Selecting and Preparing Clinical Sites for the Successful Conduct of Decentralized Clinical Trial Activities–Findings From the Trials@Home RADIAL Proof‐of‐Concept Trial

Abstract

Decentralized clinical trials (DCTs) offer opportunities to improve trial accessibility, participant convenience, and efficiency, yet may pose significant operational challenges for clinical trial sites. This paper presents the operational insights gained from selecting, training, and supporting clinical sites within the RADIAL proof‐of‐concept trial, part of the Trials@Home project. RADIAL was a multicenter, low‐intervention phase IV trial comparing conventional, hybrid, and fully decentralized approaches for individuals with type 2 diabetes mellitus across six European countries. Site selection involved detailed feasibility assessments evaluating operational capabilities, recruitment potential, technological readiness, and willingness to implement decentralized elements. Despite proactive training, including ongoing support via a centralized helpdesk, sites faced initial difficulties with technology management and participant onboarding. Contractual complexities were prominent, particularly regarding clearly delineating responsibilities and data handling in agreements involving third‐party providers. Moreover, integrating third‐party services necessitated meticulous oversight strategies and continuous stakeholder coordination to ensure regulatory compliance and efficient trial management. Our experiences underscore essential considerations for future DCT implementations: proactive stakeholder alignment; tailored, timely, and ongoing training and support; intuitive technology design informed by clinical user input; robust, centralized oversight structures; and clearly defined delegation frameworks for third‐party engagements. Addressing these operational considerations will facilitate smoother transitions toward decentralized clinical research models, maximizing their potential benefits while managing associated complexities effectively—especially for clinical site staff.

Clinical trials are essential for evidence generation for evaluating the efficacy and safety of health interventions. Traditionally, these trials have relied on in‐person site visits for key activities, such as eligibility assessment, informed consent, and data collection. While this model has served clinical research in the past, it is also accompanied by significant limitations, including geographic and logistical barriers to participation, time burden for both participants and staff, and high operational costs. ¹ , ²

Decentralized clinical trials (DCTs) have emerged as an alternative or complementary approach to conventional, site‐based clinical trials, in which some or all trial activities are centered around participants, either at home or in the vicinity, rather than at a clinical site. DCTs may be fully decentralized or may combine conventional site‐based procedures with specific decentralized trial elements. Potential benefits of DCTs include the ability to enroll participants who might not be able to participate in a trial if they need to travel to a clinical research site, convenience for participants, such as reduced travel and increased autonomy, and the potential for real‐time and real‐world data collection. ³ , ⁴ , ⁵ , ⁶ Furthermore, DCTs also hold the promise of increasing efficiency and reducing trial costs. ⁷ , ⁸ Beyond the commonly cited potential benefits for participants and overall trial efficiency, the implementation of DCTs also presents challenges. These include potential risks to the validity and reliability of remotely conducted clinical assessments and collected data, increased risk of data breaches, and concerns about weakening the physician–patient relationship at clinical sites. ⁹

Next to influencing the participant experience, DCT approaches also change the way of working for clinical site staff. This evolving role of the clinical trial site is important to consider; while DCTs may alleviate some burden for the clinical site, such as frequent in‐person visits, facility constraints, and intensive on‐site monitoring, they also require sites to adapt to new technologies, remote workflows, and decentralized coordination models. ¹⁰ Consequently, questions arise: What challenges do sites face in implementing DCTs? What support or training is required to facilitate this transition? And how can site readiness and acceptance be best evaluated at the time of site selection?

The ongoing shift toward DCTs necessitates an evaluation of the traditional site selection processes, as well as the training and support requirements. Beyond conventional criteria, evaluating both site readiness and site acceptance regarding DCT elements is crucial for successful DCT implementation, followed by effective training and support during the trial. This paper is part of a series (Box 1) on the operational insights from setting up and conducting the RADIAL trial, a proof‐of‐concept trial where participants were enrolled in either a site‐based, hybrid, or fully decentralized arm in six European countries. This design allows for learnings on operational aspects of the different trial arms. In this paper, we focus on the operational model of clinical sites that interact with participants remotely, as was used in the RADIAL trial, because it is expected that most future clinical trials will have a hybrid setup and therefore still utilize physical clinical locations. We acknowledge the existence of fully decentralized or “site‐less” trials. ¹¹ However, such models are beyond the scope of this manuscript. This paper aims to present the operational experiences in selecting, training, and supporting clinical trial sites on decentralized elements as part of the RADIAL trial.

BOX 1. RADIAL paper series.

This is a series of papers in which we share learnings gathered during setup, regulatory submission, and conduct of the RADIAL proof‐of‐concept trial, the first clinical trial studying the feasibility and acceptability of decentralized trial approaches (DCTs). The papers will focus on:

1. Introduction into DCTs and the RADIAL trial

2. Regulatory advice, interactions, and approval

3. Site selection and training for DCT

4. Recruiting and consenting decentralized trial participants

5. Supply of investigational medicinal product and study materials to decentralized trial participants

6. Technological solutions for DCTS

For each of these topics, we discuss why the topic is important in DCTs, elaborate on how we set this up in the RADIAL trial, share our experiences and learnings, contrast them against what is known from literature, and provide recommendations for future trials.

RADIAL SETUP FOR SELECTING, TRAINING, AND SUPPORTING CLINICAL TRIAL SITES ON DECENTRALIZED ELEMENTS

To investigate, among other aspects, the site‐related aspects of DCTs, the Trials@Home consortium set up the RADIAL proof‐of‐concept trial. RADIAL was a three‐arm parallel‐group, open‐label, multicenter low‐intervention phase IV trial that investigated people living with type 2 diabetes mellitus (T2DM) administering Insulin Glargine 300 U/mL within a conventional, hybrid, and fully decentralized approach. The study consisted of two parallel parts, Part A and Part B (see Figure 1 ). Part A encompassed site‐based recruitment, after which participants were randomized 1:1 to either the conventional arm or the hybrid arm. Part B consisted of a fully decentralized remote arm, with decentralized recruitment and remote electronic consent. For an extensive description of the study design, study population, and rationale for RADIAL, the reader is referred to the first paper of this series “Bringing trial activities to participants – the Trials@Home RADIAL proof‐of‐concept trial investigating decentralization of trials”. ¹² This section further details the country and site selection for the RADIAL trial, as well as the setup for site training and support.

Figure 1.

Design of the RADIAL proof‐of‐concept trial.

Selection of countries and clinical trial sites

A critical component in setting up the RADIAL trial was the selection of clinical trial sites. The selection followed a structured, multi‐step process designed to ensure a broad representation across Europe, diverse healthcare systems, legal and regulatory environments, and operational feasibility in terms of recruitment and startup timelines. Initially, 21 countries were considered, based on the insights from the consortium members, the regulatory mapping carried out by the Trials@Home project ¹³ and their recruitment potential. This country list was subsequently refined by applying knock‐out criteria regarding startup timelines, the trial volume in a country, and the acceptability of online recruitment. For 17 countries, a detailed feasibility assessment was conducted at the site level to establish the final country selection for the study.

In these 17 countries, 212 clinical sites received an electronic feasibility questionnaire to (1) assess their interest in the concept of the trial, (2) identify potential operational challenges that could affect the trial’s execution, (3) estimate recruitment potential at the site, and (4) determine the site’s preference for participating in either or both Part A and Part B. Trial‐specific details were shared with the sites on how the DCT elements that were incorporated in RADIAL had been designed for the trial. In addition, sites were asked to indicate any foreseen hurdles with the implementation of DCT approaches. A short list of potential participating countries and sites was created based on this information and various other factors, including known startup timelines, recruitment potential (based on prevalence and trial data), and availability of potential investigators.

Following the feasibility assessment, selected sites were invited to a virtual site selection visit. During the selection visits the unique aspects of the RADIAL trial and its decentralized elements were discussed. There were no specific requirements to be selected for participation in Part A. In Part B, sites were asked whether they were capable to independently conduct or support a sponsor‐managed, online recruitment method to advertise the trial across the country, for example, via social media, and have the capacity to pre‐screen the patients who would apply via the RADIAL trial website.

Ultimately, in each country, multiple sites were selected for participation in Part A. For Part B, only one site was selected per country, since the Part B site was designated to recruit nationally for this Part. Following site selection, the sites were contracted for the study by agreeing on and signing the clinical trial agreement (CTA), in which the agreements between site and sponsor on site responsibilities and oversight were documented. The CTA ensured that all parties were aligned on operational, legal, ethical, and financial expectations and needed to be signed before the clinical site staff could be trained, the site initiation visit (SIV) could be performed, and the site could be initiated.

Training and support setup for selected clinical sites

A comprehensive site training and support program was developed for RADIAL to educate clinical study staff on the trial procedures and its decentralized elements. This included training during an online investigator meeting, an on‐site SIV, and the provision of an online knowledge base and helpdesk. The investigator meeting provided an overview of the design and the research questions of the trial. During the SIV, clinical research associates (CRA) trained clinical study staff on trial‐specific procedures and provided demonstrations for all devices and other DCT elements, although time was often too limited for an in‐depth, hands‐on training. The demonstrations included the setup and use of the RADIAL study smartphone application and the connection of the study devices to this application, namely a Bluetooth‐connected glucometer and a Bluetooth‐connected insulin pen adaptor cap, placed on the insulin injector pen to transmit dosing data. Clinical study staff were also trained during the SIV on using the RADIAL electronic data capture system (EDC) and central study platform, which, in addition to data entry, functioned as a workflow engine, coordinating participant journeys, task triggers (e.g., shipment of investigational medicinal product (IMP)), and study communications with logic customized for each study arm. While most were triggered at scheduled intervals, some study milestones required manual activation to initiate specific tasks. For a complete description of the system architecture, the reader is referred to the sixth paper in this series “Operationalizing Decentralized Clinical Trials: Technology Insights and Recommendations from the Trials@Home RADIAL Proof‐of‐Concept Study”. ¹⁴ For each first SIV for each country, CRAs were accompanied by a representative of the sponsor with in‐depth knowledge of the study devices and decentralized study elements. An additional online training for the home nursing services to provide more details on the home nurse selection, qualification, and activities was provided by the third‐party home nursing provider themselves after SIV and prior to site activation.

During the trial, an online knowledge base and helpdesk provided a centralized ticketing system with multilingual support and structured documentation of all training materials (both the site‐ and participant‐facing training), such as instructional videos, user manuals, and flowcharts, in a user‐friendly format, including a search function. The knowledge base was accessible via a dedicated learning platform and introduced to site staff during the SIV. The centralized ticketing was staffed by experts who had been trained by the sponsor and technology vendors on the various systems and devices. When needed, vendor representatives were available to further support the helpdesk team. Furthermore, during the trial, continuous support and, where needed, additional training were provided by CRAs, which served as the primary point of contact for the sites.

Participant‐generated data and site oversight

In a decentralized study design, some of the responsibilities for data capture may be transferred from the clinical site to the participant. In RADIAL, data entry was performed both by the site, primarily through manual transcription from paper or electronic source records into the electronic case report forms (eCRFs) in the central study platform, and by the participant, either manually via the custom study app that was tightly integrated with the central study platform, enabling secure and synchronized data exchange between participants and the study and the site. For example, participants in the hybrid and remote arm were asked to report medical events and changes in medication use via the study app, and this information was directly incorporated into the eCRFs. These entries were subsequently reviewed by the clinical study staff and could form the basis for a (serious) adverse event (AE) report or concomitant medication report. The hypothesis was that allowing participants to report events at the moment they occurred would improve the timeliness of reporting and, in turn, enhance the accuracy of the reports, as the events would be recalled more precisely. Additionally, it could alleviate some of the burden for sites to collect these events. This novel way of collecting these reports that is very suitable for DCTs was one of the research questions evaluated in RADIAL.

Participants were asked to report these events continuously, as they occur, but were reminded to report all their events every 4 weeks during “reporting time points.” The participant would receive a reminder on their study app to confirm that all their data were up to date. Upon confirmation, investigators were responsible for reviewing this data and supplementing the data where needed, as data validity and data integrity remain the site’s responsibility. If a participant failed to confirm their data during the reporting timepoint, or if the site deemed contact with the participant necessary upon review, an unscheduled contact would be performed. For the timing of visits and reporting timepoints, the reader is referred to the first paper of this series “Bringing trial activities to participants – Trials@Home RADIAL proof‐of‐concept trial investigating decentralization of trials”. ¹²

EXPERIENCES WITH SELECTING, TRAINING, AND SUPPORTING CLINICAL TRIAL SITES ON DECENTRALIZED ELEMENTS

Experiences with site selection for a trial with decentralized trial elements

Of the 212 sites that received the feasibility questionnaire, 108 expressed their interest in participation, 80 did not respond, and 24 declined. Those sites that declined mentioned high operational burden or no access to the study population. One site declined due to concerns related to decentralized elements, such as remote oversight of the participant. Among the interested sites, 72% indicated interest to participate in either Part A or Part B, 24% were interested exclusively in Part A, and 4% were interested only in Part B. 53% of respondents considered the need for fully remote trial approaches as moderate (moderately desirable), while 42% rated it as high (highly desirable), regardless of the disease indication. The majority of clinical sites (87%) indicated that remote trial participation could be attractive for the RADIAL trial population. Based on the feasibility assessment, contracting timelines, previous experience with sites, and recruitment potential, 61 sites in 6 countries were selected for a site selection visit by video conference. After the site selection visits, 43 sites in 6 countries were selected for participation in the RADIAL trial. Main reasons for not selecting sites were competing studies, expected long contracting timelines, and budget and country recruitment target considerations. Of those sites 43 sites, 37 sites were initially activated. One extra site was identified and activated during the trial, resulting in 38 participating sites (Figure 2 ).

Figure 2.

Overview of country and site selection visits (SSV) for the RADIAL trial.

Experiences with contracting and delegation of responsibilities

In the RADIAL trial, certain site responsibilities of the sites were delegated to third parties, such as the blood sample collection, IMP accountability during home nurse visits, direct‐to‐participant shipments of IMP and IMP temperature monitoring at the central pharmacy. These parties were contracted by the CRO acting on behalf of the sponsor. In RADIAL, as in most trials, contract negotiations relied on standardized templates, and many sites preferred to use familiar contractual language or existing frameworks. In countries using national CTA templates, the templates were often not adapted to the use of decentralized elements and the use of third‐party vendor services for these elements, particularly concerning data privacy and documentation of investigator oversight. This led to different documentation approaches across countries and even across sites within a country, in the case of the UK. Whereas some clinical trial sites agreed on adding the division of tasks and responsibilities to the standard CTA template, others, primarily in the UK, preferred further arranging the delineation of tasks and responsibilities in separate agreements, specifically for the home nurse provider. Usually, a combination of a data processing agreement (DPA), to delineate the responsibilities of processing of data for which the site was the controller under GDPR, and a service level agreement (SLA), to delineate the responsibilities for trial activities and services, were used. These agreements were signed between the site’s representative and the home nursing provider. Alternatively, some sites preferred to conduct home nurse visits using their own clinical staff rather than utilizing services provided by the sponsor. Nine out of 32 sites (28.1%) in Part A opted for their internal nursing teams, primarily due to previous negative experiences with external vendors or the presence of well‐established internal procedures for home visits (Table 1 ). At times, these variations and additional discussions about these responsibilities during the contracting phase significantly increased the contracting timelines. This contributed to the overall long contracting timelines in the trial; for 17 out of 38 sites, the time exceeded 12 months.

Table 1.

Country specific data on selection of own home nurse vs. home nurse provider

Country	Site home nurse	External home nurse provider
Poland	7	0
Spain	1	8
UK	1	7
Denmark	0	2
Germany	0	2
Italy	0	4
Total	9	23

Experiences on clinical site training and support

Despite all SIVs being successfully conducted and initial site training completed as planned, sites reported challenges when enrolling their first participants. Several sites indicated that initial study visits with participants were particularly difficult and time‐consuming due to the need to install the RADIAL study application and configure the study devices. Sites encountered issues when assisting participants, particularly regarding device connectivity and system integration. Some of these issues were related to not understanding the training materials or a lack of general digital literacy skills to effectively operate different types of mobile phones and their Bluetooth connectivity. These issues could be resolved with extra support and additional training from the central helpdesk and CRAs. Other issues were more complex and could not directly be solved by the site or via extra training as they were related to backend integration between systems and were only solved after communication with the helpdesk and systems provider. There were also issues that neither the site nor the helpdesk was able to resolve due to the incompatibility of the participant’s phones with the study devices or issues with the hardware and software of the trial devices themselves. For a more elaborate overview of the technology learnings, the reader is referred to Paper 6 in this series “Operationalizing Decentralized Clinical Trials: Technology Insights and Recommendations from the Trials@Home RADIAL Proof‐of‐Concept Study”. ¹⁴

The knowledge base and helpdesk were widely used across participating sites and played a critical role in resolving technical and operational issues. Multiple sites accessed the knowledge base regularly, with many using it at least several times per month. The most frequently accessed topics in the knowledge base concerned the trial devices that participants used for the collection of data, such as a Bluetooth‐connected glucometer that synchronized data to the study app and directly into the eCRF. A total of 30 out of 38 sites contacted the helpdesk for assistance through the ticketing system. A high proportion of support requests involved technical challenges specifically related to the installation of the study apps and the connection of glucometer insulin pen adaptor cap to these apps. In response to observed needs, additional support measures were introduced, including live or standby sessions that could be requested directly by the site or via the CRA. These measures were implemented following the observation that sites required additional support during the first participant visits. The CRA continued to serve as the main point of contact for the sites. However, their ability to resolve issues was limited in some cases, particularly where technical knowledge was required. Although CRAs were trained on device use and provided demonstrations during SIVs, some sites indicated that additional training could have been beneficial. In practice, CRAs often referred sites to the helpdesk for technical issues, which occasionally led to confusion among site staff regarding roles and responsibilities in support provision and long lines of communication.

During the RADIAL trial, study staff had to manually confirm specific trial milestones for participants on the EDC platform for participants to initiate specific tasks, such as triggering time‐sensitive questionnaires or initiating study shipments. Despite the effort to emphasize the importance of these system confirmations during site training, they were not always done in a timely manner, either because it was misunderstood or forgotten. This occurred frequently in the early stages of the trial, which consequently caused delayed shipments and delayed triggering of questionnaires, thereby missing data collection windows. Additional awareness training was conducted directly via CRAs, and additional supporting documents were uploaded to the knowledge base and shared via the monthly newsletter. In addition to these efforts to increase site awareness of system confirmations, a central coordination team was set up by the sponsor and CRO to track and monitor specific trial milestones for participants and downstream to coordinate trial activities involving multiple vendors and third‐party service providers, such as home nursing, laboratory services, and the central pharmacy. The team held weekly meetings to track participant progress, verify the successful triggering of processes in backend systems, confirm correct and timely vendor order fulfillment, and ensure that necessary forms for home nursing visits were completed by sites. The proactive awareness training and supporting documents and regular central oversight together prevented delayed triggering of system checkpoints, but at the same time required extra resources from the sponsor and CRO.

Although sites were trained on the home nursing procedures during the SIV and a separate training on home nursing services was provided by the third‐party home nursing provider, many sites required a close follow‐up and extra support in the time leading up to the home nurse visit to ensure the home nurse was correctly delegated and the visit was prepared as expected. This additional training required for the home nursing services sometimes delayed site activation, as in practice the scheduling of this training was often difficult. Therefore, based on the experience during the trial, the training was changed from an online synchronous training to a recorded asynchronous training.

Experiences with oversight, participant‐generated data, and site responsibility

Participant training on event reporting, combined with reminder alerts, generally led to timely confirmation of reporting timepoints and successful self‐reporting. This successfully allowed oversight while saving time on phone calls and allowing flexibility to conduct the study activities at a convenient time for participant and site. However, the participants who missed the window for confirmation of a reporting timepoint required follow‐up using unscheduled contacts and additional training or device troubleshooting. This was perceived as a significant burden to the clinical site staff, which can also be seen from the number of unscheduled contacts. The number of unscheduled contacts per participant was 5.6 in the hybrid arm (53 participants) and 10.1 in the remote arm (albeit the overall number of participants in the remote arm was too low, 8 in total, to draw strong conclusions from this), compared with 1.9 in the conventional arm (47 participants). A total of 483 unscheduled contacts occurred in the trial. 67% of the unscheduled contacts were assistance with malfunctioning devices, additional training on trial procedures or devices, to follow up on or as a replacement for a reporting timepoint, or for disease management. In addition, unscheduled visits were also performed to follow up on a reported possible adverse event or lack of data entry (compliance), to discuss and/or schedule visits, extra lab assessment, or expected shipments (Table 2 ).

Table 2.

Reasons for unscheduled visits

Reason for unscheduled visit	Total	Conventional	Hybrid	Remote	Pre‐randomization (Part A)
Malfunctioning device	94	18	65	11	0
Additional training (device support and study procedures)	91	28	51	8	4
Reporting timepoint	72	0	59	13	0
Disease management	71	18	47	4	2
Other (extra lab, logistical, miscellaneous)	59	11	15	27	6
Compliance follow‐up	47	4	41	2	0
Missing reason	28	2	6	13	7
Adverse event	21	6	12	3	0
Total	483	87	296	81	19

Another point of attention was that sometimes sites transcribed participant‐generated data on events and concomitant medication into their own source documentation, most likely because of unclarity on whether additional source records at the site were needed for the participant‐reported events and their entries. This did not align with the initial purpose of alleviating burden for sites in reporting these events, and therefore, the protocol was amended to further detail that the transcribing of data was not required. Moreover, real‐time access to participant‐reported and device‐captured data, such as IMP dosing and glucose levels, often raised questions for clinical site staff about the extent of responsibility for continuous monitoring.

REFLECTIONS AND RECOMMENDATIONS ON SELECTING, TRAINING, AND SUPPORTING CLINICAL TRIAL SITES ON DECENTRALIZED ELEMENTS

This paper provides operational insights into the implementation of DCT elements at clinical trial sites within the multi‐country, phase IV RADIAL proof‐of‐concept trial. While DCTs offer the potential to improve accessibility, flexibility, and efficiency in clinical research, their successful execution requires a redefinition of the roles, responsibilities, and workflows of clinical trial sites. The majority of sites contacted for RADIAL expressed willingness to participate in a trial with decentralized elements, whether fully decentralized or hybrid. Furthermore, almost all sites underlined the theoretical need and attractiveness for DCTs in this patient population. In the conduct of this proof‐of‐concept trial, several key learnings for successful execution of DCTs at clinical trial sites have been identified, such as the need for clear delegation and oversight mechanisms, especially with respect to the integration of third‐party services in the trial and specialized training and support for trial sites, especially on the digital systems used and dealing with technical issues of devices experienced by participants.

The need for clear delegation and oversight mechanisms

The integration of decentralized elements and digital technologies in clinical trials introduces complex webs of responsibilities across sponsors, investigators, vendors, and sites. As Muller et al. ¹⁵ describe, this “problem of many hands” leads to shared accountability for outcomes, such as safety and data protection, making it difficult to clearly assign and oversee individual tasks. Hence, the successful implementation of DCT elements hinges on the clear delineation of roles, responsibilities, and oversight mechanisms across all stakeholders involved. Effective delegation requires comprehensive documentation, robust data handling procedures, risk assessments, and appropriate training for all involved personnel. This responsibility is especially critical when trial activities are spread across locations and systems, as with remote data collection or home‐based services. ICH GCP E6(R3) reaffirms that investigators retain ultimate responsibility for trial conduct, even when tasks are delegated or decentralized. ¹⁶ Similarly, guidance from both the FDA and EMA stresses that the sponsor and investigator must ensure adequate oversight of all delegated tasks, including those handled by third‐party vendors. ¹⁷ , ¹⁸

The RADIAL trial highlighted the operational complexities associated with the involvement of third‐party vendors contracted by the sponsor or its delegate. The clear delineation and documentation frequently resulted in prolonged discussions and consequently longer contracting timelines. This was primarily due to a lack of standardized processes at individual trial sites and the need to customize regional or national contract templates to accommodate these decentralized activities. Efficient contracting is essential for timely trial initiation and smooth trial conduct, and the use of standardized templates can significantly decrease contracting timelines. ¹⁹ Institutional policies and local regulatory constraints have similarly been reported elsewhere as significant challenges encountered by sponsors during the integration of DCT elements into clinical trials. ¹⁰ This shows that, due to the novelty of DCTs and the lack of experience in conducting DCTs by sponsors and sites, these standardized contracts and procedures have not yet been fully realized.

Alternatively, another approach is to allow sites to use their own resources to provide decentralized elements in a trial where they have the capability and capacity. For example, in the hybrid arm, several sites indicated that they preferred using their own home nursing services. As noted by Apostolaros et al., ²⁰ engaging investigators with existing in‐house capabilities to manage such services can improve compliance and efficiency. In line with this, it is recommended to offer sites, where possible, the choice to either engage a third‐party home nursing provider via the sponsor or to use their own home nursing staff. While permitting sites to utilize their own services for decentralized elements can help streamline contracting, not all sites possess the required infrastructure or resources. In the current model, home nursing was included exclusively in the hybrid trial arm, where participants were recruited through site‐based methods and generally resided relatively close to the clinical site. However, in a fully decentralized model with home nursing, participants may be located at a much greater distance from the investigational site, making it less feasible for sites to deliver these services themselves. In such cases, investigators need to be willing and prepared to integrate third‐party services into their practice. These challenges highlight the importance of sponsors and clinical sites working together to develop harmonized and scalable contracting solutions that support the evolving needs of decentralized trial models.

The need for clear agreements on participant‐entered data and participant monitoring

According to the ICH GCP guidelines, source records may include information directly entered by participants, for example, in diaries or electronic patient‐reported outcome (ePRO) data. ¹⁶ However, the accuracy, completeness, and reliability of this data must be ensured through proper oversight. This includes training of participants, applying data validation checks, and maintaining audit trails. In RADIAL, misunderstandings emerged regarding the handling of participant‐reported data, where, in some instances, sites duplicated entire entries rather than updating specific fields following clarification with the participant. As participant‐entered data are considered direct source records, such duplications risk inflating or misrepresenting the dataset. To mitigate this, sponsors should (1) clearly define which data are considered direct entries in the eCRF, (2) provide detailed guidance on handling participant‐entered data, and (3) ensure eCRF design supports straightforward review and correction.

Moreover, DCTs typically involve a more continuous flow of participant data than conventional trials, due to ongoing reporting by participants or data capture via connected devices. This may create an implicit expectation that investigators should immediately review such data, especially in the context of safety‐related events. However, the aim of incorporating DCT elements is to enable remote data collection without imposing an additional burden on investigators. Therefore, even in cases of real‐time data availability, a clearly defined, risk‐based approach to data review frequency is essential. Participants must subsequently be informed about when and how data will be reviewed by investigators to avoid unrealistic expectations, and the usual mechanisms for emergency contacts should be in place when immediate action is required. This risk‐based approach was also often raised in discussions with regulators, sites, CRAs, and auditors. For a complete overview of interactions with regulators, the reader is referred to the second paper in this series “Regulatory interactions and learnings – RADIAL the Trials@Home proof‐of‐concept trial on decentralization”. ²¹

Different needs for site training for decentralized trial elements

Sites were not always able to resolve all issues with decentralized technologies, as it is often hard to determine whether an issue is related to proper use of the device by participants—a responsibility of the research site—the unclarity or correctness of the training materials, a responsibility of the sponsor, or issues related to the hardware and software of devices, a responsibility of the vendors. Only through centralizing this support were we able to adequately support sites and detect more systemic issues. Almost all sites used the knowledge base and helpdesk, and this was deemed a vital part of the trial support structure. Although sites were trained during the trial initiation visit, several sites struggled with supporting participants during the initial visits, particularly when technical issues arose. Limited digital literacy among some site staff, combined with a lack of visibility into system interdependencies and issues with the soft‐ and hardware of the study devices, hindered their ability to resolve problems independently. This highlights the need for both enhanced training and accessible, centralized technical support in decentralized trial settings. Interviews conducted in the Trials@Home project revealed that site staff often felt underprepared to support participants with technical issues, particularly in the early stages of the trial, further underscoring the need for better training and ongoing support structures. ²² Future sponsors that implement decentralized trial elements in a clinical trial should evaluate how to centralize support, organize communications, and align actors—an effort in RADIAL realized partially through the centralized helpdesk. Furthermore, although with the assistance of the helpdesk many issues could be resolved, the large number of technological issues and the mere line of communication of the solutions for these issues placed a significant burden on sites. This resulted in many unscheduled contacts in the arms with more decentralized elements. Adequate budgeting for unscheduled contacts is essential, as fair compensation reflecting the time and effort required from site staff for DCT‐related tasks is critical to ensuring successful implementation and broader acceptance of DCTs by sites.

While training of sites on all aspects of the trial in a timely manner is important, some training activities were scheduled too early. For example, in RADIAL, home nurse training was required for site activation, yet the home nurse visits occurred only 12 weeks after participant enrollment. By the time these visits were due, much of the practical training had been forgotten, and this required extra training and support to sites. After the first occurrence of the home nurse visits for a site, the site gained enough experience and did not need further support. When many decentralized elements are included in a trial, more tailored timing of training on these elements could improve efficiency.

In addition to strong training and support, intuitive system design is crucial for reducing the burden on site staff. A user‐friendly interface can significantly minimize the need for ongoing troubleshooting and is key to successful implementation and adoption. ²² Although the RADIAL trial employed a fully decentralized system developed with input from multiple stakeholders and subjected to user acceptance testing, certain workflows were not intuitive during routine trial conduct. For instance, a critical checkbox in the eCRF was poorly positioned, leading to frequent oversight and subsequently blocking downstream data entry. These challenges align with Johnson et al., ²² who emphasize the importance of user‐centered design in improving usability and note that even initially suboptimal systems can be adapted to better meet user needs. Effective deployment of digital tools in clinical trials is not purely a technical matter but also depends on human factors, such as system usability and the digital literacy of both participants and site personnel. ²³ Often, digital systems are still under development during trial setup, leaving limited time for thorough usability testing. This can lead to implementation delays and complications that compromise efficiency and data quality.

Besides the training to improve the sites’ ability to perform the trial, it is important to acknowledge the site burden of patient‐facing technology. The top three burdens that sites experienced are managing multiple passwords, the variety of systems used, and managing the technological support with the participant. ²⁴ The importance of simplifying technology, such as enabling single sign‐on (SSO) across DCT systems, was also noted in a survey among clinical research professionals and was deemed to be a key aspect of DCT acceptance. ²⁵ Even though RADIAL integrated a lot of technology so that there was a minimal number of passwords, SSO across the different DCT systems was not provided. Additionally, often there was not enough time during the SIV to allow for sufficient hands‐on practical training with the devices and systems. We recommend that when multiple systems and devices are included in a trial, sites should be offered access to access a test or “sandbox” environment that simulates critical and non‐standard trial processes. However, the implementation of these systems is not straightforward and places additional demands on the development of these systems.

Apart from the training needs, the composition of the site team plays is important to consider when implementing DCTs. In RADIAL, the composition of the clinical study team at the site followed a conventional model, typically including a study coordinator or research nurse who served as the primary contact for both the CRA and the participants, handling technical and trial‐related issues. One key recommendation for site staffing is to plan for and account for participant support needs related to decentralized methods in the resourcing forecast. ²⁶ Many sites in RADIAL underestimated the amount of time required to support participants in remote or decentralized elements of the study. As highlighted in our final recommendations (see Table 3 ), it is crucial that at least one person within the site team has either a strong affinity with technical aspects or a very short and efficient communication line with the site’s IT department.

Table 3.

Discussion points related to site readiness and perceptions on DCTs

Topic	Items to discuss between site and sponsor
Site readiness	Have the trial‐specific details on proposed DCT elements been shared with sites and does the site have experience with them separately or in combination? What is site’s mind‐set toward the DCT elements and motivation to conduct them, based on experience or in general? Is there appropriate infrastructure supporting DCT elements (internet connectivity, protection of participant data, and eSource) Are site staff confident using the different technologies used in the trial? Can sufficient site staff be allocated, including a person with high affinity with technology/devices? Is there a CTA template which can properly be used to include DCT elements, necessary data privacy requirements and clarity about division of responsibilities and is this acceptable? Are the proposed DCT elements accepted by the site’s local regulations/management?
Responsibilities delegation and oversight	Are roles and responsibilities discussed and clearly defined, not only between sponsor and site but also regarding third‐party vendors? What is the experience and preference of site regarding the use of third‐party vendors, such as home nurse provider? Is it clear how oversight by the site/PI can be maintained and what tools are at the site’s disposal? Is the role of a participant in the direct entry of data (e.g. ePROs) clear and the site responsibility in reviewing and following up on these data?
Site training	Does the site training meet site’s needs regarding technical aspects/tools and escalation of technical issues? Is the technology sufficiently intuitive or is there more training needed? Is the site aware of the operational expectations especially in areas connected to DTCs, which are not the standard procedures of the site? Is the site aware of the need for digital patient engagement?
Site support	Is the compensation proposed sufficient to cover all activities? Is there budget flexibility ensuring that site staff is adequately compensated for the time spent on ad hoc DCT activities? Is there sufficient material provided to get acquainted with the technology, such as a sandbox environment, hands‐on training with devices and tutorials? Are there possibilities discussed to decrease the sites’ burden, for example with the use of one platform and/or single sign‐ons or the existence of a technical helpdesk?

Selecting sites for decentralized clinical trials

While sponsors are responsible for implementing intuitive systems and training can be closely tailored to the site’s needs, it is inevitable that sites must still adapt to the increasingly complex technological landscape of DCTs by ensuring proper infrastructure and staffing, including personnel with strong digital literacy competencies. Although some sites in RADIAL had previous exposure to DCT elements, many sites did not have extensive experience in conducting DCTs. Public perceptions of digital and online activities and a limited exposure of clinicians and participants to DCTs are considered DCT barriers. ⁹ This aligns with a survey among clinical research professionals that found fewer than half of the respondents had experience with DCTs, highlighting the novelty and ongoing challenges of implementing these approaches. ²⁵ Early discussions with site personnel can help tailor training and identify support needs. This includes evaluating staff technological literacy and openness to adopting new procedures. It is important to assess the site’s technological readiness and ability to communicate effectively through non‐traditional, non‐face‐to‐face channels. ²⁶ Therefore, it might be beneficial to involve the site’s IT team with a high affinity for technology in setting up the study support structure. The evolving clinical trial landscape necessitates all stakeholders to re‐calibrate their position in it and assess what is needed to successfully embark, together, upon the journey of a clinical trial containing DCT elements. Assessing whether a clinical trial site is adequately prepared to conduct a given study is essential. Tailored discussion points on site readiness and perceptions of DCTs can support sponsors, CROs, and sites during feasibility assessments and site selection (Table 3 ).

CONCLUSION

The RADIAL trial provides a comprehensive view of the operational complexities and critical success factors for implementing DCT elements at clinical trial sites. Effective deployment of DCTs requires redefined roles and responsibilities, robust oversight mechanisms, and early stakeholder engagement, particularly in integrating third‐party services and navigating regulatory and contractual complexities. While support is important in any trial, it is especially critical in decentralized trials due to their novelty for many sites, the higher technical complexity, and the different workflows involved. Our findings highlight the importance of tailored, just‐in‐time training, particularly for novel digital tools and centralized support infrastructures, such as helpdesks and oversight teams, to assist sites in managing increased technological and operational burdens. Intuitive system design, adequate user acceptance testing, and fair site compensation are essential to reduce friction during trial conduct. Furthermore, readiness assessments during site selection must go beyond infrastructure to include digital literacy, adaptability, and attitudes toward innovation. As DCTs continue to evolve, increasing experience, more intuitive systems, and improved devices are expected to ease trial conduct and reduce the burden on sites. Nonetheless, success will continue to rely on collaboration, clear communication, and effective coordination across all actors involved.

FUNDING

The Trials@Home project has received funding from the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement no.: 831458. This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation program and EFPIA.

CONFLICT OF INTEREST

LR and MH are employees of Sanofi and hold shares in this company. No conflicts of interest, financial or otherwise, are declared by the other authors.

DISCLAIMER

The research leading to these results was conducted as part of the Trials@Home consortium. This paper only reflects the personal view of the stated authors and neither IMI nor the European Union, EFPIA, or any Associated Partners are responsible for any use that may be made of the information contained herein.

Supporting information

Appendix S1.