Effects of Prevalence on Diagnostic Tests
Interpretation Continued...
Based on the differing PPV and NPV of a breast lump in different clinical settings, look at the different approach taken to women with the same physical finding, but different pre-test probabilities (or risk):
Setting | Prevalence |
PPV |
Advice |
Family Physician’s Office |
1% | 7.5% | You have a lump, most likely benign. Further investigation, likely noninvasive diagnostic imaging. |
Breast Screening Clinic | 3% | 20% |
You have a lump, probably benign. Further investigation, likely noninvasive diagnostic imaging. |
Surgical Clinic | 50% | 89% | You have a lump, probably cancerous. Further investigation needed, including immediate biopsy |
Here is an example of the different approach to the women with NO lump detected, but different baseline pre-test probabilities:
Setting |
Prevalence |
NPV |
Advice |
Family Physician’s Office |
1% | 99.8% | No lump. Reassurance. |
Breast Screening Clinic |
3% | 99.0% |
No lump. Reassurance. |
Surgical Clinic | 50% | 82% | No lump, but still some risk of cancer (~20%). Further investigation discussed, such as screening diagnostic imaging. |