encyclopedia of biopharmaceutical statistics edited by chow s-c (2003) isbn 0824742613 (paper) 1055...
TRANSCRIPT
![Page 1: Encyclopedia of biopharmaceutical statistics Edited by Chow S-C (2003) ISBN 0824742613 (paper) 1055 pages; £252; $445 ISBN 0824742621 (electronic); £153; $252 (combined purchase](https://reader031.vdocuments.mx/reader031/viewer/2022020517/575025551a28ab877eb34a87/html5/thumbnails/1.jpg)
PHARMACEUTICAL STATISTICS
Pharmaceut. Statist. 2006; 5: 75–77
Published online in Wiley InterScience (www.interscience.wiley.com).
Book Reviews
This section of the journal aims to inform you about recent
books. We target books specifically about the theory and
application of statistics in the pharmaceutical sector, though
reviews of statistical books relevant to a wider audience are also
welcome. We cover new books or substantially new editions,
and include both those aimed at statisticians and those intended
to inform scientists about statistics. We are particularly
interested in comparative reviews of books on a similar subject.
Please contact Peter Lane ([email protected]) to offer to
review a book or suggest further books that should be reviewed.
You can find a list of books already suggested on the PSI
website, at
http://www.psiweb.org/resources/resources.asp?parent
folderid=252&subgroup_id=4 and on the Wiley website, at
http://www3.interscience.wiley.com/cgi-bin/jabout/
93012805/Reviews.html
These include information about the current status of
reviews, and are updated approximately monthly; the
list is also published quarterly in SPIN, the newsletter of
the PSI.
Encyclopedia of Biopharmaceutical Statistics
Edited by Chow S-C (2003)
ISBN 0824742613 (paper) 1055 pages; £252; $445
ISBN 0824742621 (electronic); £153; $252 (combined purchase
available)
Dekker, marketed by Taylor-Francis and CRC; http:/www.
crcpress.com/
This is an impressive collection of material describing those
processes and techniques in the pharmaceutical industry that
are particularly relevant to statisticians. Most of the articles
have been written by individuals contributing just one technical
essay each, averaging about seven pages, though some articles
have two or more authors. They are mostly practitioners within
the industry, who are describing and explaining subjects with
which they have long familiarity. Don’t expect to find enlight-
enment on areas with which you are yourself already familiar,
for this is no deep manual into the complexities of statistical
methods, nor up-to-the-minute review of cutting-edge research.
But who in the industry can claim recent familiarity with a
substantial proportion of the issues that concern us? The book
serves an admirable purpose in providing readable and
informative summaries about areas of which you have not
had recent first-hand experience, including suggestions for
further and deeper reading if needed.
The second edition of this single-volume encyclopaedia
appeared quickly: only three years since the first in 2000. The
intervening time has been spent in almost doubling its size, from
72 topics occupying 535 pages to 141 occupying 1055. I found
only one substantial review of the first edition, in The Statistician
(2001; 50:101–103). The reviewer pointed to three examples of
areas that were then not covered. Two now have their own
articles – logistic regression and survival analysis – as anyone
would expect given the title. The other deficiency noted was of
guidance in specific therapeutic areas: the single article on cancer
trials has now been joined by one on vaccine trials, though there
is nothing specific on trials in other areas such as neurology or the
central nervous system. Another criticism was of inconsistency in
the graphics; the poor examples cited are regrettably still present,
though the quality of the graphs in most articles is good.
I looked up a couple of areas I have been working in recently,
and found the description clear, readable and informative,
though lacking in recent developments and any depth. The
article on ‘Dropout’ gave a well-structured overview of the issues
and described many of the available techniques, including the
usual explanation of MCAR and MAR (Missing Completely At
Random or just At Random) which many will have seen recently
at statistical meetings. I learnt a little more about selection
models, but found no mention of the latest abbreviationMMRM
(Multivariate, or Mixed, according to preference, Models for
Copyright # 2006 John Wiley & Sons, Ltd.Received 15 January 20012006
![Page 2: Encyclopedia of biopharmaceutical statistics Edited by Chow S-C (2003) ISBN 0824742613 (paper) 1055 pages; £252; $445 ISBN 0824742621 (electronic); £153; $252 (combined purchase](https://reader031.vdocuments.mx/reader031/viewer/2022020517/575025551a28ab877eb34a87/html5/thumbnails/2.jpg)
Repeated Measurements) or any reference to the influential
recent publications by Craig Mallinckrodt. Though SOLAS gets
a mention when describing multiple imputation, none is made of
the MI and MIANALYZE Procedures introduced in SAS 8.
Similarly, the article on ‘Multicenter trials’ covers the material I
am familiar with, but has no references later than 1998.
Looking at an area I have only briefly touched on once, over
five years ago, I found the article on ‘Stability matrix designs’ to
be a useful introduction to the terminology and methodology of
this area of manufacturing. There are several examples to
illustrate the text, and a reasonable list of 24 references for
further reading. I also took a look at a topic about which I
know virtually nothing in the context of clinical trials. The
short three-page article on ‘Ethnic factors’ lays out the
background of this issue briefly, and then gives a series of
assessments in various contexts; however, no references are
given for further reading.
A question that clearly needs answering for a book
of this nature is how it fits in with the competition. Wiley’s
eight-volume Encyclopedia of Biostatistics is now also in
its second edition, appearing in 2005, and can claim far
more width and depth. But at £2095 for the 6100 pages, I can
imagine that many readers will find it hard to get access.
Another alternative nowadays, of course, is to rely on search
engines on the Internet, and seek information on new topics as
and when the need arises. However, despite the ingenuity of the
algorithms, it is unlikely that you will easily find reliable,
authoritative and well-written introductions to specific issues.
So there is clearly a place for this short encyclopedia as a
general reference resource in the pharmaceutical industry.
Peter Lane
Research Statistics Unit, GlaxoSmithKline, UK
(DOI: 10.1002/pst.202)
Design of Studies for Medical Research
Machin D, Campbell MJ (2005)
ISBN 0470 844 957; 286 pages; £29.95; h45.00; $55.00
Wiley; http:/www.wiley.com/
This book provides an excellent overview, as the title states, of
the design of studies for medical research. As the blurb details
‘This book is intended for anyone working in medical research, . . .It will be ideal for anyone embarking upon research for the first
time . . ., but is of relevance to all health and allied health
professionals.’ From the point of view of a statistician, it is not a
statistics text book and does not give detail of statistical analyses.
Instead, it is a text book for statisticians, and other health
professionals involved in medical research, wishing to understand
the importance of, and wishing to implement, good study design.
It provokes the reader into considering the many aspects involved
in designing studies, and is a good resource for the inexperienced
researcher. For those new to medical research, the book provides
an introduction to study design within a wide area of human
research, detailing the design of surveys, clinical trials and
epidemiological studies, as well as studies involved with diag-
nostics and prognostic factors.
The book broadly breaks down medical research into three
areas: preclinical, clinical and epidemiological. There are 11
chapters, covering the following topics: defining evidence in
evidence-based health care; the importance of appropriate
measurements and measurement tools; principles of study-size
calculation; randomization; cross-sectional and longitudinal
studies; surveys, cohorts and case-control studies; general issues
in clinical trials; early phase clinical trials; phase III trials;
diagnosis; and prognostic factor studies. Each chapter contains a
good number of practical examples of the topic being discussed,
giving an extensive reference list, and key design features are
summarized at the end of each section as appropriate. Statistical
detail is discussed only where necessary, and tables of reference
are supplied as an appendix. For the more technical reader, there
are extra technical notes at the end of many of the chapters.
However, these expand only a little further on any statistical
detail mentioned within the chapter.
The first chapter introduces the notion of evidence and the
hierarchy of designs (in terms of the weight given to evidence
obtained) within each of preclinical, clinical and epidemiological
studies. It also introduces statistical considerations, informed
consent and ethics, study protocol and reporting. The second
chapter is concerned with the design of forms and questionnaires,
and the importance of taking appropriate measurements. It
describes different types of data and measurements, and the
layout and design of forms and questionnaires, and discusses the
issue of data collection and processing. Chapter 3 details the
importance of the study-size calculation, and gives an outline of
the basic workings required to carry out a sample-size calcula-
tion. It details significance tests, hypotheses, and type I and II
errors, and gives sample-size formulae for continuous, binary
and survival outcomes. It discusses the calculation of sample
sizes in practice, considering effect sizes, limited resources,
situations with more than one primary outcome, and internal
pilot studies, with examples to illustrate.
Chapter 4 discusses randomization within preclinical studies,
observational studies, surveys and clinical trials, and also the
practicalities of randomizing patients. Chapter 5 describes the
design of cross-sectional and longitudinal studies, and the
differences between the two. It discusses designs for cross-
sectional studies with a single group, two groups and more than
two groups, and considers repeated measures and autocorrela-
tion for longitudinal studies. Identifying groups to be compared,
achieving representative samples, selecting appropriate endpoints
Book Reviews76
Copyright # 2006 John Wiley & Sons, Ltd. Pharmaceut. Statist. 2006; 5: 75–77