You are on page 1of 9

G Model

CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS


Journal of Chromatography A, xxx (2014) xxxxxx

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Review

Youden test application in robustness assays during method


validation
Eftichia Karageorgou, Victoria Samanidou
Laboratory of Analytical Chemistry, Department of Chemistry, Aristotle University of Thessaloniki, Thessaloniki 54124, Greece

a r t i c l e i n f o a b s t r a c t

Article history: Analytical method validation is a vital step following method development for ensuring reliable and
Received 6 November 2013 accurate method performance. Among examined gures of merit, robustness/ruggedness study allows
Received in revised form 15 January 2014 us to test performance characteristics of the analytical process when operating conditions are altered
Accepted 17 January 2014
either deliberately or not. This study yields useful information, being a fundamental part of method
Available online xxx
validation. Since many experiments are required, this step is high demanding in time and consumables.
In order to avoid the difcult task of performing too many experiments the Youden test which makes use
Keywords:
of fractional factorial designs and has been proved to be a very effective approach. The main advantage
Method validation
Robustness
of Youden test is the fact that it keeps the required time and effort to a minimum, since only a limited
Ruggedness number of determinations have to be made, using combinations of the chosen investigated factors. Typical
Youden test applications of this robustness test found in literature covering a wide variety of sample matrices are
briey discussed in this review.
2014 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
1.1. Food analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
1.2. Biouids and pharmaceutical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
1.3. Environmental analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
2. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Conict of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

1. Introduction

According to the International Organization for Standardization, Method validation has been studied extensively in the literature,
method validation is dened as conrmation through examina- as well as in several guidelines from regulatory agencies, including
tion and provision of objective evidence that the requirements for Eurachem [2], the International Union of Pure and Applied Chem-
a specied intended use or application are fullled [1]. istry [3], International Conference on Harmonization (ICH) [4], the
The scientic applicability of an analytical method is required U.S. Food and Drug Administration [5] and the European Commis-
in terms of validation procedures not only in new method devel- sion Legislation through Commission Decision 2002/657/EC [6].
opment, but for established method revision, methods used in The ofcial guidelines do not present a validation experi-
different laboratories, the employment of different analysts or ment sequence because the optimal sequence may depend on the
instrumentation, and for demonstration of equivalence between method itself and the application. For quantitative analytical HPLC
two different methods as well [2]. methods, however, the following sequence has proven to be useful
(Fig. 1):

Corresponding author. Tel.: +30 2310997698; fax: +30 2310997719. 1. Selectivity


E-mail address: samanidu@chem.auth.gr (V. Samanidou). 2. Linearity and range

http://dx.doi.org/10.1016/j.chroma.2014.01.050
0021-9673/ 2014 Elsevier B.V. All rights reserved.

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050
G Model
CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS
2 E. Karageorgou, V. Samanidou / J. Chromatogr. A xxx (2014) xxxxxx

Table 1
Potential factors to be examined in the robustness testing of main separation
techniques.

Separation technique Factors

HPLC Mobile phase constituents


Mobile phase pH
Organic modier %
Buffer concentration, salt concentrations or
ionic strength
Concentration of additives (ion pairing agents)
Flow rate
Column temperature
Gradient elution: initial mobile phase
composition slope of the gradient
Column stationary phase
Column manufacturer
Wavelength of detection

Gas chromatography (GC) Injector temperature


Column temperature
Detector temperature
Temperature program (initial and nal
Fig. 1. The main factors examined during a method validation procedure. temperature)
Slope of the temperature Gradient
Gas ow-rate
3. Precision (repeatability and intermediate precision) Split or splitless conditions
4. Accuracy Liner type
Column manufacturer
5. Detection limit
Column stationary phase
6. Quantitation limit
7. Robustness/ruggedness Capillary electrophoresis (CE) Electrolyte concentration
Buffer pH
8. Stability [4].
Concentration of additives
Temperature
The knowledge of the robustness/ruggedness of analytical pro- Applied voltage
cesses is an essential feature in the analytical validation [7]. Sample injection time
Sample concentration
Robustness is dened as the susceptibility of an analytical method
Rinse times
to changes in experimental conditions which can be expressed as Wavelength of detection
a list of the sample materials, analytes, storage conditions, envi-
ronmental and/or sample preparation conditions under which the
method can be applied as presented or with specied minor mod- (d) perform the experiments in random order on a control sample
ications. For all experimental conditions which could in practice with analyte concentration halfway in the concentration range of
be subject to uctuation (e.g. stability of reagents, composition of the method scope. Youden selected a 27-4 PlackettBurman design
the sample, pH, temperature) any variations which could affect the because it enabled the study of up to seven factors in eight experi-
analytical result should be indicated. Robustness and ruggedness ments. The eight runs are split into two groups of four runs on the
are closely related terms often used interchangeably in the litera- basis of levels +1 or 1. The effect of every factor is estimated as
ture [6]. Although there is a confusion between the terms due to the difference of the mean result obtained at the level +1 from that
their similarity, ruggedness is an older term replaced by intermedi- obtained at the level 1. Once effects have been estimated, to deter-
ate precision which is a measure showing how well is the method mine whether variations have a signicant effect on the result, a sig-
performing under normal conditions from lab to lab, instrument to nicance t-test is used. The t-value is compared with the 95% con-
instrument and analyst to analyst. Robustness is the measure show- dence level two-tailed tabulated value with the degrees of freedom
ing the performance of the method when small deliberate changes coming from the precision study for each concentration [11].
are made to the normal conditions [8]. Youden and Steiner in 1975, introduced a useful robustness test
Robustness testing is performed at the end of development or to analytical chemistry [12]. These tests examine, within one single
the beginning of validation. When method robustness is found to laboratory, the susceptibility of analytical methods against small
be acceptable, the method can be further validated, and when the changes in the method parameters using experimental designs.
validation results are regarded as satisfactory, it can be routinely Such preliminary tests are less expensive than collaborative studies
applied. Otherwise, the method should be adapted or redeveloped and allow adjusting the method before it is studied collaboratively
[9]. [13].
A robustness study examines the alteration of factors impor- Youden robustness test can be performed either during the
tant for the analytical procedure. In experiments to study the main optimization of the separation technique used, or during the opti-
effects of factors, screening designs are used. Plackett and Burman mization of sample preparation. Various factors important for the
developed such designs which are suitable only when the inter- analytical procedure can be investigated. Tables 1 and 2 summarize
actions are negligible or when considering a key set of dominant the most important separation and sample preparation techniques
factors, since their designs cannot deal with factor interactions [10]. used, as well as they give examples for the potential factors that
The basic idea of Plackett and Burman designs in brief, is described can be used for the robustness study.
as follows. The strategy is based on a landmark procedure sug- Other approved methods may be used at this point. The Youden
gested by Youden in 1967: (a) identify the inuential factors; (b) for approach, however, keeps the required time and effort to a
each factor, dene the nominal and the extreme values expected in minimum. This robustness test is a fractional factorial design. Inter-
routine work and encode them as follows: nominal value = 0, high actions between the different factors cannot be detected [6].
value = +1 and low value = 1; (c) arrange the experimental design The basic idea is not to study one alteration at a time but to
by using a two level 27-4 fractional PlackettBurman design; and introduce several variations at once. As an example, let A, B, C, D, E,

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050
G Model
CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS
E. Karageorgou, V. Samanidou / J. Chromatogr. A xxx (2014) xxxxxx 3

Table 2 statistical treatment, like method-performance studies (collabo-


Potential factors to be examined in the robustness testing of common sample prepa-
rative trials), laboratory-performance studies (prociency tests),
ration techniques.
collaborative bias evaluation, inter-laboratory evaluation of to-be
Sample preparation technique Factors standard methods as well as certication studies for reference
SPE Sorbent type materials. Furthermore, new robust statistics and graphical meth-
Sorbent manufacturer ods were taken into account [13].
Sorbent mass Accuracy proles of chemical assays were introduced and
Sample mass or volume
derived from the uncertainty of the analytical result. Gonzalez et al.
Wash solvent
Elution solvent in 2006 explained the calculation of accuracy proles, based on the
Evaporation temperature estimation of the measurement uncertainty of the analytical assay
pH of sample from validation data. For the sake of illustration, a case study deal-
pH of buffer constituents in solvents
ing with the spectrouorimetric determination of quinine in tonic
MSPD Sorbent type water is explained in detail [14]. Also in 2007, the same authors
Sorbent manufacturer gave more information on the topic, through a detailed step-by-
Sorbent mass
step guide to analytical method validation, considering the most
pH of sample
pH of buffer constituents in solvents relevant procedures for checking the quality parameters of ana-
Sonication time lytical methods. Using a holistic approach, authors also explained
Evaporation temperature the estimation of measurement uncertainty and accuracy proles,
Wash solvent
which they discussed in terms of accreditation requirements and
Elution solvent
Sample mass or volume
predened acceptability limits [10].
Dejaegher et al. in their tutorial in 2009, introduced the appli-
cation of experimental designs in separation science. Method
F, G denote the nominal values for seven different factors that could optimization is often divided into screening and optimization
inuence the results, if their nominal values are changed slightly. phases. In the rst step, many factors, potentially affecting the sep-
Let their alternative values be denoted by the corresponding lower aration, are tested to identify those with the largest effects. These
case letters a, b, c, d, e, f and g. This results in 27 or 128 different are then further examined in an optimization phase, to determine
possible combinations. It is possible to choose a subset of eight of the best separation conditions. During the rst phase of method
these combinations that have a balance between capital and small optimization and during robustness testing, screening designs are
letters (Table 3). Eight determinations have to be made, which will usually applied, whereas during the second phase of optimization,
use a combination of the chosen factors (AG). The results of the response-surface designs are often used. In this tutorial, the differ-
determinations are shown in Table 3 as sz [6]. ent steps in the application of both types of design are explained
After the comparison of the averages of the capital letters (AA and elaborated with some examples [9].
to AG) with the averages of their corresponding small letters (Aa In 2013 Cuadros-Rodrguez et al. discussed a dual general pro-
to Ag), the effect of a factor is evaluated, when the difference is cedure for checking robustness/ruggedness in both intrinsic and
signicant larger than the differences of the other factors. Standard extrinsic validation. Inertia study is introduced to designate
deviation of the differences Di (SDi ) is calculated according to the this methodology in extrinsic validation, whereas precision and
equation: trueness tests are proposed inside the inertia study. Several exper-

  D2 
imental designs, depending on the goal of robustness/ruggedness
i study and the type of variables (quantitative or qualitative) are
SDi = 2
7 considered [7].
In the eld of pharmaceutical analysis one review is published
When SDi is signicantly larger than the standard deviation of in 2013 trying to assist in the planning of validation procedures in
the method carried out under within-laboratory reproducibility quantitative high-performance liquid chromatographic methods.
conditions it is a foregone conclusion that all factors together have Authors also discuss relevant approaches of various parameters.
an effect on the result, even if every single factor does not show a Finally, this article provides several up-to-date examples, which
signicant inuence and that the method is not sufciently robust are useful as an introduction to analytical validation for practical
against the chosen modications [6]. applications either in academic research or the industrial sector
Various review studies are available in the literature referring [15].
to the usefulness of validation process and Youden robustness Scope of this review article is to present characteristic appli-
test in analytical chemistry. In year 2000 Hund et al. dis- cations of Youden test, during the validation process of analytical
cussed various inter-laboratory studies performed with different methods in the eld of food, biouids and pharmaceutical as well
aims and consequently require different evaluation methods and as environmental analysis.

Table 3 1.1. Food analysis


Robustness tested by applying seven small but deliberate changes in the operating
parameters. A liquid chromatographic method with tandem mass spectro-
Parameter Experiment number metric (LCMS/MS) detection and identication is proposed for the
determination of chloramphenicol (CAP) in honey. After a prelimi-
1 2 3 4 5 6 7 8
nary dissolution in water, samples were extracted with a mixture of
A/a A A A A a a a a dichloromethane/acetone and evaporated to dryness; the follow-
B/b B B b b B B b b ing clean up was carried out on an octadecyl SPE cartridge. Method
C/c C c C c C c C c
D/d D D d d d d D D
ruggedness was estimated by means of the Youden robustness
E/e E e E e e E e E test. Seven different variables chosen within the entire analyti-
F/f F f f F F f f F cal process were taken into account in the evaluation of method
G/g G g g G g G G g ruggedness: dichloromethane concentration in the extraction mix-
Observed results s t u v w x y z
ture, extract evaporation temperature, extract evaporation mode,

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050
G Model
CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS
4 E. Karageorgou, V. Samanidou / J. Chromatogr. A xxx (2014) xxxxxx

buffer pH, SPE cartridge elution mode, SPE cartridge elution vol- capability (CC) and robustness. Robustness was evaluated through
ume, ammonium acetate concentration. The experimental design the fractional factorial Youden design by selecting seven factors as
was planned with the aim of identifying the critical points in the variables for Youden test: milk thermal treatment (raw or pasteur-
procedure. The effect of each factor was estimated according to the ized), evaporation temperature, storage at +4 C for 1 day, eluate
Youden approach. As the obtained value was less than the standard ltration, immunoafnity cartridge lot, and lot and commercial
deviation of the within-laboratory reproducibility, it was demon- company of methanol and acetonitrile. Results indicated that the
strated that all selected factors together do not signicantly affect method was not affected by slight variations of some critical factors
the analytical performance. Besides this general result, the inu- as pre-treatment and clean-up of milk samples, thermal treatment
ence of each variable on method robustness was also evaluated by and different storage conditions, as well as by major changes val-
applying the t-test and no factor had a signicant effect on rugged- ued in terms of milk produced by different species (buffalo, goat
ness. The ruggedness test demonstrated that the performance of the and sheep) [20].
method was not inuenced by minor changes in the key analytical A GCMS method for the determination of anabolic steroids
variables [16]. used as growth promoting agents in piglet feed samples has been
A conrmatory method based on high-performance liquid developed and validated according to the Commission Decision
chromatographytandem mass spectrometry (HPLCMS/MS), for 2002/657/EC criteria. The robustness was evaluated using the
the simultaneous determination of lincomycin and ve macrolide Youden test. For this purpose, minor changes of seven variables in
antibiotics in honey, was developed. The analytes were extracted, the sample preparation procedure were chosen: operator, leach-
and cleaned-up by a single solid-phase extraction step on OASIS ing volume, saponication-volume, saponication-temperature,
HLB column. Validation of the method performed according to liquidliquid extraction (LLE) volume, evaporation temperature of
European Commission Decision 2002/657/EC: linearity, specicity, the nal extract, and derivatization-time. Following the experi-
decision limit, detection capability, repeatability, reproducibility, mental plan, results indicate that Saponication-temperature is
recovery and ruggedness. The robustness test was conducted by the parameter with the biggest inuence on the robustness of the
the Youden procedure cited by Commission Decision. Seven vari- method. Moreover, all the chosen factors together do not signi-
ables were chosen and deliberately altered: the volume of dilution cantly affect the method performances [21].
buffer, pH and molarity of dilution buffer, the methanol percent- The same group of authors developed an improved sample
age during the washing steps of the SPE purication, the ammonia preparation procedure for the determination of 17 steroids (cor-
solution percentage in the elution solvent, the SPE elution volume, ticoids and androgenic anabolic steroids) in feed samples. This
and the evaporation temperature of the solvents in the nal extract. procedure is based on two reported LCUV methods. Validation was
The results indicate that the method is robust and minor but still performed meeting European Legislation criteria. Robustness of the
signicant uctuations in the operative parameters that can occur method was carried out using the Youden test described in the
during the routine application of the method have no signicant Directive 2002/657/EC. The variables chosen were: the analyst, lix-
effect on its performance characteristics [17]. iviation volume, several SPE parameters (cartridges age, ultrapure
An experimental design developed by Youden and Steiner was water pH wash, % MeOH wash, elution volume), and nal evap-
applied to micro-fermentation experiments with two different oration temperature. The obtained results indicate SPE % MeOH
grape musts, in order to verify the impact of several parameters on wash is the parameter with the larger inuence, whereas all the
the fermentation course and on avor development. The follow- chosen factors together do not signicantly affect the method per-
ing seven parameters were studied: turbidity of the must, yeast formances [22].
strain, temperature of the alcoholic fermentation, de-acidication A precise method for the determination of 10 sulphonamide
of the must, suspension of the lees, presence of oxygen and nitrogen antibiotics in egg by liquid chromatographytandem mass spec-
supplementation. The inuence of each parameter on the fermen- trometry (LCMS/MS) has been developed. Drugs were extracted
tation process is determined by calculating the mean value and the using a mixture of dichloromethane/acetone (50:50, v/v), acidi-
absolute differences between the two variants. Results indicated ed with acetic acid and then cleaned-up on a cation-exchange
the positive effects of a higher fermentation temperature on the solid-phase extraction (SPE) cartridge. The method was validated
development of 3-mercaptohexanol, an important contributor to according to European Community Guidelines, whereas robust-
the characteristic aroma of the Petite Arvine wine, whereas results ness was estimated according to the Youden test. The chosen
obtained with the vinication trials conrm previous studies, thus investigated factors are: dichloromethane in the extraction, addi-
proving the correctness of the experimental design [18]. tion of acetic acid, SPE sulphonic cartridge, acetic acid for SPE
Anethole was determined in aniseed drinks by high per- conditioning, ammonia for SPE elution, SPE elution volume and
formance liquid chromatography (HPLC). A C18 column and a ammonium acetate in the mobile phase. The results arising from the
methanolwater (80:20, v/v) mobile phase with isocratic elution eight relevant robustness experiments indicate a decrease in sul-
were used. UV detection at 257 nm was carried out. Validation famethoxazole recovery rates when the acetic acid concentration is
was performed according to the EURACHEM guidelines. In order to at low level. The uctuations introduced in the current parameter
establish the robustness of the proposed method, the inuence of were in the range between 50% around the nominal value (5%)
four factors was studied, column temperature, % of methanol in the instead of the common 10% and so they represent a heavy mod-
mobile phase, ow rate and wavelength of detection. Robustness ication in the analytical procedure. In fact, these variations can
study was performed according to Youden approach. Furthermore, hardly be included in the typical denition of minor changes and
an experimental design was performed by using a two level 27-4 fac- consequently it cannot be inferred that the variable into account
torial PlackettBurman design. There are no signicant differences has a signicant effect on method ruggedness, so that the method
between the results obtained when factors have the maximum itself is not able to withstand minor uctuations in the operative
and minimum values. Therefore, these factors do not affect to the parameters that can occur during its routine application [23].
method development so that the robustness of the method can be A sensitive multiresidue method is developed for the simul-
assessed [19]. taneous determination of tropane alkaloids in grains and seeds
An HPLC method with uorimetric detection was developed for (wheat, rye, maize, soybean, linseed). The analytes were sepa-
the determination of aatoxin M1 (AFM1) in milk. Validation was rated by isocratic HPLC and detected using a triple quadrupole
performed according to European Decision 2002/657/EC includ- mass spectrometer with electrospray ionization (ESI). For a fast
ing selectivity, recovery, precision, decision limit (CC), detection and effective clean-up procedure for oily matrices such as soybean

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050
G Model
CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS
E. Karageorgou, V. Samanidou / J. Chromatogr. A xxx (2014) xxxxxx 5

and linseed, matrix solid phase dispersion (MSPD) C18 material QuEChERS sorbent, sonication time, milk volume, volume of wash
was used to remove co-extracted non-polar components. Method solvent, volume of methanol, and volume of ACN. The results indi-
performance is presented in terms of linearity, specity, selectiv- cate that the factor that has less inuence is the sonication time.
ity, accuracy, precision and ruggedness. In this study, the Youden However sonication is considered as absolutely necessary during
approach, a fractional factorial design, to determine any critical the pre-treatment procedure because it ensures efcient contact
points was employed. The factors examined were: extraction pH, between the solid and the extractant. Change in sorbent mass inu-
centrifuge speed, evaporation temperature, buffer concentration, ences signicantly either in positive or negative way ve out of
column temperature and mobile phase ow. The test conrmed seven analytes. Furthermore, the developed method was proven to
that the method was not signicantly inuenced by the combined be robust in the assayed experimental conditions for the sum of
minor variations introduced and can be considered acceptably studied variables [27].
robust [24]. In the second method 12 -lactam antibiotics, (four penicillins
In this study, a validated HPLC method with uorescence detec- and eight cephalosporins) were effectively extracted and deter-
tion for the simultaneous quantication of hydroxytyrosol and mined from milk matrix. Method ruggedness was estimated for
tyrosol in red wines is presented. The validation of the analytical minor changes by means of the seven parameters/eight exper-
method was based on selectivity, linearity, robustness, detection iments design, also known as the Youden and Steiner test. The
and quantication limits, repeatability, and recovery. A fractional inuence of following parameters on the results were tested: mass
factorial experimental design including percentage of pH, modier of Oasis sorbent, mass of QuEChERS sorbent, sonication time, wash
in solvents, ow rate, equilibrating time, percentage of ethanol in solvent volume, amount of acetone in wash solvent which is water,
the sample and column temperature on the peak area for hydroxy- eluent volume (MeOH), and eluent volume (ACN). From the results
tyrosol and tyrosol, was performed to analyze the inuence of these arising, the factor that has the minor inuence is the mass of Oasis
six different conditions on analysis. Fractional factorial design was sorbent. On the other hand the most important parameter is the vol-
chosen instead of full factorial design to reduce the total number of ume of methanol. Any change of this factor during the elution step
necessary experiments. Non signicant differences were found in inuences signicantly either in positive or negative way ve out
the resulting peak areas for both peaks. Therefore, the developed of twelve examined antibiotics. Less negative or positive inuence
method was proven to be robust in the assayed experimental con- on target analytes have factors of the mass of QuEChERS sorbent
ditions for the six studied variables. The nal optimized method and eluent volume ACN, whereas all the chosen factors together do
allowed the analysis of 30 nonpretreated Spanish red wines to not signicantly affect the method performances [28].
evaluate their hydroxytyrosol and tyrosol contents [25]. A sensitive and selective conrmatory method for milk-residue
For the quantitative determination of polyphosphates in prod- analysis of ten quinolones and eight cephalosporins by LCMS/MS
ucts of animal origin (meat, dairy and sh products), an ion has been developed in the third method. The Youden approach was
chromatography with suppressed conductivity detection method used for the ruggedness test of the method. The following parame-
was developed. The method validation, performed according to ters were tested: mass of Plexa sorbent, mass of QuEChERS sorbent,
Regulation 882/2004/EC and Decision 657/2002/EC in terms of lin- sonication time, percentage of acetone in wash solvent, volume of
earity, specicity, precision, recovery, detection and quantication wash solvent which is water, volume of methanol, and volume of
limits and ruggedness. Method robustness was tested by using ACN. From the examined factors the mass of Plexa sorbent is the
the Youden experimental design, performed for different prod- one with the major inuence on most of the analytes. Factors of
ucts of animal origin: cooked ham, wurstel, processed cheese and less inuence appear to be the mass of QuEChERS sorbent and son-
surimi. The seven factors used as Youden test variables were the ication time, while the volume of solvents in washing and elution
matrix and six ctitious factors; therefore, the experimental design steps has the minor inuence. The method proved to be robust for
requires eight independent experiments, four with the validation all investigated factors together [29].
matrix (cooked ham) and four with each alternative matrix. Results In the last method tetracycline antibiotics were examined.
indicated that the matrix variation has no effect on the analytical Target analytes were determined by an accurate and sensitive chro-
performances, and, consequently, the method is also applicable to matographic analytical method. Method robustness was estimated
wurstel, processed cheese and surimi samples [26]. for minor changes by means of the Youden and Steiner test. Seven
Authors have had a thorough research on the presence of different factors were chosen, because of their possible critical
antimicrobials agents in milk, food of animal origin. They have inuence. These included sonication time, sorbent material, evap-
developed various multi-class conrmatory methods by liquid oration temperature, milk volume, concentration of oxalic acid,
chromatography, using analytical columns of conventional and volume of elution solvent, and nature of elution solvent. Evapora-
novel technology [2730]. Validation was fully performed accord- tion temperature inuences in a negative way all compounds and
ing European Decision 2202/657/EC criteria in terms of specicity, especially OTC. This effect was expected as tetracyclines are tem-
linearity response, trueness, precision (repeatability and between- perature sensitive. The nature of organic solvent at elution step
day reproducibility), decision limit (CCa), detection capability inuences positively all tetracyclines and the most TC in favor of
(CCb), and ruggedness. In the eld of sample preparation, an acetonitrile. Moreover, all the chosen factors together do not sig-
innovative ultrasound-assisted MSPD approach together with new nicantly affect the method performances [30].
sorbent materials was used, and all target analytes were well
resolved from complex milk matrix. 1.2. Biouids and pharmaceutical analysis
In the rst method, both penicillins and amphenicols were
studied as described. The robustness of this method was assessed For the determination of captopril in tablet dosage forms, a
according to the Youden and Steiner approach. The basic idea is high performance liquid chromatorgaphic (HPLC) procedure using
instead of studying one alteration at a time, several variations are indirect photometric detection was developed. A low capacity
introduced at once. This was performed by adopting the experi- anion-exchange column was used with potassium phthalate as the
mental design described in Decision 2002/657/EC and tested by mobile phase marker and indirect detection at 280 nm. The chro-
the introduction of seven small but deliberate changes in the matographic conditions were optimized using the Box and Behnken
operating parameters (variables) during sample preparation pro- factorial experimental design. The mobile phase related factors
cess and by the consequent assessment of their inuence on the including pH, ow rate, marker concentration, % organic modi-
method results. The changes were: mass of Strata sorbent, mass of er, along with the column temperature, detector wavelength and

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050
G Model
CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS
6 E. Karageorgou, V. Samanidou / J. Chromatogr. A xxx (2014) xxxxxx

column age were chosen as the seven variables for Youden and A chiral capillary electrophoresis (CE) method has been
Steiners robustness test. The effect of each variable was investi- developed allowing the enantiomeric separation of racemic
gated at two levels. A total of eight experiments were conducted citalopram (R-() and S-(+) citalopram) using chiral selector
in order to identify variables that may have to be controlled in as carboxymethyl--cyclodrextrin (CM--CD). The inuence of
order to obtain accurate assay results. Results showed that the chemical and instrumental parameters on the separation such as
method is robust. Using the developed method, commercially avail- cyclodextrin and buffer concentrations, buffer pH, voltage, injec-
able captopril tablets were assayed and the results were found to tion pressure was investigated, so good chiral separation of the
be comparable with those obtained by the compendial method racemic mixture was achieved in less than 4 min. A robustness
[31]. test was performed using the PlackettBurman fractional design
A capillary gas chromatographic method with ame ionization using a matrix of 15 experiments for seven factors (internal
detection (FID) has been developed for the analysis of tamoxifen, parameters) with a statistical treatment suggested by Youden and
anastrozole, and letrozole in their pharmaceutical preparations, Steiner. Factors chosen in this study were: pH phosphate buffer,
drugs used in advanced breast cancer, using clomipramine as buffer concentration, cyclodextrin concentration, voltage ramp,
the internal standard in order to achieve quantication. A test injection time, injection pressure, time of rinsed with electrolyte.
of the ruggedness of this method was carried out using the Results showed that the most critical factors in this developed
PlackettBurman fractional design with a matrix of fteen exper- electrophoretic procedure could be the decrease of cyclodextrin
iments. Factors chosen were: column head pressure, time for concentration below its optimal value and the variation of volt-
the splitless step, temperature for the splitless step, injector age ramp and injection pressure too. Method proved to be robust
temperature, detector temperature, injected volume, nal oven for all the variations tested in this study for all the operating fac-
temperature. The mean effect of each variable is the average differ- tors on every studied electrophoretic parameter. Applicability of
ence between observations made at the extreme levels and those the method has been proved in the analysis of four pharmaceutical
made at the optimal level. Mean effects and standard errors were formulations [35].
calculated using the procedures described by Youden and Steiner. Five antidepressant drugs (uoxetine, uvoxamine, citalopram,
Taking into account the results when the selected operating fac- sertraline and paroxetine) were determined by a sensitive capil-
tors were tested using PlackettBurman experimental design and lary GCMS method. For GC separation were investigated, ow
YoudenSteiner statistical analysis, this method has proved to be rate, column head pressure, injector temperature, injection split-
rugged for all the variations tested in this study. The only critical less conditions and oven temperature program. MS detection
values are, in some cases, the injected volume and the nal oven was performed in SIM mode to increase the sensitivity. The
temperature, as was to be expected [32]. PlackettBurman fractional design based on balanced incomplete
In this paper, a capillary zone electrophoresis method is used blocks was employed for the robust test of the method, using a
for the separation of six antidepressants used for the treatment design for seven factors and 15 experiments. Youden and Steiner
of mental illness. Optimum conditions for their separation were approach was used for the statistical analysis. Investigated factors
investigated. A robustness test of the proposed method was per- were: Injection temperature, splitless temperature, splitless time,
formed using the fractional factorial model of PlackettBurman, EM voltage, oven time program and the two different oven tem-
requiring 15 experiments, in which the inuence of seven factors perature programs. Results indicated that splitless temperature is
at three different levels was tested on different electrophoretic the most critical factor (but still robust) from all examined ones.
results: efciency; resolution; and corrected peak area, whereas Finally, the method was applied to the analysis of these antide-
statistical evaluation was performed by Youden and Steiners pressants in nearly all their pharmaceutical formulations, obtaining
method. The main interaction effects are produced by injection high recoveries [36].
time over the corrected peak areas and by the ionic strength of the An extraction method, using a polypropylene membrane sup-
electrolyte over efciency. However, the ruggedness obtained in all porting dihexyl ether (three-phase hollow ber-based liquid phase
cases allows use of this method by different laboratories, analysts, microextraction followed by a HPLCDAD determination) was used
or instruments without any appreciable error [33]. for the analysis of three pharmaceuticals (salicylic acid (SAC),
Paroxetine and three metabolites have been determined at ibuprofen (IBU) and diclofenac (DIC)). The extraction procedure is
clinical levels in human urine by a validated nonaqueous capil- controlled by the donor/acceptor pH and the extraction time (fac-
lary electrophoresis (NACE) method. Prior to NACE determination, tors). A full factorial design for three factors and two levels involving
the samples were puried and enriched by an extraction- eight experiments (23 ), has been used to determine the effect and
preconcentration step with a preconditioned C18 cartridge and importance of the mentioned variables on the nal result. The
eluting the compounds with methanol. Good results were obtained effects were differential quantities that are related to the response
for different aspects including stability of the solutions, linearity, change as a result of changes in one or more factors/variables. From
accuracy, and precision. Fractional factorial designs developed by the results arises, that the main factor is the acceptor pH for all the
Plackett and Burman with a matrix of 15 experiments were used analytes. The method has been successfully applied to their deter-
for a ruggedness test of the method, based on balanced incomplete mination in human urine and the results obtained demonstrated
blocks according to procedures described by Youden and Steiner. that could also be applied to the determination of the correspond-
The variables selected in three value levels in this study were: (A) ing metabolites [37].
voltage, (B) injection time, (C) concentration of NH4 OAc, (D) tem- In the next example, the robustness of a chromatographic
perature of the separation, (E) % ACN, (F) % acetic acid and (G) time method applied to quantify lumefantrine in raw material sam-
of rinse with electrolyte. The mean value of each variable is the ples was assessed using Youdens test. Seven analytical parameters
average difference between observation made at the extreme lev- were selected and small variations were induced in the nominal
els and those made at the optimal level. Mean effects and standard values of the method. Then, eight runs were performed aiming to
errors were calculated using the procedures described by Youden determine the inuence of each parameter in the nal result. The
and Steiner. The main interaction effects over paroxetine area are seven analytical parameters employed were: methanol concentra-
produced by injection time and voltage, so it is important to work tion in mobile phase, mobile phase pH, column temperature, mobile
with an internal standard. Results indicated that this analytical phase ow rate, column supplier, methanol supplier, chromato-
method for measuring paroxetine and its metabolites has proved graph model. The chromatographic method proved to be highly
to be rugged to all the variations tested [34]. robust regarding the lumefantrine content. Some parameters such

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050
G Model
CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS
E. Karageorgou, V. Samanidou / J. Chromatogr. A xxx (2014) xxxxxx 7

as mobile phase ow rate and methanol supplier presented low described before, study design was based on Youdens nonrepli-
inuence in the evaluated factors of the chromatographic method. cate plan for collaborative tests of analytical methods. Samples
Youdens test showed to be a simple and feasible procedure to were spiked with ETU at 6 concentrations levels, prepared as 3
evaluate the robustness of chromatographic methods [38]. Youden pairs. Water was extracted by passing the sample through
In this work, a three phase hollow ber-based liquid phase an absorbent matrix type tube, from which ETU was recovered with
microextraction (HF-LPME) combined with an HPLC procedure methylene chloride, and nally an aliquot of each extract is ana-
using diode array and uorescence detection has been developed lyzed by GC. In the study participated twelve laboratories. Data
for the determination of eight uoroquinolones. The robustness were analyzed using a USEPA computer program, which measured
study is based on a landmark procedure suggested by Youden. Fac- recovery and precision for ETU and compared the performance
tors investigated were pH values, for donor and for the acceptor of the method between the two water types tested. Results for
phase. Stirring time is not considered as a variable for robustness the water matrixes showed no statistically signicant differences
study due to its high optimum value and the fact that variations in [42].an interlaboratory method validation study was conducted
the order of minutes do not have signicant effects in the extraction on EPA Method 515.1, referring to the determination of chlori-
efciency. Results obtained that t values calculated for each factor nated acids in water by gas chromatography with an electron
are lower than the tabulated one, so the procedure can be con- capture detector. This method is one of the 6 pesticide meth-
sidered robust against the considered factors for all the analyzed ods developed for the USEPA National Pesticide Survey (NPS).
compounds. The proposed method was applied to the determina- Method recovery and precision for analyses of sub-ppb to low-ppb
tion of the analytes in bovine urine and in environmental water concentrations of chlorinated acids were determined in reagent
samples (surface, tap and wastewater) [39]. water and nished drinking waters. Target analytes included the
12 pesticides that were quantitatively measured in the National
1.3. Environmental analysis Pesticide Survey (bentazon, 2,4-D, 2,4-DB, 3,5-dichlorobenzoic
acid, DCPA-diacid, dicamba, dichlorprop, 5-hydroxydicamba, pen-
Edgell et al. in the early 1990s used Youden test in their vali- tachlorophenol, picloram, 2,4,5-T, and 2,4,5-TP) and 5 pesticides
dation studies on ofcial USEPA (United States Environmental (aciuorfen, chloramben, dalapon, dinoseb, and 4-nitrophenol)
Protection Agency) methods. Firstly, an interlaboratory method that were only qualitatively assessed in the National Pesticide Sur-
validation study was conducted on EPA Method 531.1. This method vey because of recognized method imprecision. The study design
refers to the measurement of N-Methylcarbamoyloximes and was based on Youdens nonreplicate plan for collaborative tests of
N-Methylcarbamates in water by HPLC with post column deriva- analytical methods. The waters were spiked with all chlorinated
tization, to determine the precision and mean recovery of 10 acids, each at 6 concentration levels, prepared as 3 Youden pairs.
carbamate pesticide compounds in reagent water and in nished Eight laboratories analyzed the water samples. Statistical analy-
drinking waters. The study was based on Youdens plan for col- sis performed with a USEPA computer program, which measured
laborative tests of analytical methods. Samples were spiked with recovery and precision for each of the 17 compounds and compared
10 carbamate pesticides at 6 concentration levels, as 3 Youden the performance of the method between water types. The of-
pairs. Eight laboratories analyzed the samples. Results were ana- cial method proved acceptable for the 12 NPS analytes recovered
lyzed using an EPA computer program, which measured precision quantitatively. Results for the 5 other NPS analytes were impre-
and recovery for each of the 10 compounds and compared the per- cise so these compounds are not included in the adopted method
formance of the method between water types. The method was [43].
acceptable for all analytes tested, whereas the results for the pooled In order to test for the presence of low pharmaceuticals in the
drinking waters were not signicantly different from the results for environment highly sensitive and selective analytical procedures
the reagent waters [40]. are required. Hollow ber-based liquid-phase microextraction (HF-
A joint U.S. Environmental Protection Agency/AOAC interlab- LPME) is a relatively new technique used in analytical chemistry
oratory method validation study was conducted on EPA Method for sample pre-treatment that offers high selectivity and sensitiv-
507. The method refers to the determination of nitrogen and ity compared to most traditional extraction techniques. This paper
phosphorus pesticides in nished drinking water by Gas chro- describes an extraction method using a polypropylene membrane
matography with a nitrogen-phosphorus detector, to determine supporting dihexyl ether for the analysis of three pharmaceuticals,
the mean recovery and precision for analyses of 45 nitrogen- or salicylic acid (SAC), ibuprofen (IBU) and diclofenac (DIC) in raw
phosphorus-containing pesticides in reagent water and nished and treated wastewaters followed by a HPLC/MSMS determina-
drinking waters. The study was based on Youdens nonreplicate tion. A full factorial design on the experimental conditions for the
plan for collaborative tests of analytical methods. Like in the HF-LPME extraction for three factors and two levels involving eight
previous method, samples were spiked with pesticides at 6 con- experiments (23 ), has been used to determine the effect and impor-
centration levels, prepared as 3 Youden pairs. Ten laboratories tance of the mentioned variables on the nal result. Factors tested
extracted the spiked waters with organic solvents and analyzed are the donor pH, acceptor pH and stirring time. As it can be seen,
an aliquot of each extract by GC. Results were analyzed using an the main factor is the acceptor pH for all the analytes. For DIC and
EPA computer program, which measured recovery and precision IBU, an increase in the acceptor pH leads to better results in the
for each of the 45 pesticides and compared the performance of extraction, while in the case of SAC leads to a decrease in the sig-
the method between water types. The ofcial method found to nal; so this pH value is a critical parameter. Stirring time seemed
be acceptable for all analytes tested except merphos, which ther- to be the less critical factor and always with a positive effect on
mally decomposed in the injection port of the gas chromatograph the extraction procedure. From the results obtained it is possible
[41]. to consider the acceptor pH as a critical factor that must be care-
A joint EPA-AOAC interlaboratory method validation study fully controlled, while the extraction procedure can be considered
was conducted on USEPA National Pesticide Survey Method 6, a robust extraction procedure [44].
which refers to the determination of ethylene thiourea (ETU) in The same group of authors, developed an electromembrane
nished drinking water by Gas chromatography with a nitrogen- extraction method, combined with a HPLC procedure using diode
phosphorus detector. The purpose of the study was to determine array and uorescence detection for the determination of six widely
and compare the mean recoveries and precision for determina- used non-steroidal anti-inammatory drugs (salicylic acid, ketoro-
tion of ETU in reagent water and nished drinking waters. As lac, ketoprofen, naproxen, diclofenac and ibuprofen) in wastewater

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050
G Model
CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS
8 E. Karageorgou, V. Samanidou / J. Chromatogr. A xxx (2014) xxxxxx

samples. The drugs were extracted from aqueous sample solu- 2. Conclusions
tions, through a supported liquid membrane. The proposed method
was successfully applied to urban wastewaters, whereas excel- The robustness/ruggedness study during method validation
lent selectivity was demonstrated as no interfering peaks were yields useful information concerning the performance of the
detected. The robustness study of the developed method was based method if minor changes in operating parameters occur unex-
on the procedure suggested by Youden. A design matrix with two pectedly. Since these changes may happen either during sample
factors in eight experiments was used where the +1 and 1 levels preparation or during analysis, a high number of experiments seem
corresponded to slight modications in the more critical variables: to be necessary, taking into consideration the most important
pH values of donor and acceptor phases and stirring time. A sig- parameters. A useful test to avoid the difcult task of perform-
nicance t-test was used to determine whether variations have a ing too many experiments can be the Youden approach which is a
signicant effect on the result. The t values were compared with fractional factorial design. By this approach the number of experi-
the corresponding critical t values (n = 4) at 5% signicance level ments is limited, however we have to keep in mind that interactions
and three degrees of freedom. The results obtained showed that the between the different factors cannot be detected. Till now this use-
procedure can be considered robust against the considered factors ful tool has been applied in many methods covering a wide range
for all the analyzed compounds [45]. of applications.
A three phase-hollow ber-based liquid phase microextraction
combined with a HPLC procedure using diode array and uo- Conict of interest
rescence detection has been developed from the same group of
authors in the two previous methods, for the determination of Authors have declared no conict of interest.
four widely used sulfonamides: sulfadiazine, sulfamerazine, sul-
famethazine, sulfamethoxazole and their main metabolites. The
References
proposed method was applied to the determination of the analytes
in environmental water samples (surface, tap and wastewater). [1] International Organization for Standardization, ISO 8402:1994, Quality-
Robustness study performed based on the procedure suggested by Vocabulary, ISO, Geneva, Switzerland, 2005.
Youden, exactly in the same way as described in [44,45]. Factors [2] Eurachem Guide: The Fitness for Purpose of Analytical Methods. A Laboratory
Guide to Method Validation and Related Topics, 1998.
tested were pH values (4.5 and 3.5 for donor phase and 12.5 and [3] M. Thompson, S.L.R. Ellison, R. Wood, Pure Appl. Chem. 74 (2000) 835.
11.5 for the acceptor phase). From the results obtained, method [4] ICH, Q2(R1). Validation of Analytical Procedures: Text and Methodology, ICH,
can be considered robust against examined factors for all target Geneva, Switzerland, 2005.
[5] U.S. Food and Drug Administration, Draft Guidance, 2000.
analytes [46]. [6] European Commission Decision 2002/657/EC, Off. J. Eur. Commun. 221 (2002)
The occurrence of cyanobacterial blooms and the production 8.
of toxins as Cylindrospermopsin (CYN), in aquatic environments [7] L. Cuadros-Rodriguez, R. Romero, J.M. Bosque-Sendra, Crit. Rev. Anal. Chem. 35
(2005) 57.
were increasing in many regions of the world due to progressive [8] http://www.vub.ac.be/fabi/tutorial/robust/guideline.doc (accessed 05.11.13).
eutrophication of water bodies. In this work, an analytical method [9] B. Dejaegher, Y. Van der Heyden, Acta Chromatogr. 21 (2009) 161.
has been developed an optimized for the determination of CYN [10] A.G. Gonzalez, M.A. Herrador, Trends Anal. Chem. 26 (2007) 227.
[11] W.J. Youden, Statistical Techniques for Collaborative Tests, Association of Of-
from Aphanizomenon ovalisporum cultures. The analytical proce-
cial Analytical Chemists (AOAC), Washington, DC, USA, 1967.
dure is based on solvent extraction followed by a purication step [12] W.J. Youden, E.H. Steiner, Statistical Manual of AOAC, Association of Ofcial
with graphitized cartridges and CYN quantication by LCMS/MS. Analytical Chemists, AOAC-I, 1975.
[13] E. Hund, D.L. Massart, J. Smeyers-Verbeke, Anal. Chim. Acta 423 (2000) 145.
The extraction and purication steps were optimized using a two-
[14] A.G. Gonzalez, M.A. Herrador, Talanta 70 (2006) 896.
level full factorial design with replications according to Youden [15] R. Bonlio, E.C. Laignier Cazedey, M.B. de Araujo, H.R. Nunes Salgado, Crit. Rev.
approach. The three factors considered, relative to the SPE proce- Anal. Chem. 42 (2012) 87.
dure, were: the batch of the graphitized carbon cartridges, the ow [16] A.F. Forti, G. Campana, A. Simonella, M. Multari, G. Scortichini, Anal. Chim. Acta
529 (2005) 257.
rate of the sample through the cartridge, and the nal redissolved [17] C. Benetti, R. Piro, G. Binato, R. Angeletti, G. Biancotto, Food Addit. Contam. 23
water volume after SPE treatment, which permits its validation. The (2006) 1099.
effect of each considered factor was estimated as the difference [18] C. Fretz, V. Rouvinez, S. Knel, J.L. Luisier, R. Amad, Vitis 45 (2006) 37.
[19] J.M. Jurado, A. Alcazar, F. Pablos, M.J. Martin, Chromatographia 64 (2006) 223.
of the mean result obtained at the high level from that obtained [20] M. Muscarella, S. Lo Magro, C. Palermo, D. Cent, Anal. Chim. Acta 594 (2007)
at the low level. Once effects have been estimated, to determine 257.
whether variations have a signicant effect on the results, a signif- [21] R. Muniz-Valencia, S.G. Ceballos-Magana, R. Gonzalo-Lumbreras, A. Santos-
Montes, R. Izquierdo-Hornillos, J. Sep. Sci. 31 (2008) 727.
icance t-test is used, and the t-values were compared with the 95% [22] R. Muniz-Valencia, S.G. Ceballos-Magana, R. Gonzalo-Lumbreras, A. Santos-
condence level with the degrees of freedom from the precision Montes, R. Izquierdo-Hornillos, J. Sep. Sci. 31 (2008) 2303.
study for each concentration. Finally, the present method can be [23] A.F. Forti, G. Scortichini, Anal. Chim. Acta 637 (2009) 214.
[24] Z. Jandric, M.N. Rathor, J. Svarc-Gajic, B.M. Maestroni, J.J. Sasanya, R. Djurica, A.
considered as robust against the three factors (at the levels xed in
Cannavan, Food Addit. Contam. 28 (2011) 1205.
the study) [47]. [25] Z. Pineiro, E. Cantos-Villar, M. Palma, B. Puertas, J. Agric. Food Chem. 59 (2011)
In the last example, a new sample preparation procedure is 11683.
[26] M. Iammarino, A. Di Taranto, Eur. Food Res. Technol. 235 (2012) 409.
described, for the determination of aminopolycarboxylic acids in
[27] E.G. Karageorgou, V.F. Samanidou, J. Sep. Sci. 34 (2011) 1893.
river water. The procedure consists of the solid-phase extraction of [28] E.G. Karageorgou, V.F. Samanidou, I.N. Papadoyannis, J. Sep. Sci. 35 (2012) 2599.
the aminopolycaroxyllic acids on activated charcoal cartridges after [29] E.G. Karageorgou, A. Myridakis, E.G. Stephanou, V.F. Samanidou, J. Sep. Sci. 36
increasing the ionic strength and acidifying the sample. Chromato- (2013) 2020.
[30] E.G. Karageorgou, M. ArmeniI, I Moschou, V.F. Samanidou, Food Chem. 150
graphic technique used was GC with mass spectrometric detection. (2014) 328.
The robustness of the method was evaluated by analyzing data [31] T. Mirza, H.S.I. Tan, J. Pharm. Biomed. Anal. 25 (2001) 39.
from sample preparation procedure, after checking seven vari- [32] J.J. Berzas, J. Rodriguez, A.M. Contento, M.P. Cabello, J. Sep. Sci. 26 (2003) 908.
[33] J. Rodriguez Flores, J.J. Berzas Nevado, A.M. Contento Salcedo, M.P. Cabello Diaz,
ables according to Youden and Steiners robustness test. Factors J. Sep. Sci. 27 (2004) 33.
tested were: pH, NaCl mass, charcoal mass, eluent volume, reagent [34] J. Rodriguez Flores, J.J. Berzas Nevado, A.M. Contento Salcedo, M.P. Cabello Diaz,
volume, derivatization reaction time and derivatization reaction Electrophoresis 25 (2004) 454.
[35] J.J. Berzas Nevado, C. Guiberteau Cabanillas, M.J. Villasenor Llerena, V.
temperature. Results indicated that the proposed method was Rodriguez Robledo, J. Chromatogr. A 1072 (2005) 249.
robust against all examined variables, and for all target analytes [36] J.J. Berzas Nevado, M.J. Villasenor Llerena, C. Guiberteau Cabanillas, V.
[48]. Rodriguez Robledo, S. Buitrago, J. Sep. Sci. 29 (2006) 103.

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050
G Model
CHROMA-355083; No. of Pages 9 ARTICLE IN PRESS
E. Karageorgou, V. Samanidou / J. Chromatogr. A xxx (2014) xxxxxx 9

[37] M. Ramos Payan, M.A. Bello Lopez, R. Fernandez-Torres, J.L. Pirez Bernal, M. [43] K.W. Edgell, E.J. Erb, E.J. Wesselman, J.E. Longbottom, J. Assoc. Off. Anal. Chem.
Callejon Mochon, Anal. Chim. Acta 653 (2009) 184. 76 (1993) 1113.
[38] I. da Costa Csar, G.A. Pianetti, Braz. J. Pharm. Sci. 45 (2009) 235. [44] M. Ramos Payan, M. Angel Bello Lopez, R. Fernandez-Torres, M. Callejon
[39] M.R. Payan, M.A. Bello Lopez, R. Fernandez-Torres, J.A. Ocana Gonzalez, M. Mochon, J.L. Gomez Ariza, Talanta 82 (2010) 854.
Callejon Mochon, J. Pharm. Biom. Anal. 55 (2011) 332. [45] M. Ramos Payan, M. Angel Bello Lopez, R. Fernandez-Torres, M. Villar Navaro,
[40] K.W. Edgell, L.A. Biederman, J.E. Longbottom, J. Assoc. Off. Anal. Chem. 74 (1991) M. Callejon Mochon, Talanta 85 (2011) 394.
309. [46] M. Ramos Payan, M. Angel Bello Lopez, R. Fernandez-Torres, M. Villar Navaro,
[41] K.W. Edgell, E.L. Jenkins, V. Lopez-Avila, J.E. Longbottom, J. Assoc. Off. Anal. M. Callejon Mochon, J. Chromatogr. B 879 (2011) 197.
Chem. 74 (1991) 295. [47] R. Guzman-Guillen, A.I. Prieto Ortega, I. Moreno, G. Gonzalez, M.E. Soria-Daz,
[42] J.E. Longbottom, K.W. Edgell, E.J. Erb, V. Lopez-Avila, J. Assoc. Off. Anal. Chem. V. Vasconcelos, A.M. Camean, Talanta 100 (2012) 356.
76 (1993) 1113. [48] J.J. Jiminez, Anal. Chim. Acta 770 (2013) 94.

Please cite this article in press as: E. Karageorgou, V. Samanidou, J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.01.050