Decentralised, pragmatic and real-world elements in trials
An interventional trial can move activities away from the investigator's location, run parts of itself inside usual clinical practice, and take some of its data from sources that were never collected for it. Real-world data can supply an external control group or support a hybrid design, though they do not suit every trial. Using these methodologies means knowing what each one is, planning around the participants' needs, engaging the people who will make it work, showing that real-world data are fit for purpose and that the sponsor and inspectors can reach them, protecting participants' privacy, and keeping oversight adequate as the arrangements multiply.
Good clinical practice across trial methodologies
Good clinical practice, as ICH E6(R3)'s principles and Annex 1 describe it, applies across clinical trial types, designs and settings, and it remains relevant when a trial uses various operational approaches, design elements and data sources, which the guideline collectively also calls methodologies. As trial designs evolve and technology advances, the application of good clinical practice should be tailored to the trial design and the technologies used. In this context a proportionate risk-based approach remains relevant and supports the use of evolving trial designs and technologies without compromising the rights, safety and well-being of participants and the reliability of the trial results. A quality by design approach should be used so that the methodologies adopted and implemented are fit for purpose and the quality and amount of information generated or collected is sufficient to support good decision making.
ICH E6(R3) Annex 2 provides additional good clinical practice considerations focusing on trials that incorporate decentralised elements, pragmatic elements or real-world data, and a trial may incorporate one or more of these methodologies. The annex is not intended to cover every methodology, and its considerations may be applied to methodologies that emerge later, provided they are applied in accordance with local regulatory requirements. The annex should not be read as an endorsement of any specific methodology, and it should be read in conjunction with the principles and Annex 1.
Decentralised elements
In ICH E6(R3) Annex 2, decentralised elements are the trial-related activities conducted outside the investigator's location. They may include trial visits and protocol-related procedures conducted at the participant's home, at local healthcare centres or in mobile medical units, and they encompass remote interactions such as video calls and the use of digital health technologies to conduct visits, perform procedures and collect data.
Digital health technologies is a broad term for the wide range of technologies supporting healthcare that collect, measure or monitor health-related data from participants or patients, such as mobile applications, wearables and sensors. Depending on the context, the term may refer to technologies used specifically in trials or to those used within broader healthcare settings, and a digital health technology used to acquire participant data for a trial is a data acquisition tool as the ICH E6(R3) glossary defines that term.
As of 2023, the most widely used definition of a decentralised trial was the FDA's: a clinical investigation in which some or all of the trial-related activities occur at a location separate from the investigator's location. The Decentralized Trials & Research Alliance glossary gives a more recent and potentially more specific definition, which expands on the FDA's to clarify that a decentralised trial uses technology, processes or services that create the opportunity to reduce or eliminate the need for participants to physically visit a traditional research site. A traditional trial with onsite visits does not automatically become a decentralised trial because it includes a device considered a decentralised component, such as a wearable or connected device: the components need to materially reduce or eliminate the need for onsite visits, not just provide an additional opportunity to collect data.
Decentralised trials are broadly subclassified as full or hybrid. A full decentralised trial does not require participants to go to trial sites at all: all trial-related activities are done at the participant's home or in another local setting. It is executed through telemedicine, mobile or local healthcare providers or mobile technologies, and so is not bound by the geographic limitations that affect traditional trials. A hybrid decentralised trial blends onsite and offsite activities, so hybrid trials can cover a range of configurations. A hybrid trial partially eliminates the requirement for participants to visit a physical trial site to perform a protocol-required event that may traditionally have taken place onsite.
As of 2023, there was a growing demand for clinical trial approaches that reduce the burden on participants and increase the recruitment and retention of a more equitable participant population. Decentralised trials are not new: as of 2023, Pfizer's REMOTE trial had started 12 years earlier. Research after the COVID-19 pandemic showed that trials using decentralised components recovered faster from its impact than trials of traditional design, and analyses have shown that using decentralised components gives substantial cost savings and enhances participation.
Pragmatic elements
Pragmatic elements integrate aspects of usual clinical practice into the design and conduct of the trial, usual clinical practice meaning the established day-to-day processes and procedures that healthcare professionals apply when caring for patients. They may include protocols whose trial processes and eligibility criteria reflect usual clinical practice in that patient population and reflect the settings where usual care is provided, such as hospitals, healthcare centres, primary physicians and local healthcare providers. They also include streamlined data collection, such as capturing only the minimum necessary data by aligning with usual clinical practice, or capturing data only at the timings of routine care, to minimise the complexity of trial conduct. Technologies such as digital health technologies, and automated data exports from electronic health records, can be used to help such streamlined data collection.
Real-world data and secondary use of data
Real-world data incorporated in an interventional trial are data relating to patient health status collected from a variety of sources outside clinical trials, such as electronic health records, registries and claims databases.
Data are often used in a trial in one of two ways: as primary data collection, when they are collected specifically for the trial, or as secondary use, when they are repurposed from existing sources collected for other purposes. Both can contribute meaningfully to the trial objectives, but each requires different considerations for data quality, data governance and regulatory acceptability. Primary data collection gathers data prospectively in accordance with the protocol, typically through trial-specific procedures such as clinical assessments, laboratory tests or participant questionnaires. Secondary use incorporates into the trial data originally collected for other purposes, and where those data reflect patient experiences, outcomes or healthcare delivery outside a trial setting they are real-world data. Such data can serve various roles in a trial, including but not limited to endpoint or outcome data or an external control, and when they are used it is essential to ensure that they are fit for their intended purpose in the trial.
External control groups
An external control group consists of subjects selected from an external source, such as existing clinical trial data or electronic health records. As of 2025, an increasingly attractive use of real-world data is within clinical development where, for example, a traditional control group would be impractical or unethical, and there they serve as the source of an external control group, giving a structured comparator where otherwise there would be none. An external control group can be considered for diseases where patient enrolment may be problematic, such as rare diseases, or where randomising patients to the control group may be unethical, such as cancer. Opportunities to incorporate real-world data in a non-randomised single-arm trial include external controls where the disease evaluation criteria are well established; historical records of vital signs collected from trial participants in different geographies, pooled or stratified by a prespecified participant characteristic; and comparison data sets where a placebo or no-treatment arm is not ethical or feasible.
The biggest challenge in using an external control is controlling bias. As of 2020, the FDA suggested the use of external controls only under certain conditions, for example when distinct treatment effects are expected between the test and external control groups. The external control should be selected from the data sources most appropriate to the study's purpose and should align as closely as possible with the study's eligibility criteria, to minimise potential confounding and selection bias. Comparison between the two groups also depends on whether similar endpoint assessments are available in both, and here external controls drawn from existing clinical studies with similar purposes may be more applicable than those from electronic health records or registries.
Hybrid designs incorporating real-world data
Hybrid study designs incorporating real-world data or evidence have been applied in clinical trials for regulatory decision-making: in exploratory new drug submissions that use real-world evidence to gain insight into clinical outcome or safety data, as methods for single-arm trials that require external controls, and in trials that need real-world evidence to satisfy post-marketing requirements for additional safety or effectiveness data to support a regulatory decision. Hybrid designs better equipped to capture long-term outcomes should harness methodologies such as decentralisation, for example trained nurses; direct-to-patient approaches, for example wearables; and databases, for example registries and claims.
Whether and how to design with real-world data
Real-world data may not be applicable to every type of trial design. Incorporating them as a design element often begins with a multifaceted discussion of considerations that include whether the treatment methodology is routine enough and the therapeutic area established enough to gather sufficient real-world data before the study starts; whether the available data are of sufficient quality for statistical comparison with data from a more traditional longitudinal study; in rare and ultra-rare diseases, whether the volume of available data is sufficient; and, where historical controls are used, whether clinical practice guidelines and data collection methods have remained consistent enough for the data to be an accurate comparator. As a design element, real-world data can broaden perspective and potentially improve regulatory decision-making. As of 2021, the FDA saw immense potential in using real-world data in designing clinical trials.
Heterogeneity in the data is the intrinsic underlying issue in most data sources, and it should be thought through in the study design and statistical analysis, in consultation with biostatistical colleagues. Study design is always a collaboration with colleagues in other functions such as biostatistics and medical affairs. In the view of one group of authors, medical writers can and should be influencers in that collaboration, raising awareness of the potential of big data in study design so that all stakeholders consider its practical use.
Participants' needs
When planning and implementing decentralised and pragmatic elements, it is important to consider the needs of the participants, including the availability of and familiarity with the technologies, and whether any burden to participants needs to be balanced with the use of such elements.
Ethics committees reviewing decentralised trials raise questions of equity and digital literacy, such as whether participants will be provided with a device, whether they know how to use it, whether they can access the internet, and whether they have the option of using a facility to access the programme. One speaker suggested that offering participants options is good practice, so that participation is not inadvertently hindered.
Engagement with interested parties
Engagement with relevant interested parties is particularly important when a trial uses various operational approaches, design elements and data sources. Their early involvement may be valuable in refining research questions, selecting meaningful interventions and outcomes, and shaping the trial design, procedures and communication strategies. Communication with them may be undertaken in various ways, taking into account ICH E8(R1).
Engaging patients, patient advocacy groups and their communities, as appropriate, can help ensure the successful integration and implementation of the methodologies. Involving patients early in the design may help refine it, including ensuring the suitability of the digital health technologies used in trials with decentralised elements, and help address issues related to the logistics of participation. This engagement may also highlight areas where additional training or support may be needed, such as digital literacy, physical ability, or a lack of access to technology that may require alternative approaches, specialised training or the provision of technology.
Engaging healthcare professionals or investigators early in the design of a trial that incorporates such methodologies is critical for its successful implementation and conduct. Early engagement can help address issues related to the infrastructure needed to conduct the trial, develop protocols that incorporate healthcare professionals' usual clinical practice when appropriate and allow the integration of data generated from usual clinical practice when such data are fit for purpose, and identify areas where training or support is needed.
Sponsors are encouraged to engage with regulatory authorities early, especially when designing and planning trials that employ various operational approaches, including technological tools, incorporate complex design elements or use real-world data sources. Early engagement will help address the appropriateness of using such methodologies in the trial design and allow timely identification of challenges and strategies for resolution.
Fitness for purpose of real-world data
Sponsors are ultimately responsible for ensuring that real-world data used in a trial are managed in a manner that ensures participant protection and the reliability of the trial results. This includes ensuring that the data are fit for purpose, that appropriate access and oversight are in place, and that their use complies with applicable privacy and data protection requirements, including consent and permission.
Annex 2's considerations for using a variety of data sources should be read in conjunction with section 3.16.1 of Annex 1 and with its section 4, on data governance for the investigator and sponsor.
The sponsor should ensure the fitness for purpose of real-world data, which can be described by their reliability and relevance. Reliability includes accuracy, completeness, provenance and traceability; relevance includes the availability of the key data elements, such as exposure, outcomes and covariates, needed to answer the specific trial question with the specific method. Attention should be given to variability in the data source, potential bias, generalisability and traceability. The sponsor should implement procedures that support adequate documentation, data linkage and quality control, while accounting for differences in how real-world data are structured, collected and managed across sources. In selecting real-world data sources, sponsors should assess whether the data are appropriate to support the specific trial endpoints and objectives and weigh the potential benefits against the limitations. These considerations should be reflected in trial planning, protocol design and oversight processes.
A variety of real-world data sources may be used in trials, including electronic health records, administrative claims data and registries. Because these data are typically generated outside the trial, sponsors generally do not have control over how they were originally collected, recorded or curated. Sponsors should assess whether a source is fit for purpose for its intended use in the trial and apply a risk-proportionate approach to the measures used to ensure appropriate data quality, integrity and governance, considering the importance of the data to the trial objectives and the potential impact of data limitations on the results.
The considerations that may inform whether a given source is fit for purpose include, but are not limited to:
- The potential variability of data formats and structure, such as different terminologies or standards, across sources.
- The lack of standardised timing of data collection and procedures, since the timing and frequency of clinical assessments in real-world data are based on clinical practice and may have been influenced by the participant's clinical status, so the protocol schedule may not match what is available.
- Comparability between the trial data and a control group using real-world data, such as comparability of baseline factors, minimisation of potential selection bias through pre-planned patient selection, and consistency in clinical evaluation methods.
- Missing data, for example because participants moved to a different healthcare system, and intercurrent events that may be difficult to capture or ascertain, such as discontinuation of treatment or an additional or alternative therapy not captured in the electronic health record, for which the annex refers to ICH E9(R1).
- The overall quality of data collected in clinical practice or registries, including operational processes, database structure and the consistency of vocabularies and coding systems.
- The de-identification methodologies, such as pseudonymisation, used to protect the privacy and confidentiality of personal information.
- The fitness-for-purpose assessment of the systems and tools used to collect and acquire the data, such as registries and digital health technologies, including their validation status as appropriate.
Multiple data sources might need to be linked to support the reliability of real-world data, including their completeness, as when data from electronic health records and claims databases are linked, or a source is linked to a mortality database to confirm outcomes. When data are linked, accurate matching to the individual should be assured, and the sponsor should ensure adequate measures to protect both data privacy and the reliability of the trial results. Linkage should be pre-specified in the protocol or protocol-related documents, including how inconsistencies between sources will be resolved in a systematic and consistent manner.
Access to individual-level data and source records
Individual-level data, pseudonymised for trial use, are datasets or listings capturing data for each individual, usually extracted from source records and structured for statistical analysis and reporting. Pseudonymised data are processed so that a specific individual cannot be identified without collating the data with additional information. Data in source records are the data in the original records, such as patient medical records, laboratory reports, imaging files and physician notes, from which individual-level data are derived. Source records are typically not pseudonymised and contain patient-identifying information as well as other information subject to data privacy and protection requirements.
To fulfil their responsibility for ensuring that real-world data are fit for purpose and ultimately support the reliability of the trial results, sponsors should be able to access individual-level data, whether generated by themselves or by other entities including service providers, and data in source records, as necessary and in accordance with applicable regulatory requirements. Sponsor access may be necessary, for example, to implement quality assurance and control. For inspection purposes, regulatory authorities may require access to individual-level data and data in source records in accordance with applicable regulatory requirements.
Where real-world data are owned and controlled by entities other than the sponsor, such as healthcare institutions, registry holders or insurers, the sponsor should ensure appropriate arrangements are in place with them to enable access to and use of the data for trial purposes. These arrangements should include sponsor access to data in source records for quality assurance and quality control in accordance with applicable regulatory requirements, and they should allow regulatory authorities to access individual-level data and source records to support inspections. The arrangements should include documented agreements that define how data will be shared and protected in accordance with applicable regulatory requirements and the institutional policies of each party.
The extent of sponsor access to individual-level data and data in source records should be proportionate to the criticality of the data. For use cases of higher criticality, such as real-world data supporting key efficacy and safety endpoints, simply assessing the fitness for purpose of the source may be insufficient to ensure the reliability of the data, so it may be necessary for sponsors, for example, to confirm whether a clinical event occurred, was assessed or was documented. In such cases sponsors should ensure access to source records to support their quality assurance and quality control, and where such access is not feasible because of constraints set by the entities that own or control the data, they should consult the relevant regulatory authorities. For use cases of lower criticality, such as exploratory endpoints, system- and process-level assessments may be sufficient, and the protocol or protocol-related documents should then describe the fitness for purpose of the data, including any relevant limitations and a justification of the assessment approach. Regardless of which entity owns or controls the data or undertakes the processing steps, sponsors remain responsible for oversight of those steps, including extraction, linkage and transformation, and for implementing quality assurance and quality control measures to ensure the data are of appropriate quality.
Privacy and data security
Remote data collection, such as remote visits, digital health technologies and extraction of data from electronic health records, requires special attention to data security vulnerabilities, including cybersecurity and data privacy. Some of the real-world data considerations may also apply to remote data collection.
Sponsors should ensure that security safeguards, including cybersecurity, are in place so that the privacy and confidentiality of participants' personal information are protected in accordance with applicable regulatory requirements. Service providers may need participants' personal information to fulfil their activities, as when, where appropriate, it is disclosed because the investigational product is shipped to participants or a home nurse is deployed. In these circumstances, sponsors and service providers should ensure that appropriate informed consent has been provided by the participant, that the personal information is protected from inadvertent disclosure, and that access to it is limited to those authorised. Sponsors should address the risk of personal information being disclosed through a data breach, for example when data from digital health technologies or real-world data are used.
Sponsor oversight
Sponsor oversight may be more complex depending on the range of methodological approaches to the trial design and conduct and the number of service providers involved. Sponsors should ensure that processes, including clear lines of communication, are in place to provide an appropriate level of oversight, so that participants' rights, safety and well-being are protected and the reliability of the trial results is ensured. Sponsor oversight involves, but is not limited to, quality control and assurance measures specifically customised to the trial and its critical to quality factors and identified risks. These measures should be implemented in proportion to the potential impact on the protection of participants or the reliability of the trial results. Sponsors should maintain appropriate oversight of service providers, including ensuring the maintenance of their essential records.