Laboratory QAQC Quality AssuranceQuality Control

1
Laboratory QA/QC (Quality Assurance/Quality
Control)

• Year 2006

2
Course Outline

• Week 1 Fundamental Statistics (Normal
  distribution, outlier and examples) ?Glossary and
  explanation
• Week 2 Detection limit (method detection limit,
  instrument detection limit) and calibration curve
• Week 3 Quality assurance/Quality control
  (definition, standard curve, control chart)
• Week 4 Quality assurance/Quality control
  (definition, standard curve, control chart)
• Week 5 Result expression, sample correction and
  others
• Week 6 Examination

3
?????????????
4
Fundamental Statistics

• Normal or Gaussian distribution If a measurement
  is repeated many times under essentially
  identical conditions, the results of each
  measurement, x, will be distributed randomly
  about a mean value (arithmetic average) because
  of uncontrollable or experimental error.
• Standard deviation? is denoted by s (number of
  estimate is finite)

5
Fundamental Statistics

• Therefore, 95 of the measurement lie between ?
  2?
• Variance of the population (s2)
• Standard error of the mean Standard deviation
  divided by the square root of the number of the
  value (s/ )
• Confidence limit 104, 6 and 14 are the limit
• Confident interval from 6 to 14

6
Normal distribution
?average, ?standard deviation
7
(No Transcript)
8
Outlier (Rejection of data)

• Theoretically, no result should be rejected
• First, calculate the statistic T
• T (XH - X)/s for a high value, or
• T (X - XL)/s for a low value.
• Second, compare the value of T with the value
  from Table 1010I for either a 5 or 1 level of
  significance

9
Examples for rejecting outliers
Find the outliers from the following data for 5
or 1 level of significance
10
Glossary

• Accuracy combination of bias and precision of
  an analytical procedure, which reflects the
  closeness of a measured value to a true value.
• Precision measure of the degree of agreement
  among replicate analyses of a sample, usually
  expressed as the standard deviation.
• Bias consistent deviation of measured values
  from the true value, caused by systematic errors
  in a procedure.

11
Glossary Instrumental detection level (IDL)

• the constituent concentration that
• Produce a signal greater than three standard
  deviation of the noise level (close to 99
  probability that it is different from the blank)
• produces a signal greater than five times the
  signal/ noise ratio of the instrument.
• For seven replicates of the sample, the mean must
  be 3.14s above the blank where is the standard
  deviation of the seven replicates.
• Referred to new_niea_pa107.doc

12
Student T test
13
(No Transcript)
14
Graphical definition of IDL
Note Determined from the blank which would be
near zero concentration
15
Glossary Method detection level (MDL)

• the constituent concentration that, when
  processed through the complete method, produces a
  signal with a 99 probability that it is different
  from the blank (zero).
• For seven replicates of the sample, the mean must
  be 3.14s above the blank where is the standard
  deviation of the seven replicates. Compute MDL
  from replicate measurements one to five times the
  actual MDL.

16
Glossary Method detection level (MDL)

• Prepare the seven portions over a periods of at
  least three days
• The initial estimated concentration of prepared
  solution is about one to five times of the MDL.
• Analyze the seven portions twice
• Compare the two population variances by F-test
• Referred to handout and new_niea_pa107.doc

17
Graphical definition of MDL
Note Determined from the concentration which
is near the expected limit of detection
18
Glossary

• Level of quantitation (LOQ)/minimum quantitation
  level (MQL)the constituent concentration that
  produces a signal sufficiently greater than the
  blank that it can be detected within specified
  levels by good laboratories during routine
  operating conditions.
• Typically, it is the concentration that produces
  a signal 10 time the LDL or two to three times of
  the MDL.

19
Glossary

• Quality assessmentprocedure for determining the
  quality of laboratory measurements by use of data
  from internal and external quality control
  measures, mostly referred for external quality
  control
• Quality assurancea definitive plan for
  laboratory operation that specifies the measures
  used to produce data of known precision and bias.
  
• Quality controlset of measures within a sample
  analysis methodology to assure that the process
  is in control.

20
Glossary

• Surrogate standarda pure compound added to a
  sample in the laboratory just before processing
  so that the overall efficiency of a method can be
  determined.
• Internal standarda pure compound added to a
  sample extract just before instrumental analysis
  to permit correction for inefficiencies.

21
Glossary Type I and Type II errors

• Type I errorWe make the mistake of rejecting
  the null hypothesis when it is true.?
  P(rejecting H0 when it is true).
• Type II errorWe make the mistake of failing to
  reject the null hypothesis when it is false.?
  P(failing to reject H0 when it is false).

22
Glossary Relative standard deviation

• Also known as the coefficient of variation (CV),
  which commonly is expressed as a percentage.
• If analyses at low concentrations yield a result
  of 10 1.5 mg/L and at high concentrations 100
  8 mg/L. the standard deviations do not appear
  comparable. However, the percent relative
  standard deviations are 100 (1.5/10) 15 and
  100 (8/100) 8. which indicate the smaller
  variability obtained by using this parameter.

23
Homework 1

• Grab your old statistics book, (if you never
  took, go to library to find it), explain the
  meaning of
• (1) t-distribution
• (2) typically, people say the reasons of 99 of
  the total measurement for normal distribution is
  when ??3? large (unlimited) or small number of
  samples (30 samples)
• (3) The meaning of F-distribution
• (4) Bring your statistic book to the class

24
Homework 2

• In the previous example, if eight aliquots are
  used. The eighth sample concentration is same as
  the first one for both sets. Find the new MDL.

25
F-distributioncompare two population variance
26
(No Transcript)
27
Calibration curve preparation

• As a minimum, measure three different dilutions
  of the standard when an analysis is initiated.
  (standard method). EPA Taiwan requires five
  different dilutions of the standard
• The lowest concentration should be Level of
  quantitation (LOQ)
• Subsequently, verify the standard curve daily by
  analyzing one or more standards within the linear
  range, as specified in the individual method.
  (preferably by different source of standard)
• Reportable analytical results are those within
  the range of the standard dilutions used.
  (standard method). Within 20 to 80 of the
  highest concentration is preferable (EPA Taiwan).

28
Calibration curve preparation

• Do not report values above the highest standard
  unless an initial demonstration of greater linear
  range has been made, no instrument parameters
  have been changed, and the value is less than 1.5
  times the highest standard.
• The lowest reportable value is the MDL, provided
  that the lowest calibration standard is less than
  10 times the MDL (It should be). If a blank is
  subtracted report the result even if it is
  negative.

29
Reporting

• Apply the MDL to reporting sample results as
  follows Report results below the MDL as "not
  detected."
• Report results above the LOQ with a value and its
  associated error

30
Calibration curve determination

• Linear model and non-linear model (rare).
• Least square method. For linear model, R2 should
  be greater than 0.995.
• Calibration factor (Internal standard) and
  Response factor (external standard). Mostly use
  for individual organic compounds (VOC, pesticide,
  GC, GC/MS, etc). The ratio of response (peak
  area) for standard(As)and amount of compound for
  standard(Ws)called calibration factor (CF, EPA
  Taiwan) or generally called response factor (RF,
  Standard method)

31
Quality Assurance

• Quality assurance (QA) is the definitive program
  for laboratory operation that specifies the
  measures required to produce defensible data of
  known precision and accuracy.
• Quality assurance (QA) includes Quality control
  and quality assessment

32
Quality Assurance Plan

• Establish a QA program and prepare a QA manual of
  plan. Include in the QA manual and associated
  documents the following items
• cover sheet with plan approval signatures,
• staff or organization responsibilities
• Sample control and documentation
• standard operating procedure for each analytical
  method (SOP)

33
Quality Assurance Plan

• analyst training requirements
• equipment preventive maintenance procedures
• calibration procedures
• corrective actions
• internal quality control activities
• performance audits
• data assessment procedures for bias and
  precision, and data reduction, validation, and
  reporting.

34
Quality Control

• Quality control (QC) may be either internal or
  external.
• Internal QC is the subject of this section All
  analysts use some QC as an intuitive effort to
  produce credible results. However, a good quality
  control program consists of at least seven
  elements
• certification of operator competence,
• recovery of known additions,
• analysis of externally supplied standards,
• analysis of reagent blanks,
• calibration with standards,
• analysis of duplicates.
• maintenance of control charts.
• ???????????????????????????????

35
Quality Control

• external QC, also known as "quality assessment,"
  is discussed in 1020C.

36
1. Certification of Operator Competence

• Before an analyst is permitted to do reportable
  work, competence in making the analysis is to be
  demonstrated.
• Make a minimum of four replicate analyses of an
  independently prepared check sample having a
  concentration between 5 and 50 times the method
  detection limit (MDL) for the analysis in that
  laboratory. General limits for acceptable work
  are shown in the following Table.

37
1. Certification of Operator Competence
38
2. Recovery of Known Additions

• Use the recovery or known additions as part of a
  regular analytical protocol. Use known additions
  to verify the absence of matrix effects. When a
  new matrix type is to be analyzed, verify the
  amount of interference.
• The sum of the duplicates and known additions
  must equal at least 10 of the number of samples.
  
• Make the known addition between 5 and 50 times
  the MDL
• or between 1 and 10 times the ambient level

39
2. Recovery of Known Additions

• Do not use a known addition above the
  demonstrated linear range of the method
• use concentrated solutions so volume change in
  sample is negligible. (new_niea_pa104.doc, P.7)
• Or calculate recovery by mass balance
• See Table 10201 for acceptable limits
• Referred to new_niea_pa104.doc

40
Example for spiked recovery

• A sample with 100 mg/L of calcium. 1000 mg/L
  standard of calcium, 5 ml, was used to spiked
  into 100 ml water sample. The result showed
  150mg/L. Find your recovery.

The recovery will be 100 if added volume is
ignored.
41
3. Analysis of Externally Supplied Standards

• As a minimum, analyze externally supplied
  standards whenever analysis of known additions
  does not result in acceptable recovery or once
  each day, whichever is more frequent
• Use laboratory control standards with a
  concentration between 5 and 50 times the MDL or
  near sample ambient levels, whichever is greater.

42
3. Analysis of Externally Supplied Standards

• Where possible, use certified reference materials
  as laboratory control standards. National
  Institute of Standards and Technology (NIST) or
  Standard Reference Materials are preferred, if
  available.
• Prepare them independently from the standards
  used for calibration

43
4. Analysis of Reagent Blanks

• Analyze reagent blanks whenever new reagents are
  used
• Analyze a minimum of 5 of the sample load as
  reagent blanks
• Analyze a reagent blank after any sample with a
  concentration greater than that of the highest
  standard or that might result in carryover from
  one sample
• Acceptance level lt 2 MDL (general used) or lt 5
  of the regulation allowed (EPA Taiwan)

44
5. Calibration with Standards

• As a minimum, measure three different dilutions
  of the standard when an analysis is initiated.
  (same as previously mentioned)
• Subsequently, verify the standard curve daily by
  analyzing one or more standards within the linear
  range, as specified in the individual method.

45
6. Analysis of Duplicates

• When most samples have measurable levels of the
  constituent being determined, analysis of
  duplicate samples is effective for assessing
  precision.
• Analyze duplicates and known additions in
  matrices representative of the samples analyzed
  in the laboratory.
• See Table 1020I for acceptable limits for
  duplicate analyses.

46
7. Control Chart

• Referred to new_niea_pa105
• Precision (range) (???????)
• Mean Chart (?????????) The mean chart for QC
  sample is constructed from the average and
  standard deviation of a specified number of
  measurements of the analytes of interest.
• Accuracy Chart (?????????)The accuracy chart for
  QC sample is constructed from the known spike and
  standard deviation of a specified number of
  measurements of the analytes of interest.

47
Precision (range) Chart(???????)
48
Precision (range) Chart(???????)

• Every 10 samples to analyze a duplicate.
• 15 pairs of duplicates are needed to constructed
  the precision chart
• For every RPD ()

49
Precision (range) Chart(???????)

• Average
• Standard deviations of those 15 pairs
• ?2s and ?3s to represent the warning level and
  control level

50
Precision (range) Chart(???????)
51
Mean Chart (?????????)

• 15 pairs of duplicates are needed to constructed
  the mean chart
• ?2s and ?3s to represent the warning level and
  control level
• S is calculated from calculated value for mean,
  (mean chart) or using percentage if the
  concentration varies (accuracy chart) .

52
Mean Chart (?????????)
53
Mean Chart (?????????)
54
Accuracy Chart (?????????)

• 15 pairs of duplicates are needed to constructed
  the mean chart
• or
• based on ways of calculation
• ?2s and ?3s to represent the warning level and
  control level

55
Accuracy Chart (?????????)
56
Accuracy Chart (?????????)
57
Chart analyses

• Control limitIf one measurement exceeds a CL,
  repeat the analysis immediately. If the repeat
  measurement is within the CL, continue analyses
  if it exceeds the CL, discontinue analyses and
  correct the problem.(????)
• Warning limitIf two out of three successive
  points exceed a WL, analyze another sample. If
  the next point is within the WL, continue
  analyses if the next point exceeds the WL,
  evaluate potential bias and correct the problem.

58
Chart analyses

• Standard deviationIf four out of five successive
  points exceed l.s, or are in decreasing or
  increasing order, analyze another sample. If the
  next point is less than 1s. or changes the order,
  continue analyses otherwise, discontinue
  analyses and correct the problem.(????)
• Trendingif seven successive samples are on the
  same side of the central line, discontinue
  analyses and correct the problem.(????)

59
Example of Chart analyses
60
Possible ways of correction action(example of a
commercial lab)
61
?????????
62
Homework

• Glucose BOD Standards. The data below are 30
  measurements on a standard glucose/glutamate
  mixture that has a theoretical BOD of 200 mg/L.
  Use these data to construct a Mean chart.
• 203 213 223 205 209 200 200 196 201 206 192 206
  185 199 201 206 196 214 189 205 201 226 207 214
  210 207 188 199 198 200

63
Quality assessment(External quality
control)(P.1-12)

• Example To evaluate the performance for a
  commercial laboratory
• Laboratory Check Samples (internal Proficiency)
• Use samples with known amounts of the
  constituent of interest supplied by an outside
  agency or blind additions prepared in-dependently
  within the laboratory to determine recovery
  achieved by an analyst.
• In general, method uncertainty will have been
  established beforehand acceptable recovery falls
  within the established uncertainty. For example,
  if the acceptable range of recovery for a
  substance is 85 to 115, then the analyst is
  expected to achieve a recovery within that range
  on all performance evaluation samples.

64
Quality assessment(External quality control)

• 2. Laboratory inter-comparison samples
• 3. Compliance Audits
• Compliance audits are conducted to evaluate
  whether the laboratory meets the applicable
  requirements of the SOP or consensus method
  claimed as followed by the laboratory.
• A recommended format with a few initial items in
  the check list is shown in Table 1020III.

65
Quality assessment(External quality control)

• 4. Quality System audits
• Quality system audits should be conducted by a
  qualified auditor who is knowledgeable about the
  section or analysis being audited.
• 5. Management review

66
Easier ways to checkChecking Correctness of
Analyses

• Anion-cation balance
• Measured TDS calculated TDS (mg/L)
• Calculated TDS to EC (electric conductivity as
  ?s) ratio (Siemen, S1/ohm-cm)

67
Expression of Results

• ????????????????ppm(10-6,?????)?ppb(10-9,?????)??
  ???????,????????,????
• ????mg/L?µg/L
• ????mg/kg?µg/kg
• ????ppm?mg/Nm3?

68
Significant Figures

• Reporting requirement
• All digits in a reported results are expected to
  be known definitely except for the last digit
• Rounding off (4?6?5????,?5????????)
• Calculation
• Several numbers are multiplied or divided (as
  few significant figures as the numbers are
  present)
• Several numbers are added or subtracted, the
  number that has the fewest decimal places is the
  one to follow

69
Sample collectionGeneral Requirement

• Bake at 450oC all bottles to be used for
  organic-analysis sampling, especially in ?g/L
  level.
• Fill the container full or leave space?(P.1-35)

70
Sample collectionTypes of Samples

• Grab sample ????(for example, influent)
• Composite sample ????
• Samples are preferred to analyze on site DO,
  residual chlorine, soluble sulfide, temperature
  and pH.

71
Sample collection Chain of Custody procedures???

• A. Sample labels
• B. Sample seals
• c. Field log book Record
• all information pertinent to a field survey or
  sampling in a bound log book. As a minimum,
  include the following in the log book purpose of
  sampling location of sampling point name and
  address of field contact producer of material
  being sampled and address, if different from
  location and type of sample.

72
Chain of Custody procedure

• d. Chain-of-custody record
• e. Sample analysis request sheet
• f. Sample delivery to laboratory
• g. Receipt and logging of sample
• h. Assignment of sample for analysis

73
Apparatus Containers

• Borosilicate glass bottle is commonly used.
• Silica and sodium maybe leached from glass but
  not plastic
• Trace metals may absorb onto the walls of glass
  containers
• Use glass containers for all the organics. (VOC,
  pesticide, oil, etc)
• Amber bottle for minimization of photodegradation
• Referred Table 1060I
• To contain or to deliver? (NIEA PA-106, P.2)

74
To contain or to deliver?

• TC (to contain) and TD (to deliver).
• Graduated cylinders (??) and volumetric
  flasks(??), are usually marked with a TC. When
  liquid is poured from a piece of glassware a
  small amount remains behind, clinging to the
  sides of the vessel. 
• Pipets (???) and burets(???), are marked with a
  TD.

75
Sample storage (some keys)

• Follow 1060I
• Purposes To retard chemical and biological
  changes that inevitably continue after sample
  collection.
• Certain cations are subject to loss by adsorption
  or ion exchange with the walls of glass
  containers.
• These include aluminum, cadmium, chromium,
  copper, iron, lead, manganese, silver, and zinc,
  which are best collected in a separate clean
  bottle and acidified with nitric acid to a pH
  below 2.0 to minimize precipitation and
  adsorption on container walls.

76
Sample storage (some keys)

• determine temperature, reduction-oxidation
  potential (ORP) and dissolved gases in situ and
  pH, specific conductance, turbidity, and
  alkalinity immediately after sample collection.
• Zero head-space is important in preservation of
  samples with volatile organic compounds and
  radon. Avoid loss of volatile materials by
  collecting sample in a completely filled
  container.

77
Sample preservation (some keys)

• Follow 1060I
• keep samples as cool as possible without
  freezing.
• Analyze samples as quickly as possible on arrival
  at the laboratory. If immediate analysis is not
  possible, preferably store at 4C.
• Use chemical preservatives only when they do not
  interfere with the analysis being made.

78
Reagent-Grade Water
What is the processes in our laboratory?
79
Reagent water quality guideline
What is the water quality in our laboratory?
80
Sample volume needed

• Referred Table 1060I.
• Most of the time, 2-L sample is needed for most
  physical and chemical analyses.

81
Safe laboratory practice

• Refer 1090 B in P.1-39

82
Supplements for statistics
83
Student-t Distribution
f(t)


?/2
?/2
(???)
0
-tc
t
tc
See any Student-t Distribution Table
84
Student-t vs. Normal Distribution

• 1. Both are symmetric bell-shaped
  distributions.
• 2. Student-t distribution has fatter tails than
  the normal.
• 3. Student-t converges to the normal for
  infinite sample.
• 4. Student-t conditional on degrees of freedom
  (df).
• 5. Normal is a good approximation of Student-t
  for the first few decimal places when df gt 30 or
  so.

85
Probability statements
P( t lt -tc ) P( t gt tc ) ???
P(-tc lt t lt tc ) 1 ? ?
86
Deriving a Confidence Interval
87
Hypothesis Test Procedures

• 1. A null hypothesis, H0.
• 2. An alternative hypothesis, H1.
• 3. A test statistic.
• 4. A rejection region.

88
The Null hypothesis
This hypothesis is a belief we maintain until
proven otherwise
89
The Alternative Hypotheses
90
The Test Statistic
For Z, T or F Test Statistic
If null hypothesis true
If null hypothesis not true t has some other
probability distribution
91
Two-Tailed Test
f(t)
reject
reject


do not reject
?/2
?/2
0
-tc
t
tc
Probability of a Type I error ?
92
One-Tailed Test
f(t)
reject


do not reject
?
0
t
tc
Note ignore left tail
Probability of a Type I error ?
93
Rejection Rules

• 1. Two-Sided Test If the value of the test
  statistic falls in the critical region in either
  tail of the t-distribution, then we reject the
  null hypothesis in favor of the alternative.
  Otherwise, we do not reject the null hypothesis.
• 2. Left-Tail TestIf the value of the test
  statistic falls in the critical region which lies
  in the left tail of the t-distribution, then we
  reject the null hypothesis in favor of the
  alternative. Otherwise, we do not reject the null
  hypothesis
• 3. Right-Tail TestIf the value of the test
  statistic falls in the critical region which lies
  in the right tail of the t-distribution, then we
  reject the null hypothesis in favor of the
  alternative. Otherwise, we do not reject the null
  hypothesis

94
Type I and Type II errors

• Type I errorWe make the mistake of rejecting
  the null hypothesis when it is true.?
  P(rejecting H0 when it is true).
• Type II errorWe make the mistake of failing to
  reject the null hypothesis when it is false.?
  P(failing to reject H0 when it is false).

95
Failure to Reject does not mean Hypothesis is
True!
When we fail to reject the null hypothesis, that
does not mean that we can conclude that the null
hypothesis is true.
Failure to reject the null hypothesis is a rather
weak conclusion since it only means that the data
are compatible with the null hypothesis.
If the null hypothesis ?20 is not rejected, then
the null hypotheses ?20.1, ?2-.2, etc. may well
be compatible with the data too, so tests of
these hypotheses would not be rejected either.
96
Format for Hypothesis Testing

• 1. Determine null and alternative hypotheses.
• 2. Specify the test statistic and its
  distribution as if the null hypothesis were
  true.
• 3. Select ? and determine the rejection region.
• 4. Calculate the sample value of test
  statistic.
• 5. State your conclusion.