Metrological basis of IQC

1
Metrological basis of IQC
Metrological basis of IQC

• Introduction
• The error concept chosen here
• The total error concept
• Total error calculations
• Instability of the analytical process
• Instability and analytical process specifications
• Instability how much can be tolerated?
• Analytical process specifications - TEa
• IQC and TEa
• The error model for IQC
• Basic formula
• Critical errors
• Graphical presentation of critical errors
• Calculation of critical errors
• Special topic The TEa problematic

2
Introduction
Metrological basis of IQC

• FOREWORD
• Currently, there is confusion about metrological
  terms. While the term error is used in the
  metrological bible (VIM 1), it is deprecated
  in one of the recent ISO concepts (GUM 2). This
  concept, even, recommends to avoid the
  distinction between random and systematic effects
  ( uncertainty concept GUM).
• Moreover, ISO uses the concept of accuracy 3
• Accuracy Combined trueness precision
• Inaccuracy Combined (untrueness note this
  term is not defined by ISO) and imprecision.
• Note Some traditions link accuracy to systematic
  error.
• This book uses the total error concept (Adapted
  from Stöckl D. Scand J Clin Lab Invest
  199656193-7)
• Total-, systematic-, and random error
• InAccuracy combined untrueness and
  imprecision
• Experimental estimates Total error, bias, SD
• Note all experimental estimates have a
  statistical uncertainty!

Classical metrological error concept distinction
of random and systematic error. Inaccuracy,
untrueness and imprecision pendents for
total-, systematic-, and random error. Note
All experimental values (bias, s) have a certain
statistical level of probability (uncertainty).
3
The total error (TE) concept
Metrological basis of IQC

• The term "total error" (TE) () is applied to
  measurement results that are influenced by random
  (RE) and systematic (SE) components of error.
• Calculation of TE for single measurements
• TE SE z RE, or TE SE z s
• (z is usually set to 1.96 or 2.58, encompassing
  95 or 99 two-tailed of a gaussian distributed
  population)
• Calculation of TE for multiple measurements
• TE SE tn-1 s/?n
• s the standard deviation, n number of
  measurements, tn-1 student's t-value for (n-1)
  degrees of freedom (e.g., at the 95 confidence
  level)

4
Instability and analytical process specifications
Metrological basis of IQC

• Instability how much can be tolerated?
• The answer is ? Poor analytical quality must not
  invalidate the medical decision!
• The conclusion is We have to know the medical
  requirements for analytical quality (analytical
  process specifications) ? Allowable total error
  (TEa).
• Analytical process specifications - TEa
• Concepts and numerical examples can be found in
  the literature below ().
• Note Optimally, TEa should be medically
  justified!
• IQC and TEa
• Observation
• The statistically defined "stop-limit" of IQC
  (TEIQC e.g., 3 s) may be much lower than
  medically relevant (e.g., triacylglycerides).
• The idea
• Let's stop the process at a medically relevant
  (or otherwise defined) allowable total error TEa
  (TEa also called "quality specification").
• Advantages
• If TEa gtgt TEStable,
• "very loose" IQC-rules can be selected.
• Prerequisite for IQC
• TEstable lt TEa

How much instability? TEIQC, max lt TEa ? work
within specifications TEIQC,max Maximum total
error guaranteed by the IQC-procedure (95 of
results should be within TEa)
5
The error model for IQC
Metrological basis of IQC

• Basic formula ()
• TE Bias ?SEcont smeas z ?REcont
  smeas
• Bias method bias, usually assumed to be zero
• ?SEcont change in systematic error that can be
  detected by the quality control procedure
• smeas measured, intrinsic standard deviation of
  the test
• ?REcont change in random error that can be
  detected by the quality control procedure
• z statistical multiplier related to the portion
  of a distribution exceeding a quality requirement
  ("defect rate")
• Note mostly, a z-value of 1.65 is selected to
  obtain a 5 defect rate (one-tailed). When pure
  random error is addressed, a z-value of 1.96 is
  selected to obtain a 5 defect rate (two-tailed).
• Critical random and systematic error definition
• The amount of error that places 5 of results
  outside TEa.
• The role of IQC is to detect process
  deterioration before gt5 of the results exceed
  TEa.
• Graphical presentation of critical errors

Introduction of an SE that places 5 of results
outside TEa D SEcrit 2
Increase of an RE that places 5 of results
outside TEa D REcrit 2 (REi intrinsic or
stable imprecision)
6
Calculation of critical errors
Metrological basis of IQC

• Assumption of imprecision and bias
• Critical random error (?REc) and critical
  systematic error (?SEc) can be calculated that
  cause measurements to exceed TEa in 5 of the
  cases ( defect rate) due to either increased
  imprecision or increased systematic error.
• Formulae for a 5 defect rate
• Case with ?SEc 0
• ?REc (TEa - bias)/1.96 smeas
• Case with ?REc 1
• ?SEc (TEa - bias)/smeas - 1.65
• Assumption of zero bias
• For the purpose of IQC, the bias of the method
  often is assumed to be zero.
• This is justified if a laboratory establishes its
  own IQC target values or if it can verify the
  system-specific target values of commercial IQC
  materials.
• Formulae for ?REc and ?SEc with bias 0
• ?REc TEa/1.96 smeas
• ?SEc (TEa/smeas) - 1.65
• Note For simplification, the QC-Validator
  program applies the default factor 2 (for RE and
  SE) for the stable process (elder versions
  1.65!). The factor may be changed by the user.

7
Special topic the TEa problematic
Special topic the TEa problematic

• Introduction
• Specifications for TEa, the concepts
• Clinical
• Biological
• Notes on IQC rules based on biological TEa
• IQC rules from biological TEa and reality
• Expert
• Summary
• Checklist TEa values

8
Introduction
Special topic the TEa problematic

• BEWARE
• There are a lot of misunderstandings when
  selecting a TEa for IQC purposes.
• We look into TEa in more detail!
• Questions
• What is a responsible choice for TEa?
• Is there a consensus about TEa?
• Investigate current concepts for establishing
  values for TEa.
• Specifications for TEa
• Current concepts for TEa
• The Table below lists, in hierarchical order,
  current concepts for establishing values for TEa.
• Sources (in hierarchical order)
• Clinical concepts (few analytes)
• Concepts based on biological variation

9
Specifications for TEa
Special topic the TEa problematic

• Inspection of proposed numbers for TEa
• EXAMPLE Serum-sodium
• When we compare the currently proposed numbers,
  we can conclude that
• There is no consensus about TEa!
• Additionally TEa may depend on
  concentration/disease

10
TE from biology
Special topic the TEa problematic

• BUT Some are too stringent for current
  technology
• Biology gives extreme low values for some
  analytes
• TEa sodium Ricos 0.9 (http//www.westgard.com/b
  iodatabase1.htm but note CVg on that site
  CVb)
• ? Apply a bottom-line
• Stöckl D. Desirable Performance criteria based
  on biological analyte variation - hindrances to
  reaching some and reasons to surpass some. Clin
  Chem 199339913-4.
• Bottom-line (for stable process!)
• CV 1
• Bias 1,5
• TE 3,2

11
Proposed numbers How realistic are they?
Special topic the TEa problematic

• The numbers proposed by clinical and biological
  concepts, usually, are desirable numbers (
  goals), which would allow medical decisions to be
  made without compromise of analytical quality.
  However, the laboratory has to work with real
  numbers state-of-the-art quality.
• Comparison of reality with goals from biology
• The Figure below compares  state-of-the-art
  quality with desired quality 1/4 of the
  reference interval.
• 1 TEa ¼ of the reference interval
• 2 State-of-the-art total CVa.
• Rules calculated with the above TEa and CV
  values by the Validator
• (90 assurance 2 materials bias Na 0.5, Trigl
  2, others 1).
• Observation
• Some state-of-the-art CVs are too big to allow
  calculation of IQC-rules from biology-derived TEa
  values, for example, for ? sodium.

12
More on expert TEa values
Special topic the TEa problematic

• Below, are shown more examples of IQC rule
  setting by use of expert TEa values (upper
  example) and by general expert opinion (lower
  example).
• Koch, Westgard, et al. Clin Chem 199036230-33
  (see also Westgard Stein. Clin Chem
  199743400-3).
• Based on expert TEa values, following rules
  were proposed
• 13.5s (n 2) for Sodium (TEa 2.9, CV
  0.5), potassium, urea nitrogen, creatinine,
  phosphate, uric acid, cholesterol, total protein,
  total bilirubin, GGT, ALP, AST, and LD.
• 12.5s (n 2) for Chloride, total CO2.
• Calcium was a problem Compared to TEa, precision
  was considered as too poor.
• Mugan, Carlson, Westgard. J Clin Immunoassay
  199417216-22.
• 12.5s (n 3) for Prolactin, ß-hCG, CEA, FSH,
  LH, TSH, ß2-microglobulin.
• ? Experts tend to simple rules choice is not
  directly related to medical requirements
• Rules selected by general opinion
• Steindel SJ, Tetrault G. Arch Pathol Lab Med
  1998122401-8.
• 2.5 2.7s rule generally
• Tetrault GA. CAP Today 1995 (April)60-1.
• 3.5 s rule generally
• ? Experts tend to simple rules choice is often
  more related to considerations about Pfr than to
  medical requirements. Sometimes, experts change
  their minds.

13
TEa and IQC
Special topic the TEa problematic

• Summary
• Different models give different TEa values
• Additionally TEa may depend on
  concentration/disease
• Many proposals do not account for the needs of
  IQC
• Some values that are recommended are beyond the
  state-of-the-art of routine methods, or give IQC
  rules with high Pfr
• Experts tend to simple rules choice is often
  more related to considerations about Pfr than to
  medical requirements. Sometimes, experts change
  their minds.
• Dont choose the most convenient TEa.
• The most useful general purpose TEa numbers
  result from the concept of biological variation.
• But, apply bottom-line values
• Checklist TEa values
• Apply TEa values from the following hierarchy
• 1. Clinical models (e.g., cholesterol glucose)
• 2. Biological variation (obtain the database)
• Bottom-line values
• No numbers from 1-2 3. Expert models

14
Control rules based on TEa
Control rules based on TEa

• Automatic selection of rules based on TEa The
  Validator
• OPSpecs-Charts
• Critical error graphs
• Selection of a control rule based on TEa with
  the Validator an example
• Other selection tools the IQC decision tool
  the TEa/CVa,tot ratio
• Summary and Conclusion

15
Selection of control rules based on TEa
Control rules based on TEa

• Tools for IQC rule decision based on TEa
• The Westgard software tools (OPSpecs chart)
• The STT IQC decision tool
• The ratio of TEa/total-CVa should be gt4
• Automatic selection of rules based on TEa The
  Validator
• Based on the TEa concept and the calculation of
  critical errors, the Validator software allows an
  automatic selection of the most appropriate
  control rule.
• OPSpecs-charts
• The Validator makes use of so-called
  OPSpecs-charts (Charts of Operational Process
  Specifications) to relate TEa and stable method
  performance (bias and imprecision) with IQC-rule
  selection.
• Based on a preselected TEa, these charts allow to
  select IQC rules for different combinations of
  bias/imprecision (see Figure).
• The charts are specific for a specific total
  error requirement, for example, a chart for a TEa
  of 5 will differ from one requiring 2.
• A selection between QC procedures for systematic
  error or random error has to be made and the
  probability level for error detection has to be
  specified, for example, 90 (80 for RE), 50,
  etc (OPSpecs-Charts for RE not in EZ rules!).
• Imprecision is plotted on the x-axis and bias is
  plotted on the y-axis. Stable imprecision and
  bias are plotted as operating point (Figure
  2/0).
• A selection of IQC-rules is presented.

16
Critical error graphs
Control rules based on TEa

• The Validator
• Additionally, it shows the power functions of the
  rules with indication of the critical error (
  critical error graphs).
• Critical error graphs are made for systematic
  and random error, separately.
• They show the power functions for the IQC rules
  that the Validator proposes.
• Additionally, a box is presented that indicates
  Pfr and Ped of the respective rules.
• The critical error is indicated by a vertical
  line. Usually, it should intersect the power
  functions at Ped values gt0.9 (90 for error
  detection).
• ?Note for the RE graph, the critical error line
  intersects most power functions at values lt0.9.
  This indicates the generally weaker power of
  these rules to detect random error!

17
Selection of a control rule based on TEa
Control rules based on TEa

• EXAMPLE
• Analyte
• Cholesterol
• TEa
• 13 From German EQA (RILIBÄK)
• Assumed stable performance
• Bias 0
• CVa,tot 2
• Apply the Validator
• See OPSpecs-chart
• ?The Validator selects a very convenient rule
  13.5s.
• Note powerful control rules are at the right in
  the OPSpecs!
• Conclusion
• If TEa is high Easy rules are selected

18
The OPSpecs chart a more detailed look
Control rules based on TEa

• The Figure shows a simplified version of an
  OPSpecs chart.
• Construction of the lines
• TE SE k RE at RE 0 SE TE
• ? all lines start at RE0SETE they stop at
  RETE/k, SE0
• According to the Westgard approach, the operating
  point of your test ( stable performance) should
  be left of the TE line with k 4. If the
  operating points is located on the right, the TE
  cannot be controlled in 90 of the cases by IQC.
• REMEMBER
• TEa (preselected!) SE k RE (SE RE are
  calculated from TEa)

19
Remark on the location of the operating point
Control rules based on TEa

• The Figures below show the expected measurement
  populations (for different k values) for
• an operating point with a low ratio SE/RE (upper
  Figure)
• an operating point with a high ratio SE/RE
  (lower Figure).

20
Other selection tools
Control rules based on TEa

• As already addressed before, different Validator
  versions may use different default k values for
  the stable process
• - Either 1,96 pure RE
• - Or 1.65 pure SE
• Usually, both components are present. In
  practice, there is a gradual move from 1.96 to
  1.65 when SE moves from 0 to SE RE. If SE gt RE,
  the multiplier 1.65 is justified. This has also
  implications for the IQC controlled process. For
  this reason, the STT variant was developed.

Problem of the OPSpecs charts When RE becomes gt
SE, one moves from the 1-sided case (k 1,65 k
4) to the 2-sided case (k 2 k 4,85)
21
The ratio of TEa/total-CVa (EXCEL-file)
Control rules based on TEa

• The ratio of TEa/total-CVa allows a quick
  estimate of test performance. This ratio should
  be gt4. However, the ratio depends strongly on the
  TEa chosen (see also discussion later special
  topic TEa problematic).
• REMARK bias is not considered by that tool!
• The EXCEL-file contains a nearly complete list of
  TEa and state-of-the-art CVa values for the
  following analyte groups
• Ion selective electrode (ISE)
• Substrates
• Enzymes
• Specific proteins
• Therapeutic drug monitoring (TDM)
• TEa criteria are
• Ricos et al., or
• 1/6th of the reference (therapeutic) interval
• CVa data
• State-of-the-art
• Summary

22
IQC policy
IQC policy

• Introduction
• Software
• Samples
• Frequency ( location) of IQC measurements
• Performance (State-of-the-art)
• IQC rule selection
• Patient release
• Process control
• Examples
• IQC rules for state-of-the-art performance
• Screening with TEa/CVa,tot
• STT IQC decision tool
• EZ rules/Validator

23
Introduction
IQC policy

• The IQC policy will be developed on the basis of
  the Belgian situation
• RECALL Koninklijk Besluit
• Art. 34. 1. The laboratory director has to
  organize IQC in all disciplines.
• 3. IQC consists of several procedures which
  allow, before the release of patient results, to
  detect all significant within- or between-day
  variations Remedial actions policy/rule
  selection
• Art. 35. 1. The frequency of control
  measurements has to be such that it can guarantee
  a clinically acceptable imprecision. This
  frequency depends on the characteristics of the
  method and/or the instrument Additional
  Praktijkrichtlijn
• Control rules used for start and for acceptance
  of a run ? at least 2 IQC events.
• Rule selection/frequency
• 2. The control material, must be stable within
  a defined period of time. Different aliquots of
  the same lot must be homogeneous
• Concentrations Praktijkrichtlijn.
• Sample requirements
• 3. For each new lot, the mean and the SD have to
  be determined. IQC materials may, at the same
  time, not be used as calibrator and control
  material Establish stable performance own
  targets!

24
Introduction
IQC policy

• On the basis of the Belgian regulation, we will
  apply all input elements that we have seen to the
  analytes ISE substrates enzymes, specific
  proteins TDM.
• The elements we have seen (see also ckecklists)
  were
• Knowledge-base
• -Statistics
• -Rule selection
• -Metrology/Error concept for IQC
• -TEa data, biological variation
• State-of-the-art performance data
• -CVa,tot, CVa,w
• -Experience from "Peer-IQC"
• -Laboratory experience
• (also Questionnaire to participants)
• Software tools
• -Laboratory IQC software
• -EZ-rules

25
IQC-software
IQC policy

• Discussion
• 
  
  
  
  .
  
  
  
  
  
  
  

26
The IQC sample
IQC policy

• Samples 2. The control material, must be
  stable within a defined period of time. Different
  aliquots of the same lot must be homogeneous.
• Concentrations Praktijkrichtlijn
• See also checklist Samples
• ? Select commercial one
• Note In principle, their target and range cannot
  be used by the laboratory (Art. 35. 3.) (3. For
  each new lot, the mean and the SD have to be
  determined).
• Consider participation in an IQC "Peer"-system
• Sample advantages due to the high number of
  participants
• System specific target means and SDs
• Low target uncertainty
• Control of sample stability
• Discussion
• 
  
  
  
  .
  
  
  

27
Frequency of measurement and location
IQC policy

• Frequency Art. 35. 1. The frequency of control
  measurements has to be such that it can guarantee
  a clinically acceptable imprecision. This
  frequency depends on the characteristics of the
  method and/or the instrument.
• Praktijkrichtlijn
• Control rules used for start and for acceptance
  of a run.
• ? Frequency at least 2 IQC events
• ? Location at least at start and end
• ? See also checklist Frequency and location
• Discussion
• 
  
  
  
  .
  
  
  

28
Stable performance
IQC policy

• Stable performance
• 3. For each new lot, the mean and the SD have to
  be determined.
• ?The laboratory has to establish its own stable
  performance data.
• REMEMBER
• The target SD ( stable imprecision) is the
  cornerstone of IQC. It deserves special
  attention. All instabilities (random and
  systematic) are compared relative to the stable
  imprecision.
• Recommendation 1
• Compare your performance with your colleagues
  (IQC-"Peer")
• Advantages due to the high number of participants
• Better IQC-sample
• Easier set-up of IQC (more reliable estimates of
  stable performance)
• Easier troubleshooting by direct comparison with
  peer
• Recommendation 2
• Compare your performance with the data from the
  manufacturer!
• See EXCEL-file for an overview about
  state-of-the-art performance data for the
  Analyte groups

29
Stable performance
IQC policy

• Discussion
• 
  
  
  
  .
  
  
  
  
  
  
  

30
IQC rule selection
IQC policy

• Generally, IQC rule selection can be done on
• Statistical basis A rule is chosen based on Pfr
  and Ped. SD-limits are taken from stable
  performance.
• TEa basis From a specification for TEa,
  critical error values can be calculated.
• From the critical error values, adequate IQC
  rules can be selected, naturally, on statistical
  basis.
• Selection tools are power functions, OPSpecs,
  critical error graphs, the TEa/CVa,tot ratio the
  IQC decision tool
• ? See also checklists basic statistics power of
  control rules metrology.
• Belgian regulation, however
• 3. IQC consists of several procedures which
  allow, before the release of patient results, to
  detect all significant within- or between-day
  variations
• Art. 35. 1. guarantee clinically acceptable
  imprecision?
• What is significant what is clinically
  acceptable?
• ? We need a goal for TEa!
• requires, in principle, IQC rule selection on
  the TEa basis.

31
Selection of TEa
IQC policy

• Discussion
• 
  
  
  
  .
  
  
  
  
  
  
  

32
Example
IQC policy

• Selection of the TEa
• We select TEa from biology
• TEa from the Ricos concept, and
• The most stringent TEa, with bottom-line
• We apply TEa to manufacturers performance
• Example
• Serum Glucose
• Ricos TEa 6.3
• CVa,tot 2
• Bias 0
• Exercise
• STT IQC decision tool

33
The modified TEa approach
IQC policy

• Apply the TEa approach (TEa/CVa,tot ratio the
  Validator/EZ rules IQC decision tool) as
  evaluation tool for current quality.
• Analytes that cannot fulfill the TE
  specifications
• ? Decide on the most stringent rule you want to
  apply (Pfr n)
• Analytes that fulfill TE very easy
• ? Decide on the most loose rule you want to apply
• All others decide whether you want to use
  individually optimized rules
• Different rules
• Same rule, but movement
• Adapt the frequency (less/more measurements)
• Problem
• Optimization always should consider individual
  test stability!
• Example 1 most stringent rule (left)
• Westgard multirule 13s / 22s / R4s / 41s
• n 4 (required for full power)
• Pfr 3?Relatively high!
• Ped for a 2.5s-shift 90
• Ped for 3 RE 90

34
The modified TEa approach
IQC policy

• Example 2 for the most stringent rule (left)
• Westgard multirule 13s / 22s
• n 4
• Pfr 1
• Ped for a 2.5s-shift 83
• Ped for 3 RE 83

35
IQC rule selection Process control
IQC policy

• Basic idea
• Independent of using IQC for patient release, we
  may want to use it for process control!
• ? For process control, each test could be
  controlled by the same IQC procedure.
• ? For process control, higher Pfr values can be
  accepted. However, Ped should be considerably
  higher than for patient release.
• Examples
• Westgard multirules with high n Ped (90) 1.8 D
  SE (see Figure) for the logic of Westgard
  multirules lower Figure
• Mean and variance rules with high n

12s/13s/22s/R4s/41s/10x(mean) Pfr 3, Ped for a
2.5s shift 90 (for n 4, without the mean
rule)
36
Rule selection
IQC policy

• Discussion
• 
  
  
  
  .
  
  
  
  
  
  
  

37
Selection of IQC rules Tools Examples
IQC policy

• Tools
• The ratio of TEa/total-CVa should be ?4
• The STT IQC decision tool
• The Westgard software tools (Validator/EZ Rules)
• EXERCISES
• Sodium
• We apply the modified TEa approach, the most
  stringent TE (with bottom-line), and the
  state-of-the-art CV data
• 1. Select TEa Bottom-line 3.2
• 2. Apply the IQC decision tool
• - Set stable Bias 0
• - Stable CV (total) 1
• 3. Sort of analyte?
• 4. Apply Validator for individual rule selection
• OUTCOME
• The IQC tool and the Validator show that sodium,
  with the values chosen, cannot be controlled
  according to the TEa concept.
• The ratio of TEa/total-CVa is 3.2.

38
IQC rule selection Applications
IQC policy

• First, we screen the analytes with the
  Ricos-TEa/CVa,tot ratio
• The tables show analytes with ratios lt4 (full
  list in EXCEL-file).?In principle, the most
  stringent IQC rule should be applied for all
  these analytes.
• Example 13s/22s (n 4)

TEa/CVa,tot ratio
TEa/CVa,tot ratio
39
IQC rule selection Applications
IQC policy

• The tables below show analytes with ratios ?4
  (full list in EXCEL-file).
• ?In principle, individually optimized IQC rules
  can be applied for all of these analytes (note
  most loose 13.5s).

TEa/CVa,tot ratio
TEa/CVa,tot ratio
40
IQC rule selection Applications
IQC policy

• Excercises with the EZ rules, using the specific
  protein examples (ratios from 4-7).
• Note all fields must be filled in (may be dummy
  values)!
• Selection criteria 2 materials, SE detection 90
• APO A1 (4.3) Rule 12.5s n4 Pfr 4
• a-1-Antitrypsin (5.1) Rule 13s/ 22s n2
  Pfr 1
• IgA (6.8) Rule 13.5s n2 Pfr lt0.1
• Example (6) Rule 13.5s n2 Pfr lt0.1
• Observation from ratios gt6, the process can be
  controlled very easy! ? 6 sigma process
• We take the 13.5s (n4) rule for all analytes
  with a TEa/CVa,tot ratio ?6

TEa/CVa,tot ratio
TEa/CVa,tot ratio
41
IQC rule selection Applications
IQC policy

• Screening with the Ricos-TEa/CVa,tot ratio all
  ?4 - lt6
• The table shows examples where optimized rules
  could be used. However,
• Consider Is it worth?
• Should one use the most stringent IQC rule?
• REMARK
• Remember process control.
• One stringent rule could be used for all
  analytes.

TEa/CVa,tot ratio
42
VARIA
IQC policy

• Calculation of an actual TE
• Rule n and patient release
• Dealing with a bias
• Rules with wide limits (e.g. 6s) lot
  variations
• Fine-tuning of IQC according to instability
• Calculation of an actual TE
• When actual performance does not satisfy the TEa
  requirement, and one selects the most stringent
  rule, one can calculate (or extract with EZ
  rules) the real TE for the actual performance and
  the rule chosen.
• Examples, IQC rule 13s/22s (n 4)
• CVa Ricos TEa Real TE
• () () ()
• Chloride 1.0 1.5 4.3
• Albumin 1.2 3.9 5.2
• Calcium 2.0 2.4 8.6
• Note with this rule (without bias), actual TE
  4.3 CVa
• Rule n and patient release
• When IQC rules with higher number of n are
  selected, and IQC measurements are done
  continuous, patient release has to be postponed
  until the full power of the rule is reached
  (cumulation needed).

43
VARIA
IQC policy

• IQC rules with wide limits (e.g. 6s) and lot
  variations
• When rules with wide limits are used, significant
  lot-to-lot variations may be recognized in the
  chart, but may not cause rule violations (see
  Figure below).
• Decide about
• the medical relevance of the variations
• whether you want to pick them up by the process
  control
• whether you address them with quality assurance
  means.
• Note Similar may happen when you overestimated
  CVa (for example, high CVa,tot/CVa,w ratios).

44
Varia
IQC policy

• Discussion
• 
  
  
  
  .
  
  
  
  
  
  
  

45
Remedial actions
IQC policy

• EXCERCISES
• Inspecting IQC charts
• Note in advance When results of controls fail
  to meet the laboratory's established criteria for
  acceptability, all patient test results obtained
  in the unacceptable run must be evaluated to
  determine if patient results have been adversely
  affected and the laboratory must take remedial
  actions to ensure the reporting of accurate and
  reliable patient test results
• Westgard philosophy (Basic QC Practices)
• Dont simply rerun the control
• Dont simply try a new bottle
• ? Identify the problem (inspect the chart) and
  correct it
• Which type of error (concentration dependent,
  SE, RE)
• Relate the error to a potential cause
• Consider common factors (temperature pipetting
  volumes kind of test kinetic/endpoint
  wavelength)
• Relate to recent changes (operator, calibration,
  reagent, IQC-material, maintenance, etc.)
• Demonstrate and document that the error was fixed
• Run controls after fixing the problem
• Acceptability limits

46
Pfr of the IQC rule and frequency of remedial
actions
IQC policy

• When you compare your actual short-to-medium term
  frequency of remedial actions with the Pfr you
  expect from the IQC rule, consider the nature of
  the CVa,tot you chose. The more variations you
  included, the lower the frequency of remedial
  actions will be.
• If actual remedial actions frequency ltlt rule Pfr
• Consider to pick up certain variations by IQC
• ? Consequence reduces your CVa,tot that you use
  for IQC
• ? May allow a different rule when the TEa concept
  was used.
• Consider that you used a wrong CVa,tot
• The state-of-the-art example
• We have 40 analytes with the rule 13s/22s, Pfr
  1 (n 4), 30 analytes with the rule 13.5s,
  Pfr 0,2 (n 4), and assume 8 IQC measurements
  per day. Then,
• ? we expect 80 chance for a false rejection
• ? Under stable conditions, we expect at maximum 1
  remedial action per day,
• More, if system is unstable
• Less, if Cva,tot gt CVa short-to-medium term.
• CVa,tot /CVa,w ratio
• Note Troubleshooting is facilitated by increased
  CVa,tot /CVa,w ratios.

47
Remedial actions
IQC policy

• Note Either the measurement procedure or the
  control procedure can be faulty.
• Release of patient results
• You suspect the QC sample was the reason for
  failure
• Take a new bottle.
• Remeasure with n 4, for example. If the IQC
  rule is not violated, assume the IQC sample
  caused the problem.
• However, realize the uncertainty of your
  estimate!
• ?Keep an eye on the process for a while.
• You suspect a calibration problem
• Recalibrate the system, however, document the
  post-calibration status by a sufficient number of
  control measurements (for example 4).
• The system is under control again when the IQC
  rule is not violated. However, realize the
  uncertainty of your estimate!
• Remeasure the patient samples between the last
  in-control event and the out-of-control event.
• Compare the mean of the new patient results with
  that of the old results. When there was a
  calibration problem, this should be reflected in
  the difference of both means.
• You suspect other problems
• Look at patient data
• Investigate all IQC levels ( IQC across
  materials), check other analytes ( IQC across
  analytes).

48
Remedial actions
IQC policy

• Process control
• Rules with higher Pfr (e.g. 2.5s, Westgard
  multirules, or others) can be used for closely
  monitoring the analytical process.
• In case that these rules are violated, patient
  results can still be reported. However, when
  time, the analytical process (calibration,
  reagent, instrument) or the IQC material should
  be investigated.
• Possibilities
• Look at patient data
• Check electronic QC data
• Investigate all IQC levels ( IQC across
  materials), check other analytes ( IQC across
  analytes).
• Invstigate related analytes (e.g., all enzymes
  could be temperature related).
• Compare the actual CV with the expected one, and
  compare the IQC results with appropriate
  IQC-"Peer" (EQA) data.
• When necessary, recalibrate the system and check
  the success of recalibration as described above.
• Check instrument (maintenance), reagents,
  operator change.
• In general, trouble-shooting should be done with
  an increased number of control measurements,
  however, higher CVa,tot/CVa,w ratios facilitate
  troubleshooting.
• Over the time, check the success of your remedial
  actions.
• CAUTION

49
Remedial actions
IQC policy

• Discussion
• 
  
  
  
  .
  
  
  
  
  
  
  

50
Internal Quality Control
IQC policy

• General discussion
• 
  
  
  
  .
  
  
  
  
  
  
  

51
Reminders
IQC policy

• Reminder 1
• Troubleshooting should be done with replicates!
• Reminder 2
• Dont forget proactive quality assurance
• A new lot (bottle) of IQC material is introduced
• The operator changed
• Instrument was calibrated or underwent major
  maintenance
• New lots (bottles) of reagents or calibrators
• Overall summary
• The guiding rule (regulation)
• Knowledge
• Basic statistics
• Power functions
• TE error concept (metrology)
• TEa (critical errors, specifications)