Understanding Biostatistics

There are many biostatistics books in print, apparently because no one book meets the need to be both (1) intelligible to a busy medical student, house officer, or clinician, and (2) sufficiently complete for those wanting a thorough treatment of the subject. The first question, therefore, is: Does this book fill a needed space in the biostatistical landscape, and if so, what is that space? Based on many years of teaching biostatistics to medical students, I would suggest a simple taxonomy for introductory biostatistics texts, which consists of two variables with two levels each. The first variable is depth, with two levels I will call "acquaintance" and "thorough understanding." The second variable is coverage of "clinical" versus "only classical" statistical topics. The adjective "clinical" is for books that cover topics such as measuring clinical error and making decisions at the bedside: sensitivity, specificity, and predictive value; receiver-operator curves; the application of Bayes' theorem to clinical decisions; indices of agreement such as kappa; decision analysis trees, measurement of vaccine efficacy; life table (survivor-ship) analyses such as that of Kaplan and Meier, and perhaps simple epidemiologic statistics such as risk ratios and odds ratios. The "only classical" statistics categories are well known. This provides a fourfold classification of biostatistics books for (1) those who want only to be acquainted with biostatistics from a strongly clinical perspective (in order to operate effectively at the bedside and understand the literature); (2) those who want only to be acquainted with classical biostatistical topics, probably for laboratory research; (3) those who want a thorough understanding of the subject from a clinical orientation; and (4) those who want a thorough understanding of classical, but not necessarily clinical, biostatistics. If one finds the right book covering classical biostatistics, it is possible to get most or all of the clinical topics from some of the newer texts on "clinical epidemiology." The Hassard book falls into category four, i.e., it provides a thorough introduction to classical biostatistics, and, although most of the examples are from the field of medicine, it does not really take a clinical orientation. Among the biostatistics books in this category, the Hassard book is the clearest and most user-friendly book I have read. And it is thorough. For example, in most introductions to biostatistics, the subject of "multiple associations" is mentioned only briefly, if at all, despite its critical role in a proper understanding of clinical …