Sensitivity and Specificity Flashcards
Accuracy
Measure of how close a test result is to the truth
- proportion of all correct tests (out of total number of tests run)
What are 3 things that affect accuracy?
- prevalence
- sensitivity
- specificity
2 x 2 table: positives and negatives
Positive Negative
+ a (TP) b (FP)
- c (FN) d (TN)
Accuracy formula
= a + d / n
______ assumes the cost of mistakes in either direction are equal
Accuracy
Precision
How close the measured values (test results) are to each other
- may not be accurate or correct, but they are close to each other
What do accuracy and precision have in common?
- describes placement of data
- summarizes data
- determine exactness
- sometimes represents the same thing
- helps determine successfulness of results
- used in physics, engineering, stats, etc
Accuracy - summary
- how close data or results are to the goal
- measure of quantity to reality
- takes into account the accepted value
- bad results would be far from the goal
- determined with a single measurement
- does not speak about quality
Precision - summary
- how close data or results are to each other
- measures how well measurements can be reproduced
- doesn’t take into account accepted value
- bad results would be scattered
- needs several measurements to be determined
- speak about quality
Sensitivity
Ability of a test to correctly identify those that have the disease
- of all animals that have the disease, what proportion test positive
Highly sensitive tests will have a low number of _______
False negatives
Specificity
Ability of a test to correctly identify those that do not have the disease
- of all the animals that do not have the disease, what proportion test negative
Highly specific tests will have a low number of ______
False positives
Concerned with the population of animals that have the disease
Sensitivity
- looking at the left side of the 2 x 2 table (a and c, TP/FN)
= a/a+c
What are reasons to get a false negative?
- natural or induced tolerance
- improper timing
- improper selection of test
- non-specific inhibitors
- antibiotic induced Ig suppression
- incomplete or blocking antibody
- analytically insensitive tests
As sensitivity decreases, the likelihood of getting _________ increases
False negatives
How to interpret sensitivity?
TP = 93
FN = 7
- out of every 100 cats, the test will correctly identify 93 cats that have the disease, and will falsely identify 7 cats as negative that actually have the disease
Concerned with the population of animals that do not have the disease
Specificity
- looking at the right side of the 2 x 2 table (b and d, FP/TN)
= d/b+d
Why might a result be a false positive?
- cross reaction
- non specific inhibitors
- non specific agglutinins
- contamination
- vaccination cross reaction
As specificity goes down, the likelihood of _______
False positives increases
What is expressed as a proportion?
Epidemiologic sensitivity
- reliance on testing to identify ill animals
How is analytical sensitivity different from epidemiologic sensitivity?
Analytical sensitivity expresses the detection limits of a test
- ability to detect very low levels of a target
Gold standard
Required to calculate sensitivity, need samples from animals that we know the true disease status of
- are properly classified as either positive or negative
- run some developed gold standard test and compare results to determine the other tests’ accuracy
Epidemiologic specificity
Ability of a test to correctly detect (classify) non-diseased animals
Analytical specificity
Ability to measure the correct substance
- not measuring particles or molecules other than the target
Proportion of all tests, both positive and negative, that are correct
Accuracy
Is accuracy a good way to assess a test?
No, it is easily manipulated
- is used in some marketing advertisements
Sensitivity of a test may be different if comparing _______
Populations in early vs late stages of disease
- false negatives may be due to very early infections