About the Discrepancy Database

Oracle Clinical checks that all collected data conforms to restrictions built into the study by question definitions and Validation and Derivation Procedures. The system creates a Discrepancy record in the Discrepancy Database, either during manual data entry or update, batch data load, or Batch Validation, if the response to a question is invalid. Discrepancies usually result from human error: either transcription errors during data entry or illegible CRFs. They may also indicate flawed logic in Validation or Derivation Procedures or in question definitions, or real, though unexpected, clinical data.

Managing discrepancies includes reviewing them, investigating their cause, and either resolving them or declaring them irresolvable, so that your clinical data is as clean and complete as possible for analysis and submission to regulatory agencies. Resolving discrepancies most often means correcting incorrectly entered patient data, but may also mean editing Validation Procedures or question definitions.

To rerun changed Procedures, recheck data against changed question definitions, and re-evaluate changed data, run Batch Validation (see Batch Validation Process).

The system stores discrepancies separately from patient data, but associates each discrepancy with its Patient, Visit, Lab and other Received DCM header information and the specific response(s) that triggered its creation. Review statuses are used to track the process of resolving discrepancies; the past and current status(es) and related comments are also stored with each discrepancy (see Single-Record View).

You can use profiles to control access to data entry update and to filter queries on the Discrepancy Database. See Selecting a Profile.

Oracle Clinical offers many utilities to help you resolve discrepancies, including:

• Viewing CRF Images: If your company has set up this feature, you can view an image of the actual clinical report form that includes the discrepant response. From the Maintain Discrepancy Database window's Special menu, select Get Image. See Browse and Update Patient Data.

• Data Clarification Forms (DCFs): To facilitate handling of large numbers of discrepancies, you can group them by review status, Investigator, site, DCM, visit, or any combination of these criteria, print a report based on the criteria you choose, and send it to an Investigator for review (see Using Data Clarification Forms). DCFs contain discrepancies related to the data of only one patient.

• Group Updates: Query for and hand-select a group of discrepancies and simultaneously update their review status, comment text, or CRF page numbers, mark them for passive review, or create DCFs for them. See Selecting a Group of Records to Change.

• Mass Changes: Use SQL statements to select a set of patient data and make the same correction to all of it at once. See Making Mass Changes.

• Reports: From the Conduct menu, select Conduct Report, then select Data Validation to run many data validation reports.

You may also use the Oracle Thesaurus Management System, which compares patient data responses to terms in standard dictionaries such as MedDRA and WHO-Drug (or your organization's internal dictionary) and, when fully integrated with Oracle Clinical, derives standard dictionary terms for Oracle Clinical responses.

To open the Maintain Discrepancy Database window, from the Conduct menu, select Data Validation, then select Discrepancy Database. To set up the system and administrative settings that control access to the discrepancy database, see the Oracle Clinical Administrator's Guide.