Assessment: Data Review and Deviation Decisions
Domain 3: Assessment
Assessment tests how you evaluate batch data, deviations, out-of-specification (OOS) results, and environmental monitoring trends, and whether the evidence supports continuation, hold, or release. The Craftsman Bio Process exam emphasizes defensible conclusions over convenient ones.
Core rules
- Apply ALCOA+ data integrity: attributable, legible, contemporaneous, original, accurate, plus complete, consistent, enduring, and available.
- Review critical process parameters and quality attributes together. A passing final result does not excuse an uncontrolled parameter excursion.
- Classify deviations by impact. Minor, major, and critical classifications drive investigation depth and batch disposition.
- Use structured root-cause methods such as 5 Whys or fishbone diagrams, and verify CAPA effectiveness rather than closing on paperwork alone.
- OOS investigations follow phases: a Phase I laboratory investigation first; a Phase II manufacturing investigation if laboratory error is excluded. Retesting to obtain a passing result is not an investigation.
- Trend environmental monitoring data; a single in-limit result can still be part of an adverse trend.
Worked example
A 72 h bioburden result is 12 CFU against a limit of 10 CFU. Correct action: treat it as OOS, place the batch on hold, and start a Phase I laboratory investigation - check sample handling, incubation conditions, and enumeration error. Do not simply retest until a passing result appears. If laboratory error is excluded, expand to a Phase II manufacturing investigation covering hold times, cleaning, and operator technique. Reasoning: the first result is the original data; retesting without investigation destroys the ability to determine cause.
Traps that cost marks
- Averaging multiple results to bring a failing value inside a limit.
- Closing a deviation without verifying CAPA effectiveness.
- Ignoring an upward environmental monitoring trend because each point is in limit.
- Treating an OOS as a laboratory problem before any Phase I review.
- Reviewing final release results while ignoring in-process excursions.