Table 1 Definitions of measures of diagnostic accuracy
Target condition | |||||||
|---|---|---|---|---|---|---|---|
Present | Absent | ||||||
Test result | + | a | b | ||||
- | c | d | |||||
Sensitivity | a/(a + c) - Proportion of true positives that are correctly identified by the test [31] | ||||||
Specificity | d/(b + d) - Proportion of true negatives that are correctly identified by the test | ||||||
Likelihood ratio (LR) | Describes how may times a person with disease is more likely to receive a particular test result than a person without disease [32] The interpretation of likelihood ratios depends very much on clinical context. Likelihood ratio for positive result (LR +) = [a/(a + c)]/[b/(b + d)] = sensitivity/(1 -specificity) Likelihood ratio for negative result (LR -) = [c/(a + c)]/[d/(b + d)] = (1 - sensitivity)/specificity | ||||||
Diagnostic odds ratio (DOR) | Used as an overall (single indicator) measure of the diagnostic accuracy of a diagnostic test. It is calculated as the odds of positivity among diseased persons, divided by the odds of positivity among non-diseased. When a test provides no diagnostic evidence then the DOR is 1.0. [33] This measure has a number of limitations: by combining sensitivity and specificity into a single indicator the relative values of the two are lost i.e. the DOR can be the same for a very high sensitivity and low specificity as for very high specificity and low sensitivity [33] Further, tests that are effective for classifying persons as having or not having the target condition have DORs that whose magnitude is much greater (e.g. 100) than usually considered as indicating strong associations in epidemiological studies. [34] DOR = [a/c]/ [b/d] = [sensitivity/(1 -specificity)]/[(1 - sensitivity)/specificity] = LR +ve/LR -ve = ad/bc | ||||||
Predictive value | Positive predictive value: proportion of patients with positive test results who are correctly diagnosed Positive predictive value (PPV) = a/ (a + b) Negative predictive value: proportion of patients with negative test results who are correctly diagnosed Negative predictive value (NPV) = d (c + d) Predictive values depend on disease prevalence, the more common a disease is, the more likely it is that a positive test result is right and a negative result is wrong. [35] | ||||||