Statistical analysis for analytical methods validations
AUTHORS:
Corneo R - Nanni V - Tizzani E
Nguyen Van Danh - Vanasia A
ASTRAZENECA SPA ITALY
A.VANASIA Srl
INTRODUCTION
In analytical chemistry every new analytical method, after the development step, must be
validated before routine use.
The validation process consist to evaluate the method results obtained with some statistical
approaches in order to define the systematic errors and application limits.
SCOPE OF VALIDATION PROCESS
The principal aim of the validation process is to discover the analytical limits, precision
and accuracy grade in order to evaluate the future analytical data obtained using the
method.
A full validation process should permit to verify if an anomalous result should be refused
or accepted. In this way we can maintain a constant analytical quality of results.
METHODOLOGY AND CONTENTS
In analytical chemistry there is a variety of analytical methods classified by analytical
methodology:
1. manual method (every operation is performed manually by analysts
2. automatic method (some or totally operation is performed by instrument).
For both these categories we will perform analytical validation. Obviously for the
automatic method the validation process is more complex and detailed because we have to
consider not only the systematic error of manual operations bur also the systematic errors
of every single part of automatic systems used. Actually in modern laboratories we
increase the automatic analysis therefore we need more efficiency for validation exercise.
The validation process is regulated by an International Guidelines or by the International
Pharmacopoeias (USP or PhEur).
Generally a validation process consist of the following study:
Linearity
Method or/and System Repeatability
Accuracy
Intermediate Precision
Quantification Limit
Detection Limit
Robustness
All the data series produced with these studies should be evaluated with a specific program
in order to apply the same approach and elaboration. Moreover it is evident that every
calculation approach must be validated before the use.
For these reasons we choose the standard SAS statistical applications. In particular we
used the GLM REG Proc. with relative output files and options statistical evaluation for
various model acceptability limits (e.g. intercept calculation and relative confidence
limits).
Some models are correlate with automated graphics procedures.
This program is divided in:
easier data entry
pick list menu
automated output
automated graphics
Main Menu
Insert new Test
Select Test
Select existing Test
Modify Tests Title
Screen of Test
Screen Accuracy of Data Entry/Modify
Accuracy
Title
TEST OF UK1 (FIRST)
Test Date
1.00)
__________
Analist
Content
Level
(%)
0.05
(0.05
____________________
(%)
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
87.03
Degrado (%)
0.684
Weight mg
% Theoric
% Practic
6.29
8.15
7.56
8.05
7.93
8.26
7.44
7.26
____________
____________
0.04271
0.05535
0.05134
0.05467
0.05385
0.05609
0.05052
0.0493
____________
____________
0.042
0.052
0.048
0.052
0.051
0.053
0.047
0.046
____________
____________
0.10
0.25
0.50
Screen Linearity of Data Entry/Modify
Linearity Title TEST OF UK1 (FIRST)
Test Date __________ Level (%) _____ Analist
% Theoric
Weight Mg
Items : 1
0.05
10.13
1
0.05
9.98
____________________ Unit MCG/ML
X Value
Y Area
2.12302
11280
2.09159
11051
2
2
0.1
0.1
19.98
20.18
4.18737
4.22929
22764
23546
3
3
0.2
0.2
40.34
40.38
8.45438
8.46277
47523
47341
4
4
0.4
0.4
80.75
80.8
16.92344
16.93395
96199
96140
5
5
0.5
0.5
101.05
100.7
21.17788
21.10453
120200
118552
6
6
0.6
0.6
121.54
121.25
25.47214
25.41137
145930
145243
7
7
0.8
0.8
161.52
161.58
33.85108
33.86366
196441
193612
8
8
1
1
202.08
201.92
42.35158
42.31805
242366
242741
9
9
____________
____________
____________
____________
____________
____________
____________
____________
10
10
____________
____________
____________
____________
____________
____________
____________
____________
Screen System Repeatability of Data Entry/Modify
System Repeatability (Precisione del Sistema)
Title
TEST OF UK1 (FIRST)
Test Date
Analist
04/08/2000
____________________
1
2
3
4
5
6
7
8
9
10
Area
2150570
2147458
2147478
2147382
2145879
2144850
2142292
2142521
2144085
2141689
Level (%)
Weight mg
90
____________
( B=Before
A=After )
Screen Intermediate Precision of Data Entry/Modify
Title
Date
Name :
Intermediate Precision
TEST OF UK1 (FIRST)
18/11/1999
18/11/1999
19/11/1999
19/11/1999
Analist 1
Analist 2
Analist 3
Analist 4
LOCATELLI
COLOMBO
Order :
1.
0.113
2.
0.107
3.
0.11
4.
0.107
5.
0.107
6.
0.11
7.
0.115
8.
0.109
9. ____________
10 ____________
0.213
0.203
0.21
0.202
0.214
0.205
0.214
0.202
____________
____________
LOCATELLI
0.108
0.111
0.109
0.111
0.11
0.11
0.11
0.115
____________
____________
COLOMBO
0.203
0.204
0.205
0.209
0.206
0.205
0.204
0.212
____________
____________
Screen Quantification Limit of Data Entry/Modify
Quantification Limit
Title
TEST OF UK1 (FIRST)
Test Date
Injection
Level
__________
Analist
p/p mg
____________________
Area
1
2
3
4
0.05
0.05
0.05
____________
7.54
7.15
8.02
____________
10932
9996
11773
____________
1
2
3
4
0.025
0.025
0.025
____________
4.17
3.65
3.96
____________
5975
5076
5813
____________
1
2
3
4
0.01
0.01
0.01
____________
1.8
2.03
3.15
____________
2168
2772
4570
____________
1
2
3
4
0.005
0.005
0.005
____________
2.48
1.86
1.57
____________
1458
1374
973
____________
1
2
3
4
____________
____________
____________
____________
____________
____________
____________
____________
____________
____________
____________
____________
Examples : Output of Linearitys Test
Assay=TEST OF UK1 (FIRST)
Model: MODEL1
Dependent Variable: YAREA
Analysis of Variance
Source
DF
Sum of
Squares
Mean
Square
Model
Error
C Total
1 95135853350
14 11297249.362
15 95147150599
95135853350
806946.383
Root MSE
Dep Mean
C.V.
898.30194
110058.06250
0.81621
R-square
Adj R-sq
F Value
Prob>F
117896.127
0.0001
0.9999
0.9999
Parameter Estimates
Variable
DF
Parameter
Estimate
Standard
Error
T for H0:
Parameter=0
Prob > |T|
INTERCEP
XVALUE
1
1
-1305.874836
5767.236825
394.49692146
16.79646991
-3.310
343.360
0.0052
0.0001
Linearity
Confidence Limit 95% of Intercept * Y = -1305.874836 +
5767.2368255 X
Assay
Lower
95th
Upper
95th
Degrees T_student Sqrt Xvalue
Freedom 97.5(%)DF Number Unit
YAREA
XVALUE
11280
11051
22764
23546
47523
47341
96199
96140
120200
118552
145930
145243
196441
193612
242366
242741
2.1230
2.0916
4.1874
4.2293
8.4544
8.4628
16.9234
16.9340
21.1779
21.1045
25.4721
25.4114
33.8511
33.8637
42.3516
42.3181
THEORY
TEST OF UK1 (FIRST) -2151.99 -459.763
14
2.14479
MCG/ML
0.05
0.05
0.10
0.10
0.20
0.20
0.40
0.40
0.50
0.50
0.60
0.60
0.80
0.80
1.00
1.00
