Reply To: sensitivity and specificity
I think you’ve got it! Don’t know the level of detail required?
Sensitivity and specificity are test-based calculations and are independent of the population.
I always think it’s useful to remember there are 2 groups within your population. Those with disease and those without and the ability of the test to perform in each of these groups is different which is where the 4 categories below come from.
True positive = Animals who have the condition and test positive for the condition.
False positive = Animals who do not have the condition but test positive for the condition
True negative = Animals who do not have the condition and test negative for the condition
False negative = Animals who have the condition and test negative for the condition
Sensitivity is the tests ability to detect those who DO have the condition.
It is in relation to positives. So for this test say we use 100 dogs (easy maths) we know have the disease as the population. 96% of them test positive, 4% are false negatives (we know they have the disease but the test fails to recognise it).
Sensitivity (True positive rate) = True positives / (true positives + false negatives)
So the proportion of the population that does have the condition as a percentage of animals that test positive. So your sensitive test is very good detecting your positives. So 96% of the time a negative is reliable.
Specificity is the tests ability to detect those who DO NOT have the condition.
It is in relation to negatives. So for this test say we use 100 dogs we know do not have the disease and run the test 94% of them test negative, 6% are false positives (we know they do not have the disease but the test thinks they do).
Specificity (True negative rate) = True negatives / (True negatives + False positives)
So the proportion of the population that does not have the condition as a percentage of animals that test negative. So 94% of the time a positive is reliable.
This test overall is pretty good as in either camp it’ll return you a reliable result.
Hopefully this helps in what is often a really confusing topic.