| |||||||||
| Bioscience book menu |
|
 | ||
| By: Mark Chang ISBN: 1584889624 Publisher: Chapman & Hall/CRC Release Date: 27 June, 2007 Bioscience book rank: 383194 | There are explosions of adaptive design papers in past several years. This book alone has included about 400 references. It is very confusing to most new researchers in this field. This book use a unified approach to treat the major hypothesis test based adaptive design methods, i.e., view different methods as some forms of stagewise p-values combinations for test statistics. Chapter 1 provides overview of adaptive designs. Chapter 2 provides background for various clinical trials including superior, non-inferiority, equivalence and dose-response trials. The unified approach is presented in chapter 3 for stopping boundary determination, adjusted p-value, early futility and efficacy stopping, expected sample-size and clinical trial duration, conditional power, and futility index. All the formulations for these operating characteristics are presented in multiple-integration forms. In the next several chapters, all the integrations for the operating characteristics are carried out for particular combinations of p-values - lead to particular statistical methods for adaptive designs. In the most cases, the book avoid to using approaches from the original papers when the ideas were first proposed to avoid confusions and reduce the amount of material to be included. Chapter 7 presented another way (conditional error approach) to look at the common and different characteristics among different methods. Almost all methods for adaptive design can be reviewed as the conditional error approach. The difference is that each method uses a different conditional error function. In the chapter, different conditional error function and conditional power formulations are summarized. Chapter 8 discusses the recursive conditional error method so that it can be used for a K-stage adaptive design. Chapters 9 to 14 discuss different types of adaptive trials using the statistical methods that have been discussed in the previous chapters. These trials include sample-size reestimation, drop-loser design, biomarker adaptive design, response-adaptive randomization, adaptive treatment switch, and multiple endpoint issues. Chapters 15 and 16 discuss Bayesian adaptive approach for clinical trials. Chapter 17 talks about implementation issues. Chapter 18 is for readers who are interested in philosophical debates.
<br />
<br />If you have not read too much adaptive design research papers, you wouldn't be confused, and you may not appreciate the unified approach in this book.
<br />
<br />For most chapters, computer programs (SAS Macro) are provided with illustrated examples from clinical trials. However, it is not the author's intention to teach to how to implement adaptive design using SAS. The main purpose to include computer programs is to provide tools that you can use to design your adaptive trials since the software for adaptive design is very expensive (some reach [..]annual license for single user). It is not a computer book. Hence the algorithm of the computer program is usually not provided your clinical trials. However, each program is written with clear logic flows and is only about a page long. It should not be a challenge to most readers who have coding knowledge. The corresponding R functions are presented in Appendix. Because they are so similar between a SAS Macro and the corresponding R function. It is wise to put one of them in the appendix. The R functions cover the typical adaptive designs. Others can be directly translated from SAS macros without any difficulties.
<br />
<br />There are exercises at end of chapters. Some are good, some are OK. This should be enhanced for the revision.
<br />
<br />For some reasons, Amozon.com does not include the sample pages from the book. I am the author of book; I think it is helpful to use this feature to provide some insight for the readers. More information can be found online, where you can obtain the table of contents and the electronic computer programs. Rank it 5 starts that could be author's bias.
<br />
<br /> The book, going by the table of contents, provides a fairly comprehensive overview of the field of adaptive designs in drug development. <br />After having read it, I am somewhat disappointed. The topics are in fact all there, and the different approaches are presented. There is no real overview on how the different approaches link together though. <br />I think that other texts like Ting ([[ASIN:0387290745 Dose Finding in Drug Development (Statistics for Biology and Health)]]) do a much better job at providing the background. <br /> <br />The code seems quite useful, but the typesetting is fairly disastrous. Most functions and macros have many parameters, and they are listed in floating text style instead of a tabular layout, making it very hard to read. <br />The code is typeset in proportional font (where monospace is standard) and does not contain any comments and documentation of particular blocks. <br /> <br />Finally, the text comprises 27 SAS programs and only 6 R programs. The SAS programs are in the corresponding chapters, the R programs are all put at the very end of the book in its own appendix chapter. The R code is of fairly low quality, suggesting that the author is a SAS user and transcribed the code into R. <br />Example of some typical and not so great R code: <br />for(i in 1:nStgs) { TSc[i] = 0} <br /> <br />So the benefit of the book might be in the SAS library. It is not in the introduction to adaptive design theory and certainly not in the small R library, making the title somewhat misleading. <br /> <br /> This book just came out but I know a lot about it and about the author before I even got a copy. In November of last year Mark Chang coauthored a book in this Chapman and Hall series that I reviewed with praise because of the importance of the topic and the way it was demonstrated to work in a variety of real problems in pharmaceutical clinical trials. This book is even better as it goes more deeply into the methodology, the controversies and the results from simulation studies. Also it is much more practical because for every case where an application is given a SAS macro is also included to allow the reader to try the methodology for himself. In March of 2007 I actually designed a two-stage adaptive design with sample size reestimation for bioequivalence trials. I met mark at a conference where he presented much of his recent work and he was instrumental in helping me through his first book and his journal articles. This book had already gone to the publisher but he realized that this important design had overlooked. He added it when the copyedited version came to him. The design and the simulations related to it are very close to what I actually used. For those who like to program in R, he provides R code corresponding to each of the SAS macros that he gave. These programs make the new methodology readily available to interested users. The book is very comprehensive in that it covers a wide variety of applications for phase 2, phase 3 and combined phase trials. With the FDAs new initiative to speed up the drug discovery process this book will be an invaluable tool to statisticians in the pharmaceutical industry who would like to learn and apply these methods that along with the group sequential methodsare gaining favor within the FDA. | |
 | ||
| By: Chap T. Le ISBN: 0471416614 Publisher: Wiley-Liss Release Date: 15 October, 2001 Bioscience book rank: 35497 | + Note: This review was originally written for another book from this author, but touches on my experience with this text as well. <br /> <br />It is mind-boggling how one can have so many mistakes in a textbook like this. This book was required for our Biostatistics II class, and I am really displeased by all of the mistakes. What is the worst offense is that some of the answers for test questions are wrong. For example, the one way ANOVA problem 7.21 has the wrong F-statistic: the wrong answer. Others in the class have complained about this text as well. This is simply amazing to me, and leaves me not trusting much of the text in this book. <br /> <br />His other book "Health and Numbers" is just as terrible; the professor that used that book for class made a list of the mistakes they found, and us students in that class found many more along the way. I ran into another statistics book that takes all of the procedures step-by-step and really made it simple for me. This book, and his other are not recommended. | |
 | ||
| By: Shein-Chung Chow, Mark Chang ISBN: 1584887761 Publisher: Chapman & Hall/CRC Release Date: 16 November, 2006 Bioscience book rank: 411920 | This book's best features are its bibliography (about 240 entries) and its broad survey and taxonomy of adaptive methods. Its publication represents an important step in popularizing adaptive trials and, thus, streamlining drug/device/biologic development pipelines.
<br />
<br />The book is, however, filled with inaccuracies on several levels: incorrect grammar and equation references, undefined symbols, a reference to a non-existent appendix, unclear language (e.g., what is the "statistical strength for rejecting Ho" on Page 150?), mathematical typos [e.g., P(x|y,theta) rather than P(y|theta) in the integrand for the posterior predictive probability distribution P(y|x)], and misapplications of statistical philosophy (e.g., using Neyman-Pearson hypothesis testing for statistical inference, identifying the p-value as a post-hoc type I error rate). In the sample I took of about 1/3 of the pages, about 120 errors occur. The book should be considered only a pre-publication "beta" version. Any second edition should receive much more attention to detail.
<br />
<br />A statistician or clinical scientist planning a potentially adaptive trial could use this book to learn about some of the aspects of a trial that can be made adaptive. The book could also help him/her to assess the assumptions and mathematical complexity of methods under consideration. However, when it comes to actually performing an analysis, one would want to use the bibliography to obtain the relevant articles and books, perhaps together with Chang's "Adaptive Design Theory and Implementation Using SAS and R" (Chapman & Hall/CRC Biostatistics).
<br />
<br />Overall, this book disappointed me. The authors should have had several more collaborators and copyeditors check their work. I meet the second author, Mark Chang, at a conference on adaptive designs. I work as a professional statistician in the pharmaceutical industry. For the past several years, at least ten, these ideas have been the topic of research and it is being investigated as a possible way to speed up drug development and its development is being encouraged by the FDA. There has not been a formal statistical text covering the existing theory and its application to clinical trials. Consequently, when we knew this was coming out we preordered it and have been studying it since it came out last November. <br /> <br />The book has lived up to expectations. Adaptive designs are very similar to group sequential designs in that they have planned times to make preliminary assessment of the trial data and then decide whether or not to continue the trial or modify the design. Adaptive designs can be more flexible than their group sequential counterparts. They even can allow changes to the protocol as long as the criteria for making such changes are mapped out in advance of the trial. <br /> <br />These methods have been controversial in the past and simulation studies are often required to determine their properties. But there has been enough development now that some designs are being applied in real trials. In fact we are considering a two stage adaptive design similar to the ones described in this text (except applied to bioequivalence). <br /> <br />Later this year Mark Chang is coming out with an applied text that include SAS macros to aid in the implementation of the methods. A preview of the manuscript was displayed at an adaptive trials conference that I attended recently. I can enthusiastically recommend that one even more than this one! However, any biostatistician working on clinical trials should have this book on his or her bookshelf. | |
 | ||
| By: Gerald van Belle, Patrick J. Heagerty, Lloyd D. Fisher, Thomas S. Lumley ISBN: 0471031852 Publisher: Wiley-Interscience Release Date: 26 July, 2004 Bioscience book rank: 347764 | The authors write well and cover most of the important topics very thoroughly. They motivate the subject very well with a number of important and "real world" examples in the first chapter.
<br />A unique feature is its detailed coverage of sample size determination in a number of contexts.
<br />
<br />The book was published in 1993 which is not recent enough to cover advances in meta analysis, resampling, Bayesian Hierarchical Models (with Markov Chain Monte Carlo Methods) and frailty models. At least bootstrap methods and meta analyses are mentioned in the book.
<br />
<br />Noteworthy are the full chapters on multiple comparison problems and discriminant analysis. This is an excellent reference book for biostatisticians.
<br />
<br /> This text is a great reference for everything from basic analysis to more advanced topics like longitudinal and time-to-event data. The book is well-written with relevant, current examples. If I had to choose one reference book, this would be it because of its depth and breadth of content (at the intermediate level). This was a terrible book. To say that it is user unfriendly is an understatement. It is wordy and too long, furthermore, it doesn't teach you what you need to know. It is difficult to reference things (for example, try looking up sensitivity in the index!) and the writing is very tangential. This is not a good book if you are just starting out. I would not receommend it. | |
 | ||
| By: Muin J. Khoury, Julian Little, Wylie Burke ISBN: 0195146743 Publisher: Oxford University Press, USA Release Date: 23 October, 2003 Bioscience book rank: 458633 | ||
 | ||
| By: James F. Jekel, David L. Katz, Joann G. Elmore ISBN: 0721690793 Publisher: Saunders Release Date: 15 September, 2001 Bioscience book rank: 571121 | Worth its weight in gold. Covers the principles of every subject area required for board certification by the American Board of Preventive Medicine. Includes numerous lists, tables, charts, and a full set of referenced questions and answers. Downright poignant considering the precipitous decline in America's public health and preventive medicine infrastructure since the day George W. Bush assumed office.
<br />
<br />Mark Gary Blumenthal, MD, MPH This is a useful study guide and reference tool for anyone who is starting in epidemiology. It gives a great overview on study designs, statistics, and a great deal of other useful information for creating, implementing and evaluating a study. I have found it extremely useful in both my studies and in the implementation of studies. I would recommend this book as a study tool in any epidemiology course! | |
 | ||
| By: Jos W. R. Twisk ISBN: 0521614988 Publisher: Cambridge University Press Release Date: 15 May, 2006 Bioscience book rank: 326476 | The book is easy to understand and presents a series of situations where multilevel models are applied, in the context of epidemiological analysis. The examples are mostly very simple, and I don't think the book will help if you want to fit models. If you want to understand them better, then it may be of use. I have some other restrictions as well - some concepts presented in chapter 2 (section 2.8.1) are in direct contradiction with MLwiN user's guide (version 2, 2005, section 7.5, p.86). My understanding is that the latter is correct. Also, a model fitted in chapter 3 with a dichotomous variable random at level 2 includes the covariance parameter, what looks wrong to me. All in all, I think there are better options for an introductory text, e. g. Snijders & Bosker (1999). A good introduction to multilevel analysis that addresses the practical issues that confront researchers. Twisk's work is very good for those who are not experts in mathematics and would like to apply this statistic method for investigating cluster effect. Through the tutorial journey mainly illustrated by MLwiN, readers can easily catch the general picture of multilevel analysis and make sense of interpretation and application. The comparison of different statistic software used in multilevel analysis and the concise text for power analysis were another two precious gifts the author gave us. | |
 | ||
| By: Geoffrey R. Norman, David L. Streiner ISBN: 1550091239 Publisher: B.C. Decker Inc. Release Date: 15 July, 2000 Bioscience book rank: 174536 | This book is written as a review for people have already taken some kind of statistics or biostatistic. The funny unique writing style keeps me focus and actually helps me to understand the material better. People who have never taken any statistic class may have a harder time. However, why would any one who has no statistic background try to read a book titles "Biostatistics." There are alway those Dummy books. Even people with no statistic background but with some mathematics background will appreciate this book. Although I have long since graduated to much less easygoing fare, I love this book. When I was a confused and clueless medical student, this was the first text that helped me to come to terms with biostatistical concepts. <br /> <br />If you have a background in psychology or other softer sciences, are facing your first biostatistics class, and find other texts a bit scary or obscure, this is definitely worth a try. <br /> <br />The jokes become annoying and the examples a bit garbled on a second read, but I still love this book because of the clarity of its visuals. I also like and highly recommend the chapters on factor analysis and survival analysis, and usually give this book to new students in our lab. I took stat 20 years ago, and decided I needed an update, particularly regarding how to calculate sample size: this book was the ticket. A lot about sample size always seemed to me to be smoke-and-mirrors, and WHEN that's so (not always, but sometimes!), Norman and Streiner admit it, saving the learner a lot of frustration. Money well spent. | |
 | ||
| By: Noel S. Weiss ISBN: 019530523X Publisher: Oxford University Press, USA Release Date: 01 June, 2006 Bioscience book rank: 781652 | This book was a reference for the course "Clinical Epidemiology" taught by Noel Weiss. I originally borrowed this from the library, but after completing the course, I have realized this is a book I want to have onhand forever. Clear, concise with excellent examples. The chapters excellently describe the subtle issues with RCTs, case control, cohort studies and those designed to measure safety. | |
 | ||
| By: Shein-Chung Chow, Jun Shao, Hansheng Wang ISBN: 1584889829 Publisher: Chapman & Hall/CRC Release Date: 22 August, 2007 Bioscience book rank: 308102 | I found the errors in the book absolutely infuriating! Many typos in the text from the first edition are maintained in the second edition. The typos in both text and formulas in the expanded information in the second edition together with gaps in the development of the information made it necessary to consult other sources to figure out what's going on. This is the second edition of a very popular book on sample size estimation that is a valuable reference for any statistician in the industry. we all need to go through such expercises at the beginning of a trial as part of the protocol development. <br /> <br />One disappointment I have with the book is that it does not delineate new topics and other changes/additions from the first edition. Often this is covered by having two Prefaces, the original one from the first edition and a new one from the second edition. The authors unfortunately did not choose to do that. So an owner of the first edition would have to scan through both books to identify the changes. <br /> <br />Another disappointment is the lack of reference to any existing software to do sample size estimation and these days there are a lot of products available. The programs nQuery Advisor and Power and Precision handle equivalence, superiority and noninferiority problems for continuous data. They also provide approximate and exact methods for binomial data. Other packages such as StatXact handle sample size estimation for exact binomial tests as well as for Fisher's Exact test. PASS, S+SeqTrial and East are packages that provide the designs and sample size stopping rules for group sequential procedures and in some cases adaptive designs. Also with the development of version 9 of SAS comes the new procedures power and glmpower that do everything that nQuery can handle. <br /> <br />The value of the book is that it develops the methodology and therefore helps with the understanding of how and when to use the various procedures. Traditional tables that use to be important for sample size calculations are now obsolete given the availability of good software tools. Although the book goes to great lengths to cover almost any application. Most of these applications can be handled these days through the available software packages. <br /> <br />I can definitely recommend this book as a fine reference on sample size estimation for the wide range of trial applications. I would only try to encourage the authors to drop the use of tables and get up to date by recommending the appropriate software for the various applications. | |
Related books in this category:
bioinformatics, computational biology, sequence analysis, protein structure, data mining, biostatistics
Main book index: all categories
|
| ||||||||