Introducing an accessible series on statistics for clinicians

For many clinicians, statistics is the equivalent of a foreign language: they may know a few words here and there from their travels, but they have never had the time to learn the language properly. As health care providers, we are increasingly being asked to engage in critical appraisal and sort through the large volume of research to help guide decision making. For many, this means reading mainly the abstract and the discussion, and glossing over the jargon in the methods and results. This is unfortunate, as the methods can obviously make or break the validity of the results and determine whether we decide that a study is valid and practice changing, or fatally flawed and pointless.

This is not a novel endeavour. The whole evidence-based medicine movement began with the Users’ guide to the medical literature series originally published in JAMA in the early 1990s and now compiled in a book.1 These articles focused mainly on study design and introduced a whole generation of practitioners to clinical epidemiology. However, with the rise of desktop statistical packages, such as SPSS, STATA and SAS, complex statistical methods have been put within the reach of many investigators. The results have generally been positive in that complex analyses can be performed by many more people, but the room for error has also increased tremendously. Therefore, the need for caution and critical appraisal is even more urgent. Many series on basic statistics for clinicians have been published — such as a primer for clinicians in the Canadian Medical Association Journal2 and the ongoing statistics notes in The BMJ3 — but we have decided to take a fresh look at this topic with the purpose of providing a concise and accessible overview of commonly used statistical tests. The series will cover a number of statistical ideas and methods commonly used in medical studies and will be published at regular intervals. It will build over successive articles, so it might be useful to read them in order (at least initially). We have chosen not to cover some topics that have already been comprehensively discussed in the literature (eg, randomised controlled trials), and to cover others that may be less familiar (eg, receiver-operator characteristic curves). Our hope is that, by the end of this series, you will have grown in practical knowledge and will be reading the methods and results of research articles and feel empowered to come to your own conclusion about the wheat and the chaff in medical literature. We welcome suggestions from readers or contributions from other authors that will help to expand the series.