Browse the corpus

While the core design principles remain unchanged in the 21st century, contemporary cohort studies are often large (n>100 000), long term, and complex, with collection and long term storage of biological samples (eg, blood, urine, stool), comprehensive exposure assessment using conventional (eg, questionnaire) and novel approaches (eg, wearables, imaging, multiomics assays), and electronic linkages to different healthcare records. However, the complexity of the study procedures and data collected warrant carefully designed IT, data management, and quality control systems, as well as cross disciplinary collaboration throughout different phases of study. There is also an increasing need for cloud computing, AI, and data science to store, process, and analyse high dimensional big (eg, whole genome sequencing) data. As such, any single institute may not have all the essential infrastructure and expertise to effectively establish and manage a large cohort study and to fully utilise the resultant data. Collaboration at every stage of cohort development with researchers nationally and internationally is essential to maximise the scientific potential and global impact of a cohort.

International collaboration on cohort studies in China typically involves contribution of scientific insights and analytical skills to specific papers; participation of Chinese cohorts in international consortiums; comparative investigations across Chinese and international cohorts; or long term partnerships between Chinese and overseas researchers, often covering the full study lifecycle. These approaches are not mutually exclusive, with some cohorts having multiple tiers of collaboration, all underpinned by openness and data sharing. Among existing cohorts involving extensive international collaboration and data sharing in China, the China Kadoorie Biobank (CKB) and UK Biobank, each with 0.5 million participants, are the leading examples, along with other large (n >50 000) Chinese cohorts (eg, Shanghai Women’s Health Study (SWHS) and Shanghai Men’s Health Studies (SMHS)) (table 1). Main large population based prospective cohort studies in adults in China Sample size of adult participants. Some cohorts included children and the corresponding sample size was not indicated in this table. Primary international collaborating institutes involved since study conceptualisation. Follow-up for mortality ended in 1999. Successions of periodic nationwide surveys conducted every 3 years as baseline; biosamples collected since 2010. Passive follow-up restricted to township or street level healthcare records but not national systems, so follow-up may be incomplete. Multi-component cohort constituted of multiple sub-cohorts of distinctive population characteristics.

Successions of periodic nationwide surveys conducted every 3 years as baseline; biosamples collected since 2010. Passive follow-up restricted to township or street level healthcare records but not national systems, so follow-up may be incomplete. Multi-component cohort constituted of multiple sub-cohorts of distinctive population characteristics. Age range or baseline survey year was not clearly reported in the published material. The CKB was established in 2004-08 with a Sino-British collaboration covering the entire project lifecycle. In particular, the Oxford based team raised initial funding and developed detailed procedures and IT systems for study conduct, quality control, management, and monitoring; both teams worked closely in undertaking pilot studies and refining the study design, procedures, and fieldwork plans; and the China based team mobilised and managed immense human and logistical resources for implementation. Several other major cohorts adopted broadly similar approaches, with, for example, the SWHS and SMHS co-established as a China-US collaboration.4 8 While the CKB benefited from experience learnt in earlier cohort studies (eg, Mexico City Prospective Study, MCPS), it also informed the UK Biobank, which was established in 2006-10. As one of the most extensively used open medical data resources worldwide, the UK Biobank has also benefited China by enhancing capacity development and enabling rapidly growing high impact research among a large number of Chinese scientists (that is, 16% of total) (fig 1).

UK Biobank, which was established in 2006-10. As one of the most extensively used open medical data resources worldwide, the UK Biobank has also benefited China by enhancing capacity development and enabling rapidly growing high impact research among a large number of Chinese scientists (that is, 16% of total) (fig 1). UK Biobank research statistics on cumulative number of active research projects by region in 2012-23 and number of peer reviewed publications by region in 2012-23

UK Biobank, which was established in 2006-10. As one of the most extensively used open medical data resources worldwide, the UK Biobank has also benefited China by enhancing capacity development and enabling rapidly growing high impact research among a large number of Chinese scientists (that is, 16% of total) (fig 1). UK Biobank research statistics on cumulative number of active research projects by region in 2012-23 and number of peer reviewed publications by region in 2012-23 Recognising the unique value of biobank studies, during 2016-18 the Chinese government invested 400 million Chinese Yuan (£41m; $56m; €48m) to establish more than 25 population and disease specific cohorts, including seven community based cohorts as part of the Million Cohort Initiative, for which the CKB was selected as a demonstration project (table 1).18 Moreover, as medical record linkages become increasingly feasible, many nationwide representative cross sectional surveys (eg, the China Chronic Disease and Risk Factor Surveillance (CCDRFS)) have been converted into prospective cohort studies.9 In 2017, the China Cohort Consortium (CCC; https://chinacohort.bjmu.edu.cn/) was established to facilitate collaboration, involving 154 cohorts of over 100 million individuals to date.19 Chinese cohorts are increasingly sharing data nationally and internationally, but most require direct collaboration on specific projects. Although the CKB has been able to share individual participant data worldwide since 2015, it only involves non-genetic data owing to policy constraints in China and the number of open access projects was much smaller compared with the UK Biobank. Altogether, cohort studies in China offer uniquely rich opportunities for international collaboration, especially if the genetic data can be shared more widely, and the post-2016 cohorts have laid promising ground for future collaboration on rapidly evolving health issues in younger generations.

Cohort studies in China can capture a different range of risk exposures, genetic architecture, disease patterns, and social context from those undertaken in other populations. These greatly diversify and enrich global population health research, shedding new light on longstanding evidence gaps that could not be properly addressed by cohorts in western high income populations alone. For example, cohort studies in China could reliably examine the associations of blood cholesterol with ischaemic heart disease at levels well below those typically seen in western populations,20 21 informing the design of statin trials in the west, which went on to demonstrate the benefits of lipid lowering at all levels.22 Using East Asian specific “flushing” variants (absent in white people), a Mendelian randomisation study in China reliably refuted a longstanding belief in the benefits of moderate alcohol drinking on the risk of cardiovascular disease.23

west, which went on to demonstrate the benefits of lipid lowering at all levels.22 Using East Asian specific “flushing” variants (absent in white people), a Mendelian randomisation study in China reliably refuted a longstanding belief in the benefits of moderate alcohol drinking on the risk of cardiovascular disease.23 Studies in China also enable investigation of specific diseases that are now relatively rare (and thus difficult to study) in western and other high income East Asia (eg, Japan) populations, as exemplified by the contrasting causal associations of low density lipoprotein cholesterol with ischaemic stroke (positive) and intracerebral haemorrhage (inverse) in Chinese adults, which informed the safe use of statins in populations with high rates of intracerebral haemorrhage.21 22 Moreover, the complex interplay of genes and environment in influencing disease risks may operate differently across populations. This was demonstrated by findings of the importance of higher homocysteine levels for increased risk of stroke, but not ischaemic heart disease, in a low folate intake Chinese population,24 in contrast to previous null findings in populations with high folate intake. Furthermore, successive large Chinese cohorts have documented the growing tobacco related mortality among men (in contrast to the declining trends among Chinese women and in high income populations), informing China’s national tobacco control policy6 25 and long term monitoring of smoking (and other exposures) in national representative cohorts (eg, CCDRFS).

horts have documented the growing tobacco related mortality among men (in contrast to the declining trends among Chinese women and in high income populations), informing China’s national tobacco control policy6 25 and long term monitoring of smoking (and other exposures) in national representative cohorts (eg, CCDRFS). International collaboration can also facilitate knowledge exchange and capacity development, improve study design and ethical standards, attract funding from multiple sources, introduce best practices, enhance data quality, and increase study visibility and global impact. At the conceptualisation stage, international collaboration helps to identify research priorities of global importance, while ensuring the study design, main objectives, and operation are tailored to the local context. At the planning and implementation stages, there could be sharing of expertise and identification of best practices in study conduct (eg, data security and management). In particular, as ethics and data management regulations often differ across countries, international collaboration would naturally require the study to be conducted to the highest standards across all countries. For study conduct, traditional paper based data collection poses major logistical and quality control challenges to large cohort studies. The CKB was the first cohort in China to be paper-free, enabled by a laptop based electronic questionnaire with built-in logic and error checks as part of a comprehensive bespoke IT system for data collection and study management.26

llection poses major logistical and quality control challenges to large cohort studies. The CKB was the first cohort in China to be paper-free, enabled by a laptop based electronic questionnaire with built-in logic and error checks as part of a comprehensive bespoke IT system for data collection and study management.26 The size and complexity of contemporary cohort studies require substantial investment not only to enrol participants but also to maintain the study long term (eg, outcome follow-up, periodic re-survey, storage of biosamples) and to turn samples into data using new technologies (eg, whole genome sequencing, multiomics). International collaboration can effectively pool funding from multiple sources. For example, the CKB was initially funded by a philanthropic seed grant from the Hong Kong-based Kadoorie Charitable Foundation, and the successful establishment of the cohort has subsequently attracted further public and charitable funding from various sources in China (eg, National Natural Science Foundation, Ministry of Science and Technology) and the UK (eg, Medical Research Council, Wellcome Trust) and industry (eg, Merck, GSK, BGI).

ation, and the successful establishment of the cohort has subsequently attracted further public and charitable funding from various sources in China (eg, National Natural Science Foundation, Ministry of Science and Technology) and the UK (eg, Medical Research Council, Wellcome Trust) and industry (eg, Merck, GSK, BGI). Similarly, as a global open access resource, the UK Biobank has attracted substantial funding from industry for large scale assays (eg, whole genome sequencing, proteomics, metabolomics),27 28 which would not be feasible with traditional national government or charitable funding. Collaboration with industry also helps to improve assay technology and reduce costs. The data generated through industry funding are also released to the wider research community after a short period of exclusive use, benefiting a large number of users globally. By charging a modest access fee (with discount for users from low income countries) for cost recovery purpose, it also helps to improve the long term sustainability of the UK Biobank leveraged by numerous research projects using the data. This “open access” model is increasingly common in large scale epidemiological research, as funders, researchers, and scientific journals recognise the data generated as common goods to humanity. Data sharing and international collaboration can also help improve the quality of research by enabling reproducible science with cross validation across and within cohorts, as well as reducing publication bias.

Clearly, international collaboration entails important challenges. Firstly, cultural and language barriers between different stakeholders are inevitable. Experiences of successful Chinese cohorts have shown clear advantages of having key personnel with diverse cultural understanding and research experience. For example, both the CKB and SWHS have benefited significantly from having at least one lead investigator of Chinese origin with research training both within and outside China, who can facilitate collaboration and resolve conflicts at all stages of study conduct. During the long term collaboration, mutual academic visits and training of junior researchers are important to strengthen relationships and ensure sustainability from the bottom up. Ethics is one key aspect that requires in-depth mutual understanding of research practices and culture, especially given the ever evolving and distinctive ethical standards and regulations internationally. The concept of informed consent may differ by population, and the wordings used in information sheets may not always be directly translatable across cultures. It is therefore essential to conduct iterative internal review, pilot testing, translation, and back translation of different study materials.

s internationally. The concept of informed consent may differ by population, and the wordings used in information sheets may not always be directly translatable across cultures. It is therefore essential to conduct iterative internal review, pilot testing, translation, and back translation of different study materials. Another major challenge lies in the research assessment culture. Researchers in China are under immense pressure from their highly competitive assessment system, which, for example, does not fully recognise shared first or senior authorships—something that is common and often inevitable in large scale collaborations. This issue also exists outside China, but there has been promising development to promote recognition beyond mere ranking of authorship by focusing on individuals’ specific contributions.29 A successful collaboration also requires a clear governance framework and agreement on data and intellectual property ownership, which may differ depending on the nature of collaboration and levels of investment from each party.

ion beyond mere ranking of authorship by focusing on individuals’ specific contributions.29 A successful collaboration also requires a clear governance framework and agreement on data and intellectual property ownership, which may differ depending on the nature of collaboration and levels of investment from each party. Major challenges also arise from rapidly evolving geopolitical tensions and regulatory requirements internationally. In 2019, the Chinese government passed a new law regulating the collection, use, and sharing, both cross border and within China, of human genetic materials and information.30 While the law serves to ensure the appropriate use of genetic data for research and clinical translation, the initial lack of detailed implementation regulations has largely stalled many international collaborations. China has also introduced cyberspace security regulation to govern cross border transfer of both commercial and research related data.

the appropriate use of genetic data for research and clinical translation, the initial lack of detailed implementation regulations has largely stalled many international collaborations. China has also introduced cyberspace security regulation to govern cross border transfer of both commercial and research related data. Similarly, with emerging geopolitical tensions, some western countries have imposed restrictions on collaboration with research institutions in China. Again, the lack of clear guidelines has resulted in confusion and widely varying interpretation, with some UK institutions, for example, now requiring application of specific data export licences for collaboration with certain Chinese institutions and enterprises. Unlike UK Biobank, the US All of Us cohort does not permit data sharing with researchers from China. These measures may have been introduced to tackle legitimate concerns about highly sensitive national security related technologies, but they also pose major obstacles to other politically neutral scientific collaboration that would benefit population worldwide.

does not permit data sharing with researchers from China. These measures may have been introduced to tackle legitimate concerns about highly sensitive national security related technologies, but they also pose major obstacles to other politically neutral scientific collaboration that would benefit population worldwide. There are also legitimate concerns about sharing personal data. The conventional approach to deliver datasets to users, even with anonymisation, still entails the risk of untraceable secondary sharing in breach of data sharing agreements, especially in places with weak contractual culture. There is also growing concern about confidential data leakage or participant re-identification with the use of large language models in processing and analysing cohort data. However, increasingly, these concerns can be minimised by using secure cloud based analysis platforms (eg, UK Biobank Research Analysis Platform (RAP)) that can avoid direct data transfer and document a clear audit trail of data utilisation. Federated analysis is also increasingly common in multicohort collaborations, with individual cohorts analysing their data locally and integrating results using meta-analysis, thereby avoiding data transfer.31

ysis Platform (RAP)) that can avoid direct data transfer and document a clear audit trail of data utilisation. Federated analysis is also increasingly common in multicohort collaborations, with individual cohorts analysing their data locally and integrating results using meta-analysis, thereby avoiding data transfer.31 The emerging major cut in medical research and foreign aid funding is also threatening international collaboration.32 Behind this is a worrying global trend of nationalism that prioritises domestic interest over international collaboration, posing considerable challenges to the funding and sustainability of major cohorts founded on international collaboration. In China, for example, there is a growing trend to discount publications based on overseas databases (eg, UK Biobank) when researchers are evaluated for funding applications—something yet to be seen elsewhere. In the UK, the Wellcome Trust no longer funds research projects in China owing to the lack of a representative office in China, as required by the recently established Chinese law governing overseas non-governmental organisation operations.

hers are evaluated for funding applications—something yet to be seen elsewhere. In the UK, the Wellcome Trust no longer funds research projects in China owing to the lack of a representative office in China, as required by the recently established Chinese law governing overseas non-governmental organisation operations. There are also many operational and data harmonisation challenges. For example, data collected under different sociocultural contexts may not be directly comparable (eg, Townsend Deprivation Index in UK Biobank versus education in the CKB, as best measures of socioeconomic status). Moreover, the quality and standards of disease diagnoses may differ across cohorts. In China, while most of the stroke diagnoses were based on brain imaging (about 95% in CKB), only about 10% of cases of chronic obstructive pulmonary disease were diagnosed by spirometry, even though a large proportion of hospital inpatient cases were found to be valid based on clinical assessment.33 34 Furthermore, laboratory results may not always be comparable, owing to differences in assay technologies used (eg, antibody versus mass spectrometry based proteomics assays). These pose major challenges for data harmonisation and combined analyses. To tackle these issues, major efforts are now underway to standardise and harmonise core data (eg, variable names, definitions, and values used) collected across major cohorts as a part of new initiatives by IHCC (https://globalgenomics.org/ihcc), which could serve as a template for future new cohorts.

mbined analyses. To tackle these issues, major efforts are now underway to standardise and harmonise core data (eg, variable names, definitions, and values used) collected across major cohorts as a part of new initiatives by IHCC (https://globalgenomics.org/ihcc), which could serve as a template for future new cohorts. In summary, large population based cohort studies are now at the forefront of population health and biomedical research, with great potential to transform the development of precision health. Despite the uncertainties and potential challenges, there are ample opportunities and value in international collaboration including Chinese cohorts and researchers ( box 1). In future, the development of cloud based research analysis platform (as already commenced in the CKB to facilitate sharing of recently completed cohort-wide whole genome sequence data) represents a way forward, which will further maximise the scientific potential of the data generated while protecting the integrity, confidentiality, and security in accordance with national and international regulations. Meanwhile, we must continue to facilitate international collaboration by enhancing knowledge transfer, attracting inward investment, improving study quality, reducing geographical barriers, and maximising the value of study resources. Capture diverse risk exposures, disease patterns, genetic architecture, and social context of global relevance Contribution to international monitoring efforts on exposure and disease patterns Sharing of expertise Identification of best practices

In summary, large population based cohort studies are now at the forefront of population health and biomedical research, with great potential to transform the development of precision health. Despite the uncertainties and potential challenges, there are ample opportunities and value in international collaboration including Chinese cohorts and researchers ( box 1). In future, the development of cloud based research analysis platform (as already commenced in the CKB to facilitate sharing of recently completed cohort-wide whole genome sequence data) represents a way forward, which will further maximise the scientific potential of the data generated while protecting the integrity, confidentiality, and security in accordance with national and international regulations. Meanwhile, we must continue to facilitate international collaboration by enhancing knowledge transfer, attracting inward investment, improving study quality, reducing geographical barriers, and maximising the value of study resources. Capture diverse risk exposures, disease patterns, genetic architecture, and social context of global relevance Contribution to international monitoring efforts on exposure and disease patterns Sharing of expertise Identification of best practices Effective pooling of funding from diverse sources Language and culture barriers, including differences in research assessment culture Different ethical standards Geopolitical tension Data security and confidentiality concerns Nationalism and funding priorities Comparability of variables and laboratory results

Identification of best practices Effective pooling of funding from diverse sources Language and culture barriers, including differences in research assessment culture Different ethical standards Geopolitical tension Data security and confidentiality concerns Nationalism and funding priorities Comparability of variables and laboratory results Large cohort studies in China can capture unique characteristics in risk exposures, disease patterns, and genetic architecture that will lead to novel discoveries and improvement in disease prevention and treatment International collaboration on China based cohort studies, which may cover different stages of study lifecycle, can help to improve study quality, attract external funding, and maximise study potential and global impact Important challenges for international collaboration include cultural differences, data comparability, funding priorities, regulatory constraints, and geopolitical tensions Good governance framework, openness, transparency, cooperation, and data sharing are key elements to the success of collaborative cohort studies

Capture diverse risk exposures, disease patterns, genetic architecture, and social context of global relevance Contribution to international monitoring efforts on exposure and disease patterns Sharing of expertise Identification of best practices Effective pooling of funding from diverse sources

Language and culture barriers, including differences in research assessment culture Different ethical standards Geopolitical tension Data security and confidentiality concerns Nationalism and funding priorities Comparability of variables and laboratory results