Concurrent Validation

FDA considers concurrent validation to be a subset of prospective validation. The Agency recognizes that in a limited number of cases it may be impossible to complete validation of an API process in a timely manner when data from replicate production runs are unavailable because: - Only a limited number of API batches intended for clinical or orphan drug products have been produced; - API batches are produced infrequently (e.g., limited market demand, complex multi-step API processes with long completion times); or - APIs batches are produced by a modified process (e.g., a validated process goes outside the proven acceptable range of a critical operating parameter and the batch is subjected to intensive analytical tests).

In such cases, firms should do all of the following: - Document the reasons for not completing process validation before shipment of the API; - Perform all of the elements of prospective validation, as discussed in Section XIII.C., exclusive of replicate production run testing, before releasing any batch for distribution; - Conduct intensive in-process monitoring and testing, along with intensive testing of each API batch, to show that each production run resulted in an API meeting its predetermined quality characteristics and specifications (such data should also be assessed under the validation protocol to determine consistency of the process);.

- Provide for the Quality Control unit to evaluate batch production records, in-process controls, and analytical data from each process run to determine whether each batch should be released.

The level of intensive in-process and final API testing should be greater than levels for validated routine production runs, and should only be reduced to routine levels after the process has been determined to be validated. In addition, data from production runs that are evaluated as part of the validation studies should encompass the operating ranges that are approved for use during routine process control.

The Agency cautions that this validation approach should be applied with discretion so as not to: - Unduly delay completion of, or avoid performing, meaningful validation; or - Distribute API batches manufactured before completion of validation for a prolonged period of time.

This approach should not be viewed as a viable alternative where the number and frequency of API production runs permit timely completion of validation prior to API distribution. If analysis of data shows that the process used to manufacture the distributed batches was not, in fact, validated, no additional batches should be distributed until corrections have been implemented and the process has been determined to be validated.