Technical Guidelines For The Use Of Real World Data In Clinical Evaluation Of Medical Devices (Trial)


Announcement on the Publication of the Technical
Guidelines for the Use of Real-World Data in Clinical
Evaluation of Medical Devices (Trial)
(No. 77 in 2020)
In order to standardize and guide the application of real-world data in clinical evaluation of medical
devices, according to the work arrangement of Scientific Action Plan on Regulation of Drugs in China
initiated by the NMPA, the NMPA has organized the formulating of the Technical Guidelines for the Use
of Real-World Data in Clinical Evaluation of Medical Devices (Trial), which is hereby issued.
It is hereby notified.
Appendix: Technical Guidelines for the Use of Real-World Data in Clinical Evaluation of Medical
Devices (Trial)
National Medical Products Administration (NMPA)
November 24, 2020
Technical Guidelines for the Use of Real-World Data in
Clinical Evaluation of Medical Devices (Trial)
The Guidelines aims to preliminarily normalize and reasonably guide the application of real-world
data in clinical evaluation of medical devices, provide technical guidance for the application and
registration of real-world data of medical devices by applicants and the technical review of such clinical
data by regulatory departments. Medical devices mentioned in the Guidelines include in-vitro diagnostic
The Guidelines is technical guidance document for the use of applicants and review personnel, which
is not enforced as laws and regulations. The Guidelines shall be used in accordance with the relevant laws
and regulations. Real-world data and real-world study are in the rapid development stage. The Guidelines is
formulated on the basis of the existing cognitive level and needs to be continuously improved and revised
in accordance with scientific development.
I. Overview
(I) Real-world data and evidence
The real-world data as described in the Guidelines refer to a variety of data collected from multiple
sources related to patients’ health status and/or routine diagnosis and treatment, in addition to traditional
clinical trials.
Researches centered on relevant scientific issues, applying multidisciplinary methods and techniques
such as epidemiology, biostatistics and evidence-based medicine in a comprehensive way, and using
real-world data, are collectively referred to as real-world study. The real world study carries out prospective
or retrospective studies by systematical collection of real-world data and application of reasonable design
and analysis methods.
Real-world evidence refers to clinical evidence related to use of medical devices and risks/benefits
formed through the analysis of real-world data, which may be valid scientific evidence for regulatory
decision-making. The quality of the data may vary greatly due to different sources of real-world data, and
not all real-world data can produce valid real-world evidence.
(II) Advantages and limitations of real-world study
In general, compared with traditional clinical trials, real-world studies are carried out in a real
environment, with relatively fewer restrictions on enrolled subjects. The sample size may be larger, and
long-term clinical outcomes are more likely to be obtained, and the study results may be better
extrapolated. Real-world study can apply a variety of data, such as hospital records data, enrollment data,
medical insurance data, and etc. Real-world studies can also be used to observe rare serious adverse events,
answer questions about the diagnosis and treatment of rare diseases, and evaluate differences of clinical
outcomes among different populations, medical environment and method of application.
The limitations of real-world study include but are not limited to: there are many sources of real-world
data and the quality of data remains to be evaluated; real-world study are often involved with multiple
biases and confounders (including selection bias, information bias, confounding, etc.), and there may be
challenges for study conclusions.
II. Common real-world data sources
Common real-world data include, but are not limited to registration data, hospital medical records,
regional health care data, medical insurance data, health archives, public monitoring data, self-reported
patient data, and data generated by mobile devices. In addition to the above, real world data can also
include data generated in the production, sale, transportation, storage, installation, use, maintenance,
delisting, disposal and other processes of medical devices (such as acceptance report, maintenance report,
user feedback, use environment, calibration records, running log, image raw data, etc.).
According to their sources and characteristics, real-world data includes but is not limited to the
following cases:
(I) Generated from the process of health medical service delivery and payment and generated for
management purposes, such as hospital electronic medical record data, medical insurance data, health
records, etc.
(II) Based on the research purpose when the database was established, the unified data standard and
data collection model are established to form and establish data resources in the routine clinical practice,
such as device registration data and etc.
III. Quality evaluation of real-world data
Good quality of real-world data is the basis of real-world study, which directly affects the strength of
evidence generated in real-world study. In the quality evaluation of real-world data, the relevance and
reliability of data should be paid attention to on the basis of following ethical principles, conforming to
regulatory requirements and ensuring data security. The relevance of data refers to whether it can
adequately answer clinical questions related to the research purpose, including whether it completely
covers the data of research population, whether a relatively unified or standardized intervention/exposure
can be formed, whether a comparable control can be set, and whether the outcome variables and
measurement results required by the study are included and whether data related to confounders can be
obtained. The reliability of data refers to the accuracy of data collection, including determining the
collection scope and collection variables before collection, formulating a data dictionary and specifying
collection methods, data transfer mode and storage medium format of data collected and etc., so as to fully
guarantee the authenticity and integrity of data. To evaluate the quality of real-world data, specific
considerations can be taken from the following aspects: