Contributed Commentary by Martin Clancy and Katie Best, Phastar 

December 5, 2025 | Decentralized clinical trials (DCTs) aim to harness the power of technology to lower the burden of clinical trial participation, increase engagement and improve outreach and diversity. However, for these benefits to be realized, we need to understand the specific challenges created by DCTs and how to mitigate them. 

Martin Clancy, principal statistician at Phastar, and Katie Best, senior statistician at Phastar, explore the challenges of DCT data analysis, the role of the statistician, and which aspects of trial design and analysis require consideration. 

DCTs Versus Traditional Trials 

DCTs aim to improve participant recruitment, enrollment, engagement, and retention by reducing or removing the need for in-person visits to traditional brick and mortar trial sites. DCTs can be fully remote, using digital healthcare technologies (DHTs) and local healthcare providers (HCPs), or partially remote with occasional visits to trial sites. 

In September last year, the FDA published recommendations for the implementation of decentralized elements in clinical trials. The guidance highlighted the potential benefits of DCTs including enhanced convenience for participants and reduced caregiver burden. The document also highlighted how DCTs could facilitate research on rare diseases and diseases affecting populations with limited mobility or limited access to traditional clinical trial sites. For investigators, DCTs can access a broader pool of patients which are more generalizable to the population of interest. 

However, there are specific considerations for both the design and analysis of DCTs. 

Mitigating Challenges When Planning a DCT 

Remote procedures introduce variability. This can lead to inconsistent data quality and outcome measurement errors. To mitigate this, it is important to standardize remote processes, provide clear, protocol-defined instructions and training for both subjects and telehealth staff. 

Randomization stratification complexity can lead to increased site or staff errors in randomization. It is necessary to simplify stratification, where possible, making it easy to assess and qualify participants in a remote setting, and remove technical barriers through staff training. 

Accessing more diverse patient populations and use of novel endpoints and digital measures can lead to inaccurate assumptions on variability and effect size, which are critical to sample size calculations. To overcome this, build adaptive sample size reassessment into the trial design and consider gathering pilot data if time and budget allow it. 

Increasingly complex data collection and the risk of missing data can lead to the potential for unclean and missing data and delayed analysis. A detailed data management plan ensuring integration of wearables, apps, and patient-reported outcomes is crucial to mitigate this challenge, along with defined trigger rules for sensitivity analysis using conservative missing data rules. Increased patient support is also vital. This should include training on effective use of wearables and apps and expanded patient engagement strategies, such as between-visit text messages or telephone reminders. 

Investigator oversight challenges can also lead to an increased risk of loss-to-follow-up and data entry delays. To mitigate this risk, limit enrollment at sites to manageable levels, implement strong remote monitoring and site support, and add caps per site at the time of randomization planning. 

For fully remote trials where the IP is shipped to the patient, an additional challenge can be defining treatment start. This can lead to misalignment between randomization and treatment initiation. Before starting the trial, clearly define the start of follow up in protocol, align with estimate strategy, and set out the risks and mitigations for delays between randomization and treatment initiation. 

Analyzing DCT Data 

A lack of patient motivation, stemming from reduced direct contact with healthcare professionals and the remote nature of visits and enrollment strategies, may contribute to increased premature withdrawal rates in decentralized clinical trials, despite best efforts to mitigate missing data. To defend the robustness of key conclusions from a DCT, any higher-than-expected rates of premature withdrawal should be proactively addressed in analysis plans prior to unblinding. Just like traditional trials, statisticians should also pay close attention to study discontinuation rates, and reason for discontinuation, throughout the study duration. 

Because DCTs often rely on remote data collection through wearables, mobile apps, and telehealth platforms, there is also a higher risk of data entry lag—or even permanent data loss—compared to more traditional trial settings. If data entry lag impacts the timely availability of key efficacy data, it can make forecasting of key milestones such as interims, database locks, and primary completion dates challenging and less predictable. Working closely with clinical operations team at the study design stage and identifying key endpoints critical to interim decision making is essential. This allows mitigation strategies to be implemented to minimize the risk of missing or delayed data. Any uncertainty around the timing of critical data availability should be closely monitored and factored into the planning of data-driven milestones. 

A key motivation for incorporating DCT elements into trial design is the desire to improve patient outreach and diversity. To ensure this is achieved, it is important to collect data that allow comparison of these outcomes against traditional trials. Sociodemographic factors can influence both trial enrollment and retention in DCTs, so the representativeness of the sample should be carefully assessed to ensure appropriate population-level inferences are made. 

More Guidance is Coming 

Further guidance on decentralized, pragmatic and real-world elements of clinical trials is coming soon. The draft of ICH E6 (R3) Annex 2 is available now, with the final adoption anticipated later this year.  

The guidance highlights the importance of quality by design and focusing efforts and resources on critical aspects of clinical trials that might impact the safety of participants and the reliability of results. The aim is to encourage innovation while avoiding unnecessary complexities. 

DCTs have the potential to extend research to a broader range of participants, creating more robust results and improving rare disease research. However, careful planning is crucial to protect data integrity and ensure we unlock the insights necessary to improve health outcomes.  

Martin Clancy, principal statistician at Phastar, has 10 years of experience in the pharmaceutical and clinical research industry. Specializing predominantly in late-phase clinical development, he has supported in the design and oversight of multiple clinical trials. Martin has supported the implementation of decentralized and remote design elements for registrational trials, ranging from partly remote clinical trials during the COVID-19 pandemic, to fully remote decentralized trials. He can be reached out at Martin.clancy@phastar.com.  

Katie Best, senior statistician at Phastar, has over 5 years' experience within Phastar. She discovered medical statistics during her Statistics MSC at Lancaster University, where she specialized in clinical trials, survival and event analysis, and longitudinal data. Over her time at Phastar, Katie has led studies in a range of therapeutic areas, including respiratory and oncology, being responsible for all statistical components of a study. Katie is also an SME for the clinical trial transparency reporting and version control groups and a member of the PSI training committee. She can be reached out at Katie.best@phastar.com.