50
Slope = 1
Quality Control 40
For Clinical Laboratories
Measured value
30 Slope = 2
20 Slope = 0.5
ANALYTICAL ERRORS 10
Slope ≠ 1.0
y-intercept = 0
0
0 10 20 30 40 50
Actual value
Dr. Tamer Soliman
Associate Professor of Clinical and Chemical Pathology
3
Analytical Errors
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Analytical Errors
Errors that occur during the testing (analytical) process.
Error = and deviation from the true value.
Random Error Systematic Error
(Imprecision) (Bias)
Constant Proportional
An error which is the same An error which is proportional to
regardless of the analyte’s concentration the analyte’s concentration.
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Random Systematic
Error Error
Imprecision Bias
Characteristics Characteristics
• No consistent pattern or direction. • Follow a consistent pattern and direction.
• Unpredictable. • Predictable.
• Difficult to detect and eliminate. • Easier to detect and correct
• Could be minimized by test repetition and • Cannot be minimized by test repetition and
averaging the values of obtained results. averaging (all the data are wrong in the same
direction).
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Random Systematic
Error Error
Imprecision Bias
Estimation Estimation
• In the term of imprecision • In the term of bias
• which is the amount of deviation of results away which is difference between the average of
from the center (mean of measurements). measurements and the true (reference) value
obtained from control or reference sample.
• RE is expressed as: • SE is expressed as:
a. Standard deviation; SD a. Bias = Mean of values – True value
b. Coefficient of variation; CV% b. Bias% = (Bias/true value)×100
• Bias may be positive (+) or negative (–)
according to the direction of SE.
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Random Systematic
Error Error
Imprecision Bias
Causes Causes
• Operator-related causes • Reagent-related causes
• Instrument-related causes • Calibration-related causes
• Environmental causes (Fluctuations) • Operator-related causes
• Instrument-related causes
• Environmental causes (Permanent changes)
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
No Errors
When performing a linear regression analysis, where: 50
the x-axis represents the measured values and Perfect Regression line
the y-axis represents the true values, Slope = 1
40 y-intercept = 0
1. Assuming there are no errors,
Measured value
the perfect regression line that represents the relation 30
between the measures and actual values has:
20
• Slope of 1.0
• y-intercept of 0
10
0
0 10 20 30 40 50
Actual value
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Random Error
When performing a linear regression analysis, where: 50
the x-axis represents the measured values and Random distribution
the y-axis represents the true values, Around the regression line
40
2. Random errors are represented by data points that
Measured value
are scattered randomly around the regression line, 30
and they do not have a consistent direction.
20
10
0
0 10 20 30 40 50
Actual value
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Constant Systematic Error
When performing a linear regression analysis, where: 50
the x-axis represents the measured values and
the y-axis represents the true values,
40
3. Constant systematic errors change the y-intercept
Measured value
(y-intercept ≠ 0), but not the slope, of the regression 30
line. (moving the regression line up and down)
• Slope = 1.0 20
• y-intercept ≠ 0
10
Slope = 1.0
y-intercept ≠ 0
0
0 10 20 30 40 50
-10 Actual value
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Proportional Systematic Error
When performing a linear regression analysis, where: 50
the x-axis represents the measured values and Slope = 1
the y-axis represents the true values,
40
4. Proportional systematic errors change the slope
Measured value
(slope ≠ 1), but not the y-intercept, of the regression 30 Slope = 2
line.
• Slope ≠ 1.0 20 Slope = 0.5
• y-intercept = 0
10
Slope ≠ 1.0
y-intercept = 0
0
0 10 20 30 40 50
Actual value
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Combined Systematic Error
When performing a linear regression analysis, where: 50
the x-axis represents the measured values and
the y-axis represents the true values,
40
Slope = 1.5
5. Combined systematic errors change both the slope y-intercept = 10
Measured value
(slope ≠1), and the y-intercept (y-intercept ≠ 0) of the 30
regression line.
20
• Slope ≠ 1.0
• y-intercept ≠ 0
10
0
0 10 20 30 40 50
Actual value
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Error
(Total Analytical Error)
• Represents the overall variation of mean of measurements from the true value
• Used as a method to evaluate the method performance.
• Calculated by the sum of SE (bias) and RE (imprecision).
Expressed in Multiplied by
absolute number, | Bias | the Z factor [or coverage factor (k)]
which is usually 1.65 to cover 90% of the RE
• TE calculation : TE = | Bias | + 1.65 × SD
%TE = | %Bias | + 1.65 × %CV
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Error
(Total Analytical Error)
Number of replicates
X Measured value
(Mean)
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Error
(Total Analytical Error)
Number of replicates
-2 SD -1 SD X 1 SD 2 SD Measured value
(Mean)
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Error
(Total Analytical Error)
Number of replicates
-2 SD -1 SD X 1 SD 2 SD Measured value
(Mean)
5% 90% 5%
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Error
(Total Analytical Error)
Number of replicates
-2 SD -1 SD X 1 SD 2 SD Measured value
(Mean)
-1.65 SD 1.65 SD
5% 90% 5%
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Error
(Total Analytical Error)
1.65 SD
RE
(Imprecision)
Number of replicates
-2 SD -1 SD X 1 SD 2 SD Measured value
(Mean)
-1.65 SD 1.65 SD
90% 5%
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Error
(Total Analytical Error)
1.65 SD
SE RE
(Bias) (Imprecision)
Number of replicates
Target -2 SD -1 SD X 1 SD 2 SD Measured value
value -1.65 SD
(Mean)
1.65 SD
90% 5%
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Error
(Total Analytical Error)
Total Error
1.65 SD
SE RE
(Bias) (Imprecision)
Number of replicates
Target -2 SD -1 SD X 1 SD 2 SD Measured value
value -1.65 SD
(Mean)
1.65 SD
90% 5%
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Allowable Error
TEa
• Refers to the maximum amount of error— both imprecision and bias combined—
that can be allowed for an assay, beyond which the analytical result is considered
invalid.
• Expressed as a percentage of the true value.
• TEa for analytes are used by the laboratories as quality goals for:
a. Evaluating a new methodology or new reagent lot for patient testing.
b. Setting QC limits for troubleshooting QC result (TE should be < TEa).
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Allowable Error
TEa
How TEa is determined?
In 1999, a conference held in Stockholm, Sweden, reached a consensus agreement that recommended a
hierarchy of 5 models for quality goals:
1. Evaluation of the effect of analytical performance on clinical outcomes.
2. Evaluation of the effect of analytical performance on clinical decision levels:
• Using data based on biological variation (within- or between-subject variations).
• Using data based on analysis of clinicians’ opinions.
3. Published professional recommendations:
• From national and international expert bodies.
• From expert local groups or individuals.
4. Performance goals set by Regulatory bodies or Organizers of EQA schemes.
5. Goals based on the current state of the art: using current data from EQA or proficiency testing
schemes and current publications on methodology to determine the analytical performance for
the specific test.
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Allowable Error
TEa
Desirable Biological Variation Database specifications
[Link]
2019 – 2024 CLIA Proposed Acceptance Limits for Proficiency Testing
[Link]
CLIA Acceptable Test Performance Criteria
[Link]
QUALITY CONTROL – Laboratory Analytical Errors Dr. Tamer Soliman
Total Allowable Error
TEa
CVI within-subject biologic variation
CVG between-subject biologic variation
I desirable specification for imprecision
B desirable specification for inaccuracy
TE desirable specification for allowable total error
Number Biological Variation Desirable specification
Analyte
of papers CVI CVg I(%) B(%) TE(%)
S- Albumin 24 3.20 4.75 1.6 1.43 4.07
U- Albumin, concentration, first morning 3 36.0 55.0 18.0 16.4 46.1
S- Aldosterone 2 29.4 40.1 14.7 12.4 36.7
U- Aldosterone 1 39.4 40.1 19.7 14.05 46.56
S- Alkaline phosphatase 22 6.45 26.1 3.23 6.72 12.04