What should a PPQ protocol contain?

Why

The FDA's 2011 guidance lists recommended protocol elements. These are the manufacturing conditions including operating parameters, processing limits and raw material inputs, and the data to be collected with when and how they will be evaluated. Next come the tests (in-process, release and characterization) with acceptance criteria for each significant step, and the sampling plan, with sampling points, number of samples and frequency, adequate for statistical confidence within and between batches and more extensive than routine.

It also covers criteria and statistical methods for intra-batch and inter-batch variability, provisions for deviations and nonconforming data, and confirmation that equipment, training and material sources are qualified. The status of analytical method validation and approval by the quality unit complete the list. Annex 15 adds critical process parameters and quality attributes with their limits, non-critical parameters to be monitored and why, equipment with calibration status, in-process controls with reasons, and how batches will be released.

The protocol must be approved before execution, and departures need approval before they happen. Data cannot be excluded without documented scientific justification, which is why that provision is written into the protocol in advance. A protocol that says only "three batches will meet specification" misses the point of PPQ, which is to show the process is consistent, not just that each batch passes release testing.

How to work it through

1. Start from the control strategy and list the CQAs and CPPs with their limits.
2. Define the tests, sampling points and number of samples for each unit operation, with the statistical rationale.
3. Write acceptance criteria that address variability within and between batches, not only specifications.
4. State how deviations, invalid data and departures from the protocol will be handled.
5. Confirm equipment, methods, training and suppliers are qualified.
6. Obtain approval from all relevant departments and QA before the first batch.

Where the answer stops holding

A protocol cannot anticipate every event, so the deviation provision matters. Statistical criteria depend on assumptions about distributions and sample sizes that small batch numbers may not support. Protocols copied from another product often keep criteria that do not fit the new process.