Definition
Quality Control (QC) is part of quality
management focused on fulfilling quality
requirements
QC is examining “control” materials of known
substances along with patient samples to
monitor the accuracy and precision of the
complete examination (analytic) process.
1
Purpose
The goal of QC is to detect errors and
correct them before patients’ results are
reported
2
Types of errors
Preanalytical Errors occur prior to specimen testing and may
include variables involving the process of obtaining a
specimen.
Analytical Errors occur during actual testing of the specimen.
Performance of tests in the laboratory is rigorously controlled,
with quality control procedures in place that markedly reduce
errors in the analytic phase of testing.
Postanalytical Errors occur after a test result is generated. In
general, such postanalytical errors occur with entry,
manipulation, and reporting of test data. If results are written
by hand or entered via keyboard, then an entry error may
occur.
3
4
5
Quantitative Examinations
Measure the quantity of a particular
substance in a sample
Measurements should be both accurate and
precise
6
Qualitative Examination Methods
Examinations that do not have numerical
results:
growth or no growth
positive or negative
reactive or non-reactive
color change
7
Semi-quantitative Examination Methods
Results are expressed as an estimate of the
measured substance:
“trace amount”, “moderate amount,” or “1+, 2+, or 3+”
number of cells per microscopic field
titters and dilutions in serologic tests
8
Quality control
important part of quality management system
goal is to identify errors and eliminate them before
reporting patient results
different methods applied for quantitative, qualitative,
and semi-quantitative results
11
Quantitative Tests
measure the quantity of a particular substance
in a sample
quality control for quantitative tests is
designed to assure that patient results are:
accurate
reliable
12
What is a Control?
material that contains the substance being
analyzed
include with patient samples when performing a
test
used to validate reliability of the test system
run after calibrating the instrument
run periodically during testing
13
Calibrators vs. Controls
14
Calibrators Controls
A substance with a specific A substance similar to
concentration. patients’ samples that
has an established
Calibrators are used to set concentration.
(calibrate) the measuring
points on a scale. Controls are used to ensure
the procedure is working
1 2 3 4 5 properly.
1 2 3 4 5
15
Characteristics of Control Materials
appropriate for the diagnostic sample
values cover medical decision points
similar to test sample
available in large quantity ideally enough
for one year
can store in small aliquots
16
Types of Control Materials
may be frozen, freeze-dried, or
chemically preserved
requires very accurate
reconstitution if this step is
necessary
17
Sources of Controls Materials
commercially prepared
made “in house”
obtained from another laboratory, usually
central or reference laboratory
18
Control Materials
Target value predetermined
ASSAYED Verify and use
Target value not predetermined
UNASSAYED
Full assay required before using
In-house pooled sera
“IN-HOUSE”
Full assay, validation
19
Choosing Control Materials
values cover medical decision points
similar to the test sample
controls are usually available in high, normal, and low ranges
20
Preparation and Storage of
Control Material
adhere to manufacturer’s
instructions
keep adequate amount
of same lot number
store correctly
21
Steps in Implementing Quantitative QC
obtain control material
run each control 20
times over 30 days
3SD
calculate mean and +/-1,2,3
Standard Deviations 2SD
1SD
Mean
1SD
2SD
3SD 22
Measurement of Variability
Variability is a normal occurrence when a control
is tested repeatedly
Affected by:
Performance
Operator Environmental characteristics of
technique conditions the measurement
The goal is to differentiate between variability
due to chance from that due to error
23
Measures of Central Tendency
Although variable, sets of data are distributed
around a central value
F
r
e
q
u
e
n
c
y
Measurement
24
Measures of Central Tendency
Mode the value which occurs with the
greatest frequency
Median the value at the center or
midpoint of the observations
Mean the calculated average of the
values
25
Not all central values are the same
Mean Mode
F Median
r
e
q
u
e
n
c
y
Measurement
26
Symbols Used in Calculations
∑ is the sum of (add data points)
n = number of data points
x1 - xn = all of the measurements
(1 through n)
__
X represents the mean
27
Calculation of Mean
X 1 X 2 X 3 ... X n
X
n
X = Mean
X1 = First measurement
X2 = Second measurement
Xn = Last measurement in series
n = Total number of measurements
28
Example
Calculation of Mean: ELISA Tests
Run controls 20 times in 30 days. Record both OD and
cut off (CO) values for each measurement.
Divide the OD by the CO (OD/CO) for each data point
or observation. This standardizes the data.
Add the ratios and divide by the number of
measurements to get the mean.
29
Data showing outlier
1. 192 mg/dL 11. 204 mg/dL
2. 194 mg/dL 12. 208 mg/dL
3. 196 mg/dL 13. 212 mg/dL
4. 196 mg/dL 14. 198 mg/dL
5. 185 mg/dL 15. 204 mg/dL
6. 196 mg/dL 16. 208 mg/dL
7. 200 mg/dL 17. 212 mg/dL
8. 200 mg/dL 18. 198 mg/dL
9. 202 mg/dL 19. 192 mg/dL
10. 270 mg/dL 20. 196 mg/dL
30
Normal distribution
all values symmetrically distributed around
the mean
characteristic “bell-shaped” curve
assumed for all quality control statistics
Frequency
mean
31
Quality Control is used to monitor
the accuracy and the precision
of the assay.
What are
accuracy and
precision?
32
Definitions
Accuracy The closeness of
measurements to the true
value
Precision The amount of variation in
the measurements
Bias The difference between the
expectation of a test result
and an accepted reference
value
33
Accuracy and Precision
Accurate Precise
and Precise but Biased Imprecise
Accurate = Precise but not Biased
34
Standard Deviation and Probability
For a set of data with a normal
distribution, a random
Frequency
measurement will fall within:
68.2%
+ 1 SD 68.3% of the time
+ 2 SD 95.5% of the time 95.5%
99.7%
+ 3 SD 99.7% of the time
-3s - 2s -1s Mean +1s
+2s +3s
35
Standard Deviation (SD)
SD is the principle measure of
variability used in the laboratory
2
(x1 x )
SD n 1
Standard Deviation – Statistical Formula
36
Coefficient of Variation
The coefficient of variation (CV) is the SD
expressed as a percentage of the mean.
SD
CV x 100 %
mean
CV is used to monitor precision
CV is used to compare methods
CV ideally should be less than 5%
37
Levey-Jennings Chart
Graphically Representing Control
Ranges
38
Statistics for Quantitative QC
assay control material at least 20
data points over a 20–30-day period
ensure procedural variation is
represented
calculate mean and + 1, 2 and 3 SD
39
Draw lines for Mean and SDs
(calculated from 20 controls)
196.5 +3SD
194.5 +2SD
192.5
+1SD
190.5 MEAN
188.5 -1SD
186.5 -2SD
184.6
-3SD
Days
40
Levey-Jennings Chart
Plot daily control measurements
196.5 +3SD
194.5 +2SD
192.5 +1SD
190.5 MEAN
188.5 -1SD
186.5 -2SD
184.6 -3SD
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Days
41
Number of Controls
Interpretation depends on number of controls run
with patients’ samples.
Good: If one control:
accept results if control is within ± 2SD
unless shift or trend
Better: If 2 levels of controls
apply Westgard multirule system
42
Detecting error
random error: variation in QC results with
no pattern- only a cause for rejection if
outside 2SDs.
systematic error: not acceptable, correct the
source of error
Examples:
shift–control on one side of the mean 6
consecutive days
trend–control moving in one direction– heading
toward an “out of control” value
43
D oetcin n
D g ’tR
ero ep
r ea tco n tro lId en tify th eP ro b lem
Random error: variation in QC results with no pattern-
12s
R4s
Imprecision : large amount of scatter about the
mean. Usually caused by errors in technique
Systematic error: Gradual change in the mean of
the control values
41s
10x
Inaccuracy = may see as a trend or a shift, usually
caused by change in the testing process
Examples:
Shift
Trend
Levey-Jennings Chart
Shift
196.5 +3SD
194.5 +2SD
192.5
+1SD
190.5 MEAN
188.5 -1SD
186.5 -2SD
184.6
-3SD
Days
46
Shift
Operator error
Failure to follow manufacturer’s instructions
An outdated procedure manual
Equipment failure
Calibration error
Sudden failure or change in light source
Change in reagent formulation
Change in reagent lot
Major instrument maintenance
Change in room temperature or humidity
Failure in sampling system
47
Levey-Jennings Chart
Trend
196.5 +3SD
194.5 +2SD
192.5
+1SD
190.5 MEAN
188.5 -1SD
186.5 -2SD
184.6
-3SD
Days
48
49
Westgard Multirule System
a “multi-rule” system developed by
Dr. James O. Westgard based on statistical
concepts
a combination of decision criteria or
rules to assess if a system is in control
use when at least 2 levels of control
are run with the examination run
cannot use with only one control
Dr. Westgard
50
Westgard Multirule System Titles
1
2S rule
Used when 2 levels of
control material are
analyzed per run.
1 rule
3S
2 rule
2S
R rule
4S
4 rule
1S
10 rule
X
51
Westgard – 12S Rule
WARNING RULE – not cause for rejecting a
run
One of two control results falls outside ±2SD
Alerts technologist to possible problems
Must then evaluate the 1 rule
3S
52
53
Westgard – 22S Rule
Two consecutive control values for the same
level fall outside of ±2SD in the same
direction, or
Both controls in the same run exceed ±2SD.
Patient results cannot be reported.
Requires corrective action
54
55
Westgard – 13S Rule
This rule identifies unacceptable random error
or possibly the beginning of a large systematic
error.
Any QC result outside ±3s violates this rule.
One measurement exceeds 3 standard
deviations either above or below the mean of
the reference range.
Indication: Inaccuracy and/or imprecision
56
57
Westgard – R4S Rule
One control exceeds the mean by –2SD, and
the other control exceeds the mean by +2SD.
The range between the two results will
therefore exceed 4 SD.
Random error has occurred, test run must be
rejected.
58
59
Westgard – 41S Rule
Rejection
four consecutive QC results for one level of
control are outside ±1SD, or
both levels of control have consecutive
results that are outside ±1SD.
60
61
Westgard – 10X Rule
requires control data from
previous runs
ten consecutive QC results for one level of
control are on one side of the mean, or
both levels of control have five consecutive
results that are on the same side of the mean
62
63
Westgard Multirule QC
64
[Link]
Activity
Using the Westgard Multirule
System, describe the control
values for each day in the
following slide.
65
66
Answers for activity
Day 21, 22, 24, 26, 27, 30, 31, 33, 34, 36-44 – in
control
Day 23, 28, 29 – 12s
Day 25 - 13s
Day 32 – 22s
Day 35 - R4s
67
Measurement Uncertainty
represents a range of values in which the true value is
reasonably expected to lie
is estimated at “95% coverage”
the more precise the method, the smaller the range of
values that will fall within 95%
for most instances, a range of + or - 2 SDs is accepted
as measurement uncertainty that is explained by
random variation
68
If QC is out of control
STOP testing
identify and correct problem
repeat testing on patient
samples and controls after correction
Do not report patient results until
problem is solved and controls
indicate
proper performance
69
Solving out-of-control problems
identify problem
refer to established
policies and procedures
for corrective action
70
Possible Problems
degradation of reagents or kits
control material degradation
operator error
failure to follow manufacturer’s instructions
an outdated procedure manual
equipment failure
calibration error
71
Summary
A quality control program for quantitative tests is essential. It
should:
monitor all quantitative tests
have written policies and procedures, followed by laboratory
staff
have a quality manager for monitoring and reviewing QC data
use statistical analysis, provide for good records
provide for troubleshooting and corrective action
72
Key Messages
A QC program allows the laboratory to differentiate
between normal variation and error.
The QC program monitors the accuracy and precision
of laboratory assays.
The results of patient testing should never be released
if the QC results for the test run do not meet the
laboratory target values.
73
Questions?
74
