Clinical Trials

A Methodologic Perspective

Steven Piantadosi

• Publisher: Wiley
• Year: 2017
• ISBN:
  9781118959220
• Edition: 3
• Topic: Mathématiques
• Subtopic: Probabilités et statistiques

Presents elements of clinical trial methods that are essential in planning, designing, conducting, analyzing, and interpreting clinical trials with the goal of improving the evidence derived from these important studies

This Third Edition builds on the text's reputation as a straightforward, detailed, and authoritative presentation of quantitative methods for clinical trials. Readers will encounter the principles of design for various types of clinical trials, and are then skillfully guided through the complete process of planning the experiment, assembling a study cohort, assessing data, and reporting results. Throughout the process, the author alerts readers to problems that may arise during the course of the trial and provides common sense solutions. All stages of therapeutic development are discussed in detail, and the methods are not restricted to a single clinical application area.

The authors bases current revisions and updates on his own experience, classroom instruction, and feedback from teachers and medical and statistical professionals involved in clinical trials. The Third Edition greatly expands its coverage, ranging from statistical principles to new and provocative topics, including alternative medicine and ethics, middle development, comparative studies, and adaptive designs. At the same time, it offers more pragmatic advice for issues such as selecting outcomes, sample size, analysis, reporting, and handling allegations of misconduct. Readers familiar with the First and Second Editions will discover revamped exercise sets; an updated and extensive reference section; new material on endpoints and the developmental pipeline, among others; and revisions of numerous sections.

In addition, this book:

• Features accessible and broad coverage of statistical design methods—the crucial building blocks of clinical trials and medical research -- now complete with new chapters on overall development, middle development, comparative studies, and adaptive designs

• Teaches readers to design clinical trials that produce valid qualitative results backed by rigorous statistical methods

• Contains an introduction and summary in each chapter to reinforce key points

• Includes discussion questions to stimulate critical thinking and help readers understand how they can apply their newfound knowledge

• Provides extensive references to direct readers to the most recent literature, and there are numerous new or revised exercises throughout the book

Clinical Trials: A Methodologic Perspective, Third Edition is a textbook accessible to advanced undergraduate students in the quantitative sciences, graduate students in public health and the life sciences, physicians training in clinical research methods, and biostatisticians and epidemiologists.

This book is accompanied by downloadable files available below under the DOWNLOADS tab.

These files include:

• MATHEMATICA program – A set of downloadable files that tracks the chapters, containing code pertaining to each.

• SAS PROGRAMS and DATA FILES used in the book.

The following software programs, included in the downloadables, were developed by the author, Steven Piantadosi, M.D., Ph.D:

• RANDOMIZATION – This program generates treatment assignments for a clinical trial using blocked stratified randomization.

• CRM – Implements the continual reassessment methods for dose finding clinical trials.

• OPTIMAL – Calculates two-stage optimal phase II designs using the Simon method.

• POWER – This is a power and sample size program for clinical trials.

Executables for installing these programs can also be found at https://risccweb.csmc.edu/biostats/.

Steven Piantadosi, MD, PhD, is the Phase One Foundation Distinguished Chair and Director of the Samuel Oschin Cancer Institute, and Professor of Medicine at Cedars-Sinai Medical Center in Los Angeles, California. Dr. Piantadosi is one of the world's leading experts in the design and analysis of clinical trials for cancer research. He has taught clinical trials methods extensively in formal courses and short venues. He has advised numerous academic programs and collaborations nationally regarding clinical trial design and conduct, and has served on external advisory boards for the National Institutes of Health and other prominent cancer programs and centers. The author of more than 260 peer-reviewed scientific articles, Dr. Piantadosi has published extensively on research results, clinical applications, and trial methodology. While his papers have contributed to many areas of oncology, he has also collaborated on diverse studies outside oncology including lung disease and degenerative neurological disease.

1
Preliminaries

1.1 Introduction

The best time to contemplate the quality of evidence from a clinical trial is before it begins. High-quality evidence about the effects of a new treatment is a consequence of good study design and execution, which themselves are the results of careful planning. This book presents elements of clinical trial methodology that are needed in planning, designing, conducting, analyzing, and assessing clinical trials with the goal of improving the evidence derived from these important studies. The topics discussed include subjects relevant to early planning and design—some that find general agreement among methodologists and others that are contentious. It is unlikely that a reader experienced with clinical trials will agree with all that I say or emphasize, but my perspective should be mainstream, internally consistent, and useful for learning.

Much of what is written about clinical trials tells us what to do. As a result, many investigators, sponsors, and regulators become proficient in standard practice, perhaps to the point that they view it as restrictive. Not as much of the literature explains why we do what is recommended. Knowing only what to do creates patterned thinking, whereas knowing why we do it allows appropriate creative exceptions and coping with atypical circumstances. I try throughout this book to emphasize the rationale for what we do in clinical trials, and specifically discourage unhelpful patterned thinking.

My further purpose is to challenge the dogma and traditions to which many young investigators have been exposed by their older colleagues. I will not stereotype and validate impressions gathered from the very imperfect culture of clinical investigation that exists nearly everywhere. In recent years especially, I have noticed some aggressive ignorance regarding the scientific principles underlying clinical trials. The goal of this book is to teach foundational principles and their sometimes complicated implications so that the sublime, ordinary, and absurd in clinical trials becomes more clear. As a general rule, I will discard cherished perspectives if they lack a strong scientific basis.

This book is not an introduction to clinical trials and has evolved to become more technical and lengthy as it has been updated through the years. Even so, I am aware from colleagues that portions of it have found their way into didactics for beginning students of clinical trials. My intent for this book is to be part of a one quarter or semester structured postgraduate course for an audience with quantitative skills and therapeutic focus. I take for granted that the audience has already completed a formal didactic introduction to clinical trials. I also expect that the audience will have completed a basic postgraduate series in biostatistics. The first edition of this book evolved over a dozen years by merging a course in experimental design with one on clinical trials. The second edition was the result of seven additional years of teaching and concomitant changes in the field. The third edition incorporates changes and needs that I perceive in the field over the past 8–10 years, and responds partly to my experiences teaching diverse clinical investigators at a second major medical school and academic medical center.

1.2 Audiences

The intended teaching audience includes both medical and statistical professionals early in their clinical research careers. Each component of the audience presents a certain dilemma. The book assumes a working knowledge of basic biostatistics. It is not really possible to get beyond the introductory ideas of clinical trials without statistics. It is also helpful if the reader understands some more advanced statistical concepts, including ideas like methods of inference, error control, lifetables, survival models, and likelihoods. Clinicians often lack this knowledge. However, modern clinical researchers are seeking the required quantitative background through formal training in clinical investigation methods or experimental therapeutics.

Biostatistics professionals require no clinical knowledge to understand the concepts in this book. Even so, fundamental principles of clinical research are helpful, including some basic human biology, research ethics, therapeutic developmental paradigms, clinical terminology and ideas of clinical pharmacology. Eventually, for the clinical trials biostatistician, immersion in a disease-oriented discipline will be vital to make the most substantive collaborative contributions as well as to identify and solve relevant methodological problems.

Uneven discussion (basic to technically complex) results from this mixed target audience and perspectives, as well as from the very nature of clinical trials. The classes that I teach using this book typically contain a mixture of biostatistics graduate students, medical doctors in specialty or subspecialty training (especially working toward a degree in clinical investigation), and other health professionals training to be sophisticated managers or consumers of clinical trials. For such an audience, the goal is to provide breadth and to write so as not to be misunderstood. This book should be supplemented with lecture and discussion, and possibly a computer lab. The reader who does not have an opportunity for formal classroom dialogue will need to explore the references more extensively. Exercises and discussion questions are provided at the end of each chapter. Most are intentionally made open-ended, with a suggestion that the student answer them in the form of a one-or two-page memorandum, as though providing an expert opinion to less-experienced investigators.

In short, this book targets clinical trialists, who are not so simple to define. Operationally, a clinical trialist is someone whose career is focused on the science of trials. A trialist uses investigational methods across disciplinary boundaries, compared to a specialist who might perform some trials in a single domain. Being true interdisciplinary experts, trialists can be derived from a number of sources: (i) quantitative or biostatistical, (ii) administrative or managerial, (iii) clinical, or (iv) ethics. Students can effectively approach the subject from any of these perspectives. Eventually however, a mature trialist will be conversant with important details from all the foundational fields.

It is common today for rigorous trialists to be strongly statistical. This is because of the fairly rapid recent pace of methods for clinical trials coming from that field, and because statistics pertains to all of the disciplines in which trials are conducted (or all of science for that matter). However, the discussion in this book does not neglect the other viewpoints that are also essential to understanding trials. Many examples herein relate to cancer because that is the primary field in which I work, but the concepts will generalize to other areas. Studying trials in different clinical disciplines is the best device for understanding principles. Unfortunately, the structure of many institutions and collaborations inhibits this.

Scientists who specialize in clinical trials are frequently dubbed “statisticians” by their clinical colleagues. I will sometimes use that term with the following warning regarding rigor: statistics is an old and broad profession. There is not a one-to-one correspondence between statisticians or biostatisticians and knowledge of clinical trials. However, trial methodologists, whether statisticians or not, are likely to know a lot about biostatistics and will be accustomed to working with statistical experts. Many trial methodologists are not statisticians at all, but evolve from epidemiologists or clinicians with a strongly quantitative orientation, as indicated above. My personal experience is that there are many excellent clinical trialists whom statisticians would label “physicians.” The stereotyping is not important. The essential idea is that the subject has many doorways.

In recent years, many biostatistics professionals have gravitated toward bioinformatics, computational biology, or high-dimensional data. I shall not take the space here to define these terms or distinguish them from research informatics or (bio)medical informatics. It is just worth pointing out that many biostatisticians have no experience with clinical trials. Technologies are of high value, but each wax and wanes with limited impact on methods of comparison.

The principles of valid comparisons in medicine have evolved over hundreds of years, and especially in the recent century. These principles have not been altered by...