A Statistically-Based Process for Auditing Clinical Data Listings

In October 1995, members of the DuPont Merck Biometrics and Quality Assurance staffs met to discuss improvements in the audit process for clinical data listings. With the advent of the International Conference on Harmonization (ICH) good clinical practices (GCPs), they were concerned about assuring that the procedures and processes for managing incoming clinical data and reporting to regulatory agencies were appropriately rigorous to ensure complete and accurate processing of clinical trial data. The following article describes a statistically-based audit plan that takes into consideration the phase of the trial, the importance of the trial to regulatory filings, and key study variables. The audit plan also includes a procedure for randomly selecting studies and patients for audit, and establishes statistically-based acceptance and rejection criteria. This audit process has been in use for six months and has resulted in more efficient planning and execution of clinical data audits. Due to the randomness of the process and high standards for acceptable quality levels (AQL), internal quality control (QC) procedures have been strengthened to assure that all data listings meet these standards.