rather than manufacturers’ projections, Beijing wants its own evidence on drug effectiveness; it will reshape how drugs are priced and reimbursed
Clinical trials tell you whether a drug works in ideal conditions. Real-world data (RWD) tells you whether it works on actual patients over time. Beijing has decided it needs the second kind of evidence, and is building the infrastructure to collect it at scale. Since 2013, RWD has expanded beyond the handful of pilot zones used to fast-track drug registration. It is being wired into the decisions that determine whether a drug stays on the national insurance list, at what price, and on what terms. This shift is inseparable from a broader ambition: to turn fragmented hospital data, a single-payer database covering 1.33bn enrollees, into a coherent national asset.
from data silos to national infrastructure
The PRC holds enormous volumes of RWD, but quality is uneven and there is a shortage of specialised institutions and contract research organisations able to convert that data into high-quality real-world evidence (RWE), said Sun Lixia 孙丽霞 Tigermed vice president. The ‘medical data silo’ problem is the reason why. Patients move between hospitals without a referral system, electronic health records and imaging reports cannot flow reliably between them, and no single institution holds a complete record. Fragmented clinical records are only half of it. A second, related problem lies with insurance data. Researchers at Nanjing University of Science and Technology point to unclear property rights and undefined circulation rules as the key bottlenecks in turning ‘held’ insurance data into usable data.
Two rounds of policy have tried to resolve this. A 2022 framework, the 20 articles on data management, separated data holding, processing and operating rights to begin resolving ownership. The National Data Administration’s data elements x three-year action plan (2024–26) then called for electronic medical record sharing across hospitals, with insurance data as the connective tissue linking records across platforms, an ambition the 14th 5-year plan for national health system informatisation also set out. Neither has closed the underlying silos: the 2022 network security measures for medical institutions, which assign data security responsibility within each institution, concluded at the end of 2025 with the problem unresolved.
RWD policy is the applied expression of this push. Beijing’s first formal pathway was approved by the State Council 2013 at the Boao Lecheng pilot zone in Hainan, where clinical data from overseas-approved products can be used and collected to support NMPA (National Medical Products Administration) when it evaluates registration or reimbursement decisions. In July 2024, the same rights were formalised for the Greater Bay Area. The real-world evaluation guidelines, together with the 79 hospitals named as ‘trusted evaluation points’ in January 2026, are the latest nodes in this network that formalises real-world evaluation as a standing national infrastructure. The data shows which drugs improve patients’ lives and which medical services offer better value, noted Ma Shaohua 马少华 Peking University Cancer Hospital party secretary.
money, evidence, politics
There are three main structural pressures to explain the big strides of RWD policies from 2025 onwards.
First, the growing strain on the national insurance fund. By end-2025, 23 percent of the PRC population were aged 60 or over. Drug spending is expected to increase as the pool of working-age contributors shrinks. As the PRC puts more emphasis on drugs for seniors, children and people with rare diseases, RCT (randomised control trial) is no longer the best trial method for ever more specific clinical questions. With RWD, the state can check whether the drugs it funds are delivering results. Zhu Yongfeng 朱永峰 NHSA (National Health Security Administration) office director frames the goal as ‘value-based purchasing’: paying for outcomes, not approvals.
Second, price talks rest almost entirely on trial data and manufacturer-submitted projections, with no independent evidence of real clinical performance. NHSA’s account of how the undisclosed price ceiling is calculated shows that assessors primarily rely on pharmacoeconomic evidence, international prices, and competitor comparisons, most of which are derived from materials submitted by the manufacturer. NHSA acknowledged that some drugs lack head-to-head comparative evidence, with experts relying instead on manufacturers’ indirect comparison data. In 2021, the NHSA commissioned Xuanwu Hospital to begin real-world studies of centrally procured generics, an early signal that Beijing wanted its own independent evidence base rather than relying on data from manufacturers. The Chinese Journal of Pharmacoepidemiology (2025, Vol. 34, Issue 6: 690–701) also urged the field to pursue multidisciplinary, multi-institutional, cross-regional cooperation to build open, sustainable data sources.
Lastly, in early 2025, 20 medical experts submitted a proposal at the Shanghai CPPCC session voicing concerns about low generic drug quality and a lack of access to imported original drugs. Led by Zheng Minhua 郑民华 Ruijin Hospital director of surgery, the disputes sharpened political urgency. Although the state’s public response seemed muted, large-scale RWD and real-world evaluation of volume-based procurement drugs were underway.
insurance integration
Pilot programs for real-world value evaluation of medical insurance launched in September 2025, targeting a national framework by end-2027. The framework applies RWE at three points: before reimbursement, where companies are encouraged to use the one-year window after trials end to build evidence; at listing, where conditionally approved drugs must commit to real-world evaluation for the NHSA to weigh in its review; and at renewal, where the NHSA checks whether a drug’s real-world benefit and value for money have matched the projections made when it was first admitted. Shi Zihai 施子海 NHSA vice director, described the goal as using RWD to objectively identify the clinical value of medical products.
The 2025 NRDL (National Reimbursement Drug List) came with a separate commercial insurance drug catalogue for innovative drugs too expensive for basic state coverage. The official formula is ‘commercial insurance blazes the trail; basic insurance follows.’ This allows private insurers to cover a wealthier group of patients and to pay closer to those new drugs’ market price. The drugs build up track records in real use, and that evidence can support manufacturers’ later bid to enter basic insurance at a negotiated price. In its first year, the catalogue took in 19 drugs from 18 companies, including five CAR-T cancer therapies priced around C￥1 million (~US$147,000, at ￥6.78/US$1) each.
Drugs on the commercial route are reserved for drugs that basic insurance cannot yet afford. As Huang Xinyu 黄心宇 NHSA Medical Service Administration Department director put it, some drugs enter the market at a high price that basic insurance struggles to cover, and the commercial catalogue offers a transition that protects innovation while still reaching patients who need the drug. However, commercial insurance buyers are wealthier and more urban than the population that basic insurance ultimately covers, so the early RWD may not exactly reflect how a drug performs in poorer or rural patients. Whether the NHSA will adjust for that gap when this evidence feeds into basic insurance pricing is not yet spelled out.
practical hurdles
Data quality is fundamental, but three further obstacles are structural.
First, the privacy law data dilemma. The Personal Information Protection Law (PIPL) treats medical data as sensitive information and contains no scientific research exemption; hence, researchers must obtain fresh patient consent for each new use. Anonymised data can be lifted outside the law’s scope, but it will not be very useful for real-world evaluation; de-identified data can be used for real-world tracking but faces strict PIPL restrictions.
Second, manufacturer influence. Hu Xiaohua 胡晓华 and Zhang Xiyu 张曦予 East and Concord Partners lawyers, have warned that company-funded studies can shade into bribery when payments to doctors are dressed up as research, or into illegal purchase of medical records when patient data is pulled from hospital systems. The proposed fix is procedural: move core evaluation to trusted hospital-based institutions rather than the drug sponsor, so a company’s stake in the product cannot shape the findings. That fix is not yet enforceable; current rules rely on the sponsor’s intent rather than a firewall between company funding and the evaluation itself.
Finally, there is a timing mismatch no guideline can easily fix. Credible RWE takes years to build, but the insurance list needs to be updated annually. The 2026 negotiation reform pushed medical insurance review before a drug is formally approved, and as Guo Xinfeng 郭新峰 Nanjing Evidence-Based Biotech observed, the competitive focus for drug companies is shifting toward clinical value validation, RWE accumulation, and long-term payment fit. Yet the September 2025 pilot never addressed how evidence maturity will be weighed against renewal deadlines when the two collide. A system that wants mature evidence before renewal but cannot wait for it will either decide too early or slow the cycle down, and both outcomes carry costs for patients and manufacturers.
data infrastructure is key
For all its gaps, this is the most serious attempt the PRC has made to put its insurance decisions on an evidence footing rather than taking manufacturers’ projections on trust. The framework is now largely built, though some of its rules are still in draft. What is not yet settled is whether the plumbing beneath it, the data that can actually move between hospitals, the PIPL that governs how RWD can be reused, the firewall that keeps company-funded studies honest, and the evidence that matures faster than a renewal deadline can bear the weight the system is putting on it. The first test comes as the pilot’s evaluation results begin to land from the end of 2026 and start feeding into reimbursement decisions, and that is the moment to watch.
the builders
Fu Chaoqi | 付超奇 NHSA Big Data Centre director
This single-payer scale is what makes the database valuable for pharmaceutical evaluation, as it draws on actual clinical and settlement records rather than controlled-trial conditions. Once rigorously de-identified, this population-level health data can provide drugmakers with high-quality real-world research support, substantially shortening patient recruitment and clinical trial timelines, lowering R&D costs, and accelerating the path to market for new drugs. Fu proposed building a ‘Personal Medical-Insurance Cloud’ on top of the existing national insurance data platform, which he argues could become the world’s largest real-world research sample library, though for now it remains a proposal moving into early pilots.
But the data must never leave the system. Instead of handing raw health records to AI companies, the AI itself is brought inside the medical-insurance platform, where it runs and trains on the data in place, and only the results are released. Disciplined data governance is what converts scale into evidence, and what turns PRC’s data advantage into a lasting competitive edge in pharmaceutical innovation.
Director of the Big Data Centre of NHSA, Fu previously served as deputy director and then director in the NHSA’s General Office. He has led the rollout of the national medical-insurance information platform and the flagship National Smart Healthcare Insurance Competition, and has championed building a ‘Personal Medical-Insurance Cloud’ to strengthen competitiveness in PRC health-focused AI agents. Fu holds a bachelor’s and master’s in engineering from Peking University.
Ding Yilei | 丁一磊 NHSA pharmaceutical prices and procurement bidding department director
The National Health Security Administration is reshaping how drug value is judged, shifting from trial data towards RWE drawn from clinical practice. The structural fix is to put the hospital, not the sponsor, in charge. Ding identifies company-initiated studies as the ones the public and the insurance system trust least, because the firm has a direct stake in the product and the results. Since hospitals are the main channel for producing real-world data, trusted evaluation institutions must take on the core evaluation work and generate credible findings. That is the principle. The procedure lags behind it. Until they are finalised, the independence of company-funded evidence rests on intent rather than an enforceable firewall.
Medicine pricing and centralised procurement director at NHSA, Ding is best known as a leading architect and public face of the PRC volume-based procurement program, which has driven steep cuts to drug and device prices, most prominently the landmark artificial-joint tender that lowered prices by some 82 percent. He frames procurement as ‘empowering’ the pharmaceutical industry and supporting innovation. He has also helped build PRC’s RWE system to evaluate the overall value of drugs.
Chen Yi 陈怡 | Tsinghua University Institute for Hospital Management professor
The insurance authority and the drug regulator approach real-world evidence from fundamentally different vantage points. Regulators ask whether a drug is safe and effective once marketed. Insurance authorities ask whether a drug delivers value that justifies sustained public funding, and at what price. That distinction matters most for drugs granted conditional approval, which often reach the market with insufficient clinical evidence and high per-course costs. Without real-world effectiveness data to substantiate their value, the payment risk to the insurance fund continues to accumulate. Value assessment should therefore be dynamic, running from initial access through to each renewal cycle, with real-world data used to balance clinical benefit against fund affordability and arrive at a more defensible negotiated price. However, not all real-world data becomes evidence once analysed. High-quality data and sound statistical methodology are the core of effective RWD gathering and real-world evaluation.
A Tsinghua University’s Institute for Hospital Management professor, Chen, previously held roles including economist at the World Bank, senior economic officer at the US Department of Labour, and vice president of corporate and government affairs at Eli Lilly and Ascletis Pharma. Her research focuses on health economics, innovative drug pricing and reimbursement, health technology assessment, real-world data, and market access policy. She holds a PhD in Economics from the University of Utah and a BA in International Economics from Fudan University. She has over 20 years of experience across government, international organisations, and the pharmaceutical industry in China and the United States.
context
Jan 2026: NHSA releases six-part draft drug evaluation guidelines, and 79 hospitals sign on as national trusted evaluation points.
Dec 2025: 2025 NRDL published with the new dual-catalog pathway; 19 high-priced drugs, including five CAR-T therapies, enter the commercial catalog
Sep 2025: NHSA launches the real-world medical insurance value evaluation pilot, targeting a national framework by 2027
Jun 2025: NHSA and the NHC issue measures to support innovative drugs, formally bringing in RWD-based value evaluation
Jan 2025: Shanghai CPPCC experts raise generic drug quality concerns, sharpening political pressure for independent evidence
Jul 2024: real-world study rights formalised in the Greater Bay Area
Jan 2024: National Data Administration and 16 agencies publish the data elements x three-year action plan (2024-26, naming healthcare a priority for unlocking data value
Jan 2020: NMPA issues PRC’s first real-world evidence guideline for drug development and review, putting RWE formally into the regulatory system
2013: Boao Lecheng pilot zone established in Hainan, the first channel for using overseas clinical data to support registration in PRC
 | A guest post by
|