Transcript ABC of Quality Control - A Problem Based Approach

ABC of Quality Control
A problem based approach
RT ERASMUS
NHLS / FACULTY OF HEALTH SCIENCES,
UNIVERSITY OF STELLENBOSCH
20th July, 2007, Bela Bela
1.What is quality control?
It is a statistical process used to monitor and evaluate the analytical process
which produces patient results. The statistical process requires regular testing
of quality control products along with patient samples comparison of quality
control results to specific statistical limit (ranges).QC results are used to validate
patient results and requires testing normal and abnormal controls for each test at
least daily to monitor the analytical process.
Regular QC testing creates a QC database. Validation occurs by comparing daily
QC results to a laboratory defined range of QC values.
Calculate the control mean values of the following:
Normal control
Test : creatine kinase
Control values are:
94, 93, 97, 95, 95, 100, 99, 100, 99, 100
What does the mean value for these control values reflect ?
2.The mean and standard deviation are the most fundamental statistics used
by the laboratory. The QC statistics for each test performed in the laboratory
are calculated from the QC database collected by regular testing of control
products.
Consequently the mean level of the control reflects the behaviour of the test
at that specific concentration.
3. Calculate mean and standard deviation of the following control
values obtained for potassium over several days
4.0
4.1
4.0
4.2
4.1
4.1
4.2
What do you understand by the results obtained ?
What do you understand by the term “standard deviation” ?
What is the standard deviation used for?
S.D S=
∑ (xn - х) 2
S
√∑ (xn- х) 2
n-1
= the sum of squares of differences between individual QC
values and the mean
n
=
number of values in the data set
х
=
28.7 mmol/L ÷ 7 = 4.1
_______________________________________________
=√ (4.0 – 4.1)2 + (4.1 – 4.1)2 + (4.0 – 4.1)2 + (4.2 – 4.1)2 + (4.1 – 4.1)2 + (4.1
– 4.1)2 + (4.2 – 4.1)2
__________________________________________
6
S
=
S
√ (– 0.1)2 + (0.0)2 + (-0.1)2 + (0.1)2 + (0)2 + (0)2 + (0.1)2
___________________________
6
__________
=
√ (0.04 ÷ 6)
=
0.082 or 0.1
This type of S.D is called between run standard deviation b/c data used to
calculate the statistics came from different analytical runs.
Standard deviation indicates the dispersion or spread of values. The
bigger the value, the greater the imprecision.Standard deviation can be used
for
(A) evaluating method performance (i.e. b/w run and within run precision)
(B) instrument evaluations (C) comparing performance with other
laboratories using the same instrument (D) monitor on going day to day
performance
Create a Levey-Jennings for the following control values obtained on
each day for serum creatine kinase.
327, 325, 321, 323, 315, 308, 304, 298, 327, 334
The mean value of the control is 350 U/L with a standard deviation of 25
U/L.
S.D is commonly used for preparing a Levey-Jennings Chart. The chart is
used to graph successive day to day quality control values.
Step 1 :
Step 2 :
Calculate decision limits. These limits are + 1s, + 2s, + 3s
from the mean
The control limits of serum creatine kinase are
+ 1s
= 350 – 25 = 325
350 + 25 = 375
+ 2s =
350 – (25) (2) = 300
350 + (25) (2) = 400
+ 3 s = 350 – (25) (3) = 275
350 + (25) (3) = 425
When an analytical process is within control, approximately 68% of all QC
values fall within + 1 standard deviation. Likewise 95.5% of all QC values fall
within +2 standard deviations (2s) of the mean. About 4.5% of all data will be
outside the +2s limits when the analytical process is in control. Approximately
99.7% of all QC values are found to be within +3 standard deviations of the
mean. As only 0.3% or 3 out of 1000 points will fall outside the +3s limits any
value outside of +3s is considered to be associated with a significant error
condition and patient results should not be reported.
Some laboratories incorrectly consider any quality control value outside its
+2s limits to be out of control, as approximately 4.5% of all valid QC values
will fall somewhere between +2 and +3 standard deviations limits.
Laboratories using +2s control limits frequently reject good runs.
5. How do you use the Levy-Jennings Chart to evaluate run quality?
Once the QC data have been plotted on the chart search for
(a)systematic error
(b)random error
Systematic error is evidenced by a change in the mean of the control value.
The change may be gradual and seen as a trend, or may be abrupt and
seen as a shift.
Trend
Usually subtle. Causes are weakening light source
gradual accumulation of debris in reagent tubing or on electrode surface
aging of reagents gradual breakdown of control materials
gradual loss of temperature control
gradual loss of filter integrity
Shifts
Abrupt changes in control mean.
Signifies a sudden positive or negative change in test system performance.
May be caused by
sudden failure of light source
change in reagent lot / formulation
sudden change in incubation temperature or room temperature/ humidity
failure in reagent dispensing or sampling
inaccurate recalibration
Random Errors :
Any +ve or –ve deviation away from calculated mean can be classified as
acceptable or unacceptable. Unacceptable random error is when any data
point is outside +3S.D.
WESTGARD RULES
These are quality control rules developed by Dr James Westgard in 1981.
Most rules are expressed as NL, where N represents the number of control
observations. Thus 13s represents a control rule which is violated when one
control observation exceeds the +3s control limits.
12s
Warning rule (single observation outside +2s)
Warns that random or systematic error may be present
Examine this with other previous values
If no source of error found, it is assumed that it is an acceptable random error.
Violation of following rules results in rejection of entire run.
13s :Identifies unacceptable random error or beginning of a large
systematic error.
22s :Identifies systematic error only.Two consecutive QC results
greater than 2s on the same side of the mean
R4s Identifies random error only and is applied only within the current run
Implies a 4s difference between control values within a single run.
e.g. in a run, 2 controls (low and high) are used.
Level 1 is + 2.8s above mean and level 2 is –1.3.
Total difference is 4.1s.
Violation of following rules does not necessarily require rejection of
analytical run. They usually identify smaller systematic error, which is often
not clinically significant. Can be eliminated by instrument maintenance or
calibration.
3ls : 3 consecutive results greater than ls on the same side as the mean
Two applications to 3ls and 4ls rule.
These are within control material or across control materials.
6. (a) What do you understand by C.V
(b) How does it differ from S.D
(c) What function does it serve in
laboratory statistics
6.Coefficient of Variation (C.V)
Is the ratio of standard deviation to the mean and
is expressed as a percentage. Allows for easier
comparison of the overall precision.
Because S.D increases with analyte
concentration, C.V is regarded as a statistical
equalizer.
C.V can also be used when comparing instrument
performance. In practice CV is used for method
comparison over different levels of controls which
may be used to identify method performance over
varying analyte ranges.