3 Epidemiology Module - Is this test useful

Alison Grant 28/8/18

Diagnostic tests, always the right answer?

WHEN YOU ARE DOING A DIAGNOSTIC TEST, YOU REALLY NEED TO THING ABOUT THE DISEASE PREVALENCE IN THE POPULATION YOU’RE TESTING!

How do work out if a test is useful? We compare this new test against our existing “gold standard”. We need to make sure its effective at identifying and removing potential cases

-	-	Disease	Disease	Disease
-	-	Positive	Negative	Total
Test	Positive	A	B	-
Test	Negative	C	D	-
Test	Total	-	-	-

Sens = A / A+C

Spec = D / D+B

PPV = A / A+B

NPV = D / C+D

sensitivity denominator = people who genuinely have disease: amongst people with the disease, how many have a positive test result

spec, amonst people without the disease, how many have a test result saying they dont have disease

ppv, amongst people with a positive test result, how many people actially have disease?

npv, amongst people with a negative test result, how many definitely don’t have disease

the predictive values are going to be affected by the population prevalence. the sens/spec are not

PPV is directly related to disease prevalence NPV is inversely related to disease prevalence