Diagnostic Testing Flashcards
what is test sensitivity?
- the probability that the test will correctly identify truly disease individuals
- among disease positive animals, the proportion that are test positive
- highly sensitive tests detect all disease-positive individuals (all negatives are truly negative)
what is test specificity?
- the probability that the test will correctly classify truly non-diseased individuals
- among disease negative individuals, the proportion that are test negative
- highly specific tests detect all disease-negative individuals (all positives are truly positive)
how do you choose between a test that has high sensitivity versus high specificity?
SnOUT: increased sensitivity finds/overshoots all positive cases, so you can be confident in a negative result
-use to ID ALL disease/infected individuals
SpIN: increased specificity finds/overshoots all negative cases, so you can be confident in a positive result
what is positive predictive value?
- the proportion of test-positive animals that are disease
- among test positive animals, the proportion that are disease positive
what is negative predictive value?
- proportion of test-negative animals that are non-disease
- among test negative animals, the proportion that are disease negative
compare sensitivity and specificity to predictive value
Sn and Sp generally considered to be constant!
-eval BEFORE performing the test
-consider cost/logistics to decide which test to perform
predictive values CHANGE
-use to interpret results AFTER the test is performed
-consider whether testing will change course of action
compare and contrast screening versus diagnostic tests
screening test:
-applied to apparently healthy individuals
-for purpose of early diagnosis of disease
-prevalence for entire population/applied to a low prevalence population
diagnostic test:
-applied to diseased individuals
-for purpose of confirming or ruling out a specific diagnosis
-prevalence for similar animals/applied to a high prevalence poopulation
DIFFERENT POPULATIONS
clinical test: anything that predicts health status better than chance
how does prevalence impact predictive values?
the higher the prevalence of disease in a population, the higher the PPV
the lower the prevalence of disease in a population, the higher the NPV
what affects predictive values?
- prevalence of disease in population
-higher prev = higher PPV
-lower prev = lower PPV - a more SPECIFIC test improves the PPV
-fewer false positive results - a more SENSITIVE test improves NPV
-fewer false negatives
in what situation would you choose a highly specific for a highly sensitive test?
highly sensitive:
-screening (ID all diseased or affected to rule OUT disease
-high consequences of disease, use as first test when you DONT want to miss a positive
highly specific:
-to rule IN disease: could be second test after a highly Sn before euthanasia or aggressive treatment
-wen you really want confidence in a positive result
describe testing in series versus testing in parallel
testing in series:
- 2 or more tests performed on a patient sequentially
- retest the positives
-any negative considered negative - improves overall Sp and PPV of testing strategy
-patient must prove it is truly diseased
-best when there are high consequences of truly positive (euth, chemo) - test result may influence next steps, often used to rule in a condition
testing in parallel:
1. 2 or more tests performed on a patient at the same time
- retest the negatives
-any positive considered positive - improves overall Sn and NPV of testing strategy
-patient must prove it is truly healthy
-use when the risk of not treating/isolating a truly positive animal is high (ex. salmonella in horses) - quickly test for multiple outcomes, often used to rule out a condition, good when prevalence is low
