BJS Bookshelf: The Art of Statistics – Learning from Data.

Statistics is a subject that, in my experience, is seldom taught in medical school. This is peculiar, given the emphasis on evidence-based medicine and the role of research. The Art of Statistics – Learning from Data by the University of Cambridge statistician Sir David Spiegelhalter is an important book in several ways. It offers an excellent introduction to statistics, without becoming too technical. The book is written in an engaging and jargon-free fashion and is replete with interesting examples and anecdotes to illustrate statistical concepts and conundrums. Without doubt, the book helps the reader to develop a statistical approach to analyzing and interpreting data.

The book encompasses fourteen separate chapters on a vast array of topics, ranging from causation, regression, algorithms, and Bayesian statistics. It begins with a chapter on basic concepts, such as categorical data and percentages. Here Spiegelhalter convincingly shows how relative risks often exaggerate risks, and how important it is to put things into context by providing the absolute risk. To illustrate this, Spiegelhalter shows how the media decided to misinterpret a report by the World Health Organizations (WHO), in which processed meat was classed as a ‘Group 1 carcinogen’, just like cigarettes and asbestos. The report, which the WHO relied on, showed that 50g of processed meat a day was associated with an increased risk of bowel cancer of 18%. The media reacted immediately, and wrote: ‘Bacon, Ham and Sausages Have the Same Cancer Risk as Cigarettes Warn Experts’. This sounds ‘worrying, but should it be?’ asks Spiegelhalter. Thereafter, Spiegelhalter shows, with simple arithmetic, that an increase of 18% relative risk corresponds to a rise from six to seven cases out of one hundred. In other words, there is one extra case of bowel cancer in all those one hundred life-time-bacon-eaters – a fact, which many of us do not find so impressive. I also found the chapters on causation and regression informative. Here Spiegelhalter examines, inter alia, the terms confounder and causation, and how an adjustment for dealing with confounders, at times, can produce paradoxical results, where the apparent direction of an association is reversed by adjusting for a confounding factor. This is known as Simpson’s paradox. Although not a textbook on statistics, The Art of Statistics, helps the reader to understand basic statistical concepts, and the manifold examples that are used to illustrate statistical concepts will hopefully convince most surgeons to read it.