data visualization

risk-based monitoring How can Sponsors harness the power of data visualization and statistics to enhance the collaboration between clinical operations and the biometrics team?

Before diving into data-driven monitoring, it is important to note that various perspectives of the personnel involved. The CRA generally focuses on Source Data Verification and pending action items while keeping in mind the project timelines. The Statistician takes on the responsibility of calculating and analyzing p-values and confidence intervals. Meanwhile, the Data Manager is concerned with turnaround time and data cleaning. Finally, the Project Manager takes a look at the overall risk.

In a traditional monitoring approach, 100 percent SDV is checked at the site with according to the monitoring plan agreed upon at the beginning of the study. It has been demonstrated that traditional ways of conducting frequent monitoring visits every four to eight weeks to investigational sites and 100 percent SDV do not necessarily result in higher data quality.

Therefore, there is a need to develop a smarter and complementary approach to achieve effective study data. In a Risk-Based or Adaptive approach, data quality and integrity as well as patient safety are achieved by combining the following approaches:

  • Supervised Analysis: based on risk indicators and definition of thresholds
  • Unsupervised Analysis: centralized statistical monitoring for fraud, trends and correlations
  • Adaptive SDV according to data quality

With the adoption of the ICH GCP E6(R2) guidelines, Sponsors need to consider a strategy that aligns monitoring with these guidelines. The ICH GCP E6(R2) guidelines are the most important change in ICH guidelines in the past 20 years, and Sponsors need to find a way of being more targeted.

This means adopting a “Quality by Design” approach and more focus on quality management, data analytics and centralized statistical monitoring.In terms of processes, this means focusing on the following areas:

  • Data Management: data quality, timely resolution and improved communication with the clinical team
  • Biostatistics: comprehensive data review via algorithms, new types of summary statistics around site data quality (composite risk score) and new levels of involvement during the study
  • Clinical Operations: transparency, site level risk, task management and audit trail of data review and actions
  • Medical Team: consolidated data availability and subject safety

Risk-Based Monitoring is an evolving framework that has moved from site visits to centralized monitoring via Excel to automation and finally to a centralized statistical monitoring approach that increasingly uses data-driven, statistical solutions.

The current state of RBM focuses on four core areas:

  • Risk Profiling to include the identification and prioritization of risks and a mitigation plan
  • Surveillance including analytics such as data visualization, statistical models and alerts
  • Centralized Decisions based on data-driven monitoring to including subject level data, site level performance and site monitoring recommendations
  • Issue Management for recording issues and taking action

ICH GCP E6(R2) Risk Based ApproachAnother important aspect to take into consideration in terms of the ICH GCP E6(R2) guidelines in the responsibility of the Sponsor for data quality. Even if the study is full outsourced to a CRO, the Guidelines state (5.2.1), “a Sponsor may transfer any or all of the Sponsor’s trial-related duties and functions to a CRO, but the ultimate responsibility for the quality and integrity of the trial data always resides with the Sponsor”.

Centralized Statistical Monitoring gives Sponsors that oversight of the quality of monitoring and data coming from sites. It enables Sponsors to identify issues such as lack of variability or implausible values that go undetected using other approaches. This approach also allows for visual exploration and facilitates the intervention at sites to improve the quality of the data. A risk score can be assigned to each site with an advanced and sophisticated statistical approach.

CROS NT has a tool that uses Principal Component Analysis (PCA) which is a multivariate analysis that converts p-value observations using statistical analysis to identify problematic sites rather than subjective thresholds. The tool also offers comprehensive data visualization including patient profiles and drill-down facility.

The value to the Sponsor is an increase in clinical data quality, optimization of on-site monitoring and a significant reduction in overall regulatory submission risk.