SlideShare a Scribd company logo
Implementing
Decision Rule Made
Simple
https://consultglp.com
Wednesday 11 August 2021 at
8.00pm (Singapore Time)
10th Free Webinar on ….
Implementing
Decision Rule
Made Simple
by
Yeoh Guan Huah
https://consultglp.com 2
Introduction
• Laboratory analysis is part of conformity assessment
• Conforming with the product specification stated in a sales contract
• Attesting the test result in compliance with a regulatory limit or tolerance level
• General statement of conformity includes:
• Conformance vs Non-conformance
• Compliance vs Non-compliance
• Acceptance vs Rejection
• Pass vs Fail
• Positive vs negative
• Presence vs absence
• A decision based on the tests and measurements must be made before issuing such statement. A set
of decision rules is to be established.
3
Introduction
• There is always a risk to make a wrong decision
• Tests done on a laboratory sample; how representative is the
sample?
• Test results are always associated with measurement
uncertainty
• Decision may be “biased” due to someone’s possible vested
interest
• Are we able to control such risk in order to make an
informed decision?
4
Definitions of “Decision Rule”
• ISO/IEC 17025:2017 clause 3.7: “Rule that describes how
measurement uncertainty is accounted for when stating conformity
with a specified requirement”
• ISO Guide 98-4, JCGM 106: “Documented rule that describes how
measurement uncertainty will be accounted for with regard to
accepting or rejecting an item, given a specified requirement and the
result of a measurement”
• Eurachem Guide on Compliance Assessment: “A documented rule
that describes how measurement uncertainty will be allocated with
regard to accepting or rejecting a product according to its
specification and the result of a measurement”
5
Therefore, decision rule is a rule taking measurement
uncertainty into account when stating conformity
with a specification value or compliance with a
regulatory limit.
When making a claim of non-conformity, the risk should be as
low as possible for false rejection, i.e., we should have high
confidence to correctly reject the test result.
6
The
requirements in
ISO/IEC 17025
for decision
rules
7
Relevant clauses of ISO/IEC 17025:2017
• Clause 6.2.6 requires that the laboratory shall authorize
personnel to “analyse results, including statements of
conformity or opinions and interpretations”. That implies only
authorized signatories are allowed to handle decision making
in this case.
• The laboratory is now required to record the decision rule
adopted taking into account the risk; such a rule will have on
reporting false positive or negative results. (Ref: clause 7.1.3)
8
Relevant clauses 7.1.3 of ISO/IEC 17025:2017
• 7.1 Review of requests, tenders and contracts
• 7.1.3 When a customer requests a statement of conformity
to a specification or standard for the test or calibration (e.g.
pass/fail, in-tolerance/out-of-tolerance) the specification or
standard, and the decision rule shall be clearly defined. Unless
inherent in the requested specification or standard, the
decision rule selected shall be communicated to, and agreed
with, the customer.
9
Relevant clauses 7.8.6 of ISO/IEC 17025:2017
• 7.8.6 : Reporting statements of conformity
• 7.8.6.1 When a statement of conformity to a specification
or standard is provided, the laboratory shall document the
decision rule employed, taking into account the level of
risk (such as false accept and false reject and statistical
assumptions) associated with the decision rule employed,
and apply the decision rule.
• NOTE: Where the decision rule is prescribed by the
customer, regulations or normative documents, a further
consideration of the level of risk is not necessary.
10
Relevant clauses 7.8.6 of ISO/IEC 17025:2017
• 7.8.6 Reporting statements of conformity (contd.)
• 7.8.6.2 The laboratory shall report on the statement of conformity, such
that the statement clearly identifies:
• a) to which results the statement of conformity applies;
• b) which specifications, standards or parts thereof are met or not met;
• c) the decision rule applied (unless it is inherent in the requested
specification or standard).
The laboratory shall report on the statement of conformity, such that the
statement clearly identifies:
• To which results the statement of conformity applies;
• Which specifications, standards or parts thereof are met or not;
• The decision rule applied
11
To make a decision rule, we must have….
• A measurand (analyte) clearly specified (i.e. specification of the
object)
• A test result (normally assuming normal distribution of test results)
• Its associated measurement uncertainty
• A specification or regulation giving upper and/or lower limits
• A decision rule
This rule can decide to take or not to take measurement uncertainty
into account, but it includes the risk of making a wrong decision,
of which the involved parties are willing to take.
12
Aims of making a decision rule
• One should be clear on what type of risk he is taking
when making a proper decision rule:
• the supplier’s (laboratory’s) risk (Type I error, ) or
• the consumer’s (customer’s) risk (Type II error, )
• From the laboratory point of view, we should be looking
at Type I () False Positive error, as we want to minimize
and control our risk in issuing a statement of conformity
that might be proven wrong later.
• We usually set the error or risk that we can afford to take,
e.g.  = 0.05 (or 5%), bearing in mind that our test results
have associated measurement uncertainties.
13
Important assumption: The
uncertainty of measurement
is represented by a normal
(Gaussian) probability
distribution function, which is
consistent with the typical
measurement results (being
assumed the applicability of
the Central Limit Theorem)
What is Type I () error?
• A type I error is caused by wrongly rejecting a true (“Pass”) situation
(statistically speaking, a null hypothesis Ho)
• This happens when the test result is found to be close to the specification
value.
• In a hypothesis (significance) testing, we write:
• Null hypothesis Ho : test result = specification value (not literally but with probability)
• Alternate hypothesis H1 : test result > or < specification value (depending on the
upper or lower limits of specification)
• Generally, we want to control our error (risk) when making a statement to
reject Ho, i.e., saying the test result fails to meet with the given specification
limit
• We normally fix such error at  = 0.05, or 5% error (or 95% confidence)
14
Today’s practice on decision rules
• In current practice, we have been making a very simple decision
rule, often based on direct comparison of measurement value
with the specification or regulatory limits.
• The reason is either that
• these limits are deemed to have assumed including the measurement
uncertainty, which is not normally true,
• or, it has been assumed that the measurement value has zero
uncertainty!
• But, due to the presence of uncertainty in all measurements,
we are actually taking more than 50% risk to allow the actual
true value of the test parameter found outside the
specification when reporting the measurement value exactly
on the specification limit. 15
>50% Risk
above
specification
limit!
Specification-
upper limit
(Maximum)
A worked example
• Suppose a test method has a measurement uncertainty (+ 0.35ppm)
of a lead Pb content near its upper feed product specification limit
of 10ppm.
• Upon analysis, a result of 9.65+0.35ppm was found in the feed
sample given with 95% confidence.
• This result indicates that the true Pb content of the sample is
somewhere within the range: 9.30ppm to 10.00ppm with 95%
confidence
• It means that there is 2.5% chance that the true Pb content is
outside both ends of this range. 16
The meaning of Measurement Uncertainty
◼ For a measured value of 9.65ppm with an uncertainty of
0.35ppm, we shall write :
◼ 9.65 + 0.35 ppm
◼ The uncertainty range covers :
◼ 9.30 – 10.00 ppm with 95% confidence that the true value will
lie somewhere within:
◼ 9.30 9.65 10.00 ppm
◼ (True Value)
17
=0.025 uncertainty
=0.025 uncertainty
A worked example
• Remember that this uncertainty
(U) of +0.35 ppm comes from
• 2 x 0.175 ppm
• where 2 is the coverage factor
for normal distribution at about
95% confidence and,
• 0.175 ppm is the combined
standard uncertainty (u) from
the MU evaluation study.
• To be exact, the coverage factor
should be 1.96 for 95.00%
confidence for two-tailed
distribution situation (+).
By MS Excel functions:
-1.9600 =NORM.INV(0.025,0,1)
1.9600 =NORM.INV(0.975,0,1)
18
Use of Excel functions for normal distribution
Excel® entry
-1.960 =NORM.INV(0.025,0,1)
1.960 =NORM.INV(0.975,0,1)
0.975 =NORM.DIST(1.96,0,1,TRUE)
0.950 =NORM.DIST(1.645,0,1,TRUE)
“=NORM.INV(probability, mean, standard_deviation)”
“=NORM.DIST(x, mean, standard_deviation, cumulative)”
Proportion of area under the curve from - to
z- value. When mean = 0 and SD = 1, x = z.
19
z =
z =
A worked example
• Hence, if we were to decide on 9.65 ppm as our critical value to
make a “Pass” statement of conformity with confidence, we are
actually taking a 2.5% risk, instead of 5.0%, to be proven wrong if
the same sample were to be tested elsewhere by the same
procedure.
• But we are prepared to take a usual 5% risk in making such
statement.
• What is the critical value if we wish to make a decision rule to
take a 5% risk (or making a 5% error in rejecting the test result)
instead?
20
A worked example
• Remember the product
specification’s upper limit is 10.0
ppm Pb
• Our concern therefore is how
much uncertainty of our test
result is in the rejection zone
above the 10 ppm limit
• Hence, the coverage factor should
be chosen at the right-tailed
normal distribution instead of
two-tailed distribution.
• This coverage factor = 1.645,
instead of 1.960 under normal
probability distribution.
By MS Excel function:
-1.6449 =NORM.INV(0.05,0,1)
1.6449 =NORM.INV(0.95,0,1)
21
A worked example
• Therefore, the critical value for us to take a 5% risk to reject the result
against the upper specification limit is therefore:
• 10.00 – 1.645 x 0.175 or 10.00 – 0.288 or 9.71 ppm Pb
• That means our decision rule can be stated as below:
• “A pass statement shall be reported when the test result is equal or
smaller than 9.71 ppm Pb content in the feeds with 95% confidence. A
provisional or conditional pass can be reported when the test result is
above 9.71 ppm Pb but below 10.00 ppm Pb. A fail statement is to be
declared for test results above 10.00 ppm Pb. ”
22
In summary, graphical effect of the decision rules:
Specification:
10.0 ppm Max
9.65 ppm
9.71 ppm
U = 0.35 ppm U = 0.29 ppm
2.5% Risk 5.0% Risk
Acceptance Zone
Rejection Zone
23
Use of Guard Band g with acceptance limits
• ILAC-G8, Eurolab documents suggest the use of Guard
Band g which is the current trend of thinking.
• Guard band (g) is defined as: “interval between a
tolerance limit and a corresponding acceptance limit
where length g = |TL – AL|.”
Upper Tolerance Limit TL
Lower Tolerance Limit TL
Acceptance Limit AL
Acceptance Limit AL
g
g
Nominal
24
A graphical example of areas defined for a specification
interval with guard bands
+ U
25
Example 1A of areas defined for a specification interval with
guard bands inside the set limits – conservative or stringent
acceptance zone (Type I ) model
26
In this model, the risk taken is 2.5% at  = 0.025 only at upper or lower specification limits
Example 1B of areas defined for a specification interval with
guard bands inside the set limits – conservative or stringent
acceptance zone (Type I ) model
In this model, the risk taken is 5% at  = 0.05 only at upper or lower specification limits
27
Example 2 of areas defined for a specification interval with
guard bands outside the set limits – Relax rejection zone
(Type II  ) model
28
Example 3 of areas created by considering the upper and lower
tolerance limits given by the customer with the laboratory’s
own measurement uncertainty
29
Similar approaches for microbiological counts
• Microbiological analysis is based on counting the number of colonies in
terms of CFU (Colony Forming Units) after incubation with appropriate
agar media.
• Discrete counting does not follow normal probability distribution.
• A logarithmic transformation of the counts is to be made in order to
assume a normal distribution for continuous data
• For example, for a count of 100 CFU, Log10(100) = 2.000
• In MU study of microbiological counts, the combined standard
uncertainty is expressed in the form of logarithmic relative standard
deviation, Log10, rel
• Same principles of factor 1.645 are applied like those in chemical analysis.
30
The binary
approach
• Applicable for qualitative “Pass/Fail”, “Present/Absent”,
etc. testing
• When you take two samples for testing and if both pass,
you report “Pass”
• If one passes and another fails, then you need to take
two more samples for analysis
• If both pass, it is determined as a pass
• If either or both fail, it fails.
• This approach has implicitly taken uncertainty into
account and is therefore the defined “decision rule”
31
In reporting ….
• The laboratory shall :
• apply the decision rule
• Ensure its report to contain either the decision rule used or a reference to a standard
or agreed decision rule
• Listing the uncertainty of measurement result is optional, depending on customer’s
liking
• Making a disclaimer statement, e.g. by ILAC Guide 8.03 “The statement(s) of
compliance with specification (or requirement) is based on a 95% coverage probability
for the expanded uncertainty of the measurement results on which the decision of
compliance is based”.
• There must be enough detail so that if another lab is tasked with taking the same
measurements and using the same decision rule, that they will come to the same
conclusion about a “pass” or “fail”, in order to avoid any legal implication.
32
Decision rule documentation, the lab shall ….
33
document the decision rules
employed taking into account the
level of risk associated with the
decision rule employed (i.e. false
accept and false reject) and statistical
assumptions associated with the
decision rule employed;
prepare a list of critical values for the
test parameters with Type I error α =
0.05, i.e. 95% confidence, normally
after some statistical hypothesis
analysis
take time to clearly document what
has been agreed with the clients and
ensure that lab procedures allow for
the correct treatment of the readings
and uncertainties
might need to change test procedures
to cater for the decision rule that will
be applied such as a review of
measurement uncertainty level. It has
to be such that anybody in the lab
will know what is required and what
the limits are.
Parting words ….
• Decision rule construction requires four components:
• the measurement result,
• its measurement uncertainty,
• the specification limit or limits, and
• the acceptable level of the probability of making
each type of wrong decision.
• Choice of a decision rule is a business consideration
that takes into account the cost of rejecting an in-
specification product, the cost of accepting an out-of-
specification product, the uncertainty associated with
the measurement, the cost of making the measurement,
and client’s relationship for future business.
34
i.e., the lab must agree with the customer on its decision rule and
communicate clearly about the level of risk associated with this
decision rule.
Conclusions
• Issuance of a certificate of conformity or a statement of
compliance allows the laboratory to carry a fair bit of
responsibility.
• The laboratory must be confident on its own measurement
uncertainty estimation.
• If a statement of conformity is required by the clients, a lot
more detail is required to be discussed before carrying out
the laboratory analysis.
• ISO/IEC 17025 requires the Lab to be very clear about the
criteria used and must communicate with and be agreed by
the customer, defined in the test report, certificate of
conformance or statement of compliance to avoid future
disputes.
35
Useful References
• ILAC-G8:09/2019 : Guidelines on Decision Rules and Statements
of Conformity
• EuroLab Technical Report No. 01/2017 : Decision rules applied to
conformity assessment
• UKAS LAB 48 Edition 3 (June 2020) : Decision rules and
statements of conformity
• OIML Guide G 19 (2017) : The role of measurement uncertainty in
conformity assessment decisions in legal metrology
• JCGM 106:2012 : Evaluation of measurement data – The role of
measurement uncertainty in conformity assessment
36

More Related Content

What's hot

ISO 17025
ISO 17025 ISO 17025
ISO 17025
Akma Ija
 
Estimation of Measurement Uncertainty in Labs: a requirement for ISO 17025 Ac...
Estimation of Measurement Uncertainty in Labs: a requirement for ISO 17025 Ac...Estimation of Measurement Uncertainty in Labs: a requirement for ISO 17025 Ac...
Estimation of Measurement Uncertainty in Labs: a requirement for ISO 17025 Ac...
PECB
 
ISO/IEC 17025 2017 Check List
ISO/IEC 17025 2017 Check ListISO/IEC 17025 2017 Check List
ISO/IEC 17025 2017 Check List
Mohamed Salama Elimam
 
Method verification
Method verificationMethod verification
Method verification
Robert Farnham
 
Iso 17025 standard
Iso 17025 standardIso 17025 standard
Iso 17025 standard
Christian Silmaro
 
ISO 17025.pptx
ISO 17025.pptxISO 17025.pptx
ISO 17025.pptx
MohsinALi566341
 
ISO/IEC 17025:2017 Required documentation
ISO/IEC 17025:2017 Required documentationISO/IEC 17025:2017 Required documentation
ISO/IEC 17025:2017 Required documentation
Mohamed Salama Elimam
 
What is the decision rule all about?
What is the decision rule all about?What is the decision rule all about?
What is the decision rule all about?
Maryam Sourani
 
Testing Laboratory Practices and ISO 17025
Testing Laboratory Practices and ISO 17025Testing Laboratory Practices and ISO 17025
Testing Laboratory Practices and ISO 17025
quality_management
 
Measurement Uncertainty
Measurement UncertaintyMeasurement Uncertainty
Measurement Uncertainty
GLP Consulting Singapore
 
ISO/IEC 17025:2017 Training ppt
ISO/IEC 17025:2017 Training pptISO/IEC 17025:2017 Training ppt
ISO/IEC 17025:2017 Training ppt
Global Manager Group
 
ISO 17025 Accreditation Detail Review
ISO 17025 Accreditation Detail Review ISO 17025 Accreditation Detail Review
ISO 17025 Accreditation Detail Review
Abdul Rahman
 
Measurement Uncertainty
Measurement UncertaintyMeasurement Uncertainty
Measurement Uncertainty
DenizErtekin2
 
Introduction to measurement uncertainty
Introduction to measurement uncertaintyIntroduction to measurement uncertainty
Introduction to measurement uncertainty
Vinoth Kumar K
 
Quality Objectives
Quality ObjectivesQuality Objectives
F iso-17025
F iso-17025F iso-17025
F iso-17025
Ahmedjet
 
Analytical method validation
Analytical method validation Analytical method validation
Analytical method validation
Dr. Amit Gangwal Jain (MPharm., PhD.)
 
What Documentation Required for ISO 17025:2017 Accreditation?
What Documentation Required for ISO 17025:2017 Accreditation?What Documentation Required for ISO 17025:2017 Accreditation?
What Documentation Required for ISO 17025:2017 Accreditation?
Global Manager Group
 
Guide25 vs ISO/IEC17025
Guide25 vs ISO/IEC17025Guide25 vs ISO/IEC17025
Guide25 vs ISO/IEC17025
SEREE NET
 
Key factors and main change over in iso 17025 2017
Key factors and main change over in  iso 17025 2017Key factors and main change over in  iso 17025 2017
Key factors and main change over in iso 17025 2017
Dr.Lenin raja
 

What's hot (20)

ISO 17025
ISO 17025 ISO 17025
ISO 17025
 
Estimation of Measurement Uncertainty in Labs: a requirement for ISO 17025 Ac...
Estimation of Measurement Uncertainty in Labs: a requirement for ISO 17025 Ac...Estimation of Measurement Uncertainty in Labs: a requirement for ISO 17025 Ac...
Estimation of Measurement Uncertainty in Labs: a requirement for ISO 17025 Ac...
 
ISO/IEC 17025 2017 Check List
ISO/IEC 17025 2017 Check ListISO/IEC 17025 2017 Check List
ISO/IEC 17025 2017 Check List
 
Method verification
Method verificationMethod verification
Method verification
 
Iso 17025 standard
Iso 17025 standardIso 17025 standard
Iso 17025 standard
 
ISO 17025.pptx
ISO 17025.pptxISO 17025.pptx
ISO 17025.pptx
 
ISO/IEC 17025:2017 Required documentation
ISO/IEC 17025:2017 Required documentationISO/IEC 17025:2017 Required documentation
ISO/IEC 17025:2017 Required documentation
 
What is the decision rule all about?
What is the decision rule all about?What is the decision rule all about?
What is the decision rule all about?
 
Testing Laboratory Practices and ISO 17025
Testing Laboratory Practices and ISO 17025Testing Laboratory Practices and ISO 17025
Testing Laboratory Practices and ISO 17025
 
Measurement Uncertainty
Measurement UncertaintyMeasurement Uncertainty
Measurement Uncertainty
 
ISO/IEC 17025:2017 Training ppt
ISO/IEC 17025:2017 Training pptISO/IEC 17025:2017 Training ppt
ISO/IEC 17025:2017 Training ppt
 
ISO 17025 Accreditation Detail Review
ISO 17025 Accreditation Detail Review ISO 17025 Accreditation Detail Review
ISO 17025 Accreditation Detail Review
 
Measurement Uncertainty
Measurement UncertaintyMeasurement Uncertainty
Measurement Uncertainty
 
Introduction to measurement uncertainty
Introduction to measurement uncertaintyIntroduction to measurement uncertainty
Introduction to measurement uncertainty
 
Quality Objectives
Quality ObjectivesQuality Objectives
Quality Objectives
 
F iso-17025
F iso-17025F iso-17025
F iso-17025
 
Analytical method validation
Analytical method validation Analytical method validation
Analytical method validation
 
What Documentation Required for ISO 17025:2017 Accreditation?
What Documentation Required for ISO 17025:2017 Accreditation?What Documentation Required for ISO 17025:2017 Accreditation?
What Documentation Required for ISO 17025:2017 Accreditation?
 
Guide25 vs ISO/IEC17025
Guide25 vs ISO/IEC17025Guide25 vs ISO/IEC17025
Guide25 vs ISO/IEC17025
 
Key factors and main change over in iso 17025 2017
Key factors and main change over in  iso 17025 2017Key factors and main change over in  iso 17025 2017
Key factors and main change over in iso 17025 2017
 

Similar to Implementing decision rule made simple

5.22.18_Statements-Conformity (1).pdf
5.22.18_Statements-Conformity (1).pdf5.22.18_Statements-Conformity (1).pdf
5.22.18_Statements-Conformity (1).pdf
AnujPandey899047
 
Basic QC Statistics - Improving Laboratory Performance Through Quality Contro...
Basic QC Statistics - Improving Laboratory Performance Through Quality Contro...Basic QC Statistics - Improving Laboratory Performance Through Quality Contro...
Basic QC Statistics - Improving Laboratory Performance Through Quality Contro...
Randox
 
How to Measure Uncertainty
How to Measure UncertaintyHow to Measure Uncertainty
How to Measure Uncertainty
Randox
 
Uncertainty (ASCP)
Uncertainty (ASCP)Uncertainty (ASCP)
Uncertainty (ASCP)
Zeeshan Siddique
 
Qpa -inspection and test sampling plan
Qpa -inspection and test sampling planQpa -inspection and test sampling plan
Qpa -inspection and test sampling plan
SalmanLatif14
 
Validation of qualitative lab test methods
Validation of qualitative lab test  methods Validation of qualitative lab test  methods
Validation of qualitative lab test methods
Mostafa Mahmoud
 
Acceptance sampling
Acceptance samplingAcceptance sampling
Acceptance sampling
SIBENDU SURAJEET JENA
 
Method validation
Method validationMethod validation
Method validation
DrHinal
 
(NASA_P.Reese_J.Harben)_Implementing_Strategies_for_Risk_Mitigation_(NCSLI-2011)
(NASA_P.Reese_J.Harben)_Implementing_Strategies_for_Risk_Mitigation_(NCSLI-2011)(NASA_P.Reese_J.Harben)_Implementing_Strategies_for_Risk_Mitigation_(NCSLI-2011)
(NASA_P.Reese_J.Harben)_Implementing_Strategies_for_Risk_Mitigation_(NCSLI-2011)
Paul Reese
 
Quality Assurance of Laboratory Test Results based on ISO/IEC 17025
Quality Assurance of Laboratory Test Results based on ISO/IEC 17025Quality Assurance of Laboratory Test Results based on ISO/IEC 17025
Quality Assurance of Laboratory Test Results based on ISO/IEC 17025
PECB
 
Validity andreliability
Validity andreliabilityValidity andreliability
Validity andreliability
nuwan udugampala
 
Designing an appropriate qc design procedure for your lab 5 mar15
Designing an appropriate qc design procedure for your lab 5 mar15Designing an appropriate qc design procedure for your lab 5 mar15
Designing an appropriate qc design procedure for your lab 5 mar15
Randox
 
Designing an appropriate QC procedure for your laboratory
Designing an appropriate QC procedure for your laboratoryDesigning an appropriate QC procedure for your laboratory
Designing an appropriate QC procedure for your laboratory
Randox
 
Data analysis ( Bio-statistic )
Data analysis ( Bio-statistic )Data analysis ( Bio-statistic )
Data analysis ( Bio-statistic )
Amany Elsayed
 
Section 7 Analyzing our Marketing Test, Survey Results .docx
Section 7 Analyzing our Marketing Test, Survey Results .docxSection 7 Analyzing our Marketing Test, Survey Results .docx
Section 7 Analyzing our Marketing Test, Survey Results .docx
kenjordan97598
 
Presentation-on-Handling-OOS-Investigations-Regulatory-Expectations-Dipesh-Sh...
Presentation-on-Handling-OOS-Investigations-Regulatory-Expectations-Dipesh-Sh...Presentation-on-Handling-OOS-Investigations-Regulatory-Expectations-Dipesh-Sh...
Presentation-on-Handling-OOS-Investigations-Regulatory-Expectations-Dipesh-Sh...
Tom Aspinall
 
FDA OOS investigation India (Out of Specifications).pdf
FDA OOS investigation India (Out of Specifications).pdfFDA OOS investigation India (Out of Specifications).pdf
FDA OOS investigation India (Out of Specifications).pdf
midohamada2
 
Measurement risk and the impact on your processes
Measurement risk and the impact on your processes  Measurement risk and the impact on your processes
Measurement risk and the impact on your processes
Transcat
 
App. Of Stat. Tools
App. Of Stat. ToolsApp. Of Stat. Tools
App. Of Stat. Tools
Denny Thayil
 
Metrology & The Consequences of Bad Measurement Decisions
Metrology & The Consequences of Bad Measurement DecisionsMetrology & The Consequences of Bad Measurement Decisions
Metrology & The Consequences of Bad Measurement Decisions
Rick Hogan
 

Similar to Implementing decision rule made simple (20)

5.22.18_Statements-Conformity (1).pdf
5.22.18_Statements-Conformity (1).pdf5.22.18_Statements-Conformity (1).pdf
5.22.18_Statements-Conformity (1).pdf
 
Basic QC Statistics - Improving Laboratory Performance Through Quality Contro...
Basic QC Statistics - Improving Laboratory Performance Through Quality Contro...Basic QC Statistics - Improving Laboratory Performance Through Quality Contro...
Basic QC Statistics - Improving Laboratory Performance Through Quality Contro...
 
How to Measure Uncertainty
How to Measure UncertaintyHow to Measure Uncertainty
How to Measure Uncertainty
 
Uncertainty (ASCP)
Uncertainty (ASCP)Uncertainty (ASCP)
Uncertainty (ASCP)
 
Qpa -inspection and test sampling plan
Qpa -inspection and test sampling planQpa -inspection and test sampling plan
Qpa -inspection and test sampling plan
 
Validation of qualitative lab test methods
Validation of qualitative lab test  methods Validation of qualitative lab test  methods
Validation of qualitative lab test methods
 
Acceptance sampling
Acceptance samplingAcceptance sampling
Acceptance sampling
 
Method validation
Method validationMethod validation
Method validation
 
(NASA_P.Reese_J.Harben)_Implementing_Strategies_for_Risk_Mitigation_(NCSLI-2011)
(NASA_P.Reese_J.Harben)_Implementing_Strategies_for_Risk_Mitigation_(NCSLI-2011)(NASA_P.Reese_J.Harben)_Implementing_Strategies_for_Risk_Mitigation_(NCSLI-2011)
(NASA_P.Reese_J.Harben)_Implementing_Strategies_for_Risk_Mitigation_(NCSLI-2011)
 
Quality Assurance of Laboratory Test Results based on ISO/IEC 17025
Quality Assurance of Laboratory Test Results based on ISO/IEC 17025Quality Assurance of Laboratory Test Results based on ISO/IEC 17025
Quality Assurance of Laboratory Test Results based on ISO/IEC 17025
 
Validity andreliability
Validity andreliabilityValidity andreliability
Validity andreliability
 
Designing an appropriate qc design procedure for your lab 5 mar15
Designing an appropriate qc design procedure for your lab 5 mar15Designing an appropriate qc design procedure for your lab 5 mar15
Designing an appropriate qc design procedure for your lab 5 mar15
 
Designing an appropriate QC procedure for your laboratory
Designing an appropriate QC procedure for your laboratoryDesigning an appropriate QC procedure for your laboratory
Designing an appropriate QC procedure for your laboratory
 
Data analysis ( Bio-statistic )
Data analysis ( Bio-statistic )Data analysis ( Bio-statistic )
Data analysis ( Bio-statistic )
 
Section 7 Analyzing our Marketing Test, Survey Results .docx
Section 7 Analyzing our Marketing Test, Survey Results .docxSection 7 Analyzing our Marketing Test, Survey Results .docx
Section 7 Analyzing our Marketing Test, Survey Results .docx
 
Presentation-on-Handling-OOS-Investigations-Regulatory-Expectations-Dipesh-Sh...
Presentation-on-Handling-OOS-Investigations-Regulatory-Expectations-Dipesh-Sh...Presentation-on-Handling-OOS-Investigations-Regulatory-Expectations-Dipesh-Sh...
Presentation-on-Handling-OOS-Investigations-Regulatory-Expectations-Dipesh-Sh...
 
FDA OOS investigation India (Out of Specifications).pdf
FDA OOS investigation India (Out of Specifications).pdfFDA OOS investigation India (Out of Specifications).pdf
FDA OOS investigation India (Out of Specifications).pdf
 
Measurement risk and the impact on your processes
Measurement risk and the impact on your processes  Measurement risk and the impact on your processes
Measurement risk and the impact on your processes
 
App. Of Stat. Tools
App. Of Stat. ToolsApp. Of Stat. Tools
App. Of Stat. Tools
 
Metrology & The Consequences of Bad Measurement Decisions
Metrology & The Consequences of Bad Measurement DecisionsMetrology & The Consequences of Bad Measurement Decisions
Metrology & The Consequences of Bad Measurement Decisions
 

More from GH Yeoh

Common mistakes in measurement uncertainty calculations
Common mistakes in measurement uncertainty calculationsCommon mistakes in measurement uncertainty calculations
Common mistakes in measurement uncertainty calculations
GH Yeoh
 
Worked examples of sampling uncertainty evaluation
Worked examples of sampling uncertainty evaluationWorked examples of sampling uncertainty evaluation
Worked examples of sampling uncertainty evaluation
GH Yeoh
 
Biological variation as an uncertainty component
Biological variation as an uncertainty componentBiological variation as an uncertainty component
Biological variation as an uncertainty component
GH Yeoh
 
Basic calculations of measurement uncertainty in medical testing
Basic calculations of measurement uncertainty in medical testingBasic calculations of measurement uncertainty in medical testing
Basic calculations of measurement uncertainty in medical testing
GH Yeoh
 
A note and graphical illustration of type II error
A note and graphical illustration of type II errorA note and graphical illustration of type II error
A note and graphical illustration of type II error
GH Yeoh
 
Concept of Limit of Detection (LOD)
Concept of Limit of Detection (LOD)Concept of Limit of Detection (LOD)
Concept of Limit of Detection (LOD)
GH Yeoh
 
Presenting our training courses
Presenting our training coursesPresenting our training courses
Presenting our training courses
GH Yeoh
 

More from GH Yeoh (7)

Common mistakes in measurement uncertainty calculations
Common mistakes in measurement uncertainty calculationsCommon mistakes in measurement uncertainty calculations
Common mistakes in measurement uncertainty calculations
 
Worked examples of sampling uncertainty evaluation
Worked examples of sampling uncertainty evaluationWorked examples of sampling uncertainty evaluation
Worked examples of sampling uncertainty evaluation
 
Biological variation as an uncertainty component
Biological variation as an uncertainty componentBiological variation as an uncertainty component
Biological variation as an uncertainty component
 
Basic calculations of measurement uncertainty in medical testing
Basic calculations of measurement uncertainty in medical testingBasic calculations of measurement uncertainty in medical testing
Basic calculations of measurement uncertainty in medical testing
 
A note and graphical illustration of type II error
A note and graphical illustration of type II errorA note and graphical illustration of type II error
A note and graphical illustration of type II error
 
Concept of Limit of Detection (LOD)
Concept of Limit of Detection (LOD)Concept of Limit of Detection (LOD)
Concept of Limit of Detection (LOD)
 
Presenting our training courses
Presenting our training coursesPresenting our training courses
Presenting our training courses
 

Recently uploaded

如何办理(uvic毕业证书)维多利亚大学毕业证本科学位证书原版一模一样
如何办理(uvic毕业证书)维多利亚大学毕业证本科学位证书原版一模一样如何办理(uvic毕业证书)维多利亚大学毕业证本科学位证书原版一模一样
如何办理(uvic毕业证书)维多利亚大学毕业证本科学位证书原版一模一样
yqqaatn0
 
SAR of Medicinal Chemistry 1st by dk.pdf
SAR of Medicinal Chemistry 1st by dk.pdfSAR of Medicinal Chemistry 1st by dk.pdf
SAR of Medicinal Chemistry 1st by dk.pdf
KrushnaDarade1
 
Micronuclei test.M.sc.zoology.fisheries.
Micronuclei test.M.sc.zoology.fisheries.Micronuclei test.M.sc.zoology.fisheries.
Micronuclei test.M.sc.zoology.fisheries.
Aditi Bajpai
 
BREEDING METHODS FOR DISEASE RESISTANCE.pptx
BREEDING METHODS FOR DISEASE RESISTANCE.pptxBREEDING METHODS FOR DISEASE RESISTANCE.pptx
BREEDING METHODS FOR DISEASE RESISTANCE.pptx
RASHMI M G
 
Comparing Evolved Extractive Text Summary Scores of Bidirectional Encoder Rep...
Comparing Evolved Extractive Text Summary Scores of Bidirectional Encoder Rep...Comparing Evolved Extractive Text Summary Scores of Bidirectional Encoder Rep...
Comparing Evolved Extractive Text Summary Scores of Bidirectional Encoder Rep...
University of Maribor
 
3D Hybrid PIC simulation of the plasma expansion (ISSS-14)
3D Hybrid PIC simulation of the plasma expansion (ISSS-14)3D Hybrid PIC simulation of the plasma expansion (ISSS-14)
3D Hybrid PIC simulation of the plasma expansion (ISSS-14)
David Osipyan
 
What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.
moosaasad1975
 
Eukaryotic Transcription Presentation.pptx
Eukaryotic Transcription Presentation.pptxEukaryotic Transcription Presentation.pptx
Eukaryotic Transcription Presentation.pptx
RitabrataSarkar3
 
Phenomics assisted breeding in crop improvement
Phenomics assisted breeding in crop improvementPhenomics assisted breeding in crop improvement
Phenomics assisted breeding in crop improvement
IshaGoswami9
 
NuGOweek 2024 Ghent programme overview flyer
NuGOweek 2024 Ghent programme overview flyerNuGOweek 2024 Ghent programme overview flyer
NuGOweek 2024 Ghent programme overview flyer
pablovgd
 
Sharlene Leurig - Enabling Onsite Water Use with Net Zero Water
Sharlene Leurig - Enabling Onsite Water Use with Net Zero WaterSharlene Leurig - Enabling Onsite Water Use with Net Zero Water
Sharlene Leurig - Enabling Onsite Water Use with Net Zero Water
Texas Alliance of Groundwater Districts
 
Shallowest Oil Discovery of Turkiye.pptx
Shallowest Oil Discovery of Turkiye.pptxShallowest Oil Discovery of Turkiye.pptx
Shallowest Oil Discovery of Turkiye.pptx
Gokturk Mehmet Dilci
 
Nucleic Acid-its structural and functional complexity.
Nucleic Acid-its structural and functional complexity.Nucleic Acid-its structural and functional complexity.
Nucleic Acid-its structural and functional complexity.
Nistarini College, Purulia (W.B) India
 
Nucleophilic Addition of carbonyl compounds.pptx
Nucleophilic Addition of carbonyl  compounds.pptxNucleophilic Addition of carbonyl  compounds.pptx
Nucleophilic Addition of carbonyl compounds.pptx
SSR02
 
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptxThe use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
MAGOTI ERNEST
 
molar-distalization in orthodontics-seminar.pptx
molar-distalization in orthodontics-seminar.pptxmolar-distalization in orthodontics-seminar.pptx
molar-distalization in orthodontics-seminar.pptx
Anagha Prasad
 
Equivariant neural networks and representation theory
Equivariant neural networks and representation theoryEquivariant neural networks and representation theory
Equivariant neural networks and representation theory
Daniel Tubbenhauer
 
ESR spectroscopy in liquid food and beverages.pptx
ESR spectroscopy in liquid food and beverages.pptxESR spectroscopy in liquid food and beverages.pptx
ESR spectroscopy in liquid food and beverages.pptx
PRIYANKA PATEL
 
Unlocking the mysteries of reproduction: Exploring fecundity and gonadosomati...
Unlocking the mysteries of reproduction: Exploring fecundity and gonadosomati...Unlocking the mysteries of reproduction: Exploring fecundity and gonadosomati...
Unlocking the mysteries of reproduction: Exploring fecundity and gonadosomati...
AbdullaAlAsif1
 
Medical Orthopedic PowerPoint Templates.pptx
Medical Orthopedic PowerPoint Templates.pptxMedical Orthopedic PowerPoint Templates.pptx
Medical Orthopedic PowerPoint Templates.pptx
terusbelajar5
 

Recently uploaded (20)

如何办理(uvic毕业证书)维多利亚大学毕业证本科学位证书原版一模一样
如何办理(uvic毕业证书)维多利亚大学毕业证本科学位证书原版一模一样如何办理(uvic毕业证书)维多利亚大学毕业证本科学位证书原版一模一样
如何办理(uvic毕业证书)维多利亚大学毕业证本科学位证书原版一模一样
 
SAR of Medicinal Chemistry 1st by dk.pdf
SAR of Medicinal Chemistry 1st by dk.pdfSAR of Medicinal Chemistry 1st by dk.pdf
SAR of Medicinal Chemistry 1st by dk.pdf
 
Micronuclei test.M.sc.zoology.fisheries.
Micronuclei test.M.sc.zoology.fisheries.Micronuclei test.M.sc.zoology.fisheries.
Micronuclei test.M.sc.zoology.fisheries.
 
BREEDING METHODS FOR DISEASE RESISTANCE.pptx
BREEDING METHODS FOR DISEASE RESISTANCE.pptxBREEDING METHODS FOR DISEASE RESISTANCE.pptx
BREEDING METHODS FOR DISEASE RESISTANCE.pptx
 
Comparing Evolved Extractive Text Summary Scores of Bidirectional Encoder Rep...
Comparing Evolved Extractive Text Summary Scores of Bidirectional Encoder Rep...Comparing Evolved Extractive Text Summary Scores of Bidirectional Encoder Rep...
Comparing Evolved Extractive Text Summary Scores of Bidirectional Encoder Rep...
 
3D Hybrid PIC simulation of the plasma expansion (ISSS-14)
3D Hybrid PIC simulation of the plasma expansion (ISSS-14)3D Hybrid PIC simulation of the plasma expansion (ISSS-14)
3D Hybrid PIC simulation of the plasma expansion (ISSS-14)
 
What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.
 
Eukaryotic Transcription Presentation.pptx
Eukaryotic Transcription Presentation.pptxEukaryotic Transcription Presentation.pptx
Eukaryotic Transcription Presentation.pptx
 
Phenomics assisted breeding in crop improvement
Phenomics assisted breeding in crop improvementPhenomics assisted breeding in crop improvement
Phenomics assisted breeding in crop improvement
 
NuGOweek 2024 Ghent programme overview flyer
NuGOweek 2024 Ghent programme overview flyerNuGOweek 2024 Ghent programme overview flyer
NuGOweek 2024 Ghent programme overview flyer
 
Sharlene Leurig - Enabling Onsite Water Use with Net Zero Water
Sharlene Leurig - Enabling Onsite Water Use with Net Zero WaterSharlene Leurig - Enabling Onsite Water Use with Net Zero Water
Sharlene Leurig - Enabling Onsite Water Use with Net Zero Water
 
Shallowest Oil Discovery of Turkiye.pptx
Shallowest Oil Discovery of Turkiye.pptxShallowest Oil Discovery of Turkiye.pptx
Shallowest Oil Discovery of Turkiye.pptx
 
Nucleic Acid-its structural and functional complexity.
Nucleic Acid-its structural and functional complexity.Nucleic Acid-its structural and functional complexity.
Nucleic Acid-its structural and functional complexity.
 
Nucleophilic Addition of carbonyl compounds.pptx
Nucleophilic Addition of carbonyl  compounds.pptxNucleophilic Addition of carbonyl  compounds.pptx
Nucleophilic Addition of carbonyl compounds.pptx
 
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptxThe use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
 
molar-distalization in orthodontics-seminar.pptx
molar-distalization in orthodontics-seminar.pptxmolar-distalization in orthodontics-seminar.pptx
molar-distalization in orthodontics-seminar.pptx
 
Equivariant neural networks and representation theory
Equivariant neural networks and representation theoryEquivariant neural networks and representation theory
Equivariant neural networks and representation theory
 
ESR spectroscopy in liquid food and beverages.pptx
ESR spectroscopy in liquid food and beverages.pptxESR spectroscopy in liquid food and beverages.pptx
ESR spectroscopy in liquid food and beverages.pptx
 
Unlocking the mysteries of reproduction: Exploring fecundity and gonadosomati...
Unlocking the mysteries of reproduction: Exploring fecundity and gonadosomati...Unlocking the mysteries of reproduction: Exploring fecundity and gonadosomati...
Unlocking the mysteries of reproduction: Exploring fecundity and gonadosomati...
 
Medical Orthopedic PowerPoint Templates.pptx
Medical Orthopedic PowerPoint Templates.pptxMedical Orthopedic PowerPoint Templates.pptx
Medical Orthopedic PowerPoint Templates.pptx
 

Implementing decision rule made simple

  • 1. Implementing Decision Rule Made Simple https://consultglp.com Wednesday 11 August 2021 at 8.00pm (Singapore Time) 10th Free Webinar on ….
  • 2. Implementing Decision Rule Made Simple by Yeoh Guan Huah https://consultglp.com 2
  • 3. Introduction • Laboratory analysis is part of conformity assessment • Conforming with the product specification stated in a sales contract • Attesting the test result in compliance with a regulatory limit or tolerance level • General statement of conformity includes: • Conformance vs Non-conformance • Compliance vs Non-compliance • Acceptance vs Rejection • Pass vs Fail • Positive vs negative • Presence vs absence • A decision based on the tests and measurements must be made before issuing such statement. A set of decision rules is to be established. 3
  • 4. Introduction • There is always a risk to make a wrong decision • Tests done on a laboratory sample; how representative is the sample? • Test results are always associated with measurement uncertainty • Decision may be “biased” due to someone’s possible vested interest • Are we able to control such risk in order to make an informed decision? 4
  • 5. Definitions of “Decision Rule” • ISO/IEC 17025:2017 clause 3.7: “Rule that describes how measurement uncertainty is accounted for when stating conformity with a specified requirement” • ISO Guide 98-4, JCGM 106: “Documented rule that describes how measurement uncertainty will be accounted for with regard to accepting or rejecting an item, given a specified requirement and the result of a measurement” • Eurachem Guide on Compliance Assessment: “A documented rule that describes how measurement uncertainty will be allocated with regard to accepting or rejecting a product according to its specification and the result of a measurement” 5
  • 6. Therefore, decision rule is a rule taking measurement uncertainty into account when stating conformity with a specification value or compliance with a regulatory limit. When making a claim of non-conformity, the risk should be as low as possible for false rejection, i.e., we should have high confidence to correctly reject the test result. 6
  • 8. Relevant clauses of ISO/IEC 17025:2017 • Clause 6.2.6 requires that the laboratory shall authorize personnel to “analyse results, including statements of conformity or opinions and interpretations”. That implies only authorized signatories are allowed to handle decision making in this case. • The laboratory is now required to record the decision rule adopted taking into account the risk; such a rule will have on reporting false positive or negative results. (Ref: clause 7.1.3) 8
  • 9. Relevant clauses 7.1.3 of ISO/IEC 17025:2017 • 7.1 Review of requests, tenders and contracts • 7.1.3 When a customer requests a statement of conformity to a specification or standard for the test or calibration (e.g. pass/fail, in-tolerance/out-of-tolerance) the specification or standard, and the decision rule shall be clearly defined. Unless inherent in the requested specification or standard, the decision rule selected shall be communicated to, and agreed with, the customer. 9
  • 10. Relevant clauses 7.8.6 of ISO/IEC 17025:2017 • 7.8.6 : Reporting statements of conformity • 7.8.6.1 When a statement of conformity to a specification or standard is provided, the laboratory shall document the decision rule employed, taking into account the level of risk (such as false accept and false reject and statistical assumptions) associated with the decision rule employed, and apply the decision rule. • NOTE: Where the decision rule is prescribed by the customer, regulations or normative documents, a further consideration of the level of risk is not necessary. 10
  • 11. Relevant clauses 7.8.6 of ISO/IEC 17025:2017 • 7.8.6 Reporting statements of conformity (contd.) • 7.8.6.2 The laboratory shall report on the statement of conformity, such that the statement clearly identifies: • a) to which results the statement of conformity applies; • b) which specifications, standards or parts thereof are met or not met; • c) the decision rule applied (unless it is inherent in the requested specification or standard). The laboratory shall report on the statement of conformity, such that the statement clearly identifies: • To which results the statement of conformity applies; • Which specifications, standards or parts thereof are met or not; • The decision rule applied 11
  • 12. To make a decision rule, we must have…. • A measurand (analyte) clearly specified (i.e. specification of the object) • A test result (normally assuming normal distribution of test results) • Its associated measurement uncertainty • A specification or regulation giving upper and/or lower limits • A decision rule This rule can decide to take or not to take measurement uncertainty into account, but it includes the risk of making a wrong decision, of which the involved parties are willing to take. 12
  • 13. Aims of making a decision rule • One should be clear on what type of risk he is taking when making a proper decision rule: • the supplier’s (laboratory’s) risk (Type I error, ) or • the consumer’s (customer’s) risk (Type II error, ) • From the laboratory point of view, we should be looking at Type I () False Positive error, as we want to minimize and control our risk in issuing a statement of conformity that might be proven wrong later. • We usually set the error or risk that we can afford to take, e.g.  = 0.05 (or 5%), bearing in mind that our test results have associated measurement uncertainties. 13 Important assumption: The uncertainty of measurement is represented by a normal (Gaussian) probability distribution function, which is consistent with the typical measurement results (being assumed the applicability of the Central Limit Theorem)
  • 14. What is Type I () error? • A type I error is caused by wrongly rejecting a true (“Pass”) situation (statistically speaking, a null hypothesis Ho) • This happens when the test result is found to be close to the specification value. • In a hypothesis (significance) testing, we write: • Null hypothesis Ho : test result = specification value (not literally but with probability) • Alternate hypothesis H1 : test result > or < specification value (depending on the upper or lower limits of specification) • Generally, we want to control our error (risk) when making a statement to reject Ho, i.e., saying the test result fails to meet with the given specification limit • We normally fix such error at  = 0.05, or 5% error (or 95% confidence) 14
  • 15. Today’s practice on decision rules • In current practice, we have been making a very simple decision rule, often based on direct comparison of measurement value with the specification or regulatory limits. • The reason is either that • these limits are deemed to have assumed including the measurement uncertainty, which is not normally true, • or, it has been assumed that the measurement value has zero uncertainty! • But, due to the presence of uncertainty in all measurements, we are actually taking more than 50% risk to allow the actual true value of the test parameter found outside the specification when reporting the measurement value exactly on the specification limit. 15 >50% Risk above specification limit! Specification- upper limit (Maximum)
  • 16. A worked example • Suppose a test method has a measurement uncertainty (+ 0.35ppm) of a lead Pb content near its upper feed product specification limit of 10ppm. • Upon analysis, a result of 9.65+0.35ppm was found in the feed sample given with 95% confidence. • This result indicates that the true Pb content of the sample is somewhere within the range: 9.30ppm to 10.00ppm with 95% confidence • It means that there is 2.5% chance that the true Pb content is outside both ends of this range. 16
  • 17. The meaning of Measurement Uncertainty ◼ For a measured value of 9.65ppm with an uncertainty of 0.35ppm, we shall write : ◼ 9.65 + 0.35 ppm ◼ The uncertainty range covers : ◼ 9.30 – 10.00 ppm with 95% confidence that the true value will lie somewhere within: ◼ 9.30 9.65 10.00 ppm ◼ (True Value) 17 =0.025 uncertainty =0.025 uncertainty
  • 18. A worked example • Remember that this uncertainty (U) of +0.35 ppm comes from • 2 x 0.175 ppm • where 2 is the coverage factor for normal distribution at about 95% confidence and, • 0.175 ppm is the combined standard uncertainty (u) from the MU evaluation study. • To be exact, the coverage factor should be 1.96 for 95.00% confidence for two-tailed distribution situation (+). By MS Excel functions: -1.9600 =NORM.INV(0.025,0,1) 1.9600 =NORM.INV(0.975,0,1) 18
  • 19. Use of Excel functions for normal distribution Excel® entry -1.960 =NORM.INV(0.025,0,1) 1.960 =NORM.INV(0.975,0,1) 0.975 =NORM.DIST(1.96,0,1,TRUE) 0.950 =NORM.DIST(1.645,0,1,TRUE) “=NORM.INV(probability, mean, standard_deviation)” “=NORM.DIST(x, mean, standard_deviation, cumulative)” Proportion of area under the curve from - to z- value. When mean = 0 and SD = 1, x = z. 19 z = z =
  • 20. A worked example • Hence, if we were to decide on 9.65 ppm as our critical value to make a “Pass” statement of conformity with confidence, we are actually taking a 2.5% risk, instead of 5.0%, to be proven wrong if the same sample were to be tested elsewhere by the same procedure. • But we are prepared to take a usual 5% risk in making such statement. • What is the critical value if we wish to make a decision rule to take a 5% risk (or making a 5% error in rejecting the test result) instead? 20
  • 21. A worked example • Remember the product specification’s upper limit is 10.0 ppm Pb • Our concern therefore is how much uncertainty of our test result is in the rejection zone above the 10 ppm limit • Hence, the coverage factor should be chosen at the right-tailed normal distribution instead of two-tailed distribution. • This coverage factor = 1.645, instead of 1.960 under normal probability distribution. By MS Excel function: -1.6449 =NORM.INV(0.05,0,1) 1.6449 =NORM.INV(0.95,0,1) 21
  • 22. A worked example • Therefore, the critical value for us to take a 5% risk to reject the result against the upper specification limit is therefore: • 10.00 – 1.645 x 0.175 or 10.00 – 0.288 or 9.71 ppm Pb • That means our decision rule can be stated as below: • “A pass statement shall be reported when the test result is equal or smaller than 9.71 ppm Pb content in the feeds with 95% confidence. A provisional or conditional pass can be reported when the test result is above 9.71 ppm Pb but below 10.00 ppm Pb. A fail statement is to be declared for test results above 10.00 ppm Pb. ” 22
  • 23. In summary, graphical effect of the decision rules: Specification: 10.0 ppm Max 9.65 ppm 9.71 ppm U = 0.35 ppm U = 0.29 ppm 2.5% Risk 5.0% Risk Acceptance Zone Rejection Zone 23
  • 24. Use of Guard Band g with acceptance limits • ILAC-G8, Eurolab documents suggest the use of Guard Band g which is the current trend of thinking. • Guard band (g) is defined as: “interval between a tolerance limit and a corresponding acceptance limit where length g = |TL – AL|.” Upper Tolerance Limit TL Lower Tolerance Limit TL Acceptance Limit AL Acceptance Limit AL g g Nominal 24
  • 25. A graphical example of areas defined for a specification interval with guard bands + U 25
  • 26. Example 1A of areas defined for a specification interval with guard bands inside the set limits – conservative or stringent acceptance zone (Type I ) model 26 In this model, the risk taken is 2.5% at  = 0.025 only at upper or lower specification limits
  • 27. Example 1B of areas defined for a specification interval with guard bands inside the set limits – conservative or stringent acceptance zone (Type I ) model In this model, the risk taken is 5% at  = 0.05 only at upper or lower specification limits 27
  • 28. Example 2 of areas defined for a specification interval with guard bands outside the set limits – Relax rejection zone (Type II  ) model 28
  • 29. Example 3 of areas created by considering the upper and lower tolerance limits given by the customer with the laboratory’s own measurement uncertainty 29
  • 30. Similar approaches for microbiological counts • Microbiological analysis is based on counting the number of colonies in terms of CFU (Colony Forming Units) after incubation with appropriate agar media. • Discrete counting does not follow normal probability distribution. • A logarithmic transformation of the counts is to be made in order to assume a normal distribution for continuous data • For example, for a count of 100 CFU, Log10(100) = 2.000 • In MU study of microbiological counts, the combined standard uncertainty is expressed in the form of logarithmic relative standard deviation, Log10, rel • Same principles of factor 1.645 are applied like those in chemical analysis. 30
  • 31. The binary approach • Applicable for qualitative “Pass/Fail”, “Present/Absent”, etc. testing • When you take two samples for testing and if both pass, you report “Pass” • If one passes and another fails, then you need to take two more samples for analysis • If both pass, it is determined as a pass • If either or both fail, it fails. • This approach has implicitly taken uncertainty into account and is therefore the defined “decision rule” 31
  • 32. In reporting …. • The laboratory shall : • apply the decision rule • Ensure its report to contain either the decision rule used or a reference to a standard or agreed decision rule • Listing the uncertainty of measurement result is optional, depending on customer’s liking • Making a disclaimer statement, e.g. by ILAC Guide 8.03 “The statement(s) of compliance with specification (or requirement) is based on a 95% coverage probability for the expanded uncertainty of the measurement results on which the decision of compliance is based”. • There must be enough detail so that if another lab is tasked with taking the same measurements and using the same decision rule, that they will come to the same conclusion about a “pass” or “fail”, in order to avoid any legal implication. 32
  • 33. Decision rule documentation, the lab shall …. 33 document the decision rules employed taking into account the level of risk associated with the decision rule employed (i.e. false accept and false reject) and statistical assumptions associated with the decision rule employed; prepare a list of critical values for the test parameters with Type I error α = 0.05, i.e. 95% confidence, normally after some statistical hypothesis analysis take time to clearly document what has been agreed with the clients and ensure that lab procedures allow for the correct treatment of the readings and uncertainties might need to change test procedures to cater for the decision rule that will be applied such as a review of measurement uncertainty level. It has to be such that anybody in the lab will know what is required and what the limits are.
  • 34. Parting words …. • Decision rule construction requires four components: • the measurement result, • its measurement uncertainty, • the specification limit or limits, and • the acceptable level of the probability of making each type of wrong decision. • Choice of a decision rule is a business consideration that takes into account the cost of rejecting an in- specification product, the cost of accepting an out-of- specification product, the uncertainty associated with the measurement, the cost of making the measurement, and client’s relationship for future business. 34
  • 35. i.e., the lab must agree with the customer on its decision rule and communicate clearly about the level of risk associated with this decision rule. Conclusions • Issuance of a certificate of conformity or a statement of compliance allows the laboratory to carry a fair bit of responsibility. • The laboratory must be confident on its own measurement uncertainty estimation. • If a statement of conformity is required by the clients, a lot more detail is required to be discussed before carrying out the laboratory analysis. • ISO/IEC 17025 requires the Lab to be very clear about the criteria used and must communicate with and be agreed by the customer, defined in the test report, certificate of conformance or statement of compliance to avoid future disputes. 35
  • 36. Useful References • ILAC-G8:09/2019 : Guidelines on Decision Rules and Statements of Conformity • EuroLab Technical Report No. 01/2017 : Decision rules applied to conformity assessment • UKAS LAB 48 Edition 3 (June 2020) : Decision rules and statements of conformity • OIML Guide G 19 (2017) : The role of measurement uncertainty in conformity assessment decisions in legal metrology • JCGM 106:2012 : Evaluation of measurement data – The role of measurement uncertainty in conformity assessment 36