Central site monitoring: Results from a test of accuracy in identifying trials and sites failing Food and Drug Administration inspection. (April 2014)

• Record Type: Journal Article
• Authors: Lindblad, Anne S; Manukyan, Zorayr; Purohit-Sheth, Tejashri; Gensler, Gary; Okwesili, Paul; Meeker-O'Connell, Ann; Ball, Leslie; Marler, John R
• Is Part Of: Clinical trials. Volume 11:Number 2(2014)
• Page Start: 205
• Page End: 217
• Publication Date: 2014-04
• Subjects: 615.5072405
• DOI: 10.1177/1740774513508028
• Languages: English
• ISSNs: 1740-7745

Abstract

Background: Site monitoring and source document verification account for 15%‒30% of clinical trial costs. An alternative is to streamline site monitoring to focus on correcting trial-specific risks identified by central data monitoring. This risk-based approach could preserve or even improve the quality of clinical trial data and human subject protection compared to site monitoring focused primarily on source document verification. Purpose: To determine whether a central review by statisticians using data submitted to the Food and Drug Administration (FDA) by clinical trial sponsors can identify problem sites and trials that failed FDA site inspections. Methods: An independent Analysis Center (AC) analyzed data from four anonymous new drug applications (NDAs) where FDA had performed site inspections overseen by FDA's Office of Scientific Investigations (OSI). FDA team members in the OSI chose the four NDAs from among all NDAs with data in Study Data Tabulation Model (SDTM) format. Two of the NDAs had data that OSI had deemed unreliable in support of the application after FDA site inspections identified serious data integrity problems. The other two NDAs had clinical data that OSI deemed reliable after site inspections. At the outset, the AC knew only that the experimental design specified two NDAs with significant problems. FDA gave the AC no information about which NDAs had problems, how many sites were inspected, or how many were found to have problems until after the AC analysis was complete. The AC evaluated randomization balance, enrollment patterns, study visit scheduling, variability of reported data, and last digit reference. The AC classified sites as 'High Concern', 'Moderate Concern', 'Mild Concern', or 'No Concern'. Results: The AC correctly identified the two NDAs with data deemed unreliable by OSI. In addition, central data analysis correctly identified 5 of 6 (83%) sites for which FDA recommended rejection of data and 13 of 15 sites (87%) for which any regulatory deviations were identified during inspection. Of the six sites for which OSI reviewed inspections and found no deviations, the central process flagged four at the lowest level of concern, one at a moderate level, and one was not flagged. Limitations: Central data monitoring during the conduct of a trial while data checking was in progress was not evaluated. Conclusion: Systematic central monitoring of clinical trial data can identify problems at the same trials and sites identified during FDA site inspections. Central data monitoring in conjunction with an overall monitoring process that adapts to identify risks as a trial progresses has the potential to reduce the frequency of site visits while increasing data integrity and decreasing trial costs compared to processes that are dependent primarily on source documentation.