Clinical Research and Regulatory Affairs

ISSN: 1060-1333 (Print) 1532-2521 (Online) Journal homepage: http://www.tandfonline.com/loi/icrr20

Design of Experiments: Concept and Applications

of Plackett Burman Design

K. Vanaja & R.H. Shobha Rani

To cite this article: K. Vanaja & R.H. Shobha Rani (2007) Design of Experiments: Concept and
Applications of Plackett Burman Design, Clinical Research and Regulatory Affairs, 24:1, 1-23,
DOI: 10.1080/10601330701220520

To link to this article: http://dx.doi.org/10.1080/10601330701220520

Published online: 10 Oct 2008.

Submit your article to this journal

Article views: 128

View related articles

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=icrr20

Download by: [Universitaetsbibliothek Tuebingen] Date: 02 June 2016, At: 07:18

 Clinical Research and Regulatory Affairs, 24(1):1–23, (2007)
Copyright © Informa Healthcare
ISSN 1060-1333 print/1532-2521 online
DOI: 10.1080/10601330701220520

DESIGN OF EXPERIMENTS: CONCEPT AND APPLICATIONS

1532-2521
1060-1333
LCRR
Clinical Research and Regulatory Affairs
Affairs, Vol. 24, No. 1, April 2007: pp. 1–41

OF PLACKETT BURMAN DESIGN

K. Vanaja
Concept
K. Vanajaand
andApplications
R.H. ShobhaofRani
Plackett Burman Design

䡺 Al-Ameen College of Pharmacy, Hosur Road, Bangalore, India

Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

R.H. Shobha Rani 䡺 Al-Ameen College of Pharmacy, Hosur Road, Bangalore, India

䡺 Application of statistics has broadened far beyond its origin to various areas of research and one
among them is the design of experiments. Design of experiments gathers the maximum amount of
information in the lowest number of analyses. In order to optimize a process and maintain repeat-
ability, screening designs are performed. A two-level, mathematically derived multiples of four
screening design, popularly known as Plackett-Burman screening design, is thoroughly reviewed in
this paper. Application of this design in various areas of research such as formulation development
and analytical development has also been focused to explore its potential in screening unbiased esti-
mates of all main effects in the smallest design.

Keywords Plackett-Burman, Screening Design, Design of Experiments, Statistical

Designs

INTRODUCTION
Statistics is a mathematical science pertaining to the collection, analysis,
interpretation and presentation of data, which is applicable to a wide variety of
academic disciplines [1]. It is concerned with models of data from experi-
ments in which one compares the effects of two or more treatments. The
design of experiments involves all stages in the choice of which experiments
are conducted to test hypotheses. That includes the choice of experimental
subjects, operations to be carried out and measurements to be made, as well as
the choice of measuring instruments. All of that is strongly dependent on the
state of knowledge in the field and on the technology available to make mea-
surements. Many experiments are now feasible that could not be carried out
many years ago. Application of statistics has broadened far beyond its origin to
various areas of research, one of them being the design of experiments [2].

Address correspondence to Dr. Shobha Rani R.H., Professor and Head, Department of Pharmacy
Practice, Al-Ameen College of Pharmacy, Hosur road, Near Lal-bagh main gate, Bangalore-560027,
India. E-mail: [email protected]

1
 2 K. Vanaja and R.H. Shobha Rani

DESIGN OF EXPERIMENTS – AN OVERVIEW

Statistical experimental design, also known as design of experiments
(DOE), is the methodology of how to conduct and plan experiments in
order to extract the maximum amount of information with the lowest num-
ber of analyses [3]. A designed experiment is a tool or set of tools used for
gathering test data. Typical characteristics of an experimental design are
planned testing, data analysis approach, simultaneous factor variability and
scientific approach [4].
Advantages of the design of experiments:
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

• Helps to handle experimental error.

• Helps to determine the important variables that need to be controlled
and find the unimportant variables that need not be controlled.
• Eliminates the confounding of effects whereby the effect of design vari-
ables is mixed up.
• Helps to measure interactions, which is very important.
• Allows extrapolation of data and search for the best possible product
within the test variable ranges.
• Allows plotting graphs to depict how variables are related and what level
of variables give the optimum product. Use of statistical models shows us
the interrelationship between variables.

DOE has been applied in many functional areas, one being research to
quantify the inter-relationship between variables and to screen a large num-
ber of variables to identify important ones. Another area is product devel-
opment whereby it is used to improve products through the reformulation
and development of new products. Design of experiments is based on
sound and logical statistical principles. All principles are based on statistics,
that is proven by a reliable method, backed by trial and error. A formula for
a successful product starts with the experimenter’s knowledge of the prod-
uct and a good test design and ends with statistical modeling.
As quality is not created by chance, it requires a logical approach to
experimentation to understand which variables affect quality and consis-
tency. The solution is the design of experiments, a theme which runs right
through the FDA’s draft guidelines on PAT (Process Analytical Technol-
ogy). Applying DOE throughout a pharmaceutical process is the key to
quality by design – it maximizes the information content of a small number
of experiments. Pharmaceutical processes are complex systems affected by
many factors. DOE identifies optimal formulation and process conditions
for these systems and provides understanding of the underlying relation-
ships. This provides the basis for evaluating different scenarios (i.e. learn-
ing) throughout a product’s lifecycle. Traditional one-factor-at-a-time
 Concept and Applications of Plackett Burman Design 3

approaches do not work because they fail to take into account the factor
interactions.
With statistical DOE having numerous advantages it helps to identify
the optimum response. DOE is applied in experimental situations where
several independent variables potentially impact one or more response vari-
able [5]. The experimenter controls the independent variable in a designed
experiment, while the response variable is an observed output of the exper-
iment. Changing more than one variable simultaneously, rather than
changing one variable at a time, leads to effective results. Interactions
between variables can cause problems that none can see until change has
been made. The miracle of DOE is that several factors are varied simulta-
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

neously, yet each factor is evaluated independently.

Statistical DOE is an efficient procedure for setting up a detailed exper-
imental plan in advance of doing any experiments. Well-chosen designs
maximize the amount of information that can be obtained for a given
amount of experimental effort.
There are several steps common to most DOE methods. The first is to
define the problem to be solved. The second step is to list the factors that
might affect the way the process operates. The third step is to conduct exper-
iments to study the factors in different combinations. The last step is to
choose the combination that yields the best results. Process documentation
can then be used to train personnel in the most efficient method. Hence in
an experiment, at the initial stage, a researcher may have numerous control
factors that have to be studied. To determine initially which of these has an
impact on the response variable, a screening design is performed [6].

SCREENING DESIGNS
A ‘screening design’ refers to an experimental design which can be
applied when a large number of potential causative factors have to be exam-
ined, to identify the most important ones that may have an effect on one or
more responses of interest. This will reduce the number of factors to be
investigated in further experimentation. In order to eliminate unimportant
factors before investing time and money in a more elaborate experiment,
screening could be performed. The screening design has a number of valu-
able features: It helps to improve the quality control process by determining
the upper and lower control limits of a certain variable. Process can be
refined by identifying the influencing factors in a less expensive way [7].
Minimizing the number of experiments while maximizing information is
the ultimate goal [8]. Another feature is that the quality of the product can
be improved through a structured approach, which keeps the ideas and
information in an understandable and readable format. Results can be
checked in an efficient and reliable manner as it is a mathematical expression.
 4 K. Vanaja and R.H. Shobha Rani

Information gained can be used to optimize a process, and the repeatability

of a process can be maintained [9, 10].
The strategy [11] in all screening experiments is as follows:

1. Identification of the need to run the screening design.

2. Determination of the practicality of number of runs – the trade-off
between the information gained versus the expense of the experiments.
3. All the variables are noted and feasibility performed.

A screening run is the one which isolates those input factors that are most
important to the output. Output could be from the software used for
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

screening designs and the researcher’s own knowledge of the system, with
the cost factors in mind. This leads to the exclusion of those factors deemed
to be of minor significance to the desired output, based on the results of
the screening experiments.

PLACKETT-BURMAN SCREENING DESIGNS

A popular class of screening designs is the Plackett-Burman design
(PBD), developed by R.L. Plackett and J.P. Burman in 1946. It was designed
to improve the quality control process that could be used to study the
effects of design parameters on the system state so that intelligent decisions
can be made. Plackett and Burman (PB) devised orthogonal arrays are use-
ful for screening, which yield unbiased estimates of all main effects in the
smallest design possible. Various number or ‘n’ factors can be screened in
an ‘n + 1’ run PB design. A characteristic feature is that the sample size is a
multiple of four rather than a power of two (4k observations with k = 1,
2,……n). PB designs are used to investigate n–1 variables in n experiments
proposing experimental designs for more than seven factors and especially
for n × 4 experiments, i.e., 8, 12, 16, 20, etc., that are suitable for studying
up to 7, 11, 15, 19, etc., factors respectively. Such designs are known as satu-
rated designs. The main advantage of saturated designs is the minimum
number of observations needed to calculate an effect for a certain factor. A
selection of two-level Plackett-Burman designs is equal to the saturated frac-
tional factorial designs. This means that seven factors are analyzed with frac-
tional factorial (27-4) and with a PBD both requiring eight observations. To
study 11 factors a PBD is used with 12 runs, whereas the fractional factorial
designs require 16 observations. Thereby PBD require fewer experiments
than the highly fractionated factorial designs that include the same number
of factors. The projective property of the PB design is that it allows the
experimenter to follow up an initial design with runs which allow an effi-
cient separation of main effects and interaction effects [9, 11, 12, 13, and
14]. The disadvantage of PB design is that the aliasing pattern is much
 Concept and Applications of Plackett Burman Design 5

more complex, each main effect is aliased with every two-way interaction
not involving that effect. Lack of fit is difficult to assess, and first-order
effects may be confounded with interaction effects. PB designs are Resolu-
tion III designs with the attribute of requiring the lowest number of runs,
but do not allow the estimation of interactions between factors; it can iden-
tify the significant main factors that make up the possible significant inter-
actions. Further analysis of the important main factors would allow the
analyst to identify and estimate the significant interaction terms. Therefore,
the use of a Plackett-Burman design is appropriate for screening [8].
Various steps involved in a screening design [15] can be summarized as:
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

• Select the factors

• Define the levels for the factors
• Select the responses to be measured
• Generate a design matrix of PBD
• Randomize (block) and perform the experiments described in the exper-
imental set up
• Replicate the design
• Develop a model
• Statistical or graphical analysis of the effects
• Interpret and conclude from the statistical analysis
• Recommend possible improvements and if necessary apply higher resolu-
tion designs
• Build verification products

Selection of Factors, Factor Levels and Responses

The first critical step in a screening design is selecting the factors, defin-
ing their levels and responses which have to be measured. Selection of fac-
tors is based on the experience of the researcher, who chooses one or
several responses and discovers where in the experimental space, derivatives
of the response in respect of each factor can be obtained [12]. Appropriate
knowledge of the physical properties of factors and a thorough understand-
ing of experimental design increases the information gained from the
experiments. Factor levels have to be achievable in combination with each
other, capable of set and reset reproducibly between different designs of
experiments [15]. If the factor range is too small, the variation of the
response is too small and influence of the experimental error on the
response is more; on the other hand if the range is too large, the first degree
model used to interpret the results of the experimental design would not be
valid. Therefore, in a screening PBD, a range between low and high levels
of each factor is generally small in comparison with the ranges used during
the optimization phase [12].
 6 K. Vanaja and R.H. Shobha Rani

From the experiments, a number of responses can be determined

depending on the experimenter’s requirement and area of research. For
instance, an experimenter would be interested in responses describing
quantitative responses such as drug content, drug release in the area of for-
mulation development or peak areas, and peak heights in chromatographic
methods. Qualitative response factors such as resolution and relative reten-
tion can also be considered [16].

Design Matrix
After selection of factors and their levels, a design matrix is generated
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

(for example, see Table 1). Plackett Burman Screening Design (PBSD) is
used to indicate two level fractional factorials, although more levels are pos-
sible. It allows efficient estimation of main effects of all factors being
explored. An example of PBD with 12 runs and 11 factors is presented in
Table 1. The columns represent factors with degrees of freedom equal to
the number of levels in the column. The elements in the columns specify
the levels high level (+1) and low level (−1) to be set for factors for the
given experiment. The rows in the matrix contain the process run. Out of
these runs, their effect on the variable can be studied. The number of runs
goes down columns. The post processing of results and the main effect of
interaction can be determined. Any variable having an effect on another
variable can be used. The prediction of changes is more accurate in the
presence of more variables. PBD is particularly cyclical and the table of 11
factor 12 experimental design is obtained from a first line which describes
the first experimental run and in Table 1 [+ − + − − − + + + − +] where the
‘+’ sign denotes the factor in its high level and the sign ‘−’ denotes the fac-
tor in its low level. The second line is obtained by moving the minus sign at
the far right to the beginning of the next line and sliding the rest of the

TABLE 1 An example of a 12-run Plackett Burman design

Trial X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11

1 + − + − − − + + + − +
2 + + − + − − − + + + −
3 − + + − + − − − + + +
4 + − + + − + − − − + +
5 + + − + + − + − − − +
6 + + + − + + − + − − −
7 − + + + − + + − + − −
8 − − + + + − + + − + −
9 − − − + + + − + + − +
10 + − − − + + + − + + −
11 − + − − − + + + − + +
12 − − − − − − − − − − −
 Concept and Applications of Plackett Burman Design 7

signs from the previous row one place along. The rest of the matrix is filled
appropriately in the same way. Experiments 2–12 are obtained by noting all
the cyclical permutations of this line. It is possible to verify that each factor
is examined at six high and six low levels, and it is also possible to verify that
the main factor is not confounded when the effects are determined. With
the matrix a mathematical formula is obtained which gives the effect of
change in variable on the total model.
Modified PBDs are also reported [17] to examine the influence of a
number of selected factors at three levels. The three levels under consider-
ation were a nominal level and two extreme levels. The nominal level (level
0) is the level of the factor and the extreme levels (levels −1 and +1) are the
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

levels deviating from the nominal one. The first type is a centered PBD,
which is carried out at the extreme levels, i.e. −1 and +1, but is called ‘cen-
tered’ because the values of the extreme levels are smaller and larger than
the nominal value. The nominal value is centrally situated in the interval
between the two extreme levels. The second type of a similar kind is the
reflected PBD consisting of the principle two two-level designs. The first is
performed with the nominal level and one extreme level for each factor,
and the second with the nominal level and the other extreme level. In a
reflected PBD the levels ‘0’ and ‘+1’ are entered once for (−) and (+); then
one enters the levels ‘0’ and ‘−1’ for (−) and (+) [17].
In a PBD which is saturated, ‘n’ factors can be screened in ‘n + 1’ run
that means 11 factors can be studied with only 12 trials. The result obtained
would not be sufficient; as the mathematical model assumes only that the
responses vary linearly in respect to each factor. Such a response obtained
will be illustrated by a straight line between the two measured responses. In
certain cases response could also be a curve which remains inside the mea-
sured result or a curve which jumps outside the measured response result.
Then, it is imperative to add a central point or points to classical ‘PBD’
designs which are then called “Augmented Plackett and Burman” designs.
Several central points are generally added instead of one, to obtain an esti-
mate of the method repeatability [18]. All these characteristic features of
PBD give the researcher an important measure in controlling a process.

Randomization and Replication

Design of experiments or a screening design is normally performed in a
random sequence. For practical reasons experiments are sorted by one or
more factors. But randomization is the running of test parts in random
order. It is the opposite of running tests systematically. The running of tests
randomly prevents the confounding of effects that can happen when tests
are run in a standard order [15, 19]. On the other hand replication is
the running of one or more test parts under the same conditions. Each
 8 K. Vanaja and R.H. Shobha Rani

combination of factor levels in the design is normally run more than once.
Repeated examination of test parts builds confidence in the test results and
enables us to compute the pure error and statistical significance of test
results. The variability of measurements can be computed within each
unique combination of factor levels. This variability gives an indication of
the random error in the measurements (e.g., due to uncontrolled factors,
unreliability of the measurement instrument, etc.) because the replicated
observations are taken under identical conditions (settings of factor levels).
Once the PBSD matrix is obtained, the researcher has completed per-
forming the experiments, and the responses are recorded, the next step
would be to analyze the results, which is the vital part in understanding the
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

estimated response surface.

STATISTICAL ANALYSIS OF RESULTS

Regression Coefficients and Analysis of Variance
Upon collection of all the runs and calculated responses the calculation
of regression coefficients is initiated. PBDs are based on Hadamard matri-
ces [18] and the results are interpreted using the first degree polynomial
model:

y = a 0 + a1x1 + a 2 x 2 + ........... + a11x11

Where ‘y’ stands for the predicted response, x1 – x11 stands for the settings
(factors), a1 – a11 are the respective coefficients and a0 stands for the inter-
cept of mean. As PBD is a saturated design, the main effect estimates do not
show standard errors and all the degrees of freedom are used to estimate the
factor main effects. After the estimation of the factor regression coefficients,
the determination of significant factors affecting the dependent variables of
interest (responses) is followed by the analysis of the variance (ANOVA).
In order to determine which of the factors significantly affected
the dependent variable of interest ANOVA is performed. ANOVA presents
the sum of squares (SS) which is used to estimate the factor main effects,
the F-ratios (F) as the ratio of the respective mean-square effect (MS) and
the mean-square error. ‘P’-probability values indicate the significant factors
affecting the response.

Pure Error and Lack of Fit

If the experimental design is at least partially replicated, then one can
estimate the error variability for the experiment from the variability of the
replicated runs. Since the measurements are taken under identical condi-
tions, that is, at identical settings of the factor levels, the estimate of the
 Concept and Applications of Plackett Burman Design 9

error variability from those runs is independent of whether or not the

"true" model is linear or non-linear in nature.

Graphical Representation of Results

Diagnostic Plots of Residuals
To start with, before accepting a particular "model" that includes a par-
ticular number of effects, the distribution of the residual values is exam-
ined. These are computed as the difference between the predicted values
(as predicted by the current model) and the observed values. In this plot
the actual residual values are plotted along the horizontal X-axis; the verti-
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

cal Y-axis shows the expected normal values for the respective values, after
they are rank-ordered. If all values fall onto a straight line, then one can be
satisfied that the residuals follow normal distribution. A graphical represen-
tation is shown in Figure 1 as generated by the software Design Expert V.
6.0.5 © Stat ease Inc., U.S.A.

Pareto Chart of Effects

The Pareto chart of effects is often an effective tool for communicating
the results of an experiment, in particular to laymen. In this graph, the
ANOVA effect estimates are sorted from the largest absolute value to the
smallest absolute value. The magnitude of each effect is represented by a

FIGURE 1 Graphical representation of diagnostic plot of residuals.

 10 K. Vanaja and R.H. Shobha Rani
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

FIGURE 2 Representation of a Pareto chart.

column, and often, a line going across the columns indicates how large an
effect has to be (i.e., how long a column must be) to be statistically signifi-
cant (Figure 2).

Detection of Significant Sources

Another useful, albeit more technical summary graph, is the normal
probability plot of the estimates. Because the number of degrees of free-
dom for the error term is small in saturated designs, the power of classical
ANOVA will be too low. For this reason a graphical tool, the half normal
plot, can be used to which the algorithms of Length and Dong are applied
to identify possible significant effects and to estimate standard deviation of
the effects. Significant effects in half normal plots are detected through
visual inspection. A graphical representation is shown in Figure 3 as gener-
ated by the software Design Expert V. 6.0.5 © Stat ease Inc., U.S.A. The
slope of the line through the effects assumed to be non-significant gives an
estimate of the standard deviation (σ) of the error [12].

Interaction Plots
A general graph for showing the means is the standard interaction plot,
where the means are indicated by points connected by lines as shown in
Figure 4. This plot is particularly useful when there are significant interac-
tion effects in the model.

With the insight to Plackett-Burman screening design, we further have

reviewed its application in various areas of pharmaceutical research focus-
ing mainly on formulation development and analytical development.
 Concept and Applications of Plackett Burman Design 11
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

FIGURE 3 A graphical representation of a Half Normal plot.

APPLICATIONS
The aim of pharmaceutical development is to design a quality product
and its manufacturing process to consistently deliver the intended perfor-
mance of the product. The information and knowledge gained from phar-
maceutical development studies and manufacturing experience provide
scientific understanding to support the establishment of the design space,
specifications, and manufacturing controls [20].
During pharmaceutical development studies leading to enhanced knowl-
edge of product performance over a wider range of material attributes, pro-
cessing options and process parameters, the inclusion of design of experiments
provides an opportunity to demonstrate a higher degree of understanding of
material attributes, manufacturing processes and their controls [20].

In Formulation Development
In the pharmaceutical industry, the most concentrated use of screening
designs is mainly found in the product and process development functions,
 12 K. Vanaja and R.H. Shobha Rani
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

FIGURE 4 A graphical representation of one factor plot.

including analytical method development, because of the increasing com-

plexity in the optimization of formulation and validation of analytical meth-
ods. In the evolving environment of research, experimental screening
designs continues to gain acceptance as a valuable development tool. A
screening design such as PBSD being a statistical design helps in isolating
and understanding the most significant factors affecting the desired
response. Hence, applications of PBSD in various areas of formulation
development such as transdermal drug delivery, osmotic drug delivery, lipo-
somal drug delivery, enteric coating, tablet coating including dissolution and
excipient-drug interaction in solid dosage form have been reviewed here.
An upcoming area of formulation is transdermal drug delivery system
(TDDS), in which a particular type of TDDS known as drug-in-adhesive
matrix system has been investigated. As sample preparation is a pivotal part
of an analytical method in any formulation system, a 10 run experimental
PBD was used to study the effect of sample solvent strength, sample solvent
volume, shaking time and shaking frequency on extraction variations. Main
effect plots and interaction plots were generated using the software
Minitab. Further, in order to obtain an optimized solvent strength and sol-
vent volume, 4 × 5 general factorial designs were performed [21].
Plackett Burman design was employed to formulate chitosan films for
screening the process and formulation variables that produced a significant
effect on the flux of 5-fluorouracil and indomethacin. Further permeation
of both drugs across additional films was prepared according to central
 Concept and Applications of Plackett Burman Design 13

composite design, and the data obtained was used for generating equations
for optimizing the film formulation [22].
A study employing PBSD was conducted by S.V. Sastry and M.A. Khan to
screen the formulation variables and identify the main effects of the vari-
able in the development of the Atenolol osmotically controlled delivery sys-
tem coated with cellulose acetate pseudolatex. Seven factor, 12-run PBD
was generated using X-stat®. Factors were analyzed via the polynomial
equation using ANOVA, and the significance of the ratio of mean square
variation due to the regression co-efficient and residual error was tested.
From the study, it was concluded that PBD provided the magnitude and
direction of the main effect of several formulation and process variables
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

[23]. The same study was continued with a secondary objective to identify a
suitable antioxidant for incorporating in the formulation as the drug
undergoes metal-catalyzed oxidative degradation, a seven-factor, 12-run
Plackett-Burman screening design employed to study the main effects of
the amounts of Polyox N10 and N80, Carbopol 934P and 974P, sodium
chloride, orifice size, and % coating weight gain. The response variable was
the cumulative percent of drug released in 12 h, Y3, with constraints on lag
time Y1 and time for 50% drug released Y2 amount of sodium chloride [24].
Critical processing variables for coating actives onto tablets were identi-
fied through statistically designed experiments. Six process variables investi-
gated were inlet airflow, pan speed, inlet air temperature, coating time,
atomization pressure and fan pressure. The potential of these variables to
significantly affect content uniformity and percent drug loading were exam-
ined using PBD. Although a simple linear model screened the critical pro-
cess variables, the authors suggest the necessity of a higher polynomial
model to optimize the current design [25].
Delivery systems mainly use liposomes which can entrap or anchor
many substances. Design of an efficient liposomal preparation is a multivari-
ate procedure in which many factors could affect the desired liposomal
properties. Isolation of the significant from the insignificant factors is time
consuming and costly, and it would impossible for a formulator to guess the
effects of factors on the final results. L.L. Yannis [14] has studied the appli-
cation of PBD to extract valuable information for the efficient design of
liposomes entrapping vitamins with the lowest number of experiments.
Liposomal formulations of riboflavin were prepared with the stabilization
ratio as the response variable. 11 factors affecting their effectiveness as stabi-
lization systems were identified and studied. Important factors were identi-
fied using the Pareto chart of effects; a multiple linear equation was used to
predict each response for new factor settings; and significant factors affect-
ing dependent variables were analyzed by ANOVA. Statistical results
obtained proved the efficiency of PBSD in identifying the significant factors
and halved the number of factors making the drawing of conclusions easier.
 14 K. Vanaja and R.H. Shobha Rani

Another study employed PBD to develop and optimize a novel

cross-linked calcium-aluminium-alginate-pectinate oilisphere complex for
in vitro-specific release of Mentha Piperita. Different levels of alginate and/
or pectin were exposed to varying reaction times in an aqueous calcium
chloride/aluminium chloride/aluminium sulfate crosslinking solution.
Matrix resilience, hardness, total fracture energy, erosion, and pH-dependent
in vitro release were the responses measured from the 14 formulation run.
The total fracture energy values generated from PBD were further pre-
dicted using a neural model as the application of Artificial Neural Networks
(ANN) in formulation design and development is increasing. The mean
differences between the experimental and predicted values for total frac-
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

ture energy obtained from PBD and neural model showed no significant
differences. The measured total fracture energy response was closely
related to the predicted values of the optimization formulation, demon-
strating the feasibility of optimization procedures in developing the cross-
linked matrix [26].
The dissolution test has emerged in the pharmaceutical field as a very
important tool to characterize drug product performance. Statistical exper-
imental design methodologies have been evaluated to study the influence
of critical dissolution test variables on ketoprofen release from an
extended-release model dosage form. A duplicate 4 run PBD was gener-
ated, and statistical analysis of experimental data was performed. An experi-
mental domain was set up considering the factors as per FDA guidelines,
with pH, volume of dissolution medium and stirring speed as the factor vari-
ables and dissolution efficiency as the response variable. Results were ana-
lyzed graphically to point out the important factors affecting the considered
response and help select the better factor: pH and paddle rotation speed
were found to be significant and medium volume was not significant to min-
imize the response. It was concluded that PBD was suitable to point out the
significant factors affecting standard deviation of the drug in the consid-
ered experimental domain leading to optimal repeatability of the dissolu-
tion experiment [27].
A seven-factor, 12-run Plackett-Burman screening design was employed
to evaluate the main effects of homogenization time X1, rate of water addi-
tion X2, amount of polymer X3, amount of precipitating solution X4, con-
centration of electrolytes X5, compression pressure X6, and the
concentration of lubricant X7 on the rate of drug release preparation of
extended release naproxen tablets with Eudragit L100-55. The mathemati-
cal relationship for the percent of naproxen dissolved in 12 h (Y5) with var-
ious factors yielded a polynomial equation [28].
A Plackett-Burman screening design and 23 factorial design were
employed to study how drug type, microcrystalline cellulose (intra or extra),
filler type (lactose or dicalcium phosphate), disintegrant type (sodium
 Concept and Applications of Plackett Burman Design 15

starch glycolate or croscarmellose sodium), proportion of magnesium stear-

ate, and impeller speed affect tablet hardness, disintegration time, and dis-
solution using two model drugs which were chosen based on their
solubility: metoprolol tartarate and hydrochlorothiazide [29].
A Dual Controlled Gastrointestinal Therapeutic System of Salmon Cal-
citonin was formulated, and screening of process and formulation variables
was carried out using PB design. A seven-factor, 12-run Plackett-Burman
screening technique was employed to evaluate the effects of orifice size,
coating level, amounts of sodium chloride, Polyox® N10 and N80 and
Carbopol® 934P and 974P on drug release. Response variables were cumu-
lative percent released in 24 hr with constraints on time for 25% and 50%
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

drug release. Factors showing maximum influence on drug release were

amounts of Carbopol® 934P and Polyox® N10 in the drug layer, orifice size
and coating level showing negative effects [30].
The Plackett-Burman screening experiment was initially used in evaluat-
ing multiple design features to predict those that have a statistically signifi-
cant effect on the sublimation rate. The results from the Plackett-Burman
screening experiment reflected that vial diameter has the greatest impact
on the sublimation rate [31].
Plackett-Burman designs were applied as a screening method to evalu-
ate the most significant factors with the fewest experiments. Central com-
posite 2(3)+ star designs were performed to evaluate the response surfaces.
The methods were fully validated and were applied successfully to the deter-
mination of indapamide in pure and pharmaceutical forms with accuracy
and precision [32].
Identification of stabilizing and destabilizing effects of excipient-drug
interaction in solid dosage form was carried out using PBD. Compatibility
of pyridoxial hydrochloride with tableting excipients was investigated by iso-
thermal stress testing of multicomponent mixture and DSC. A compatibility
screening procedure efficiently detected positive and negative drug-excipi-
ent interaction [33].
To optimize the process parameters in fluidized bed granulation, a
Plackett-Burman design was applied to screen the inlet air temperature, the
inlet flow rate, the spray rate, the nozzle air pressure, the nozzle spray diam-
eter, and the nozzle position. The Plackett-Burman design showed that the
key process parameters were the inlet flow rate and the spray rate and prob-
ably also the inlet air temperature. In continuation, a fractional factorial
design (25-2) was applied to screen the remaining parameters plus the noz-
zle aircap position and spraying time interval. The fractional factorial
design showed that the nozzle air pressure was also important. As the target
values for the granule yield and the geometric mean granule size were
reached during the screening experiments, the authors considered that fur-
ther optimization was not necessary [34].
 16 K. Vanaja and R.H. Shobha Rani

Arrhenius equation and Plackett-Burman design were applied for stability

testing and drug-excipient compatibility of formulations, respectively. Plackett-
Burman experimental design was used to investigate the compatibility of
ascorbic acid with various syrup excipients. A 20-run Plackett-Burman design
was used to study the effect of excipients on the stability of ascorbic acid.
The Plackett-Burman design showed the ability to identify both destabiliz-
ing and stabilizing factors. Thus, it was recommended for preformulation
compatibility study. Authors concluded that these methods provide a help-
ful tool for pharmaceutical scientists to develop formulations [35].
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

In Analytical Development
Drug development and post-approval of drug products necessitates the
generation of reliable and accurate data using analytical methods which
include chromatographic and non-chromatographic techniques. Method
validation is a regulatory requirement, and performance parameters are
used to evaluate the performance of analytical methods. Ruggedness is a
measure of the reproducibility of test results under normal operational
conditions from laboratory to laboratory and from analyst to analyst. The
purpose of ruggedness testing is to determine the variables affecting the
susceptibility of the method and a way to control it. In this regard, the
American Society for Testing and Materials (ASTM) guidance recommends
the use of Plackett-Burman screening design to determine the ruggedness
of the analytical method [36].
Robustness studies of racemic citalopram by capillary electrophoresis
were conducted in a pharmaceutical formulation using PBD. A design
matrix consisting of 15 experiments for seven internal parameters with sta-
tistical treatment was performed. Results of the levels variation effects were
checked on critical electrophoretic responses such as resolution between
enantiomers, efficacy of each compound and corrected peak areas for each
analyte. Application of PBD provided a fast and sensitive method for separa-
tion and quantification of chiral isomers in the racemic mixtures [37].
With the goal of obtaining an efficient and fast separation, a 8-run PB
matrix was used to carry out robustness testing of a capillary electrophoretic
method developed for enantio resolution of salbutamol. Three experi-
ments at the central level of each factor were also included to test the model
linearity and obtain an estimate of experimental variance. Considered factors
affecting the response were evaluated by ANOVA, statistical analysis of co-
efficients and graphic analysis [38].
Plackett-Burman experimental design was studied in the chiral separa-
tion of clenbuterol using capillary electrophoresis (CE). This saturated frac-
tional design approach was used to simultaneously investigate the variables
such as pH, cyclodextrin concentration, electrolyte ionic strength, methanol
 Concept and Applications of Plackett Burman Design 17

concentration, and injection time, each at three levels. It was concluded

that Plackett-Burman experimental design offers of rapid means for testing
robustness of chiral CE methods or for the simultaneous screening of
experimental variables [39].
Two-level screening designs, such as Plackett-Burman and fractional fac-
torial design, were applied for the robustness testing. The concentrations of
ciprofloxacin and ciprofloxacin impurity C in the working standard solu-
tion for ruggedness testing were approximate, in order to monitor the
responses to small changes in the process variables. Testing of the stability
of ciprofloxacin in pharmaceutical preparations showed that the proposed
method can be successfully applied in studies of the stability of ciprofloxa-
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

cin and that ciprofloxacin impurity C is the main degradation product of

ciprofloxacin [40].
Robustness tests were performed on a HPLC assay to determine simulta-
neously ketoconazole and formaldehyde in an anti-dandruff shampoo. A
PBD with 16 experiments for determining formaldehyde was chosen to esti-
mate the effects of 11 real factors and four dummies on four responses, i.e.
concentration to formaldehyde according to peak area (Fa), tailing factor
(Asf), the retention factor (k), of the formaldehyde peak, and the resolu-
tion between reagent and the formaldehyde peaks (Rs). The same was also
examined with PBD having six factors and five dummy factors for ketocona-
zole on the response, i.e. content of ketoconazole, retention factor (k), tail-
ing factors (Asf), of ketaconazole peak. Statistical analysis revealed that the
method was robust and could be used for fast routine analysis [41, 42].
A high-performance liquid chromatography stability study of malonyl-
coenzyme A was performed using statistical experimental designs. Plackett-
Burman screening design was first used for selecting the most important
variables, with which a central composite design was constructed. The use-
fulness of this approach was demonstrated by obtaining kinetic data from
the mathematical function and by the evaluation of the cessation of the reac-
tion procedure in the activity assay of acetyl-coenzyme A carboxylase [43].
A three-step method development/optimization strategy for HPLC
assay/impurity methods for pharmaceuticals using Plackett-Burman design
was described. The purpose of the 16-injection set experiments was to evalu-
ate nine method factors with regard to method precision, accuracy, sensitiv-
ity and specificity. The results provided logical justifications in selecting
method parameters such as column temperature, detection wavelength,
injection volume, and sample solvent. In data analysis, instead of the tradi-
tional mathematical manipulations, graphical methods were used to exam-
ine and present data by creating the so-called main effect plots [44].
Two-centered PBD and reflected PBD were performed in a 12-run
experiment having six factors and five dummies on the responses, i.e. nor-
malized effects and critical effects on the tailing factor of a drug substance.
 18 K. Vanaja and R.H. Shobha Rani

Comparison of the results from both approaches showed that the normal
probability plots and statistical interpretation gave similar results. The
authors concluded that the use of dummy factors in PBD seems to be
appropriate to estimate the experimental error in a design [45].
An optimization step with PBD of 16-injection set experiments was
investigated with the purpose of evaluating nine method factors with regard
to method precision, accuracy, sensitivity and specificity. The design was
selected based on the robustness of the testing strategies. The results were
analyzed graphically by the generation of the main effect plots. Optimized
method conditions were obtained by analyzing the response data and with
the help of the response optimizer [46].
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

A pressurized fluid extraction (PFE) and normal-phase-high perfor-

mance liquid chromatography (NP-HPLC) method was proposed for the
determination of additives in polyethylene films. A study of pressurized
fluid extraction PFE variables was performed using a Plackett-Burman (PB)
experimental design for screening and a central composite design (CCD)
for optimizing the main variables obtained from the Pareto charts. The
optimized method was applied to determine additives in several polyethyl-
ene films [47].
Supercritical Fluid Extraction (SFE) is reported to be useful in the
selective removal of analytes in different types of samples, minimizing sam-
ple handling and reducing the use of environmentally toxic solvents. The
use of SFE in removing N-nitrosamines from different samples of tobacco
and food was investigated using PBD to study the influence of several
parameters during the process of recovery. A 12-run randomized PBD at
two levels was performed to study the influence of CO2 pressure, extraction
temperature, static and dynamic extraction time, restrictor temperature
and volume of modifier on evaluated responses, i.e. mass loss and recovery.
Pressure, extraction temperature, and dynamic extraction time were found
to be significant factors as revealed by normal probability plot. Six factors
reduced to three led to the inclusion of these three factors in a further
higher order design such as central composite design [48].
Plackett-Burman screening design with fold-over was carried out to dis-
tinguish the significant factors affecting the solid phase extraction (SPE)
procedure. As this part occurred in the determination of only three statisti-
cally relevant parameters, requiring consecutive optimization, a rotatable
central composite design was applied to define the response surface as a
function of the significant parameters and to choose the optimal conditions
for the SPE in phase II of this study [49].
Plackett-Burman design for screening and a central composite design
(CCD) for optimizing the significant variables were applied for the determi-
nation of fuel dialkyl ethers and BTEX in water using headspace solid-phase
microextraction and gas chromatography-flame ionization detection. The
 Concept and Applications of Plackett Burman Design 19

optimized method was applied to the analysis of the surface waters of the
Gipuzkoa (North Spain) rivers, marinas and fishing harbors [50].
Established as a highly efficient separation technique alternative to
HPLC, Micellar electrokinetic chromatography (MEKC) is a separation
based on the differential partitioning of analytes between a micellar phase
and a surrounding electrolyte. In order to study the influence of experi-
mental parameters on separation efficiency, selectivity and retention factor
of MEKC, PBSD strategy was adopted. It was applied to the optimization of
the separation of phenols and chiral separation of (+)-1-(9-anthryl)-2-propyl
chloroformate (APOC)-derivatized amino acids. PBD was applied to test the
ruggedness of the MEKC method developed to determine sildenafil, vard-
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

enafil and tadalafil. The influence of six factors at three different levels was
tested on the resolution and corrected peak area. The mean effects and
standard errors were calculated using the procedures of Youden and
Steiner. Statistical treatment of the analytical method proved to be rugged
to all the variations tested confirming the application of PBD in ruggedness
testing [51, 52].
An initial screening of the experimental parameters was performed
using a Plackett-Burman design to multivariate evaluation of separation
electrolytes for micellar electrokinetic chromatograpy. Significant parame-
ters were further evaluated using full factorial designs. The total resolution
of the separation is calculated and used as response. The proposed scheme
has been applied to the optimization of the separation of phenols and the
chiral separation of (+)-1-(9-anthryl)-2-propyl chloroformate-derivatized
amino acids [53].

Regulatory Framework
Food and Drug Administration has developed an innovative approach
for helping pharmaceutical manufacturing by encouraging the voluntary
development and implementation of innovative pharmaceutical develop-
ment, manufacturing and quality assurance. The agency considers PAT to
be a system for designing, analyzing and controlling manufacturing
through timely measurements of critical quality and performance attributes
of raw and in-process materials and processes, with the goal of ensuring
final product quality. Using the approach of building quality into products,
this guidance highlights the necessity for process understanding and oppor-
tunities for improving manufacturing efficiencies through innovation. In
order to support and justify flexible regulatory paths for innovation in man-
ufacturing and post-approval changes, scientific understanding of the rele-
vant multi-factorial relationships and means to evaluate the applicability of
this knowledge in different scenarios is required. This is achieved through
the use of multivariate mathematical approaches, such as statistical design
 20 K. Vanaja and R.H. Shobha Rani

of experiments, response surface methodologies, process simulation and

pattern recognition tools. To emphasize a strong link between product
design and process development an effective control of all critical quality
attributes is essential. The design and optimization of drug formulations
and manufacturing processes within the PAT framework [20] recommends
the following steps:

• Identify and measure critical material and process attributes relating to

product quality.
• Design a process measurement system to allow real time or near real time
monitoring of all critical attributes
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

• Design process controls that provide adjustments to ensure control of all

critical attributes.
• Develop mathematical relationships between product quality attributes
and measurements of critical material and process attributes.

In order to provide a quality decision, more rigorous statistical princi-

ples are required. This statistical principle would define acceptance criteria
for end-point attributes measurement and sampling strategies. Thus, when
used appropriately, these statistical design tools would enable the identifica-
tion and evaluation of product and process variables that may be critical to
product quality and performance.

CONCLUSION
Multivariate mathematical approaches, such as statistical design of
experiments, response surface methodologies, process simulation and pat-
tern recognition tools help to support and justify flexible regulatory paths
for innovation in manufacturing and post-approval changes. Screening
designs used in the early stages of research and development helps the
researcher to identify the significant factors for a large-scale simulation with
a relatively small number of runs. Performing an initial screening of inputs
in an experimental design of least resolution would fulfill researchers’ ulti-
mate objective of developing a model with emphasis placed on identifying
the active main effects and factor interactions. One such screening design is
PBSD, which is a saturated design where the number of experiments is the
number of factors minus 1. As PB design is a multiple of four rather than
power of two cyclic designs it generates the most information for the least
amount of work involving the fewest runs. It identifies the important factors
and the study’s main effects and all the data is used to estimate each effect.
By allowing to study a large number of factors involved and examining cur-
rent processes, PBD identifies the factors to be examined more intensely
allowing intelligent decision making for future research and development.
 Concept and Applications of Plackett Burman Design 21

ACKNOWLEDGEMENT
Authors wish to thank Principal and Management of Al-Ameer College of Pharmacy, India for all
the facilities rendered. We would also like to acknowledge Indian Council for Medical Research (ICMR)
for providing Senior Research Fellowship to one of the authors.

REFERENCES
[1] http://www.mathscareers.org.uk (accessed September 2006).
[2] Jacquez JA. Design of Experiments. Pergamon 1998; 3358(2): 259–279.
[3] Decaestecker TN, Lambert WE, Peteghem Carlos HV, Deforce D, Van Bocxlaer JF. Optimization of
solid-phase extraction for a liquid chromatographic-tandem mass spectrometric general unknown
screening procedure by means of computational techniques. J Chrom A 2004; 1056: 57–65.
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

[4] http://www.doesinc.com/knowledge.htm (accessed September 2006).

[5] Myers RW. Response Surface Methodology. In: Encyclopedia of Biopharmaceutical Statistics. Taylor
and Francis, 2003: 858–869.
[6] Webb TS, Bauer KW Kr. Comparison of analysis strategies for screening designs in large-scale com-
puter simulation models, Proceedings of the Winter Simulation Conference, Tew JD, Mannivannan S,
Sadowski DA, Seila AF, Eds, 1994: 305–311.
[7] Murphy RJ. Screening design. In: Encyclopedia of Biopharmaceutical Statistics. Taylor and Francis,
2003: 920.
[8] Trocine L, Malone CL. Finding important independent variables through screening designs: A
comparison of methods, Proceedings of the Winter Simulation ConferenceJoines JA, Barton RR,
Kang K, Fishwick PA, Eds, 2000: 749–754.
[9] http://www.itl.nist.gov/div898/handbook (accessed June, 2006).
[10] Prvan T, Street DJ. An annotated bibliography of application papers using certain classes of frac-
tional factorial and related designs. J Stat Plan Inf 2002; 245–269.
[11] Wass JA. Formal experimental design and analysis for immunochemical product development.
Part I. IVD Technology Magazine 1997, September.
[12] Nijhuis A, van der Knaap HCM, Jond SDE, Vandeginste BGM. Strategy for ruggedness tests in
chromatographic method validation. Anal chim Acta 1999; 391: 187–202.
[13] Plackett RL and Burman JP. The design of optimum multifactorial experiments. Biometrika 1946;
33:305–325.
[14] Yannis LL. A Plackett-Burnam screening design directs the efficient formulation of multicompo-
nent DRV liposomes. J Pharm Biomed Anal 2001; 26: 255–263.
[15] Leeuwen van JA. Vandeginste BGM, Kateman M, Mulholland, Cleland A. An expert system for the
choice of factors for a ruggedness test in liquid chromatography. Anal Chim Acta 1990; 228: 145–153.
[16] Heyden YV, Questier F, Massart DL. A ruggedness test strategy for procedure related factors:
experimental set-up and interpretation. J Pharm Biomed Anal 1998; 17: 153–168.
[17] Heyden YV, Luypaert K, Hartmann C, Massart DL, Hoogmartens J, Beer DJ. Ruggedness tests on
the High Performance Liquid Chromatography ass of the United States Pharmacoepia XXII for
tetracycline hydrochloride. A comparison of experimental designs and statistical interpretations.
Anal Chim Acta 1995; 312: 245–262.
[18] Goupy Jacques. What kind of experimental design for finding and checking robustness of analyti-
cal methods? Anal Chim Acta 2005; 544: 184–190.
[19] http://DFSS for Thermal Management Screening DoEs.htm (accessed September 2006).
[20] http://www.fda.gov/cber/guidelines.htm (accessed January 2007).
[21] Weiyong L, David N, Henrik TR, Kudan P, Tridarsh S. Sample preparation optimization for assay
of active pharmaceutical ingredients in a transdermal drug delivery system using experimental
designs. J Pharm Biomed Anal 2005; 37: 493–498.
[22] Vikas R, Kumar B, Dinesh G, Ashok KT. Sodium citrate cross-linked chitosan films:optimization as
substitute for human/rat/rabbit epidermal sheets. J. Pharm Pharmaceut Sci 2005; 8(1): 10–17.
[23] Sastry SV and Khan MA. Aqueous based polymeric dispersion: Plackett-Burman design for screen-
ing of formulation variables of Atenolol Gastrointestinal Therapeutic System. Pharm Acta Hel
1998; 73: 105–112.
 22 K. Vanaja and R.H. Shobha Rani

[24] Khan MA, Sastry SV, Vaithiyalingam SR, Agarwal V, Nazzal S, Reddy IK. Captopril gastrointestinal
therapeutic system coated with cellulose acetate pseudolatex: evaluation of main effects of several
formulation variables. Int J Pharm 2000 Jan 5; 193(2): 147–56.
[25] Rege BD, Gawel J, Kou JJ. Identification of critical process variables for coating actives onto tablets
via statistically designed experiments. Int J Pharm 2002; 237: 87–94.
[26] Siobanda W, Pillay v, Danckwerts MP, Viljoen AM, Vuuren SV and Riaz A. Experimental Design for
the Formulation and Optimization of Novel Cross-Linked Oilispheres Developed for In Vitro Site-
Specific Release of Mentha piperita Oil. AAPS PharmSciTech 2004; 5(1): 1–14.
[27] Furlanetto S, Maestrelli F, Orlandini S, Pinzauti S, Mura P. Optimization of dissolution test preci-
sion for a ketoprofen oral extended-release product. J Pharm Biomed Anal 2003; 32: 159–65.
[28] Zaghloul AA, Faltinek J, Vaithiyalingam SR, Reddy IK, Khan MA. Naproxen-Eudragit micro-
spheres: screening of process and formulation variables for the preparation of extended release
tablets. Pharmazie 2001 Apr; 56(4): 321–4.
[29] Li JZ, Rekhi GS, Augsburger LL, Shangraw RF. The role of intra- and extragranular microcrystal-
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

line cellulose in tablet dissolution. Pharm Dev Technol 1996 Dec; 1(4): 343–55.
[30] Rakhi BS, Mohammad N, Indra KR, Mansoor AK. A Dual Controlled Gastrointestinal Therapeutic
System of Salmon Calcitonin. II. Screening of Process and Formulation Variables. Clin Res Reg Aff
2004; 21(3–4): 231–238.
[31] Anthony C, Kerryann S. Statistical Evaluation of Vial Design Features That Influence Sublimation
Rates during Primary Drying. Pharm Res 2004; 21(3): 536–542.
[32] Rodriguez JC, Barciela J, Garcia S, Herrero C, Pena RM. Optimization of two flow-injection spec-
trophotometric methods for the determination of indapamide in pharmaceutical dosage forms.
J AOAC Int 2005; 88(4): 1148–54.
[33] Dürig T and Fassihi AR. Identification of stabilizing and destabilizing effects of excipient-drug
interactions in solid dosage form design. Int J Pharm 1993; 97(1–3): 161–170.
[34] Rambali B, Baert L, Thone D, Massart DL. Using experimental design to optimize the process
parameters in fluidized bed granulation. Drug Dev Ind Pharm. 2001 Jan; 27(1): 47–55.
[35] Srisagul S. Application of Arrhenius Equation and Plackett-Burman Design to Ascorbic Acid Syrup
Development. Naresuan University Journal 2004; 12 (2): 1–121.
[36] Donald DH and Mumtaz S. Development and validation of HPLC stability indicating assays. In:
Jens TC, Rhodes CT, Eds. Drug Stability, Principles and Practices. 3d ed Marcel Dekker, 2005:
329–384.
[37] Nevado JJB, Cabanillas CG, Llerena MJV, Robledo VR. Enantiomeric determination, validation
and robustness studies of racemic citalopram in pharmaceutical formulations by capillary electro-
phoresis. J Chrom A 200; 1072: 249–257.
[38] Gotti R, Furlanetto V, Cavrini V, Pinzauti S. Design of experiments for capillary electrophoretic
enantioresolution of salbutamol using dermatan sulfate. J Chrom A 2000; 875: 411–422.
[39] Rogan, MM Altria KD and Goodall DM. Plackett-Burman experimental design in chiral analysis
using capillary electrophoresis. Chromatographia 1994; 38(11–12): 723–729.
[40] Predrag S, et al., Ruggedness testing of an HPLC method for the determination of ciprofloxacin
2005. J Serb Chem Soc; 70(7): 979–986.
[41] Dejaegher B, Capron X, Smeyers-Verbeke J, Vander YH. Randomization tests to identify
significant effects in experimental designs for robustness testing. Anal Chim Acta 2006; 564:
184–200.
[42] Nguyen MNA, Tallieu L, Plaizier-Vercammen J, Massart DL, Vander HY. Validation of an HPLC
method on short columns to assay ketoconazole and formaldehyde in shampoo. J Pharm Biomed
Anal 2003; 32: 1–19.
[43] De Spiegeleer BM, Sintobin K, Desmet J. High performance liquid chromatography stability study
of malonyl-coenzyme A, using statistical experimental designs. Biomed Chromatogr 1989 Sep;
3(5): 213–6.
[44] Li W, Rasmussen HT. Strategy for developing and optimizing liquid chromatography methods in
pharmaceutical development using computer-assisted screening and Plackett-Burman experimen-
tal design. J Chromatogr A 2003 Oct 24; 1016(2): 165–80.
[45] Vander HY, Hartmann C, Massart DL, Michel L, Kiechle P, Erni F. Ruggedness tests for a high-per-
formance liquid chromatographic assay: comparison of an evaluation at two and three levels by
using two-level Plackett-Burman designs. Anal Chim Acta 1995; 316: 15–26.
 Concept and Applications of Plackett Burman Design 23

[46] Weiyong L and Henrik TR. Strategy for developing and optimizing liquid chromatography meth-
ods in pharmaceutical development using computer-assisted screening and Plackett–Burman
experimental design. J Chrom A 2003; 1016(2): 165–180.
[47] Garrido-Lopez A, Tena MT. Experimental design approach for the optimisation of pressurised
fluid extraction of additives from polyethylene films J Chromatogr A 2005 Dec 16; 1099(1–2):
75–83.
[48] Reche F, Garrigos MC, Marin ML, Canto A, Jimenez A. Optimization of parameters for the super-
critical fluid extraction in the determination of N-nitrosamines in rubbers. J Chrom A 2002; 963:
419–26.
[49] Decaestecker TN, Lambert WE, Van Peteghem CH, Deforce D, Van Bocxlaer JF. Optimization of
solid-phase extraction for a liquid chromatographic-tandem mass spectrometric general unknown
screening procedure by means of computational techniques. J Chromatogr A 2004 Nov 12;
1056(1–2): 57–65.
[50] Arambarri I, Lasa M, Garcia R, Millan E. Determination of fuel dialkyl ethers and BTEX in water
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016

using headspace solid-phase microextraction and gas chromatography-flame ionization detection.

J Chromatogr A. 2004 Apr 16; 1033(2): 193–203.
[51] Shahla M, Gunnar T, Bo K, Optimization of resolution in micellar electrokinetic chromatography
by multivariate evaluation of electrolytes, J of Chrom A 2001; 907: 267–277.
[52] Flores JR, Nevado JJB, Penalvo GC, Diez NM. Development of a Micellar electrokinetic capillary
chromatography method for the determination of three drugs employed in the erectile dysfunc-
tion therapy. J Chrom B 2004; 811: 231–263.
[53] Mikaeli S, Thorsen G, Karlberg B. Optimisation of resolution in micellar electrokinetic chromatogra-
phy by multivariate evaluation of electrolutes. J Chromatogr A 2001 Jan 12; 907(1–2): 267–77.
[54] http://www.fda.gov/CVM/guidance (accessed December 2006).
Downloaded by [Universitaetsbibliothek Tuebingen] at 07:18 02 June 2016