Clean data continuously
We’ve found that the traditional approach to data handling, which involved gearing up the analyst team to “clean” data before the database lock, is obsolete for CIDs. Trials with adaptive elements require precise planning for the data that will be needed at each stage. That data must be monitored, analyzed, and cleaned on an ongoing basis. Throughout these trials, sponsors must communicate with independent data monitoring committees (IDMCs) so that they can review the data and make timely decisions about stopping, modifying, or expanding the study to a new indication or patient population.
Continuous data cleaning requires tight coordination between data teams and clinical teams. At Parexel, we have adopted a data-cleaning approach that can deliver interim analyses to IDMCs with a minimal lag time after the last patient’s assessment. We have learned by experience that this is a prerequisite for IDMCs to make timely decisions. Smaller companies sometimes underestimate the data monitoring requirements of running an adaptive trial, and the frequent need for data cuts can quickly overwhelm their in-house staff.
At Parexel, we find that only about 3% of data needs correcting or changing due to the source verification, data cleaning, and query process. Rather than deploying resources and expending effort on all the data, Parexel’s risk-based approach to data monitoring concentrates on critical factors, such as the project-specific QTLs and KRIs defined at the study’s start. We continually interrogate the data to spot emerging risks, mitigate them, and make decisions to protect the integrity of the dataset.