I'm a doctor. All tests have something called a sensitivity, ie. the % of true positives identified from all positives and specificity, ie. the % of true negatives identified from all negatives.
Let's say the sensitivity is 95% and the sensitivity is 99%, we'll first run 100000 tests in a population where 1/100 has the condition and one then one where 1/1000 has the condition
1000 have condition 99000 not, we find 950 TruPos, 50 FalseNeg and 98010 TruNeg and 990 FalseNeg
950/(990+950)=49% of Pos are TruPos and 98010/(98010+50)=99% of Neg are TruNeg
100 have condition and we find 95 TruPos and 5 FalseNeg and 98901 TruNeg and 999 FalsePos
95/(999+95)=8.6% of positives are TruPos and 98901/(98901+5)=99.99% of negatives are TruNeg
So the the positive predictive value ( TruePos/AllPos) is greatly dependent on the prevalence or chance before testing of having a certain condition.
So just testing everyone with no or minor symptoms leads to many false positives
I'm a doctor. All tests have something called a sensitivity, ie. the % of true positives identified from all positives and specificity, ie. the % of true negatives identified from all negatives.
Let's say the sensitivity is 95% and the sensitivity is 99%, we'll first run 100000 tests in a population where 1/100 has the condition and one then one where 1/1000 has the condition
1000 have condition 99000 not, we find 950 TruPos, 50 FalseNeg and 98010 TruNeg and 990 FalseNeg
950/(990+950)=49% of Pos are TruPos and 98010/(98010+50)=99% of Neg are TruNeg
100 have condition and we find 95 TruPos and 5 FalseNeg and 98901 TruNeg and 999 FalsePos
95/(999+95)=8.6% of positives are TruPos and 98901/(98901+5)=99.99% of negatives are TruNeg
So the the positive predictive value ( TruePos/AllPos) is greatly dependent on the prevalence or chance before testing of having a certain condition.
So just testing everyone with no or minor symptoms leads to many false positives