6 months. Current smoking status was self-reported by the participants. Family history of CVD was defined as a parent or sibling with a history of coronary heart disease, myocardial infarction, or stroke.
The physical examination was conducted by professionally trained investigators. Waist circumference, which was measured with an inelastic tape measure, was defined as the midpoint between the lower rib and the superior border of the iliac crest at minimum breathing. The blood pressure of the participants was measured with electronic sphygmomanometers (OMRON HEM-7051, Omron (Dalian Co., Ltd.)). Each individual was measured twice with an interval of 30 s. The average was taken as the final blood pressure result.
A 5 ml fasting blood sample was collected from each subject. Blood glucose, high-density lipoprotein cholesterol, and total lipoprotein cholesterol were determined by the modified hexokinase enzymatic method using the Japanese Olympus AV2700 biochemical automatic analyzer in the Biochemical Laboratory of the First Affiliated Hospital of Shihezi University School of Medicine.
CVD events in the study cohort were determined from patients' hospital medical records, questionnaires, and social security records. Questionnaires were used to follow up with the subjects, and the disease information of the subjects was collected and checked with the hospital social security data and medical record information. If the subjects died during the follow-up period, the family members will be asked about the time of death, the place of death, and the cause of death, and then this information will be checked against the information records provided by the hospital.
All participants signed informed consent. This study was approved by the Ethics Committee of the First Affiliated Hospital of Shihezi University School of Medicine (No. SHZ2010LL01).
CVD outcome definitions
In this study, CVD was defined as an individual's first diagnosis of non-fatal acute myocardial infarction, death from coronary heart disease, and fatal or non-fatal stroke.
Acute myocardial infarction was defined as an increase in biochemical markers of myocardial necrosis with ischemic symptoms, pathological Q waves, ST-segment elevation or depression, or coronary intervention. Coronary heart disease deaths include all fatal events due to myocardial infarction or other coronary death. Stroke was defined as an ischemic or hemorrhagic attack. If more than one CVD event occurred during follow-up, only the first CVD event was included as an outcome event.
Framingham risk score equations
We use the Framingham CVD prediction model developed in 2008 for people aged 30–74 [5]. The model equations are as follows: Male = 1-(0.9431^exp (age *3.06117 + TC *1.12370–0.93263*HDL-C + 1.93303*SBP + 0.65451*smoking status + 0.57367*Diabetes-23.9802); Female = 1-(0.9747^exp (age *2.32888 + 1.20904*TC -0.70833*HDL-C + 2.76157*SBP + 0.52873*smoking status + 0.69154*Diabetes-26.1931). We performed a simple calibration of the Framingham model using means of risk factors in this population and risk of morbidity [10].
Statistical analysis
Continuous variables that satisfy the normal distribution are described by the mean ± standard deviation. Continuous variables that do not satisfy the normal distribution are described by the median and interquartile range. Categorical variables are described by the sample size and percentage. Since the follow-up period of this study was five years, only the five-year CVD actual risk and the five-year predicted risk were calculated in this study.
Calculate eCRF using gender-specific equations. Female(METs) = 14.7873 + (age × 0.1159) – (age2 × 0.0017) – (BMI × 0.1534) – (waist circumference × 0.0088) – (resting heart rate × 0.364) + (physical activity [active vs inactive] × 0.5987) – (smoking [yes vs no] × 0.2994); eCRF in male(METs) = 21.2870 + (age × 0.1654) – (age2 × 0.0023) – (BMI × 0.2318) – (waist circumference × 0.0337) – (resting heart rate × 0.0390) + (physical activity [active vs inactive] × 0.6351) – (smoking [yes vs no] × 0.4263) [11]. BMI indicates body mass index (calculated as weight in kilograms divided by height in meters squared).
WC and eCRF were grouped by tertiles, with the lowest group serving as the reference group. The log-rank test was used to compare the risk of CVD morbidity among eCRF groups and WC groups. We performed pairwise comparisons among the three groups and used a Bonferroni-corrected P-value (P = 0.017) to ensure the accuracy of the log-rank between-group test results (https://www.graphpad.com/support/faq/after-doing-logrank-analysis-on-three-or-more-survival-curves-can-i-perform-multiple-tests-for-differences-between-pairs-of-curves/).
A univariate COX proportional hazards regression was used to analyze the association between WC, eCRF, and CVD risk. Age, educational status, career, marital status, exercise status, smoking, alcohol consumption, TC, and HDL-C were adjusted as confounders during multivariate COX proportional hazards regression analysis. Exploring the association between eCRF, WC, and CVD risk in this population using a restricted cubic spline with 4 knots, with knots equally distributed. We take the point in the restrictive cubic spline where the direction of HR change changes as a rough value for the change in the variable.
This study calibrated the Framingham original model by mean levels of risk factors and the five-year risk of CVD in this population. The WC and eCRF were added to the Framingham model for model adjustment. After introducing new risk factors, use bootstrap 1000 times to internally validate the model. The discrimination of the model is evaluated using AUC(Area Under Curve), NRI(Net Reclassification Index), and IDI(Integrated Discrimination Improvement) metrics. We calculated categorical NRI and continuous NRI separately, using 10% and 20% as risk cut-off points for categorical NRI. We use the Delong test to analyze whether there is a difference in AUC between the model after adding the new variable and the original model. Model calibration is evaluated using pseudo R2 values. High pseudo R2 values indicate better discrimination. We choose the model with the best predictive performance and build an online risk calculator based on the coefficients for each risk factor.
SPSS and R software were used for data analysis. JAVA Script was used to build an online risk calculator.