Probability in Diagnostic Tests

Question 1

Define sensitivity.

Answer 1

test is POSITIVE in those WITH disease [TP]/ [all people w/ disease] most negative results are true (test more likely to give FP) SNOUT: Sensitive test with Negative result rules OUT disease

Question 2

Define specificity.

Answer 2

test is NEGATIVE in those WITHOUT disease [TN]/ [all people w/o disease] most positive results are true (test more likely to give FN) SNIP: Specific test with Positive results rules IN disease

Question 3

Descrive a Receiver-Operator Characteristic (ROC) curve.

Answer 3

1) sensitivity y-axis, false positive rate x-axis 2) good test crowds upper left corner (little loss of specificity until sensitivity is very high)

Question 4

Define positive predictive value.

Answer 4

Given the test was positive, what’s the probability I have the disease? TP / [TP + FP]

Question 5

Define negative predictive value.

Answer 5

Given the negative test, what’s the probability I don’t have disease? TN / [TN + FN]

Question 6

Relationship between prevalence and predictive value.

Answer 6

As prevalence increases, + PV increases, -PV decreases

Question 7

What is a reference interval?

Answer 7

range of values intended to include 95% ofpersons assumed to be disease free