Evaluation of Diagnostic Tests

Question 1

tests

Answer 1

anything used to distinguish health from disease

Question 2

“gray area” of test results

Answer 2

overlap in titer levels in healthy vs disease animals creates a “gray area” for how to determine test cutoff values

healthy animal: high titers can be caused by vaccination, maternal Ab, and field exposure

diseased animals: low titers can be caused by early stage of infection, immunosuppression, and individual variation

Question 3

single test cutoffs

Answer 3

classifies a test result as positive or negative

Question 4

two test cutoffs

Answer 4

classifies test results as positive, negative, or suspect/indeterminate (gray area values)

Question 5

sensitivity

Answer 5

proportion of diseased animals that test positive

high sensitivity tests: catch ALL disease positive animals
- high rate of false positives
- low rate of false negatives

have confidence in the NEGATIVE result

sensitivity = TP / TP+FN

Question 6

specificity

Answer 6

proportion of healthy animals that test negative

high specificity tests: catch ALL disease negative animals
- high rate of false negatives
- low rate of false positives

have confidence in the POSITIVE result

specificity = TN / TN+FP

Question 7

predictive properties

Answer 7

used when we do not know the disease status of the animal but we do know the test result
- can determine disease status

Question 8

what are predictive properties affected by

Answer 8

1. prevalence
2. diagnostic test properties

Question 9

how does prevalence affect PPV and NPV

Answer 9

high prevalence = increase PPV, decrease NPV

low prevalence = decrease PPV, increase NPV

Question 10

how do diagnostic test properties affect predictive properties

Answer 10

high sensitivity = low PPV, high NPV

high specificity = high PPV, low NPV

Question 11

positive predictive values

Answer 11

proportion of test positive animals that are infected

most predicted by SPECIFICITY

PPV = TP / TP+FP

Question 12

negative predictive values

Answer 12

proportion of test negative animals that are healthy

most predicted by SENSITIVITY

NPV = TN / TN+FN

Question 13

prevalence

Answer 13

total # disease / total population

P = (TP + FN) / (TP+FN+TN+FP)

influences how we interpret gray area titers

Question 14

what test is best for screening/disease surveillance

Answer 14

high sensitivity tests

rule out tests -
will capture ALL disease positives and can be confident that the test negative animals are truly negative

has a LOW false negative rate - have confidence in the negative result

Question 15

what test is best for confirmatory testing

Answer 15

high specificity tests

rule in tests -
will capture ALL disease negatives and can be confident that the test positive animals are truly positive

has a LOW false positive rate - have confidence in the positive result

Question 16

if a test has a low sensitivity, how can it be used in a herd setting

Answer 16

test many animals - more likely to find a true positive animal within a herd if there is a larger sample number

Question 17

parallel/simultaneous testing

Answer 17

requiring two different diagnostic tests to confirm/rule out diagnosis

ex. Johne’s disease
- animal is considered positive if tests positive on ELISA, PCR or both
- animal is considered negative if tests negative on ELISA and PCR (both)

net GAIN in sensitivity
net LOSS in specificity

Question 18

series/sequential testing

Answer 18

doing additional testing not hose that test positive on initial screening test

ex. EIA
- animals that test positive on ELISA are then tested on AGID
- AGID has higher specificity –> confirmatory test

net LOSS in sensitivity
net GAIN in specificity