Reliability of analytical data depends on appropriate quality assurance methods. This holds true for environmental and food monitoring, materials analysis, but also for analytics in biotechnology or in medicine (in-vitro diagnostics, point-of-care testing). The authors of this best-selling title describe a proven and integrated concept for quality assurance. The new edition is enhanced by a chapter on the important topical subject of measurement uncertainty, plus a CD-ROM with interactive examples in the form of Excel-spreadsheets. These allow readers to gain an even better comprehension of the statistical procedures for quality assurance while also incorporating their own data. Following an introduction, the text goes on to elucidate the 4-phase model of analytical quality assurance: establishing a new analytical process, preparative quality assurance, routine quality assurance and external analytical quality assurance. Besides updating the relevant references, the authors incorporated the latest international standards in the field. From reviews of the previous edition: "extremely informative and comprehensive" Accreditation and Quality Assurance "The book successfully combines the role of authoritative textbook and reference manual, with that of a practical handbook and guide to the selection of an appropriate AQA strategy." The Analyst "This book presents a unique, four-phase strategy for quality assurance." Journal of the American Chemical Society

The late Werner Funk was professor at the University of Applied Science in Gießen-Friedberg (Germany).
Vera Dammann is a senior scientist at the University of Applied Science in Gießen-Friedberg (Germany) since 1978. She has been working in areas such as computing, electrical engineering, electrical and biophysical measurement technology, and clinical engineering. She is project manager for the international study course 'clinical engineering' and a member of the board of the German association for biomedical instrumentation.
Gerhild Donnevert is laboratory engineer in the laboratory for analytical chemistry at the University of Applied Science in Gießen-Friedberg (Germany). She is an active member of the German standardization committees for the examination of water, wastewater and sludge as well as of the Technical Committee 147 'Water Quality' of the International Standards Organization ISO.

Quality Assurance in Analytical Chemistry

Applications in Environmental, Food and Materials Analysis, Biotechnology, and Medical EngineeringBy Werner Funk Vera Dammann Gerhild Donnevert Sarah Iannelli Eric Iannelli

John Wiley & Sons

Copyright © 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
All right reserved.
ISBN: 978-3-527-31114-9

Chapter One

Phase I: Establishing a New Analytical Procedure

1.1 Introduction

1.1.1 Objectives of Phase I

Before a new analytical procedure, especially one requiring calibration, can be used for routine analysis, the individual steps need not only be determined but, where necessary, they must also be optimized, and the entire fundamental analytical procedure must be verified for its performance. The performance characteristics obtained in this way are documented and/or published with the description of the analysis and they form the basis for later quality assurance in routine analysis.

1.1.2 When Are Characteristic Data Obtained?

The statistical methods described below find their application primarily during the establishment of a new analytical process. In addition, they are suitable for providing the analyst with information about recent analytical quality achieved during the testing and training phase of an analytical process that is in need of calibration (see also Section 2.3).

The process data (linear calibration function including precision measures) should be determined anew with each new calibration of the analytical process, for example after:

• changing reagents (new batches),

• technical intervention in analytical equipment (after technical modifications, maintenance and repair, for example bulb changes in photometry),

• changes in staff,

• etc.

1.1.3 The Progression of Phase I

Phase I consists primarily of the following five main steps (see Scheme 1-1):

• preparation of the calibration experiment,

• testing for

– linearity of the calibration function

– precision (variance homogeneity, absence of outliers, securing the lower bounds of the range)

• final determination of the calibration data,

• testing for the influence of the individual process steps or for matrix effects on the calibration data,

• testing for the influence of time on the analytical process.

Of special importance is the carefully conducted correction of the analytical process during the testing phase the moment the test results of a part of the system indicate an unacceptable analytical quality. If all possibilities for improving quality within the framework of the analytical process are exhausted and have not led to an improvement, in most cases a restriction of the range (reduction of the highest substance concentration chosen) can lead to an acceptable precision and to linearity of the calibration function as well as to the required variance homogeneity.

1.1.4 Results of Phase I; Statistical Data

Phase I of analytical quality assurance lays the foundation for later routine analysis by providing quality data. Above all, this includes:

• the range tested,

• the coefficients of the calibration function:

– in the case of a first-order calibration function (y = a + bx): axis intercept a and slope b (characteristic of the sensitivity of the analytical procedure),

– in the case of a second-order calibration function (y = a + bx + cx²): axis intercept a, coefficient b of the linear term, as well as the coefficient c of the quadratic term; the sensitivity E of the analytical process determined from the function,

• the standard deviation of the procedure, s_xo, as an absolute measure of precision for the calibration, and

• the process variation coefficient, V_xo, as a relative measure of precision.

In addition, the general evaluation of the analytical process also documents the following:

• decision limit, x_DL, as the substance content that produces a measurement larger than the blank value with a probability of error α (e.g., α = 5%),

• minimum detectable value, x_MDV, as the substance content that is larger than the blank value with a probability of error β (e.g., β = 5%),

• limit of quantification, x_LQ, as the substance content that can be determined with a maximum allowable relative result uncertainty,

• auxiliary test value, x_a, for the validation of a suitable range,

• recognized constant and/or proportional systematic deviations with certain matrices, as well as an indication of the time dependency of the analytical results, the time required, and other special remarks.

1.2 Calibration of the Fundamental Analytical Procedure (Fundamental Calibration)

For analytical processes in need of or capable of being calibrated, the application of physical measurement principles does not lead directly to an analytical result; the observations only represent the result of a physical measurement, which must be converted into the analytical result using data obtained empirically using a calibration experiment [101].

The use of an analysis function

Analytical result – function of the observation

or in mathematical notation:

[??] = f([??])

(with [??] as the observation and [??] as the substance content/analytical result) is based on the application of the calibration function obtained from the calibration experiment

observation = function of substance content or

y = f (x)

(with x = substance content of the standard sample and y = accompanying observation)

and the precision data therefrom for the determination of an unknown substance content in a sample. After solving for x, the calibration function then becomes the analytical function, which, after inserting the observation of the tested sample, [??], gives the analytical result, [??] [101].

A fundamental calibration represents the calibration of the fundamental analytical procedure, or, in other words, no sample preparation steps such as extraction or work-up are performed, only standards in pure solvents (e.g., distilled water) are analyzed.

1.2.1 Establishment of an Analytical Range

Every calibration begins with the choice of a preliminary range as determined by:

a) The objective of the application as related to practical conditions. This should cover a large range of applications. In addition, the mean of the range should be roughly equal to the sample concentration most often expected, as far as this is possible in individual cases.

b) The technical possibilities.

b1) The measured values at the lower end of the range must be significantly different from the process blank values. A lower range limit is only useful when it is at least the same or larger than the minimum detectable value (see Section 1.3.2) of the procedure. In addition, dilution and concentration steps must be easily and flawlessly realized.

b2) The required analytical precision must be achievable throughout the entire range (see also "limit of quantification", Section 1.3.3). Since the imprecision of an analysis increases absolutely with increasing substance content, the range in question must not be chosen too large. If another range is necessary for routine analysis, then this should...