Data Quality, Integrity and Risk Management considerations in Clinical Trials

Organizers: Rakhi Kilaru (PPD)
Chair: Rakhi Kilaru (PPD)
Vice Chair: Richard Payne (Lilly)

Frank Rockhold (Duke)
Andrew Hartley (PPD)
Kelci Miclaus (Veeva Systems)
James Vaughan (PAREXEL)


Title: Quality by Design: Implications for Pragmatic Clinical Trials
Speaker: Frank W. Rockhold (Duke)

Data from “real world” clinical practice and drug utilization – outside of clinical trials – is regarded as a pragmatic source of evidence with high potential to clinical research. Robust RWD will not only leverage increasing volumes of data, but weave together different sources of data, such as clinical data, registries, and electronic health records, to bridge the gap between efficacy and effectiveness and enhance the efficiency of clinical research. Notably, US, Japanese and EU regulatory agencies, have launched major initiatives to address issues regarding the use of real-world data (RWD) to inform regulatory decision making encapsulated in the concepts of “Quality by Design”. Although many challenges and limitations remain with the use of RWD, there have also been many successful case studies. In this talk the challenges of synthesizing real-world data and randomized clinical trials data to generate well designed quality clinical trials to facilitate more efficient clinical research and clinical decision-making. Examples on using RWD to design randomized clinical trials and a case study will be shared.

Title: Some Statistical Tools for Assessing and Predicting Data Quality in Clinical Trials
Speaker: Andrew M. Hartley (PPD)

Data are generally considered high quality if they are “fit for [their] intended uses in operations, decision making and planning.” ¹ High quality data are data that closely approximate their intended underlying quantities (estimands) and satisfy reasonable distributional assumptions. High data quality ensures accurate and valid assessments of efficacy, safety and tolerability, and increases statistical power and precision, facilitating favorable regulatory outcomes. Clinical statisticians and programmers can assist researchers in checking and improving data quality, by assessing the probabilities of data errors, of deviations from assumptions, and of satisfactory data quality at the completion of clinical trials. They can also estimate expected benefits of interventions such as additional data cleaning and training sites in study procedures; such estimations may aid organizations in selecting the most beneficial interventions. This presentation outlines some predictive, inferential and decision-analytic tools for such tasks

Title: Clinical Data Integrity Signal Detection
Speaker: Kelci Miclaus (Veeva Systems)

Methodology for clinical operational integrity has shown great advancement as regulatory guidance moves away from strict 100% source data verification to allow reporting meaningful/impactful signals of quality. In this presentation we discuss several methods of clinical subject and clinical site anomaly detection, including detection of unusual patterns in clinical findings data using empirical site distributions and multivariate outlier detection to supervised risk-based monitoring techniques. We will also discuss anomaly detection where patterns across clinical domains are used to assess data completeness in clinical trials. Finally we focus on the integral part data visualization plays in assessing and understanding data integrity signals in clinical trials research.

Title: Quality by Design: Risk Based Management
Speaker: James  Vaughan (PAREXEL)

As the industry’s acceptance and understanding of Risk Based Monitoring has evolved, Parexel identified the need to incorporate a robust solution for central statistical monitoring of clinical trial data into our Risk Based Services. This is being accomplished through the implementation of complimentary technology platforms: the Parexel Data Driven Monitoring Tool, used to review and manage clinical site activity, and CluePoints, used for the central statistical analysis of scientific patient data. This presentation will focus on Parexel’s approach to Risk Based Management, the objective being the identification of issues which may indicate emerging trends, patterns, and/or systemic risks to subject safety and the reliability of trial results. Examples of identified risks from both technology platforms will be shared.