Centralized Monitoring in Clinical Trials: Everything You Should Know

What Is Centralized Monitoring?

Centralized monitoring is a key component of Risk-Based Quality Management (RBQM) that enables proactive detection of quality-related risks, both pre-identified and unanticipated, during clinical trials. Centralized monitoring improves data quality, enhances patient safety, and ensures regulatory compliance by leveraging statistical and data-driven methodologies.

The Three Pillars of Centralized Monitoring

Centralized monitoring is built on three primary elements:

1. Statistical Data Monitoring (SDM)
2. Key Risk Indicators (KRIs)
3. Quality Tolerance Limits (QTLs)

Each of these components plays a unique role in identifying, assessing, and mitigating risk in clinical trials. Let’s explore them in detail.

Table of Contents

• Statistical Data Monitoring (SDM): Uncovering Hidden Risks
• Key Risk Indicators (KRIs): Targeting Known Risks
• Quality Tolerance Limits (QTLs): Defining Acceptable Variability
• How SDM, KRIs & QTLs Work Together
• Regulatory Guidelines & Best Practices for Centralized Monitoring in Clinical Trials
• How KRIs Improve Data Quality in Clinical Trials
• How SDM Enhances Data Quality by Detecting Hidden Risks
• Comparing Centralized Monitoring with Traditional Monitoring Methods in Clinical Trials
• The Future of Centralized Monitoring in Clinical Trials with Artificial Intelligence (AI) & Machine Learning (ML)
• Challenges & Opportunities in Implementing Centralized Monitoring for Clinical Trials

Statistical Data Monitoring (SDM): Uncovering Hidden Risks

What Is SDM?

Statistical Data Monitoring (SDM) is an unsupervised, data-driven approach to identify unanticipated clinical trial risks. Unlike traditional monitoring methods, which rely on predefined assumptions, SDM systematically evaluates all collected clinical data to detect anomalies and inconsistencies.

How SDM Works

SDM utilizes advanced statistical models to:

• Identify atypical patterns across patients, investigational centers, and regions.
• Detect potential fraud, miscalibrated equipment, protocol deviations, and unintentional errors.
• Generate risk signals based on statistical tests, prioritizing “at-risk” sites for further investigation.

Key Risk Indicators (KRIs): Targeting Known Risks

What Are KRIs?

Key Risk Indicators (KRIs) are predefined metrics highlighting potential clinical and operational risks in a study. Unlike SDM, which uncovers unknown risks, KRIs focus on monitoring specific, expected risk factors. KRIs should be monitored continuously; trends over time can provide early warnings of site-level issues. Choose KRIs relevant to the therapeutic area and trial design, focusing on those that affect patient safety or trial validity.

Examples of Common KRIs

• Protocol Deviations per Site: A high rate of deviations may indicate training deficiencies or investigator non-compliance.
• Screen Failure Rates: A sudden increase in screen failures could signal a misinterpretation of eligibility criteria.
• Adverse Event (AE) Reporting Rates: Disparities in AE reporting across sites may indicate underreporting or differences in site management.

Quality Tolerance Limits (QTLs): Defining Acceptable Variability

What Are QTLs?

Quality Tolerance Limits (QTLs) are study-level thresholds set to detect systemic issues that could impact trial integrity. They serve as a safeguard to ensure that critical parameters remain within an acceptable range. While KRIs focus on site-level or operational risks, QTLs are predefined study-wide limits requiring corrective action if breached.

Examples of QTLs

• Patient Discontinuation Rate: A predefined threshold ensures excessive dropouts trigger an investigation.
• Protocol Compliance: If deviations exceed an acceptable range, intervention is required.
• Efficacy Endpoint Variability: Significant deviations in primary endpoint measurements across sites may indicate inconsistencies.

How SDM, KRIs & QTLs Work Together

While each component serves a distinct purpose, their true power lies in integration. SDM uncovers hidden risks that KRIs and QTLs might not detect. KRIs focus on predefined risk areas, complementing SDM’s exploratory nature. And QTLs provide a structured framework for identifying systemic study risks. When combined, these elements offer a comprehensive risk management strategy that enhances data quality and regulatory compliance.

Regulatory Guidelines & Best Practices for Centralized Monitoring in Clinical Trials

RBQM and centralized monitoring are backed by the world’s most significant clinical development regulators. RBQM was encouraged by both the Food & Drug Administration (FDA) and the European Medicines Agency (EMA) through guidances published as early as 2011. It was more firmly incorporated as a GCP expectation in 2016 with the publication of ICH E6 (R2) and has been further reinforced with the publication of ICH E8 (R1) and ICH E6 (R3).

ICH E6 (R2) emphasizes the need to develop risk-based approaches to monitoring clinical trials, including a clear expectation for centralized monitoring. It also supports the need to document centralized monitoring activities and encourages sponsors to utilize new technologies to make clinical research safer and more efficient.¹ For an in-depth look at the evolution, benefits, and future of RBQM, download “The Ultimate Guide to RBQM.”

How Key Risk Indicators (KRIs) Improve Data Quality in Clinical Trials

KRIs are vital to centralized monitoring and are adept at quickly pinpointing issues and discerning trends.

In each study, around ten to twenty different KRIs are created, each with a predefined value range. They mainly focus on the trial’s reliability and compliance and include metrics such as adverse event reporting rates, protocol deviation rates, and data entry timelines. Sites breaching these thresholds during ongoing data review will be earmarked as high risk and flagged for follow-up.

To quantify the potential benefits of KRIs during clinical trials, CluePoints analyzed 212 studies, comprising a total of 1,676 sites having risk signals on at least one of the nine commonly used KRIs covering a broad range of risks. 83% of the site KRIs improved at the closure of the risk signal, as measured by an improved P-value between signal creation and closure.²

How Statistical Data Monitoring (SDM) Enhances Data Quality by Detecting Hidden Risks

SDM can reveal unexpected risks to data quality by assessing hundreds of variables collected during the clinical trial and converting the mass of P-values computed for each site into a global score, named the site Data Inconsistency Score (DIS). This score is regularly calculated for each clinical trial site as the study data accumulates. A site is considered ‘at-risk’ or atypical when its DIS reaches the significant threshold (DIS ≥ 1.3 corresponding to a P-value > 0.05).

In a bid to quantify the potential benefits of SDM during clinical trials, CluePoints analyzed data from 159 studies involving a total of 1,111 atypical sites with a significant DIS. In 83% of the sites, the DIS improved when all risk signals created at the site level were closed by the study team.³

Comparing Centralized Monitoring with Traditional Monitoring Methods in Clinical Trials

As part of the same project, the team retrospectively analyzed two studies that did not use any centralized monitoring tools, such as SDM or KRI, and compared their data to those from the 159 studies that did use SDM. They found that forty-three of the non-centralized monitoring sites would have had a significant DIS during the trial if SDM analysis had been conducted. Among those forty-three sites, only 56% had an improved DIS at the end of the clinical trial, which is much lower than the 83% of sites using SDM.³

Additionally, a Transcelerate analysis confirmed that traditional monitoring provides a relatively low return on investment. This analysis of more than 1,100 trials by 50-plus sponsors found that just 3.7% of EDC data is corrected after initial site entry. Of these, just 1.1% of data is amended as a result of SDV/SDR, with 1.4% being the result of EDC auto-queries and 1.2% the result of all other reviews, such as data management, biostatistician review, or medical/safety monitoring. In addition, it was unclear whether the detected errors would’ve had any bearing on data quality or patient safety.⁴

The Future of Centralized Monitoring in Clinical Trials with Artificial Intelligence (AI) & Machine Learning (ML)

The application of AI and ML to centralized monitoring is already driving further efficiency. The “risk signals” used to monitor and track investigations and corrective action contain vast amounts of free text entered by various users as they document their findings. Therefore, using natural language processing (NLP) and deep learning techniques, CluePoints screens all this information and flags risk signals that lack the required documentation. By running this algorithm retrospectively on all risk signals processed prior to the release of this ML algorithm, we observed that 40% of the signals weren’t clearly documented. This has been reduced to 30% since the release, or a 25% overall improvement in signal documentation.⁵

Challenges & Opportunities in Implementing Centralized Monitoring for Clinical Trials

By enabling sponsors to identify potential risks and take timely corrective action—before issues can negatively impact data reliability—centralized monitoring can help sponsors give their studies the best chance of success. Regular data reviews can safeguard patient well-being and maintain the integrity of results throughout the trial, contributing to inspection readiness and accelerating overall timelines.

Yet hurdles to adoption remain, and chief among them is change management. Traditional methods, such as on-site monitoring, are heavily embedded in our processes and mindsets. It means that successfully implementing change requires both a bottom-up and top-down approach that demonstrates the value of centralized monitoring techniques, provides easy-to-use systems, and, crucially, secures buy-in from all levels at all functions. Want to learn more? Contact us to continue exploring centralized monitoring in clinical trials.

References:

1. Integrated Addendum to ICH E6(R1): Guideline for good clinical practice: E6(R2) [Internet]. 2017. https://www.ich.org/page/efficacy-guidelines. Accessed 27 July 2022
2. de Viron, S. et al. Does Central Monitoring Lead to Higher Quality? An Analysis of Key Risk Indicator Outcomes. Ther Innov Regul Sci 57, 295–303 (2023). https://doi.org/10.1007/s43441-022-00470-5
3. de Viron, S. et al. Does Central Statistical Monitoring Improve Data Quality? An Analysis of 1,111 Sites in 159 Clinical Trials. Ther Innov Regul Sci (2024). https://doi.org/10.1007/s43441-024-00613-w
4. Sheetz, N. et al. Evaluating Source Data Verification as a Quality Control Measure in Clinical Trials. Ther Innov Regul Sci 48, 671–680 (2014). https://doi.org/10.1177/2168479014554400
5. Young, S. Using Machine Learning and NLP to Improve Central Monitoring Documentation. Applied Clinical Trials. December 2023.