CRF Design for a Complex Dosing Regimen
A Phase I study uses a dose-escalation design with three dosing schedules: once daily (QD), twice daily (BID), and three times daily (TID). The protocol allows dose modifications at the investigator's discretion, including dose reduction, interruption, and re-escalation. The CRF currently has a single 'Dose' field and a 'Frequency' field with a free-text option. The data manager notices that sites are entering dose changes in the comments field rather than in structured fields. The study has 120 subjects enrolled across 15 sites. The database is due to be locked in 8 weeks.
1. What are the key CRF design changes needed to capture dose modifications accurately, and what are the risks of making these changes at this stage?
Read the worked answer
The CRF needs structured fields for each dose modification: date of change, previous dose, new dose, reason for modification (coded list), and whether the modification was a reduction, interruption, or re-escalation. A repeating log form for dose modifications is preferable to a single field. The frequency field should be a coded list (QD, BID, TID) rather than free text. Risks of changing the CRF at this stage include: (1) sites may need to re-enter historical data, which is time-consuming and error-prone; (2) the database build may require validation and testing, potentially delaying lock; (3) mid-study changes can introduce inconsistency between data collected before and after the change. Mitigation: implement the change with a clear effective date, provide training to sites, and use a data clarification process to migrate existing free-text comments into the new structured fields. The data manager should assess whether the change is essential for analysis; if the free-text comments contain the necessary information, a post-hoc coding step may be sufficient.
2. How should the data manager handle the existing free-text dose modification data?
Read the worked answer
The data manager should first extract all free-text comments related to dose modifications and review them for completeness. A query should be sent to each site asking them to transcribe the dose modification details into the new structured fields, using the comments as a reference. The data manager should provide a clear instruction and a deadline. If sites do not respond in time, the data manager may need to escalate to the CRA. The original free-text comments should be retained in the database for audit purposes. The data manager should also perform a quality check comparing the structured entries against the free-text to ensure accuracy. If the volume is large, a sampling approach may be used, but all entries should be reviewed for critical fields like dose amount and date.