Best Practices: Using Checks to Automate Data Cleaning
Veeva Clinical Data offers several tools for automating data validation. In Veeva DQS, Checks automate repetitive data review tasks that do not require human judgment. By continuously evaluating study data, DQS Checks automatically create and close queries, or identify protocol deviations across EDC and third-party sources. Incorporating DQS Checks into your data review and cleaning strategy eliminates unnecessary manual review, allowing data managers and medical reviewers to focus on discrepancies that require their attention and more detailed review.
Use the following decision guide and best practices to identify when and how to effectively deploy DQS Checks.
Review Methods in Veeva Clinical Data
The Veeva Clinical Data system has multiple methods for identifying discrepancies and creating queries to ask sites to confirm or correct data.
- Veeva EDC Rules trigger simple system actions based on the data that is entered.
- Veeva DQS Checks evaluate more complex but deterministic data relationships and automatically open queries or identify protocol deviations.
- When human judgment is required before opening a query, you can use targeted review listings in Veeva DQS. Targeted review listings are centered on a single review objective, allowing data managers and medical reviewers to focus on discrepant records.
Depending on the reality of the study implementation, and your goal for the data cleaning process, one of these methods will best meet your needs.
| Review Method | Manual | Mechanism | Optimal Use Case |
|---|---|---|---|
| EDC Rules | No | Programmed in Veeva EDC, with actions automatically triggered based on entered data. | Same-form or simple cross-form validation. |
| DQS Checks | No | Configured in Veeva DQS, with query and protocol deviation actions triggered. | Complex logic, third-party or lab data integration, rules that require coded terms, building in time for the site to enter data before a query is opened. |
| DQS Review Listings | Yes | Configured in Veeva DQS, Review listings require human evaluation and manual actions. | Any discrepancies that require human judgement before creating a query or identifying a protocol deviation. |
Selecting the Right Review Method
When choosing which review method best suits your goals, consider the following:
- Data origin: does the discrepancy involve only EDC data, or is non-EDC data involved?
- Logic complexity: is the logic highly complex, does it reference multiple Forms, or could criteria change requiring study design changes if an EDC Rule is used?
- Manual review: Is clinical review needed before creating a query?
- Other considerations: Does the logic require use of coded terms? Do you want to build in a grace period for sites to complete data entry before the query is created?
In most cases, you should use an EDC Rule to create site queries whenever technically feasible. Similarly, we recommend that all validation to identify protocol deviations be performed in DQS. EDC rules execute immediately during data entry, delivering the fastest real-time feedback to site personnel before forms are submitted. Follow the decision framework outlined here to determine whether to use an EDC Rule, a DQS Check, or a DQS Review Listing.
Evaluate Data Origin and Review Objective
When choosing the review method that best suits your situation, you should first consider the origin of the data and your objective. The following table outlines the recommended method based on certain data origin and review objective criteria:
| Data Origin | Review Objective | Recommended Method | Benefit |
|---|---|---|---|
| EDC Data Only | Real-time site validation during data entry | EDC Rule | Provides immediate query to site. |
| EDC Data Only | Complex logic, coded terms, or delayed grace periods | DQS Check | Executes in DQS without requiring study design changes. |
| EDC Data Only | Discrepancy requires human evaluation | DQS Review Listing | Allows reviewers to review before manually creating a query. |
| Non-EDC / 3PD | Discrepancy automatically requires a query 100% of the time | DQS Check | Continuously scans incoming third-party data on an hourly schedule. |
| Non-EDC / 3PD | Discrepancy requires human evaluation | DQS Review Listing | Allows reviewers review before manually creating a query. |
Determine Logic Complexity
The complexity of the logic you want to use when evaluating data will also impact your choice of review method. While EDC Rules are suitable for same-form or cross-form checks, more complex logic may require the use of DQS Checks. The following table outlines the recommended review method depending on your review scenario and validation criteria.
| Review Scenario | Evaluation Criteria | Recommended Method | Benefit |
|---|---|---|---|
| Low-complexity | Straightforward same-form or multi-form validations. | EDC Rules | Immediate feedback to site. |
| Complex | Multi-form logic, or complex calculations. | DQS Check | Advanced logic capabilities. |
| Study design change risk | Logic would require significant study design changes. DQS Checks can be created and updated without a post-go-live change in EDC. | DQS Check | No impact to study design. |
Confirm Whether Manual Review Is Needed
While Veeva DQS Checks greatly reduce time spent on manual review, when human judgment is required, you should create a review listing. In Veeva DQS, review listings can be targeted around a single review objective, only outputting discrepant data for streamlined and focused review. The following table outlines the recommended review method depending on the type of discrepancy and the evaluation criteria.
| Discrepancy Type | Evaluation Criteria | Recommended Method | Benefit |
|---|---|---|---|
| Predictable discrepancies | Data inconsistency always requires a query or Protocol Deviation every time it occurs | DQS Check | Automates query or PD generation without requiring human intervention. |
| Qualitative or subjective discrepancies | Discrepancy requires clinical evaluation, medical interpretation, or cross-referencing before issuing a query | DQS Review Listing | Allows reviewers to evaluate context and directly issue manual queries or Listing Defined or Confirm Value Quick Queries. |
Other Considerations
Other considerations about when to use a DQS Check versus an EDC Rule or a Review Listing include the following:
- Coded terms: medical dictionary coding is often unavailable within EDC Rules. If your validation involves coded terms, such as flagging overlapping adverse events and concomitant medications, use a DQS Check.
- Grace period: EDC Rules are triggered on data entry or Form submission. In some workflows, such as when Form B is required after Form A is submitted, an EDC rule can create instant system queries while site staff are still entering data. Configuring a DQS Check can allow for a grace period for site staff to complete data entry, without creating additional unnecessary queries.