Current issues of ACP Journal Club are published in Annals of Internal Medicine

Resource Corner

How to Read a Paper


ACP J Club. 2001 Sep-Oct;135:A14. doi:10.7326/ACPJC-2001-135-2-A14

How to Read a Paper aims to introduce clinical and nonclinical readers to the basics of evidence-based medicine. It is structured around a series of articles first published in BMJ in 1997 and later collected into a book, but several changes have been made to it since its first edition.

With 12 chapters and 4 appendices, Greenhalgh introduces the reader to the reasons for practicing evidence-based medicine, searching the literature, research methods, and how to critically appraise a paper. Specific chapters discuss how to appraise different study types (therapy, diagnosis, systematic reviews, guidelines, economics, and qualitative research), statistics, and common sources of bias in medical research. The final chapter addresses how to implement evidence in clinical practice. The 4 appendices contain critical appraisal checklists, search filters for MEDLINE, and a summary of mathematical terms used in clinical trials.

This book is clearly written and easy to read. It offers several examples and avoids technical jargon and complex mathematical concepts: The chapter on statistics gives some examples of how authors may try and “trick” readers with statistics, before proceeding to describe the major types of statistical tests and when they should be used. The chapter on diagnosis introduces a potentially complex topic by using the analogy of a “jury” (a diagnostic test) in predicting whether someone is guilty or innocent of murder (the target disorder). Once the basic concepts have been established, the chapter proceeds with a clinical example of urine glucose testing for diabetes mellitus. Certain concepts, however, are dealt with more superficially. Asking clinical questions, the use of biomedical databases (other than MEDLINE), and the appraisal of etiologic and prognostic studies receive only brief mention. Many mathematical concepts are not given their common names. For instance, the terms “control event rate” (CER), “experimental event rate” (EER), and “patient expected event rate” (PEER) are not listed in the index or in a glossary, which may be confusing to readers.

In summary, How to Read a Paper is a clear, well-written introduction to evidence-based medicine and uses interesting and relevant examples from clinical practice. Although certain areas could be covered in more depth, each chapter contains references to many other sources of information. It is likely to appeal to a wide audience but perhaps especially to persons who are nervous about mathematics or to those new to the field of evidence-based medicine.

Andrew McIntosh, MB, ChB, MPhil
Royal Edinburgh Hospital
Edinburgh, Scotland, UK


Clinical Usefulness: 4 of 5 stars

Greenhalgh T. How to Read a Paper. 2d ed. London: BMJ Books; 2000.

How to Read a Paper can be purchased online at for £16.95.