Applied Survival Analysis: Regression Modeling of Time to Event Data
Rao, P. V., Journal of the American Statistical Association
David W. HOSMER Jr. and Stanley LEMESHOW. New York: Wiley 1999. ISBN 0-471-15410-5. xiii + 386 PP. $79.95.
This book is patterned after the text Applied Logistic Regression, by the same authors. A data-based approach is taken to provide a description of regression models appropriate for survival data arising in health-related studies. The book is a good source of information about proportional hazards (PH) models and contains an up-to-date list of references to alternative methods for analyzing survival data.
There are a number of fine texts on survival analysis. In addition to the now-classical text by Kalbfleisch and Prentice (1980), these include texts by Miller (1981), Cox and Oakes (1984), Fleming and Harrington (1991), Lee (1992), Anderson, Borgan, Gill, and Keiding (1993), Collett (1994), and Marubini and Valsecchi (1995), among others. The texts of Kalbfleisch and Prentice, Fleming and Harrington, and Anderson et al. provide detailed accounts of survival analysis at fairly advanced theoretical level, with the latter two books taking the modern approach based on counting-process theory. The texts by Miller and Cox and Qakes are written at a lower theoretical level and provide concise treatments of the topic suitable for a one-semester course on survival analysis at the undergraduate and graduate levels.
The level of writing in Applied Survival Analysis is similar to that in the texts of Lee, Collett, and Marubini and Valsecchi, but unlike the latter texts, the former does not address issues pertaining to clinical trials. Nevertheless, the book is a welcome addition to the literature because of its emphasis on tying survival analysis regression methods to methods based on standard linear regression analysis. The authors' writing style and their emphasis on detailed interpretation of results on the basis of real data analysis makes the book an ideal source of reference for practicing statisticians. The book can also be used as a text for a course in applied survival analysis for graduate students with good background in linear and logistic regression analyses.
The book contains nine chapters and three appendices. After introducing some basic concepts such as censoring and exponential regression models, Chapter 1 describes three datasets that are used throughout the remainder of the text. Chapter 2 introduces standard methods of estimation and hypothesis testing involving one or more survival functions. The chapter introduces the use of the Kaplan-Meier estimate and the weighted logrank statistic.
Chapters 3 and 4 are devoted to fitting and interpreting proportional hazards (PH) models, and Chapters 5 and 6 describe covariate selection and residual analyses on the basis of PH model. …