Case studies on the use of central statistical monitoring and interventions to optimize data quality in clinical trials

Abstract

Purpose: QDSS is an analytics-driven centralized monitoring system that uses a Statistical Process Control (SPC) approach with added clinical surveillance. The objective of this study was to review recent cases that arose in the course of Central Statistical Monitoring and intervention to illustrate the nature of the data quality issues that can be observed and corrected using this approach. Methods: SPC charts were generated with upper and lower limits based on the distribution of the data, which is updated as the study progresses. Red signals indicate an “out of control” data point and green signals represent data points that are within the control limits. Cases with red signals that illustrate the full lifecycle of QDSS were chosen to examine the process of signal detection, root cause analysis, intervention, and follow-up. Results: Three cases with “out of control” data points were selected. Data was visualized using SPC charts such as an X-bar control chart. The out of control data points signaled a root-cause analysis to begin. Further examination into related data discrepancies, discussions with study staff, and review of study protocols resulted in the identification of the root cause of the poor data quality, for example, a lack of training for new study staff that entered part way through the study. Based on the root cause, interventions were suggested to the study staff, such as retraining new staff members. After the implementation of the intervention, the data was re-examined to confirm the resolution of the “out of control” data points. Conclusions: Centralized statistical monitoring is an important tool in clinical trials to detect data outliers and anomalies that may have an effect on study outcomes. In all three cases studies were threatened by a data quality risk in their primary or secondary endpoints that were easily corrected. These case studies illustrate the power of QDSS to effectively identify and resolve data aberrancies to increase assay sensitivity.