Overview of rational testing

Pathology, imaging and other tests have inherent limitations and risks and must be used thoughtfully, with consideration of the clinical context. Rational testing always considers both the qualities of the test (including sensitivity and specificity, and predictive value and likelihood ratio, which are derived from these) and the ‘pretest probability’ of the condition.

The pretest probability is determined from the prevalence of the condition in patients with particular clinical features (eg patient presenting with fatigue) or, if this data is not available, the prevalence of the condition in the population (eg patients presenting in primary care). For a guide to the prevalence of conditions associated with fatigue in different age groups, see A guide to the prevalence of conditions associated with fatigue in children and adolescents for children and adolescents, and A guide to the prevalence of conditions associated with fatigue in adults for adults.

The specificity of a test refers to its ability to correctly identify people who do not have the disease (‘true negatives’). The sensitivity of a test refers to its ability to correctly identify those with the disease (‘true positives’). The likelihood ratio helps assess how useful a test is to rule in or rule out a specific disease. The likelihood ratio for a positive test result (LR+) is the ratio of ‘true positives’ and ‘false positives’, calculated as sensitivity/(1-specificity); it is an indication of how much more likely the patient is to have the specific disease after a positive test result. Likelihood ratios are properties of particular tests and therefore usually independent of disease prevalence in a population.

‘Post-test probability’ utilises a combination of patient- and test-related factors to estimate whether a specific test is likely to contribute to a diagnosis. Appendix 16.3 shows worked examples for breast cancer and iron deficiency, to illustrate how post-test probability is calculated. In these worked examples, post-test probability is calculated using odds and likelihood ratios. However, as many practitioners intuitively consider probabilities (or risks) rather than odds, a nomogram1 is available to estimate post-test probability from pretest probability, without needing to convert probabilities to odds and then odds back to probabilities.

For links to more detailed information about the principles of rational testing, see Further reading: rational testing.

For definitions of statistical terms, see Fatigue: glossary of statistical terms.

For discussion about investigations in patients with fatigue, see When to investigate fatigue.

1 Deeks JJ, Altman DG. Diagnostic tests 4: likelihood ratios. BMJ 2004;329(7458):168-9. URLReturn