Educational information for adults 21+. This article is not medical advice. Kiody does not sell concentrated 7-OH.
The short answer
A certificate of analysis is a snapshot. A well-maintained history of
lot-specific COAs can become a moving picture.
One COA may show that the tested sample met selected specifications
on one occasion. A series of comparable results may reveal something
different: gradual drift, a sudden step change, greater variability,
repeated results near a limit, a new supplier pattern, a seasonal shift,
a laboratory change or an apparent improvement that is really caused by
a different reporting basis.
Trend analysis can help a quality team ask better questions. It
cannot prove that every unit in a lot is identical, that an untested lot
is acceptable, that a product is effective, that a vendor is trustworthy
in every respect or that a product is legal in a particular place.
Before results are compared, they must be made meaningfully
comparable. At minimum, confirm the product type, lot, analyte, unit,
reporting basis, method, laboratory, sample source and relevant
detection or quantitation limit. Comparing 1.4% mitragynine
in dry leaf with 14 mg/mL in a liquid extract is not trend
analysis. Neither is placing “ND,” “<LOQ” and zero into the same
spreadsheet as though they mean the same thing.
The central rule is simple:
first establish comparability → then visualize the sequence → then investigate patterns → never replace lot release with a trend
Why a COA history matters
Imagine two suppliers. Each presents a current COA that passes its
stated specifications. Supplier A has 24 consecutive lot results
clustered comfortably within the established operating range, with
stable methods and complete lot matching. Supplier B has only one
report, no previous history and several values close to upper
limits.
The current reports may both say “pass,” but the evidence is not
equally informative.
A historical record can help answer questions such as:
- Is the reported mitragynine concentration broadly consistent for the
same product category? - Is 7-OH changing relative to mitragynine, total alkaloids or dry
weight? - Are contaminant results moving toward an action limit even though
they still pass? - Did a new raw-material origin, supplier, harvest period or process
coincide with a shift? - Did results change after a laboratory, method or reporting-limit
change? - Are failures concentrated in one product format, line, facility or
season? - Does a product claim remain supported across multiple lots, or only
by a favorable report? - Are missing results, changing panels or unexplained unit changes
breaking the evidence chain?
These questions are useful because specifications and trend signals
answer different questions. A specification asks whether a result meets
a defined requirement. A trend asks whether the pattern of results has
changed or deserves attention.
A passing result can still be unusual. An unusual result can still
pass. A failing result is not rescued because it fits a historical
pattern.
The five records
that should not be confused
1. The public COA
A public COA is usually a selected summary. It may identify the
laboratory, sample, lot, methods, analytes, results, units and pass/fail
conclusions. It rarely contains all raw data, sampling records,
calculations, laboratory controls, deviations or internal quality
decisions.
Public COAs can support transparency, but they are not the complete
batch record.
2. The specification
A specification is a documented requirement or limit. Under the
dietary-supplement current good manufacturing practice framework, 21 CFR
§111.70 describes specifications for components, in-process production,
labels and packaging, and finished products. Finished-product
specifications address identity, purity, strength, composition and
contamination limits that may lead to adulteration.
FDA states that kratom is not lawfully marketed as a dietary
supplement. Kiody therefore should not imply that Part 111 makes a
kratom product FDA approved or lawful. The framework is cited here
because it provides a concrete quality-system model for defining and
verifying product requirements.
A specification should have a documented scientific or regulatory
basis. It should not be invented after a laboratory reports a
result.
3. The control limit
A statistical control limit is calculated from an appropriate
historical data set to help identify behavior that is unusual for a
process. It is not the same as a regulatory or product
specification.
For a conventional Shewhart chart, a center line and upper and lower
control limits are derived from process data under defined assumptions.
NIST explains that measurements outside control limits are treated as
out-of-control signals for the measurement process it describes.
That does not mean every kratom COA series is suitable for a textbook
control chart. Botanical lots may come from different farms, seasons,
suppliers, formulations and sample plans. Combining unlike material can
create misleading limits. A quality team must define a rational data
family before calculating anything.
4. The warning or
internal action limit
A company may set an internal warning level inside a formal
specification so that it can investigate drift before a failure occurs.
This is a management tool, not automatically a legal threshold.
For example, a manufacturer may create an internal review trigger
when a contaminant result moves unusually close to its established
specification. The trigger should be documented, applied consistently
and revised through change control—not improvised only after an
inconvenient result appears.
5. The target or normal
operating range
A target describes the intended value or central operating region.
Some characteristics have two-sided targets. Others have only an upper
or lower requirement. Natural botanical composition may not have a
narrow target unless the product is blended or standardized to one.
“Typical” should not be presented as a guarantee. A historical
average is also not a legal safe harbor.
Specifications
and control limits are not interchangeable
This distinction is the heart of responsible trend analysis.
| Question | Specification | Statistical control limit |
|---|---|---|
| What does it describe? | A requirement the material or product must meet | Expected variation in a defined, stable data-generating process |
| Where does it come from? | Regulation, safety assessment, product design, customer requirement or other documented basis |
Calculations using suitable historical data and defined assumptions |
| Does a value outside it fail? | It may trigger rejection, investigation or another documented disposition |
It signals unusual process behavior; it does not automatically equal a product failure |
| Can it be moved to make data pass? | No; changes require documented scientific and quality review | It may be recalculated only through a controlled, justified process |
| Is it always two-sided? | No; contaminants often use an upper limit | Depends on chart and purpose |
Four situations illustrate the difference:
- Inside control limits and inside specification: The
result is not statistically unusual and meets the requirement.
Lot-specific release review is still required. - Outside control limits but inside specification:
The lot may pass, but the shift deserves investigation. A supplier,
method, analyst, season or process may have changed. - Inside control limits but outside specification:
The process may be consistently producing unacceptable material.
“Stable” does not mean “acceptable.” - Outside both: The result requires formal handling
under the applicable deviation, out-of-specification and material-review
procedures.
A chart cannot convert a failure into a pass. A specification cannot
explain why a process changed.
What should be
trended for kratom products?
The answer depends on the product and quality plan. A useful program
starts with attributes connected to identity, composition,
contamination, stability and legal thresholds.
Alkaloid results
Potential fields include:
- mitragynine;
- 7-hydroxymitragynine;
- other alkaloids actually within the laboratory method’s validated
scope; - total alkaloids when the term is explicitly defined;
- ratio of 7-OH to total alkaloids when a rule or specification uses
that denominator; - 7-OH on a dry-weight basis where relevant;
- alkaloid amount per serving when supported by net contents and
serving information; and - difference between labeled claim and measured result, where a lawful
and supported quantitative claim exists.
Do not calculate “total alkaloids” by adding whichever analytes
happen to appear on a report unless the governing definition permits
that calculation. Do not treat percent of total alkaloids, percent of
dry product, ppm and milligrams per serving as equivalent.
Microbiological results
Trend quantitative counts separately from qualitative pathogen
findings. Total aerobic count, yeast and mold, bile-tolerant
gram-negative organisms and other enumeration results use different
methods and meanings from “detected/not detected” pathogen tests.
A sequence of values reported as <10,
<100 and ND cannot be averaged responsibly
until the laboratory’s reporting conventions, units, sample amounts and
limits are understood. A negative Salmonella result for one tested
analytical portion does not prove that every gram in the lot is pathogen
free.
Elemental contaminants
Lead, arsenic, cadmium and mercury should be separate series. Total
arsenic should not be mixed with inorganic arsenic. Results reported in
µg/g, mg/kg and ppm may be numerically equivalent in some solid-matrix
contexts, but the basis still needs confirmation. Serving-based exposure
calculations also require an accurate serving mass.
Near a threshold, measurement uncertainty, rounding and reporting
basis can materially affect interpretation.
Pesticides,
mycotoxins and residual solvents
These panels often produce censored data: many results are below a
reporting limit. A lower reporting limit in a newer method can make it
appear that contamination has “increased” simply because the laboratory
can now report smaller values.
Trend individual analytes when possible. Do not turn “panel passed”
into a numerical time series. Record changes in the analyte list, method
and reporting limits.
Moisture and water activity
Moisture content and water activity are different properties. They
should never share a chart. Moisture may be reported by loss on drying,
oven methods, Karl Fischer measurement or another defined method. Water
activity is a ratio related to available water and requires its own
instrument and procedure.
Changes can be useful for evaluating drying, packaging, storage and
stability, but low water activity does not establish the absence of
pathogens or mycotoxins.
Physical and packaging
attributes
A mature system may trend:
- capsule fill weight;
- net-content checks;
- blend or unit uniformity results;
- sieve or particle-size measures;
- seal failures;
- package leaks;
- metal-detector or foreign-material events;
- label-reconciliation discrepancies;
- damaged deliveries;
- customer complaints by lot; and
- returns associated with package integrity.
These records should remain linked to the exact product, packaging
configuration, equipment and lot.
Build a
comparable data set before making a chart
Trend analysis usually fails because unlike results are combined, not
because someone chose the wrong chart color.
Keep product families
rational
At minimum, separate:
- whole or cut leaf;
- plain botanical powder;
- pure-leaf capsules;
- blended botanical leaf;
- extracts;
- enhanced leaf;
- liquids and beverages;
- concentrated 7-OH products; and
- manufactured derivatives.
Kiody’s approximately 500 mg pure-leaf capsules should not be
combined with extract capsules simply because both use a capsule shell.
Package format does not determine chemical composition.
Use the same analyte
identity
Mitragynine, 7-OH, MGPI, MGM-15 and MGM-16 are not interchangeable
names. A general “alkaloids” column erases important distinctions.
The federal legal status also differs. As of September 3, 2026, the
federal threshold proceeding for 7-OH remains pending, while MGPI,
MGM-15 and MGM-16 are separately controlled under a temporary Schedule I
order. A trend database needs analyte-specific fields so it does not
hide a controlled compound inside a category total.
Normalize units
only when conversion is valid
For a solid product:
1% = 10,000 ppm0.01% = 100 ppm1 mg/g = 0.1% = 1,000 ppm
These arithmetic relationships do not resolve the reporting basis. A
dry-weight result, an as-received result and a percentage of total
alkaloids use different denominators.
For liquids, converting mg/mL to a mass percentage requires density
or another justified basis. Do not assume 1 mL equals 1 g unless the
actual method and product support that assumption.
Preserve censored results
ND, <LOD, <LOQ and
<reporting limit are not zero.
If a laboratory reports <0.010%, the measurement
supports a conclusion below that stated boundary under the method’s
rules. It does not establish that the true amount is 0.000%. Replacing
every nonquantified result with zero biases averages downward. Replacing
each with the reporting limit biases them upward.
Sophisticated statistical approaches for censored data exist, but the
public-facing interpretation can remain simpler: preserve the original
qualifier, plot the reporting limit when helpful and avoid false
precision.
Record method and
laboratory changes
A step change may come from the product—or from the measurement
system.
Record at least:
- laboratory name and location;
- method identifier and revision;
- instrument technique;
- sample-preparation revision;
- reporting basis;
- LOD, LOQ or reporting limit;
- uncertainty statement when available;
- accreditation scope status;
- analyst or run identifier when available; and
- date the change took effect.
Do not splice two method eras together without marking the
transition. If equivalence has not been demonstrated, analyze them as
separate series.
Confirm the sample
represents the lot
A trend of precise measurements is still misleading if the samples
were convenience grabs from different locations, submitted by different
parties or mislabeled.
The data record should connect each result to:
- the defined lot;
- sampling plan;
- sample collector;
- date and location of collection;
- number and position of increments;
- composite or discrete-sample status;
- sample seal;
- custody transfers; and
- laboratory accession record.
Trend analysis cannot repair missing chain of custody.
A practical 12-step
trend-review process
Step 1: Define the decision
Write the question before selecting data. Examples include “Has
incoming leaf moisture shifted since the supplier change?” or “Are 7-OH
results moving toward the internal review level?”
Avoid broad questions such as “Is quality improving?” until “quality”
is broken into measurable attributes.
Step 2: Define the data
family
Choose a comparable product, supplier, material type, process,
facility, method and reporting basis. Document inclusions and
exclusions.
Step 3: Collect source
records
Use original laboratory reports and controlled batch records. Do not
scrape rounded values from marketing images if the underlying reports
are available.
Step 4: Reconcile identity
Match product name, lot code, sample identifier, test dates,
laboratory accession and report version. Exclude unresolved mismatches
and investigate them separately.
Step 5: Standardize valid
units
Convert only when mathematically and scientifically justified. Retain
both the original result and normalized value so the transformation is
auditable.
Step 6: Mark qualifiers and
limits
Preserve ND, <LOD, <LOQ,
estimated values and amended-report status. Capture the limit that
applied to each result.
Step 7: Plot in
chronological or process order
A simple run chart is often the best first view. Use manufacture or
sampling order when it better reflects the process than report-issue
date.
Step 8: Add specifications
separately
Draw formal specification lines and label their basis. Do not call
them control limits.
Step 9: Evaluate the pattern
Look for a sustained upward or downward movement, clustering,
repeated near-limit results, sudden level change, increasing spread,
isolated point, seasonal pattern or missing-data period.
NIST describes trend detection as evaluating whether sequential data
depart from what would be expected without a trend. Visual inspection is
useful for screening; formal statistical tests require appropriate
assumptions and expertise.
Step 10: Check for
measurement-system causes
Review laboratory, method, reference-standard, calibration,
reporting-limit and sample-preparation changes before attributing the
signal to product manufacturing.
Step 11: Investigate with
context
Compare supplier, origin, harvest period, treatment, equipment,
operator, environment, packaging and storage records. A chart identifies
a question; it rarely proves the cause.
Step 12: Document
the decision and follow-up
Record the reviewer, date, signal, records examined, assessment,
action, owner and due date. Preserve any decision to separate data sets,
revise an alert or re-establish a baseline through controlled
change.
Run
charts, control charts and capability: use the right tool
Run chart
A run chart plots sequential results, often with a median or other
reference. It is easy to understand and useful for finding visible
shifts, cycles and trends. It does not require pretending that a diverse
botanical supply chain is one stable process.
For many vendor or consumer reviews, this is enough.
Control chart
A control chart adds statistically derived limits. It is most
defensible when the underlying process, sampling and measurement system
are sufficiently stable and comparable.
NIST’s measurement-process example uses historical check-standard
data and warns that control limits depend on long-term process
variability. Applying those exact sample-size or limit rules to kratom
lot results without justification would be inappropriate. The NIST
material teaches principles; it is not a kratom-specific regulatory
standard.
If chart rules are used—such as points outside limits, long runs on
one side or sustained slopes—they should be selected before viewing the
next result. Too many overlapping rules create false alarms.
CUSUM and EWMA charts
Cumulative-sum and exponentially weighted moving-average charts can
be more sensitive to small sustained shifts than a basic Shewhart chart.
That sensitivity can be useful for laboratory check standards or a
highly repeatable manufacturing process.
It can also be misleading when a data series has many supplier or
seasonal changes. These tools belong in a qualified statistical program,
not in a decorative dashboard.
Process capability
Capability indices compare the spread and centering of a stable
process with specification limits. NIST notes that common indices such
as Cp and Cpk assume a sufficiently large data
set and often assume normality.
Do not calculate a capability index from six COAs, censored
contaminant data or a mixture of different products and methods. A
capability value does not excuse any individual failure and does not
establish consumer safety.
Five fictional review
examples
These examples are educational. The values and products are fictional
and are not Kiody test results.
Example 1:
Passing mitragynine result after a step change
Twelve plain-leaf powder lots from one supplier report mitragynine
between 1.15% and 1.34% on a dry-weight basis. The next three report
1.62%, 1.65% and 1.61%. All meet the company’s broad composition
specification.
The correct response is not “failed,” but it is also not “nothing
happened.” Review origin, harvest period, blending records, sample plan,
method revision and laboratory controls. The sustained step change may
represent different source material or measurement behavior.
Example
2: Apparent 7-OH rise caused by reporting limits
Older reports show 7-OH as <0.01%. A new laboratory
reports 0.004%, 0.006% and 0.005% using a lower quantitation limit.
Entering each older result as zero creates an apparent increase.
Entering each as 0.01% creates an apparent decrease. Neither conclusion
is justified. Mark the laboratory transition, retain censored values and
avoid combining the eras without a scientifically supported
approach.
Example 3:
Microbial counts after a treatment change
A treated powder line historically reports low total aerobic counts.
After equipment maintenance, counts remain below the specification but
rise across four lots.
The pattern may justify review of post-treatment handling, equipment
reassembly, sanitation, environmental monitoring, packaging delay and
sample timing. It does not prove pathogens are present, and it does not
justify changing the specification to match the new results.
Example 4: Lead
result near an upper limit
Ten leaf lots show lead results well below the established limit. Two
new lots from the same source are close to the upper limit and have
overlapping measurement uncertainty.
The quality team should use its pre-established decision rule,
confirm units and reporting basis, review the sampling plan, consider
laboratory discussion and assess related lots. Rounding a borderline
result until it appears comfortably below the limit is not a sound
decision rule.
Example 5:
Extract and leaf mixed in one chart
A dashboard shows dramatic “alkaloid inconsistency.” Examination
reveals that it combines plain leaf percentages, extract milligrams per
gram and liquid milligrams per milliliter.
The chart has no defensible meaning. Separate product families and
reporting bases. Do not normalize products into a supposed “leaf
equivalent” unless the conversion has a defined, supported purpose and
inputs.
Twenty warning
signs in a COA trend program
- One product’s COA is repeatedly used to represent later lots.
- Leaf, extract and enhanced products appear in one series.
- Dry-weight and as-received results are mixed.
ND,<LODand<LOQare
all entered as zero.- A laboratory change is not marked.
- The method identifier disappears from the data record.
- Reporting limits change without explanation.
- “Pass” is converted into a numerical value.
- Total arsenic and inorganic arsenic are combined.
- Yeast-and-mold counts are treated as mycotoxin results.
- Microbial enumeration and pathogen absence are placed on the same
scale. - Specification lines are labeled as control limits.
- Control limits are recalculated immediately after an unusual result
to hide the signal. - Failed lots are deleted from the history.
- Retested values replace original results without preserving the
investigation. - Only favorable analytes are trended while panel gaps are
ignored. - Supplier, origin or process changes are absent from the chart.
- Results are compared without lot and chain-of-custody
reconciliation. - A chart is used to make medical or efficacy claims.
- A passing historical pattern is treated as proof that an untested
current lot is acceptable.
What a
trustworthy trend record should contain
A practical record may include:
- product name;
- product category;
- formulation or blend code;
- finished-product lot;
- source-material lot or lots;
- supplier;
- origin, if verified and relevant;
- manufacture date;
- sample date;
- sample plan identifier;
- collector;
- sample location or increment pattern;
- composite or discrete status;
- custody record reference;
- laboratory name;
- laboratory location;
- accession number;
- report number and version;
- report issue date;
- analyte;
- original result;
- original unit;
- reporting basis;
- qualifier;
- LOD;
- LOQ;
- reporting limit;
- normalized result, if valid;
- normalization formula;
- method identifier;
- method revision;
- instrument technique;
- uncertainty, when reported;
- specification;
- specification source;
- pass/fail decision;
- internal warning level, if applicable;
- deviation or OOS number;
- retest or resample status;
- treatment or process condition;
- packaging configuration;
- storage condition;
- supplier/process/method change marker;
- complaint or return link;
- trend signal;
- investigation summary;
- disposition;
- corrective or preventive action;
- reviewer and approval; and
- review date and next review date.
Not every field belongs on a public webpage. The point is that
defensible interpretation requires more context than a row of
numbers.
How consumers can
compare several public COAs
A consumer does not need statistical software to ask useful
questions.
- Confirm every report is for the same product type.
- Match the package lot to the current report.
- Check that the laboratory and report appear authentic.
- Compare the same analyte across reports.
- Confirm units and reporting basis match.
- Note method or laboratory changes.
- Preserve “less than” and “not detected” qualifiers.
- Look for missing panels as well as changing values.
- Treat natural botanical variation as possible, not automatically
suspicious. - Ask the seller to explain major unexplained changes.
Consumers should not use a COA history to self-prescribe or calculate
a personalized dose. A measured alkaloid value does not establish a safe
amount for a particular person, and a labeled serving is not a medical
recommendation.
How
vendors can publish useful history without creating false certainty
A transparent public archive should make the current lot easy to find
while preserving prior reports. Good practices include:
- searchable lot codes;
- report dates and version numbers;
- clear product categories;
- direct explanation of units;
- conspicuous method or laboratory changes;
- archived amended reports rather than silent replacement;
- links to guides on sampling, uncertainty and reporting limits;
- no “zero contaminant” claim based on ND;
- no promise that every future lot will match a historical average;
and - a clear route for reporting a product-quality concern.
Do not cherry-pick the strongest alkaloid result as the permanent
product description. Do not publish statistical limits that consumers
could mistake for legal or safety standards without explaining their
purpose.
Trend
signals should lead to investigation, not storytelling
When a pattern changes, several explanations may be plausible:
- natural botanical variation;
- source-region or harvest shift;
- supplier substitution;
- blending change;
- moisture change;
- treatment or drying change;
- sample-plan change;
- storage or packaging effect;
- laboratory or method change;
- calibration or reference-standard issue;
- transcription or unit-conversion error; or
- a genuine process problem.
The chart does not choose among them. Investigation should compare
independent records and consider whether related lots or distributed
products may be affected.
NIST’s guidance on out-of-control measurement signals lists
possibilities such as setup errors, recording errors, changes in
reference artifacts, instrumentation degradation, environmental
conditions and operator effects. These examples reinforce an important
point: an unusual analytical result can come from the measurement
process as well as the product.
Repeating a test without a written reason is not a complete
investigation. If an original result is invalidated, the scientific
reason and supporting evidence should remain in the record. Resampling
also asks a different question from retesting the same prepared or
retained sample.
Trend review does not
replace lot release
Every lot decision needs its own evidence. A historical pattern may
influence sampling frequency, supplier oversight or internal review, but
it should not become a shortcut that ignores the current lot.
The Part 111 model requires scientifically valid methods appropriate
for their intended use and written records supporting quality decisions.
Its record-retention rule generally requires Part 111 records to be kept
for one year past a shelf-life date, if used, or two years beyond
distribution of the last associated batch. Again, citing this framework
does not mean FDA has approved kratom or recognizes it as a lawful
dietary supplement.
A complete release review may include:
- approved specifications;
- representative sampling;
- identity evidence;
- required contaminant and composition tests;
- laboratory-control review;
- batch-production records;
- packaging and label review;
- deviation and investigation status;
- reserve-sample status; and
- documented quality-unit disposition.
The trend history adds context. It is not the release
authorization.
Product-category cautions
Plain botanical leaf and
powder
Natural variation is expected. Trend comparable supplier and product
families without turning color names or strain names into chemical
guarantees. A red, green or white marketing name does not by itself
establish a standardized alkaloid profile.
Pure-leaf capsules
Trend the leaf composition separately from capsule fill weight and
shell or ingredient attributes. Kiody’s approximately 500 mg capsules
contain pure botanical leaf rather than extract; that description should
be confirmed against the current product record before publication.
Extracts
Extraction ratio, yield, standardized percentage and milligrams per
unit answer different questions. A “10x” name is not a result. Trend
measured analytes using clearly defined units and matrices.
Enhanced leaf
Enhanced leaf combines botanical material with added concentrate or
extract. Blend uniformity becomes central. One composite sample can hide
unit-to-unit variation.
Concentrated
7-OH and manufactured derivatives
Do not place these in an ordinary-leaf data family. Their
composition, risk, legal status and analytical needs differ. Kiody does
not sell concentrated 7-OH, MGPI, MGM-15 or MGM-16.
Current federal
note as of September 3, 2026
Analytical results and legal categories must remain separate:
- The federal proceeding concerning 7-OH above a specified
threshold remains pending. HHS extended the public-comment
deadline through September 10, 2026. The request for
information is not a final federal scheduling rule. - DEA’s separate temporary order placed MGPI, MGM-15 and
MGM-16 in Schedule I effective August 26, 2026, through
August 26, 2028, unless extended or made permanent. - DOJ updated its announcement on September 1, 2026
to say it will exercise enforcement discretion when only incidental
trace MGPI is confirmed in a product otherwise consistent with botanical
kratom. DOJ expressly states that this is not a legal
exemption, creates no numerical trace threshold and does not
apply to MGM-15, MGM-16 or manufactured, concentrated, fortified or
intentionally added MGPI. - State and local rules may independently restrict ordinary botanical
leaf, extracts, enhanced products, 7-OH or related substances. A passing
COA does not establish permission to sell, possess or ship a product in
a jurisdiction.
Because these rules are changing quickly, publication review should
verify the federal notices and Kiody’s nationwide leaf and 7-OH trackers
on the day the article goes live.
Frequently asked questions
1. What is kratom COA trend
analysis?
It is the structured comparison of comparable laboratory and quality
results across ordered lots or time periods to identify shifts, drift,
variability and recurring issues.
2. Is
one passing COA enough to prove consistent quality?
No. It is evidence about the tested sample and listed attributes. It
does not establish consistency across previous, current or future
lots.
3. Can
two passing lots still be meaningfully different?
Yes. Both may meet specifications while showing a change that
deserves review. Passing and typical are different concepts.
4. What is a run chart?
It is a plot of results in sequence, often with a reference such as a
median. It is a useful first view of shifts and trends.
5. What is a control chart?
It is a statistical chart with a center line and calculated limits
intended to identify unusual behavior in a defined process. It requires
appropriate data and assumptions.
6. Are control
limits the same as safety limits?
No. Control limits come from process behavior. Safety, legal or
product specifications come from a separately documented basis.
7.
Does an out-of-control result automatically fail the lot?
No. It signals unusual behavior that needs investigation. The lot
must still be judged against applicable specifications and quality
procedures.
8. Can a
stable process still make unacceptable product?
Yes. A process can be consistent but centered outside a required
specification.
9. Should ND be entered as
zero?
No. “Not detected” means the method did not detect the analyte under
stated conditions. It does not prove absolute absence.
10. Can <LOQ
results be averaged?
Not as ordinary precise numbers. They are censored observations.
Preserve the qualifier and obtain qualified statistical help if a
numerical analysis is necessary.
11. Can I compare percent
with ppm?
Sometimes, if the matrix and basis match. For solids, 1% equals
10,000 ppm, but dry-weight, as-received and total-alkaloid bases are
different denominators.
12. Can mg/mL be converted
to percent?
Only with adequate information, including the intended mass or volume
basis and, when needed, product density. Do not assume every liquid has
water’s density.
13. Why mark laboratory
changes?
Different methods, instruments, preparation procedures and reporting
limits can create an apparent shift even if the product did not
change.
14. How many
lots are needed for a control chart?
There is no universal kratom number. It depends on chart type,
process, variability and assumptions. A simple run chart is often more
honest when data are limited.
15. Does
natural variation make trend analysis useless?
No. It makes careful grouping and interpretation more important.
Natural variation should not become a blanket excuse for unexplained
changes.
16.
Can trend analysis prove every package in a lot is the same?
No. That depends on representative sampling, blend or fill uniformity
and process controls. One lot result cannot prove every unit is
identical.
17.
Can a seller use the best historical COA for every lot?
No. Consumers should be able to match the report to the lot they
have. A favorable old report should not stand in for current
evidence.
18. Can
trends prove a product is safe or effective?
No. Trends describe measured attributes and process behavior. They do
not establish clinical safety, therapeutic benefit or suitability for a
particular person.
19. Does a COA
establish that shipping is legal?
No. Federal, state and local rules must be checked independently.
Some jurisdictions categorically restrict botanical leaf; others
distinguish leaf from concentrated or synthetic products.
20. What should a consumer
ask for?
Ask for the current lot-specific COA, the reporting basis, laboratory
and method information, and an explanation of material changes from
prior comparable lots.
Key takeaways
- A COA is a snapshot; comparable lot histories can reveal drift and
step changes. - Specifications, control limits, warning levels and targets are
different tools. - A passing result can still be unusual, and a stable process can
still fail specifications. - Compare only like products, analytes, units, bases, methods and
sample contexts. - Preserve
ND,<LODand
<LOQ; do not convert them automatically to zero. - Mark supplier, method, laboratory and reporting-limit changes.
- Use run charts before reaching for complex statistical tools.
- A trend signal begins an investigation; it does not prove the
cause. - Trend analysis never replaces current-lot testing, batch review or
legal review. - Botanical leaf must remain distinct from extracts, enhanced
products, concentrated 7-OH and manufactured derivatives.
Sources
- 21
CFR §111.70 — Specifications - 21
CFR §111.75 — Determining whether specifications are met - 21
CFR §111.320 — Appropriate, scientifically valid laboratory
methods - 21
CFR Part 111 Subpart P — Records and recordkeeping - NIST/SEMATECH
e-Handbook — Trends in sequential process or product data - NIST/SEMATECH
e-Handbook — Shewhart control chart - NIST/SEMATECH
e-Handbook — Remedial actions for out-of-control signals - NIST/SEMATECH
e-Handbook — Process capability - FDA’s
current kratom page - HHS
7-OH threshold proceeding and September 10, 2026 comment
deadline - DEA
temporary Schedule I order for MGPI, MGM-15 and MGM-16 - DOJ
September 1, 2026 incidental-trace MGPI clarification