Validation and Verification of Analytical Testing Methods Used for Tobacco Products

Administrative guidance, not enacted legislation. This captured publisher edition preserves the listed document and attachments; later developments are not independently consolidated.

Validation and Verification of
Analytical Testing Methods Used
for Tobacco Products
Guidance for Industry
Comments may be submitted at any time for Agency consideration. Electronic comments may be
submitted to https://www.regulations.gov. Alternatively, submit written comments to the
Dockets Management Staff, Food and Drug Administration, 5630 Fishers Lane, Room 1061,
Rockville, MD 20852. All comments should be identified with docket number FDA-2021-D-
0756.
For questions regarding this guidance, contact the Center for Tobacco Products at (Tel) 1-877-
CTP-1373 (1-877-287-1373) Monday-Friday, 9 a.m. – 4 p.m. EDT.
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Tobacco Products
January 2025

Contains Nonbinding Recommendations
i
Table of Contents
I. 	INTRODUCTION............................................................................................................. 1
II. 	BACKGROUND ............................................................................................................... 2
III. 	DEFINITIONS .................................................................................................................. 3
IV. 	TOBACCO PRODUCT ANALYTICAL TEST METHODS ....................................... 6
A. 	General Recommendations ........................................................................................................... 6
1. Accuracy and Precision ................................................................................................................... 8
2. Selectivity ....................................................................................................................................... 11
3. Linearity and Range ....................................................................................................................... 12
a. Limit of Quantification (LOQ) ....................................................................................................... 13
b. Limit of Detection (LOD) ............................................................................................................... 13
B. 	Analytical Test Method Development ....................................................................................... 13
C. 	Analytical Test Method Development Recommendations........................................................ 13
D. 	Statistical Analysis ....................................................................................................................... 14
V. 	ANALYTICAL MEASUREMENT PROCEDURE .................................................... 14
A. 	General Recommendations ......................................................................................................... 15
B. 	AMP Specific Recommendations................................................................................................ 15
VI. 	TOBACCO PRODUCT ANALYTICAL SAMPLING PROCEDURES................... 18
A. 	General Recommendations ......................................................................................................... 18
B. ASP Specific Recommendations .................................................................................................... 18
VII. 	ANALYTICAL TEST METHOD VALIDATION ACCEPTANCE CRITERIA .... 20
A. 	Total Error Acceptance Ranges ................................................................................................. 20
VIII. 	REFERENCE STANDARDS ........................................................................................ 21
A. 	General Reference Standards ..................................................................................................... 21
B. 	Tobacco Product Reference Standards ...................................................................................... 21
IX. 	ANALYTICAL TEST METHOD VERIFICATION OF TOBACCO PRODUCTS 21

Contains Nonbinding Recommendations
1
Validation and Verification of Analytical
Testing Methods Used for Tobacco Products
Guidance for Industry1
This guidance represents the current thinking of the Food and Drug Administration (FDA or Agency) on
this topic. It does not establish any rights for any person and is not binding on FDA or the public. You
can use an alternative approach if it satisfies the requirements of the applicable statutes and regulations.
To discuss an alternative approach, contact the FDA staff responsible for this guidance as listed on the
title page.
I. 	INTRODUCTION
FDA is issuing this guidance to provide information and recommendations on how tobacco
product manufacturers can produce validation and verification data for the analytical procedures
and methods used to support regulatory submissions, such as substantial equivalence applications
(SE), premarket tobacco product applications (PMTA), and modified risk tobacco product
applications (MRTPA), for finished tobacco products. The recommendations address analytical
testing of tobacco product constituents, ingredients, and additives, as well as chemical stability
testing of finished tobacco products. The principles in this guidance may also be useful for
finished tobacco product testing and reporting of harmful and potentially harmful constituents
(HPHCs) in tobacco products and tobacco smoke, aerosol, vapor, gas, extract or any other
emission. Applications often have data and information to support analytical methods used for
testing, and this guidance is intended to help tobacco product manufacturers and analytical
laboratories assemble and present scientifically valid data and information to support the
analytical methodologies used for regulatory submissions of finished tobacco products. The
recommendations in this guidance will help FDA better understand and assess the reliability of
the data presented in regulatory submissions.
FDA acknowledges that other validation procedures and recommendations may differ from those
in this guidance. For example, there may be alternative approaches to matters such as the
number of replicates for validation procedures, appropriate acceptance criteria, and statistical
tests, among other things. Such alternative validation procedures and recommendations may be
useful considerations for applicants to generate data and information in support of analytical
methodologies used for regulatory submissions to FDA. Notwithstanding, all validation
information in support of submissions to FDA should provide data to demonstrate that the
1 This guidance was prepared by the Office of Science, Office of Compliance and Enforcement, and Office of
Regulations in the Center for Tobacco Products at FDA.

Contains Nonbinding Recommendations
2
analytical test methods are suitable and reliable for their intended purposes. The use of the words
acceptable and acceptability in this guidance refer to the relative reliability of test results; they
do not mean that FDA will or will not accept the study or regulatory submission.
In general, FDA’s guidance documents do not establish legally enforceable responsibilities.
Instead, guidances describe the Agency’s current thinking on a topic and should be viewed only
as recommendations, unless specific regulatory or statutory requirements are cited. The use of
the word should in Agency guidances means that something is suggested or recommended,
but not required.
II. 	BACKGROUND
The Family Smoking Prevention and Tobacco Control Act (Pub. L. 111–31) (Tobacco Control
Act), enacted on June 22, 2009, amended the Federal Food, Drug, and Cosmetic Act (FD&C
Act) and provided FDA with the authority to regulate the manufacture, marketing, and
distribution of cigarettes, cigarette tobacco, roll-your-own (RYO) tobacco, and smokeless
tobacco products to protect the public health and to reduce tobacco use by minors. The Tobacco
Control Act additionally gave FDA authority to issue regulations deeming other tobacco
products to be subject to FDA's authority. In May of 2016, FDA issued a final rule deeming all
products that meet the statutory definition of "tobacco product" in the FD&C Act, except
accessories of such tobacco products, to be subject to Chapter IX of the FD&C Act. The
Consolidated Appropriations Act of 2022 (the Appropriations Act), enacted on March 15, 2022,
amended the definition of the term “tobacco product” in section 201(rr) of the FD&C Act to
include tobacco products containing nicotine from any source. The Appropriations Act further
specified that the term “tobacco product” in regulations and guidances issued, in whole or in
part, under the FD&C Act shall have the meaning of and shall be deemed amended to reflect the
meaning of the amended definition in section 201(rr). As of April 14, 2022, manufacturers,
distributors, importers, and retailers of tobacco products containing non-tobacco nicotine
(NTN)—that is, nicotine not made or derived from tobacco, such as synthetic nicotine—must
comply with applicable requirements under the FD&C Act and implementing regulations.
The FD&C Act requires, among other things, premarket review for new tobacco products and
modified risk tobacco products [see Sections 910 and 911 of the FD&C Act, as well as 21 CFR
part 11142 and 21 CFR part 11073], and the reporting of harmful and potentially harmful
constituents under section 904. Regulatory submissions often contain data from analytical
testing, such as data about ingredients, constituents, and additives. In standard practice,
analytical testing is accomplished using validated analytical methods. In these cases, the
applicant will want to use analytical methods that are sufficiently precise, accurate, selective,
sensitive and have suitable linearity and range. FDA supports the use of national and
international standards for analytical test methods, e.g., Cooperation Centre for Scientific
Research Relative to Tobacco (CORESTA), and demonstration of competency to perform
testing, e.g., ISO 17025, for the analysis of finished tobacco products. This guidance provides
2 Premarket Tobacco Product Applications and Recordkeeping Requirements, (86 FR 55300, October 5, 2021)
available at https://www.govinfo.gov/content/pkg/FR-2021-10-05/pdf/2021-21011.pdf
3 Substantial Equivalence Reports; Food and Drug Administration Actions on Substantial Equivalence Reports, (86
FR 55224, October 5, 2021) available at https://www.govinfo.gov/content/pkg/FR-2021-10-05/pdf/2021-21009.pdf

Contains Nonbinding Recommendations
3
recommendations to ensure that the methods being validated will improve the efficiency of
FDA’s application review process. Validation involves documenting, using specific laboratory
investigations, that the performance characteristics of the method, including the analytical
measurement procedure and analytical sampling procedure, are suitable and reliable for the
intended analytical applications, in terms of precision, accuracy, selectivity, and sensitivity.
III. 	DEFINITIONS
For purposes of this guidance only, the following definitions apply:
Accuracy of an analytical method describes the closeness of mean test results obtained by the
analytical method to the true value (concentration or mass) of the analyte. The accuracy of the
method approximates the determinate error in a measurement.
Analytical measurement procedure (AMP) is a detailed step-by-step instruction that describes
sample preparation from a stock solution or homogeneous sample solution through to the
instrument measurement of the samples and standards.
Analytical sampling procedure (ASP) describes a detailed step-by-step sampling of the tobacco
product for the analyte(s) of interest. This procedure describes all steps necessary to prepare a
stock solution or homogeneous sample solution suitable for the intended analytical measurement
procedure. [Note: this is different from the sampling procedure which describes the steps taken
to obtain a representative sample of the tobacco product under test. The sampling procedure is
not further described in this document.]
Analytical test method (Analytical method) consists of the analytical sampling procedure and
analytical measurement procedure.
Certified reference standard, in the context of this guidance, is defined as a single analyte of
interest that is commercially available from a trusted, third-party source and has a certificate of
analysis (COA) accompanying the analyte. A certified reference standard is used for quality
control and quality assurance during preparation and analysis of test material. Examples are
nicotine and N-nitrosonornicotine (NNN).
Certified tobacco reference standard, in the context of this guidance, is a tobacco product with
known and defined characteristics that is commercially available from a trusted, third-party
source and has a COA accompanying the product. A certified tobacco reference standard is used
for quality control and quality assurance during preparation and analysis of test
material. Examples are NIST SRM 3222 for cigarette tobacco filler, University of Kentucky
1R6F cigarette reference, and University of Kentucky 1C1 cigar reference.
Homogenous sample is a portion of the tobacco material under test that is internally consistent
and representative of the bulk of the tobacco material.
Homogenous sample solution is a solution prepared from the homogeneous sample through all
required sampling steps (e.g., grinding, extraction, dilution) prior to testing.

Contains Nonbinding Recommendations
4
Intermediate precision is a measurement of the degree of variability in a measurement or set of
measurements within a laboratory which may be typical of expected changes in the laboratory.
Changes that are expected to occur include temporal changes (e.g., across different days),
collection of data by different individuals, and collection using different instruments and
sampling setups.
Limit of detection is the lowest analyte concentration that can be reliably differentiated from the
background noise.
Limit of quantification is the lowest analyte concentration that can be quantified by the analytical
test method with acceptable precision and accuracy.
Linearity is the ability of a method, within a certain range of analyte quantity, to provide an
instrumental response proportional to the quantity of the analyte in the test sample. An
acceptable linearity comparison does not necessarily require a straight-line correspondence.
Performance acceptance criteria (Performance criteria) are the pre-determined requirements,
defined by the analyst, that describe acceptable values for the validation or verification
parameters of the method or procedure subject to examination.
Precision of an analytical test method describes the closeness of individual measures of an
analyte when a procedure is applied repeatedly to multiple aliquots of a single homogeneous
solution of an analyte. Precision approximates the indeterminate error in a measurement, in that
a greater precision of a measurement denotes a lower indeterminate error of that measurement.
Precision is a combination of repeatability, intermediate precision, reproducibility, and
robustness. [Note: reproducibility is not a subject of this guidance because it is based on
interlaboratory variability]
Range is the interval of analyte concentrations over which the method provides suitable accuracy
and precision.
Repeatability is precision obtained under observational conditions where independent test results
are obtained with the same method on identical test items in the same test facility by the same
operator using the same equipment within short intervals of time.
Resolution is the chromatographic separation of two components in a mixture, calculated by:
Rs = 1.18 x (t R2 – t R1 )/(W1,h/2 + W 2,h/2 ),4
where t R2 and t R1 are the retention times of the two components; and W 1,h/2 and W 2,h/2 are the
corresponding peak widths at the half height.
Robustness is the measure of an analytical test method’s capacity to remain unaffected by small,
but deliberate, variations in method parameters and provides an indication of the reliability of the
method during normal usage.
4 United States Pharmacopeia, Chapter 621, Chromatography. Harmonized standard, December 1, 2022. Accessed
on January 4, 2024, https://www.usp.org/harmonization-standards/pdg/excipients/chromatography

Contains Nonbinding Recommendations
5
Sample matrix (Matrix) is everything that is part of the sample except for the analyte(s) of
interest.
Selectivity is the ability of an analytical test method to differentiate and quantify the analyte of
interest in the presence of other matrix components present in the sample. Selectivity should be
established at the limit of quantification.
Sensitivity is the change in response of the analytical test method in relation to a change in
concentration of the analyte.
Stock solution is a concentrated solution of an analyte that may be diluted with a solvent to lower
concentrations for use in testing. Note: this is not the homogeneous sample solution.
Tobacco product reference standard is a tobacco product with known characteristics and is in the
same category and subcategory of the test material. It is used for quality control and quality
assurance during preparation and analysis of test material. It may be certified or uncertified.
Total Error Probability is a statistical approach of combining the accuracy error and precision
error and is considered in terms of a probability function. This probability function demonstrates
that, given these errors, the likelihood that the true value of the analyte can be determined is
within a user-defined confidence interval.
Validation means a process of demonstrating or confirming that the analytical test method is
suitable and reliable for its intended purpose. The suitability and reliability of a method is
determined by comparing the results of measurements of the validation parameters to the
performance criteria of the intended purpose. Validation of an analytical test method applies to a
specific laboratory, for a specific tobacco product formulation, and equipment performing the
analytical test method for an intended use over a reasonable period of time. A product
formulation is a full statement of the components or parts, materials, ingredients, additives,
constituents, properties, and the principles of operation, of a tobacco product. A reasonable
period of time is a relative term based on the criteria for their specific application. A change in
product formulation can result in measurements that are outside of the validated ranges of the
AMP due to differences in the sample matrix, resulting in a method that is not suitable for the
intended purpose. A validated method can be extended to other tobacco product formulations,
different laboratories within a single company or facility, and across minor changes in equipment
(similar to acceptable adjustments described in USP-NF Chromatography <621>) through a
verification study.
Validation parameters include accuracy (recovery), precision (repeatability, intermediate
precision, and robustness), linearity/range, selectivity, and sensitivity (limit of quantification and
limit of detection).
Verification means the demonstration of a laboratory’s ability to successfully meet performance
criteria established for an analytical test method previously validated in the laboratory
performing the verification. Verification is typically recommended following a change to one of
the procedures in a method or a change to the tobacco product being tested. The extent of

Contains Nonbinding Recommendations
6
verification studies needed varies depending upon the extent of the change(s) to the method, and
may include accuracy and precision, selectivity, limit of quantification, or as appropriate, other
critical validation parameters.
IV. 	TOBACCO PRODUCT ANALYTICAL TEST METHODS
A. 	General Recommendations
FDA recommends tobacco product manufacturers consider using testing facilities that are
accredited and compliant with the standards listed in this guidance when conducting validation
and verification of analytical test methods used to support a reporting submission, record, or
document. Use of these facilities provides greater confidence in the consistency of the results
that may be expected from that facility. For certain information provided in a PMTA or SE
Report, the submission must contain documentation showing that the laboratory or laboratories is
(or are) accredited by a nationally or internationally recognized external accreditation
organization.5 These records are useful for checking the veracity of data.
At the most basic level, validation of an analytical test method is a systematic evaluation of the
errors inherent in the measurement of a sample. An analytical test method may be determined to
be suitable for its intended purpose when the calculated total error falls below the acceptable
error of the measurement. An analytical test method may be determined to be reliable by
evaluating the total error relative to acceptable error over a statistically valid period of time. A
statistically valid period of time is dependent upon the analytical test method employed. There
are different statistical tests that are used to evaluate trends and changes over time. The
validation procedure described in this guidance provides a means to demonstrate the suitability
and reliability of an analytical test method. Other validation procedures, e.g., ICH6, FDA7, USP8,
may also be considered and useful to perform validation of an analytical test method.
Validation of an analytical test method includes two separate major components, the analytical
measurement procedure validation and the analytical sampling procedure validation. The two
major components may be modular in nature. The analytical measurement procedure may be
used for a number of analytes and may be applied to different tobacco product matrices.
Likewise, the analytical sampling procedure validation may also be applied to different analytes
and different tobacco product categories and sub-categories.
5 See Premarket Tobacco Product Applications and Recordkeeping Requirements, (86 FR 55300, October 5, 2021),
§§ 1114.7(i)(1)(v)(G), (i)(2)(vii)(C) and (i)(4)(i); § 1114.7(j)(2)(vi)(C); Substantial Equivalence Reports; Food and
Drug Administration Actions on Substantial Equivalence Reports, (86 FR 55224, October 5, 2021), §
1107.19(d)(1)(ii)(G) and § 1107.19(f)(4)(iv).
6 International Conference on Harmonization (ICH), Validation of Analytical Procedures: Text and Methodology,
Q2(R1), Technical Requirements for Registration of Pharmaceutical s for Human Use, 2005.
7 U.S. Food and Drug Administration, Guidelines for the Validation of Chemical Methods in Food, Feed, Cosmetics,
and Veterinary Products, 3 rd Edition, 2019; U.S. Food and Drug Administration, Foods Program, Guidelines for the
Validation of Analytical Methods for the Detection of Microbial Pathogens in Foods and Feeds, Edition 3, 2019.
U.S. Food and Drug Administration, Foods Program, Guidelines for the Validation of Analytical Methods for
Nucleic Acid Sequence-Based Analysis of Food, Feed, Cosmetics and Veterinary Products, Edition 1.2, 2023;
U.S. Department of Health and Human Services Food and Drug Administration, Center for Drug Evaluation and
Research (CDER), Center for Biologics Evaluation and Research (CBER), Analytical Procedures and Methods
Validation for Drugs and Biologics, Guidance for Industry, 2015.
8 United States Pharmacopeia (USP), Analytical Procedure Lifecycle, <1220>, 2022.

Contains Nonbinding Recommendations
7
The analytical test method validation is intended to define the performance limitations of the
analytical test method. The performance limitations are defined by the performance criteria of
the analytical test method, which are defined prior to the beginning of method development and
the subsequent validation. The performance criteria typically include information such as the
target analyte, an approximation of the range of concentrations of the analyte in the sample, the
intended purpose of the analytical test method (e.g., qualitative, quantitative, major component,
minor component, etc.), and the number of samples to be analyzed. Performance criteria are
typically described through validation parameters with associated acceptance criteria.
The validation parameters that are recommended for a validation study of an analytical test
method include: accuracy, precision, selectivity, sensitivity, linearity, and range. These
parameters help ensure that the procedures used to report constituents, ingredients, additives, and
stability testing are suitable, accurate, and precise enough to produce reliable results over time,
and that they are sensitive and selective enough to ensure that the reported values are
representative of the actual product under test. Results obtained from a validated test method
demonstrate that the characteristics reported are representative of the actual product under test,
and the calculation of exposure and risks associated with those characteristics can be
appropriately considered. An example of performance criteria for an analytical test method is
described in Table 1.
Table 1: Hypothetical Example Performance Criteria for Nicotine Assay of Tobacco Filler9
Validation Parameter(s) 	Measurement Approach 	Acceptance Criteria
Accuracy and Repeatability 	Area under the
curve/integration
Total Error Probability10 Not
More than 0.05 (conditions:
Lower Limit: 85%
Upper Limit: 115%)
Accuracy and Intermediate
Precision
Area under the
curve/integration
Total Error Probability Not
More than 0.05 (conditions:
Lower Limit: 85%
Upper Limit: 115%)
Robustness 	Area under the
curve/integration
Total Error Probability Not
More than 0.05 (conditions:
Lower Limit: 85%
Upper Limit: 115%)
Selectivity 	Resolution 	Not Less Than 1.5
Limit of Quantification 	Serial Dilution (μg/g) 	Not More Than 0.1 (μg/g)
Limit of Detection 	S/N approximation (μg/g) 	Not More Than 0.05 (μg/g)
Linearity 	5 concentrations, 5
replicates at each
concentration
Total Error Probability Not
More than 0.05 per
concentration (conditions:
Lower Limit: 85%
Upper Limit: 115%)
9 Note: this table uses acceptance criteria based on the approaches described more fully in the remainder of this
guidance.
10 See section IV(A)(1)(a).

Contains Nonbinding Recommendations
8
Measurements for each analyte in the matrix should be independently validated. Generally,
analytical measurement procedure validations are conducted using certified reference standards
in compatible solutions (see Section V). Note that the analytical measurement procedure
encompasses all sample manipulation steps (typically only dilution) beginning from the
homogenous sample solution.
The analytical sampling procedure validation is intended to define the performance limitations of
the sample collection and manipulation necessary to obtain a homogenous sample solution.
Performance criteria of the analytical sampling procedure are also described through validation
parameters with associated acceptance criteria. The validation parameters that are recommended
for a validation study of an analytical sampling procedure include: accuracy, precision,
sensitivity, linearity, and range.
Throughout this guidance, a recommended number of replicates for a given validation parameter
is provided. These recommendations help to substantiate the results of the type of testing
conducted.11Although other validation procedures may differ in number of replicates
recommended, generally, more replicates provide better estimations of the true value being
measured.12 The reliability of the results of the validation are directly affected by the number of
replicates collected. The number of replicates necessary to obtain an acceptable probability for
validation purposes does not affect the number of replicates that should be used during the
routine usage of the method or procedure. However, in cases where the material or the method is
more variable than the acceptance ranges, increasing the number of replicates is a general
practice and is recommended. Given the wide array of methods and procedures that can be
validated, it is important to tailor the number of replicates used for validation purposes to the test
method or procedure (analytical sampling, analytical measurement) being validated. Some
analytical test method validations may be performed with fewer replicates than what is
recommended in this guidance, while others may need a larger number of replicates to meet the
performance criteria established for the method or procedure being validated.
1. 	Accuracy and Precision
The accuracy of an analytical test method describes the closeness of mean test results obtained
by the method to the true value (concentration or mass) of the analyte. The accuracy
approximates the determinate error in a measurement. Accuracy is determined by replicate
analysis of samples containing known amounts of the analyte. Measuring accuracy generally
11 Mark T. Stauffer, Calibration and Validation of Analytical Methods - A Sampling of Current Approaches,
IntechOpen Publishing, 2018, available at https://doi.org/10.5772/intechopen.69918; Standard Operating
Procedure for Validation of Analytical Methods of Tobacco Product Contents and Emissions, SOP02, World Health
Organization, 2017, available at https://www.who.int/publications/i/item/standard-operating-procedure-for-
validation-of-analytical-methods-of-tobacco-product-contents-and-emissions; S. Chandran, R. S. P. Singh,
Comparison of various international guidelines for analytical method validation, Pharmazie, 2007,
https://doi:10.1691/ph.2007.1.5064; Ghulam A. Shabir, Step-by-Step Analytical Methods Validation and Protocol in
the Quality System Compliance Industry, Institute of Validation Technology, 2004; Marc Andre Goulet and Denis
Cousineau, “The Power of Replicated Measures to Increase Statistical Power,” Advances in Methods and Practices
in Psychological Science, 2019, available at https://doi.org/10.1177/2515245919849434;
12 DeBievre, P. and H. Gunzler, Validation in Chemical Measurement, Springer Science & Business Media, 2005,
available at https://doi.org/10.1007/b138530;

Contains Nonbinding Recommendations
9
means that several tests should be done at any given concentration of a known amount of
analyte. Generally, seven or more determinations per concentration can provide sufficient data
to evaluate accuracy.
In general, the more replicates used the better the statistical analysis and, therefore, the better the
subsequent determination of variability of the assay to determine a specific amount of an analyte.
More replicates provide more data points and a greater number of data points allows greater
accuracy and certainty in the conclusion.
For accuracy, as opposed to precision, more replicates are needed over a range of concentrations
of analyte because of the nature of the error being determined. Accuracy measures errors that
are always present in every measurement. Every measurement also contains random error that
changes with each measurement. Random errors can be either positive or negative, and therefore
a larger collection of values tends to cause these errors to cancel themselves out of the
calculation, allowing the underlying accuracy to be determined. Therefore, more replicates
should generally be used to calculate accuracy than what is used to calculate precision, or
linearity.
The precision of an analytical test method describes the closeness of individual measures of an
analyte when the procedure is applied repeatedly to multiple aliquots of a single homogeneous
solution of an analyte. Precision approximates the indeterminate error in a measurement. The
precision value is calculated as the coefficient of variation (CV) of the measurement and is
approximately the same as the expected analyte concentration in the intended sample (referred to
as “target concentration” hereafter). The CV is often referred to as the percent relative standard
deviation (%RSD) and the terms are used interchangeably in most laboratories. Precision can be
further subdivided into repeatability (ISO-r), reproducibility (ISO-R), and intermediate precision,
each of which is evaluated differently depending upon whether the analytical measurement
procedure, analytical sampling procedure, or analytical test method is being examined. These
concepts are discussed in greater detail in sections V, VI, and VII, respectively. Generally, five
or more determinations per concentration can provide sufficient data to evaluate precision. In
the case of a method with high variability, additional replicates are beneficial and are
recommended.
Because every measurement is a summation of the true value, the determinate error, and the
indeterminate error, knowledge of all errors is important for an approximation of the true value.
Thus, a special effort should be made to isolate and quantify the relative amounts of the two error
types from the true value. This is generally accomplished by measuring replicate solutions of a
certified reference standard at a known concentration. Depending upon which portion of the
validation is being considered (e.g., accuracy, precision), the standard solutions are treated
differently, and the number of replicate solutions will change. This is discussed further in
sections V, VI, and VII.
Consider an example in which an analyst made two sets of replicate solutions using a certified
reference standard, in this case nicotine. One solution set has a known concentration of 1.0
μg/mL and the other set has a known concentration of 10.0 μg/mL. The analyst then made the
measurements shown in Table 2.

Contains Nonbinding Recommendations
10
Table 2: Hypothetical Example: test results from a method with 1.0 μg/mL and 10.0 μg/mL of
certified nicotine reference standard
Replicate 	Measured concentration (μg/mL)
(known concentration = 1.0
μg/mL)
Measured concentration (μg/mL)
(known concentration = 10.0
μg/mL)
1 	1.1 	10.5
2 	0.8 	10.5
3 	1.3 	10.4
4 	1.0 	10.2
5 	0.7 	10.6
6 	0.7 	10.1
7 	1.0 	10.0
Mean 	0.94 	10.33
% Recovery 	94.3% 	103.3%
Std. Dev. 	0.223 	0.229
RSD (also
known as
Coefficient
of
Variation,
CV)
23.6% 	2.2%
Based on these measurements, the accuracy of the 1.0 μg/mL set would be calculated as a %
recovery of 94.3% and precision would be calculated as 23.6% RSD. The 10.0 μg/mL set would
have a % recovery of 103.3% and a precision of 2.2% RSD. Because it is unclear if either or
both data sets would have an acceptable total error of measurement, the total error of the
measurement may be calculated using the formula below.
a. 	Total Error Evaluation
The accuracy and precision values may be considered individually or in a total error context.
Typically, analytical test methods represent a compromise and balance among sensitivity,
accuracy, and precision. Therefore, setting acceptance criteria based upon consideration of
precision and accuracy in isolation may result in a validation expectation that cannot be met.
Instead, a total error evaluation provides a better means to demonstrate the suitability of
procedures and methods for their intended purposes. A total error calculation is not a simple
summation of the precision and accuracy values. Instead, total error is considered in terms of a
probability function. This probability function is the likelihood that the true value of an analyte
at the target concentration will pass the acceptance criteria of the analyte at a user defined
confidence level.13 FDA recommends the use of a 95% confidence level as this is the accepted
scientifically valid standard. This probability function is calculated as:
Total Error Probability = 1- (t(qU,df) − t(qL,df)),
13 Ermer, J. and P. W. Nethercote, Method Validation in Pharmaceutical Analysis. Singapore, Wiley-VCH, 2014.
Note: The equation in Ermer calculates the probability that the results are within the acceptance criteria; the
equation above 1- (t(q U,df) − t(q L ,df)), calculates the probability that the results are outside the acceptance criteria.

Contains Nonbinding Recommendations
11
where:
t(q,df) = cumulative Student's t distribution with degrees of freedom (df) evaluated at
quantile, q
qU = (Upper-Mean) / (RSD*√(1 + 1/n))
qL = (Lower-Mean)) / (RSD*√(1 + 1/n))
df = n − 1
n = number of replicate standard solutions prepared
Upper = upper limit of the generally accepted range of the analyte (e.g., 102.0%)
Lower = lower limit of the generally accepted range of the analyte (e.g., 98.0%)
Mean = % recovery = average measured value/known value*100, report as %
RSD = standard deviation of a measurement/average measured value of the
measurements*100, report as %
When calculated in this manner, a total error probability of not more than 0.05 indicates the
precision and accuracy are suitable for their intended purposes.14 The selection of suitable upper
and lower acceptance criteria of the measurement are critical decisions and are discussed more
fully in sections V, VI, and VII.
As an example of the application of total error probability, consider the application of the
performance criteria included in Table 1 to the data collected in Table 2, assuming that this data
is a repeatability experiment. Table 1 states that the upper and lower values are 115% and 85%,
respectively, with an acceptance criteria of a total error probability not more than 0.05. In the
formula above, the mean value of the first data set (0.94 μg/mL) is calculated as a percentage of
the known value, which is 94.3%, the RSD is calculated as 23.6% (see Table 2). The resulting
total error probability would be 0.42, which does not meet the stated acceptance criteria.
Therefore, this method would be considered unsuitable for its intended application (the
evaluation of lower concentrations of nicotine). However, the data from the second set would
result in a mean value of 103.3% and the RSD is 2.2%, which would yield a total error
probability of 0.001. This meets the performance criteria and therefore demonstrates that the
method may be suitable (pending the evaluation of the other performance criteria) for the
evaluation of higher concentrations of nicotine.
2. Selectivity
Selectivity is the ability of an analytical test method to differentiate and quantify the analyte in
the presence of other components in a sample. Selectivity is evaluated depending upon the type
of procedure being validated. For chromatographic procedures, which make up the majority of
procedures used in tobacco chemistry, the evaluation of the resolution is made between the
analyte and interferences, including background disturbances, enantiomers, similar chemical
species, and unidentified co-eluting species. If the method is intended to quantify more than one
analyte at a time in the sample, FDA recommends each analyte be tested to ensure that there is
no interference caused by the analytes and other chromatographic features. A resolution of
14 An example for performing this calculation is to use the following formula in Microsoft® Excel®: = 1-
(T.DIST((Upper-Mean)/(SQRT(1+1/number of reps)*RSD))-T.DIST((Lower-Mean)/(SQRT(1+1/number of
reps)*RSD)))

Contains Nonbinding Recommendations
12
greater than 1.5 is generally considered to be acceptable by chromatographers. However, the
lack of chromatographic resolution does not necessarily indicate that an analytical test method
fails validation, especially if the analytical measurement procedure uses a sufficiently specific
detector (like a mass spectrometer, multichannel UV detector, serial detection techniques (e.g.,
LC-UV-MS)). When chromatographic resolution cannot be or is not achieved, FDA
recommends other approaches be considered. One possible alternative approach is the
evaluation of test mixtures of important constituents with known concentrations for bias caused
by the interference (the resolution of interferences is discussed in further detail in section V.B).
The additional bias found in these experiments may be added to the accuracy value and the total
error probability of the method is calculated. For example, one approach is to take the bias, e.g.,
uncertainty of NIST 3222, and add it to the total error equation as follows:
1-(T.DIST(((Upper+uncertainty)-Mean)/(SQRT(1+1/number of reps)*RSD))-
T.DIST(((Lower-uncertainty)-Mean)/(SQRT(1+1/number of reps)*RSD)))
3. Linearity and Range
The linearity and range of a method or procedure in the validation experiment is defined as the
range of concentrations for which an acceptable level of precision and accuracy can be achieved.
The acceptability of the measurements at each concentration may be evaluated using the total
error probability described above. The acceptance criteria are specific to the procedure and
method being validated and are discussed further in section VII.
A calibration (standard) curve is the relationship between instrument response and known
concentrations of the analyte. This type of curve is typically plotted during the method
development phase and is then regularly plotted as a component of system suitability evaluations
on an on-going basis. This helps to assure that the analytical test method is functioning as
validated and reduces loss of samples or time because of problems found after testing is
completed. FDA recommends calibration curves be generated for each analyte in the sample and
include at least five concentrations spread across the intended calibration range of the method,
with three concentrations in the linear portion of the curve and two approaching the asymptotic
portions. During method development, it is further recommended that a sufficient number of
samples at each concentration be used to adequately define the relationship between
concentration and response. However, calibration curves typically employed for routine use are
less rigorous and lack the accepted international guidelines for the scientific stringency necessary
for the evaluation of linearity and range components of the validation activity.15
As part of validation, you should use the number of replicates that would support the fitness of
the procedure across the range of analyte concentrations. A recommendation is to use five or
more replicates of a defined concentration of an analyte solution at five or more different
concentrations. If the cumulative number of replicates over multiple, daily calibration curves
equals five replicates at five concentrations, this may be an alternative approach. The
15 For additional information, see Seyed Mojtaba Moosavi and Sussan Ghassabian, Linearity Of Calibration Curves
For Analytical Methods: A Review Of Criteria For Assessment Of Method Reliability, 2018, available at
https://www.intechopen.com/books/calibration-and-validation-of-analytical-methods-a-sampling-of-current-
approaches/linearity-of-calibration-curves-for-analytical-methods-a-review-of-criteria-for-assessment-of-method

Contains Nonbinding Recommendations
13
concentrations selected should span the entire working range of the procedure or method. The
range of concentrations should include the limit of quantification, the target concentration level,
the highest concentration to which the method/procedure will likely be used, and at least two
additional concentrations. The 25 measured values (five replicates of five concentrations) would
be used to calculate a best fit line, R2
, slope, and intercept. However, FDA considers that using
R2 as a pass/fail criterion for validation does not adequately characterize the total error of the
measurements.
a. 	Limit of Quantification (LOQ)
The limit of quantification is the lowest concentration for which replicate solutions can be
measured with acceptable precision and accuracy. The acceptability of the concentrations may
be evaluated using the total error probability described above. The acceptance criteria are
specific to the procedure and method being validated and are discussed further in section VII.
The LOQ approximation using a signal-to-noise ratio may be used as a way to approximate the
starting point for LOQ determination; however, one should be aware that relying solely on the
signal-to-noise ratio to determine LOQ may be inaccurate and lead to the use of procedures that
are not suitable for their intended purpose.
b. 	Limit of Detection (LOD)
The limit of detection is the lowest concentration at which an analyte can be reliably
differentiated from the background noise. The traditional approach is approximating this using
three times the signal-to-noise ratio. The LOD can also be empirically determined through serial
dilutions of a standard or spiked analyte solution until a predetermined total error probability is
found, but the outcomes of this approach are not generally better than the traditional signal-to-
noise approach.
B. 	Analytical Test Method Development
Analytical test method development should (1) satisfy the objectives of the analytical test
method and (2) meet method performance criteria. All analytical test methods are developed to
measure the specified analytes over a specific concentration range. The method development
and establishment phase generally define the limits of the linear dynamic range (linearity and
sensitivity), selectivity, robustness of the method, and the solution stability and storage
conditions. It is important to note that method development and method validation are different
activities. The method development may provide an insight into the boundary conditions of a
procedure; however, only the evaluation as part of a validation will demonstrate the suitability of
the procedure for the intended purpose. Once the method has been established, any data (such as
selectivity measurements, or replicate measurements) collected during its development may be
used as part of a validation study, where reasonable. Note also that substantive changes made to
the procedures between method establishment and validation completion may affect the
applicability of the data collected and may therefore invalidate the validation.
C. 	Analytical Test Method Development Recommendations
When analytical testing is used to comply with a requirement in chapter IX of the FD&C Act or
FDA’s implementing regulations, submitters should use an analytical test method that has been
validated (see section V of this guidance for reference). Analytical test methods should be
performed under optimized conditions and within the limitations of the analytical test method.

Contains Nonbinding Recommendations
14
Analytical test methods should meet system suitability requirements (if applicable) to ensure that
the complete testing system (including instruments, reagents, equipment, and laboratory
personnel) is suitable for the intended purpose of analysis. Doing so will help ensure the data
produced and submitted to FDA is accurate and reliable for FDA to use in making
determinations regarding requirements of the FD&C Act or implementing regulations.
Certified reference standards should be used when executing system suitability testing,
validation, and verification of analytical test methods (e.g., calibration standards, internal
standards, system suitability standards, and controls). To have optimal method suitability, FDA
recommends a certified tobacco reference standard be used when available.
System suitability acceptance criteria should be determined during the analytical test method
development or optimization phase. Thereafter, analysts should determine that a method meets
the system suitability criteria before accepting the results generated from an analytical test
method. The frequency and acceptance criteria of system suitability evaluations should be
determined by the analyst, with an understanding of the risk of discounting analytical data
obtained while the system was not shown to be suitable. Typically, laboratories determine
system suitability daily and have check standards interspersed during longer sample runs.
Laboratories should document activities and ensure that method and data quality criteria are met,
including operational maintenance and performance verification of instruments; certificates of
analysis for reagents and standards; and properly labeled laboratory solutions with established
expiration dates. All reference standards and solutions that have expired should not be used for
any regulatory analytical testing, including validation and verification, as these solutions can add
a level of uncertainty and lead to unreliable results.
Laboratories should use a laboratory reagent blank with every analytical test to detect potential
contamination during the test material preparation and analysis process. The reagent blank or
blanks are used to identify any cross contamination and interference prior to samples being run.
The analyte being measured should be absent or below the limit of detection in the laboratory
reagent blank for the particular method used. If the analyte being measured in the test material is
detected in the laboratory reagent blank, this would skew the amount of unknown analyte in the
test material, creating unreliable results. As such, it is recommended that the results of the
analysis utilizing contaminated test material should not be reported and the cause(s) of the
contamination should be properly diagnosed and corrected before the analysis is repeated on the
laboratory reagent blank and test material.
D. 	Statistical Analysis
Statistical analyses should be used to evaluate samples, analyze data, and both verify and
validate performance characteristics. Statistical analyses of data generated during verification
should be evaluated against the predetermined analytical test method validation performance
criteria. All statistical procedures and parameters used in the analysis of the data should be
based on sound statistical principles suitable for the intended evaluation.
V. 	ANALYTICAL MEASUREMENT PROCEDURE

Contains Nonbinding Recommendations
15
A. 	General Recommendations
Analytical measurement procedure (AMP) validation experiments are conducted using solutions
of the analyte or analytes in representative matrices (matrices that are similar to the composition
of a sample that will be subject to the procedure, e.g., NIST 1593a Tomato leaves, NIST 3222
cigarette tobacco filler.). The analytical measurement procedure starts with a solution
immediately following the final extraction/solubilization of the analyte from the sample matrix
(e.g., extracts from a Cambridge filter pad, extract from tobacco filler, DNPH solution, etc.).
Therefore, all filtrations, centrifugations, liquid/liquid extractions, and dilutions should not be
considered as error sources in the evaluation of an analytical measurement procedure. For
example, the extraction efficiency, grinding, and puffing protocols should not be included in this
step, because they are components of the analytical sampling procedure validation. Replicates
for AMP validation are prepared from separate aliquots of the collected final extraction/
solubilization solution.
The validation parameters recommended to complete the validation of the AMP include:
accuracy, precision (repeatability, intermediate precision, robustness), selectivity, linearity,
range, and sensitivity (limit of quantification, and limit of detection).
B. 	AMP Specific Recommendations
Accuracy measurements for the AMP are collected using replicate standard solutions in a
representative matrix (see section V.A above). The accuracy is based on replicate spiked
solutions (see section IV.A.1 above) and is typically reported as the % recovery; % recovery is
calculated from the mean value of the analyte as measured in the replicates as compared to the
known concentration of the analyte in the solution. For example, the data included in Table 2
above presents the % recovery of solutions having a known concentration. The % recovery is
calculated by dividing the mean measured concentration by the known concentration and
multiplying the result by 100%. Generally, increasing the number of replicates will result in a
better estimation of any systematic error in the system and therefore is a better estimate of the
true value. The number of replicates should always be enough to meet the predetermined criteria
of the validation. The analyst should consider the effects of replicates prior to starting
validation.
Repeatability (precision) measurements for the AMP may be collected from the measurements
used in the accuracy parameter above. The repeatability is reported as a % RSD and is
calculated from the mean value and the standard deviation of the measurements. An example of
this approach is demonstrated in the examples of total error probability above. Alternatively, in
cases where an analyst wishes to minimize the number of replicates and the method is expected
to be well within the expected performance and acceptance criteria, calibration curves collected
on seven different days could be used for the accuracy, repeatability, linearity, and intermediate
precision measurements. However, this approach does represent an increased risk of failing a
total error probability calculation because of variability introduced by day-to-day variability.
Intermediate precision for the AMP is intended to probe the typical changes that may occur with
an actively used procedure in a single laboratory. Therefore, recommended components of the
intermediate precision experiment include changes caused by differences in instruments,
analysts, and different days. In cases where there is only one instrument in a laboratory, it is

Contains Nonbinding Recommendations
16
possible to focus the intermediate precision experiment only on the analyst and day. Regardless
of the combination of the variables, FDA recommends six independent scenarios be used to
complete the analysis; six independent measurements taken in six independent conditions (day,
instrument, and/or analyst) generally provides an adequate model of the errors expected.16 Each
scenario should be separately collected from a different AMP sample manipulation. Each of
these sets should start by utilizing the sample manipulations from the same working solution.
Examples statistically demonstrating some considerations for intermediate precision conditions:
Assuming a laboratory has two qualified analysts (A and B), two calibrated instruments (α and β)
and several days (numbers 1-6) in which to complete this experiment. The laboratory might
instruct:
• 	Analyst A to use instrument α on six days (Aα1, Aα2, Aα3, Aα4, Aα5, Aα6)
• 	Analyst A to use instruments α and β on three days (Aα1, Aβ1, Aα2, Aβ2, Aα3, Aβ3)
• 	Analysts A and B to use instrument α on three days (Aα1, Bα1, Aα2, Bα2, Aα3, Bα3)
• 	Analysts A and B to use instruments α and β on two days (Aα1, Bα1, Aβ1, Bβ1, Aα2,
Bα2, Aβ2, Bβ2)
• 	Other variations may also be suitable
If the laboratory instructs the following, it would likely not provide an adequate number of
independent scenarios:
• 	Analyst A to use instrument α for three days and Analyst B to use instrument β on same
three days (Aα1, Aα2, Aα3) and (Bβ1, Bβ2, Bβ3). The two combinations are
independent but are not part of the same sample set and, therefore, would be two sets of
three determinations, rather than one set of six.
• 	Analyst A to use instrument α, twice a day for three days. Multiple experiments by a
single analyst typically do not provide insight into the longer-term variability that this
experiment attempts to emulate.
• 	Analyst A and B to use instrument α for two days (Aα1, Bα1, Aα2, Bα2)
Intermediate precision is reported as a % RSD and is calculated from the mean value across all
conditions tested and the standard deviation of the measurements in the same manner as the
repeatability evaluation described above.
Robustness is typically evaluated using a set of experiments intended to identify the boundaries
of acceptable instrument setting adjustments that can be made without causing a change in the
AMP. In these experiments, the analyst systematically modifies instrument conditions and
settings until the resulting data no longer meet the performance criteria or exhibit other issues
(e.g., peak fronting, baseline anomalies, peak broadening, peak splitting, etc.). In many cases,
changes to certain settings may have no detrimental effects on the AMP. In these cases, the
validation report should note the conditions that were adjusted and the extent to which the
16 Ermer, J. and P. W. Nethercote. See footnote 13.

Contains Nonbinding Recommendations
17
adjustments did not affect or change the resulting data, with special attention given to the
precision, accuracy, and selectivity. In cases where the adjustments do result in changes to the
data, especially regarding performance criteria, the condition or setting of each adjustment and
the resulting changes should also be reported so that this information can be considered if the
validated procedure is extended to other products or analytical applications.
Selectivity of an AMP may be accomplished through the evaluation of solutions containing the
analyte and the potential interferences. Where an interference is detected, it would be useful to
describe how the interference will be addressed. In cases of a chromatographic AMP, selectivity
can be obtained by demonstrating acceptable chromatography through the analysis of resolution
between a pair of peaks in the chromatogram (these two peaks are termed the critical pair). The
critical pair may be the analyte and the nearest eluting interferent identified during method
development. Typically, a solution containing the two compounds of the critical pair are
prepared from certified reference standards or is a standard mixture. This solution is used as a
system suitability solution and is often also used as a system check between sample injections.
In some cases, the critical pair may be two other peaks (neither are the analyte) found to elute
close to each other. When a critical pair does not include the analyte, data showing that the
analyte peak is resolved should be provided. Where a multi-dimensional (multi-channel)
detector is used (such as a photodiode array or tandem mass spectrometer) a demonstration that
interfering compounds do not result in a determinate error should be provided.
The linearity and range of the AMP are typically evaluated at five or more concentrations over
the linear dynamic range identified during method development. During the validation of the
AMP, the solutions measured should be prepared from certified reference standards of the
analyte in matrix matched solvents. In this case, the matched matrix to be used should be the
solvent blank. The data obtained in the collection of the LOQ and repeatability values may be
used in the linearity and range determinations. The acceptability of the linearity and range may
be evaluated by calculating the total error probability at each concentration using the
repeatability values from Table 3 below. The target concentration of the procedures will
generally be assumed to be the mid-level concentration of the calibration curve (e.g., the 3 rd
concentration in a five-point calibration curve) included in the linearity and range data sets. The
slope, intercept, and degree of fit calculation (such as R2 ) may be reported but are not generally
used in determining the validity of the AMP because these parameters are only loosely correlated
to measurements of error.17
The limit of quantification of the AMP is typically determined through serial dilutions of low
concentrations of the analyte in solvent blank. As indicated above, the LOQ is directly related to
the target concentration of the procedure. For example, if an AMP is intended to measure an
analyte at 100 μg/mL, an acceptable LOQ will be different from one intended for a 100 ng/mL,
as validation is a determination of the suitability of a method for the intended purpose. This is
because the expected amount of error is greater at lower concentrations, which is reflected in the
total error probability. See section VII for further discussion of setting acceptance criteria.
Where the AMP is intended for multiple target concentrations, LOQs for each target
concentration may be useful and are recommended.
17 Ermer, J. and P. W. Nethercote. See footnote 13.

Contains Nonbinding Recommendations
18
The limit of detection is calculated using a solvent blank for the AMP validation. The calculated
LOD represents the lowest amount of analyte that can be detected without typical matrix effects
or sampling errors. This LOD represents what can be achieved by a typical laboratory and
defines the limits of the measurement system and solution preparation processes.
VI. 	TOBACCO PRODUCT ANALYTICAL SAMPLING PROCEDURES
A. 	General Recommendations
The analytical sampling procedure (ASP) in most cases will be the source of the greatest
variability in measurements that cannot be credited to the variability of the tobacco product
itself. Validation of the ASP cannot occur without an associated AMP. Generally, any AMP
that meets the acceptance criteria of the ASP validation should provide equivalent results to an
ASP. The suitability criteria of the ASP differ slightly from that of the AMP, because of the
increased number of error sources encountered during sampling. For example, an analytical
sampling procedure intended to measure tar in cigarette smoke will contribute greater error to the
measurement than the AMP and, therefore, the acceptable error of the analytical test method will
be defined by the ASP.
Evaluations of the ASP are based on the effects of both the sampling system and any
interferences contributed by the tobacco product that affect the analytical measurement.
Therefore, a validation of the ASP should include comparative measurements of sample
extracts/smoke from a certified tobacco reference standard (e.g., 1R6F, CRP1) and of the
tobacco product category that is the subject of the validation. The use of these reference
products allows comparison to known values, thereby reducing the number errors and better
reflecting the reliability and suitability of the procedure (see discussion in section VIII(B) on
creating a tobacco reference standard when a certified tobacco reference standard is not
available). ASPs are generally developed to be applicable to all tobacco products in the same
category (e.g., smokeless tobacco). However, the target concentrations of analytes can differ
dramatically and differences in ingredients and product design may lead to substantial
differences in the types and amounts of interferences collected by the ASP. Thus, each ASP
should be validated and included for each tobacco product analytical test method in each
regulatory submission for a tobacco product. The topic of verification is discussed at greater
length in section IX.
The validation parameters recommended to complete the validation of the analytical sampling
procedure generally include accuracy and precision (repeatability, intermediate precision,
robustness), using the certified tobacco reference standard if available.
B. ASP Specific Recommendations
Generally, accuracy measurements for the ASP are made using a spike and recovery approach.
It can be difficult to prepare a solution from a tobacco product reference standard that does not
contain the analyte. Therefore, the accuracy value should be a relative content comparison. The
tobacco product reference standard is smoked/aerosolized/extracted and the resulting solution
(after extraction manipulation) serves as the blank sample solution. The blank sample solution is
then spiked (“fortified”) using a certified reference standard of the analyte at approximately the
target concentration. The difference between the blank (unfortified) sample solution and the

Contains Nonbinding Recommendations
19
fortified sample solution represents the accuracy calculated as a % recovery. The accuracy is
based on replicate spiked solutions within the AMP linear dynamic range. Generally, you should
use the number of replicates that would support the fitness of the procedure. A recommendation
is that no fewer than seven replicates are sufficient to provide accuracy information.
Repeatability measurements for the ASP may be collected from the measurements used in the
accuracy parameter above. The repeatability is reported as a % RSD and is calculated from the
mean value and the standard deviation of the measurements.
Intermediate precision for the ASP is intended to probe the typical changes that may occur with
the procedure in a single laboratory. Therefore, changes caused by differences in instruments,
analysts, and different days are individual components to include in the intermediate precision
experiment. Because ASP is intended to characterize the sources of error in the sampling
procedure, each of the precision measures collected is specific to the sample collection. In cases
where cigarette smoking is the sampling procedure, the instrument is the smoking machine. For
example, multi-port smoking machines have the potential to add a variable systematic sampling
error when there are differences in the ports. Therefore, when a multi-port smoking machine is
used, it would be appropriate to evaluate the systematic error present in the smoking machine
used to collect HPHC data. Where this evaluation indicates a difference in systematic error it
would be useful to describe how differences in the ports are addressed in a regulatory submission
for a tobacco product. In cases where there is only one instrument in your laboratory, it is
possible to focus the intermediate precision experiment only on the analyst and day. Regardless
of the combination of the variables, we recommend that at least six independent scenarios be
studied to complete the analysis as this generally provides an adequate model of the errors
expected (see the description and examples in the AMP section above for further details). As
suggested in the example, each scenario might be collected from a different ASP sample
manipulation (with corrections as appropriate), and separately collected.
Robustness is typically evaluated using a set of experiments that seeks to identify the boundaries
of acceptable instrument setting adjustments that can be made without causing a change in the
ASP (e.g., vaping/aerosolizing regimen for new and emerging products, homogenizing
smokeless tobacco). In these experiments, the analyst systematically modifies instrument
conditions and settings until the resulting data no longer meet the performance criteria or the
result exhibits other significant issues. For example, evaluating the robustness of an ASP by
altering sample-extraction times and sample-preparation ratios, is acceptable. In many cases,
changes to a certain setting have no detrimental effects on the ASP. In these cases, validation
information in support of submissions to FDA should note the conditions that were adjusted and
the extent to which the adjustments yielded no changes to the analytical assessment. In cases
where the adjustments do result in changes to the data, especially regarding performance criteria,
the condition or setting of each adjustment and the resulting changes should also be reported.
This portion of the validation is typically concluded as part of the method development. When
available from that source, no additional data is needed, but the results should be reported from
the method development information.
The limit of detection can be calculated using a sample blank, or by a spike and recovery
approach, as described in the accuracy for the ASP validation, above. The calculated LOD

Contains Nonbinding Recommendations
20
represents the lowest amount of analyte that can be detected with representative matrix effects
and sampling errors. This LOD represents the practical limits that a method can achieve for the
product and method being validated.
VII. 	ANALYTICAL TEST METHOD VALIDATION ACCEPTANCE CRITERIA
A critical component of a validation experiment is the determination of the acceptance criteria
necessary to demonstrate that the validation procedure is suitable for its intended purpose. This
determination should be made prior to the beginning of the validation experiments and should be
consistent across methods utilized for validation. The acceptance criteria selected for a
validation need not be associated with the acceptance criteria of a tobacco product. For
reference, the validation acceptance criteria applied in the evaluation of FDA-developed methods
are based upon the work of William Horwitz18 and are applied in a manner similar to that
described in the FDA Food Program Validation guidance.19
A. 	Total Error Acceptance Ranges
Generally, FDA recommends total error probability to evaluate the acceptability of the measured
error components for a given analytical test method. The total error probability calculation is a
statistical calculation that determines the probability that a measurement having the target
concentration would fall within an acceptance criterion. When using this approach, the total
error acceptance criterion should be defined prior to beginning the validation study and should be
linked to concentration of the analyte in the sample matrix and criticality of the result. The lower
the concentration, the wider the acceptance criteria should become. For example, nicotine in
burley tobacco leaf is generally between 4 to 12 milligrams per gram of tobacco. However, an
assay may measure as low as 0.1 milligram and as high as 30 milligrams and at either limit, it is
scientifically recognized that accuracy will suffer. It is important to get the highest accuracy
within the known range in which the analyte is likely to be seen. An approach to estimate the
acceptance criteria is provided by the Horwitz-Thompson equation and the repeatability.20 As
indicated in the prior sections, total error measurements for AMP validation (including accuracy,
repeatability, intermediate precision, robustness, and LOQ) use the repeatability value RSD to
calculate the upper and lower values. Total error measurements for the ASP validation
(including accuracy, repeatability, intermediate precision, robustness, and LOQ) use the
reproducibility value RSD to calculate the upper and lower values. The values provided in the
table below are for illustrative purposes only and do not represent acceptable differences
between tobacco products or acceptable levels of HPHCs or other constituents in a tobacco
product for any reporting submission, record, or document under chapter IX of the FD&C Act or
associated regulations.
18 Horwitz, W., The variability of AOAC methods of analysis as used in analytical chemistry. J. Assoc. Off. Anal.
Chem. 60: 1355-1363, 1997.
19 FDA. Foods Program. Guidelines for the Validation of Chemical Methods in Food, Feed, Cosmetics, and
Veterinary Products, 2019, https://www.fda.gov/media/81810/download.
20 Massart, D. L., et al., ‟Benchmarking for Analytical Methods: The Horwitz Curve.” LCGC Europe 18(10): 528,
2005.

Contains Nonbinding Recommendations
21
Table 3: Concentration and variability criteria
Concentration
of Analyte
(%)
Concentration
of Analyte
(ppm or ppb)
Concentration
with w/w
units (mg/g or
μg/g)
Concentration
Fraction
Suggested ASP
Limits
(Reproducibility
Value from
Horwitz,
RSD, in %)*
Suggested
AMP Limits
(Repeatability
Value from
Horwitz,
RSD, in %)*
10 	100,000 ppm 	100 mg/g 	0.1 	2.8 	1.4
1 	10,000 ppm 	10 mg/g 	0.01 	4 	2
0.1 	1000 ppm 	1 mg/g 	0.001 	5.7 	2.8
0.05 	500 ppm 	500 μg/g 	0.0005 	6 	3
0.01 	100 ppm 	100 μg/g 	0.0001 	8 	4
0.001 	10 ppm 	10 μg/g 	0.00001 	11 	6
0.0001 	1 ppm 	1 μg/g 	0.000001 	16 	8
0.00001 	100 ppb 	0.1 μg/g 	0.0000001 	22 	11
<0.00001 	< 100 ppb 	< 0.1 μg/g 	< 0.0000001 	22 	22
*The RSD in Table 3 are estimates only and are not the RSD from the measured analyte (see footnote 20 reference, Massart,
et al.)
VIII. 	REFERENCE STANDARDS
A. 	General Reference Standards
All reference standards should be suitable for their intended use. Additionally, FDA
recommends that all reference standards and prepared solutions should be properly labeled and
stored with established expiration dates. Reference standards and solutions that have expired
should not be used for any regulatory analytical testing, including validation and verification.
Additionally, all reference standards and prepared solutions should be accurate and reliable.
B. 	Tobacco Product Reference Standards
A tobacco product reference standard is a tobacco product with known characteristics and is in
the same category and subcategory of the test material. It is used for quality control and quality
assurance during preparation and analysis of test material. It may be certified or uncertified.
FDA recommends use of certified tobacco reference standards, when available. If a certified
tobacco reference standard is unavailable, a tobacco product reference standard can be
constructed in-house. The tobacco product reference standard should be for the same product
type as the test material. In addition, the tobacco product reference standard should be
homogeneous and used concurrently with every test to verify accurate determination of the
analyte(s).
IX. 	ANALYTICAL TEST METHOD VERIFICATION OF TOBACCO PRODUCTS
Once a method has been validated, any subsequent substantial21 change results in a “new”
method that should be independently validated to ensure that the changes did not adversely affect
21 Examples of the types of changes that could be considered “substantial” are described at length for
chromatographic procedures in the current USP/NF under General Chapter Chromatography <621>, section “system
suitability.”

Contains Nonbinding Recommendations
22
the suitability of the method. The extent of the validation for the “new” method depends upon
the extent of change.
Generally, validation should be conducted relative to a reference product within the same
product category and subcategory (e.g., cigarette, filtered) to the product that is to be tested. The
completed validation is applicable to the tobacco product evaluated in the validation process and
can be extended to other products within the same product category and subcategory through a
verification process. 	For example, validation of an analytical test method using a reference
cigarette (such as 1R6F) does not mean that the validated procedure is suitable for the testing of
other cigarette products. However, the validated method can be demonstrated to be applicable to
another cigarette brand by completing a verification process. The verification process should
demonstrate that the analyte of interest (e.g., nicotine or other HPHC) falls within the linear
dynamic range of the validated method and that no interferences are present in the product under
test that were not found in the reference product. A recommended approach to verification is
described below however, alternative approaches may be used.
Verification is a shortened form of a validation procedure that is used to demonstrate that a
previously validated method is suitable for use with another tobacco product in the same
category or in some cases where different instruments are used (typically in cases where the
instrument comes from a different vendor). The verification parameters recommended include
accuracy and selectivity of the sample tobacco product subject to the applied method. Additional
verification parameters may be added at the analyst’s discretion. Verification should be
completed for each analyte intended to be measured with the analytical test method to
demonstrate the assay performs as expected and provides consistent accurate results.
Verification is similar to the ASP validation approach. As with validation, performance criteria
as stringent as those typically used for purposes of validation should be determined prior to
conducting the verification experiments.
Generally, accuracy measurements for verification are made using a spike recovery approach.
Often it is difficult to prepare a solution from a reference tobacco product that does not contain
the analyte. Therefore, the accuracy value should be a relative content comparison. The
reference product is smoked/extracted and the resulting solution (after extraction manipulation)
serves as the blank sample solution. The blank sample solution is then spiked using a certified
reference standard of the analyte at approximately the target concentration. The difference
between the blank (unfortified) sample solution and the spiked (fortified) sample solution
represents the accuracy calculated as a % recovery. The accuracy is based on replicate spiked
solutions. The validation approach is described in section IV of this guidance.
Selectivity for verification may be measured by evaluating solutions containing analyte and
potential interferences. Where an interference is detected, it would be useful to describe how the
interference will be addressed. In cases of a chromatographic AMP, selectivity can be obtained
through the demonstration of acceptable chromatography through the analysis of resolution
between a pair of peaks in the chromatogram (these two peaks are termed the critical pair). The
critical pair may be the analyte and the nearest eluting interferent identified during method
development. Typically, a solution containing the two compounds of the critical pair are
prepared from certified reference standards or is a standard mixture. This solution is used as a

Contains Nonbinding Recommendations
23
system suitability solution and is often used as a system check between sample injections. In
some cases, the critical pair may be two other peaks (neither are the analyte) found to elute close
to each other. When a critical pair does not include the analyte, data showing that the analyte
peak is resolved should be documented. Where a multi-dimensional (multi-channel) detector is
used (such as a photodiode array or precursor or product ion mass spectrum of the analyte peak),
the analyst should be able to demonstrate that interfering compounds do not result in a
determinate error.