Choosing the random error. The smaller these values

Choosing a new analyzer is an important decision
and a challenge for any laboratory. Before a new instrument or technique/method
can be used for testing of patient samples, it must undergo evaluation and
validation. A measurement that is hundred percent accurate and hundred percent
precise is ideal. But, test methodology, laboratory instruments, reagents used,
and laboratory operations all contribute to variations in results. Accuracy
and precision reveal reliability of test results. Many statistical procedures
are available for analysing of collected data, which help to get insight into
the agreement between instruments. Thus, in Bland-Altman statistical analysis
of assessing agreement between two methods, the mean of differences (bias) reflects
inaccuracy or the systematic error, and standard deviation (SD) reflects imprecision or the random
error. The smaller these values are, the more reliable are
the results. The Bland-Altman method also allows estimation of confidence
limits for the bias or limits of agreement. But, this method does not indicate
whether those limits are clinically acceptable or not. Therefore, clinically acceptable
limits of agreement must be defined, and these should be based on clinical
necessity (Hanneman SK, 2008; Giavarina D, 2015). In this
work the limits of agreement were evaluated according to CLIA requirements for
Analytical Quality. In linear regression analysis the significance of correlation between two methods is expressed by P-value. The
P-value for each independent variable
tests the null hypothesis, whether the independent variable has no correlation
with the dependent variable. If the P-value
for an independent variable is less than the significance level, then the data
provide enough evidence to reject the null hypothesis and the variable is
statistically significant. Analysis of data by Passing-Bablok
regression does not require assumption of Gaussian
errors distribution. This procedure allows assessing of systematic and
proportional differences between two analyzers, which is the case when the
intercept differs significantly from zero and the slope differs significantly
from one. Analytical sensitivity reflects
the lower limit of detection. Thus, a test/method with 100% sensitivity
identifies all positive results and classifies them correctly as positive.
Analytical specificity is the ability of a method to detect the substance of
interest and no other substance (Lalkhen AG & McCluskey A, 2008).