Automation And Visualization Of PK/PD Data Reconciliation In Clinical Trial

PK/PD data plays important role in clinical trial. In early clinical trial, PK/PD can help to understand the efficacy and safety of drug and provide basis for critical decisions in clinical trial (e.g."go/no go", dose and administration route selection). But PK/PD analysis results are prone to exposing dosing information which can lead to breaking blind, so it is collected and managed in double-route-double-database way which is different from other CRF data. The information of samples (such as collection date, time, PKUSMID, etc.) is recorded on the CRF and stored in study database, while the samples are analyzed by vendor or analytical lab and the analysis results were loaded in a separate database with strictly limited access to prevent breaking blind. During this process, the PK/PD data is transferred through different departments and databases and it is easy to have missing, inconsistency or other data issues, so data managers need to conduct PK/PD data reconciliation.Due to the access limit and complicate logical relationship, PK/PD reconciliation can’t be done through edit check, so it is conducted through manual review currently, which can be time-consuming with unavoidable human error. At the same time, PK/PD data is closely related to other modules, such as dosing, so several modules are involved in PK/PD reconciliation, and multi-data source operation is tricky.This research created PK/PD reconciliation tool with SAS code to reconcile and mark data issues automatically and VB to create user-friendly interface to visually control PK/PD reconciliation process based on the summary of data issues in8664PK/PD raw data. Then the clinical data of Pfizer was used, including897simulated PK/PD data,8664PK/PD data in completed study and2820PK/PD data in ongoing study, to compare the efficiency and accuracy of PK/PD reconciliation tool and single manual review.There are575PK/PD data issues in9completed clinical trials and the incidence rate is6.6%, mainly including:missing (60.3%), inconsistency (19.5%), out of time window (12.0%) and logical error (8.3%). Based on the summary, PK/PD reconciliation tool was created with easy operation process. When it is applied for different studies, users can choose or enter study-specific variables and select datasets need to be reconciled through browse, the tool will transfer the variables to SAS macro and run SAS code to reconcile data automatically. There is no need to change SAS code, so the misoperation in SAS code can be avoid and there is no requirement for SAS knowlesge in users. The verification result showed that comparing with manual review, PK/PD reconciliation tool improved the accuracy by7.4%(P=0.05) while shorten the time by86.8%(P=0.0002). So PK/PD reconciliation tool can improve the work efficiency of PK/PD data reconciliation process.