Abstract
Objective: Suggesting patient-oriented ratio measures of diagnostic tests analogous to the likelihood ratios. Study Design and Setting: Algebraic equation based on Bayes' theorem. Results: Two new measures are suggested: the positive predictive ratio (PPR), relating the positive predictive value (PPV) to the false negative rate in the patient population, and the negative predictive ratio (NPR), relating the negative predictive value (NPV) to the false positive rate in the patient population. The new measures are analogous to likelihood ratios. These measures are calculable from the familiar 2×2 table in a selected study population or in a target (general) population. Examples are given, using published data. Conclusion: PPR adds to the information given by the PPV, and NPR adds to the information given by the NPV. PPR and NPR may be more appropriate in the clinical setting than likelihood ratios.
| Original language | English |
|---|---|
| Pages (from-to) | 450-457 |
| Number of pages | 8 |
| Journal | Journal of Clinical Epidemiology |
| Volume | 58 |
| Issue number | 5 |
| DOIs | |
| State | Published - May 2005 |
Keywords
- Bayes' theorem
- Clinical
- Diagnosis
- Patients
- Predictive value
- Test
ASJC Scopus subject areas
- Epidemiology