For each test performed on a reference population, it is important to calculate the sensitivity, the specificity, the positive predictive value, and the negative predictive value in order to determine how useful the test is for detecting a disease or characteristic in the target population. If we want to use a test to determine a specific characteristic in a population sample, we need to know:
- How likely is the test to detect the presence of a feature in someone having such feature (sensitivity)?
- How likely is the test to detect the absence of a feature in someone not having such feature (specificity)?
- How likely is a person who turns out positive to the test will have really this characteristic (positive predictive value)?
-
How likely is a person who turns out negative to the test he will not have really this characteristic (negative predictive value)?
It is very important to calculate these values for determine whether a test is useful for measuring a specific characteristic in a reference population. This article will explain how to calculate these values.
Steps
Method 1 of 1: Perform your calculations
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 1 Step 1. Choose and define a population to test, for example 1,000 patients in a medical clinic
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 2 Step 2. Define the disease or feature of interest, such as syphilis
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 3 Step 3. Obtain the best documented test example to determine disease prevalence or feature, such as a darkfield microscopic observation of the presence of the "Treponema pallidum" bacterium in a syphilitic ulcer sample, in collaboration with clinical results
Use the sample test to determine who owns the trait and who doesn't. As a demonstration, we will assume that 100 people have the feature and 900 do not.
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 4 Step 4. Obtain a test on the characteristic you are interested in determining the sensitivity, specificity, positive predictive value, and negative predictive value for the reference population, and run this test on all sample components of the selected population
For example, let's assume this is a Rapid Plasma Reagin (RPR) test for determining syphilis. Use it to test the 1000 people in the sample.
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 5 Step 5. To find the number of people who have the trait (as determined by the sample test), write down the number of people who tested positive and the number of people who tested negative
Do the same for people who do not possess the trait (as determined by the sample test). This will result in four numbers. People who possess the trait and who have tested positive are to be considered true positives (PVs). People who do not possess the characteristic and have tested negative are to be considered false negatives (FN). People who do not possess the trait and have tested positive are to be considered false positives (FP). People who do not possess the characteristic and have tested negative are to be considered true negatives (VN). For example, let's say you ran the RPR test on 1000 patients. Among the 100 patients with syphilis, 95 of these tested positive, and 5 tested negative. Among the 900 patients without syphilis, 90 tested positive and 810 tested negative. In this case, VP = 95, FN = 5, FP = 90, and VN = 810.
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 6 Step 6. To calculate the sensitivity, divide PV by (PV + FN)
In the above case, this would equate to 95 / (95 + 5) = 95%. Sensitivity tells us how likely the test will be positive for someone who possesses the characteristic. Of all the people who possess the trait, what proportion will be positive? A 95% sensitivity is a pretty good result.
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 7 Step 7. To calculate specificity, divide VN by (FP + VN)
In the above case, this would equate to 810 / (90 + 810) = 90%. Specificity tells us how likely the test will be negative for someone who does not possess the characteristic. Of all the people who do not possess the trait, what proportion will be negative? A specificity of 90% is a pretty good result.
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 8 Step 8. To calculate the positive predictive value (PPV), divide PV by (PV + FP)
In the above case, this would equate to 95 / (95 + 90) = 51.4%. The positive predictive value tells us how likely someone will have the characteristic if the test is positive. Of all those who test positive, what proportion does the characteristic really possess? A PPV of 51.4% means that if you test positive, you have a 51.4% chance of having the disease.
Calculate Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value Step 9 Step 9. To calculate the negative predictive value (NPV), divide NN by (NN + FN)
In the above case, this would equate to 810 / (810 + 5) = 99.4%. The negative predictive value tells us how likely someone will not have the characteristic if the test is negative. Of all those who test negative, what percentage do not really possess the characteristic? An NPV of 99.4% means that if you test negative, you have a 99.4% chance of not having the disease.
Advice
- Good detection tests have high sensitivity, because the goal is to determine all who possess the characteristic. Tests with high sensitivity are useful for to exclude diseases or characteristics if they are negative. ("SNOUT": acronym for SeNsitivity-rule OUT).
- There precision, or efficiency, represents the percentage of results correctly identified by the test, i.e. (true positives + true negatives) / total test results = (PV + NV) / (PV + NV + FP + FN).
- Try drawing a 2x2 table to make things easier.
- Good confirmatory tests have a high specificity, because the goal is to have a test that is specific, avoiding mislabeling those who test positive for the characteristic but who do not actually have it. Tests with a very high specificity are useful for confirm the diseases or characteristics if they are positive ("SPIN": SPecificity-rule IN).
- Know that sensitivity and specificity are intrinsic properties of a given test, and that Not depend on the reference population, in other words these two values should remain unchanged when the same test is applied to different populations.
- Try to understand these concepts well.
- The positive predictive value and the negative predictive value, on the other hand, depend on the prevalence of the characteristic in a reference population. The rarer the trait, the lower the positive predictive value and the higher the negative predictive value (because the pretest probability for a rare trait is lower). Conversely, the more common the characteristic, the higher the positive predictive value and the lower the negative predictive value (because the pretest probability for a common characteristic is higher).
