Competing Risks: A Practical Perspective (Statistics in Practice) [Hardcover]

In the medical world competing risks problems are common. Many times people avoid using competing risks methodology because they are not well understood and not well implemented in software.

I have found this book to be a very good introduction to competing risks survival analysis. I particularly liked how sample S-Plus/R code was given at the end of each chapter to demonstrate how to implement the methodologies presented. A major asset is that the author provides functions for calculating quantities not found in the "cmprsk" library for R. The book is well written with a minimum of mathematical detail, which I appreciated being an MD and not a PhD. I suspect that those people looking for a mathematical development will be disappointed. The last chapter contains some problems that the reader can attempt to solve if he wishes. The nice thing is that these problems all have worked out solutions, a feature that I really appreciated.

I did have some critiques of this book, however:

A) I think that the author should have spent a bit more time in chapters 2 and 3 developing the concepts of the various survival analysis functions. For instance, in a 4-5 page sequence in chapter 3 the author presents (1)the cumulative incidence function, (2) the overall distribution function, (3) the subsurvivor function, (4) the subdensity function, (5) the subhazard, (6) the overall hazard, (7) the hazard function of the subdistribution, (8) the cumulative subhazard function, (9) the multivariate joint survivor function, (10) the marginal survivor function, (11) the subdistribution function and (12) the hazard of the marginal distribution. This a common fault that I have found in many survival textbooks and is one major source of reader confusion. How many hazards and distributions can one reader reliably digest in a page of text??? Rather, I would have appreciated a slow presentation of these functions, one at a time, with an example of its calculation using real data and with some intuition as to its meaning and interpretation. Graphs of each would be very useful. This would mean a chapter 3 that is 2-3 times longer but much more useful, especially since understanding these quantities is critical to understanding the rest of the book.

B)The book should have dealt with some of the more complicated aspects of the competing risks model. For instance, how does the Fine-Gray model perform with (1) time-dependent covariates, (2) missing data, (3) longitudinal data and repeated measures? There are publications on many of these tough subjects that suggest that the competing risks model may not be interpreted in the same manner as the usual Cox model.

C) I think that the book should have provided more contrasts between routine survival analyses and the competing risks versions of these tests. Again, the goal is intuition about the implications, subtleties and complexities of the various ways to look at survival data.

Overall, I highly recommend this book to individuals looking to get a good introductory understanding of the competing risks framework and gain some practical knowledge on how to perform these analyses.