Customer Reviews

Sample Size Calculations in Clinical Research (Chapman & Hall/CRC Biostatistics Series)

5 star:		(0)
4 star:		(1)
3 star:		(2)
2 star:		(0)
1 star:		(0)

 

 

The authors of this book have a great deal of clinical trial experience in the pharmaceutical industry as well as strong academic backgrounds. For the clinical trial statistician there is now a rich supply of software products to aid in the determination of sample size for a variety of modeling situations. So knowing formulas is no longer important. What is important is to understand the basis for the formulas. This book provides the industrial perspective and the main fixed sample size designs. In this industry trials are constructed to show superiority, noninferiority and equivalence. These three distinct appoaches lead to different results because the null and alternative hypotheses change as you change your goal from superiority to equivalence.

This book makes that important distinction and is very scholarly, providing many of the relevant references. Although most clinical trials are still parallel design randomized controlled trials with fixed sample size, there are more and more trials that allow for sequential decisionmaking and hence the actual total sample size can be subject to randomness. The group sequential trials have been the most successful in this regard. But now there are also more flexible "adaptive designs" that are being used. For group sequential designs see the text by Jennison and Turnbull and for the adaptive designs Chow and Chang and a more recent applied text by Chang are very good sources of information. Software packages that are available to do group sequential and adaptive designs are East by Cytel, Seq+Trials by Insightful Corp., PASS by Number Crunchersand ADDPLAN by a German Company. Also statisticians like Mark Chang and Keaven Anderson have created their own routines for adaptive designs using the R programming language.

At first browse, this book looks a bit like one of Julius Bendat's excellent texts on time series analysis: dense in formulas, but rewarding the wade.

And then I tried to actually _work through_ their examples. A formula-rich book is NO place for typos.

I don't mind when the text uses "lossed" for "lost;" I can quickly figure out what was meant. I resent having to do forensics to rebuild what formulas and/or results I should have seen in the examples.

That three-star rating reflects two things: the potential this book could have had, and my expectation that sooner or later there will be an ERRATA listing that helps sort this beast out.

I am a PhD statistician working in a pharma company. My overall impression about this book is that it is not a handbook for use on a daily basis for clinical trialists. I expected the book to be more insightful about sample size calculations. Examples are very artificial, not taken from real-life.

The book has an academic flavor, however the intended audience is clinical trials practicioners. It would be much better to start each chapter with a couple of strong examples, then description of the methodology, and finally sample size calculations.

I think this book needs a lot of improvement before it can be used as a good reference