OOS vs OOT: Handling Results Without Overreacting
Introduction
Not every strange result is a failure. OOS needs containment and investigation; OOT needs statistical thinking and trend awareness. Overreacting either way burns time and credibility.
Definitions
- OOS (Out‑of‑Specification): Result outside a predefined acceptance criterion.
- OOT (Out‑of‑Trend): Result within spec but abnormal relative to historical data.
Handling OOS (simplified)
- Immediate checks for obvious/assignable error
- Retest per SOP (pre‑defined, not ad‑hoc)
- Full investigation if unresolved
- Disposition and potential CAPA
Handling OOT (simplified)
- Confirm data integrity and calculations
- Analyze trend and context (environment, lot, operator)
- Scientific assessment and monitoring plan
- Consider change control if a process shift is suspected
Pitfalls
- Retesting until pass without rationale
- Treating OOT like OOS (or vice versa)
Documentation
- Data sets reviewed, hypotheses, tests performed
- Statistical rationale and conclusions
- Links to CAPA/change if applicable
How an eQMS helps
AI Assist: compresses long analytical narratives into a clean executive summary
Attachment control and versioned data
Trails on every change
