Kratom Out-of-Specification Results: Investigations, Retesting and Batch Decisions

September 1, 2026

A kratom out-of-specification result creates a decision point—not permission to keep testing until a passing number appears.

The first result may reflect a real problem in the lot. It may also reflect a documented laboratory error, a nonrepresentative sample, a calculation mistake, contamination during preparation or a method limitation. The purpose of an investigation is to determine what the evidence supports while preserving the original result, sample history and batch status.

A certificate of analysis should not be replaced simply because a second test looks better. Nor should every unexpected result automatically condemn a lot without review. A defensible system quarantines the affected material, investigates promptly, tests only under a written and scientifically justified plan, and assigns disposition through authorized quality personnel.

This guide focuses primarily on ordinary botanical Mitragyna speciosa leaf powder and pure-leaf capsules, including Kiody’s approximately 500 mg capsules. Extracts, beverages, enhanced products and concentrated 7-hydroxymitragynine products are different matrices and may require different investigation and sampling plans.

Kiody serves adults 21+ and does not sell concentrated 7-OH. This guide is educational and non-medical. It does not diagnose, treat, cure or prevent any condition, create a universal investigation procedure, replace legal advice or establish Kiody’s private product specifications.

The short answer

A useful out-of-specification investigation follows six principles:

Control the material. Keep the lot and related material in quarantine while the result is unresolved.
Preserve the evidence. Retain the original data, calculations, sample, preparation history and instrument records.
Investigate before retesting. Review laboratory and production causes before generating more numbers.
Use a written hypothesis. Additional testing should answer a defined question, not search randomly for a passing result.
Consider every valid result. A later passing result does not erase a valid failure.
Separate investigation from disposition. Quality personnel decide whether to reject, reprocess or otherwise dispose of material under the applicable procedure and evidence.

The central rule is simple: do not test a lot into compliance.

What counts as an unexpected result?

Terms vary among laboratories and quality systems. A written procedure should define them.

Out of specification

An out-of-specification, or OOS, result falls outside an established specification or acceptance criterion. Examples might include a pathogen detection when the specification requires absence, a heavy-metal result above a product limit, or an alkaloid result outside an established composition range.

The specification must exist before the result is evaluated. Moving a limit after seeing the data is not an investigation.

Out of trend

An out-of-trend, or OOT, result may still meet the specification but differ meaningfully from historical or expected behavior. A botanical lot that normally falls within a narrow moisture range might remain technically acceptable yet shift sharply upward. OOT results can reveal process drift, supplier change, sample mix-up or method instability before a formal failure occurs.

Trend limits are not necessarily product specifications. The record should identify which rule triggered the review.

Atypical or unexpected result

An atypical result may conflict with related data, appearance, prior testing or scientific expectations without crossing a formal limit. Examples include a detected analyte that was absent from all supplier records, an impossible calculation relationship, or two replicate preparations that disagree sharply.

Invalid result

An invalid result is one that a scientifically sound investigation concludes should not be used as a valid measure of the sample. Examples can include a proven dilution error, documented instrument malfunction, sample spill, contamination demonstrated by a blank or transcription error.

“Analyst error” is not an adequate conclusion by itself. The investigation should identify the specific error, the evidence connecting it to the result and the corrective action.

Inconclusive result

An inconclusive result does not provide enough reliable information to make the intended decision. A damaged sample, insufficient test portion, method inhibition or conflicting valid evidence may lead to an inconclusive status. Inconclusive does not mean passing.

Near-limit result

A result close to a specification requires careful review of units, rounding, uncertainty, method precision and the written decision rule. It is not automatically OOS, but it should not be manipulated through selective decimal handling.

The Part 111 material-review baseline

FDA’s dietary-supplement current good manufacturing practice rule provides a direct quality-system framework for specification failures.

Under 21 CFR §111.113, quality-control personnel must conduct a material review and make a disposition decision when a specification is not met. The same requirement can be triggered by a batch deviation, an unanticipated occurrence that may lead to adulteration, calibration information suggesting a batch-quality problem or a returned dietary supplement.

Section 111.113 states that material failing a specification must be rejected unless quality-control personnel approve a permitted treatment, in-process adjustment or reprocessing. The person conducting the material review must document the review and decision at the time of performance.

21 CFR §111.123 assigns quality-control personnel responsibility for reviewing batch records, conducting required material reviews, approving or rejecting reprocessing, determining whether finished-product specifications are met and approving or rejecting finished batches for distribution. It does not allow a batch that fails product specifications simply to be released because an explanation sounds plausible.

21 CFR §111.140 requires written quality-control procedures and documentation of material reviews, disposition decisions and follow-up. The record must be included in the appropriate batch production record.

These are binding dietary-supplement CGMP provisions for firms and operations within their scope. Applying them as a quality benchmark does not mean FDA has approved kratom, Kiody or any particular product.

FDA also publishes Investigating Out-of-Specification Test Results for Pharmaceutical Production. That document addresses drug production and contains nonbinding recommendations; it is not a kratom-specific dietary-supplement rule. Its staged approach—laboratory assessment, expanded investigation, scientifically justified retesting and full evaluation of data—offers useful technical concepts when a botanical quality procedure is designed.

Immediate actions after an unexpected result

1. Keep the lot in quarantine

Do not release, ship, relabel, blend or otherwise dispose of the affected material while the result is unresolved. Confirm the physical quantity and location of the lot and any sublots, work-in-process, finished packages or related shipments.

If some of the lot was already distributed, escalation may include a distribution hold, complaint review, health-hazard evaluation or recall assessment under appropriate procedures. This guide does not make those decisions automatically.

2. Notify authorized personnel

The laboratory, quality unit and relevant production personnel should know that an investigation has opened. Sales pressure or inventory need should not determine whether the result is accepted.

3. Preserve original records

Preserve:

raw instrument files and audit trails;
chromatograms, spectra, images or plate records;
sample weights, dilutions and calculations;
preparation worksheets and analyst notes;
calibration and system-suitability data;
blanks, controls, spikes and duplicates;
sample seals and chain of custody;
remaining laboratory sample and retains;
environmental and equipment conditions; and
communications relevant to the result.

Do not overwrite, backdate or silently replace the original COA. If a corrected COA is later justified, it should remain traceable to the prior version and reason for correction.

4. Assess immediate scope

Ask whether the issue could affect:

only one analytical preparation;
the submitted laboratory sample;
one container or pallet;
the entire lot;
other lots tested in the same run;
other products using the same component;
batches produced on shared equipment; or
already distributed material.

The first scope assessment can change as evidence develops.

Phase I: the laboratory investigation

The initial laboratory review should be prompt and evidence based. It asks whether an assignable analytical cause exists.

Confirm the sample identity

Review the product name, lot number, sample ID, seal, container, receipt condition and chain of custody. A mislabeled jar or switched worksheet can produce a technically correct result for the wrong material.

Review the method and version

Confirm that the correct approved method, version and specification were used. Check whether the product matrix and analyte fall within the method’s validated or verified scope.

Review sample preparation

Examine weights, extraction time, solvent, dilution steps, mixing, digestion, filtration, incubation and test-portion size. Recalculate every dilution factor and unit conversion.

An error should be demonstrated, not assumed. A difficult extraction is not automatically an analyst error.

Review reference materials and reagents

Confirm identity, purity assignment, expiration or retest date, preparation, storage and traceability of standards, controls, media and reagents. A standard-preparation error can bias an entire sequence.

Review instrument and system performance

Evaluate calibration, maintenance, instrument logs, system suitability, control charts, pressure, temperature, detector response and alarms. Look for drift, carryover, leaks, contamination, power interruption or failed checks.

Passing system suitability is relevant but not conclusive. It cannot prove that the sample was prepared correctly or that the method performed adequately in that matrix.

Review controls and neighboring samples

Blanks, spikes, duplicates, positive controls, negative controls and continuing calibration checks can locate problems. Neighboring high-concentration samples may suggest carryover. Multiple unrelated failures can suggest a run-wide issue.

Do not use production samples to condition the system secretly or discard exploratory injections without preserving the data.

Interview the analyst

The analyst should describe what occurred while the work is fresh. Questions should be neutral and specific. “You must have made a mistake” encourages speculation. Better questions address unusual observations, interruptions, deviations, instrument behavior and procedure clarity.

Decide whether an assignable cause is demonstrated

If a specific laboratory cause is proven and scientifically explains the result, the result may be invalidated under the written procedure. The investigation should document:

the identified cause;
the evidence;
why the cause explains the result;
the impact on other samples or runs;
corrective action; and
the authorized decision.

If no assignable laboratory cause is found, the investigation should expand. “No obvious error” is not a reason to erase the result.

Phase II: production, process and sampling review

An unexplained result requires examination beyond the laboratory.

Lot definition and genealogy

Reconstruct supplier lots, receiving records, production dates, equipment, operators, packaging runs, rework and distribution. Confirm that the tested material belongs to the claimed population.

Sampling plan

Determine who sampled the lot, which containers were selected, how increments were collected, whether a composite was created and whether the plan was appropriate for the analyte.

A method can be valid while a sample is not representative. A top-of-bag scoop cannot automatically establish the condition of an entire multi-container lot.

Manufacturing and blending

Review milling, blending time, equipment loading, fill sequence, start-middle-end data and potential segregation. Botanical powders may vary by particle size, moisture or localized contamination.

Sanitation and environment

For microbiology or foreign-material issues, review sanitation records, environmental monitoring, roof leaks, water events, pest observations, traffic patterns and equipment-cleaning history.

Components and packaging

Examine whether the same supplier component appears in other batches and whether packaging damage could have affected moisture or contamination. A failed finished-product result can arise from an incoming component, process or post-process exposure.

Review complaints, deviations, prior OOS/OOT results, supplier performance and stability data. An apparently isolated result may be the latest point in a broader pattern.

Retesting: answer a hypothesis, not a hope

Retesting means analyzing additional portions of the original laboratory sample or another retained portion under a predefined plan. It can be appropriate when the investigation identifies a scientifically plausible question.

Examples include:

repeating a preparation after a proven dilution error;
testing retained individual increments after a composite failure;
comparing a result using a justified orthogonal method;
evaluating possible matrix inhibition with suitable controls; or
confirming an unexpected detection under a defined confirmation procedure.

Before retesting, the protocol should state:

the hypothesis;
the sample to be used;
the method and analyst;
the number of determinations;
the acceptance and invalidation criteria;
how original and new results will be evaluated; and
who approved the plan.

The number of retests should not change after the results begin to appear.

Resampling is not the same as retesting

Retesting uses additional test portions from the original sample or retained laboratory material. Resampling collects new material from the lot.

Resampling can answer whether the first sample represented the lot, but it also creates a new selection opportunity. It should be used only under a justified, approved plan.

A new sample should not be collected merely because the first sample failed. The investigation should explain why the original sampling may have been inadequate and how the new design corrects the suspected limitation.

If contamination may be localized, a broad resampling plan can help characterize distribution. An average composite may be inappropriate because it can dilute a hotspot. Retaining individual increments can preserve location information.

Why averaging can mislead

Averaging valid passing and failing results can conceal variation.

Suppose a specification has an upper limit and three valid results are 0.7, 0.8 and 1.4 in fictional units. An average of 0.97 does not make the 1.4 result disappear. Whether replicates may be averaged depends on the method, specification and predefined calculation procedure—not on whether the mean passes.

Replicate injections from one preparation are different from independent sample preparations. Pooling them without distinction can understate real variability.

When results conflict, the investigation should ask why. Possible explanations include sample heterogeneity, preparation variability, method precision, contamination or data error. “Average passes” is not a root cause.

Near-limit results, rounding and uncertainty

Near a decision boundary, small choices can change the classification:

dry-weight versus as-is basis;
percent of product weight versus percent of total alkaloids;
milligrams per gram versus milligrams per serving;
stated versus actual serving mass;
number of decimal places;
rounding before or after calculation; and
treatment of measurement uncertainty.

A written rule should specify the reportable unit, basis and rounding convention before testing.

Do not add decimals beyond method capability. A result of 0.0496% may display as 0.050% under one convention and 0.0496% under another, but the underlying uncertainty remains. A boundary decision should follow the governing specification and approved procedure, not the desired commercial outcome.

An in-specification near-limit result may warrant an OOT review, tighter monitoring or supplier action even when it is not formally OOS.

Test-category considerations

Alkaloid composition

Review reference standards, analyte identity, separation of related compounds, extraction recovery, calibration, dry-weight correction and units. Confirm that the method’s LOQ and range support the decision point.

A panel labeled “mitragynine and 7-OH” does not establish that mitragynine pseudoindoxyl, MGM-15 or MGM-16 were measured.

Microbiology

Microbial contamination can be heterogeneous. A negative retest from another portion does not necessarily invalidate a pathogen-positive result. The investigation should examine method controls, confirmation, test-portion size, sampling and localized contamination.

“Not detected” is not the same as sterile. It is tied to the method and portion tested.

Heavy metals

Review digestion, blanks, certified reference material, spikes, spectral interferences and contamination. A high result may reflect a localized foreign particle, actual lot contamination or laboratory contamination. Each possibility requires evidence.

Pesticides and mycotoxins

Multi-residue analytes can have different recoveries and reporting limits. Mycotoxins and some pesticide residues may be unevenly distributed, making the sampling investigation central.

Moisture and water activity

Confirm that the same property and method are being compared. Loss on drying, Karl Fischer moisture and water activity are not interchangeable. Package damage or humidity exposure can create container-specific differences.

Identity

An identity failure cannot be corrected by potency retesting. Detecting mitragynine does not necessarily authenticate the complete botanical. Under 21 CFR §111.75, reliance on a supplier certificate does not replace required component identity testing within the Part 111 framework.

Root cause, corrective action and preventive action

An investigation should go beyond naming the immediate event.

If a dilution was wrong, ask why:

Was the worksheet unclear?
Did the software allow an invalid factor?
Was training incomplete?
Were units mixed?
Was independent review missing?
Has the same error occurred elsewhere?

Corrective action addresses the detected problem. Preventive action reduces recurrence or broader risk. Possible actions can include procedure revision, training, software controls, method improvement, supplier qualification changes, sampling redesign, equipment repair or enhanced monitoring.

Effectiveness should be checked. Closing an investigation immediately after retraining without confirming improvement leaves the system vulnerable.

Disposition options

The investigation supports disposition but does not predetermine it.

Reject

Reject when the material fails applicable specifications and no scientifically valid, permitted corrective pathway is approved. Rejection should include controlled segregation and documented final handling.

Reprocess, treat or adjust

Under 21 CFR §111.90, reprocessing, treatment or an in-process adjustment requires a quality-control material review, a scientifically valid reason and approval, and must be permitted by §111.77. The resulting batch still requires quality-control approval and compliance before release.

This is not permission to dilute a failed contaminant result into compliance. The legal and scientific permissibility of any action depends on the material, failure and procedure.

Return to supplier

Returning rejected material may be appropriate under supplier and quality procedures, but it does not erase the failure. Supplier corrective action and impact on other lots may still require review.

Release

Release should occur only when authorized quality personnel conclude that all specifications are met and the full record supports the decision. A valid unexplained OOS result cannot be ignored because later tests pass.

A ten-gate investigation workflow

Gate 1: Open and contain

Assign an investigation number, quarantine affected material and identify immediate scope.

Gate 2: Preserve data

Secure original records, sample material, retains, audit trails and communication.

Gate 3: Laboratory assessment

Review sample identity, method, calculations, standards, reagents, controls, instrument and analyst observations.

Gate 4: Assignable-cause decision

Determine whether a specific, scientifically supported laboratory cause exists. Document the evidence.

Gate 5: Expanded investigation

Review sampling, manufacturing, sanitation, components, equipment, packaging, complaints and trends.

Gate 6: Additional-testing protocol

If justified, approve a predefined retesting or resampling plan tied to a hypothesis.

Gate 7: Data evaluation

Evaluate original and new results together. Do not selectively omit unfavorable valid data.

Gate 8: Root cause and scope

Identify root cause when possible and assess impact on related lots, methods or products.

Gate 9: CAPA and effectiveness

Assign corrective and preventive actions, owners, due dates and effectiveness checks.

Gate 10: Material disposition

Quality personnel document rejection, approved reprocessing or release and ensure the batch record is complete.

Five fictional investigation examples

These examples are entirely fictional. Their values are not Kiody specifications, universal limits or legal thresholds.

Example 1: Proven dilution error

An alkaloid result fails high. The worksheet shows that the analyst entered a tenfold dilution as fivefold, and the raw response and independent calculation confirm the mistake.

Review: A specific cause is demonstrated. The original reported calculation can be invalidated under the procedure, but the raw data and investigation remain. A predefined repeat preparation confirms performance, and the worksheet control is corrected.

Example 2: Passing retest with no cause

The first heavy-metal result exceeds the specification. Two later preparations pass, but the laboratory finds no error in the original test.

Review: The valid failure remains part of the evidence. The investigation should expand to sampling, digestion variability, localized contamination and lot heterogeneity. Averaging the three results is not an adequate disposition rationale.

Example 3: Pathogen positive followed by negative

A Salmonella screen is confirmed positive. A second test portion is negative.

Review: The negative does not erase the positive. Microbial contamination can be unevenly distributed. Review controls, confirmation, sample history, sanitation, lot distribution and the need for broader investigation.

Example 4: Near-limit 7-OH calculation

A fictional result is reported on an as-is basis, while the specification is written on a dry-weight basis. Moisture correction changes the classification.

Review: The method and specification must use the same basis. Recalculate under the approved rule, document the correction and examine why the mismatch reached review.

Example 5: Failed composite, retained increments available

A ten-container composite fails a mycotoxin specification. The laboratory retained each increment separately.

Review: Testing the individual retained increments under a written plan can help locate the source and define scope. It does not automatically convert the composite failure into a pass.

Twenty investigation warning signs

The lot is released while the investigation remains open.
The original COA disappears after retesting.
Retesting begins before a hypothesis is written.
The number of retests increases after failures appear.
Only passing retests enter the final report.
Valid failures are averaged with passing results without justification.
“Analyst error” appears without a specific documented error.
A supervisor assumes laboratory error because the result is inconvenient.
System suitability is treated as proof that sample preparation was correct.
Raw data or audit trails are unavailable.
The method version cannot be identified.
Powder validation is applied to an extract or beverage without review.
A new sample is collected merely because the first sample failed.
Resampling does not correct the suspected sampling weakness.
Units or denominators change during the investigation.
“Not detected” is translated into absolute absence.
Supplier history is used to dismiss a current failure.
CAPA ends with retraining and no effectiveness check.
Related lots or distributed material are never assessed.
Commercial urgency determines the disposition.

One warning sign does not prove misconduct. It identifies a gap that should be resolved before the batch decision is trusted.

A proposed 44-field investigation record

Investigation number
Date and time opened
Reporter
Product name
Product format
Internal lot number
Supplier lot number
Quantity and container count
Current quarantine location
Test and analyte
Specification
Original result
Units and basis
Laboratory name
Sample ID
Method number and version
Analyst
Test date
Original COA version
Immediate notifications
Distributed quantity, if any
Initial scope
Chain-of-custody review
Sample-condition review
Calculation review
Standard and reagent review
Instrument and calibration review
System-suitability review
Control and blank review
Analyst interview
Assignable laboratory cause
Production-record review
Sampling-plan review
Sanitation and environment review
Supplier and component review
Complaint and trend review
Additional-testing hypothesis
Approved retest or resample protocol
Additional results
Root cause
Related-lot impact
CAPA and effectiveness check
Material disposition and rationale
Quality approver, signature and date

The record should link controlled source documents rather than copying uncontrolled fragments.

Current federal 7-OH and derivative context

As of September 1, 2026, DEA’s federal 7-OH threshold action remains proposed rather than effective. HHS extended the related information-request comment period through September 10, 2026 in an August 26 notice. Kiody’s nationwide 7-OH law tracker separates current rules from pending proposals and should be checked for later changes.

Separately, mitragynine pseudoindoxyl, MGM-15 and MGM-16 entered federal Schedule I effective August 26, 2026, through DEA’s temporary scheduling order.

An OOS procedure should not describe a proposed federal threshold as current law. It should also not treat a “mitragynine and 7-OH” test as evidence that the three scheduled derivatives were tested.

State and local rules can use different analytes, percentages, ratios, product categories and age or registration requirements. Product-quality disposition does not replace destination-specific shipping review.

Frequently asked questions

1. Does one failed result automatically prove the entire lot is defective?

Not always. It triggers containment, review and a disposition process. The result may represent the lot, localized material or a demonstrated laboratory problem. The investigation determines what the evidence supports.

2. Can the laboratory simply rerun the sample?

Not merely to seek a pass. Additional testing should follow a predefined, scientifically justified plan tied to a hypothesis.

3. Does a passing retest erase the original failure?

No. A valid original result remains part of the record. It can be invalidated only when a scientifically sound investigation demonstrates a specific cause.

4. What is testing into compliance?

It is repeated or selective testing designed to obtain enough passing results to ignore an unfavorable valid result rather than understand it.

5. Is analyst error a valid root cause?

Only when the specific error is demonstrated and explains the result. The investigation should also ask why the system allowed the error.

6. What is the difference between retesting and resampling?

Retesting analyzes another portion of the original sample or retained laboratory material. Resampling collects new material from the lot.

7. When can resampling be appropriate?

When evidence suggests the original sample may not have represented the lot and a new approved design addresses that suspected weakness.

8. Can passing and failing results be averaged?

Not simply to produce a passing mean. Averaging must follow the predefined method and specification. Conflicting valid results require investigation.

9. Is an out-of-trend result the same as OOS?

No. OOT may meet the formal specification but differ from expected history. It can still justify investigation or monitoring.

10. Does a near-limit result fail automatically?

Not necessarily. Apply the written specification, units, basis, rounding and decision rule. Method uncertainty and trend risk may still warrant review.

11. Why keep the lot in quarantine?

Quarantine prevents unresolved material from being used or distributed while the investigation and disposition remain incomplete.

12. Can a corrected COA replace the original?

A corrected COA may be issued when justified, but it should remain traceable to the original result, version and reason for correction.

13. Does a negative pathogen retest invalidate a positive result?

No. Contamination may be heterogeneous. Controls, confirmation, sampling and sanitation evidence must be evaluated together.

14. Can a supplier’s passing COA override a finished-product failure?

No. The records describe different samples and potentially different production stages. The conflict should be investigated.

15. Does reprocessing guarantee the lot can be released?

No. Reprocessing must be scientifically justified, permitted, approved and followed by evidence that all applicable specifications are met.

16. Who makes the final disposition decision?

Authorized quality-control personnel under the applicable quality system—not the salesperson, laboratory alone or inventory manager.

17. Does a completed OOS investigation make the product legal to ship everywhere?

No. Product quality, registration, labeling, age rules and destination-specific law remain separate.

18. Does Kiody sell concentrated 7-OH?

No. Kiody serves adults 21+ and does not sell concentrated 7-OH.

The practical standard: preserve the failure and follow the evidence

A trustworthy investigation does not begin with the assumption that the laboratory must be wrong or that the lot must be rejected. It begins by controlling the material and preserving what happened.

The original result, raw data, sample history, method controls, production record and follow-up testing form one evidence set. Each additional test should have a reason. Each invalidated result should have a demonstrated cause. Each disposition should be authorized, documented and connected to the complete record.

That discipline protects more than one batch decision. It reveals supplier drift, sampling weaknesses, process failures and laboratory problems before they repeat. It also prevents a passing number from becoming more important than the truth the quality system is supposed to find.