The architecture of diagnostic researchBMJ 2002; 324 doi: http://dx.doi.org/10.1136/bmj.324.7336.539 (Published 02 March 2002) Cite this as: BMJ 2002;324:539
- D L Sackett, professor (email@example.com)a,
- R B Haynes, directorb
- a Trout Research and Education Centre at Irish Lake, RR1, Markdale, ON, Canada N0C 1H0
- b Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, ON, Canada L8N 3Z5
- Correspondence to: D L Sackett
This is the second in a series of five articles
Considerable effort has been expended at the interface between clinical medicine and scientific methods to achieve the maximum validity and usefulness of diagnostic tests. This article focuses on the specific kinds of questions that arise in diagnostic research and the study architectures (the conversions of these clinical questions into appropriate research designs) used to answer them. As an example we shall take shall take assessment of the value of the plasma concentration of B-type natriuretic peptide (BNP) in the diagnosis of left ventricular dysfunction.1 Randomised controlled trials are dealt with elsewhere.
As in other forms of clinical research, there are several different ways studying the potential or real diagnostic value of a physical sign or laboratory test, and each is appropriate to one kind of question and inappropriate for others. Among the possible questions about the relation between a putative diagnostic test and a target disorder (for example, the concentration of BNP and left ventricular dysfunction), four are most relevant.
Diagnostic studies should match methods to diagnostic questions
Do test results in affected patients differ from those in normal individuals?
Are patients with certain test results more likely to have the target disorder?
Do test results distinguish patients with and without the target disorder among those in whom it is clinically sensible to suspect the disorder?
Do patients undergoing the diagnostic test fare better than similar untested patients?
The keys to validity in diagnostic test studies are
independent, blind comparison of test results with a reference standard among a consecutive series of patients suspected (but not known) to have the target disorder
inclusion of missing and indeterminate results
replication of studies in other settings
Both specificity and sensitivity may change as the same diagnostic test is applied in primary, …