Introduction to Sensitivity and Specificity
These concepts refer to characteristics of a diagnostic test.
- Sensitivity (sometimes called the true positive rate) is the percentage of patients who have a disease that test positive on the test.
- Specificity (sometimes called the true negative rate) is the percentage of patients who don’t have the disease who test negative on the test.
A perfect test would have 100% sensitivity and 100% specificity. It would make no mistakes. 100% sensitivity means that it would not miss any patients who have the disease. 100% specificity means that it would not erroneously classify anyone who is disease-free as having the disease.
Calculating Sensitivity and Specificity
The calculations for sensitivity and specificity are fairly straightforward, but it’s easy to get confused by the 2 x 2 tables. If you set the table up properly, these calculations are made down the columns (as opposed to across the rows). It may also help to think of sensitivity as the true positive rate and specificity as the true negative rate.
Tradeoffs between sensitivity and specificity
For most tests, if you increase sensitivity, specificity will drop. And vice versa. While it is possible to have a test that has both 100% sensitivity and 100% specificity, chances are that in those cases distinguishing between who has disease and who doesn’t is so obvious that you didn’t need the test in the first place.
ROC curves
One way of plotting the sensitivity and specificity of a test is through using Receiver Operating Characteristic (ROC) curves. I know I say receiver OPERATOR characteristic in the video, but that’s not right. It’s OPERATING. You still get the idea.
Screening tests
I’m going to redo this video, some of the information about bias may be confusing at the end. So just ignore that for now.
Predictive Values
We can also calculate predictive values from the 2×2 table. With sensitivity and specificity, we looked at how well a test performed either in patients who had the disease (sensitivity) or in patients who didn’t have the disease (specificity). If we shift our focus from the disease to the test, we can look at predictive values.
- Positive predictive value (sometimes called precision) is the percentage of the time that a positive test correctly identifies people who have the disease.
- Negative predictive value is the percentage of time that a negative test correctly identifies people without the disease.
The shift from sensitivity and specificity to predictive values provides us with information that may be more relevant to patients. When we have a patient sitting in front of us and a test comes back as “positive,” positive predictive value tells us how likely it is that the patient has the disease. If the test comes back as negative, negative predictive value tells us how likely it is that the patient is illness free.
A problem with predictive values is that they are very sensitive to the prevalence of the disease in the population.
Test your comprehension
With this sensitivity, specificity and predictive values problem set.
patwari 