Three layers of control can safeguard data integrity in complex atopic dermatitis trials

Atopic dermatitis (AD) clinical trials present unique, complex challenges (Table 1). Oncologists use validated biomarkers and CT scans to diagnose, stage, and evaluate response in cancer patients. However, dermatologists must depend on subjective clinical observations to determine AD severity, progression, and response, which can affect the validity and interpretability of results. For example, inaccurate or inconsistent AD assessments may increase placebo response rates, masking a drug’s true effectiveness. 

AD patients often face mental health problems like anxiety, depression, and sleep deprivation caused by intense itching, which can impact protocol adherence, patient-reported outcomes (PROs), and worsen skin symptoms. Symptoms can be subtle and change rapidly, making it difficult for investigators to determine if an AD patient qualifies for a trial and leading to high screen failure (SF) rates. Additionally, because AD is a chronic disease with unpredictable cycles of flares and remissions, patients may drop out of long-term trials.

In the past five years, Parexel has completed 21 AD clinical trials, enrolling more than 4,800 patients at 2,000 sites. We conducted the pivotal clinical trial program for the first biologic approved to treat AD. Our main focus in AD trials is ensuring data quality. Three layers of data quality control can help reduce risks caused by subjectivity, the placebo effect, comorbidities, and high SF rates. Additionally, involving patients in trial design and execution can boost retention.

1. Train principal investigators and raters

Ensuring rater consistency is essential for AD trials, as high inter- and intra-rater variability can impact key efficacy data and hinder regulatory approval. There are more than 60 measures of AD severity, but only 16 have been validated, and none are considered gold standards. Without objective lab tests or biomarkers, principal investigators (PIs) and raters must assess visual signs such as erythema (redness) and induration (skin thickening). However, estimating the Body Surface Area (BSA) affected by AD is difficult when lesions are flat, mild, or blend with non-lesional skin. Additionally, patients’ skin pigmentation can influence scoring.

The most commonly used assessments in clinical trials include the Eczema Area and Severity Index (EASI), the Validated Investigator's Global Assessment – Atopic Dermatitis (vIGA-AD), and BSA. Often, clinical trial protocols use a specific assessment that requires specialized training. At Parexel, we train (and retrain) physicians and raters on disease severity scoring, even if they use these assessments in their daily practice. 

We have a specialized team of certified dermatologists, including a global therapeutic area lead and regional MDs in the United States, Europe, and China, who are dual-trained in clinical practice and clinical trial operations. They work peer-to-peer with PIs and raters during the training process.

Rater certification and consistency quizzes are part of our standardized training on subjective assessment tools like EASI and vIGA-AD. They help us identify and fix scoring differences that are specific to raters or sites before starting enrollment. To prevent unconscious bias, raters first score vIGA-AD, then EASI. These initial assessments serve as benchmarks for rater performance and reveal any gaps, allowing targeted extra training. This ensures all raters—primary and backup—are ready to follow the protocol-specific scoring standards. Training at least two raters (one primary and one backup) at each site avoids data loss and reduces assessment bias when the primary rater is unavailable, helping to maintain the essential continuity of care.

A strong patient-provider relationship, combined with the natural, chronic, and cyclical nature of AD symptoms, often results in notable (and sometimes inconsistent) perceived improvements in the placebo group, which can threaten trial power. This placebo effect can be substantial, with rates reported as high as 40%, making it very difficult for the investigational product group to achieve statistical significance over the control group. In our PI and rater training, we emphasize the potential risks of high placebo response rates and the importance of distinguishing fluctuations in the natural course of AD from the benefits or possible side effects of the treatment.
 

Table 1. Common challenges and solutions for AD trials 

2. Monitor data in real time 

Monitoring patient data in real-time is essential to identify anomalies in outcome measures, such as unexpected or unrealistic fluctuations in disease activity or a lack of expected variability, and to understand the contextual factors behind subjective scores. At Parexel, we utilize the Elluminate digital platform to track and analyze patient-level data throughout studies. Using Elluminate, we can monitor EASI, BSA, vIGA-AD, and safety data, and visualize them on graphs for review. This allows us to detect signals, outliers, or illogical trends and take appropriate action. We verify results and outcomes, communicate with sites, and implement corrective measures as needed.

Real-time monitoring and site follow-up are handled by our in-house team of dermatologists, who are subject-matter experts (SMEs). SMEs can interpret complex, subjective data fluctuations with clinical nuance that generalist monitors might overlook. For example, if we identify a high variability between raters at a site, we review the details. How much time was spent on assessments: two minutes or one hour? Both are red flags. Assessments should typically take between 10 and 15 minutes. Did the assessment take too long because someone left the recorder on afterward? If a patient’s disease score changes unexpectedly, is it because they violated the protocol by using background concomitant medications? Could it be due to UV exposure, which can either worsen or improve AD? Did the patient introduce new OTC medications? Investigating data discrepancies involves examining every possibility. It requires extra effort, specialized knowledge, and persistent follow-up to resolve these issues.

Red flags prompt an immediate, proactive site inquiry to check for protocol deviations or external influences, clarify deviations, and prevent compromised data from entering the database. We know when to step in, what questions to ask, and who to ask. If necessary, we have used central adjudication of photographs and machine learning to identify any subgroups with high placebo response rates. We quickly flag sites where the response rate is consistently above expectations as potentially having a high placebo response or a data quality problem. Real-time monitoring of all trial data is essential to controlling rater variability and high placebo response rates.

3. Surveil statistical quality 

At Parexel, we use centralized statistical tools to detect site- and trial-level data inconsistencies that are not obvious or easy to spot with standard methods. For instance, we use CluePoints, as used by the FDA, which can recognize patterns such as an investigator consistently scoring endpoints in the same way, a lack of expected fluctuation in AD, or unusual data aggregation that might indicate systematic scoring bias or even fraud, potentially threatening the data set's integrity.
CluePoints can detect key risk indicators (KRIs) early, including:

• Sites reporting overall higher or lower response rates than expected;
• eDiary non-completion;
• Patients with large changes or no change in AD assessment scores between visits;
• Response variability per site;
• Treatment-Emergent Adverse Events (TEAEs) rate – under-or over-reporting;
• Out-of-range vital signs; and
• Lab results (collected, but the sample was not analyzed due to poor quality).

Once KRIs are identified, we develop solutions. If patients are not completing their electronic diaries, we retrain them and follow up with sites. If scores remain unchanged, we verify data accuracy and analyze trends. If TEAEs are high, we monitor and retrain sites as needed, and if vital signs are out of range, we investigate and ensure proper equipment use. 

We recently worked on planning and executing a series of Phase 2b, pivotal Phase 3, and open-label extension studies over seven years in a dermatology indication closely related to AD. The program required continuous medical oversight and attention to detail, including timely responses to lab alerts and other urgent study-related issues, monthly medical meetings to assess trends, and responses to regulators’ questions. Our medical team performed adverse event (AE) and concomitant medication listing reviews, updated AE terms through database lock, and maintained thorough recordkeeping, which contributed to successful submissions and approvals in many countries, including the United States. The sponsor’s Chief Medical Officer said the regulatory submissions contained the “cleanest database” she’d ever seen. 

Patient-guided trial design and execution are key to retention

At Parexel, we leverage our medical and operational expertise to review protocols and challenge unrealistic visit schedules, such as weekly Phase 3 visits, which can cause patient burnout and retention issues. Recently, we partnered with a sponsor to conduct a complex international Phase 1 AD trial with 15 patient cohorts. The trial involved an intense schedule of assessments, including extensive pharmacokinetics sampling and blood draws over 12 hours, with 15 study visits and a one-day visit window. Patient- and investigator-reported assessments used different tools, and patients’ eDiary entries had strict timing requirements, impacting compliance. To ensure the trial’s success, we amended the protocol by removing two study visits and excessive blood draws, reducing patient burden and improving enrollment. To promote consistency between patient and investigator assessments, we increased oversight of data verification and provided additional training for patients and site staff. We also offered extra training and support to sites on the importance of diary compliance. The result? We achieved first- and last-patient-in milestones on time.

We always recommend that sponsors reduce risk and boost compliance by offering standard, high-quality, protocol-approved emollients at no cost to all patients. This step serves two purposes: first, it prevents patients from using unauthorized, potentially treatment-confounding OTC or herbal products they see online; second, it provides a clear financial benefit to patients, increasing overall motivation and retention in the study. Many patients can't afford regular medical care or high-quality moisturizers; therefore, this patient-focused, supportive approach can help improve enrollment and retention rates. 

AD patients, especially children, often face significant mental health challenges, such as anxiety and depression due to the visible symptoms of the disease, associated stigmas, and chronic sleep problems. This requires careful attention during trial management. Since mental health challenges are a common and well-known comorbidity of AD, excluding all patients with mental health issues is impractical and would greatly limit enrollment. Therefore, we recommend that sponsors exclude only those patients with uncontrolled mental health conditions or severe psychological diagnoses. This approach helps preserve the patient pool while ensuring trial integrity.

Integrating mental health assessments, which are subjective and need consistent rating, into the study enables SMEs to track deterioration throughout the research. Medical teams can proactively determine if a patient needs mental health support, as a decline in psychological well-being may trigger worsening AD symptoms. By including mental health assessments during the trial, sponsors can observe how the skin disease advances alongside mental health.

Electronic diary (eDiary) compliance is a common challenge in dermatology studies. Our clinical team has extensive experience in actively monitoring eDiary completion to ensure patients are contacted and reminded to finish diaries before data entry is locked. This approach significantly reduces missed data entries.

A three-layer approach manages key risks

A three-layer, proactive approach involving training, real-time monitoring, and statistical surveillance enables sponsors to effectively manage the core risks of AD trials. This comprehensive quality strategy reduces the risk of high placebo response rates, produces clean databases, shortens data lock and regulatory approval timelines, and provides sponsors with a clear competitive edge in AD. Targeted patient support and strategic protocol design can enhance retention. To optimize AD studies, our clinical team offers sponsors feedback on inclusion and exclusion criteria, rescue medications, and visit schedules in the protocol, starting at the bid defense stage.
 

Return to Insights Center