ASTM E3239-21

This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles for the
Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.
Designation: E3239 − 21
Standard Guide for
Using Statistical Process Control Principles for Routine
Dosimetry in Radiation Processing
This standard is issued under the fixed designation E3239; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (´) indicates an editorial change since the last revision or reapproval.
INTRODUCTION
Statistical process control (SPC) is one part of the rationale used to establish rules for conformance
assessment of the radiation process and processed products. The underlying rationale for product
conformance assessment as it relates to the radiation process has three components: a qualified state
that is demonstrated to be capable/reliable in terms of achieving the processed product specification
limits; SPC applied to routine process monitoring data demonstrating no change to the qualified state,
thatis,astateofstatisticalcontrol,andtheapplicationofasimpleacceptancerule;istheprocessresult
within the process specification limits. This document provides information on the application of SPC
to radiation processing with examples which include capability/reliability assessments.
1. Scope ization established in the Decision on Principles for the
Development of International Standards, Guides and Recom-
1.1 This document provides guidance for the statistical
mendations issued by the World Trade Organization Technical
analysis of the irradiation process from dosimetric data.
Barriers to Trade (TBT) Committee.
1.2 This document is one of a set of guides and practices
that provide recommendations for properly implementing do-
2. Referenced Documents
simetry in radiation processing. It is intended to be read in 3
2.1 ASTM Standards:
conjunction with ISO/ASTM 52628 and ISO/ASTM 52303.
E122 Practice for Calculating Sample Size to Estimate,With
1.3 This document employs a set of standard statistical Specified Precision, the Average for a Characteristic of a
methodsandisintendedtobereadinconjunctionwithPractice
Lot or Process
E2586, Practice E2281, Practice E2587, and ASTM Manual E456 Terminology Relating to Quality and Statistics
MNL7 .
E2281 Practice for Process Capability and Performance
Measurement
1.4 This guide is applicable to high-energy electron beam,
E2586 Practice for Calculating and Using Basic Statistics
X-ray and gamma-ray irradiation processes.
E2587 Practice for Use of Control Charts in Statistical
1.5 This document assumes user knowledge of statistics,
Process Control
radiation processing, and radiation dosimetry. (See AnnexA6)
E3083 Terminology Relating to Radiation Processing: Do-
1.6 This standard does not purport to address all of the
simetry and Applications
safety concerns, if any, associated with its use. It is the
2.2 ISO/ASTM Standards:
responsibility of the user of this standard to establish appro-
51261 Practice for Calibration of Routine Dosimetry Sys-
priate safety, health, and environmental practices and deter-
tems for Radiation Processing
mine the applicability of regulatory limitations prior to use.
51608 Practice for Dosimetry in an X-ray (Bremsstrahlung)
1.7 This international standard was developed in accor-
Facility for Radiation Processing at Energies between 50
dance with internationally recognized principles on standard-
keV and 7.5 MeV
51649 Practice for Dosimetry in an Electron Beam Facility
for Radiation Processing at Energies Between 300 keV
This guide is under the jurisdiction of ASTM Committee E61 on Radiation
and 25 MeV
Processing and is the direct responsibility of Subcommittee E61.03 on Dosimetry
Application.
Current edition approved April 1, 2021. Published July 2021. DOI:10.1520/
E3239-21. For referenced ASTM standards, visit the ASTM website, www.astm.org, or
S. Luko, ed., Presentation of Data and Control Chart Analysis, 9th ed., West contact ASTM Customer Service at service@astm.org. For Annual Book of ASTM
Conshohocken, PA, ASTM International, 2018, https://doi.org/10.1520/MNL7- Standards volume information, refer to the standard’s Document Summary page on
9TH-EB. the ASTM website.
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
E3239 − 21
51702 Practice for Dosimetry in a Gamma Facility for 3.1.3.1 Discussion—Variation in the process output is inher-
Radiation Processing ent in all processes. The inherent variation of the process
51707 Guide for Estimation of Measurement Uncertainty in output results from multiple sources both dependent and
Dosimetry for Radiation Processing independent contributing to the overall process output varia-
52303 Guide forAbsorbed-Dose Mapping in Radiation Pro- tion. These inherent sources which are always present to a
cessing Facilities greater or lesser extent are referred to as common causes.
52628 Practice for Dosimetry in Radiation Processing When only common cause sources of variation are present in
52701 Guide for Performance Characterization of Dosim- the process output, the process is considered to be stable and in
eters and Dosimetry Systems for Use in Radiation Pro- astateofstatisticalcontrol.Othersourcesoftheprocessoutput
cessing variation occurring intermittently result from sources that are
4 not always present and not inherent to the process and are not
2.3 ISO Documents:
predictable within statistical limits are referred to as special
ISO 11137-1 Sterilization of health care products — Radia-
causes. (See 3.1.32.)
tion — Part 1: Requirements for development, validation
and routine control of a sterilization process for medical 3.1.4 confidence interval—an interval estimate [L,U] with
the statistics L and U as limits for the parameter θ and with
devices
ISO 11137-2 Sterilization of health care products — Radia- confidence level 1-α, where the probability Pr(L≤θ≤U) ≥ 1-α.
tion — Part 2: Establishing the sterilization dose 3.1.4.1 Discussion—The confidence level, 1-α, reflects the
ISO 11137-3 Sterilization of health care products — Radia- proportion of cases that the confidence interval [L,U] would
tion — Part 3: Guidance on dosimetric aspects of contain or cover the true parameter value of θ in a series of
development, validation and routine control repeated random samples under identical conditions. Once L
ISO 3534-1 Statistics — Vocabulary and symbols — Part 1: and U are given values, the resulting confidence interval either
General statistical terms and terms used in probability does or does not contain it. In this sense “confidence” applies
ISO 3534-2 Statistic-Vocabulary and symbols-Part 2: Ap- not to the particular interval but only to the long run proportion
plied statistics of cases when repeating the procedure many times
ISO 11462-1 Guidelines for implementation of statistical
3.1.5 control chart—chart on which are plotted a statistical
process control (SPC) – Part 1: Elements of SPC
measure of subgroup versus time of sampling along with limits
ISO 16269-6 Statistical interpretation of data – Part 6:
based on the statistical distribution of that measure so as to
Determination of statistical tolerance intervals
indicate how much common, or chance cause variation is
2.4 ICRU Report inherent in the process or product.
ICRU Report 85a Fundamental Quantities and Units for
3.1.6 control limits—limits on a control chart that are used
Ionizing Radiation
as criteria for signaling the need for action or judging whether
a set of data does or does not indicate a state of statistical
3. Terminology
control based on a prescribed degree of risk.
3.1.6.1 Discussion—For example, typical three-sigma limits
3.1 Definitions:
carry a risk of 0.135 % of being out of control (on one side of
3.1.1 absorbed-dose mapping—measurement of absorbed
the center line) when the process is actually in control and the
dosewithinanirradiatedproducttoproduceaone,twoorthree
statistic has a normal distribution.
dimensional distribution of absorbed dose, thus rendering a
map of absorbed-dose values.
3.1.7 dose map, dose mapping—See absorbed-dose map-
3.1.1.1 Discussion—For a process load, such a dose map is
ping.
obtained using dosimeters placed at specified locations within
3.1.8 dose uniformity ratio—ratio of the maximum to the
the process load.
minimum absorbed dose within the irradiated product.
3.1.2 assignable cause—factor that contributes to the varia-
3.1.8.1 Discussion—The concept is also referred to as the
tion in a process or process output that is feasible to detect and
max/min dose ratio. Product generally refers to the process
identify.
load.
3.1.2.1 Discussion—Many factors will contribute to process
3.1.9 irradiation container—holder in which process load is
output variation, but it may not be feasible (economically or
transported through the irradiator.
otherwise) to identify some of them.
3.1.9.1 Discussion—An irradiation container is often re-
3.1.3 common cause—source of inherent random variation
ferred to simply as “container” and can be a carrier, cart, tray,
in a process (output) which is predictable within statistical
product carton, pallet, product package or other holder.
limits (also called random cause and chance cause).
3.1.10 long term standard deviation, σ ,—sample standard
LT
deviation of all individual (observed) values taken over a long
period of time.
Available from International Organization for Standardization (ISO), ISO
3.1.11 lower control limit, LCL—minimum value of the
Central Secretariat, Chemin de Blandonnet 8, CP 401, 1214 Vernier, Geneva,
Switzerland, https://www.iso.org.
control chart statistic that indicates statistical control.
Available from International Commission on Radiation Units and Measure-
3.1.12 lower specification limit, LSL—specification limit
ments (ICRU), 7910 Woodmont Ave., Suite 400, Bethesda, MD 20841-3095,
http://www.icru.org. that defines the lower limiting value.
E3239 − 21
3.1.12.1 Discussion—In terminal sterilization radiation 3.1.26 rationalsubgroup—subgroupchosentominimizethe
processing, the LSL is the sterilization dose per ISO 11137- 2. variability within subgroups and maximize the variability
between subgroups.
3.1.13 population—the totality of items or units of material
under consideration. 3.1.27 routine monitoring position—position where ab-
sorbed dose is monitored during routine processing to ensure
3.1.14 population parameter—summary measure of the val-
that the product is receiving the absorbed dose specified for the
ues of some characteristic of a population.
process.
3.1.15 process capability—statistical estimate of the out-
3.1.27.1 Discussion—This position may be a location of
come (output) of a characteristic from a process that has been
minimum or maximum dose in the process load or it may be an
demonstrated to be in a state of statistical control and which
alternate convenient location in, on or near the process load
describes that process’s ability to realize a characteristic that
where the relationship of the dose at this position to the
will fulfil the requirements for that characteristic.
minimum and maximum dose has been established.
3.1.16 process capability index—an index describing pro-
3.1.28 routine processing lot—in radiation processing,a
cess capability in relation to a specified tolerance.
user defined group of product irradiated together or sequen-
tially on or in an uninterrupted set of irradiation containers,
3.1.17 process load—volume of material with a specified
characterized by one set of routine dosimetry results and one
product loading configuration irradiated as a single entity.
set of processing conditions.
3.1.18 process output—a measurable and monitored charac-
3.1.28.1 Discussion—This is often termed a “batch”, or a
teristic which is output from the process.
“run”, or a “lot”, where such terms are usually defined locally.
3.1.18.1 Discussion—In radiation processing, the measured
3.1.29 sample—agroupofobservationsortestresults,taken
characteristic is typically the absorbed dose. However, it could
from a larger collection of observations or test results, which
also be characteristics such as dwell time, conveyor speed,
serves to provide information that may be used as a basis for
beam energy, beam width, or beam current.
making a decision concerning the larger collection.
3.1.19 process parameters (irradiator parameters)—
3.1.29.1 Discussion—Subsetofapopulationmadeupofone
specified values for the process variables.
or more sampling units.
3.1.19.1 Discussion—The specification for a sterilization
3.1.30 sample statistic—summary measure of the observed
process parameter that may include its allowable tolerances.
values of a sample.
3.1.19.2 Discussion—Examples of process parameters are
cycle time and process pathway for gamma and conveyor
3.1.31 short term standard deviation, σ —the inherent
ST
speed, beam current and scan width in electron beam. variation present when a process is operating in a state of
statistical control, expressed in terms of standard deviation.
3.1.20 process performance—statistical measure of the out-
come of a characteristic from a process that may not have been 3.1.32 special cause—source of intermittent variation in a
demonstrated to be in a state of statistical control. process output.
3.1.32.1 Discussion—Sometimes “special cause” is taken to
3.1.21 process performance index—index describing pro-
be synonymous with “assignable cause.” However, a distinc-
cess performance in relation to a specified tolerance.
tion should be recognized. A special cause is assignable only
3.1.22 process quality requirement—a confidence level cor-
when it is specifically identified.Also a common cause may be
responding to a specified maximum acceptable risk.
assignable.
3.1.23 process target dose—the expected dose or dose range 3.1.32.2 Discussion—A special cause arises because of
at the dose monitoring location(s) of a routine processing lot
specific circumstances which are not always present. As such,
for a given set of process parameters. in a process subject to special causes, the magnitude of the
process output variation from time to time is unpredictable,
3.1.23.1 Discussion—The relationship correlating an ex-
that is, not predictable within statistical limits
pected dose (or dose range) and a set of process parameters at
a desired level of confidence can be developed generally from 3.1.32.3 Discussion—In radiation processing an example of
irradiator OQ and specifically from PQ.
a special cause may be a source interrupt during processing.
3.1.24 processing category—group of different products
3.1.33 specification limit—limiting value stated for a char-
that can be processed together. acteristic.
3.1.24.1 Discussion—Processing categories can be based 3.1.33.1 Discussion—In radiation processing an example of
on, for instance, composition, density or dose requirements. a specification limit is the dose specification limit, the Upper
Specification maximum product dose limit (USL – see 3.1.42)
3.1.25 process specification—revision controlled document
and the Lower Specification minimum product dose limit (LSL
thatincludesproductspecificdetails,specifiestheproductdose
– see 3.1.12). Other examples of specification limits are the
requirements and provides necessary process instructions to be
limits associated with a process parameter (see 3.1.19)
used for routine processing of product.
3.1.25.1 Discussion—See ISO 11137 Part 1, Sections 9.4.3 3.1.34 stable process—prorocess in a state of statistical
and9.4.4forguidanceinestablishingprocessspecificationsfor control; process condition when all special causes of variation
sterilization applications. have been removed.
E3239 − 21
3.1.34.1 Discussion—Observed variation can then be attrib- 3.1.38.6 Discussion—Strata partitions are defined based on
uted to random (common) causes. Such a process will gener- the population characteristic under study and the experimental
ally behave as though the results are simple random samples design of the study.
from the same population.
3.1.38.7 Discussion—In dose mapping, strata are denoted
3.1.34.2 Discussion—This state does not imply that the
by dose magnitudes; maximum dose strata, equivalent maxi-
random variation is large or small, but rather that the variation
mum dose strata, minimum dose strata, equivalent minimum
is predictable within statistical limits.
dose strata, and intermediate dose strata so that no population
element is excluded.
3.1.34.3 Discussion—In radiation processes, the operation
of the irradiator due to cost considerations or efficiency
3.1.38.8 Discussion—In radiation processing, strata can be
considerations may define the standard process to include
defined by characteristics that represent known common cause
special causes, thus defining some special causes as anticipated
sources of variation acting on the routine process and ex-
common causes, for example, partially filled irradiation con-
pressed in the process output summary statistics. An example
tainers. In doing so, the process output characterization sam-
of this are partially filled irradiation containers or variation of
pling must include sampling of these special causes to ensure
loading configuration geometries and leading/trailing edge
the statistical model of the process output accurately captures
effects.
these sources of variation.
3.1.39 subgroup average, x¯ —average for the ith subgroup
i
3.1.35 standard deviation—of a population, σ, the square
in an X-bar chart.
root of the average or expected value of the squared deviation
3.1.40 subgroup standard deviation, s,—sample standard
i
of a variable from its mean; - of a sample, s, the square root of
deviation of the observations for the ith subgroup in an s-chart.
the sum of the squared deviations of the observed values in the
sample divided by the sample size minus 1.
3.1.41 upper control limit, UCL— maximum value of the
control chart statistic that indicates statistical control.
3.1.36 standard error—standard deviation of the population
of values of a sample statistic in repeated sampling or an
3.1.42 upper specification limit, USL—specification limit
estimate of it.
that defines the upper limiting value.
3.1.36.1 Discussion—If the standard error of a statistic is
3.1.42.1 Discussion—In terminal sterilization radiation
estimated, it will itself be a statistic with some variance that is
processing, the USL is the maximum acceptable dose per ISO
dependent on the sample size (for further description of the
11137-3.
concept of standard error, see E2586, subsection 6.19).
3.2 Definitions of other terms used in this standard that
3.1.37 Statistical Process Control (SPC)—set of techniques
pertain to quality and statistics may be found in Terminology
for improving the quality of process output by reducing
E456. Definitions of other terms used in this standard that
variability through the use of one or more control charts and a
pertain to radiation measurement and dosimetry may be found
corrective action strategy used to bring the process back into a
in Terminology E3083. Definitions in Terminology E3083 are
state of control.
compatible with ICRU 85a; that document, therefore, may be
3.1.38 stratified sampling—sampling in which the popula-
used as an alternative reference.
tion to be sampled is first divided into mutually exclusive
subsets or strata, and independent samples are taken within
4. Significance and Use
each stratum.
4.1 Control charts are the primary process monitoring tool
3.1.38.1 Discussion—Strata partitions are collectively ex-
in SPC for radiation processing. The general objectives of
haustive (no population element is excluded).
implementing a SPC program with control charts are to:
3.1.38.2 Discussion—A stratified sampling method is con-
4.1.1 Increase knowledge of the process,
ducted as a proportionate allocation or an optimum allocation.
4.1.2 Control the process to provide a targeted or required
Stratified sampling ensures that at least one observation is
process output,
selected from each strata.
4.1.3 Reduce variation of the process output or in other
3.1.38.3 Discussion—Proportionate allocation ensures the
ways improve the performance of a process, and
sample size from each stratum is proportionate to the popula-
tion size of the stratum. Proportionate allocation ensures the 4.1.4 Identify single process run results that are outside of
estimate of the overall population mean is equal to the
established control limits but may be within the USL and LSL
unweighted sample average. limits.
3.1.38.4 Discussion—Optimal allocation ensures larger
4.2 These objectives when achieved:
samples are taken in the strata with the greatest variability
4.2.1 Reduce costs through reduction of losses due to scrap,
relative to the population.
rework, and investigation time,
3.1.38.5 Discussion—Stratified sampling will nearly always
4.2.2 Improve consistency of the process output,
provide a greater precision/reliability than random sampling
4.2.3 Facilitate preventive process adjustments, and
for estimating population parameters. The greater the differ-
ence between strata, the greater the gain in precision/reliability 4.2.4 Provide evidence of accurate process targeting and
compared to random sampling. process performance; state of statistical control.
E3239 − 21
5. Prerequesites other where the analysis is used to compare the process output
dose to a qualified a priori statistical model (termed control
5.1 The dosimetry system has been calibrated in accordance
standard given).
with ISO/ASTM 51261 and the user’s measurement manage-
NOTE 4—In radiation processing, the ‘standard’ in the context of
ment system; see ISO/ASTM 52628. These standard practices
‘control standard given’ is the aimed-at value or process target (dose)
and user requirements establish traceable dosimetry with a
based on characterized relationships established in prerequisite studies,
defined level of uncertainty appropriate for the conditions of see 5.2, 7.2.1.1, and 7.2.1.2 where the latter two are the basis of the a
priori statistical model. The standard value may be an experience value
use.
based on representative a priori data, or an economic value established on
5.2 Irradiator installation qualification (IQ), irradiator op-
consideration of needs of service and cost of production, or a desired or
‘aimed-at’value. Examples deriving the standard value of a process level
erational qualification (OQ), and performance qualification
and examples of process targeting are given in A3.3 and A3.4.
dose mapping (PQ) have been completed and user documented
performance acceptance criteria have been met.
6.4 Sampling is the collection of data from a number of
observations that is purportedly representative of a larger
5.3 Implementation of appropriate product processing pro-
grouping or population.
cedures to provide control and management of the process
6.4.1 Sampling is performed to collect data from which the
inputs within their normally expected or specified limits. Such
process level and process variation expectations are derived.
procedures are part of an SPC control plan (see ISO 11462-1).
This sampling is the data collected and used in the evaluation
NOTE 1—In radiation processing, process inputs embody a number of
factors and characteristics each with specification limits that ensure phase of SPC control chart implementation. (See 6.6.1 and 7.3)
process output (for example, dose) meets expectation. The specific factors
The population parameter µ is estimated with the sample mean
and characteristics will vary due to differences in product definition,
x¯ and the population standard deviation σ is estimated with the
process definition, radiation source and irradiator control systems, quali-
sample standard deviation s.
fied control parameters values from performance qualification of the
product or product family and the resulting common cause sources of 6.4.2 The quality of the sample representation of the popu-
variation present or acting on the process.
lation is dependent on the sampling procedure and sample size.
An appropriate sampling procedure for radiation processing
6. Overview – Control Charts
addressesprocessstratificationbysamplingfromallstratawith
a sufficient frequency. (See 3.1.38.)
6.1 This section provides a general description of control
charts using dose measurements. Section 7 provides guidance
6.5 The output of a radiation process is evaluated for two
specific to the application of control charts for radiation
characteristics: a process level (subgroup average dose see
processes.
7.4.1 and Note 10) and process variation (variation of the
individual dose measurements used to calculate the subgroup
6.2 Acontrol chart is the SPC analysis tool for trending and
average dose). SPC chart trending consists of a two chart pair,
evaluating a process based on the process output (for example,
either an X-bar/s-chart pair or an X-bar/R-chart pair.
dose). The control chart is composed of three parts; the center
6.5.1 The X-bar/s-chart pair evaluates the process output
line, control limits above (UCL – upper control limit) and
level (subgroup average dose) with the x-bar chart and the
below (LCL – lower control limit) the center line, and the plot
process variation with the s-chart (variation of the individual
of the process output dosimetric data.
dose measurements used to calculate the subgroup average
NOTE 2—Dose measurements can be plotted as the measured value or
dose).
asaresidualvaluerepresentingthedifferencebetweenthemeasuredvalue
6.5.2 The X-bar/R-chart pair evaluates the process output
and the standard. Dose measurements can also be plotted as normalized
dose values, for example, dose rate corrected for exposure process level (subgroup average dose) with the x-bar chart and the
parameters like cycle time in gamma or conveyor speed in electron beam.
process variation with the R-chart (range of the individual dose
NOTE3—SPCcanalsobeappliedtonon-dosimetricprocessmonitoring
measurements used to calculate the subgroup average dose).
output data that is functionally related to a process output dose. Examples
of non-dosimetric process monitoring output are irradiation process
6.6 Control chart implementation as part of a SPC plan
control parameters, such as beam current, conveyor speed, scan rate, pulse
consists of three phases; process evaluation, process
rate, etc.
monitoring, and process improvement.
6.2.1 The center line is the value of the standard given. (See
6.6.1 Process evaluation is the derivation of the a priori
6.3 and Note 4 for description of “standard given”).
statistical model of process output.
6.2.2 The control limits are the 63σ statistical limits which
6.6.1.1 The process evaluation phase consists of the collec-
estimate the extent of random variation about the standard
tion of process output sample data either from performance
given (center line) due to common cause sources acting on the
qualification dose mapping data or historic processing output
process.
data (see Annex A3 and Annex A4) used to determine:
6.2.3 In some cases, alert/warning limits are also used
6.6.1.2 Current state of the process performance (statistical
which are similar to 63σ control limits but at lower coverage
model characterizing the process output).
levels(higherlevelsofsignificance, α),forexample, 62σ.(See
6.6.1.3 Appropriate control limits for the process level chart
Section 8 and Annex A2).
(x¯-chart) and the process variability chart (either R-chart or
s-chart).
6.3 There are two purposes for control chart analysis. One
where the analysis is used to derive a statistical model of the 6.6.2 Process monitoring is control chart trending of the
process output dose (termed control no standard given) and the process output.
E3239 − 21
6.6.2.1 Theprocessmonitoringphaseconsistsofcontinuous accrual characteristics; however, it may not be sufficient to
monitoring and control chart trending of a process for any achieve the user’s requirements for estimating process output
signal that a change in the state of control may have occurred. populationparameters(µand σ)fromsampleparameters(x¯ and
6.6.3 Process improvement is the restoration of the process s). For guidance on sample size (n) see Annex A8.
to the qualified state.
NOTE 7—Alarger number of samples (N) will improve the estimates of
6.6.3.1 The process improvement phase consists of investi-
the‘between/reproducibility’sampleofprocessvariation.Alargernumber
gation and when appropriate, correction of a special cause
of sample replicates (n) will improve the estimates of the ‘within/
repeatability’ sample of process variation.
signal observed in process output.
NOTE 8—Conditions of use in 7.3.1 refer to the common cause sources
NOTE 5—After the initial process evaluation phase, the process moni-
of variation in the routine process. If, for example, partially filled
toring phase starts. When process monitoring identifies the occurrence of
irradiation containers are intended to be used, stratified sampling will
a likely special cause event, the process improvement phase starts. The
include partially filled irradiation containers in the process from which
conclusion of the process improvement phase occurs when either a special
samples of the process are taken.
cause signal becomes an assignable cause and the assignable cause is
7.3.2 Dose Map Data Source:
mitigated returning the process to a state of control or the special cause
signal is determined to be a Type I error, for example, risk of signal 7.3.2.1 Performance qualification dose mapping is con-
associatedwiththeuserselected α.Theprocessmonitoringphaseresumes
ducted to determine the dose distribution throughout the
at the conclusion of the process improvement phase.
process load (see ISO/ASTM 52303). The dose map data then
represents the low doses, high doses, and intermediate doses
7. Radiation Process Specific Considerations
for the range of dose delivered to the processed product. The
7.1 General:
range of doses can be grouped, for example, high doses and
7.1.1 SPC in radiation processing is a means of demonstrat-
lowdoses,orintheextreme,toanumberofgroupsequaltothe
ing the process target dose is achieved within statistical control
number of dose map locations.
limits, that is, the process output (product dose or other
NOTE 9—Dose map grid locations and grouping of dose map grid
monitored measurement that is directly correlated to product
locations are defined by the user based on the user’s design of experiment
dose) is representative of a sample drawn from the a priori
and intended analysis of the variable(s) of interest. (See ISO/ASTM
statistical model (standard given, see 6.3 and Note 4).
52303.)
7.1.2 Process targeting is based on the relationship between
7.3.2.2 Different groups of dose map dose values have
the irradiator control parameter values and the realized dose of
different process level values and may have different process
the process output. This relationship is independent of Upper
variation values which are often a consideration when defining
Specification Limits (USL) and Lower Specification Limit
a group of dose map dose values. One example of grouping is
(LSL) of product(s) processed.
equivalency in the case of a minimum detectable difference,
NOTE 6—Aprocess that is in a state of control can fail to meet product MDD. (See ISO/ASTM 52303 for information on minimum
LSL or USL requirements if the process is not targeted appropriately. This
detectable difference.)
type of failure is a failure to appropriately target the process or accurately
7.3.2.3 One limitation of sampling the process output from
assess the process capability or reliability to meet product LSL and USL.
performance qualification dose map data is the number of
(See A3.2 and A3.3).
sample replicates (n). (See 6.4.2 and Note 7.)The process level
7.1.3 A process capability or reliability assessment corre-
associated with the minimum dose and maximum dose from
lates the process targeting used, the a priori statistical model of
N=1 and n=3 in radiation processing is usually a good estimate
processed product dose, and the LSL and USL product require-
of the level. However, the process variation estimate with N=1
ments. An acceptable process capability or reliability assess-
and n=3 may be inaccurate (1).
ment result provides evidence that product processed will meet
7.3.2.4 The potential inaccuracy of the estimate of the
product LSL and USL specifications for the process in a state of
repeatability of/within a sample (N=1) of process variation can
control at the level of confidence used in the capability or
be mitigated by using a c4 correction factor (see Annex A5),
reliability assessment.
using a t-distribution factor, or in some cases a pooled standard
7.2 Control—Standard Given:
deviation of a partition, for example, where the partition
7.2.1 In radiation processing, SPC charts are analyzed in the
represents the low dose and equivalent low doses based on a
context of control standard given; which determines whether
minimum detectable difference partition.
observed process output level and variation (see 6.5 through
7.3.2.5 Another limitation of sampling the process output
6.5.2) differ from a standard value by an amount greater than
from performance qualification dose map data from a single
should be attributed to random chance. The a priori statistical
sample (N=1) is that no information is collected for the
model of a radiation process output is based on sampling from:
‘between/reproducibility’ of a sample (between samples N ,
7.2.1.1 Performance qualification dose mapping, or
N , N , …) of process variation. (See Note 7).
2 3
7.2.1.2 Historic process output.
7.3.3 Historic Process Output Source:
7.3.3.1 Process output data in terms of product minimum
7.3 Sampling the Irradiation Process—Evaluation Phase
and maximum dose can be used to prepare accurate unbiased
7.3.1 Samples should be taken from the actual process
intended to be routinely used (conditions of use).
7.3.1.1 Historically, Performance Qualification dose map-
ping has used a single sample (N=1) with three replicates
The boldface numbers in parentheses refer to a list of references at the end of
(n=3). This may be sufficient to determine some product dose this standard.
E3239 − 21
estimates of the process level ‘standard given’and the process 8. Interpretation of Control Chart
variation ‘standard given.’ (See 6.3 and Note 4.)
8.1 General:
7.3.3.2 The assumption of a state of control is based on the
8.1.1 In radiation processing, the process level and process
processlevelandprocessvariationstandards,theircorrespond-
variation control chart pair are used to:
ing control limits, and process output values occurring within
8.1.1.1 Provide documented evidence of the state of statis-
the control limits.
tical control of the process; process output observations falling
within predicted statistical limits (control limits).
7.3.3.3 For this to be the case, all rational subgroups (see
7.4.1 and Note 10) should achieve the process target within the 8.1.1.2 Signal the user to likely special cause events; pro-
cess output observations falling outside of predicted statistical
statistical limits defined by the control limits.
limits (control limits, alert/warning limits).
7.3.3.4 A one-way analysis of variance (ANOVA) can be
used to prepare estimates of the process level and process
NOTE 13—Signaling can also be represented by additional interpreta-
tion rules, see 8.2 and Note 14.
variation standard and demonstrate the samples (N1, N2, N3,
…) were sampled during a state of control. (See Annex A4).
8.2 The interpretation of the control chart data represents
evaluation of the observational data in comparison to the
7.4 Plotting Control Chart Data—Monitoring Phase
control chart control limits (and potentially alert/warning
7.4.1 Radiation processing consists of sequences of indi-
limits) to detect special cause events.
vidual processing runs. The individual processing run repre-
8.2.1 Special cause events are signaled:
sents a rational subgroup, N. The runs (rational subgroups) are
8.2.2 If the subgroup observation exceeds the 3σ control
monitored in accordance with a routine process monitoring
limits.
dosimetry practice that specifies a monitoring location(s) and
8.2.3 If the subgroup observation exceeds the 2σ alert/
monitoring frequency resulting in ‘n’ replicate measurements
warning limits at a frequency greater than is predicted by the
for each subgroup.
confidence level of the alert/warning limits.
8.2.4 The user may choose to identify additional rules based
NOTE 10—The nomenclature for the rational subgroup and replicate
measurements corresponds to sampling nomenclature, that is, N samples
on their process for the interpretation of control chart data.The
(processing runs) of n replicate measurements or observations (monitored
following are examples of additional rules that some users may
irradiation containers). The sample size adjustment applied to statistical
choose to use when appropriate for their process:
computations is the number of measurements/observations, n. (See Annex
8.2.4.1 Twooutofthreeconsecutiveobservationsoutsideof
A5.)
defined alert/warning limit on the same side of the center line.
7.4.2 When the monitoring frequency represents a near
8.2.4.2 Four out of five consecutive observations fall out-
constant sampling rate of the subgroup, control chart control
side of a 1σ limit on the same side of the center line.
limits will have a constant value. When the monitoring
8.2.4.3 Nine consecutive observations on the same side of
frequency does not represent a near constant sampling rate of
the center line.
the subgroup, control chart limits must be updated for each
NOTE 14—The rules in 8.2.2 and 8.2.3 are generally applicable to all
subgroup. (See examples 4 and 6 in Chapter 3 of MNL7 for
processes. However, the application of additional rules such as those
identified in 8.2.4.1 through 8.2.4.3 may vary from user to user. Users
sample size adjustments for control limits).
whose process have either a large number of common cause source(s) of
variation or a few common cause sources of variation that are very large
NOTE 11—Generally, in radiation processing when routine monitoring
in magnitude are more likely to benefit from implementing additional
frequency does not represent a near constant sampling rate, this typically
rulesoralert/warninglimitsintermsofidentifyingandeliminatingspecial
occurs in an off-product reference point routine monitoring practice
cause sources of process variation and reducing common cause source
process, that is, an off-product reference location that precedes the run and
process variation. A thorough discussion of chart types and rules for
an off-product reference location that follows the run where run size is
interpretation can be found in Refs (2 and 3).
variable subgroup to subgroup.
8.3 Special cause events, when signaled by the identified
7.4.3 Data used in the SPC chart trending of the process
rules, are investigated to determine the root cause, and to
level may need to be normalized for processing control
identify corrective action and preventative action to return the
parameters such as:
process to a state of control.
7.4.3.1 Cycle timer setting and activity for gamma (see
8.3.1 If the assignable cause is a systematic change that
Annex A4).
cannot be corrected, that is, an engineering change to the
7.4.3.2 Conveyance speed, beam current, and scan for
irradiator altering the irradiator performance or an irreversible
electron beam.
change to the product, an update to the process level and
variation standard and their respective control limits is neces-
NOTE 12—The fundamental dose delivery control parameters are the
sary. This may necessitate a repeat in whole or in part of OQ
radiation field intensity and duration of exposure. Typically, a single
primary control parameter is used to control duration of exposure
or PQ.
regulating dose delivery magnitude, for example, cycle timer setting in
gamma and conveyor speed in electron beam and X-ray. However, 9. Keywords
secondary parameters representative of the radiation source must also be
9.1 common cause; control chart; control limit; process
considered in some instances, for example, a source activity value in
improvement; radiation processing; rational subgroup; special
gamma, and two values in electron beam and X-ray; beam current and
scan (analogous to source activity) and pulse rate in pulsed systems. cause; state of statistical control; statistical process control
E3239 − 21
ANNEXES
(Mandatory Information)
A1. PROCESS CONTROL, PROCESS TARGET, PROCESS CAPABILITY, PERFORMANCE, AND RELIABILITY
A1.1 Scope A1.4 Process Capability, Performance, and Reliability
A1.4.1 To determine whether an identified process target
A1.1.1 This annex describes the concepts, relationships and
provides a process added value that meets the process specifi-
application of Process Control, Process Target, Process
cation limit(s), a sample of the process output is used to
Capability, Process Performance, and Process Reliability with
prepare either a capability, performance or reliability estimate.
respect to radiation processing.
A1.4.2 Process capability and performance:
A1.2 Process Control
A1.4.2.1 Process capability (PC) or process performance
A1.2.1 A process is conducted to produce a product with a
(PP) is defined as the range of the process added value. The
desired or targeted process added value (product dose). The standard practice is to estimate the process capability or
process added value is defined in terms of a process specifi- process performance as a ‘6-sigma’ relationship.
cation or specification range; a dose range specification in
PC 5 6σ (A1.1)
ST
radiation processing consisting of an USL – upper specification
PP 5 6σ (A1.2)
LT
limit (maximum acceptable dose per ISO 11137-3) and an LSL
–lowerspecificationlimit(sterilizationdoseperISO11137-2).
The process capability or performance is related to the
The quantity of process added value (dose) is controlled by the
process specification as an index (Cp or Pp) calculated as the
irradiator process control parameter(s). The process output
ratio of a process specification range (USL-LSL) to the process
(processed product) to be defined as successful (conforming) added value output range (6σ).
has a process added value (dose) that falls within the process
USL2 LSL
Cp 5 (A1.3)
specification range (dose range specification). A common
6σ
LT
interest for all processes is whether the process has the ability
USL2 LSL
to produce an output that will conform to the product specifi-
Pp 5 (A1.4)
6σ
ST
cation. This ability of the irradiation process to produce a
conforming product dose can be estimated in several ways; a
A1.4.2.2 Process capability (Cp) and process performance
process capability index, a process performance index, or a
(Pp) are estimates under the assumption the process target and
process reliability estimate. (See A1.4.) These measures rep-
process output are centered within the process specification
resent the quality of the process output or the degree of process
limits. Similar estimates where the assumption that the process
control with respect to product dose specifications; to what
targetandprocessoutputarenotcenteredaredenotedwithCpk
degree the process can routinely produce a conforming product
and Ppk. When the process target and process output are not
dose. These types of evaluations allow the user to assess a
centered within the specification limit, the capability or perfor-
process output and the process targeting in the context of a
mance of the process with respect to the USL and LSL are not
product dose range specification.
the same; see A3.2 example. To account for this, the index (Cp
or Pp) is split into two separate single-sided estimates (see Eq
A1.3 Process Target
A1.5 and EqA1.6). These estimates consider the product dose
average against a single or double sided specification limit
A1.3.1 The irradiation process is targeted through specify-
estimating to what degree the process can routinely produce an
ing process control parameter values, that is, cycle timer
average product dose within the product dose specification
setting, process pathway, conveyance velocity, beam current,
limits.
...