Diagnosis Flashcards
Pre-test probability
The probability of the target condition being present before the results of a diagnostic test are known. PREVALENCE
More extreme PTP require higher liklihood ratios to change them significantly
Liklihood ratio
The extent to which a new piece of information changes probabilities.
A result of the sensitivity and specificity of a test.
Closer to 1= less change
closer to 0= greater decrease in probability
closer to infinity=greater increase in probability
Post-test probability
Probability of the target condition being present after the results of a diagnostic test are known
What makes a good test?
It changes our pre-test probability enough to alter decision making. This is evaluated by specificity and sensitivity
Sensitivity
How much of the time a test is positive in those who have the disease= TP/(TP+FN)
Specificity
How much of the time a test is negative in those who do not have the disease= TN/(TN+FP)
Positive Liklihood ratio
How much PTP changes given a positive test. = Sens/(1-spec)
Closer to infinity is more impactful. Greater than 10 is LARGE change
Negative Liklihood ratio
How much PTP changes given a negative test. = (1-sens)/spec
Closer to 0 is more impactful.
When can we add tests together?
When they are independent of each other. When “adding” tests together you multiply their liklihood ratios
Verification bias
Type of measurement bias in which the results of a diagnostic test influence whether the gold standard procedure is used to verify the test result
What happens to NPV/PPV/LR when we increase the prevalence of a disease?
- This will increase the PPV
- Decrease the NPV
- LR is unchanged
What are multilevel likelihood ratios?
They are constructed from big chunks of data, NOT from STRATIFIED. This makes them LESS accurate. Can calculate using 2x2 table
