This unique book presents a framework for the strategy and methodology of diagnostic research, in relation to its relevance for practice. Now in its second edition The Evidence Base of Clinical Diagnosis has been fully revised and extended with new chapters covering the STARD guidelines (STAndards for the Reporting of Diagnostic accuracy studies) and the multivariable analysis of diagnostic data.

With contributions from leading international experts in evidence-based medicine, this book is an indispensable guide on how to conduct and interpret studies in clinical diagnosis. It will serve as a valuable resource for all investigators who want to embark on diagnostic research and for clinicians, practitioners and students who want to learn more about its principles and the relevant methodological options available.

Andre Knottnerus, Professor of General Practice, Maastricht University. He is involved in the Cochrane movement for drawing up tools for accurate diagnostic testing

Frank Buntinx, Professor, Department of Public Health, Leuven University. He is one time chair of the Cochrane Collaboration Diagnostic and Screening Tests Working Group

Evidence Base of Clinical Diagnosis

Theory and Methods of Diagnostic Research

John Wiley & Sons

Copyright © 2009 Blackwell Publishing
All right reserved.
ISBN: 978-1-4051-5787-2

Chapter One

General introduction: evaluation of diagnostic procedures

J. Andr Knottnerus, Frank Buntinx, and Chris van Weel

Summary box

Whereas the development of diagnostic technologies has greatly accelerated, the methodology of diagnostic research still lags behind that of evaluation of treatment.

Diagnostic research appears to be more comprehensive and complex than treatment research as it evaluates the connection between diagnostic and prognostic assessment with choosing optimal interventions.

Objectives of diagnostic testing are (1) detecting or excluding disorders, (2) contributing to further diagnostic and therapeutic management, (3) assessing prognosis, (4) monitoring clinical course, and (5) measuring general health or fitness.

Methodological challenges include dealing with complex relations, the "gold standard" problem, spectrum and selection bias, "soft" outcome measures, observer variability and bias, optimizing clinical relevance, appropriate sample size, and rapid progress of applicable knowledge over time.

Choosing the appropriate study design depends on the research question; the most important designs are the cross-sectional study (to determine the accuracy and added discriminatory value of diagnostic procedures) and the randomized controlled trial (to evaluate the clinical impact of [additional] testing).

To synthesize the results of various studies on the same topic, diagnostic systematic reviews and meta-analyses are powerful tools.

To make the step from research to practice, clinical decision analysis, cost-effectiveness studies, and quality-of-care research, including implementation studies, are important.

Introduction

The development and introduction of new diagnostic technologies have accelerated greatly over the past few decades. This is reflected in a substantial expansion of research on diagnostic tests and of publications on diagnostic research methodology. However, the evaluation of diagnostic techniques is far from being as established as the evaluation of therapies.

Apart from regulations for obvious safety aspects, at present-unlike the situation with regard to drugs-there are no widely accepted formal requirements for validity and effectiveness that diagnostic tests must meet to be accepted or retained as a routine part of health care. This is related to another point: in spite of important early initiatives the methodology for evaluation of diagnostics is not yet as crystallized as the deeply rooted principles of the randomized controlled trial on therapeutic effectiveness and of etiologic study designs. It is not surprising, then, that serious methodological flaws are often found in published diagnostic studies.

An additional challenge is the comprehensiveness of diagnostic research. Diagnostic evaluation is the first crucial medical intervention in an episode of illness, labeling symptoms and complaints as health problems, and indicating possible disease and its prognosis. Effective and efficient therapy-also including reassurance, "watchful waiting," and supporting patient self-efficacy-depends to a large extent on an accurate interpretation of (early) symptoms and the outcome of the diagnostic process. Accordingly, as diagnosis covers not only test accuracy but is also the basis for prognosis and appropriate treatment choice, diagnostic research is in fact much more complex than treatment research.

A special point of concern is that the funding of diagnostic evaluation studies is not well organized, especially if the research is not focused on particular body systems or disorders covered by strong research foundations. Rather than being limited to a particular body system, diagnostic evaluation studies frequently start from a complaint, a clinical problem, or certain tests.

Because the quality of diagnostic procedures is indicative for the quality of health care as a whole, it is vital to overcome the shortfall in standards, methodology, and funding. Accurate evaluation of diagnostic performance will contribute to the prevention of unjustified treatment, lack of treatment or mistreatment, and unnecessary costs. In this context, important steps forward toward the professionalizing of diagnostic studies have already been made with the work on the architecture of diagnostic studies (see also Chapters 1, 2, and 14), the Standards for Reporting of Diagnostic Accuracy studies (STARD) (see Chapter 9), and QUADAS, a tool for the Quality Assessment of Diagnostic Accuracy Studies included in systematic reviews. As a general background, this introductory chapter presents an overview of the objectives of diagnostic testing and evaluation research, important methodological challenges, and research design options.

Objectives

Diagnostic testing can be seen as the collection of additional information intended to (further) clarify the character and prognosis of the patient's condition and can include patients' characteristics, symptoms and signs, history and physical examination items, or additional tests using laboratory or other technical facilities. A "test" not only must be considered but also the specific question the test is supposed to answer. Therefore, the performance of tests must be evaluated in accordance with their intended objectives. Objectives may include:

Detecting or excluding disorders, by increasing diagnostic certainty as to their presence or absence. This can only be achieved if the test has sufficient discrimination. Table 1.1 shows the most common measures of discrimination. Most of these can be simply derived from a 2 2 table comparing the test result with the diagnostic reference standard, as demonstrated by the example of ankle trauma. A more elaborate and comprehensive explanation of how to calculate these and other measures from collected data is presented in Chapter 7. Examples of tests for which such measures have been assessed are given in Table 1.2. Such a representation allows various tests for the same purpose to be compared. This can show, for example, that less invasive tests (such as ultrasonography) may be as good as or even better diagnostically than more invasive or hazardous ones (e.g., angiography). Also, it can be shown that history data (e.g., change in bowel habit) may be at least as valuable as laboratory data. What is important is not just the discrimination per se, but rather what a test may add to what cheaper and less invasive diagnostics already provided to the diagnostic process. This is relevant, for instance, in assessing the added value of liver function tests to history taking and physical examination in ill-defined, nonspecific complaints. Finally, using the measures defined in Table 1.1 allows the comparison of the value of a test in different settings, for example, general practice versus the hospital emergency department.

Contributing to the decision-making process with regard to further diagnostic and therapeutic management, including the indications for therapy (e.g., by determining the localization and shape of a lesion) and choosing the preferred therapeutic approach.

Assessing prognosis on the basis of the nature and severity of diagnostic findings. This is a starting point for planning the clinical follow-up and for informing and-if justified-reassuring the...