Answer Framework:
OOS: Any analytical result that falls outside established acceptance criteria (registered specifications or compendial limits).
OOT: A result that remains within registered specifications but deviates noticeably from historical batch trends, stability projection curves, or statistical process limits (e.g., beyond $\pm 3\sigma$).
Approach: OOTs serve as early warning signals in stability testing or continued process verification (Stage 3). QA triggers proactive root-cause analysis before an actual out-of-specification failure occurs.