Article Text
PDF
Abstract
Introduction Health equity is an important indicator measuring social development and solidarity. However, there is a paucity in nationwide studies into the inequity in health-related quality of life (HRQoL) in mainland China, in particular using the most recent data measuring HRQoL using the EuroQol 5-Dimension-5 Level (EQ-5D-5L). This study aimed to address the gap in the literature by estimating and decomposing income-related inequality of the utility index (UI) of EQ-5D-5L in mainland China.
Methods Data were extracted from the Psychology and Behaviour Investigation of Chinese Residents (2022), including 19 738 respondents over the age of 18 years. HRQoL was assessed by the UI of the EQ-5D-5L. Concentration index (CI) was calculated to measure the degree of income-related inequality in the UI. The contributions of individual, behavioural and context characteristics to the CI were estimated using the Wagstaff decomposition method.
Results The CI of the EQ-5D-5L UI reached 0.0103, indicating pro-rich inequality in HRQoL. Individual characteristics made the greatest contribution to the CI (57.68%), followed by context characteristics (0.60%) and health behaviours (−3.28%). The contribution of individual characteristics was mainly attributable to disparities in the enabling (26.86%) and need factors (23.86%), with the chronic conditions (15.76%), health literacy (15.56%) and average household income (15.24%) as the top three contributors. Educational level (−5.24%) was the top negative contributor, followed by commercial (−1.43%) and basic medical insurance (−0.56%). Higher inequality was found in the least developed rural (CI=0.0140) and western regions (CI=0.0134).
Conclusion Pro-rich inequality in HRQoL is evident in mainland China. Targeted interventions need to prioritise measures that aim at reducing disparities in chronic conditions, health literacy and income.
- Cross-sectional survey
- Public Health
- Health policy
- Health policies and all other topics
Data availability statement
Data are available on reasonable request. Data are not publicly available but may be requested from the authors on reasonable request. Data used in this study were extracted from the Psychology and Behaviour Investigation of Chinese Residents (PBICR) conducted by the corresponding author YW.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
We conducted a search on PubMed, Embase, Web of Science, Scopus, and CNKI (China National Knowledge Infrastructure) for articles published from the inception of the databases up to September 20th, 2023, using the search terms (("Health-Related Quality of Life" OR HRQoL OR EQ-5D) AND (Equit* OR Equalit* OR Inequit* OR Inequalit* OR Disparit*)) to capture studies reporting on health inequality.
Inequalities in health-related quality of life (HRQoL) across countries and regions, as well as the contribution of income and socioeconomic status to these inequalities, have been well-documented.
Most of the current research on HRQoL inequality in China primarily examined specific regions or populations, such as Shaanxi, Shandong, Tibet, and the elderly, and dates back several years.
The EQ-5D instruments have been widely used globally for measuring HRQoL. However, of the seven studies on HRQoL inequality using EQ-5D instruments in mainland China, five employed the EQ-5D-3L scale which has higher ceiling effects compared to the EQ-5D-5L.
In summary, there is currently no nationwide study investigating HRQoL inequality in mainland China using EQ-5D-5L among the entire population.
WHAT THIS STUDY ADDS
This study used the latest nationwide data, collected between 20 June 2022 and 31 August 2022, which measured HRQoL using the EQ-5D-5L.
The findings revealed the persistence of pro-rich inequality in HRQoL in mainland China, with the western (underdeveloped) region and rural areas exhibiting higher levels of inequality.
Disparities in chronic conditions, health literacy and income were identified as the primary contributors to the inequality in HRQoL
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The study is instrumental in informing government policies and developing targeted interventions aimed at reducing HRQoL inequality in China, which is critical for advancing universal health coverage for all citizens.
The healthcare system should focus on addressing disparities in health literacy of the public and better managing chronic conditions.
The successful implementation of these measures relies on a robust primary care and community health sector.
Reducing income-related health inequalities remains a significant challenge in China.
These issues are particularly important in less developed western regions and rural areas.
Introduction
All member states of the United Nations (UN) endorsed ‘The 2030 Agenda for Sustainable Development’ in 2015, with 17 Sustainable Development Goals (SDGs) at its core promising to leave no one behind.1 The SDG-3 aims to ensure healthy lives and promote well-being for all at all ages: Every individual and community, irrespective of their circumstances, should receive the health services they need without risking financial hardship.2 Therefore, health equity is the core and ethical requirement of public health efforts.3 Health inequity not only jeopardises population health outcomes, but also violates people’s moral standards.4 Unfortunately, health inequalities are widespread both within and across countries.5 Achieving health equity has become an overarching goal of health systems in both developed and developing countries.6
Health-related quality of life (HRQoL) is a multidimensional concept measuring health outcomes from the consumer perspective. It is defined as ‘how well a person functions in their life and his or her perceived well-being in physical, mental and social domains of health’.7 HRQoL measurements cover aspects of one’s personal, family, work and social function.8–10 Common instruments assessing HRQoL include the WHO Quality of Life Measurement Scale (WHOQOL-100 and WHOQOL-BREF), the EuroQol 5-Dimension (EQ-5D-3L, EQ-5D-5L and EQ-5D-Y), the Short-Form 36 (SF-36, SF-12 and SF-6), the Health Utility Index (HUI2 and HUI3), the 15-dimensional measure (15D) and the Quality of Well-being Scale Self-Administered Form.11 The EQ-5D family is the most commonly used instrument in assessing HRQoL.12 Empirical evidence shows that HRQoL is associated with a wide range of factors. Apart from the objective health indicators (e.g., chronic conditions), HRQoL also varies by gender, age, marital status, household income and social status.13 14
There has been an increasing research attention to socioeconomic inequity in HRQoL among general populations,15 16 as well as in older people17 and those living with chronic conditions.18 However, there is a paucity in the literature documenting HRQoL inequality in China despite serious concerns over regional and socioeconomic inequalities in health. Previous studies attempted to use the EQ-5D-3L to assess inequality in HRQoL, but were often hampered by the profound ceiling effects (a substantial proportion of respondents report no health problems and achieve the maximum score).6 19–21 Most also restricted study participants to certain groups (e.g., elderly22 and employees23) or in a special region (e.g., Shaanxi,24 25 Shandong22 and Tibet26). This study aimed to address the gap in the literature by estimating and decomposing inequality of HRQoL assessed by the EQ-5D-5L in a nationwide sample in mainland China.
Methods
Data collection
Data were extracted from the Psychology and Behaviour Investigation of Chinese Residents (PBICR) conducted by the School of Public Health of Peking University.27 The PBICR adopted a multistage stratified quota sampling strategy to select study participants, covering 780 residential communities from 202 districts/counties in the 31 provinces/regions in mainland China (22 provinces, 4 municipalities directly overseen by the central government and 5 autonomous regions).
The survey was conducted over the period from 20 June 2022 to 31 August 2022 when China continued to implement a zero-tolerance policy for COVID-19. Each study participant was approached by a trained investigator face to face and invited to self-complete the online survey through the platform Wen Juan Xing (https://www.wjx.cn/). Each team of investigators was responsible for collecting 100–200 questionnaires, with each team member contributing 30–90. Eligible participants included those who were 12 years or older, resided in the survey location permanently with no more than 1-month absence, and were able to understand the survey questions and provide informed consent. Those who had serious mental health disorders, cognitive impairment or were involved in other similar research projects were excluded.
In total, 31 499 eligible residents accessed the survey and 22 630 (71.80%) returned their responses. Two investigators cross-checked each entry of data to identify logical errors and subsequently removed 714 (3.20%) records. This current study further restricted study participants to those aged 18 years and above, resulting in a final sample of 19 738 for data analyses. This was because the Chinese version of EQ-5D-5L descriptive system and utility value set in mainland China was developed based on the studies of those aged 18 years and above,28 29 and the existing studies in the USA,30 New Zealand31 and Russia32 also restricted study participants to those aged 18 years and above.
Role of the funding source
The research funder had no role in the study design, data collection, analysis and interpretation, decision to publish or writing of the manuscript.
Patient and public involvement
This study did not involve patients. The public were not involved in the design, reporting or dissemination plans of our research. The privacy and confidentiality of the participants were protected during and after the data collection.
Variables and measures
The 2022 PBICR collected data regarding individual sociodemographic characteristics, personal health and behaviours, and family and social environments.
Dependent variable
The outcome indicator, HRQoL, was assessed using the EQ-5D-5L developed by the European Society for Quality of Life (EuroQol), which is the most widely used instrument for assessing HRQoL of general populations across the world.33 Its Chinese version has been validated,28 including a population value set for estimating utility index (UI).29 The EQ-5D-5L has stronger discriminatory power and better convergent validity compared with its predecessor EQ-5D-3L, with a lower ceiling effect: 60.50% had an UI score of 1.000 for the EQ-5D-5L compared with 72.08% for the EQ-5D-3L.34 35
The UI, which is a major feature of the EQ-5D instrument,12 reflects a combination of health functioning in the five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. It was derived from population preferences for different health states.36 This study adopted the Chinese value set for the EQ-5D-5L, which has a range from −0.391 to 1.000, with 1.000 indicating the best possible health and a negative score indicating a state worse than death.29
Independent variables
HRQoL is determined by many factors, including those biological, social and economic.37 38 The social determinants of health model have been developed to explore the fundamental causes of health issues.22 26 39 According to Andersen, contributors to health outcomes can be categorised into individual, behavioural and context characteristics.40
Individual characteristics: Individual characteristics can be further classified as predisposing, need and enabling factors. In this study, gender (male, female) and age (18–44, 45–64, 65+ years) measured the predisposing factors. Need factors were indicated by medically diagnosed chronic conditions (0, 1, 2 or more) and sleep quality (poor: 0–4, fare: 5–8, good: 9–12) rated on a three-point scale containing four items derived from the Pittsburgh Sleep Quality Index (sleep onset latency, sleep duration, sleep disturbances and perceived quality).41 Marital status (unmarried, married, divorced/widowed), employment status (employed, student, retired, unemployed), educational attainment (illiteracy or elementary, junior high school, high/vocational school, university), health literacy (low: 0–9, medium: 10–18, high: 19–27) measured by the SF Health Literacy Survey questionnaire,27 per capita household income (≤¥1000, ¥1001–¥2000, ¥2001–¥3000, ¥3001–¥4000, ¥4001–¥5000, ¥5001–¥6000, ¥6001–¥9000, ¥9001–¥12 000, ¥12 001–¥15 000 and ≥¥15 001), and health insurance coverage (no, yes) were deemed as enabling factors.
Health behaviours: The PBICR assessed current status of respondents in smoking (no, yes), alcohol drinking (no, yes), tea drinking (no, yes), self-reported total water intake (<1500 mL/day, ≥1500 mL/day),42 and physical activity scores measured by metabolic equivalent of task using the International Physical Activity Questionnaire (very poor: ≤4000, poor: 4001–8000, acceptable: 8001–12 000, good: 12 001–16 000, very good: ≥16 001).43 Respondents were also asked whether they were taking nutritional supplements (e.g., protein, mineral elements, vitamins) over the past year (no, yes).
Context characteristics: The PBICR recorded the residential location of the respondents and whether they were under lockdown at the time in response to COVID-19 outbreaks (no, yes). Respondents were also asked to report exposure (no, yes, not sure) to industry pollution within 5 km of their home. Significant regional disparities in socioeconomic development exist in China. The eastern (developed) region and urban areas have accumulated more wealth and health resources than their central (developing), western (underdeveloped) and rural counterparts.44 45 Exposure to industrial pollution and public health emergency response measures has profound impacts on health needs and healthcare use of people.46
Statistical analysis
Concentration index
Income-related inequality in HRQoL was assessed by concentration index (CI). CI is both intuitive and comprehensive, which is sensitive to distributional changes. It takes a value between −1 and +1, with 0 representing non-existence of income-related inequality.47 A positive CI indicates pro-rich, while a negative CI indicates pro-poor.
where C is CI, 
denotes the HRQoL of the i-th individual, μ is the mean of overall HRQoL, 
represents the fractional rank of income distribution, which was divided into 10 groups ranging from poorest to richest based on per capita household income.
Decomposition of CI
HRQoL (y) was determined by multiple factors:
where 
represents the independent variable 
for each individual i, 
is the marginal effect of each independent variable 
, 
stands for the error.
The decomposition of the total CI of HRQoL can be written as follows47:
where 
is the mean of 
; 
is the mean of HRQoL; 
is the CI of independent variable 
; 
is elasticity, which reflects the relative importance of each independent variable’s CI. 
is the total contribution of the independent variables; 
is the generalised CI of 
.
Horizontal inequality index
Horizontal inequity index (HI) indicates inequality among people who have the same unavoidable health determinants.6 47 In this current study, unavoidable determinants indicate the individual predisposing factors of gender and age. HI removed contributions of the unavoidable variables from the overall CI, reflecting the contributions of the avoidable determinants including the individual need factors and enabling factors, health behaviours and the context characteristics.
To test the robustness of the statistical results, we also performed decomposing analyses on the CI of the Visual Analogue Scale (VAS) results embedded in the EQ-5D-5L scale.
Results
Characteristics of respondents
The quota sampling strategy led to over-representation of western residents due to the large geographic catchment in western China despite its low population size: more questionnaires (42.33%) were returned from the western sample than from the eastern (32.95%) and central (24.72%) ones. About 30.49% of respondents lived in rural areas. Nearly half (49.47%) were male and 14.15% were older than 65 years, which resemble the national demographic distributions.48
The vast majority of respondents completed high school education (68.60%), lived with no chronic conditions (74.56%), enrolled in basic social health insurance (92.13%) but not in commercial medical insurance (94.75%), reported no close exposure to industry pollution (70.18%) and were not experiencing community lockdown at the time (96.56%). Income distribution was dispersed, with 33.01% having a per capita monthly household income below ¥3000 (US$410) compared with 25.82% earning over ¥6000 (US$821). About one-third (32.56%) of respondents drank alcohol, 19.51% smoked cigarettes and 41.82% took nutritional supplements (table 1). online supplemental file S1 and S2 report the characteristics of respondents stratified by region and place of residence.
Supplemental material
Characteristics of study participants
HRQoL and its determinants
The EQ-5D-5L had a Cronbach’s alpha coefficient of 0.8070 and a Kaiser-Meyer-Olkin coefficient of 0.8030 in this study (online supplemental file S3). The respondents obtained an average score of 0.9397 (SD=0.1367) for UI and 62.35% had an UI score of 1.000. The UI scores decreased by age (p<0.0001). Women had higher UI than men (0.9433 vs 0.9361). Rural respondents had lower UI than their urban counterparts (0.9308 vs 0.9436, p<0.0001), while those in the central region had the highest UI (0.9433). The UI scores also varied by marital status and employment status (p<0.0001). Overall, higher socioeconomic status, healthier states, lower risk behaviours and healthier environments were associated with higher UI (table 2). online supplemental file S4 and S5 report the EQ-5D-5L UI results by the context characteristics. The percentages of study participants reporting health problems are presented in online supplemental file S6–S11.
EQ-5D-5L utility index of study participants
CI of respondents
The CI (0.0103) of the UI showed pro-rich inequality, with lower HRQoL being concentrated among those with lower income. The western regions had the highest CI (0.0134). Rural areas had higher CI than their urban counterpart (0.0140 vs 0.0081, p<0.0001), except for the eastern region (table 3).
Concentration index (CI) for EQ-5D-5L utility index
Decomposition of the CI
Individual characteristics contributed the largest proportion to the CI (57.68%), followed by the context characteristics (0.60%) and health behaviours (−3.28%). The contribution of individual characteristics was mainly attributable to disparities in the enabling (26.86%) and need factors (23.86%). Educational attainment (−5.24%) was the top negative contributor, followed by commercial (−1.43%) and basic medical insurance (−0.56%) (table 4).
Decomposing concentration index (CI) of EQ-5D-5L utility index by place of residence
Individual characteristics contributed the largest proportion to the CI in both rural (51.27%) and urban areas (50.41%). Income (17.56%), chronic conditions (13.08%) and sleep (8.73%) were the top contributors to the CI in rural areas. By contrast, health literacy (16.14%), chronic conditions (15.46%) and income (13.83%) were the top three contributors to the CI in urban areas. Income disparities played a less profound role for the CI in urban areas (13.83%) than in rural areas (17.56%), while health literacy played a more significant role for the CI in urban areas (16.14%) than in rural areas (6.00%) (table 4).
Individual characteristics contributed the largest proportion to the CI in all of the regions: eastern (33.73%), central (48.70%) and western (62.72%). The top three contributors to the CI in the eastern region were chronic conditions (10.11%), income (9.20%) and sleep (6.81%), compared with chronic conditions (19.34%), health literacy (15.10%), and educational level (−11.97%) in the central region, and income (19.13%), chronic conditions (17.10%), and health literacy (12.68%) in the western region (table 5).
Decomposing concentration index (CI) of EQ-5D-5L utility index by regions
Horizontal inequality index
The pro-rich inequality was mainly attributable to the avoidable determinants as indicated by the smaller changes in HI (table 6) in comparison with the CI (0.0096). Rural respondents had higher HI than their urban counterparts (0.0133 vs 0.0077). The western region exhibited the highest HI value (0.0125), surpassing both the eastern (0.0091) and the central regions (0.0086).
Horizontal inequality index (HI) of EQ-5D-5L utility index
Discussion
This current study revealed an CI of 0.0103 for the EQ-5D UI, which is slightly lower than that reported in the Korea (0.035 for EQ-5D UI) in 2012,49 and much lower than those in Chile (0.063 for EQ-5D UI) in 2017 and Iran (0.299 for EQ-5D UI) in 2018.50 51 The CI revealed in this study is also lower than those reported in previous studies in China: 0.024 in urban and 0.048 in rural areas in 2015,22 and 0.022 in Tibet in 2014.26 However, the CI revealed in this study is still higher than those (0.0076 for rural and 0.0093 for urban) reported in Shaanxi province in 2013.6
Disparities in chronic conditions, health literacy and income are the top three contributors to inequality in UI according to the findings of this current study. These factors have been consistently reported in the literature as major contributors of health inequality in China.6 26 52 Higher income not only contributes directly to higher use of healthcare and better health outcomes, but is also associated with differences in other health determinants such as health behaviours and context characteristics,52 both of which can aggravate health inequality.53–55 Higher income is associated with better access to health resources,56 higher use of healthcare services57 and less health problems.58 59 However, people with a lower income may have a lower expectation on health, which may prevent poor people from taking action to promote health.
Disparities in health literacy contribute positively to the inequality in HRQoL. Health literacy is an important determinant of health and a realistic way for individuals to participate in their own medical care and health promotion.60 People with higher health literacy may engage more in health-promoting activities to maintain fitness, especially during COVID-19.61 Higher health literacy is also associated with higher income and higher expectations on health and well-being,62 63 which helps widen income-related health inequalities.
Disparities in chronic conditions widen the inequality in HRQoL. Chronic conditions have profound implications on the physical, mental and social functioning of people that lead to lower HRQoL.64 China is experiencing significant increase in burdens of chronic diseases thanks to its dramatic industrialisation and urbanisation in recent years.65 Healthcare needs in people with chronic conditions are usually higher compared with the general population. However, access to care for chronic conditions is not equal due to financial barriers imposed by user charges despite almost universal coverage of social health insurance.66 This can further exacerbate the existing disparities in HRQoL associated with chronic conditions.67
We found in this study that educational attainment has a negative contribution to inequality in HRQoL, although higher education itself is associated with higher HRQoL. Educational disparities in China have shrunk dramatically over the past few decades.68 Meanwhile, higher educational attainment may lead to higher risk of sedentary lifestyle and higher health expectations.69 These may partly explain the negative contribution of education to inequality in HRQoL.
The findings of this current study verify the continuous existence of regional and urban–rural health inequalities. We found that the western underdeveloped regions and rural areas have higher CI than their eastern (developed), central (developing) and urban counterparts. Income disparities played a much more profound role for the CI in the western regions and rural areas. Previous studies also identified higher health inequality in western China in comparison with eastern and central China.5 70 The rapid urbanisation process in China has been accompanied with large scales of internal migration, which led to the serious polarisation in underdeveloped areas, resulting in increased ‘left behind’ children and elderly in China.71 72 Overall, however, regional and urban–rural disparities have now played a less significant role in inequality of HRQoL than others: their contributions to CI do not exceed 3.00% according to the findings of our current study. On the one hand, the improvement of transportation infrastructure, especially the high-speed rail has led to increased availability of medical and health resources, reducing the health gap across urban–rural areas.73 China’s efforts to address regional inequality may have taken effects. Indeed, like in many high-income countries,74 75 health equality has become one of the major policy goals in recent health system development in China.76
Our current study shows that the unavoidable factors (age and gender) make limited contribution to inequality in HRQoL, although remaining a significant role. With the increase of age, HRQoL declines.10 77 Older people tend to be poorer, in particular in rural areas.78 However, these factors can only explain a small percentage of CI, which highlights the importance of addressing the avoidable factors.
Over the past few decades, China has rapidly expanded its health insurance coverage.79 80 Health insurance is indeed helpful for reducing inequalities in HRQoL, as indicated by the findings of this current study. However, the contribution of health insurance to inequalities in HRQoL is relatively smaller compared with other factors.
It is important to note that different measurement tools may lead to different results in the CI. We have, therefore, reported our findings using the VAS measurement in supplementary files (online supplemental file S12–S17). We found that the VAS scores varied by educational attainment generally, but the UI did not. By contrast, the UI scores varied by commercial insurance coverage, but the VAS did not. VAS reflects an individual’s perception of overall health, which is closely associated with one’s expectation on health. Clearly, individual values are not always aligned with population endorsed values. People with different educational attainments may give different VAS scores to the same health condition due to different expectations. However, the benefit of health insurance does not consider individual differences in perceptions and expectations.
The findings of this study have significant policy implications. Despite China’s extensive efforts to achieve over 95% basic social health insurance coverage,79 80 questions have arisen about its effectiveness in reducing health inequalities.52 Some researchers argue that health insurance programmes may benefit wealthier populations more, especially when out-of-pocket payments are still required. The affordability of healthcare services can be improved for all, but the poor are more likely to forgo necessary services due to financial barriers associated with out-of-pocket payments. A universal top-down health policy approach might inadvertently exacerbate inequity in such circumstances. As a result, priority should be given to disadvantaged populations. Our study’s findings highlight the need for targeted interventions within the healthcare sector, focusing on reducing disparities in chronic conditions and health literacy. These issues are particularly important in less developed western regions and rural areas, where income disparities also play a significant role. Strengthening the primary care sector is essential, as it is well positioned to address health literacy gaps and manage chronic conditions effectively. Universal access to health information is essential to achieving universal health coverage and empirical evidence demonstrates that using educational tools such as short message service reminders can significantly enhance people’s quality of life.81 82
Limitations
There are several limitations in this current study. First, this study sample excluded those aged 17 years and below; therefore, the conclusion may not be generalisable to the whole population. Second, owing to the cross-sectional data, inference of causality needs to be cautious. Finally, the limitation of sampling methods led to over-representation of western residents and this issue should be noted when applying the results of this study.
Conclusion
There remains pro-rich inequality in HRQoL in mainland China. The top contributors to inequality in HRQoL are disparities in chronic conditions, health literacy and income. Although the western (underdeveloped) regions and rural areas have higher inequality in HRQoL, regional disparities are no longer a major concern. Indeed, over the past few decades, regional health disparities have been shrinking due to great efforts in universal health insurance coverage and infrastructure development thanks to the unprecedented socioeconomic development in China. This study expands our understanding on HRQoL inequalities in China and offers some insights into potential strategies for reducing HRQoL inequalities, which is critical for achieving adequate health universal coverage.
Data availability statement
Data are available on reasonable request. Data are not publicly available but may be requested from the authors on reasonable request. Data used in this study were extracted from the Psychology and Behaviour Investigation of Chinese Residents (PBICR) conducted by the corresponding author YW.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and this study was approved by the Ethics Research Committee of the Health Culture Research Center of Shaanxi (JKWH-2022-02). Implied informed consent was obtained prior to each survey. Participants gave informed consent to participate in the study before taking part.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
QY and XZ are joint first authors.
Handling editor Lei Si
Twitter @liuchaojie
Contributors QY contributed to the study design, data analyses, data interpretation and drafting of the manuscript. XZ contributed to data analyses, data interpretation and drafting of the manuscript. YW contributed to data analyses and data interpretation of the manuscript. CL contributed to the interpretation of the results and writing of the manuscript. All authors have read and approved the final version of the manuscript. QY is responsible for the overall content as guarantor.
Funding The study was funded by the National Natural Science Foundation of China (72174149), Humanity, and Social Science Foundation from the Ministry of Education of China (21YJAZH102), and the Key Research Institute Project of Humanity and Social Science of the Ministry of Education of China (1203-413100050).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.