permitted window

Duplicated entries: Subject entered multiple times in EDC system

Key Data Cleaning Techniques

1. Edit Checks and Validation Rules

Edit checks are predefined logical conditions programmed into the EDC system. They automatically flag invalid or inconsistent data during entry. Types include:

• Range checks (e.g., age between 18–65)
• Date logic checks (e.g., visit date after screening)
• Cross-field logic (e.g., if “Yes” to Adverse Event, then Event Description is required)

2. Manual Data Review

Clinical Data Managers (CDMs) or CRAs review data manually to detect discrepancies not captured by automated checks. This includes:

• Checking for narrative consistency in adverse events
• Reviewing lab trends over time
• Confirming consistency in visit dates and dosing intervals

Manual review requires training in GMP quality control principles and familiarity with protocol nuances.

3. Query Management

When inconsistencies are detected, queries are raised to the site via the EDC system. Effective query management includes:

• Clear, concise wording of queries
• Timely follow-up and closure
• Root cause identification for recurrent issues

4. Source Data Verification (SDV)

SDV ensures that data in the CRF matches the original source documents (e.g., patient medical records). Monitors perform SDV either 100% or based on a risk-based monitoring strategy.

According to Pharma SOP templates, SDV processes should be well-documented and follow GCP guidelines.

5. Data Reconciliation

This involves matching data across multiple systems such as:

• CRF vs lab data
• SAE database vs AE fields in the CRF
• IVRS/IWRS (randomization systems) vs dosing records

Automated reconciliation tools can flag mismatches that require manual resolution and documentation.

Tools Used in Data Cleaning

• EDC Platforms (e.g., Medidata Rave, Oracle InForm)
• Clinical Trial Management Systems (CTMS)
• ePRO/eCOA platforms
• Excel or SAS for data export and analysis
• Custom scripts and macros for automated checks

Documentation and Compliance

All data cleaning activities should be traceable. Maintain:

• Data Cleaning Log
• Query Tracking Sheets
• SDV Reports
• Audit Trail Reports from the EDC

These are critical during audits and inspections and support compliance with Stability Studies requirements for reliable data storage and documentation.

Best Practices for Efficient Data Cleaning

1. Develop a Data Management Plan (DMP) that outlines cleaning processes
2. Conduct mid-study reviews to detect and prevent accumulating errors
3. Train sites in accurate data entry and protocol compliance
4. Involve biostatisticians early to align with analysis plans
5. Use standardized coding dictionaries (e.g., MedDRA, WHO-DD)

Challenges in Data Cleaning

• Over-reliance on automated checks without manual review
• High query volumes that delay database lock
• Inadequate site training and misinterpretation of CRFs
• Protocol amendments that affect data consistency

Conclusion

Data cleaning is a multi-layered process that involves technology, expertise, and meticulous attention to detail. By applying the right techniques—from edit checks and query management to SDV and reconciliation—clinical teams can ensure high-quality datasets that withstand regulatory scrutiny and support reliable trial outcomes. Integrating these methods with robust documentation and stakeholder training is key to achieving clinical data excellence.