OBJECTIVES:

To focus on critical data and to allow prompt detection for data inconsistencies on Early Access Program (EAP) studies and Non-Interventional Studies (NIS), by setting up a Risk Based Data Management (RBDM) approach with anticipated and periodic Visual Data Quality Review.

METHODS:

Test phase ran from July to December 2013 and RBDM was implemented since 2014 on Primary Data Collection NIS (PDC-NIS), Secondary Data Use NIS (SDU-NIS) or EAP studies. Visualization meeting are planned every 3 months from patients’ inclusion up to database lock. Data visualization is supported by TIBCO Spotfire® version 5.5 to release a visual presentation of clinical data. This visualization is based on SAS derived database resulting from the raw data. In order to standardize data visualization across studies, common database structure and derivation rules have been defined.

RESULTS:

Since 2014, 15 studies with RBDM were implemented: 13 PDC-NIS, 1 SDU-NIS, 1 EAP. 8 (53%) studies were carried out in oncology, 3 (20%) rheumatology, 2 (13%) anemia, 1 (7%) hematology and 1 (7%) HER2 diagnostic testing. Annualized meeting rate per study is in average 6 data visualization. Data review meeting duration decreased from 8 hours (before RBDM) to 2 hours using spotfire®. In average 52 graphs are presented to describe clinical data: 3.1% of graphs for demographic characteristics, 8.7% for patient’s vital status, 12.5% study treatment, 8.4% early termination, 20.2% for laboratory data, 11.5 % safety data and 27.8% dedicated to study specificities. All graphics produced allows interactivity from macro to micro visualization.

CONCLUSIONS:

Added value of RBDM allows highlighting inconsistencies on clinical data faster and earlier than with classical data review. Moving from static (paper review) to interactive visualization allows a significant saving in time to clean and lock database combined with high data quality. Interactivity is a key element by selecting subsets of data and identifying easily outliers.