How to calculate sensitivity, specificity, positive predictive value and negative predictive value

Video: Sensitivity, Specificity, PPV, NPV

Content

Steps
Method 1 of 1: Make Your Own Count
Tips
Warnings

In any test carried out on a given population, it is important to calculate sensitivity, specificity, positive predictive value and negative predictive value in order to determine how useful this test is in diagnosing a disease or characteristics of a given population group. If we want to use this test to investigate the characteristics of a selected population, we need to know:

How likely is the test to detect Availability signs in humans with characteristic features (sensitivity)?
How likely is the test to detect absence signs in humans without characteristic features (specificity)?
What is the probability of a person with positive the test result is actually there is signs (positive predictive value)?
What is the probability of a person with negative the test result is actually No signs (negative predictive value)?

It is very important to calculate these values in order to determine if a test is helpful in assessing the characteristics of a given population... In this article, we will show you how to calculate these values.

Steps

Method 1 of 1: Make Your Own Count

1 Construct a sample of the population, for example 1000 patients in a clinic.
2 Identify the disease or signs you are researching, such as syphilis.
3 Conduct a reliable gold standard test to determine the prevalence of disease or signs, such as information on the presence of bacteria pale treponema, obtained using a dark-field microscope, taking into account the clinical picture. Use a gold standard test to determine who has and who does not. For clarity, let's assume that 100 subjects have them, but 900 do not.
4 Design a test for the sensitivity, specificity, positive predictive value and negative predictive value of the population of interest, and test a sample of the population. For example, let's say this is a rapid plasma reagent (RPR) test for syphilis. Use it to sample 1000 people.
5 Of those with symptoms (as established by the gold standard), write down the number of people with positive and negative results. Test people who show no signs in the same way (as established by the gold standard). You will receive four digits. People with symptoms AND a positive result are true positive (PI)... People with symptoms AND negative results are false negative (LO)... People with no signs AND a positive result are false positive (LP)... People with no signs AND a negative result are true negative (IR)... For clarity, let's say you tested 1000 patients with RPR. 95 out of 100 patients with syphilis tested positive and 5 negative. Of the 900 patients who did not have syphilis, 90 tested positive and 810 negative. In this case, PI = 95, LO = 5, LP = 90 and IO = 810.
6 To calculate the sensitivity, divide the PI by (PI + LO). In the above case, we get 95 / (95 + 5) = 95%. Sensitivity tells us how likely a test is to test positive in a person with the symptoms.Among people with the symptoms, what proportion will test positive? A sensitivity of 95% is pretty good.
7 To calculate specificity, divide RO by (LP + RO). In the above case, we get 810 / (90 + 810) = 90%. Specificity tells us how likely the test is to test negative in a person who has no symptoms. Among people with no symptoms, what proportion will get a negative result? A specificity of 90% is pretty good.
8 To calculate the positive predictive value (PPV), divide PI by (PI + LP). In the above case, we get 95 / (95 + 90) = 51.4%. Positive predictive value tells us how likely a person with a positive test result will have the symptoms. Among people who test positive, what proportion actually have the symptoms? A PPV of 51.4% means that if you test positive, there is a 51.4% chance that you are actually sick.
9 To calculate the negative predictive value (NPV), divide the RO by (RO + LO). In the above case, we get 810 / (810 + 5) = 99.4%. Negative predictive value tells us how likely a person with a negative test result will have no symptoms. Among people who test negative, what proportion are truly symptomless? An HMO of 99.4% means that if you test negative, there is a 99.4% chance that you are not sick.

Tips

Good screening tests are highly sensitive and help identify patients who have symptoms. High sensitivity tests are useful in differential diagnosis diseases or signs if they are negative. ("SNOUT": sensitivity deviation)
Accuracy or efficacy is the percentage of test results accurately established by the test, that is, (true positive + true negative) / overall test results = (PI + RO) / (PI + RO + LP + LO).
Try to draw a contingency table to make it easier for yourself.
Remember that sensitivity and specificity are intrinsic properties of a given test that not depend on the given population group, that is, if the test is carried out on different population groups, these two values should remain unchanged.
Good control tests have a high specificity so that testing will not make mistakes in identifying patients with symptoms. High sensitivity tests are useful in diagnostics diseases or signs, if they show a positive result. ("SPIN": approval of specificity)
On the other hand, positive predictive value and negative predictive value depend on the level of prevalence of signs among the selected population group. The less common the signs, the lower the positive predictive value and the higher the negative predictive value (since the prevalence is lower in cases where the signs are less common). Conversely, the more frequent the signs are, the higher the positive predictive value and the lower the negative predictive value (since the prevalence is higher in cases where the signs are more common).
Try to understand these definitions well.