Statistical Quality
Control in
clinical Laboratories
What is Quality Control?
Quality control in the medical laboratory is a statistical
process used to monitor and evaluate the analytical
process that produces patient results.
SQC is process that minimizes the variability of a
procedure” although it would be wiser to define SQC as
“The process that focuses on revealing any deviations
from well defined standards.
SQC can be used in every automated production, like the
laboratory determinations which are performed by
biomedical analyzers
When a diagnostic test is performed in the medical laboratory,
the outcome of the test is a result.
The result may be a patient result or it may be a quality
control (QC) result.
The result may be quantitative (a number) or qualitative
(positive or negative) or semi-quantitative (limited to a few
different values)
Internal and External SQC
Random and Systematic errors must be detected at an early stage and then every
effort should be taken in order to minimize them. The strategy for their detection
consists of specific SQC methods which are divided in two categories:
Internal Quality Control (IQC). It concludes all SQC methods which are performed every
day by the laboratory personnel with the laboratory’s materials and equipment. It checks
primarily the precision (repeatability or reproducibility) of the method.
External Quality Control (EQC). It concludes all SQC methods which are performed
periodically (i.e. every month, every two months, twice a year) by the laboratory personnel
with the contribution of an external center (referral laboratory, scientific associations,
diagnostic industry etc.). It checks primarily the accuracy of the laboratory’s analytical
methods.
Quality Control Products
Control materials (or simply “controls”) are all the materials which can be used for
error
detection in SQC methods. Although this term is considered equal to “control
samples”,
several SQC methods have been deployed based on patients’ results.
Control samples are pools of biological fluids (serum, whole blood, urine or other
materials). They contain analytes which are determined by the laboratory, ideally in
concentrations/activities close to the decision limits where medical action is required.
In internal and external SQC, it is common practice that laboratories use two or three
different control samples which contain different quantities of analytes e.g. low,
normal, high concentrations/activities.
Control samples with the same analytes but different concentrations/activities are
called “levels”. Different levels check the performance of laboratory methods across all
their measuring range. In most cases control samples are manufactured by analyzers’
or reagents’ manufacturers, but they can also be made by the laboratory personnel.
A normal control product contains normal levels for the analytes
being tested.
An abnormal control product contains the analytes at a
concentration above or below the normal range for the analytes.
For example, the normal range for a potassium level is about 3.5 –
5.0 mmol/L. A normal control would contain potassium at a level
within this range. An abnormal control would contain potassium at
a level below 3.5 mmol/L or above
5.0 mmol/L.
Regular Testing
Good laboratory practice requires testing normal and abnormal
controls for each test at least daily to monitor the analytical
process.
If the test is stable for less than 24 hours or some change has
occurred which could potentially affect the test stability, controls
should be assayed more frequently.
Regular testing of quality control products creates a QC database
that the laboratory uses to validate the test system. Validation
occurs by comparing daily QC results to a laboratory-defined range
of QC values.
The lab-defined range is calculated from QC data collected from
testing of normal and abnormal controls.
Quality Control Results
In Table 1, there are two ranges reported.
The acceptable range for the Level I (Normal Control) is 3.7 – 4.3
mmol/L.
The range for Level II (Abnormal Control) is 6.7 – 7.3 mmol/L.
When the daily QC result obtained for the normal control is compared
to the range calculated for the normal control, it becomes apparent
that each result lies somewhere within the expected range. This
indicates that the analytical process is “in control” at the normal level
on that day of testing.
When the daily QC result for the abnormal control (high potassium) is
compared to the defined range for the abnormal control, the
analytical process is shown to be “in control” for each day of testing
except for the last day (11/7).
On November 1 through November 6, both controls were “in control”
and patient values could be reliably reported.
However, the laboratory was “out of control” for abnormal high
potassiums on November 7 because the value obtained for the
QC material (8.0 mmol/L) was outside the acceptable range (6.7 –
7.3 mmol/L).
This means that some error occurred which may have made some
patient results unreliable. The laboratory should not report any
patient samples with an abnormally high potassium result until
the error is resolved and the abnormally high sample(s) are re-
tested.
Levey-Jennings Charts & Westgard Rules
Levey-Jennings Charts
The Levey-Jennings chart is used to graph successive
(run-to-run or day-to-day) quality control values.
Calculating a Mean [x]
The mean (or average) is the laboratory’s best estimate of the
analyte’s true value for a specific level of control.
To calculate a mean for a specific
level of control, first, add all the
values collected for that control.
Then divide the sum of these
values by the total number of
values.
For instance, to calculate the mean for the normal control (Level I)
in Table 1, find the sum of the data {4.0, 4.1, 4.0, 4.2, 4.1, 4.1, 4.2}.
The sum [] is 28.7 mmol/L.
The number of values is 7 (n = 7). Therefore, the mean for the
normal potassium control in Table 1 from November 1–7 is 4.1
mmol/L (or 28.7 mmol/L divided by 7).
Calculating a Standard Deviation [s]
Standard deviation is a statistic that quantifies how close
numerical values (i.e., QC values) are in relation to each other.
To calculate the standard deviation for the normal level of control (Level I) in
Table 1,
begin by calculating the mean [ x ]:
x = 4.0 + 4.1 + 4.0 + 4.2 + 4.1 + 4.1 + 4.2 mmol/L ÷ 7
x = 28.7 mmol/L ÷ 7
x = 4.1 mmol/L
Standard deviation is commonly used for preparing
Levey-Jennings (L-J or LJ) charts.
A chart is created for each test and level of control.
The first step is to calculate decision limits.
These limits are ±1s, ±2s and ±3s from the mean.
The mean for the Level I potassium control in Table 1 is 4.1 mmol/L
and the standard deviation is 0.1 mmol/L.8 Formula 3 provides
examples on how ±1s, ±2s and ±3s quality control limits are
calculated.
Types of laboratory errors and mistakes
In laboratory practice many non-conforming results may appear. These results
are divided in two major categories:
• Errors: Non-conforming results with “statistical meaning”. This category includes
all the “wrong” laboratory measures due to non-human action.
• Mistakes: Non-conforming results with “no statistical meaning”. This category
contains all the human errors e.g. mixing up samples.
Another classification of errors and mistakes is based on the time and the stage
they appeared in laboratory practice.
1. The pre-analytical stage encompasses all the procedures which are take place
before the analysis of the patients’ samples on the automated analyzers (e.g.
blood drawing, sample transportation, centrifugation, dilutions etc).
2. The analytical stage includes the analytical methods
3. The post-analytical stage refers to transmission of data from analyzers to the
LIS, validation of results that have been produced and posting of the results to
physicians or patients.
Common analytical errors
1. Expired reagents which may lead to erroneous results.
2. Expired controls or calibrators.
3. Calibration curve time-out elapsed.
4. Failure in sampling system.
5. Failure in aspiration system of reagents.
6. Changes in analyzer’s photometric unit / flow cell /
measuring unit.
7. Any other analyzer’s failure Common analytical errors.
Important definitions
Accuracy refers to the closeness of a measured value to a
standard or known value.
Precision refers to the closeness of two or more
measurements to each other.
Types of analytical errors
Systematic Error
Systematic error is evidenced by a change in the mean of the
control values.
The change in the mean may be gradual and demonstrated as
a trend in control values or it may be abrupt and
demonstrated as a shift in control values.
Trend Shift
A trend indicates a gradual loss of reliability in Abrupt changes in the control mean are
the test system. Trends are usually subtle. defined as shifts. Shifts in QC data
Causes of trending may include: represent a sudden and dramatic positive
or negative change in test system
performance. Shifts may be caused by:
▪ Deterioration of the instrument light ▪ Sudden failure or change in the light
source. source.
▪ Gradual accumulation of debris in ▪ Change in reagent formulation.
sample/reagent tubing. ▪ Change of reagent lot.
▪ Gradual accumulation of debris on ▪ Major instrument maintenance.
▪ electrode surfaces. ▪ Sudden change in incubation
▪ Aging of reagents. temperature (enzymes only).
▪ Gradual deterioration of control materials. ▪ Change in room temperature or
▪ Gradual deterioration of incubation humidity.
chamber temperature (enzymes only). ▪ Failure in the sampling system.
▪ Gradual deterioration of light filter ▪ Failure in reagent dispense system.
integrity. ▪ Inaccurate calibration/recalibration.
▪ Gradual deterioration of calibration.
Westgard Rules
Six basic rules are made by James westgard in 1981.
These rules are used individually or in combination to evaluate
the quality of analytical runs.
12s
RULE
12s.
One control value lies between μ+2s/μ+3s or
between μ-2s/μ-3s. It is only a warning rule.
13s
RULE 13s.
One control value lies over μ+3s or under μ-3s. This criterion
is sensitive to the detection of random errors. The results
should be blocked and not reported to the patients. The run6
is rejected.
22s 22s.
RULE
Two successive control values lie between +2s and +3s or
between -2s and -3s. It defines a systematic error. The results
should be blocked and not reported to the patients.
R4s
RULE R4s.
The distance of two successive control values, values, is over
4s. It is a criterion sensitive to random errors. The results
should be blocked and not reported to the patients
41s
RULE 41s.
Four successive control values lie between μ+1s and μ+3s or
between μ-1s and μ-3s. It defines a systematic error. The
results should be blocked and not reported to the patients.
10x
RULE 10x
reject when 10 consecutive control measurements
fall on one side of the mean