Clinical Chemistry 47: 2189-2190, 2001;
(Clinical Chemistry. 2001;47:2189-2190.)
© 2001 American Association for Clinical Chemistry, Inc.
|
Book, Software, and Web Site Reviews |
Diagnostic Strategies for Common Medical Problems. Edgar R. Black, Donald R. Bordely, Thomas G. Tape, and Robert J. Panzer, eds. Philadelphia: American College of Physicians–American Society of Internal Medicine, 1999, 674 pp., $45.00, Paperback. ISBN 0-943126-74-6.
M. Desmond Burke
Weill Cornell Center, New York Presbyterian Hospital, 525 E. 68th St., New York, NY 10021
This is the second edition of a text that seeks to blend information on the operating characteristics of laboratory (and other) tests with clinical estimates of pretest probability to provide a rational basis for the use of technology in clinical decision making. The first edition, published in 1991, grew out of an article entitled "Selection and Interpretation of Diagnostic Tests and Procedures: Principles and Applications", published as a supplement to the Annals of Internal Medicine (1981;94:553–600) by members of the General Medicine Unit at the University of Rochester. This followed the publication of Beyond Normality: The Predictive Value and Efficiency of Medical Diagnosis by Galen and Gambino (New York: Wiley, 1975), the monograph largely responsible for introducing the notion of "predictive value" to the laboratory community. The initial response to the quantitative and probabilistic approaches suggested in these publications was one of interest tempered by the realization that readily available data on the operating characteristics (sensitivity and specificity) of methods in use were largely unavailable. Things have improved considerably since then.
The current edition of Diagnostic Strategies begins with four chapters on the basic aspects of quantitative decision making and their application to the clinical decision-making process. As in the previous edition, the terms pre- and posttest probability replace prevalence and predictive value. These terms are preferable to predictive value, particularly negative predictive value, in that they make it easier to appreciate how sensitivity, specificity, and likelihood ratio data change the initial probabilities of particular diagnostic hypotheses. The effects of changing cutoff points, use of odds and likelihood ratios, combination testing, and in particular, the concept of posttest probabilities crossing "thresholds" and thereby confirming or excluding diagnostic hypotheses are all presented in lucid clinically relevant language. These are followed by 51 chapters, each dealing with a common clinical problem, e.g., myocardial infarction, pleural effusions, and hypercalcemia. With some exceptions, the editors have succeeded in establishing a uniform approach to the handling of each clinical problem. Each chapter begins with a highlighted "Key Points" section consisting of two bulleted lists, headed "Pretest Probabilities" and "Diagnostic Strategies". This is followed by sections on background information, estimating pretest probabilities, diagnostic tests, and diagnostic strategies. Shown in tabular form in the sections on diagnostic strategies are data on sensitivity, specificity, likelihood ratio, and posttest probabilities at degrees of low, intermediate, and high pretest probability. Each chapter ends with illustrative short case examples.
This book is clearly intended for the practicing physician with the goal of promoting rational, cost-effective test use. The inclusion of imaging tests in this current edition should be of value, not only to practicing physicians, but also to laboratory personnel less familiar with this component of diagnostic testing. From the point of view of the laboratory-based scientist or physician, the main value of the book may lie not so much in the test characteristic data provided as in the insight the text provides into the way clinicians use those data against the uncertain background of clinical practice. On a more basic level, the text includes much up-to-date information that should be helpful to laboratory personnel in their everyday dealings with clinicians. With that said, it should be pointed out that, although improved, the availability of information on test characteristics and pretest probabilities is still incomplete, and the difficulties inherent in assigning "rule-in" and "rule-out" thresholds are still largely unresolved. This means that many of the chapters do not quite live up to the expectations outlined in the initial chapters. Nevertheless, in the opinion of this reviewer, they represent the best that is available on the subject. I would recommend this book to all trainees, scientists, and physicians in laboratory medicine as representing the most clearly written, practical approach to the use of technology directed to the solution of commonly encountered clinical problems.
