Probability comparison
13-English version
Probability comparison of obtaining a difference in less than 10% regarding an analyte homeostatic condition, between two procedures, with an specific biologic variability: CVI%
Procedure A:
Result from a single test, performed on a single sample, obtained with an exact procedure, without error.
Procedure B:
Average between two results, from two samples withdrawn within a time period and obtained with a procedures which analytical variation is defined by the variation coefficient CVA% = 5%.
The probability calculation results are shown in the table below:
Where:
% CVB0% :
Biochemical variation coefficient percent when calculated with an analytical coefficient percent : CVA % = 0%
Biochemical variation coefficient percent when calculated with an analytical coefficient percent : CVA % = 0%
· % CVB0% = (CVI 2 + 0 2)0.5 = SE% of the measurement with a
single sample
single sample
(SE% : Standard error percent)
%CVB5%/ 20.5:
Biochemical variation coefficient percent for two (2) samples , when
calculated with an anlytical coefficient percent: CVA% = 5%
Biochemical variation coefficient percent for two (2) samples , when
calculated with an anlytical coefficient percent: CVA% = 5%
· %CVB5%/ 20.5 = (CVI 2 + 5 2) 0.5% / 20.5
This expression is the Standard Error % (SE%) of the measurement,
expressed as the average of both results.
expressed as the average of both results.
%P0%10% :
Probability that, the result obtained with a single test, on a single sample and performed by an exact method, does not differ in more than 10% of the homeostatic condition of the studied analyte.
Probability that, the result obtained with a single test, on a single sample and performed by an exact method, does not differ in more than 10% of the homeostatic condition of the studied analyte.
%P5% x2 10% :
Probability that, the informed value as the average of two results obtained, from two samples, with an analytical procedure with CVA%=5%,
does not differ in more than 10% of the homeostatic condition of the studied
analyte.
Probability that, the informed value as the average of two results obtained, from two samples, with an analytical procedure with CVA%=5%,
does not differ in more than 10% of the homeostatic condition of the studied
analyte.
analyte
|
% CVB0%
|
%CVB5%/20.5
|
%P0%10%
|
%P5% x2 10%
|
Urea
|
12.3
|
9.4
|
58.2
|
71.54
|
Glucose
|
5.7
|
5.4
|
92.0
|
93.9
|
Cholesterol
|
5.4
|
5.2
|
93.6
|
94.5
|
HDL CHOL
|
7.1
|
6.1
|
84.1
|
89.7
|
LDL CHOL
|
8.3
|
6.9
|
77.0
|
85.6
|
Urate
|
9.0
|
7.3
|
73.3
|
82.9
|
Creatinine
|
5.3
|
5.2
|
94.1
|
94.8
|
Protein. total
|
2.7
|
4.0
|
100.0
|
98.7
|
Albumin
|
3.1
|
4.2
|
100.0
|
98.4
|
Triglyceride
|
20.9
|
15.2
|
36.9
|
49.1
|
Iron
|
26.5
|
19.1
|
29.6
|
39.7
|
Bilirubin Total
|
23.8
|
17.2
|
32.6
|
43.8
|
Sodium
|
0.7
|
3.6
|
100.0
|
99.5
|
Potassium
|
4.8
|
4.9
|
96.2
|
95.9
|
Chloride
|
1.2
|
3.6
|
100.0
|
99.4
|
Calcium
|
1.9
|
3.8
|
100.0
|
91.8
|
Magnesium
|
3.6
|
4.4
|
99.5
|
97.9
|
Phosphate
|
8.5
|
7.0
|
76.2
|
84.7
|
GPT-ALT
|
24.3
|
17.5
|
31.8
|
43.1
|
GOT-AST
|
11.9
|
9.1
|
59.9
|
72.9
|
GGT
|
13.8
|
10.4
|
52.8
|
66.3
|
Alkaline phosphatase,
|
10
|
7.9
|
68.3
|
79.6
|
LDH
|
8.6
|
7.0
|
75.4
|
84.4
|
CPK
|
22.8
|
16.5
|
34.0
|
45.8
|
Amylase (pancreatic)
|
11.7
|
9.0
|
60.5
|
73.3
|
analyte
|
% CVB0%
|
%CVB5%/20.5
|
%P0%10%
|
%P5% x2 10%
|
17-Hydroxyprogesterone
|
19.6
|
14.3
|
39.0
|
51.6
|
Androstendione
|
11.1
|
8.6
|
62.2
|
75.4
|
Cortisol
|
20.9
|
15.2
|
36.9
|
49.1
|
Estradiol
|
18.1
|
13.3
|
41.8
|
54.7
|
SHBG
|
12.1
|
9.3
|
59.3
|
72.0
|
DHEA-S
|
4.2
|
4.6
|
98.3
|
97.0
|
Testosterone total
|
9.3
|
7.5
|
72.0
|
82.0
|
Free testosterone
|
9.3
|
7.5
|
72.0
|
82.0
|
analyte
|
% CVB0%
|
%CVB5%/20.5
|
%P0%10%
|
%P5% x2 10%
|
CEA
|
12.7
|
9.6
|
57.0
|
70.2
|
AFP
|
12
|
9.2
|
59.3
|
72.4
|
ferritin
|
14.2
|
10.6
|
51.6
|
65.3
|
CA 15-3
|
6.1
|
5.6
|
89.9
|
92.7
|
CA 19-9
|
16
|
11.9
|
47.1
|
59.9
|
CA 125
|
24.7
|
17.8
|
31.1
|
42.5
|
PSA TOTAL
|
18.1
|
13.3
|
41.8
|
54.7
|
analyte
|
% CVB0%
|
%CVB5%/20.5
|
%P0%10%
|
%P5% x2 10%
|
LH(males)
|
14.5
|
10.8
|
51.0
|
64.2
|
FSH(males)
|
8.7
|
7.1
|
75.0
|
84.1
|
Prolactin(men)
|
6.9
|
6.0
|
85.3
|
90.3
|
Insulin
|
21.1
|
15.3
|
36.2
|
48.4
|
analyte
|
% CVB0%
|
%CVB5%/20.5
|
%P0%10%
|
%P5% x2 10%
|
T 3
|
8.7
|
7.1
|
75.0
|
84.1
|
T4
|
4.9
|
4.9
|
95.9
|
95.7
|
TSH
|
19.3
|
14.1
|
39.7
|
52.2
|
T4 free
|
5.7
|
5.4
|
92.0
|
93.9
|
T3 free
|
7.9
|
6.6
|
79.6
|
86.9
|
Conclusions:
Probability calculations (P) show that:
The probability of getting close, to the true value of the homeostatic condition of an analyte, is higher when processing two samples, with certain degree of an acceptable analytical error, than with a single test on a single sample, even though the result was obtained performing a method without errors!!!
The reason for those analytes, where this statement is not fulfilled, is that the chosen CVA% = 5% for the study, does not comply with the quality requirements for those analytes measurements.
Remember the quality specifications that CVA% should not be greater than 0.5 times the CVI%.
For instance: for albumin, the quality specifications are not fulfilled, since CVI% = 3.1%, and the CVA% of 5.0 % of the used method for the study, is higher than the expected limit of 1.6 %, see:
