Statistical Issues in Drug Development 3rd Edition

Visit the link below to download the full version of this book:

https://medidownload.com/product/statistical-issues-in-drug-development-3rd-edit
ion/

Click Download Now

To Victoria, Helen and Mark
 Contents

Preface to the Third Edition xi

Preface to the Second Edition xv

Preface to the First Edition xix

Acknowledgementsxxiii

1 Introduction 1
1.1 Drug Development 1
1.2 The Role of Statistics in Drug Development 2
1.3 The Object of this Book 3
1.4 The Author’s Knowledge of Statistics in Drug Development 4
1.5 The Reader and Her or His Knowledge of Statistics 5
1.6 How to Use the Book 6

Part 1 Four Views of Statistics in Drug Development:

Historical, Methodological, Technical and Professional 9

2 A Brief and Superficial History of Statistics for Drug Developers 11

2.1 Introduction 11
2.2 Early Probabilists 12
2.3 James Bernoulli (1654–1705) 13
2.4 John Arbuthnott (1667–1753) 14
2.5 The Mathematics of Probability in the Late Seventeenth,
the Eighteenth, and Early Nineteenth Centuries 16
2.6 Thomas Bayes (1701–1761) 17
2.7 Adolphe Quetelet (1796–1874) 18
2.8 George Biddell Airy (1801–1892) 18
2.9 Francis Galton (1822–1911) 18
2.10 Karl Pearson (1857–1936) 19
2.11 ‘Student’ (1876–1937) 19
2.12 R.A. Fisher (1890–1962) 19
2.13 Modern Mathematical Statistics 20
2.14 Medical Statistics 21
2.15 Statistics in Clinical Trials Today 22
2.16 The Current Debate 23
2.17 A Living Science 24
2.18 Further Reading 25
viii Contents

3 Design and Interpretation of Clinical Trials as Seen by a Statistician 31

3.1 Prefatory Warning 31
3.2 Introduction 31
3.3 Defining Effects 32
3.4 Practical Problems in Using the Counterfactual Argument 32
3.5 Regression to the Mean 33
3.6 Control in Clinical Trials 37
3.7 Randomization 38
3.8 Blinding 40
3.9 Using Concomitant Observations 42
3.10 Measuring Treatment Effects 43
3.11 Data Generation Models 44
3.12 In Conclusion 45
3.13 Further Reading 45

4 Probability, Bayes, P‐values, Tests of Hypotheses and

Confidence Intervals 49
4.1 Introduction 49
4.2 An Example 50
4.3 Odds and Sods 50
4.4 The Bayesian Solution to the Example 51
4.5 Why Don’t We Regularly Use the Bayesian Approach in Clinical Trials? 52
4.6 A Frequentist Approach 53
4.7 Hypothesis Testing in Controlled Clinical Trials 54
4.8 Significance Tests and P‐values55
4.9 Confidence Intervals and Limits and Credible Intervals 56
4.10 Some Bayesian Criticism of the Frequentist Approach 57
4.11 Decision Theory 57
4.12 Conclusion 58
4.13 Further Reading 58

5 The Work of the Pharmaceutical Statistician 61

5.1 Prefatory Remarks 61
5.2 Introduction 62
5.3 In the Beginning 63
5.4 The Trial Protocol 63
5.5 The Statistician’s Role in Planning the Protocol 64
5.6 Sample Size Determination 65
5.7 Other Important Design Issues 66
5.8 Randomization 66
5.9 Data Collection Preview 67
5.10 Performing the Trial 67
5.11 Data Analysis Preview 67
5.12 Analysis and Reporting 68
5.13 Other Activities 69
5.14 Statistical Research 69
5.15 Further Reading 70
 Contents ix

Part 2 Statistical Issues: Debatable and Controversial

Topics in Drug Development 75

6 Allocating Treatments to Patients in Clinical Trials 77

6.1 Background 77
6.2 Issues 79
6.A Technical Appendix 104

7 Baselines and Covariate Information 109

7.1 Background 109
7.2 Issues 112
7.A Technical Appendix 129

8 The Measurement of Treatment Effects 133

8.1 Background 133
8.2 Issues 134
8.A Technical Appendix 156

9 Demographic Subgroups: Representation

and Analysis159
9.1 Background 159
9.2 Issues 161
9.A Technical Appendix 176

10 Multiplicity 181
10.1 Background 181
10.2 Issues 182
10.A Technical Appendix 200

11 Intention‐to‐Treat, Missing Data and Related Matters 203

11.1 Background 203
11.2 Issues 205
11.A Technical Appendix 222

12 One‐Sided and Two‐Sided Tests and Other Issues to Do with

Significance and P‐values 225
12.1 Background 225
12.2 Issues 226

13 Determining the Sample Size 241

13.1 Background 241
13.2 Issues 244

14 Multicentre Trials 265

14.1 Background 265
14.2 Issues 266
14.A Technical Appendix 285
x Contents

15 Active Control Equivalence Studies 289

15.1 Background 289
15.2 Issues 291
15.A Technical Appendix 304

16 Meta‐Analysis 307
16.1 Background 307
16.2 Issues 309
16.A Technical Appendix 335

17 Cross‐Over Trials 339

17.1 Background 339
17.2 Issues 341

18 n‐of‐1 Trials 355

18.1 Background 355
18.2 Issues 357

19 Sequential and Flexible Trials 367

19.1 Background 367
19.2 Issues 375

20 Dose‐Finding 393
20.1 Background 393
20.2 Issues 395

21 Concerning Pharmacokinetics and Pharmacodynamics 419

21.1 Background 419
21.2 Issues 426

22 Bioequivalence Studies 447

22.1 Background 447
22.2 Issues 449

23 Safety Data, Harms, Drug Monitoring and Pharmaco‐Epidemiology 473

23.1 Background 473
23.2 Issues 480

24 Pharmaco‐economics and Portfolio Management 507

24.1 Background 507
24.2 Issues 509

25 Concerning Pharmacogenetics, Pharmacogenomics

and Personalized Medicine 541
25.1 Background 541
25.2 Issues 547
25.A Technical Appendix 569

Glossary571

Index603
 Preface to the Third Edition

I have a special affection for The Statistics in Practice series. My first book, Cross‐over Trials
in Clinical Research was the first to be published in the series and I vividly remember travel-
ling from Switzerland to Harpenden to see Vic Barnett (then at Rothamsted) and get
advice on writing the book. This was my first visit to Harpenden. By the time the first edi-
tion of Statistical Issues in Drug Development was published, I myself was living in
Harpenden and commuting to University College London (UCL) and in a period of several
years that followed, during which I was on the council of the Royal Statistical Society, had
frequent contact with Vic who was the treasurer. By the time the second edition was pub-
lished, I was proud to have become one of the editors who took over from Vic and men-
tioned it in the preface. Sadly, Vic died in 2014. He continues to be missed. I am pleased to
acknowledge my debt to him in helping to get me started in writing monographs and in
founding the series that made it possible.
New books continue to appear in the series and these are now too numerous to merit a
separate mention in the preface but many are referenced in the text. I am delighted also,
that Geert Molenberghs, himself a contributor to the series, has now taken over from me
as a series editor. I am sure the series is in safe hands and wish him and future authors
every success.
Concerning Statistical Issues in Drug Development itself, as with the second edition, and
following my motto that, ‘statistics is not just for Christmas but for life’, I have decided to
give a statistical impression of what additions there have been to the book. Again, there
are two figures. The first, shows Fig P3.1 the number of words by chapter and edition and
the second, Fig P3.2 the difference in words by chapter. In both cases the division of the
book into the minor first section (Chapters 1–5), which give a basic coverage of the subject
in terms of various perspectives and the major part, consisting of 20 in‐depth chapters on
issues grouped by theme, is marked. This is a good point to mention one further change. In
the first two editions all the figures were produced in Mathcad®, long a favourite package
of mine and one I used to help me in various theoretical investigations I undertook. Using
the same package for both made it easy to produce the figures. However, although Mathcad
has excellent default settings in graphics, it has limited flexibility and I took the decision to
replace all the figures using Gentstat®, also long a favourite package of mine. The number
of figures has also been increased from 66 to 98 (excluding the two in the preface), in
some cases to better illustrate old topics but in many cases to accompany new ones.
As regards new topics covered, they include cluster randomization, historical controls,
responder analysis, studies in children, post‐hoc tests, estimands, publication bias, the
‘replication crisis’, invalid inversion, severity, random effects in network meta‐analysis,
xii Preface to the Third Edition

Edition
Third
Second

Glossary
Division

15
Words (thousands)

Ed. 3 mean

10
Ed. 2 mean

0
5 10 15 20 25
Chapter

Figure P3.1 Words per chapters for the second and third editions.

Division

5
Extra words (thousands)

3
Mean
Glossary

0
5 10 15 20 25
Chapter

Figure P3.2 Additional length of third edition compared to the second.

 Preface to the Third Edition xiii

explained variation, meta‐analysis of sequential trials, individual patient data meta‐­

analysis, cross‐over trials for absorbing binary endpoints, contralateral studies, Dawid’s
selection paradox, the MCP‐mod approach to dose‐finding, limits of quantitation, dissolu-
tion studies, biosimilars, collapsibility of measures, zero frequencies in meta‐analysis of
binary data, drug–drug interactions, choice of comparators and studies in network meta‐­
analysis and using heteroscedasticity to identify variation in response.
I received help from many people, to all of whom I owe thanks. New co‐authors I have
acquired since the second edition and whose work is cited include Olivier Collignon, Susi
Schmitz and Anna Schritz from my group at the Luxembourg Institute of Health, PhD
and MSc students Artur Araujo, Andisheh Bakhshi, Boikanyo Makubate, Edith Jude‐Eze
and Jim Weir and other collaborators Vlad Anisimov, Val Fedorov, Philip Dawid, Hans
Hockey, Nick Holford, Roger Lewis and Alan Philips and I am also grateful for continued
collaboration with Rosemary Bailey, Sheila Bird, Carl‐Fredrik Burman, Andy Grieve and
Steven Julious. Others who kindly involved me in multi‐author projects they led were Jordi
Cortes, Carole Gamble, Jen Gewandter, Sander Greenland, Dennis Lendrem and Christine
McNamee. Although we did not publish together, I have used and cited Nicola Greenlaw’s
MSc and Emmanuel Baah’s PhD theses with me and am grateful to them for their work.
I have a particular thankyou to give to Dieter Hilgers, who led the IDeAl project (Hilgers,
R.D. et al. 2018), in which I was involved, and to the EU FP7 programme which funded the
work under grant 602552. Some work on the Simplicity Complexity and Modelling
(SCAM) project (Christie, M. et al. 2011), for which I was the lead investigator is also used
and I am grateful to the EPSRC for funding. Other funding from ICON and Boehringer‐
Ingelheim for work in which I was involved is also gratefully acknowledged.
I also thank my many colleagues in both IDeAl and SCAM for their collaboration and
members of my group at the Luxembourg Institute of Health for stimulating conversa-
tions over the seven years of my time there (2011–2018), in particular my deputy Michel
Vaillant. I thank the following persons who gave generous help on particular chapters or
provided me with suggestions or relevant papers: Fiona Baines, John Belmont, Tanja
Berger, Björn Bornkamp, Frank Bretz, Olivier Collignon, George Davey Smith, Stephen
Evans, Val Fedorov, Tim Friede, Georgi Giorgiev, Björn Holzhauer, Mats Karlsson, Simon
Kirby, Huw Llewelyn, Mohammad Mansournia, Peter McCullagh, Kathrin Möllenhoff,
David Norris, Sabine Preussler, Richard Riley, Patrick Royston, Claudia Schmoor, Julia
Singer, Harry Southworth, Theo Stijnen, Alex Thompson, Peter Westfall, John Whittaker
and Christina Yap.
A particular thanks is due to philosopher of science Deborah Mayo. She suggested some
years ago that I contribute occasional blogs to her website. I have continued to do this for
many years now, with many of the topics covered having relevance to this book. I am also
grateful to Jean Miller, who works with Deborah, for her expert help in putting these up.
Thanks to Deborah and Jean, all I have had to do is write! An index of my blogs is kept here
http://www.senns.uk/Blogs.html and the reader may find that these contain additional
relevant material. Deborah has also made me re‐think about frequentist statistics in a
more positive way. The reader who wishes to know more should consult (Mayo, D. 2018).
I am grateful to Carl‐Frederik (Caffe) Burman, a frequent collaborator, for pointing out
that I need to issue a warning. The warning was already (thanks to him) in the second
edition but who reads previous prefaces? Thus, I shall give it again. Steven Piantadosi’s
excellent book on clinical trials (Piantadosi, S. 2017), for example, has much by the way
of positive advice for trialists with section headings such as Increase the sample size for non‐
adherence or Prevent methodologic errors, with sound discussion as to what needs to be
done. On the other hand, in this book, the reader will occasionally encounter headings
xiv Preface to the Third Edition

such as It is essential that the noninferiority margin be prespecificed or You cannot pool different
studies or Small trials are unethical. These are usually an indication that I am not in 100%
agreement with such statements but take issue with them, as indeed is appropriate in a
book with this title.
The objective of the book remains to aid dialogue between statisticians and life‐­scientists
and others working in drug development. Statistics, and in particular pharmaceutical sta-
tistics, is a subject that has given me great pleasure. I hope that the reader can share some
of this. The reader who finds bite‐size statistics interesting can follow me on Twitter at
@stephensenn.

References

Christie, M., Cliffe, A., Dawid, A.P., and Senn, S. (eds.) (2011). Simplicity, Complexity and Modelling.
Chichester: Wiley.
Hilgers, R.D., Bogdan, M., Burman, C.F. et al. (2018). Lessons learned from IDeAl – 33 recommen-
dations from the IDeAl‐net about design and analysis of small population clinical trials. Orphanet
Journal of Rare Diseases 13 (1): 77.
Mayo, D. (2018). Statistical Inference as Severe Testing: How to Get Beyond the Statistics Wars.
Cambridge: Cambridge University Press.
Piantadosi, S. (2017). Clinical Trials: A Methodologic Perspective, 3e. New York: Wiley.
 Preface to the Second Edition

There have been many developments since the first edition of this book and it was high
time for a second. My own period working in the pharmaceutical industry is now a distant
memory but the 10 years working as an academic since the first edition has had its com-
pensations. I have been fortunate enough to be able to consult for many pharmaceutical
companies during this time and this has certainly widened my appreciation of the work
that statisticians do within the industry and the problems they face.
Alas, this appreciation is not shared by all. Many take it as almost axiomatic that statis-
tical analysis carried out within the pharmaceutical industry is necessarily inferior to that
carried out elsewhere. Indeed, one medical journal has even gone so far as to make it a
requirement for publication that analyses from the pharmaceutical industry should be
confirmed by an academic statistician, a policy which is as impractical as it is illogical.
Two related developments since the first edition, one of which is personal, are highly
relevant. The first is that I have been honoured to succeed Vic Barnett as an editor for
Wiley’s Statistics in Practice series, in which this book appears. The second is that the series
itself, so ably founded by Vic and Helen Ramsey, has been growing steadily and since the
first edition now has attracted a number of further volumes that are highly relevant to
this one. The chapters that have particularly benefited are listed with the relevant refer-
ences as follows.
Chapter 6 Allocating treatments to patients in clinical trials (Berger 2005)
Chapter 7 Baselines and covariate information (Berger 2005)
Chapter 11 Intention to treat, missing data and related matters (Molenberghs and
Kenward 2007)
Chapter 16 Meta‐analysis (Whitehead 2002)
Chapter 19 Sequential trials (Ellenberg et al. 2003)
Chapter 20 Dose‐finding (Chevret 2006)
Chapter 22 Bioequivalence studies (Hauschke et al. 2007)
Chapter 23 Safety data, harms, drug‐monitoring and pharmacoepidemiology
(Lui 2004)
Chapter 24 Pharmacoeconomics and portfolio management (Parmigiani 2002; Willan
and Briggs 2006)
Also extremely useful are two books on Bayesian methods (O’Hagan et al. 2006;
Spiegelhalter et al. 2003) and Brown and Prescott’s book on mixed models (Brown and
Prescott 2006), which is already in a second edition. The books on survival analysis and
sequential analysis by Marubini and Valsecchi (1995) and Whitehead (1997) remain
highly relevant, of course, as does my own on cross‐over trials, (Senn 2002), also now in
a second edition.
xvi Preface to the Second Edition

All chapters have been brought up to date in the new edition and in particular, there is
extensive reference to various guidelines of the International Conference of Harmonization
that have been issued since the first edition, in particular ICH E9, Statistical Principles for
Clinical Trials. A new chapter on pharmaco‐genetics has been added. For the reader in pos-
session of a first edition who wishes to know whether to splash out on a second, Figure P.1
may be helpful. (This is partly so that I can underline the fact that nothing in life, not even
an author’s preface, should be exempt from statistics!) This compares the two editions
chapter by chapter in terms of their length in thousands of words.
The major division of the book occurs after Chapter 5, when material introducing sta-
tistics in drug development from historical, philosophical, technical and professional
perspectives is succeeded by single‐issue chapters. Figure P.2 shows even more clearly

Division
15 First edition
Second edition
Words (thousands)

10

0 5 10 15 20 25
Chapter

Figure P.1 Comparison of the first and second editions in length.

Division
8

6
Words (thousands)

2 Mean

0 5 10 15 20 25
Chapter

Figure P.2 Additional length (in thousand of words) of the second edition compared to the first.
The mean increase (excluding Chapter 25) is 2000 words.
 Preface to the Second Edition xvii

that much of the additional material has gone into chapters in Part 2. Apart from the
wholly new chapter on pharmacogenetics in particular, chapters on baselines, measur-
ing effects, intention to treat and missing data, equivalence, meta‐analysis, sequential
analysis, dose‐finding, pharmaco‐epidemiology and pharmaco‐economics have had
extensive additions.
It is appropriate for me to repeat a general warning about the book that Carl‐Fredrik
Burman has drawn to my attention. A number of the section headings contain state-
ments of position. For example, Chapter 7 has a section, ‘The propensity score is a supe-
rior alternative to adjusting for confounders than analysis of covariance’. You will get a
very misleading impression of the message of the book if you take these as being my posi-
tion. This is a book about issues and where such statements are made it is nearly always
because, as is the case with this one, I wish to take issue with them.
I am grateful to Frank Bretz, Diane Elbourne, Paul Gallo, Oliver Keene, Dieter
Hauschke, Nick Holford, Paul Johnson, Vincent Macaulay, Helmut Schütz, Helen Senn
and John Sorkin for comments, and to Mateo Aboy, Nick Holford, Jerry Nedelman, Luis
Pereira and Michael Talias for providing copies of their papers. Since the first edition, I
have acquired several new co‐authors whose work is reflected in this edition: my PhD
students Dimitris Lambrou and Sally Lee and also, Pina D’Angelo, Frank Bretz, Carl‐
Fredrik Burman, Angelika Caputo, Farkad Ezzett, Erika Graf, Emmanuel Lesaffre, Frank
Harrell, Hans van Houwelingen, William Mezanotte, Christopher Miller, Diane Potvin,
Peter Regan and Nicoletta Rosati, whom I thank. I am also extremely grateful to
Andy Grieve, for con­tinued collaboration and to my PhD students Steven Julious, Andy
Garrett for their work. I thank Kathryn Sharples, Susan Barclay, Beth Dufour and
Simon Lightfoot at Wiley and also Len Cegielka and Cherline Daniel for work on
­preparing the book.
I continue to hope, of course, that this book will aid dialogue between statisticians and
life‐scientists within the pharmaceutical industry but also hope that it will contribute to a
wider appreciation of the interesting challenges that statisticians within the pharmaceu-
tical industry face and the seriousness with which they are met. I hope that the reader
finds both stimulation and enjoyment in encountering these challenges.

References

Berger, V.W. (2005). Selection Bias and Covariate Imbalances in Randomized Clinical Trials. Chichester:
Wiley.
Brown, H. and Prescott, R. (2006). Applied Mixed Models in Medicine. Chichester: Wiley.
Chevret, S. (ed.) (2006). Statistical Methods for Dose‐Finding Experiments. Chichester: Wiley.
Ellenberg, S., Fleming, T., and DeMets, D. (2003). Data Monitoring Committees in Clinical Trials:
A Practical Perspective. Chichester: Wiley.
Hauschke, D., Steinijans, V., and Pigeot, I. (2007). Bioequivalence Studies in Drug Development:
Methods and Applications. Chichester: Wiley.
Lui, K.‐J. (2004). Statistical Estimation of Epidemiological Risk. Hoboken: Wiley.
Marubini, E. and Valsecchi, M.G. (1995). Analysing Survival Data from Clinical Trials and Observational
Studies. Chichester: Wiley.
Molenberghs, G. and Kenward, M.G. (2007). Missing Data in Clinical Studies. Chichester: Wiley.
O’Hagan, A., Buck, C.E., Daneshkah, A. et al. (2006). Uncertain Judgements. Chichester: Wiley.
Parmigiani, G. (2002). Modeling in Medical Decision Making: A Bayesian Approach. Chichester: Wiley.
Senn, S.J. (2002). Cross‐Over Trials in Clinical Research, 2e. Chichester: Wiley.
xviii Preface to the Second Edition

Spiegelhalter, D.J., Abrams, K.R., and Myles, J.P. (2003). Bayesian Approaches to Clinical Trials and
Health‐Care Evaluation. Chichester: Wiley.
Whitehead, J. (1997). The Design and Analysis of Sequential Trials, Revised 2nd ed. Chichester: Wiley.
Whitehead, A. (2002). Meta‐Analysis of Controlled Clinical Trials. Chichester: Wiley.
Willan, A.R. and Briggs, A.H. (2006). Statistical Analysis of Cost‐Effectiveness Data. Chichester:
Wiley.
 Preface to the First Edition

The conscientious mathematician acts in this respect like the lady who is a conscientious shopper.
Wishing to satisfy herself of the quality of a fabric, she wants to see it and to touch it. Intuitive insight
and formal proof are two different ways of perceiving the truth, comparable to the perception of a mate-
rial object through two different senses, sight and touch.
Polya, How to Solve It

When I first started work in the Swiss pharmaceutical industry in 1987, I had already
been working as a statistician for 12 : 3 years in the National Health Service in England
and then 9 years lecturing in Scotland. What struck me then was how much I still had to
learn, not only about medicine, pharmacology and drug development in general, but also
about my own subject, statistics. Working with other professionals, many of them experts
in fields of which I knew nothing, called for a different approach to statistics in at least
four ways. First, I had to examine what I ‘knew’ in order to establish what was useful and
what was not. Second, I had to supplement it with knowledge of many fields of statistical
theory which were new to me (design and analysis of cross‐over trials was an example).
Third, I had to pay attention to the scientific fields of my colleagues, in particular pharma-
cology, and try to work out the implications in statistical terms. Fourth, I had to repay my
colleagues’ willingness to explain their sciences to me by making my statistical points in
terms that were clear to them.
In doing the latter, it became clear to me that there were many things which I myself
had taken for granted which were debatable. I also came to the conclusion that many of
the difficulties with statistical ideas lie deeper than statistics itself (an example would be
notions of causality). Furthermore, I became convinced that statisticians pay their non‐
statistical colleagues a disservice if they try to gloss over genuine disagreements. This
book is an attempt to present many of the statistical issues in drug development in a way
which is comprehensible to life scientists working in drug development whilst avoiding
false consensus. The emphasis will be on the intuition which Polya, although himself a
distinguished mathematician, valued so highly. Although addressed to the life‐scientist it
is my hope that many statisticians, in particular those studying medical statistics or
embarking on a career in drug development, will also find it useful. Above all I hope that
it will help communication between the disciplines: a process by which the statistician
stands to benefit as much as any other professional in drug development. I cannot pre-
tend, however, to be objective on all issues and am not even sure of what such objectivity
might consist. In my defence, however, I think that I may justly claim, that if all view-
points are not given equal care and consideration, the reader will at least come away with
a wider awareness that other views exist, than would have been the case in a more con-
ventional approach.
xx Preface to the First Edition

The book is in two sections. The first, and by far the shorter section, is based on a statis-
tics appreciation course which I gave a number of times to my colleagues at Ciba‐Geigy. In
addition to a brief introduction it consists of four chapters, each of which takes a different
view, of statistics in drug development: historical, ‘philosophical’, technical and profes-
sional. I recommend that every reader will either read these chapters or satisfy her or
himself that the material is familiar. The second and larger section is loosely based on a
course (but less technical than that course) which I give to students on the Postgraduate
Diploma in Statistics at Neuchatel University and consists of 19 chapters of varying
length which may be read in almost any order and consulted as considered desirable. Each
chapter consists of a brief background statement and then a second section split into a
number of issues. These are sometimes presented as true open issues and sometimes as
positions which one might take on an issue. (Where the latter is the case, the reader should
be warned that I usually disagree with the position taken.) At the end of each chapter are
a number of references. Sometimes these will have been referred to explicitly but in some
cases they are merely listed because they are useful additional reading.
When it comes to handling statistics, the life‐scientist in drug development has two
­particular advantages over his or her colleagues elsewhere. First, (s)he will, through the
nature of the work, come into frequent contact with statistical problems. Consequently, a
basic familiarity with some essential concepts will be obtained. Second, technical statistical
matters will be handled as a matter of course by the statisticians assigned to the various
drug development projects. Life scientists working elsewhere will not always be so f­ ortunate
as to have resident experts whom they may consult. This provides a further justification for
my decision to concentrate on issues rather than technicalities.
I find the subject matter fascinating, of course, but am aware that the reader will not
always share my enthusiasm. I have tried to leaven the mix by adding chapter quotations
and even the occasional joke or anecdote, ignoring Sterne’s warning that: Tis no extrava-
gant arithmetic to say, that for every ten jokes, − thou has got an hundred enemies; and till thou
hast gone on, and raised a swarm of wasps about thine ears, and art half stung to death by them,
thou wilt never be convinced it is so.
This book would have been impossible to write without the help of many statisticians
and life scientists from whom I received my education at Ciba‐Geigy (now merged with
Sandoz to form Novartis). In particular, amongst statisticians, I have to thank my former
colleagues Farkad Ezzet, Hans‐Peter Graf, Andy Grieve, Walter Kremers, Gunther Mehring,
Amy Racine and Jakob Schenker for many helpful discussions during my time at Ciba‐
Geigy, as well as the various members of my own group, Bernhard Bablok, Nathalie Ezzet,
Friedhelm Hornig, Rolf Meinert, Erhard Quebe‐Fehling, Peter Sacares, Denise Till,
Elizabeth Wehrle and Albert Widmer for shared work over the years, and also others in
Switzerland, the USA and elsewhere, too numerous to mention. I also learned a great deal
from the life scientists with whom I collaborated and I would particularly like to thank
Reto Brambilla, Giovanni Della Cioppa, Brigitte Franke, Francesco Patalano and Bill
Richardson for sharing the joys and pains of drug development with me. I also owe a par-
ticular thank you to William Jenkins who, in introducing me to the problem of portfolio
management, led me to a wider appreciation of the role of statistics in drug development,
and to Anders Hove, Keith Widdowson, Marc Cohen, Bill Huebner, Ronald Steele of CIBA‐
Geigy and Peter Regan, formerly of Strategic Decisions Group for working on this problem
with me. My UCL colleagues, Vern Farewell and Rebecca Hardy made helpful comments
on various chapters as did Leon Aarons, Peter Bauer, Michael Budde, Stephen Evans,
Farkad Ezzet, Dieter Hauschke, Oliver Keene, Walter Kremers, John Lewis, Bill Richardson,
 Preface to the First Edition xxi

Joachim Röhmel, Mark Sculpher and John Whitehead. I thank Lew Sheiner for permission
to reproduce figure 22.1 from one of his papers. I am also grateful to Peter Bauer, Roger
Berger, Michael Dewey, Farkad Ezzet, Nancy Geller, Andy Grieve, Miranda Mugford and
Wendy Ungar for giving me access to (as yet) unpublished work of theirs and to Yadolah
Dodge, professor of statistics at the University of Neuchatel and the students on the post-
graduate diploma there for the opportunity to expound this subject. Thanks are also due
to Guernsey McPearson for contributing quotations and other material.