Multimorbidity is becoming a priority for global health research and highlights the urgent need for action(Reference Skou, Mair and Fortin1). Cardiometabolic multimorbidity (CMM) is a common pattern of multimorbidity, defined as the coexistence of two or more cardiometabolic diseases (CMD), including myocardial infarction (MI), stroke and type 2 diabetes (T2D)(Reference Di Angelantonio, Kaptoge and Wormser2). CMM is increasingly common(Reference Cheng, Ma and Ouyang3) and has more than doubled over 5 years in China(Reference Zhang, Tang and Shen4). CMM accelerates both cognitive decline and dementia progression(Reference Dove and Xu5,Reference Tai, Veldsman and Lyall6) and was associated with severe disabilities(Reference Otieno, Asiki and Aheto7,Reference He, Tu and Chen8) and multiplicative mortality risk(Reference Di Angelantonio, Kaptoge and Wormser2). In China, CMM, the most popular pattern, also posed the highest threat to mortality and warranted specific attention(Reference Fan, Sun and Yu9).
A high-quality diet can reduce the risk of CVD; many professional heart societies also promote healthy dietary patterns, such as the Mediterranean diet and Dietary Approaches to Stop Hypertension (DASH) diet(Reference Pagidipati, Taub and Ostfeld10). Following a healthy diet was estimated to prevent 42·4 % of CVD annually(Reference Bundy, Zhu and Ning11). However, most present studies on dietary patterns focused on a single CMD; few studies have been conducted on dietary patterns and CMM risk. Only one prospective study can be mentioned; this study was conducted among 2873 British men aged 60–79 and did not find a significant association of baseline Elderly Dietary Index with risk of CMM. The result may be related to the low statistical power due to the small sample size(Reference Wang, Schmidt and Lennon12). Further work with larger cohorts and more general populations is needed to better understand the relationship between dietary patterns and the risk of CMM.
The Chinese healthy diet is the only recommended healthy diet by the Dietary Guide for Chinese Residents (2022). It is represented by the dietary pattern of Jiang Nan and southeast coastal areas (Zhejiang, Shanghai, Jiangsu, Fujian, Guangdong) of China, with characteristics as follows: diverse foods; rich vegetables and fruits; frequent consumption of fish, shrimp and other aquatic products; frequent consumption of soybean products and milk; and light cooking with little salt. Residents living with this dietary pattern have a lower incidence and mortality of hypertension and CVD and have higher life expectancy(Reference Zhang, Wu and Zhang13–15). The Chinese healthy diet is similar to the popular healthy diet abroad, which may be a healthy diet pattern suitable for the prevention and control of chronic diseases in the Chinese population. Nevertheless, associations between a Chinese healthy diet and the dynamics of CMM development have yet to be investigated.
To address the knowledge gap, we conducted the first prospective analysis of a Chinese healthy diet with the risk of CMM development among 18 720 Chinese adults with a median follow-up of approximately 7·3 years. We quantified dietary pattern scores from three consecutive 24 h dietary surveys at baseline. We primarily assessed whether the Chinese healthy diet was associated with the decreased risk of CMM development, and we secondarily explored the relative weights of food components.
Methods
Study design and participants
The present study used data from the China Health and Nutrition Survey (CHNS). CHNS is an ongoing population-based prospective cohort study, covering detailed standard dietary survey data and individual lifestyle and anthropometric and clinical data in fifteen provinces of China (the area east of the Hu Huanyong Line in China)(Reference Popkin, Du and Zhai16). This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the Institutional Review Boards of the Chinese Center for Disease Control and Prevention (no. 2018-012 and no. 201524) and the University of North Carolina at Chapel Hill (no. 07–1963). Written informed consent was obtained from all subjects.
In this study, we included adult participants aged ≥ 18 years who participated in the CHNS 1997–2018, covering the most recent complete survey data. We excluded participants diagnosed with T2D, MI or stroke at baseline and those with missing information on CMD or dietary intake. We also excluded participants with implausible energy intake information (> 5000 or < 800 kcal/d), leaving a total of 18 720 participants for the primary analysis (Fig. 1).
Flowchart of participant selection. Solid boxes indicate participants excluded based on predefined criteria, and dashed boxes represent the exclusion criteria. CHNS, China Health and Nutrition Survey; MI, myocardial infarction; T2D, type 2 diabetes; CMD, cardiometabolic disease.
Assessment of dietary intake
All food intakes for participants, including meals and snacks, were obtained through a 24-h dietary recall for 3 consecutive days (one weekend day and two weekdays). The 3-d 24-h recall method has been proven to be applicable in the dietary survey of adult populations in China(Reference Li, He and Zhai17). Field investigators were experienced professionals who had worked on other national surveys and received uniform training before the investigation. During the dietary survey, investigators used food pictures or models to assist respondents in recalling their dietary intakes, and each participant was also given paper meal record forms and received standard instructions for recording their daily intake of various foods so as to minimise recall bias. The consumption of food condiments (e.g. edible oils and salt) was weighed during the same 3-d period. Data on dietary energy intake were calculated based on food consumption data and the Chinese Food Composition Table(Reference Yang18).
Assessment of the Chinese healthy diet
In this study, we selected the recommended Chinese healthy diet in China as the main exposure factor. We calculated dietary scores to measure adherence to the Chinese healthy diet. We divided the food consumption into five groups for each food component of the Chinese healthy diet. Positive scoring (0–5 points) was adopted for fresh vegetables, fruits, aquatic products, legumes, milk and food diversity; reverse scoring (5–1 points) was adopted for salt, according to the actual intake ranking. Then, we added the scores of each food component to get the total score of the dietary pattern. For food components that had not been consumed, we gave a score of 0. This dietary score method has been widely used in nutritional research, such as the DASH score(Reference Fung, Chiuve and McCullough19). The Food Variety Score was adopted as the evaluation index for food diversity. Based on the ‘Chinese Dietary Guidelines’, four groups of foods – grains, starchy foods and legumes; vegetables and fruits; livestock and fish products and eggs; and dairy products and beans and nuts – were selected to form the Food Variety Score(Reference Zhu, He and Su20). More details were provided in online Supplementary Table 1.
Assessment of outcomes
Incident type 2 diabetes (T2D), MI and stroke cases were ascertained from self-reported information and blood biochemical results. In the survey, participants were asked to report their previous history of any of the above diseases diagnosed by a doctor and the specific age when they were first diagnosed with a questionnaire-based interview. In addition to the above two self-reported questions, there were two other criteria for the diagnosis of T2D. An additional question was asked about any of the following treatments: special diet, weight control, oral medicine, injection of insulin, Chinese traditional medicine, home remedies or qigong (or spiritual treatment), and a diagnosis of T2D was confirmed if at least one of the seven answers was yes. Fasting blood glucose ≥ 7·0 mmol/l or HbA1c ≥ 6·5 % from blood sample data were also used as the diagnosis criterion for T2D(21). First cardiometabolic disease (FCMD) was defined as the first incidence of any of T2D, MI and stroke during follow-up. CMM was defined as the coexistence of two or three CMD, including T2D, MI and stroke(Reference Di Angelantonio, Kaptoge and Wormser2,Reference Canoy, Tran and Zottoli22) . Follow-up was censored at participants’ loss to follow-up or the last date with available disease information, whichever came first.
Assessment of covariates
According to previous literature reported(Reference Han, Hu and Yu23), the following eleven covariates were retrieved from the baseline questionnaires and physical examinations: age (year, continuous), sex (male/female, categorical), urbanicity (urban/rural, categorical), marital status (single/married or cohabiting/divorce, widowhood or separation, categorical), highest education (combination the years of education and the highest degree, categorised into primary school and below, middle school, high school and above), annual per capita household income (calculated from various types of household income and personal income in the survey, categorised to three groups by tertiles), physical activity (the sum of leisure, household, occupational and traffic physical activities, categorised to three groups by tertiles) and BMI (calculated from physical examination, kg/m2, continuous). Smoking an average of at least one cigarette per day before or during the survey period was defined as a smoker. Smoking status (never/former/current, categorical), alcohol intake (grouped by whether or not they drank in the past year and the frequencies they drank in the past month, none/1–2 times a month/1–4 times a week/almost every day) and total energy intake (calculated from dietary survey, kcal/d, continuous) were also included.
Statistical analysis
We categorised the dietary pattern scores into quintiles for the entire study population. Baseline characteristics were described as the means (standard deviations) for continuous variables and percentages for categorical variables according to quintiles of dietary pattern scores.
We first used the traditional Cox proportional hazards model to estimate the associations of dietary pattern scores with each disease state (FCMD, CMM). No violation was observed when we used the Schoenfeld residuals method to check the proportional hazard assumptions. Models were adjusted for all eleven covariates described above.
Then we used the multistate models (MSM) to estimate the association of dietary pattern scores with transitions of CMM progression (Fig. 2), from free of CMD to FCMD and CMM, with the lowest quintile of the dietary pattern score as the reference group. MSM applies to survival data with several types of events, especially in co-morbidity studies and can describe the competing risk of transitions between different states(Reference Putter, Fiocco and Geskus24–Reference Zhang, Liu and Wang26). For participants who entered different states on the same date, we set the theoretical entry date of the former state as the entry date of the latter state minus 0·5 d(Reference Jiang, Tian and Liu27). To explore the effects of dietary patterns on specific FCMD linked to a risk of developing CMM, we conducted a more detailed analysis on various transitions from baseline to CMM. These involved FCMD were divided into three diseases: T2D, MI and stroke, as shown in pattern B (Fig. 2 and online Supplementary Table 2). If a participant was diagnosed with at least two CMD on the same date, we theoretically assumed that the order of disease occurrence was T2D, MI and stroke, and we subtracted 0·5 d for the earlier disease(Reference Jiang, Tian and Liu27). Models were adjusted for all the covariates mentioned. To assess the linear trend across quintiles of dietary pattern scores, we assigned a median value to each quintile and modelled the median score as a continuous variable.
Numbers (percentages) of participants in transition pattern A from baseline to FCMD and CMM and transition pattern B from baseline to one of T2D, MI and stroke and then to CMM. Pattern A: I, baseline to FCMD; II, FCMD to CMM. Pattern B: Ia–Ic, IIa–IIc are six subtransitions of stages I and II according to the types of FCMD. FCMD, first cardiometabolic disease; CMM, cardiometabolic multimorbidity; T2D, type 2 diabetes; MI, myocardial infarction.
We used the quantile G-computation method to analyse the relative contributions of food components to the transition of CMM progression. The quantile G-computation method estimates the relative weights of food components for positive and negative effects and has been widely used in the analysis of mixed exposures(Reference Keil, Buckley and O’Brien28,Reference Li, Guo and Jiang29) . To explore whether the weights of each food component were statistically significant, we fitted Cox regression models and MSM to include all food components and covariates as the independent variables and adopted the same model variables and quartile grouping settings as the quantile G-computation model.
To examine potential effect modifiers, we conducted a stratified analysis for the effects of dietary pattern on CMM in populations with different characteristics. To test the robustness of the analysis results, a sensitivity analysis was performed. First, we used logistic regression models to explore the relationship between dietary pattern scores and the risk of CMM and progression. Then, we grouped the dietary pattern scores into quartiles and again examined whether the relationships between dietary pattern and the outcomes of concern were affected. We performed all statistical analyses using R version 4.3.1 and considered a two-sided P value < 0·05 as statistically significant.
Results
Characteristics of the participants
Baseline characteristics of the whole population and subgroups of the lowest and highest quintiles of the Chinese healthy diet scores are shown in Table 1. The mean age was 43·6 (±15·0) years; 52·7 % of the participants were females, and 38·2 % of the participants were urban residents. More than two-thirds of the participants were non-smokers and non-drinkers. Participants with higher adherence to a Chinese healthy diet tended to be females, urban residents and non-smokers and have higher education and household income but lower physical activity.
Baseline characteristics according to quintiles of dietary pattern scores *
Q1 and Q5, the lowest and highest quintiles of dietary pattern scores.
* Data are presented as mean (sd) for continuous variables and percentages (%) for categorical variables.
† Dietary scores are presented as median (25th, 75th percentiles).
‡ Missing values are observed for the highest education, annual per capita household income, BMI and alcohol intake. The number of missing values per variable is equal to the total number of participants (18 720) minus the number presented in this column.
§ Groups of annual per capita household income and physical activity are according to 33·3rd, 66·6th percentiles.
Of the 18 720 participants, the median follow-up time from baseline to FCMD was 7·21 years and from baseline to CMM was 7·34 years. Among these participants, 2214 (11·8 %) experienced at least one CMD, of whom 1634 developed T2D, 247 experienced MI and 333 experienced stroke (Fig. 2). Of all incident CMD patients, 156 (transition II, 7·0 %) developed CMM, with a median follow-up time of 1·54 years from FCMD to CMM.
Chinese healthy diet and cardiometabolic multimorbidity
Transition pattern A was our primary model of CMM development (Fig. 2). We examined the relationships of the Chinese healthy diet with FCMD, CMM and transition from FCMD to CMM. As shown in Fig. 3, by comparing the highest with the lowest quintiles, models adjusted all eleven covariates showed that a Chinese healthy diet was associated with lower risks of CMM and transition from FCMD to CMM, respectively. Similar protection estimates were observed for Chinese healthy diet with a 55 % lower risk of transition from baseline to CMM (HR = 0·45, 95 % CI: 0·23, 0·87) and a 60 % lower risk of transition from FCMD to CMM (HR = 0·40, 95 % CI: 0·20, 0·81). Consistent results were observed when including Chinese healthy diet scores as continuous variables in the model.
Associations of quintiles of dietary pattern scores with the risks of first cardiometabolic disease (FCMD), cardiometabolic multimorbidity (CMM) and transition from FCMD to CMM using the Cox regression model and multistate model. Analyses were adjusted for gender, age, urbanicity, marital status, highest education attained, household income, smoking and drinking status, physical activity, BMI and total energy intake. Q2–Q5 represent the second to fifth quintiles of dietary pattern scores. The filled purple dots represent adjusted odds ratios, and the horizontal purple lines represent 95 % CI.
Food component analysis may provide further explanations for the protective effect of the Chinese healthy diet (Fig. 4). Overall, the results of food component analysis suggest that fresh fruits may play the most significant roles in delaying CMM and transition from FCMD to CMM. In the Chinese healthy diet, fresh fruits were identified as the most beneficial food components, accounting for 42·8 and 43·0 %, respectively, for beneficial associations with CMM and FCMD to CMM progression (online Supplementary Table 3).
Relative weight of each food component associated with cardiometabolic multimorbidity (CMM) and transition from first cardiometabolic disease (FCMD) to CMM. Analyses were adjusted for gender, age, urbanicity, marital status, highest education attained, household income, smoking and drinking status, physical activity, BMI and total energy intake. The x-axis represents the relative weight of each food component. The cyan bars represent food components with a negative coefficient in the model and statistically significant associations, while the grey bars represent food components with no statistically significant association with the risk of CMM or from FCMD to CMM in the model.
Stratified and sensitivity analyses
Considering the small sample size of transition from FCMD to CMM, we conducted a stratification analysis of the relationship between a Chinese healthy diet and CMM (Fig. 5). We yielded essentially unchanged associations between a Chinese healthy diet and the risks of CMM among participants married or cohabiting and those with higher household income. The associations were moderately stronger among the people aged > 45 years, non-smokers, non-drinkers and those with low physical activity as well as normal BMI.
Stratified analysis of estimated associations between a Chinese healthy diet and the risk of cardiometabolic multimorbidity (CMM) by comparing the highest with the lowest quintiles. Analyses were adjusted for gender, age, urbanicity, marital status, highest education attained, household income, smoking and drinking status, physical activity, BMI and total energy intake, with exclusion of the stratified variable as appropriate. The filled purple dots represent adjusted odds ratios, and the horizontal purple lines represent 95 % CI.
In sensitivity analyses, our results were robust when we used the logistic regression model to explore the associations between dietary patterns and CMM (online Supplementary Table 4). Moreover, when we divided the dietary pattern scores according to quartiles, compared with Q1, Q4 of the Chinese healthy diet also showed protective effects for delaying CMM and transition from FCMD to CMM (online Supplementary Table 5).
Discussion
Summary of main results
Leveraging a 21-year prospective study of 18 720 adults in the CHNS study covering detailed standard dietary survey data, we examined the traditional Chinese healthy dietary pattern in relation to CMM using MSM, an approach that allows the evaluation of disease progression within a unified analytic framework. We found that a Chinese healthy diet was associated with lower risks of CMM and transition from FCMD to CMM, respectively, which highlights its unique role in characterising the routes of CMM development. Of all food components, fresh fruits were robustly associated with the CMM development and provided the highest weight of protection effects. It should be noted that the associations between a Chinese healthy diet and CMM were inconsistent in various subgroups. Our study suggests that a Chinese healthy diet would be a superior dietary recommendation to reduce CMM risks in settings similar to the CHNS study.
Chinese healthy diet reduces the risk of cardiometabolic multimorbidity
To date, there is a lack of studies that specifically explore the temporal associations between a Chinese healthy diet and the dynamic progression of multimorbidity. Limited existing studies evaluated only the occurrence of a single disease(Reference Wang30,Reference Luo, Wang and Sun31) and showed an inverse association with CVD risks, resulting in the lack of ability to involve the potential stages of disease that can adequately address disease–disease interactions. To our knowledge, the present study is the first to reveal the critical role of the Chinese healthy diet on CMM dynamics, and it suggests that the Chinese healthy diet may have profound and long-lasting effects on the CMM. What’s more, we found only one research on the association between the Chinese healthy diet and health outcomes, and the result revealed the association with a reduced risk of hypertension(Reference Wang30). Our study can add high-level scientific evidence for the health effects of the Chinese healthy diet based on a cohort study.
Eating diverse types of foods is an internationally accepted recommendation for a healthy diet. Previous studies(Reference Roche, Creed-Kanashiro and Tuesta32–Reference Dello Russo, Formisano and Lauria34) pointed out that food diversity played a crucial role in fostering high-quality diets; high food diversity means the intake of a desirable variety of food and nutrients, leading to a balanced diet, especially in the elderly. Improving food diversity was also an important dietary strategy to reduce the risk of metabolic diseases(Reference Vadiveloo, Parekh and Mattei35–Reference Sarkar, Walker-Swaney and Shetty37). Food diversity included in the Chinese healthy diet may be one of the reasons for the protective effect.
Fresh fruits are the top protective contributor
It is well acknowledged that many foods have protective effects against CVD(Reference Mozaffarian38,Reference Badimon, Chagas and Chiva-Blanch39) . However, in this study, fresh fruits manifested the strongest association with CMM and transition from FCMD to CMM among all food components, suggesting that fresh fruits are the predominant CMM-risk-reduced food of the Chinese healthy diet in the CHNS adults. This result was not surprising and similar to earlier studies. Data from China National Nutrition Surveys indicated that low fruit consumption was associated with the largest number of estimated cardiometabolic deaths(Reference He, Li and Yang40). A previous large-scale, population-based Chinese cohort also showed that a greater fruit rather than vegetable consumption was associated with a lower risk of CVD mortality(Reference Wang, Liu and Li41). A prospective cohort study in eighteen countries (Prospective Urban Rural Epidemiology [PURE])(Reference Miller, Mente and Dehghan42) highlighted that, when examined separately, fruit intake was associated with lower risk of cardiovascular, non-cardiovascular and total mortality, while legume and vegetable intakes had no effect on cardiovascular risk. It could be potentially explained that, in the Chinese traditional diet, fruit is predominantly consumed raw, whereas vegetables may be consumed raw or cooked. Research indicates that cooking vegetables may lead to the degradation of nutrient and enzyme content and potentially generate harmful byproducts(Reference Link and Potter43). The existing literature regarding the association of dairy food consumption with CVD-related health outcomes remains controversial(Reference Guo, Astrup and Lovegrove44,Reference Chen, Zhang and Hu45) . A meta-analysis involving twenty-nine prospective cohort studies with 938 465 participants did not establish a significant association between dairy product intake and the risk of CVD(Reference Guo, Astrup and Lovegrove44), which is consistent with the findings of the present study. Our study indicated that increasing fresh fruit intake should be the top priority dietary improvement measure for preventing CMM risks in Chinese adults.
Disparities in different stratifications
Socio-economic status represents significant challenges and barriers to adopting a heart-healthy diet. Research in a representative US sample(Reference Shahu, Okunrintemi and Tibuakuu46) has demonstrated that individuals with lower income levels are less likely to receive dietary counselling aimed at CVD prevention. Compared with high- and middle-income families, individuals from lower socio-economic backgrounds tend to consume fewer fruits, vegetables and whole grains, while being more inclined to consume highly processed meats and sugary beverages(Reference Metcalfe, McCaffrey and Schumacher47). Sharing family meals can be a crucial factor in promoting cardiovascular health, and a positive atmosphere during mealtimes has been associated with improved dietary quality(Reference Vedanthan, Bansilal and Soto48). These findings may explain why stronger inverse associations between adherence to the Chinese healthy diet and CMM risks were observed among participants with higher household incomes and those who are married or cohabiting. Furthermore, in our study, the health effect of the Chinese healthy diet was found only in people aged over 45 years, which may be related to higher adherence to a healthy diet in older people(Reference Alkerwi, Vernier and Sauvageot49). Our stratified results suggest that challenges in addressing disparities in diet-related CMM need to be considered.
Strengths and limitations
This study has several strengths. To our knowledge, this is the first prospective study to explore the effects of a Chinese healthy diet on the longitudinal progression of CMM using the MSM in the Chinese population. The present results could provide important clinical significance for both primary and secondary prevention of CMM. Second, the Chinese healthy diet was derived from the traditional eating habits of Chinese residents; adherence to this Chinese healthy diet will be higher for Chinese residents than other healthy diets popular abroad. It could also provide a reference for healthy diets in other East Asian countries. In addition, the combination of the household condiment weighing method and the individual 3-d 24-h dietary survey method in CHNS can help to obtain more accurate individual dietary intake data, which is suitable for the investigation of complex food intake of non-standard portions in Chinese residents(Reference Zhai, Ma and Yu50).
Nonetheless, limitations are worth noting when interpreting the results of the present study. First, the sample size of the study was not large enough to have sufficient power to study the progress of each FCMD to CMM. Future studies with multiple cohorts or larger populations are needed to underscore specific routes of CMM development in the Chinese healthy diet. Second, in addition to diabetes, the outcomes we are concerned about lack objective clinical data for diagnoses, which may underestimate the number of outcome events in the CMM. However, only 3 % events of CMM were found among Chinese adults aged 30–79 (older than our study population) during a median follow-up time of 11·2 years (longer than our study) from the China Kadoorie Biobank data with 461 047 participants(Reference Han, Hu and Yu23). Another limitation pertains to the multistate algorithm, wherein we theoretically outlined the sequence of events for participants who reported two or more events occurring on the same day. This may marginally weaken the robustness of the MSM, but the impact is expected to be minimal, as the incidence of such cases is relatively low (n 74, constituting approximately 3·3 % of the FCMD participants). Due to the established correlation between stress and the development of cardiovascular and metabolic diseases, the omission of stress-related variables in this study should also be mentioned.
Conclusion
In conclusion, using the 21-year data from CHNS, we found that the Chinese healthy diet could reduce the risk of CMM, including the risk from health to CMM and from FCMD to CMM. Among all food components, fresh fruit intake is the top protective contributor for lowering these risks. Our study also suggests that demographic and socio-economic disparities in diet-related CMM need to be considered. Our study reveals the important role of a Chinese healthy diet in the first and secondary prevention of CMM and may provide evidence for dietary guidance of CMM prevention in China. Efforts should continue to be made to develop and validate a scoring system for the Chinese healthy diet.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114525105011.
Acknowledgements
We acknowledge all participants in this study and the staff responsible for the CHNS at the National Institute for Nutrition and Health, Chinese Center for Disease Control and Prevention and the Carolina Population Center at the University of North Carolina at Chapel Hill. Appreciations are expressed to the Core Competence Training Program of Epidemiology and Health Statistics, Chinese Center for Disease Control and Prevention. We thank the West China School of Public Health, Sichuan University, for statistical methods assistance.
This work was supported by the National Key R&D Program of China (2021YFE0114200) and the National Natural Science Foundation of China (82373681). This research used data from the CHNS. We are grateful for the research grant funding from the National Institutes of Health (R01-HD30880, DK056350, R24 HD050924, R01-HD38700) from 1997 to 2019. This study was also funded by the Ministry of Finance of the People’s Republic of China from 2015 to 2017 (13103110700015005).
The authors’ responsibilities were as follows. Conceptualisation: X. F. Z., C. S., X. X., H. J. W., Y. N. H.; methodology: X. F. Z., C. S., X. X., Q. Q. W., N. Z., Y. Z.; investigation: H. J. W., C. S., J. G. Z., X. F. J., F. F. H., Y. F. O. Y., B. Z.; formal analysis X. F. Z.; visualisation: X. F. Z., H. M. Z.; writing – original draft: X. F. Z., C. S.; writing – review and editing: X. F. Z., C. S., X. X., Q. Q. W., H. J. W., Y. N. H., F. X. G., Y. L. W., X. Z.; resources: H. J. W., B. Z., C. S.; and all authors read and approved the final manuscript.
The authors declare that they have no conflict of interest.
This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the Institutional Review Boards of the Chinese Center for Disease Control and Prevention (no. 2018-012 and no. 201524) and the University of North Carolina at Chapel Hill (no. 07–1963). Written informed consent was obtained from all subjects.
Data described in the manuscript will be made available upon request.
Consent for publication: Not applicable.
