2x2 Stuff
Top left/top left+top right
How do you calculate a positive predictive value?
Top left and top right/total population
How do you calculate apparent prevalence from a 2x2?
Bottom right/bottom right+bottom left
How do you calculate the negative predictive value?
Top left and bottom left/total
How do you calculate true prevalence from a 2x2?
True - good is based on sensitivity and specificity, useful is based on predictive values
True or false - how good a test is does not change, but how useful a test is does change.
True
True or false - if sensitivity is 100%, all negative tests are truly disease negative.
False - all positive tests must be diseased
True or false - when specificity is 100%, there will still be some false positives.
Screening new herd animals, dealing with infectious/zoonotic disease, dealing with a fast progressing disease, or animal trade
What are 4 scenarios when high test sensitivity would be beneficial?
Sensitivity, specificity, and expected prevalence/pre-test probability
What are the three things that post-predictive values depend on?
Negative predictive value and 1-positive predictive value
What are the two measures of the probability that an animal is disease free following a test?
Positive predictive value or 1-negative predictive value
What are the two measures of the probability that an animal is diseased following a test?
Diseases with invasive or aggressive treatments, diseases where euthanasia is considered upon a positive result, or reportable diseases are involved
What are three scenarios where high test specificity would be beneficial?
Probability of true disease status from known test results
What do post-test probabilities tell us?
When sensitivity is 100%, a negative test rules out disease
What does SnNOUT represent?
When specificity is 100%, a positive test will rule a disease in
What does SpPIN represent?
Proportion of total population that tests positive
What is apparent prevalence?
Proportion of truly diseased animals that test positive
What is test sensitivity referring to?
Proportion of truly non-diseased animals that test negative
What is test specificity referring to?
Proportion of test-negatives that are truly disease free
What is the negative predictive value?
Proportion of test-positives that are infected or diseased
What is the positive predictive value?
Probability of disease after we have the result
What is the post-test probability of a disease?
Probability that the animal has the disease prior to the next test
What is the pre-test probability of disease?
Fewer false negatives
What is the result of a high sensitivity relating to false positives/negatives?
Few false positives
What is the result of a high specificity test relating to false negatives/positives?
Proportion of total population that is truly diseased
What is true prevalence?
Predictive values that approximate post-test disease probabilities
What is used to evaluate the usefulness of a test?
Apparent prevalence, specificity, and senstitivity
What three things need to be known in order to calculate true prevalence from apparent prevalence?
Interpreting and evaluating test results or determining if a test is useful
When are predictive values used?
To evaluate the goodness of a test
When are specificity and sensitivity used?
When pre-test probability is high/low and the result is the opposite
When are tests most confusing?
When there is the greatest amount of pre-test uncertainty (near .5)
When are tests the most useful?
When false positives equal false negatives
When is the only time that true prevalence is equal to apparent prevalence?
Positive predictive value
Which post-test probability is affected by false positives?
Negative predictive value
Which post-test probability is impacted by false negatives?
Positive predictive and negative predictive values
Which probabilities do change for tests? a. specificity b. sensitivity c. positive predictive value d. negative predictive value
Specificity and sensitivity
Which probabilities do not change for tests? a. specificity b. sensitivity c. positive predictive value d. negative predictive value
Not all test results will tell us the same thing
Why do predictive values vary while specificity and senstivity do not?
False positives and false negatives exist
Why is apparent prevalence not a good estimate of true prevalence?