Every sponsor decision, from dose selection to portfolio prioritisation, depends on whether the evidence generated in a trial is based on reliable data. If data confidence is low, uncertainty can make patient safety oversight harder, slow regulatory dialogue, and raise the cost of each subsequent decision.
Data reliability here means more than the accuracy of what is collected. It begins with study design, because a trial can only produce dependable evidence if its endpoints, safety assessments, sampling schedules and dose levels are chosen to answer the right question. A flawlessly executed study that measures the wrong things will still leave a sponsor short of the evidence they need.
Therefore, to achieve reliable data, trial quality should be built into the study design, rather than inspected at the end. ICH E6(R3), the current Good Clinical Practice guideline, emphasises a culture of quality, critical-to-quality factors, and proportionate, risk-based approaches throughout trial planning, conduct, and reporting. It acknowledges that reliable data is not only clean data but also complete, attributable, timely, and contextually well understood enough to support a decision.
For sponsors, the practical challenge is deciding which signals matter most, especially in first-in-human studies. Early trials are designed to establish safety, tolerability, and pharmacokinetics in humans. When proof-of-concept elements are included, they may also provide the first signal of efficacy. Unreliable data at this stage can push a sponsor towards the wrong dose, the wrong Phase 2 design or the wrong conclusion about whether the investigational product should continue. Consequently, discontinued, suspended, or withdrawn trials were most commonly recorded in Phase II, with 8,683 trials between 2021 and 2026, compared with 4,085 in Phase I and 2,318 in Phase III.[1] For sponsors, the implication is that first-in-human studies should be designed to surface safety, tolerability or exposure concerns early, before those issues are carried into larger and more expensive Phase II programmes.
Monitoring for data confidence
GlobalData clinical trials database figures show the scale of trial discontinuation risk. Across 77,819 suspended, terminated or withdrawn trial records since January 2021, low accrual rate was cited in 9,410 cases, lack of efficacy in 2,222, adverse events in 1,203 and protocol deviation in 202. A further 44,078 records were listed as unspecified, 13,367 as other, 2,712 as business or strategic decisions, 2,288 as financial reasons, 1,625 as product discontinuation, 425 as regulatory issues, 127 as positive outcome, 112 as ethical considerations and 48 as data collection challenges.
. With most trials discontinuing in Phase II, these statistics demonstrate the need for sponsors to monitor data reliability as early as possible.
Sponsors need to focus on indicators that leave gaps in the data to miss risk to patient safety or decision-critical endpoints. A poorly designed study can miss primary or secondary endpoint data but designing it correctly but missing the blood draw results in the same issue. In early-phase trials, this can include missed pharmacokinetic draws, incomplete safety assessments, or inconsistent timing against protocol-defined windows. A blood draw recorded seconds late may be manageable if the actual time is captured accurately, but a missed draw is materially different. A mixed-up sample between participants or time points is more serious still.
Protocol deviations should also be interpreted by impact, not simply volume. Sponsors need to understand whether deviations are administrative, endpoint-relevant, or safety-relevant. Trends are also important. For example, a single mistake might be isolated, but recurring issues across different cohorts, shifts, or sites suggest a process weakness that could need corrective and preventive measures.
Query management is another useful measure of data reliability, and fast responses alone are not enough. Sponsors should assess whether query resolution demonstrates genuine oversight, source understanding, and root-cause thinking. Superficial query closure can create the appearance of control while leaving the underlying risk intact.
Finally, a structured Corrective Action Preventive Action (CAPA) process must be in place. Sponsors should ask whether the root cause has been identified, whether the action is proportionate, and whether recurrence is being monitored. The European Medicines Agency’s (EMA) risk-based quality management paper defines quality risk management as the systematic assessment, control, communication, and review of risks associated with clinical trial planning and conduct.
The impact of Phase 1 data
Late-stage trials attract attention because they are larger, more expensive, and often pivotal for approval. Yet Phase 1 data is important because it shapes the entire development path.
Early-phase studies come with better controls that late-phase studies spread across many small investigator sites. Participants may be residents in a clinical unit with a controlled environment, assessments are performed by fulltime research staff, staff are always around to record an adverse event and dosing is closely supervised. Later-stage trials introduce more variability, including home-based reporting, outpatient visits and self-administration of the investigational product.
This doesn’t imply that Phase 1 data is risk-free, but sponsors have fewer reasons to ignore avoidable risk indicators. When datasets are small, a few errors can alter the apparent safety profile or pharmacokinetic interpretation. The reliability threshold should therefore be high from the start.
This is why the design stage is crucial. Sponsors need a detailed understanding of toxicology, nonclinical efficacy, and likely exposure-response assumptions before the first participant is dosed. The study should test doses expected to be efficacious while allowing a safety buffer for variables such as food effects or drug-drug interactions. Just as importantly, assumptions should be reviewed as human data emerges. Nonclinical-to-clinical predictions can be imperfect, so real-time data review is essential.
Selecting a provider that can safeguard decision-critical data
To find the right provider, sponsors need to look for sites that will partner up front to help design a scientifically sound and operationally feasible study. That will help make sure the right data is collected consistently. But as quality issues do arise, they should also examine how quality events are captured and trended. A robust provider should be able to show that events are recorded consistently, investigated promptly, linked to clear root-cause analyses, and then monitored for recurrence across cohorts, studies, and sites. This helps sponsors identify patterns, such as repeated missed samples, timing errors, or incomplete source documentation.
A strong partnership starts with collaboration, as even the strongest KPI framework will fail if site teams are reluctant to engage with sponsors on emerging risks. Nucleus Network’s global model is built around unified oversight and shared governance across its sites, supported by harmonised processes, shared best practices, and coordinated project delivery.
Nucleus Network is the only global provider dedicated to early phase clinical trials, operating Phase 1 facilities in Brisbane, Melbourne, Minneapolis and London, with 350 Phase 1 beds, 2,500 Phase 1 clinical trials conducted and an in-house database of 700,000 dedicated participants. Its London site is a purpose-built Phase 1 facility specialising in first-in-human studies and healthy and patient volunteer research.
The company’s approach addresses quality by design and a robust quality management system. Nucleus will work extensively with sponsors on their design, pulling from their years of experience in designing and conducting early phase clinical trials. Their quality system includes an electronic quality management system that captures quality events, tracks investigations, documents root-cause analysis, and monitors CAPA implementation. The organisation also uses weekly, monthly, and quarterly quality data reviews, supported by AI-enabled trending and lookbacks for repeat issues. Those reports are reviewed by senior leadership and executive teams, with lessons from one study used to refine processes across sites. This cross-portfolio learning means that data reliability is protected by a combination of protocol design, trained staff, monitoring discipline, query handling, quality governance and a culture that treats small signals as potential early warnings.
In early-phase development, that discipline can be the difference between a decision made with confidence and one made in uncertainty. To find out more, download the free whitepaper below.
[1] GlobalData clinical trials database, accessed 08/06/2026
