- Shopping Bag ( 0 items )
Doody's Review ServiceReviewer: Howard M. Kravitz, DO, MPH (Rush University Medical Center)
Description: This slim book is written by S. Nassir Ghaemi, MD, MPH, to introduce clinicians, residents, and students to the science behind the art of reading and critically appraising research studies.
Purpose: Dr. Ghaemi describes himself as a "clinical researcher in psychiatry," who has written this book "for clinicians and researchers in the mental health professions." Statisticians are categorically excluded from his audience because he expects they will "find it unsatisfying." The purpose is to teach readers how to understand, not to do, statistics. In sharing his insights and opinions, Dr. Ghaemi wants readers to understand conceptually what the numbers mean, to become better consumers of the literature rather than data technicians.
Audience: The target audience includes clinicians and researchers in the mental health professions, as well as residents and students. Although the content draws from the psychiatric and psychopharmacologic literature, the basic learning points will inform practitioners in all fields of medicine who need to read and apply the literature to patient care. Dr. Ghaemi notes: "one cannot be a good clinician unless one understands research." Learning to "navigate the scientific literature" bridges these two worlds.
Features: The 18 chapters are grouped into six sections (2 to 4 chapters per section): basic concepts, bias, chance, causation, the limits of statistics, and the politics of statistics. The book, a relatively short 137 pages, ends with a brief appendix on regression models and multivariable analysis. All references are compiled in a separate section. Historical perspectives and examples bring real-life relevancy to the book. The chapter on Bayesian statistics is particularly well written. True to the author's word, numbers make only cameo appearances. In total, I counted 10 figures, half in the chapter on Bayesian statistics, and six tables. The index is only moderately helpful. Clinical trials are described in a chapter on hypothesis testing, but an important omission is a chapter on research design, an essential journal club topic. Another concern is the author's sometimes and seemingly idiosyncratic use of terms, for example, "confounding bias." Whereas bias involves error in measuring a variable, confounding involves error in interpreting what may be an accurate measurement; conflating these two terms can be, well, confounding.
Assessment: This is an explication of Dr. Ghaemi's philosophy of numbers in the service of science. In reflecting on his learning and experiences, he tends to reveal his leanings, sometimes to the point of dogmatism. While informative, Dr. Ghaemi's nuanced explanations and reasoning might be better appreciated by persons who have some familiarity with the content. Overall, the book is one of the more thought-provoking of its genre, and reminiscent of the writings of Kenneth Rothman.