SIST-TP ISO/TR 22305:2008
Cigarettes - Measurement of nicotine-free dry particulate matter, nicotine, water and carbon monoxide in cigarette smoke - Analysis of data from collaborative studies reporting relationships between repeatability, reproducibility and tolerances
Cigarettes - Measurement of nicotine-free dry particulate matter, nicotine, water and carbon monoxide in cigarette smoke - Analysis of data from collaborative studies reporting relationships between repeatability, reproducibility and tolerances
This Technical Report records the data and conclusions from a review of international collaborative studies to establish the tolerance for checks of the carbon monoxide yields declared by cigarette manufacturers for their products, as specified in ISO 8243.
Cigarettes — Détermination de la matière particulaire anhydre et exempte de nicotine, de la nicotine, de l'eau et du monoxyde de carbone dans la fumée de cigarette — Analyse des données provenant d'études collectives et traitant des relations entre la répétabilité, la reproductibilité et les tolérances
Cigarete - Merjenje proste suhe snovi nikotina, nikotina, vode in ogljikovega monoksida v cigaretnem dimu - Analiza podatkov primerjalnih študij s poročanjem razmerij med ponovljivostjo, obnovljivostjo in tolerancami
General Information
Buy Standard
Standards Content (Sample)
TECHNICAL ISO/TR
REPORT 22305
First edition
2006-06-01
Cigarettes — Measurement of nicotine-
free dry particulate matter, nicotine, water
and carbon monoxide in cigarette
smoke — Analysis of data from
collaborative studies reporting
relationships between repeatability,
reproducibility and tolerances
Cigarettes — Détermination de la matière particulaire anhydre et
exempte de nicotine, de la nicotine, de l'eau et du monoxyde de
carbone dans la fumée de cigarette — Analyse des données provenant
d'études collectives et traitant des relations entre la répétabilité, la
reproductibilité et les tolérances
Reference number
ISO/TR 22305:2006(E)
©
ISO 2006
---------------------- Page: 1 ----------------------
ISO/TR 22305:2006(E)
PDF disclaimer
This PDF file may contain embedded typefaces. In accordance with Adobe's licensing policy, this file may be printed or viewed but
shall not be edited unless the typefaces which are embedded are licensed to and installed on the computer performing the editing. In
downloading this file, parties accept therein the responsibility of not infringing Adobe's licensing policy. The ISO Central Secretariat
accepts no liability in this area.
Adobe is a trademark of Adobe Systems Incorporated.
Details of the software products used to create this PDF file can be found in the General Info relative to the file; the PDF-creation
parameters were optimized for printing. Every care has been taken to ensure that the file is suitable for use by ISO member bodies. In
the unlikely event that a problem relating to it is found, please inform the Central Secretariat at the address given below.
© ISO 2006
All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized in any form or by any means,
electronic or mechanical, including photocopying and microfilm, without permission in writing from either ISO at the address below or
ISO's member body in the country of the requester.
ISO copyright office
Case postale 56 • CH-1211 Geneva 20
Tel. + 41 22 749 01 11
Fax + 41 22 749 09 47
E-mail copyright@iso.org
Web www.iso.org
Published in Switzerland
ii © ISO 2006 – All rights reserved
---------------------- Page: 2 ----------------------
ISO/TR 22305:2006(E)
Contents Page
Foreword. iv
Introduction . v
1 Scope.1
2 Statistical functions for repeatability (r), reproducibility (R) of yield measurements and
compliance tolerances for declared smoke constituent yields .1
2.1 Statistical functions for repeatability (r) and reproducibility (R) .1
2.2 A statistical model for compliance tolerances .2
3 Sources of data.3
3.1 International collaborative studies .3
3.2 UK Department of Health Cigarette Survey data .4
4 Comparison of 2003 CORESTA Collaborative Study data with those previously reported.4
4.1 General.4
4.2 Comparison of repeatability r values from CCS-03 with other collaborative studies.4
20
4.3 Comparison of reproducibility R values from CCS-03 with other collaborative studies .5
20
4.3.1 General.5
4.3.2 Relationship between reproducibility R and smoke constituent yield .5
20
4.4 Comparison of R reproducibility values from collaborative studies with measurement
100
tolerances estimated from the UK Department of Health Cigarette Survey data .5
5 Review of information relevant to setting a compliance tolerance for carbon monoxide.6
5.1 General.6
5.2 Compliance data for current tolerances.6
5.3 Confidence intervals associated with yield measurements.6
5.4 Statistical models .7
5.5 Prediction of a tolerance for CO from the relative variability in their reproducibility values.8
6 Conclusions .8
7 Recommendations.8
Annex A (informative) Background considerations on the choice of sampling procedures .31
Annex B (informative) The determination of carbon monoxide in cigarette smoke — Problems in
the evaluation of results .33
Annex C (informative) Proposals from the UK Tobacco Manufacturers Association for a
practicable tolerance for verifying cigarette packet declarations of carbon monoxide
(March, 2002).41
Annex D (informative) Analysis of bias measurements from the UK Department of Health
Cigarette Survey .59
Annex E (informative) ASIA COLLABORATIVE STUDY #11 2002/2003 .71
Annex F (informative) 2003 CORESTA Collaborative Study Report CORESTA study for the
estimation of the repeatability and reproducibility of the measurement of nicotine-free
particulate matter, nicotine and CO in smoke using the ISO smoking methods,
September 2003 .86
Bibliography .146
© ISO 2006 – All rights reserved iii
---------------------- Page: 3 ----------------------
ISO/TR 22305:2006(E)
Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards bodies
(ISO member bodies). The work of preparing International Standards is normally carried out through ISO
technical committees. Each member body interested in a subject for which a technical committee has been
established has the right to be represented on that committee. International organizations, governmental and
non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely with the
International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization.
International Standards are drafted in accordance with the rules given in the ISO/IEC Directives, Part 2.
The main task of technical committees is to prepare International Standards. Draft International Standards
adopted by the technical committees are circulated to the member bodies for voting. Publication as an
International Standard requires approval by at least 75 % of the member bodies casting a vote.
In exceptional circumstances, when a technical committee has collected data of a different kind from that
which is normally published as an International Standard (“state of the art”, for example), it may decide by a
simple majority vote of its participating members to publish a Technical Report. A Technical Report is entirely
informative in nature and does not have to be reviewed until the data it provides are considered to be no
longer valid or useful.
Attention is drawn to the possibility that some of the elements of this document may be the subject of patent
rights. ISO shall not be held responsible for identifying any or all such patent rights.
ISO/TR 22305 was prepared by Technical Committee ISO/TC 126, Tobacco and tobacco products.
iv © ISO 2006 – All rights reserved
---------------------- Page: 4 ----------------------
ISO/TR 22305:2006(E)
Introduction
0.1 Summary
The purpose of this Technical Report is to review the smoke yield data provided to Working Group
ISO/TC 126/WG 8 “Confidence intervals for the determination of carbon monoxide” and to use it as the basis
for proposing a tolerance for checks of declared carbon monoxide yields.
There are many laboratories around the world routinely measuring the nicotine-free dry particulate matter
(NFDPM), nicotine and carbon monoxide yield of cigarette brands. They can, in general, be divided into two
types: those run by cigarette manufacturers for quality monitoring and those run or contracted by regulators to
check the yield information provided by manufacturers.
These laboratories need to assess their performance against others to ensure the reliability of their
measurements. Their wide geographical spread limits such assessments on a national basis, so that
international collaborative studies provide the most practical means and generate data sets on a regular basis.
In addition to allowing individual laboratories to rank their measurements relative to others, the studies also
1)
establish confidence intervals (CIs) for the repeatability (r ) of the measurements in a single laboratory and
20
2)
reproducibility (R ) in different laboratories. The reported r and R values from each study have been used in
20
isolation but when combined, as in this report, provide a means of assessing if newly reported values are
outside the expected range. The values from the latest 2003 CORESTA study are compared in this way and
found to be within the previously reported range of values but at the lower end. There is no hard evidence,
therefore, that the harmonization work on smoking machines has reduced the variability in CO yield
measurements, but the data have been shown to be as good as the best previously reported. For this reason,
and because it was a large study including all current designs of smoking machine, it provides the most
appropriate data for estimating compliance tolerances.
3)
The measurement CIs represented by r and R provide a starting point for estimating the tolerances
20 20
relevant to compliance checks on the yield information provided by manufacturers. They need to be combined
4)
with additional information on testing and reporting as well as the inherent variability in the product
associated with routine cigarette manufacturing. The statistical model given in this report is designed to
incorporate all the relevant information to estimate compliance tolerances. The model is based upon the within
and between laboratory standard deviations for tests of 100 cigarettes, together with additional terms to
account for rounding the declared values and to include the product variability. A weakness in the model
approach stems from the lack of data for estimating the terms relating to product variability, the only source of
data being the UK Department of Health Survey, which is not specifically designed to provide such data. For
this reason the model has been used in this report without including the product terms and the calculated
5)
tolerance values [R ] compared with those from an alternative indirect prediction. Obviously, the
100+rndg
R values are lower than the true compliance tolerance since they do not include the product terms.
1
00+rndg
The simplest indirect way of predicting a CO tolerance is from the measurement variability relative to NFDPM,
for which an accepted tolerance exists. The ratio of the R values calculated from the CORESTA 2003
100+rndg
Study data was used for this purpose.
1) Based on tests of 20 cigarettes.
2) Based on tests of 20 cigarettes.
3) ISO 8243 has always included tolerances for NFDPM and nicotine but an interim CO tolerance was added in 2003
whilst ISO/TC 126/WG 8 considered a permanent value.
4) See ISO 4387 and ISO 8243.
5) Based on tests of 100 cigarettes with allowance for rounding the declared value.
© ISO 2006 – All rights reserved v
---------------------- Page: 5 ----------------------
ISO/TR 22305:2006(E)
6)
ISO 8243 provides procedures, and tolerances , for sampling both ‘over a period of time’, which is
recommended, and ‘at one point in time’. Tolerances derived from both the statistical model and ratio methods
for ‘over a period of time’ sampling are summarized below.
Parameter evaluated Carbon monoxide tolerance
20 % with a minimum of 1,5 mg
R
or
100+rndg
25 % with a minimum of 1 mg
R ratio
100+rndg
22 % with a minimum of 1,5 mg
(CO/NFDPM)
It is recommended that the compliance tolerance for CO be set at 20 % for ‘over a period of time’ sampling,
and 25 % for ‘at one point in time’ sampling, with a minimum value of 1,5 mg. This recommendation implies a
corresponding amendment of ISO 8243.
It is further recommended that the tolerances and minimum values are reviewed when compliance rates are
established from regulatory checks. It is possible that such data may only become available in the UK and
may take two or three years to assemble.
0.2 General Information
Methods of measurement specified in ISO Standards require estimates of repeatability (r) and reproducibility
[1]
(R). These are normally derived from a collaborative study conforming to the guidelines in ISO 5725-1 and
[2]
ISO 5725-2 involving as many laboratories as possible.
There is a particular problem in obtaining estimates when the measurement results in the destruction of the
product sample, for example, cigarettes or fuel for internal combustion engines. If laboratories are measuring
the physical dimensions of, say, metal nuts and their bolts, measurements can be made on the same items by
one operator within a laboratory (repeatability) and by different operators in many laboratories (reproducibility).
In this example it is always the same set of nuts and bolts which is measured throughout the experiment.
For cigarette smoke constituent determinations, the situation is entirely different. The cigarettes, once
sampled and smoked, produce a set of smoke constituent estimates, each of which is perfectly valid (provided
that the standard methods have been followed) but which cannot be repeated or confirmed. The only possible
check between data is to compare them with an accepted range of yield measurements.
[3] [4]
A series of ISO Standards exists to condition the cigarettes , to specify the smoking machine for routine
[5] [6] [7] [8]
analytical smoking and to measure smoke nicotine , smoke water and smoke carbon monoxide
[9]
(CO) .
Variation in the final yield of smoke constituent arises from all these procedures but also from manufacture of
the product (see Annex A) and from the methods of sampling. These factors require the use of special
procedures in collaborative tests on cigarette products. Product variability is minimized by the testing of
matched samples, usually taken from a single small batch production, in each participating laboratory. The
samples, therefore, do not include the normal product variability and are not representative of any individual
brand.
The r and R values from collaborative studies are thus essentially estimates of measurement variability on
near identical samples. They cannot be used directly as a tolerance for compliance checks of cigarette brands
where other sources of variability must be taken into account.
6) The ‘over a period of time’ tolerances are 15 % for NFDPM and nicotine, and 20 % for CO. The tolerances when
sampling at ‘one point in time’ are increased to 20 % for NFDPM and nicotine and 25 % for CO. In both cases, a minimum
value of 1 mg applies to NFDPM and CO and 0,1 mg nicotine.
vi © ISO 2006 – All rights reserved
---------------------- Page: 6 ----------------------
ISO/TR 22305:2006(E)
0.3 Sampling a population of cigarettes manufactured for sale
[10]
ISO 8243 specifies methods for sampling a population of cigarettes manufactured for sale. It also includes
the expected tolerances when cigarettes brands are so sampled and when smoke components are measured
using the standards detailed above.
Increasing international interest and in particular the EU Directive 2001/37/EC requiring the declaration of CO
yield on cigarette packs showed that revision of this standard was urgent. ISO/TC 126 therefore decided in
2003 to set up a working group WG 8 with the task of first making a revision to add a tolerance for CO to the
1991 edition of the standard, and then to continue to revise and if possible, simplify the text of the standard.
The first task was accomplished and ISO 8243 was published in 2003 as a minor revision. The tolerance for
CO was included on the basis of existing studies showing the need for a higher tolerance than for NFDPM.
However, further collaborative studies were conducted concurrently and the purpose of this Technical Report
is to record the data from these studies and to compare them with other sources of data not previously
reported in the ISO domain.
Any further revision will then have the most comprehensive data upon which to specify the tolerances for
nicotine-free dry particulate matter (NFDPM), nicotine and carbon monoxide.
0.4 Development of smoking machines
Pressures on laboratory efficiency and the need for greater flexibility in changing smoking parameters and
types of smoke traps, have led to the development of smoking machines of differing designs, although
meeting the requirements of ISO 3308. Evidence based on reproducibility values in ISO standards and other
sources (see Annexes B, C, D) has shown that CO measurements are more variable than NFDPM (a smoke
7)
constituent of a similar level of yield). The various members of CORESTA have assisted the manufacturers
of smoking machines to better harmonize the operating conditions of the machines by evaluating the effect of
modifications through collaborative studies. Such development has been found necessary to improve the
agreement between smoke determinations on matched samples of cigarettes from different designs (all within
the ISO 3308 specification) of smoking machines, a procedure which has been called ‘harmonization’. As a
final check on the harmonization a CORESTA Collaborative Study was set up in 2003, the details of which are
given in Annex F.
7) CORESTA: Cooperation Centre for Scientific Research Relative to Tobacco (Centre de Coopération pour les
Recherches Scientifiques Relatives au Tabac)
© ISO 2006 – All rights reserved vii
---------------------- Page: 7 ----------------------
TECHNICAL REPORT ISO/TR 22305:2006(E)
Cigarettes — Measurement of nicotine-free dry particulate
matter, nicotine, water and carbon monoxide in cigarette
smoke — Analysis of data from collaborative studies reporting
relationships between repeatability, reproducibility and
tolerances
1 Scope
This Technical Report records the data and conclusions from a review of international collaborative studies to
establish the tolerance for checks of the carbon monoxide yields declared by cigarette manufacturers for their
products, as specified in ISO 8243.
2 Statistical functions for repeatability (r), reproducibility (R) of yield measurements
and compliance tolerances for declared smoke constituent yields
2.1 Statistical functions for repeatability (r) and reproducibility (R)
ISO 5725-1 and ISO 5725-2, present the general principles for collaborative tests and give methods for the
determination of r and R.
In the present context, a collaborative test essentially entails the recruitment of as many laboratories as
possible (8 - 15 is common to provide a reasonable level of confidence in r and R, according to
ISO 5725-1:1994; 6.3.4), using ISO standard methods and procedures under repeatability conditions, to
measure matched cigarette samples covering the normal range (normally 5 different samples, according to
ISO 5725-1:1994; 6.4.1) obtained from a short production run in order to minimize the product variability (‘If
different items are to be used in different laboratories, then they shall be selected in such a way as they can
be presumed to be identical for practical purposes.’, ISO 5725-1:1994, 6.4.2).
As noted earlier, ISO requires that estimates of r and R shall be included in each standard which details a
measurement procedure. In the present standards for the determination of NFDPM (ISO 4387), nicotine
(ISO 10315) and carbon monoxide (ISO 8454), the r and R values are calculated as
r = 2,8 * s
w
2 2 ½
and R = 2,8 * [s + s ]
b w
where
s is the repeatability standard deviation between mean values of 20 cigarettes, with ± r representing
w
95 % confidence intervals on the difference between two mean values (of 20 cigarettes), determined in
one laboratory from matched samples by one operator using the same equipment within the shortest
feasible period of time;
s is the standard deviation between laboratories, with ± R representing 95 % confidence intervals on
b
the difference between mean values (again, of 20 cigarettes from matched samples), determined in two
different laboratories by different operators using different equipment.
© ISO 2006 – All rights reserved 1
---------------------- Page: 8 ----------------------
ISO/TR 22305:2006(E)
NOTE For reasons of statistical validity, it is necessary that these statistics be calculated from replicate data points,
each based on mean values of a fixed number of cigarettes for both linear and rotary smoking machines: 20 in this
instance. For a linear smoking machine, therefore, a single mean value is formed by averaging over the results from
smoking 4 channels, of 5 cigarettes, on the same smoking run. For a rotary machine, this equates to one smoking run.
Repeatability and reproducibility values based on the testing of 20 cigarettes are designated by r and R , respectively.
20 20
It should also be noted that prior to the final calculations to produce estimates of s and s , the data should be
b w
screened for possible ‘outliers’; that is, extremely high or low results relative to the large majority of the data
which, if retained, would erroneously inflate the values of r and R. Various approaches for identifying outliers
within a laboratory data set are specified in ISO 5725-2:1994 and certain techniques are recommended.
However, the standard does not recommend tests for identifying outlying laboratories, but recognises the
need for informed judgement. Clause 7.2.5 states ‘This part of ISO 5725 does not provide a statistical test by
which suspected laboratories may be judged. The primary decision should be the responsibility of the
statistical expert, ….’. Obviously suspect data is best removed if confirmed to be technically suspect by the
reporting laboratory. The consequence of removing too many results would be to erroneously reduce the
estimates of r and R; and most crucially, R would be under-estimated if results for complete laboratories were
unnecessarily removed. There is obviously a need for a cautious approach of this nature, which can result in
suspect values being reported; and some are highlighted in Tables 2 and 3.
2.2 A statistical model for compliance tolerances
It is important to appreciate that the 95 % confidence intervals based on r and R alone, would be too low if
20 20
applied in the context of checking on-pack declarations of NFDPM, nicotine and carbon monoxide. Two main
components are missing:
⎯ that due to the effects of rounding, to declare the on-pack values for NFDPM and carbon monoxide to the
nearest integer and nicotine to one decimal place, and
⎯ the component associated with the product (namely, longer-term product variability and the possible
interaction between different product designs and their measurement by separate laboratories).
In the case of rounding, this can be calculated by assuming that the ‘errors’ follow a rectangular distribution.
For example, if a mean value is to be corrected to the nearest integer, the errors would be evenly distributed
between – 0,5 and + 0,5. It follows (from mathematical analysis of this distribution-function) that a variance of
0,083 needs to be included for NFDPM and carbon monoxide when rounding is to the nearest integer, with a
variance of 0,00083 for nicotine when rounded to 1 decimal place.
Obtaining estimates of the additional product-related components of variance is not so straight forward. Ideally,
the collaborative studies carried out to estimate r and R , would have been replicated on numerous and
20 20
separate production runs of the brands tested. In the absence of this, the only data available for gaining
insight into the additional product–related statistical variation is that from the UK Department of Health Survey,
for which sampling and testing take place over a 12-month period and conform with ISO 8243. Results from
this survey are discussed in 3.2 and 4.4 of this document and a related technical paper is provided in Annex C.
In Annex C, a statistical model is presented to extend the calculated reproducibility value to include the
additional variance components due to rounding of the on-pack declared values of NFDPM, nicotine and
carbon monoxide, and those related to the product itself. This is reproduced below, firstly to illustrate the way
in which the different components of variance are combined for the purpose of estimating 95 % confidence
intervals for checking on-pack declarations and, secondly, to indicate the need for wider intervals when
sampling and testing occurs at one point in time rather than on a number of occasions over a period of time.
2 2 ½
95 % CI = ± 2 {2 [(P ± s )/5 +s ± P ] + Rounding} (1)
w b
L
where
P is the variance due to product variability over time;
2
s s the variance due to between laboratory differences for individual brands;
b
2 © ISO 2006 – All rights reserved
---------------------- Page: 9 ----------------------
ISO/TR 22305:2006(E)
P is the variance due to interaction between measurements by separate laboratories of different
L
brands;
2
s is the repeatability variance to the basis of 20 cigarettes.
w
Rounding is the variance associated with rounding.
NOTE This model assumes that the mean values obtained by a manufacturer (for determining the packet
declaration) and by a would-be regulator (for checking purposes) are each based on the results from machine-smokings of
100 cigarettes, i.e. data from the smoking of 20 cigarettes on samples obtained on each of 5 separate occasions of
production.
If the additional product-related variance components are removed from (1), the model represents the
reproducibility R for tests of 100 cigarettes.
100
½
2 2
R = ± 2 {2 [s /5 +s ]} (2)
100 w b
The effect of including the additional variance due to rounding in (2) above can be seen by comparing the R
100
values (Tables 4 and 5) with the R values (Tables 6 and 7). The increase due to rounding, whilst being
100+rndg
of practical importance, is small in comparison to the measurement variability. Its greatest impact is at the low
end of the yield range, as it diminishes with increasing yield.
The above statistical model (1) should be seen as a simplification of the full relationship for estimating the
tolerance. With a more extensive data-set, involving more laboratories than take part in the UK Survey, it may
well be shown that additional components of variance relating to measurements and the product over a period
of time should be included in this model. Even so, it does serve to show the way in which separate key
elements can be combined and indicates that when ‘spot-checks’ are made on packet declarations, the 95 %
confidence intervals will be higher than when sampling and testing is carried out on a number of occasions
over a period of time. The component P (and other possible time-related components not included in the
above model) would not be divided by 5 (i.e. the number of separate occasions of sampling and testing).
3 Sources of data
3.1 International collaborative studies
International collaborative studies take place on a regular basis, both to provide r and R values and to allow
laboratories to assess their performance against others. The latter is an essential part of validating the output
of smoke-testing laboratories (a requirement of ISO 17025
...
SLOVENSKI STANDARD
SIST-TP ISO/TR 22305:2008
01-julij-2008
&LJDUHWH0HUMHQMHSURVWHVXKHVQRYLQLNRWLQDQLNRWLQDYRGHLQRJOMLNRYHJD
PRQRNVLGDYFLJDUHWQHPGLPX$QDOL]DSRGDWNRYSULPHUMDOQLKãWXGLMVSRURþDQMHP
UD]PHULMPHGSRQRYOMLYRVWMRREQRYOMLYRVWMRLQWROHUDQFDPL
Cigarettes - Measurement of nicotine-free dry particulate matter, nicotine, water and
carbon monoxide in cigarette smoke - Analysis of data from collaborative studies
reporting relationships between repeatability, reproducibility and tolerances
Ta slovenski standard je istoveten z: ISO/TR 22305:2006
ICS:
65.160 7REDNWREDþQLL]GHONLLQ Tobacco, tobacco products
RSUHPD and related equipment
SIST-TP ISO/TR 22305:2008 en
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.
---------------------- Page: 1 ----------------------
TECHNICAL ISO/TR
REPORT 22305
First edition
2006-06-01
Cigarettes — Measurement of nicotine-
free dry particulate matter, nicotine, water
and carbon monoxide in cigarette
smoke — Analysis of data from
collaborative studies reporting
relationships between repeatability,
reproducibility and tolerances
Cigarettes — Détermination de la matière particulaire anhydre et
exempte de nicotine, de la nicotine, de l'eau et du monoxyde de
carbone dans la fumée de cigarette — Analyse des données provenant
d'études collectives et traitant des relations entre la répétabilité, la
reproductibilité et les tolérances
Reference number
ISO/TR 22305:2006(E)
©
ISO 2006
---------------------- Page: 2 ----------------------
ISO/TR 22305:2006(E)
PDF disclaimer
This PDF file may contain embedded typefaces. In accordance with Adobe's licensing policy, this file may be printed or viewed but
shall not be edited unless the typefaces which are embedded are licensed to and installed on the computer performing the editing. In
downloading this file, parties accept therein the responsibility of not infringing Adobe's licensing policy. The ISO Central Secretariat
accepts no liability in this area.
Adobe is a trademark of Adobe Systems Incorporated.
Details of the software products used to create this PDF file can be found in the General Info relative to the file; the PDF-creation
parameters were optimized for printing. Every care has been taken to ensure that the file is suitable for use by ISO member bodies. In
the unlikely event that a problem relating to it is found, please inform the Central Secretariat at the address given below.
© ISO 2006
All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized in any form or by any means,
electronic or mechanical, including photocopying and microfilm, without permission in writing from either ISO at the address below or
ISO's member body in the country of the requester.
ISO copyright office
Case postale 56 • CH-1211 Geneva 20
Tel. + 41 22 749 01 11
Fax + 41 22 749 09 47
E-mail copyright@iso.org
Web www.iso.org
Published in Switzerland
ii © ISO 2006 – All rights reserved
---------------------- Page: 3 ----------------------
ISO/TR 22305:2006(E)
Contents Page
Foreword. iv
Introduction . v
1 Scope.1
2 Statistical functions for repeatability (r), reproducibility (R) of yield measurements and
compliance tolerances for declared smoke constituent yields .1
2.1 Statistical functions for repeatability (r) and reproducibility (R) .1
2.2 A statistical model for compliance tolerances .2
3 Sources of data.3
3.1 International collaborative studies .3
3.2 UK Department of Health Cigarette Survey data .4
4 Comparison of 2003 CORESTA Collaborative Study data with those previously reported.4
4.1 General.4
4.2 Comparison of repeatability r values from CCS-03 with other collaborative studies.4
20
4.3 Comparison of reproducibility R values from CCS-03 with other collaborative studies .5
20
4.3.1 General.5
4.3.2 Relationship between reproducibility R and smoke constituent yield .5
20
4.4 Comparison of R reproducibility values from collaborative studies with measurement
100
tolerances estimated from the UK Department of Health Cigarette Survey data .5
5 Review of information relevant to setting a compliance tolerance for carbon monoxide.6
5.1 General.6
5.2 Compliance data for current tolerances.6
5.3 Confidence intervals associated with yield measurements.6
5.4 Statistical models .7
5.5 Prediction of a tolerance for CO from the relative variability in their reproducibility values.8
6 Conclusions .8
7 Recommendations.8
Annex A (informative) Background considerations on the choice of sampling procedures .31
Annex B (informative) The determination of carbon monoxide in cigarette smoke — Problems in
the evaluation of results .33
Annex C (informative) Proposals from the UK Tobacco Manufacturers Association for a
practicable tolerance for verifying cigarette packet declarations of carbon monoxide
(March, 2002).41
Annex D (informative) Analysis of bias measurements from the UK Department of Health
Cigarette Survey .59
Annex E (informative) ASIA COLLABORATIVE STUDY #11 2002/2003 .71
Annex F (informative) 2003 CORESTA Collaborative Study Report CORESTA study for the
estimation of the repeatability and reproducibility of the measurement of nicotine-free
particulate matter, nicotine and CO in smoke using the ISO smoking methods,
September 2003 .86
Bibliography .146
© ISO 2006 – All rights reserved iii
---------------------- Page: 4 ----------------------
ISO/TR 22305:2006(E)
Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards bodies
(ISO member bodies). The work of preparing International Standards is normally carried out through ISO
technical committees. Each member body interested in a subject for which a technical committee has been
established has the right to be represented on that committee. International organizations, governmental and
non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely with the
International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization.
International Standards are drafted in accordance with the rules given in the ISO/IEC Directives, Part 2.
The main task of technical committees is to prepare International Standards. Draft International Standards
adopted by the technical committees are circulated to the member bodies for voting. Publication as an
International Standard requires approval by at least 75 % of the member bodies casting a vote.
In exceptional circumstances, when a technical committee has collected data of a different kind from that
which is normally published as an International Standard (“state of the art”, for example), it may decide by a
simple majority vote of its participating members to publish a Technical Report. A Technical Report is entirely
informative in nature and does not have to be reviewed until the data it provides are considered to be no
longer valid or useful.
Attention is drawn to the possibility that some of the elements of this document may be the subject of patent
rights. ISO shall not be held responsible for identifying any or all such patent rights.
ISO/TR 22305 was prepared by Technical Committee ISO/TC 126, Tobacco and tobacco products.
iv © ISO 2006 – All rights reserved
---------------------- Page: 5 ----------------------
ISO/TR 22305:2006(E)
Introduction
0.1 Summary
The purpose of this Technical Report is to review the smoke yield data provided to Working Group
ISO/TC 126/WG 8 “Confidence intervals for the determination of carbon monoxide” and to use it as the basis
for proposing a tolerance for checks of declared carbon monoxide yields.
There are many laboratories around the world routinely measuring the nicotine-free dry particulate matter
(NFDPM), nicotine and carbon monoxide yield of cigarette brands. They can, in general, be divided into two
types: those run by cigarette manufacturers for quality monitoring and those run or contracted by regulators to
check the yield information provided by manufacturers.
These laboratories need to assess their performance against others to ensure the reliability of their
measurements. Their wide geographical spread limits such assessments on a national basis, so that
international collaborative studies provide the most practical means and generate data sets on a regular basis.
In addition to allowing individual laboratories to rank their measurements relative to others, the studies also
1)
establish confidence intervals (CIs) for the repeatability (r ) of the measurements in a single laboratory and
20
2)
reproducibility (R ) in different laboratories. The reported r and R values from each study have been used in
20
isolation but when combined, as in this report, provide a means of assessing if newly reported values are
outside the expected range. The values from the latest 2003 CORESTA study are compared in this way and
found to be within the previously reported range of values but at the lower end. There is no hard evidence,
therefore, that the harmonization work on smoking machines has reduced the variability in CO yield
measurements, but the data have been shown to be as good as the best previously reported. For this reason,
and because it was a large study including all current designs of smoking machine, it provides the most
appropriate data for estimating compliance tolerances.
3)
The measurement CIs represented by r and R provide a starting point for estimating the tolerances
20 20
relevant to compliance checks on the yield information provided by manufacturers. They need to be combined
4)
with additional information on testing and reporting as well as the inherent variability in the product
associated with routine cigarette manufacturing. The statistical model given in this report is designed to
incorporate all the relevant information to estimate compliance tolerances. The model is based upon the within
and between laboratory standard deviations for tests of 100 cigarettes, together with additional terms to
account for rounding the declared values and to include the product variability. A weakness in the model
approach stems from the lack of data for estimating the terms relating to product variability, the only source of
data being the UK Department of Health Survey, which is not specifically designed to provide such data. For
this reason the model has been used in this report without including the product terms and the calculated
5)
tolerance values [R ] compared with those from an alternative indirect prediction. Obviously, the
100+rndg
R values are lower than the true compliance tolerance since they do not include the product terms.
1
00+rndg
The simplest indirect way of predicting a CO tolerance is from the measurement variability relative to NFDPM,
for which an accepted tolerance exists. The ratio of the R values calculated from the CORESTA 2003
100+rndg
Study data was used for this purpose.
1) Based on tests of 20 cigarettes.
2) Based on tests of 20 cigarettes.
3) ISO 8243 has always included tolerances for NFDPM and nicotine but an interim CO tolerance was added in 2003
whilst ISO/TC 126/WG 8 considered a permanent value.
4) See ISO 4387 and ISO 8243.
5) Based on tests of 100 cigarettes with allowance for rounding the declared value.
© ISO 2006 – All rights reserved v
---------------------- Page: 6 ----------------------
ISO/TR 22305:2006(E)
6)
ISO 8243 provides procedures, and tolerances , for sampling both ‘over a period of time’, which is
recommended, and ‘at one point in time’. Tolerances derived from both the statistical model and ratio methods
for ‘over a period of time’ sampling are summarized below.
Parameter evaluated Carbon monoxide tolerance
20 % with a minimum of 1,5 mg
R
or
100+rndg
25 % with a minimum of 1 mg
R ratio
100+rndg
22 % with a minimum of 1,5 mg
(CO/NFDPM)
It is recommended that the compliance tolerance for CO be set at 20 % for ‘over a period of time’ sampling,
and 25 % for ‘at one point in time’ sampling, with a minimum value of 1,5 mg. This recommendation implies a
corresponding amendment of ISO 8243.
It is further recommended that the tolerances and minimum values are reviewed when compliance rates are
established from regulatory checks. It is possible that such data may only become available in the UK and
may take two or three years to assemble.
0.2 General Information
Methods of measurement specified in ISO Standards require estimates of repeatability (r) and reproducibility
[1]
(R). These are normally derived from a collaborative study conforming to the guidelines in ISO 5725-1 and
[2]
ISO 5725-2 involving as many laboratories as possible.
There is a particular problem in obtaining estimates when the measurement results in the destruction of the
product sample, for example, cigarettes or fuel for internal combustion engines. If laboratories are measuring
the physical dimensions of, say, metal nuts and their bolts, measurements can be made on the same items by
one operator within a laboratory (repeatability) and by different operators in many laboratories (reproducibility).
In this example it is always the same set of nuts and bolts which is measured throughout the experiment.
For cigarette smoke constituent determinations, the situation is entirely different. The cigarettes, once
sampled and smoked, produce a set of smoke constituent estimates, each of which is perfectly valid (provided
that the standard methods have been followed) but which cannot be repeated or confirmed. The only possible
check between data is to compare them with an accepted range of yield measurements.
[3] [4]
A series of ISO Standards exists to condition the cigarettes , to specify the smoking machine for routine
[5] [6] [7] [8]
analytical smoking and to measure smoke nicotine , smoke water and smoke carbon monoxide
[9]
(CO) .
Variation in the final yield of smoke constituent arises from all these procedures but also from manufacture of
the product (see Annex A) and from the methods of sampling. These factors require the use of special
procedures in collaborative tests on cigarette products. Product variability is minimized by the testing of
matched samples, usually taken from a single small batch production, in each participating laboratory. The
samples, therefore, do not include the normal product variability and are not representative of any individual
brand.
The r and R values from collaborative studies are thus essentially estimates of measurement variability on
near identical samples. They cannot be used directly as a tolerance for compliance checks of cigarette brands
where other sources of variability must be taken into account.
6) The ‘over a period of time’ tolerances are 15 % for NFDPM and nicotine, and 20 % for CO. The tolerances when
sampling at ‘one point in time’ are increased to 20 % for NFDPM and nicotine and 25 % for CO. In both cases, a minimum
value of 1 mg applies to NFDPM and CO and 0,1 mg nicotine.
vi © ISO 2006 – All rights reserved
---------------------- Page: 7 ----------------------
ISO/TR 22305:2006(E)
0.3 Sampling a population of cigarettes manufactured for sale
[10]
ISO 8243 specifies methods for sampling a population of cigarettes manufactured for sale. It also includes
the expected tolerances when cigarettes brands are so sampled and when smoke components are measured
using the standards detailed above.
Increasing international interest and in particular the EU Directive 2001/37/EC requiring the declaration of CO
yield on cigarette packs showed that revision of this standard was urgent. ISO/TC 126 therefore decided in
2003 to set up a working group WG 8 with the task of first making a revision to add a tolerance for CO to the
1991 edition of the standard, and then to continue to revise and if possible, simplify the text of the standard.
The first task was accomplished and ISO 8243 was published in 2003 as a minor revision. The tolerance for
CO was included on the basis of existing studies showing the need for a higher tolerance than for NFDPM.
However, further collaborative studies were conducted concurrently and the purpose of this Technical Report
is to record the data from these studies and to compare them with other sources of data not previously
reported in the ISO domain.
Any further revision will then have the most comprehensive data upon which to specify the tolerances for
nicotine-free dry particulate matter (NFDPM), nicotine and carbon monoxide.
0.4 Development of smoking machines
Pressures on laboratory efficiency and the need for greater flexibility in changing smoking parameters and
types of smoke traps, have led to the development of smoking machines of differing designs, although
meeting the requirements of ISO 3308. Evidence based on reproducibility values in ISO standards and other
sources (see Annexes B, C, D) has shown that CO measurements are more variable than NFDPM (a smoke
7)
constituent of a similar level of yield). The various members of CORESTA have assisted the manufacturers
of smoking machines to better harmonize the operating conditions of the machines by evaluating the effect of
modifications through collaborative studies. Such development has been found necessary to improve the
agreement between smoke determinations on matched samples of cigarettes from different designs (all within
the ISO 3308 specification) of smoking machines, a procedure which has been called ‘harmonization’. As a
final check on the harmonization a CORESTA Collaborative Study was set up in 2003, the details of which are
given in Annex F.
7) CORESTA: Cooperation Centre for Scientific Research Relative to Tobacco (Centre de Coopération pour les
Recherches Scientifiques Relatives au Tabac)
© ISO 2006 – All rights reserved vii
---------------------- Page: 8 ----------------------
TECHNICAL REPORT ISO/TR 22305:2006(E)
Cigarettes — Measurement of nicotine-free dry particulate
matter, nicotine, water and carbon monoxide in cigarette
smoke — Analysis of data from collaborative studies reporting
relationships between repeatability, reproducibility and
tolerances
1 Scope
This Technical Report records the data and conclusions from a review of international collaborative studies to
establish the tolerance for checks of the carbon monoxide yields declared by cigarette manufacturers for their
products, as specified in ISO 8243.
2 Statistical functions for repeatability (r), reproducibility (R) of yield measurements
and compliance tolerances for declared smoke constituent yields
2.1 Statistical functions for repeatability (r) and reproducibility (R)
ISO 5725-1 and ISO 5725-2, present the general principles for collaborative tests and give methods for the
determination of r and R.
In the present context, a collaborative test essentially entails the recruitment of as many laboratories as
possible (8 - 15 is common to provide a reasonable level of confidence in r and R, according to
ISO 5725-1:1994; 6.3.4), using ISO standard methods and procedures under repeatability conditions, to
measure matched cigarette samples covering the normal range (normally 5 different samples, according to
ISO 5725-1:1994; 6.4.1) obtained from a short production run in order to minimize the product variability (‘If
different items are to be used in different laboratories, then they shall be selected in such a way as they can
be presumed to be identical for practical purposes.’, ISO 5725-1:1994, 6.4.2).
As noted earlier, ISO requires that estimates of r and R shall be included in each standard which details a
measurement procedure. In the present standards for the determination of NFDPM (ISO 4387), nicotine
(ISO 10315) and carbon monoxide (ISO 8454), the r and R values are calculated as
r = 2,8 * s
w
2 2 ½
and R = 2,8 * [s + s ]
b w
where
s is the repeatability standard deviation between mean values of 20 cigarettes, with ± r representing
w
95 % confidence intervals on the difference between two mean values (of 20 cigarettes), determined in
one laboratory from matched samples by one operator using the same equipment within the shortest
feasible period of time;
s is the standard deviation between laboratories, with ± R representing 95 % confidence intervals on
b
the difference between mean values (again, of 20 cigarettes from matched samples), determined in two
different laboratories by different operators using different equipment.
© ISO 2006 – All rights reserved 1
---------------------- Page: 9 ----------------------
ISO/TR 22305:2006(E)
NOTE For reasons of statistical validity, it is necessary that these statistics be calculated from replicate data points,
each based on mean values of a fixed number of cigarettes for both linear and rotary smoking machines: 20 in this
instance. For a linear smoking machine, therefore, a single mean value is formed by averaging over the results from
smoking 4 channels, of 5 cigarettes, on the same smoking run. For a rotary machine, this equates to one smoking run.
Repeatability and reproducibility values based on the testing of 20 cigarettes are designated by r and R , respectively.
20 20
It should also be noted that prior to the final calculations to produce estimates of s and s , the data should be
b w
screened for possible ‘outliers’; that is, extremely high or low results relative to the large majority of the data
which, if retained, would erroneously inflate the values of r and R. Various approaches for identifying outliers
within a laboratory data set are specified in ISO 5725-2:1994 and certain techniques are recommended.
However, the standard does not recommend tests for identifying outlying laboratories, but recognises the
need for informed judgement. Clause 7.2.5 states ‘This part of ISO 5725 does not provide a statistical test by
which suspected laboratories may be judged. The primary decision should be the responsibility of the
statistical expert, ….’. Obviously suspect data is best removed if confirmed to be technically suspect by the
reporting laboratory. The consequence of removing too many results would be to erroneously reduce the
estimates of r and R; and most crucially, R would be under-estimated if results for complete laboratories were
unnecessarily removed. There is obviously a need for a cautious approach of this nature, which can result in
suspect values being reported; and some are highlighted in Tables 2 and 3.
2.2 A statistical model for compliance tolerances
It is important to appreciate that the 95 % confidence intervals based on r and R alone, would be too low if
20 20
applied in the context of checking on-pack declarations of NFDPM, nicotine and carbon monoxide. Two main
components are missing:
⎯ that due to the effects of rounding, to declare the on-pack values for NFDPM and carbon monoxide to the
nearest integer and nicotine to one decimal place, and
⎯ the component associated with the product (namely, longer-term product variability and the possible
interaction between different product designs and their measurement by separate laboratories).
In the case of rounding, this can be calculated by assuming that the ‘errors’ follow a rectangular distribution.
For example, if a mean value is to be corrected to the nearest integer, the errors would be evenly distributed
between – 0,5 and + 0,5. It follows (from mathematical analysis of this distribution-function) that a variance of
0,083 needs to be included for NFDPM and carbon monoxide when rounding is to the nearest integer, with a
variance of 0,00083 for nicotine when rounded to 1 decimal place.
Obtaining estimates of the additional product-related components of variance is not so straight forward. Ideally,
the collaborative studies carried out to estimate r and R , would have been replicated on numerous and
20 20
separate production runs of the brands tested. In the absence of this, the only data available for gaining
insight into the additional product–related statistical variation is that from the UK Department of Health Survey,
for which sampling and testing take place over a 12-month period and conform with ISO 8243. Results from
this survey are discussed in 3.2 and 4.4 of this document and a related technical paper is provided in Annex C.
In Annex C, a statistical model is presented to extend the calculated reproducibility value to include the
additional variance components due to rounding of the on-pack declared values of NFDPM, nicotine and
carbon monoxide, and those related to the product itself. This is reproduced below, firstly to illustrate the way
in which the different components of variance are combined for the purpose of estimating 95 % confidence
intervals for checking on-pack declarations and, secondly, to indicate the need for wider intervals when
sampling and testing occurs at one point in time rather than on a number of occasions over a period of time.
2 2 ½
95 % CI = ± 2 {2 [(P ± s )/5 +s ± P ] + Rounding} (1)
w b
L
where
P is the variance due to product variability over time;
2
s s the variance due to between laboratory differences for individual brands;
b
2 © ISO 2006 – All rights reserved
---------------------- Page: 10 ----------------------
ISO/TR 22305:2006(E)
P is the variance due to interaction between measurements by separate laboratories of different
L
brands;
2
s is the repeatability variance to the basis of 20 cigarettes.
w
Rounding is the variance associated with rounding.
NOTE This model assumes that the mean values obtained by a manufacturer (for determining the packet
declaration) and by a would-be regulator (for checking purposes) are each based on the results from machine-smokings of
100 cigarettes, i.e. data from the smoking of 20 cigarettes on samples obtained on each of 5 separate occasions of
production.
If the additional product-related variance components are removed from (1), the model represents the
reproducibility R for tests of 100 cigarettes.
100
½
2 2
R = ± 2 {2 [s /5 +s ]} (2)
100 w b
The effect of including the additional variance due to rounding in (2) above can be seen by comparing the R
100
values (Tables 4 and 5) with the R values (Tables 6 and 7). The increase due to rounding, whilst being
100+rndg
of practical importance, is small in comparison to the measurement variability. Its greatest impact is at the low
end of the yield range, as it diminishes with increasing yield.
The above statistical model (1) should be seen as a simplification of the full relationship for estimating the
tolerance. With a more extensive data-set, involving more laboratories than take part in the UK Survey, it may
well be shown that additional components of variance relating to measurements and the product
...
Questions, Comments and Discussion
Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.