Synopsis
This text is designed for laboratory workers and clinicians but should also be useful for other healthcare workers. The authors assume that the reader has access to a computer with statistical software for inspection and analysis of data. The book has four main parts. Chapter 1 describes the different kinds of data and statistical analyses that can be applied to them. Topics include how to assess the shape of distributions, how to transform data, when to use parametric and non-parametric tests, and how to check data input and deal with outliers. Chapter 2 deals with the inaccuracy, imprecision, detection limits, analytical goals and other aspects of analytical methods used in laboratory medicine, and explains how to compare both quantitative and qualitative analytical methods. Chapter 3 describes how to obtain and use data for diagnosis and for specificity of tests and how to design parallel and sequential test strategies. It discusses the use of ROC analyses, Baye's theorem and likelihood ratios and describes how to determine the Number Needed to Treat. Chapter 4 deals with study design, analysis and reporting in medical research.
Topics include the types of study needed for evidence-based medicine and for clinical audit and the practical aspects of research project management. It provides advice on the presentation of abstracts, posters, papers and theses and how to respond to referee's reports. The book ends with a broad review of the use of personal computers and software in medical science. The authors also include brief biographies of five scientists who made major contributions to the development of medical statistics and evidence-based medicine - Johann Gauss, R.A. Fisher, Edwards Deming, Austin Bradford Hill and Archie Cochrane.
