Browse the corpus

Epidemiological research in China stands at a critical juncture where cohort studies should guide the next era of health gains. But cohorts cannot be a means to merely generate mountains of dormant data; they should lead to the development of novel research methods to investigate unresolved questions. The Nature Index 2025 research leaders report9 ranks China first in contributions to high quality science journals; eight of the index’s top 10 institutions are now Chinese, a remarkable ascent in the past decade. Although China excels in fields such as materials science and physics, its clinical and epidemiological research are less impactful relative to overall research volume.9 The exponential growth in research output is, on the face of it, commendable, but a looming threat is poor quality research and exploitation of open access datasets.

excels in fields such as materials science and physics, its clinical and epidemiological research are less impactful relative to overall research volume.9 The exponential growth in research output is, on the face of it, commendable, but a looming threat is poor quality research and exploitation of open access datasets. For example, Hemani and colleagues document an exponential rise in mendelian randomisation publications, reaching around 16 000 PubMed results by mid-2025 since the formalisation of the approach in 2003. Most of the increase is accounted for by papers with authors affiliated with Chinese institutions. For one journal, this is mirrored in submissions: the proportion of all submissions coming from China is increasing substantially and rapidly, and the proportion of all submissions from China that are mendelian randomisation studies is also higher than from elsewhere.10 Many of the mendelian randomisation submissions exhibit signs of low quality, such as analysing relationships with limited justification, failing to discuss the gene-environment equivalence principle, neglecting STROBE-MR reporting guidelines, and containing methodological errors.10

s also higher than from elsewhere.10 Many of the mendelian randomisation submissions exhibit signs of low quality, such as analysing relationships with limited justification, failing to discuss the gene-environment equivalence principle, neglecting STROBE-MR reporting guidelines, and containing methodological errors.10 China seems to have two overlapping research ecosystems: one exemplifies high calibre work, with researchers making major methodological advances and developing impactful large scale cohorts; the other reflects the intense pressure faced by academics and clinicians worldwide to publish. This second context often leads to formulaic, non-hypothesis-driven analyses that mine large datasets for correlations without rigorous biological frameworks, and, sometimes facilitated by AI tools for text generation, result in questionable “discoveries.”10 For cohort studies, these same pressures must be mitigated to avoid initiating redundant “me too” studies of limited added value, ensuring resources focus on innovative, high quality designs that build on existing strengths and push forward the boundaries of biomedical research.

result in questionable “discoveries.”10 For cohort studies, these same pressures must be mitigated to avoid initiating redundant “me too” studies of limited added value, ensuring resources focus on innovative, high quality designs that build on existing strengths and push forward the boundaries of biomedical research. One thing is certain amid China’s achievements and growing pains: continued international collaboration is essential, as China is as much a part of the world as its cohort studies are integral to global health. Already a model for other nations in areas such as reducing child mortality and expanding healthcare access, China, if guided by robust, well designed cohorts, stands to benefit the entire world through its use of large scale research to drive health progress.