OOS & OOT Investigations
An out-of-specification (OOS) result — a test result that falls outside the established acceptance criteria — is one of the highest-stakes events in a quality-control laboratory, because how it is investigated determines whether an unsafe batch is caught or a good batch is wrongly rejected (or, worse, a bad one wrongly released). The discipline is a structured, two-phase investigation that establishes whether the result is real before any decision is made, and it is governed in the US by a specific FDA guidance that grew out of the landmark court case that shaped the whole area. Out-of-trend (OOT) results — within spec but drifting — are the early-warning companion. This page covers both; the deviation machinery they feed is the [deviation management](/topics/deviation-management) explainer.
What an explainer is not
A topic explainer is SPEQ’s synthesis of what a practice involves, cited to the standards that govern it. It does not reproduce their text, and it does not determine which of them apply to your product or process.
[ POSITION IN THE FRAMEWORK ]
7 DIMENSIONS · 21 LINKSAn out-of-specification result is a question, not a verdict: a two-phase investigation decides whether the test or the process was wrong, across the GMP and quality-system disciplines, and retesting into compliance is the classic finding.
06 · QUALITY MATURITY — OOS & OOT INVESTIGATIONS, REACTIVE TO ADAPTIVE
An inconvenient result is retested until it passes; the initial OOS is discarded on 'probable analyst error' and no trend is watched.
An OOS SOP with two phases exists, but Phase I invalidations rest on generic justifications and OOT trending is absent.
Phase I finds a documented, scientifically justified cause before any retest; no cause means escalation to a full Phase II with root cause.
OOT limits from historical data catch drift while results still pass; investigations bound batch impact and feed CAPA and product quality review.
Statistically trended data pre-empts failures; OOS/OOT handling is a defensible, data-integrity-sound record, not a way to make results disappear.
SPEQ’s shared five-stage progression, labelled synthesis — not the FDA QMM rating scale. Where does your organization sit? Score your quality system →
07 · REGULATORY & EVIDENCE
GOVERNING STANDARDS · 3
Derived from the 3 standards SPEQ maps to this subject, across 3 regulatory bodies: FDA, ICH, PIC/S.
RECORDS & OBJECTIVE EVIDENCE
- Two-phase OOS investigation records with test solutions retained for Phase I
- A documented, scientifically justified assignable cause before any retest
- Phase II investigations reaching batch record, process, and other affected batches
- OOT limits derived from historical data and the trending behind them
- Batch-disposition decisions reflecting confirmed OOS results
COMMON INSPECTION FINDINGS
- Retesting into compliance without a justified assignable cause
- OOS results invalidated on generic 'probable analyst error'
- Averaging a failing original with a passing retest
- No Phase II when Phase I found no laboratory cause
- No OOT trending, so drift is only caught at outright failure
An OOS result is a question, not a verdict
The foundational mindset is that an OOS result does not, by itself, mean the batch failed — it means something produced a result outside specification, and that something might be the *product* or might be the *test*. Jumping straight from "OOS result" to "reject the batch" or, more dangerously, to "the test must be wrong, retest until it passes" both skip the investigation that is the entire point. The FDA guidance on investigating OOS results (finalised in 2006, with a Level 2 revision in 2022) exists precisely because the temptation to explain away an inconvenient result is so strong, and the framework it sets is designed to remove that discretion.
The historical anchor is the 1993 *United States v. Barr Laboratories* decision, which established much of the thinking now codified: a firm cannot simply retest into compliance, an initial OOS result cannot be discarded without a documented, scientifically justified cause, and averaging a passing retest with a failing original to reach a passing mean is not acceptable. That case is why the modern OOS investigation is a defined procedure rather than a matter of laboratory judgement — the rules exist to stop the result the analyst does not like from quietly disappearing.
Phase I: is the result even real?
The investigation runs in two phases, and Phase I is the laboratory investigation: before anything else, determine whether the OOS result is attributable to an identifiable laboratory error. The analyst and supervisor review the testing — the method, the calculation, the instrument, the standards, the sample preparation — looking for an assignable cause: a mis-dilution, a failed system suitability, a transposed value, a malfunctioning instrument. Critically, this review must happen **before the test solutions are discarded**, because they are the evidence, and it must be documented as it goes.
The rules on what Phase I can and cannot do are strict. A retest requires a *scientifically justified, documented* assignable cause — you may not retest merely because you dislike the result. Where a clear laboratory error is found, the OOS result may be invalidated and the retest becomes the reportable result, with the error documented and, where relevant, subject to CAPA so it does not recur. Where **no** assignable laboratory cause is found, the OOS result cannot be invalidated — and the investigation must escalate to Phase II. The most common and serious finding in this whole area is invalidating an OOS result on a weak or generic justification ("probable analyst error") to make it disappear; an inspector reading a pattern of conveniently invalidated OOS results is looking at a laboratory managing its data rather than its quality.
Phase II: the full-scale investigation
When Phase I finds no assignable laboratory cause, the OOS result must be treated as potentially valid, and the investigation widens into a full-scale (Phase II) investigation that reaches into manufacturing. Now the question is whether the process, not the test, produced the result: a review of the batch record, the process, other batches and products that could be affected, and the raw materials, conducted as a formal deviation investigation with a genuine root-cause analysis. Any additional laboratory testing in Phase II (such as a defined, pre-approved retesting or resampling plan) must be scientifically sound and decided in advance, not improvised to chase a passing answer.
The batch-disposition decision is the documented outcome of this investigation — and a confirmed OOS result on a validly manufactured batch is a genuine failing result that the batch record must reflect, whatever the commercial cost. The investigation also has to bound its own scope: the same root cause may implicate other batches, and product already released may be affected, so Phase II must reach far enough to answer "what else is exposed?" before it closes. This is where OOS handling connects to the wider quality system — it feeds deviation management and CAPA, and its integrity is a data-integrity concern in its own right, because an OOS investigation is a contemporaneous record of something that went wrong.
OOT: catching the drift before it fails
An **out-of-trend (OOT)** result is different from OOS and just as important: it is a result that is still *within* specification but is atypical compared with the expected pattern — a stability data point drifting toward a limit, a release assay creeping across batches, a value inconsistent with historical performance. OOT is defined relative to a trend, not a fixed limit, so detecting it requires statistically informed limits (control or alert limits derived from historical data), not just the specification. A result can be perfectly in-spec and still be a clear OOT signal that something is changing.
The value of OOT monitoring is that it is a *leading* indicator: it catches the drift while the results are still passing, giving the chance to investigate and correct before a batch actually fails specification or a stability trend forces a shelf-life reduction. A quality system that only reacts to OOS results is inherently late — it waits for failure — whereas one that also trends for OOT is watching the approach to failure. The two are complementary: OOT investigation is the early warning; OOS investigation is the response when the warning was missed or the change was abrupt.
FREQUENTLY ASKED
What is an OOS result and does it mean the batch failed?
An out-of-specification (OOS) result is a test result outside the established acceptance criteria. It does not by itself mean the batch failed — it means something produced a result outside spec, and that something might be the product or the test. The investigation exists to determine which before any disposition decision is made. Jumping straight to rejection, or to "the test must be wrong," skips the entire point.
What are the two phases of an OOS investigation?
Phase I is the laboratory investigation: determine, before discarding the test solutions, whether the result is attributable to an identifiable, documented laboratory error. If a scientifically justified assignable cause is found, the result may be invalidated and a retest performed; if not, the result cannot be invalidated. Phase II is the full-scale investigation into manufacturing — batch record, process, other batches, raw materials — treating the OOS as potentially valid, with a documented root cause and a batch-disposition decision.
Why can’t you just retest an OOS result until it passes?
Because retesting into compliance is prohibited — a principle codified after the 1993 United States v. Barr Laboratories decision and the FDA OOS guidance (2006, Level 2 revision 2022). A retest requires a documented, scientifically justified assignable cause; you may not retest merely because you dislike the result, discard an initial OOS without justified cause, or average a failing original with a passing retest. Invalidating OOS results on weak justifications is the classic finding.
What is the difference between OOS and OOT?
An OOS (out-of-specification) result falls outside the acceptance criteria. An OOT (out-of-trend) result is still within specification but atypical relative to the expected pattern — a stability point drifting toward a limit, an assay creeping across batches. OOT is defined against a trend using statistically informed limits, not the specification, and it is a leading indicator that catches drift while results still pass, before an actual OOS failure.