Regulations and compliance · Practical guide

OOS result, invalid test or missing data: classifying findings

An out-of-specification result, an invalid test and missing data are distinct findings that may occur together. A later passing value does not automatically invalidate an OOS result. A missing record gives no measured OOS value. The laboratory and QA should secure the available evidence and establish the basis for each decision.

Table of contents

At a glance

When is a finding an invalid test, a missing dataset or an out-of-specification result?

Distinguish the observed result, validity of the test and completeness of its data. In the FDA scope discussed, OOS means a result outside established specifications or acceptance criteria. Invalidation needs a sound scientific basis. Missing data require investigation of their cause and impact. Preserve the original finding and investigation in the required record context. Repeats or averages must not obscure an unresolved original result.

  • Evaluate results status, test validity, and data completeness separately.
  • Invalidation is not based solely on a later appropriate value.
  • Investigation, potential product impact, and further testing decisions are clearly traceable and justifiable.

1. Start with Method, Specification and Scope

The FDA OOS Guidance in the May 2022 revision addresses specifically chemistry-based laboratory tests for drugs within the CDER domain. It is not a generic procedure applicable to every biological assay, every clinical measurement, or every technical software test result. The specific method and applicable quality framework must be clearly defined.

In the field of medicinal products for human use, EU-GMP Chapter 6 describes requirements for testing methods, records and the examination of conspicuous results. The EU and FDA sources support a careful investigation, but are not interchangeable in all details. Therefore, prior to classification, define which result is evaluated against which criterion according to which method.

Sources: [1] [2]

2. Separate three evaluation axes

OOS first describes a relationship between result and defined criterion. Invalid, on the other hand, describes a well-founded assessment of the usability of an audit or a result. Missing data describes a gap in the existing evidence. None of these statements automatically replaces the others.

Our own matrix assists in the initial classification. It is not a procedure for automated batch decision-making. An out-of-specification (OOS) result may be attributable to a clearly documented laboratory error. A missing data set may have contained a previously generated OOS result or may stem from a test that was never performed. Without investigation, this distinction remains unresolved.

Worksheet for distinguishing findings
ObservationWhat is demonstrated by this isWhat has not yet been demonstratedNext question
Value outside the criterion The corresponding result was obtained Cause and final product decision Is the inspection correct; what explains the finding?
Documented execution error A specific event may have influenced the inspection. Automatic invalidation of all results Which results are causally affected?
Required file missing The existing evidence is incomplete Whether the sample has passed or failed Was the item tested, recorded, transferred or deleted?
Later value fits Another value meets the criterion The original finding is contradicted. Which scientific explanation links both?

Sources: [1] [3] [2]

3. Document the initial conditions prior to further testing

Maintain records of results, method, sample, equipment used, and relevant procedural conditions. Preserve required data and investigation context. In the case of unexpected results, available sample aliquots or solutions for root cause analysis may be valuable, provided stability and the method permit.

Avoid a rapid restart that obscures the original state. For electronic systems, sequence information, relevant events, or existing files can contribute to investigation. The specific information required depends on the individual finding. A comprehensive list of all technical logs is equally inappropriate, as is the blanket rejection of all data from a non-conforming test.

Sources: [1] [3] [4]

4. Example: three similar reports, three investigations

In the example, a laboratory performs a chemical content assay with an assumed acceptance range of 95.0 to 105.0 percent. Case A yields 94.0 percent with no immediately apparent execution error. Case B involves a documented loss of sample solution prior to completion of the assay. In Case C, the result file is missing, although the task is displayed as completed.

These cases are not aggregated into a single error code. The table outlines the technical investigation assignment. The example does not refer to any specific test method; the numbers are provided solely for clear and unambiguous differentiation.

Example: three completed laboratory test results
CaseInitial findingAppropriate investigationPremature conclusion that the evidence does not support
A: Result 94.0 percent within the range of 95.0–105.0 percent Laboratory evaluation; without clear cause further examination by method A value recorded later within the range renders 94.0 automatically invalid.
B: Implementation Sample solution partially spilled prior to completion, event documented Rationale for impact on testing and existing data; define subsequent actions All existing data may be deleted
C: Completeness Result file missing, processing status available Scope, storage, transmission, and any other relevant data to be clarified Missing file automatically indicates OOS or passed
Joint Conclusion Causes and implications may differ. Investigation, evidence collection, and decision-making integrated A common green status replaces the investigation assessment

Sources: [1] [3] [2]

5. Provide a scientific rationale for a declaration of invalidity

The investigation is required under § 211.192 of the US Drug CGMP regulations. The FDA OOS Guidance clarifies when a single result may be rejected: a observed and documented event must reasonably be determinable as the cause of the OOS result. A subsequent appropriate value or a hypothesis regarding a operator error is insufficient for this determination.

Assess each documented event according to its impact. Loss of a sample solution before the test is complete calls for a different response from an unexplained low result. Identify the affected steps, data and conclusions. Retain the original record and the reasoned assessment. Marking a result invalid is not authority to delete it.

Sources: [1] [3] [5]

6. Repeat testing with a clear question statement

Additional tests may support the investigation. The FDA distinguishes, for example, retesting from the original sample material and a new sample collection. The decision requires a scientific rationale and an appropriate method. Repeating the test until a suitable value appears does not constitute a credible root cause analysis.

In Example A, a further value of 98.0 percent could arise. Although the average of 94.0 and 98.0 is 96.0 percent, this calculation does not render the original out-of-specification (OOS) finding invalid and does not, by itself, justify an acceptable product decision. The calculation of a methodologically defined mean and the retrospective blending of OOS and repeat values represent distinct scenarios.

Sources: [1]

7. Investigate missing data without preconclusion

In case C, it must first be established whether the test was performed and whether records have been generated. A missing transfer may require a different corrective action than an interrupted measurement or an unauthorized deletion. The investigation must neither automatically assume tampering nor dismiss the gap as merely an IT issue.

Check the scope: Are additional orders, time periods, or interfaces affected? Can the required statement be reconstructed from appropriate original documents or controlled copies? If the basis for a quality decision remains incomplete, this limitation must be explicitly addressed in the subsequent decision. A newly generated file is not automatically a substitute for the missing original context.

Sources: [3] [4] [2]

8. Checklist for traceable conclusion

The conclusion integrates the findings, investigation, and decision. It should indicate which results are used, which are not used despite being justified, and which data gaps remain. The decision regarding product or batch follows the relevant quality procedure and cannot be derived solely from a simple inspection table.

The following internal reference document supports the discussion between laboratory, QA, and IT. It separates distinct investigation questions while maintaining awareness of their potential interrelationships. A confirmed error may additionally necessitate an evaluation of other affected tests or products.

  • Method, specification, and applicable scope defined.
  • Observation of root cause and interpretation separately.
  • Original data and investigation context secured.
  • Assess invalidation against a specific scientific justification.
  • Additional tests with defined objective and methodology.
  • Missing data and potential scope assessed.
  • Usage of results, open items, and follow-up actions are clearly decided and traceable.

Sources: [1] [3] [2]

Project experience and collaboration

Preserve the initial finding before selecting the investigation path

With the laboratory, QA and the system team, we distinguish what a prepared finding actually shows: a measured value outside specification, a demonstrated test malfunction or a data gap. A known technical fault needs different evidence from an unexplained result deviation. We examine the records and connect system events to the substantive investigation. The authorised parties decide on invalidation, product impact and further action based on the complete investigation.

Frequently asked questions about Distinguishing OOS results, invalid tests and missing data

Is every invalid test result considered an out-of-specification (OOS) result?

No. An invalid or incompletely performed test may result in no usable result value. An OOS assumes a value falls outside a predefined criterion. Both scenarios may be interrelated, but must be distinguished based on their actual evidential basis.

Sources: [1]

Is a missing file automatically considered an out-of-specification?

No. The missing file initially creates a data gap. Investigate whether it has been verified which data were generated and why they are missing. Without this clarification, neither a valid result nor an out-of-specification (OOS) value can be derived solely from the absence of data.

Sources: [3] [4]

Can an appropriate repeat value substitute the first result?

Not without robust scientific justification. A later value does not automatically explain the cause of the initial finding. The investigation must justify the use of the results and consider all relevant findings. Testing into compliance is not an appropriate approach.

Sources: [1]

When can a laboratory error lead to a withdrawal of approval?

When an investigation identifies a specific cause and the link between that cause and the affected result is demonstrable, a general assumption is insufficient. Additionally, document the scope and required corrective actions to ensure the same error is not overlooked in future testing.

Sources: [1] [3]

Must invalid data be deleted?

No. Invalidation and storage are distinct decisions. The original finding and its justified evaluation remain retained within the required documentation context. Specific storage requirements must be determined separately; an unlimited general deadline is not asserted here.

Sources: [3] [4]

Why is the average 96.0 percent not enough in the example of this article?

It arises from 94.0 and 98.0 percent, but does not address the root cause of the original out-of-specification (OOS) result. Post-hoc averaging using repeated values may mask the relevant deviation. Methodologically intended mean calculation must be evaluated according to its own criteria and principles.

Sources: [1]

Must every data gap be evaluated as intentional manipulation?

No. Causes can be technical, organizational, or behavior-related. A factual investigation must establish the actual root cause and impact. It must neither prematurely assume intent nor exclude the possibility of data integrity compromise without proper evaluation.

Sources: [3]

Does the FDA Out-of-Specification Guidance apply to all laboratory test methods?

Your focus is on chemistry-based drug testing within the CDER domain. This approach is not universally applicable to all biological assays, PAT methods, or clinical measurements. Determine the appropriate framework and specific method prior to adopting individual recommendations.

Sources: [1]

Is rejecting a batch enough to close the issue?

No. In the US Drug CGMP domain, § 211.192 requires a thorough investigation that may also include other batches of the same medicinal product and other medicinal products, if they could be related to the defect. Conclusions and corrective actions must be documented in writing. The FDA OOS Guidance confirms that the rejection of a batch does not substitute for such an investigation. Therefore, a previously made product decision does not conclude the root cause and scope determination.

Sources: [5] [1]

What must IT contribute to the absence of laboratory records?

IT can support traceability of creation, storage, transmission, and potential restoration. The meaning in the process of the absence of proof should be assessed by the laboratory and the relevant quality responsibility personnel through a structured process. Successful technical restoration alone does not constitute complete and usable validation evidence.

Sources: [3] [4] [2]

Sources and review date

Source date: 7 October 2026. Regulatory requirements, guidelines, and industry practices must be categorized separately. Tables, figures, and case examples are to be considered standalone reference materials; they do not constitute completed product testing.

  1. FDA: Investigating Out-of-Specification Test Results for Pharmaceutical ProductionFinal Guidance, Level 2 Revision May 2022; CDER, chemistry-based drug product testing.
  2. EU-GMP Chapter 6 – Quality ControlRevision 28.03.2014, effective from 01.10.2014; current human medicinal product index, particularly §§6.7–6.10 and 6.15–6.17.
  3. FDA: Data Integrity and Compliance With Drug CGMP – Questions and AnswersFDA Data Integrity Q&A, December 2018: Question 2 regarding the retention of invalidated results in the complete record, 9 concerning appropriate copies, and 12 on data storage including incomplete or discontinued entry.
  4. 21 CFR §211.194 – Laboratory recordseCFR notice until 05.10.2026; paragraphs (a)(1)–(8), (b)–(e).
  5. eCFR: 21 CFR § 211.192 – Production record reviewRead on 18 October 2026: Investigation of unresolved deviations and specification violations, expansion to potentially affected additional batches and products, written conclusions and follow-up actions. The FDA OOS Guidance explains this procedure, but does not supersede it.

Sources checked: .

Daniel Herrmann, Daniel Herrmann Consulting

Daniel Herrmann Consulting

Jointly develop the specific evidence process

We support the laboratory, QA, and IT teams in distinguishing clear results and data gaps, and in establishing a traceable investigation and documentation process.

  • Computer system validation →

    Bring your GxP systems into validated operation, from laboratory software, MES and eQMS to ERP. We define the scope, plan risk-based tests using GAMP 5 and hand the evidence to QA for release.

Send a project enquiry Customer testimonials →