What is spike recovery? Assessing Assay Accuracy in Method Validation
Spike recovery is the measured increase in analyte attributable to a known addition to a sample, expressed as a percentage of the amount added on the same measurement basis.

On this page · 11 sections
- 1What spike recovery measures—and what it leaves unproven
- 2ICH Q2(R2): accuracy evidence across the reportable range
- 3Designing three levels, independent preparations and appropriate controls
- 4Calculating percent recovery: two cases with native-background correction
- 5Matrix effects are not the same as standard addition
- 6Setting recovery acceptance criteria without universal percentage limits
- 7Extraction efficiency: comparing pre-extraction and post-extraction spikes
- 8Laboratory evidence, 21 CFR 211.194 and product-specific regulatory scope
- 9When recovery fails: validation decisions and OOS investigations
- 10Moving study evidence between systems without losing calculation history
- 11A practical checklist for implementation, review and change assessment
How does Spike Recovery (Assay Accuracy) apply to your shop floor?
Pick your industry and scale — Ask V5 rewrites the definition in your context, gives a worked example, and shows what V5 does on day one.
01What spike recovery measures—and what it leaves unproven
Spike recovery measures how much of a known analyte addition an analytical procedure detects in a test sample, expressed relative to the amount added after accounting for any native analyte.
Related terms include spike and recovery for the experiment, percent recovery for its numerical result, and recovery study for the planned set of preparations and measurements.
The result concerns the added analyte under the tested conditions: the selected matrix, concentration, fortification point and procedure. Favorable recovery does not independently establish specificity or precision. Nor does it prove equivalent recovery of analyte already embedded in the original sample; a freshly added standard may behave differently from material incorporated during manufacture or naturally present in a botanical.
| Term | Main question | Boundary |
|---|---|---|
| Spike recovery | How much added analyte is measured? | Tested spike conditions only |
| Analytical accuracy | Agreement with an accepted reference value? | Broader than one recovery result |
| Precision | Agreement among repeated determinations? | Consistency, not necessarily correctness |
| Extraction recovery | Retention through sample preparation? | Preparation losses, not total assay accuracy |
| Matrix effect | Alteration of analytical response by matrix components? | Response influence, not a calibration technique |
| Standard addition | Calibration using graded additions to the sample? | Calibration approach, not proof of extraction |
An individual result is therefore one observation, not an accuracy conclusion for the method’s entire intended use. Supporting that conclusion calls for a study addressing relevant concentrations, matrices and preparation conditions, with sufficient independent determinations to judge performance against predefined criteria. In assessing a scientifically valid method, ask whether the experiment represents the analytical task—not merely whether its recovery number looks favorable.
02ICH Q2(R2): accuracy evidence across the reportable range
ICH Q2(R2), section 3.3.1 recommends establishing accuracy across the reportable range under regular analytical-procedure conditions, including the sample matrix and the described preparation steps. The practical question is therefore not simply whether an added standard can be measured, but whether the experiment represents how routine samples become reported results.
For a recovery study, ask: do the selected matrices, analyte levels and preparation conditions represent the decisions the procedure will support? As a study-design practice, map each selected condition to its intended use. A convenient matrix or concentration should not substitute for a scientifically justified selection. This mapping also gives reviewers a useful starting point during analytical method transfer.
Section 3.3.1.4 identifies mean percent recovery of a known added amount of analyte as one accuracy-reporting approach. It recommends accompanying the result with an appropriate confidence interval or justified alternative statistical interval. Percent recovery thus connects the known addition to the accuracy summary; interval interpretation and acceptance decisions are addressed in the acceptance-criteria section.
Spiking is not the only route to accuracy evidence. Section 3.3.1 allows accuracy to be inferred in certain cases once precision, response within the range and specificity have been established. That qualification matters: it is not a blanket exemption from demonstrating accuracy.
Keep the status and scope of ICH Q2 clear. This is pharmaceutical-validation guidance; section 1.2 identifies release and stability testing of commercial drug substances and products as its scope. It is not a universal legal mandate to run spike-recovery studies for every regulated product. Product-specific legal applicability is addressed in the laboratory-records section.
03Designing three levels, independent preparations and appropriate controls
ICH Q2(R2), section 3.3.1.4 gives three concentrations with three replicates each of the full analytical procedure as an example for assessing accuracy across the reportable range—not a mandatory design for every method. The following operational recommendations are study-design practice, adapted to the method and matrix.
Choose low, middle and high levels against the actual reportable range and intended decision: routine assay, impurity quantitation or another defined use. Connect that rationale to analytical procedure development. Label each level explicitly as amount added, total sample concentration or concentration presented to the instrument, with units and dilution mapping; these are not interchangeable.
Define the experimental unit before execution: one independently fortified sample aliquot taken through the full procedure. This example produces nine independent preparations. Three injections from each of three preparations do not substitute for nine preparations. Specify whether injection repeats assess injection variability or are averaged into one preparation result.
| Preparation group | Fortification stage | Replication | Purpose |
|---|---|---|---|
| Unspiked native control | None; matched solvent addition | Paired aliquots | Estimate native background |
| Relevant blank | No analyte added | Protocol-defined | Detect reagent or matrix interference |
| Reference preparation | Direct reference dilution | Protocol-defined | Provide known-concentration comparison |
| Low, middle, high fortified levels | Before sample processing | Three independent preparations per level | Assess full-procedure recovery |
An analyte-free placebo is useful when it represents the formulation matrix without native analyte. Use paired native controls when recovery requires subtraction of endogenous analyte, especially where background varies among aliquots. A reagent blank does not replace that control.
Match sample aliquots, final volumes, solvent composition and handling times between fortified and control preparations. Capture reference identity, assigned potency, stock concentration, spike volume and relevant stability conditions through contemporaneous recording. Account for spike solvent in controls rather than allowing fortification to change extraction conditions.
Avoid choosing a tiny added amount against a large native background without evaluating uncertainty: recovery then depends on the small difference between two large measurements, so subtraction uncertainty can dominate the result.
- Define intended useSet matrix, analyte, range and reporting units.
- Approve protocolFix preparations, criteria and exception rules before execution.
- Prepare controlsLink standards, native controls and independent fortified samples.
- Run procedureInclude extraction and preserve original instrument results.
- Calculate RecoveryReconcile dilution, baseline subtraction and individual determinations.
- Investigate ExceptionsRetain initial evidence and justify any repeat work.
- Review studyAssess intervals, document conclusions and approve intended use.
04Calculating percent recovery: two cases with native-background correction
Use a background-corrected calculation to isolate the measured contribution of the spike:
Recovery (%) = [(C_spiked − C_native,matched) / C_added] × 100.
C_spiked is the measured analyte concentration in the spiked preparation. C_native,matched is the native analyte contribution corresponding to the sample aliquot and dilution represented in that preparation. C_added is the known added analyte concentration at that same final volume. All terms must share a common concentration basis; amounts may instead be used if every term represents the same preparation on a common amount basis.
Case 1: matched final volumes
Assume matched preparations containing the same sample aliquot are each made to 10.00 mL. The unspiked preparation measures 10.0 µg/mL. Adding 0.200 mL of a 1,000 µg/mL stock contributes 200 µg, equivalent to 20.0 µg/mL at the final volume. The spiked preparation measures 29.2 µg/mL. The reference material stock potency is already accounted for.
Recovery = [(29.2 − 10.0) / 20.0] × 100 = 96.0%.
Case 2: fortification dilutes the native sample
Assume additive volumes, no transfer loss and a sample measuring 100.0 µg/mL before fortification. Combine 9.00 mL of sample with 1.00 mL of a 200.0 µg/mL stock. At 10.00 mL final volume, the native contribution is (9.00 × 100.0) / 10.00 = 90.0 µg/mL; the addition is (1.00 × 200.0) / 10.00 = 20.0 µg/mL.
A measured result of 109.0 µg/mL gives [(109.0 − 90.0) / 20.0] × 100 = 95.0%. Incorrectly subtracting 100.0 µg/mL would give 45.0%. Both cases are illustrative results, not acceptance limits.
Calculation order
- Normalize units and dilution.
- Establish the matched native contribution.
- Calculate recovery for each independent preparation.
- Summarize by spike level.
Retain unrounded intermediate values and apply the reporting convention only at the end. Using total expected analyte as the denominator answers a different question: measured total divided by expected total assesses total-result agreement, not recovery of the known addition.
05Matrix effects are not the same as standard addition
A matrix effect changes analytical response because of accompanying sample components, such as salts, excipients or co-extractives. Standard addition is a calibration technique: known amounts of analyte are added to sample aliquots to create a series from which native analyte concentration is estimated.
A simple linear standard-addition approach assumes an appropriate linear response model, controlled dilution and comparable response behavior for native and added analyte. Matrix concentration must remain comparable across aliquots, and additive background interference must be addressed. A single fortified sample is not a standard-addition calibration.
- Solvent standards establish response without sample matrix; use suitable materials supported by reference standard qualification.
- Matrix-matched standards help assess response differences against solvent standards at equivalent analyte concentrations, provided the matrix is representative and native background is accounted for.
- Post-extraction additions probe response in the prepared extract without testing losses during earlier extraction. Compare their incremental response with equivalent additions in solvent.
Matrix-matched calibration can compensate for some response differences; it does not demonstrate their absence. Standard addition cannot reconstruct analyte lost before the additions were made.
Changing matrix lots, matrix concentration or analyte concentration can challenge calibration representativeness or the response model. Test these changes deliberately rather than carrying forward a correction unchanged. Selectivity problems need separate evidence: co-eluting degradants, for example, call for discrimination demonstrated through a stability-indicating method, not another recovery percentage that may conceal their contribution.
06Setting recovery acceptance criteria without universal percentage limits
ICH Q2(R2), section 3.3 describes independently evaluating accuracy and precision, each with a predefined acceptance criterion. This is guidance, not a universal recovery limit. In laboratory practice, the method owner should justify criteria against intended use, concentration, matrix variability, reference uncertainty and the consequences of analytical bias.
Do not borrow a percentage window simply because another method tests pharmaceuticals, foods, supplements or botanicals. Connect the rationale to the method’s measurement uncertainty and the decisions its results support.
| Analytical situation | Main uncertainty | Criterion-design consideration |
|---|---|---|
| Clean placebo assay | Reference assignment and preparation | Relate allowable bias to assay decisions and reference uncertainty. |
| Native-analyte-rich matrix | Background subtraction | Account for uncertainty when the added amount is small relative to native content. |
| Trace-level measurement | Blank contribution and low signal | Justify concentration-dependent criteria against reporting needs. |
| Heterogeneous botanical preparation | Subsampling and matrix variability | Represent relevant material variability rather than only a convenient homogeneous preparation. |
ICH Q2(R2), section 3.3.1.4 recommends reporting accuracy with an appropriate 100(1−α)% confidence interval, or a justified alternative statistical interval. The observed interval should be compatible with the corresponding accuracy acceptance criteria unless otherwise justified.
Specify the decision rules before evaluating results, including which concentration levels are assessed separately:
- Mean recovery: evaluates average bias; averaging can hide opposing errors.
- Individual results: define any limits for individual preparations separately.
- Variability: specify the dispersion measure and its criterion, rather than inferring precision from an acceptable mean.
- Interval: define its type, confidence level and compatibility rule; document any justification for a different interpretation.
A confidence interval for mean recovery describes uncertainty in the estimated mean. It is not a bound containing every future result; acceptable mean recovery alone therefore does not establish acceptable individual-result performance.
07Extraction efficiency: comparing pre-extraction and post-extraction spikes
A pre-extraction spike passes through sample preparation; a post-extraction spike is added to the resulting extract. Comparing them can help distinguish preparation-associated losses from changes in detector response. However, a soluble spike added to tablets, powders or botanical tissue may not reproduce the accessibility, binding or distribution of native analyte. High spike recovery therefore does not establish complete native-analyte extraction. Define comminution method and particle size, spike equilibration time, solvent composition, solvent-to-sample ratio, mixing, extraction duration and temperature as study-design variables.
Worked case 3: separating two response reductions
Assume equal analyte additions, equal final volumes and injection volumes, matched extracted matrices, proportional detector response and background-subtracted integrated responses. The neat standard gives 100,000 arbitrary integrated-response units; the post-extraction spike gives 80,000 units; the pre-extraction spike gives 60,000 units.
| Diagnostic ratio | Calculation |
|---|---|
| Post-extraction-to-neat response factor | 80,000 / 100,000 × 100 = 80% |
| Apparent extraction recovery | 60,000 / 80,000 × 100 = 75% |
| Overall process response relative to neat standard | 60,000 / 100,000 × 100 = 60% |
The factors reconcile: 0.80 × 0.75 = 0.60. Under these assumptions, the ratios suggest a response reduction associated with the extracted matrix and an additional loss associated with preparation. They do not identify the loss mechanism. Do not use this decomposition when matrix composition, spike behavior, stability or detector response differs between preparations in ways that invalidate the comparison.
When fortification cannot represent release from the original material, support the study with native-analyte evidence: a suitable reference material or a scientifically justified comparison procedure. If preparation-induced degradation is plausible, a suitable stability-indicating method can help distinguish degradation from incomplete extraction.
08Laboratory evidence, 21 CFR 211.194 and product-specific regulatory scope
Start with the product, jurisdiction and analytical decision—not a generic recovery-study checklist. For U.S. pharmaceutical work, check the current text of 21 CFR 211.194, Laboratory records and 21 CFR 211.165, Testing and release for product applicability and relevant obligations. Where Part 211 applies, 211.194(a) requires laboratory records to “include complete data derived from all tests necessary to assure compliance with established specifications and standards, including examinations and assays”. For a recovery experiment run as part of such testing, that means (a)(2) a statement of each method used and the location of data showing it meets proper standards of accuracy and reliability; (a)(4) a complete record of all data secured, including graphs, charts and spectra, identified to the material and lot tested; (a)(5) a record of all calculations, “including units of measure, conversion factors, and equivalency factors”, which covers the spike-amount and percent-recovery arithmetic; (a)(6) a statement of results and how they compare with established standards; and (a)(7)–(8) the tester’s and a second reviewer’s initials or signature. Under 211.194(b), any modification of an established method needs records with the reason and data showing the modified method is at least as accurate and reliable, which is where recovery data often appear in practice.
| Product and jurisdiction | Reference or framework | Status | Scope question |
|---|---|---|---|
| U.S. pharmaceutical work | 21 CFR 211.194; 211.165 | Binding regulation where applicable | Which laboratory-record and testing/release obligations apply to this product and activity? |
| Pharmaceutical work in other jurisdictions | Local requirements; ICH Q2(R2) | Local law distinguished from ICH guidance | Which local requirements and guidance implementation govern this analytical decision? |
| Supplements and food | Product-specific rules and analytical guidance | Determine applicable law and guidance separately | Which category and jurisdiction govern, without importing pharmaceutical requirements? |
| Devices and chemicals | Product-specific testing frameworks | Determine applicable law and guidance separately | Which requirements govern the intended analytical use and resulting decision? |
ICH Q2(R2), section 1.2 addresses analytical procedures for commercial drug-substance and drug-product release and stability testing. It is guidance, not a universal cross-sector mandate.
Recommended practice: a reconstructable evidence package
The following record design is recommended practice, not a list of fields required by the CFR provisions above:
- Connect protocol and method version to sample identity, matrix lot and preparation identity.
- Record reference lot and potency, stock and spike quantities, sample mass or aliquot, final volume, fortification stage and native control.
- Link raw instrument data to calculations, calculation version, level summaries, exceptions and review conclusions.
Evidence relationship diagram: protocol + method version → identified preparations and controls → raw data → versioned calculations → level summaries → reviewer decision. Link exceptions or a deviation to the affected preparation, calculation and conclusion. These arrows describe relationships, not a mandatory database schema; another analyst should be able to reconstruct the conclusion.
Keep method-validation evidence distinct from routine batch-test results. Where a study supports a method used for batch decisions, preserve the study-to-method link and connect routine results to the relevant batch record. Do not assign every validation preparation to a production batch.
Validation Study
Defines the approved validation scope.
- study_id
- method_version
- protocol_version
- criteria_version
- reportable_range
- reporting_unit
Reference Stock
Identifies the material added.
- stock_id
- reference_lot
- assigned_potency
- stock_concentration
- concentration_unit
Preparation
Links matrix, fortification and native control.
- preparation_id
- study_id
- stock_id
- matrix_lot
- native_control_id
- spike_stage
- spike_volume_mL
- final_volume_mL
Determination
Retains original result and calculation lineage.
- result_id
- preparation_id
- raw_data_uri
- result_concentration
- dilution_factor
- recovery_pct
- calculation_version
Review decision
Connects exceptions with the final assessment.
- review_id
- study_id
- exception_id
- reviewer_id
- reviewed_at
- decision
09When recovery fails: validation decisions and OOS investigations
Failure against a validation protocol’s recovery criterion is not automatically a product out-of-specification result: the latter concerns a result outside an applicable product specification. Their relationship depends on the study’s purpose and evidence linking the analytical issue to product results. A validation failure may nevertheless call earlier product results into question.
Investigation sequence: laboratory practice, not a prescribed legal procedure
- Retain initial raw data, preparations where stable, calculations and processing history; include relevant changes in the audit trail review.
- Confirm sample and standard identities, spike amounts, units, dilution factors and native-background inputs.
- Assess blanks, unspiked samples, independent standards and system suitability before choosing an explanation.
- Formulate a testable cause. Document why any repeat preparation or injection can distinguish that cause, what outcome is expected and how all results will be evaluated.
| Suspected failure mode | Discriminating check |
|---|---|
| Calculation or preparation error | Independently reconstruct calculations and reconcile recorded weights and volumes. |
| Reference-material problem | Compare with an independently prepared, suitably characterised reference material. |
| Matrix-dependent response | Compare matched post-extraction spikes with solvent standards. |
| Analyte loss or instability | Compare fortification timing and fresh versus held preparations. |
| Interference | Examine blanks and confirm selectivity using an orthogonal measurement. |
Selective replacement of unfavorable results hides variability; retrospective widening of criteria removes the planned decision boundary; averaging away a known problem obscures its cause. Retain the original conclusion and distinguish justified follow-up work from the initial study.
Assess whether affected methods, matrices or earlier results need additional review, and document the scope and rationale. Do not automatically use observed recovery as a correction factor for reported product results: any correction strategy needs its own technical justification and review under the applicable quality system.
10Moving study evidence between systems without losing calculation history
Use the workflow diagram to track progress from protocol approval through final study review, rather than treating file transfer as completion. At each handoff, retain the connection between preparation records, acquisition evidence, recovery calculations and exception assessments. The controls below are implementation recommendations, not regulatory requirements or statements of product features.
Assign system responsibilities explicitly. An instrument data system can retain acquisition evidence; a laboratory system can organize samples and results; a quality system can connect investigations and decisions. Design MES-to-LIMS integration around sample and result identities, and MES-to-QMS integration around linked exceptions and their disposition. These are architecture choices, not assumptions about any product.
- Carry stable study, sample, preparation and acquisition identifiers, explicit units, and method and calculation versions with each result.
- Reconcile expected and received preparation counts by level and control type. Count independent preparations separately from repeat injections; identify missing native controls before accepting background-corrected recovery.
- Detect duplicate imports by record identity and version, not filename alone. Flag partial transfers and mismatched concentration units; document conversions rather than relabelling values.
- Compare the calculation version used with the approved study rule. A transferred result calculated against an obsolete rule remains unresolved even when every file arrives.
Hold unresolved records from decision use. Preserve the original value, record the correction reason and responsible actor, and recalculate through a controlled version. Reassess affected level summaries and study conclusions, not just the corrected row. Use an audit-trail review workflow to make that history reviewable. Avoid silent overwriting: successful transfer confirms delivery, not scientific approval.
11A practical checklist for implementation, review and change assessment
Use this practical method-owner checklist before accepting the study for its intended use. It adds no legal requirements. Evidence pointers should identify controlled records and locations, not reproduce their contents.
| Review gate | Question | Evidence pointer |
|---|---|---|
| Intended use and matrix coverage | Does the conclusion match the actual matrix, range and preparation steps studied? | Protocol scope; final review record |
| Independent preparations | Are preparations independent as planned, and are all planned preparations accounted for? | Protocol design; preparation register |
| Reference and fortification traceability | Can each spike be traced to its reference lot, assigned content, stock preparation and addition? | Reference certificate; fortification records |
| Matched-background calculations | Does the native control represent the same sample amount? Do units reconcile? Can each recovery be reconstructed? | Paired preparation records; raw results; versioned calculations |
| Predefined decision rules | Were criteria and treatment of individual results defined before evaluation and applied consistently? | Approved protocol; calculation output; review record |
| Complete exception assessment | Are missing, invalidated, repeated and excluded results explained, with their impact assessed? | Raw-result inventory; exception assessments; reviewer disposition |
Implementation readiness
Record the reviewer’s decision and any restrictions before implementation. Where calculations changed, use the audit trail to connect the reviewed version to its inputs and superseded results.
Ongoing change assessment
Through change control, assess whether matrix changes, reference-material changes, extraction changes or altered reporting ranges affect the evidence’s relevance. Document whether additional work is needed and why; this checklist sets no universal retesting interval.
Close with a bounded conclusion: what the study supports, the conditions it covers and any unresolved limitations.
Frequently asked questions
Q.Can spike recovery be meaningful when the sample's true analyte concentration is unknown?+
Yes. Recovery can assess the measured increment from a known addition without establishing the sample's true native concentration. Calculate recovery as 100 × (spiked result − unspiked result) ÷ added concentration, with all terms on the same dilution-corrected basis. For example, assuming matched aliquots and negligible volume change, results of 21 mg/L and 12 mg/L after adding 10 mg/L give 90% recovery. This tests recovery of added analyte, not necessarily native analyte extraction; baseline variability also contributes to measurement uncertainty.
Q.How should recovery be calculated when the unspiked result is below the quantitation limit?+
Do not silently substitute zero, half the quantitation limit, or the quantitation limit itself. A below-limit result does not establish an exact baseline concentration. Define the handling approach before evaluating recovery: use a suitably sensitive procedure to estimate the baseline, demonstrate that its contribution is negligible for the intended decision, or evaluate scientifically supported baseline bounds. Document how that choice affects the recovery conclusion. A LIMS reporting convention should not become the scientific justification for a numeric substitution.
Q.Does a passing average recovery justify accepting one unusually low replicate?+
Not by itself. Apply the protocol's predefined decision rules, including any individual-result and variability criteria; do not invent a new exclusion rule after seeing the data. Investigate whether the low result reflects preparation error, instability, interference, or genuine method variability, retaining it unless exclusion has a documented scientific basis. ICH Q2(R2), section 3.3.1.4 recommends an accuracy interval compatible with acceptance criteria unless otherwise justified. A passing mean therefore does not end the assessment; an unusual replicate may warrant out-of-trend handling.
Q.Can repeated injections of one fortified vial count as independent recovery replicates?+
No. Repeated injections share the same fortification, extraction, dilution, and vial preparation, so they measure injection-related variation rather than independent preparation variability. For full-procedure recovery replicates, prepare separate test portions and carry each through the relevant preparation steps. Multiple injections can still help diagnose instrument variability, but label them as technical repeats and distinguish their count from the number of independent preparations. Record preparation and injection identifiers separately in the chromatography record so reviewers can reconstruct the study design.
Q.When can a surrogate matrix support an accuracy study, and which differences need justification?+
A surrogate matrix can support a scientifically justified study when it represents the sample properties that influence recovery. Compare composition, pH, viscosity, binding, extraction behavior, and detector suppression or enhancement; justify relevant differences with comparative data. Do not assume an analyte-free substitute behaves like the product. ICH Q2(R2), section 3.3.1.2 describes adding analyte to the test sample when expected components cannot all be reproduced. Evidence from excipient compatibility studies may inform matrix selection, but cannot replace analytical recovery comparisons.
Q.Can recovery acceptance criteria be carried over to a new matrix or a transferred method?+
Treat existing criteria as a proposal, not an automatic inheritance. The method owner should justify their suitability for the receiving laboratory, intended use, concentration range, and matrix. Compare extraction efficiency, interferences, sample handling, equipment, and achievable variability. A new matrix may need additional recovery experiments even when instrument settings remain unchanged. During analytical method transfer, define the evidence needed to retain or revise criteria before testing; widening limits merely to accommodate disappointing transfer results is not a scientific justification.
Q.Should every production batch include a spike recovery test?+
Not as a default laboratory practice. Decide whether routine spikes are useful through the method's control strategy, considering matrix variability, extraction risk, and the consequences of an undetected analytical bias. A validation recovery study and a routine batch-associated spike serve different purposes: one characterizes performance across the intended range, while the other checks selected conditions during use. Where routine spikes are specified, define their frequency, levels, and failure actions. Keep these decisions distinct from system suitability testing.
Q.Does a failed recovery study automatically make associated product results out of specification?+
No. Failure against a recovery criterion concerns analytical performance; a product result outside its specification is a separate finding. Investigate whether the recovery failure undermines associated results, considering shared preparations, standards, reagents, matrix effects, and the timing of the failure. Do not automatically invalidate product results or accept them merely because their reported values pass. Use the applicable OOS investigation process where appropriate, document the impact assessment, and leave result acceptance and product disposition to authorized human reviewers.
Primary sources
Want to see how Spike Recovery (Assay Accuracy) could fit into your own records and workflows? Explore the related V5 pages or talk to our team about what applies to your operation.
