If you encountered a significant discrepancy in patient data during a clinical trial, what steps would you take to address it?