of test analyte. If the replicated values agree closely with each other, the measurements
are said to be ofhigh precision(orlow imprecision). In contrast, if the values diverge,
the measurements are said to be ofpoororlow precision(orhigh imprecision). In
analytical biochemical work the normal aim is to develop a method that has as high a
precision as possible within the general objectives of the investigation. However,
precision commonly varies over the analytical range (see below) and over periods of
time. As a consequence, precision may be expressed as eitherwithin-batchorbetween-
batch. Within-batch precision is the variability when the same test sample is analysed
repeatedly during the same batch of analyses on the same day. Between-batch
precision is the variability when the same test sample is analysed repeatedly during
different batches of analyses over a period of time. Since there is more opportunity for
the analytical conditions to change for the assessment of between-batch precision,
it is the higher of the two types of assessment. Results that are of high precision may
nevertheless be a poor estimate of the ‘true’ value (i.e. oflow accuracyorhigh bias)
because of the presence of unidentified errors. Methods for the assessment of precision
of a data set are discussed below. The term imprecision is preferred in particular by
clinical biochemists since they believe that it best describes the variability that occurs
in replicated analyses.- Accuracy(also calledtrueness,biasandinaccuracy): This is the difference between
the mean of a set of analytical measurements on the same sample of test analyte and
the ‘true’ value for the test sample. As previously pointed out, the ‘true’ value is
normally unknown except in the case of standard measurements. In other cases
accuracy has to be assessed indirectly by use of an internationally agreed reference
method and/or by the use of external quality assessment schemes (see above) and/or
by the use of population statistics that are discussed below. - Detection limit(also calledsensitivity): This is the smallest concentration of the test
analyte that can be distinguished from zero with a defined degree of confidence.
Concentrations below this limit should simply be reported as ‘less than the detection
limit’. All methods have their individual detection limits for a given analyte and this
may be one of the factors that influence the choice of a specific analytical method for
a given study. Thus the Bradford, Lowry and bicinchoninic acid methods for the
measurements of proteins have detection limits of 20, 10 and 0.5mg protein cm^3
respectively. In clinical biochemical measurements, sensitivity is often defined as the
ability of the method to detect the analyte without giving false negatives (see Section
16.1.2). - Analytical range:This is the range of concentrations of the test analyte that can be
measured reproducibly, the lower end of the range being the detection limit. In most
cases the analytical range is defined by an appropriate calibration curve (see Section
1.4.6). As previously pointed out, the precision of the method may vary across the
range. - Analytical specificity(also calledselectivity): This is a measure of the extent to which
other substances that may be present in the sample of test analyte may interfere
with the analysis and therefore lead to a falsely high or low value. A simple example is
the ability of a method to measure glucose in the presence of other hexoses such as
mannose and galactose. In clinical biochemical measurements, selectivity is an index
19 1.4 Quantitative biochemical measurements