Flashcards in  Diagnostic Research Design Deck (34)
Do you need a diagnosis if pretest probability is 100%?
No, it is not needed
Define: Treatment Threshold
Probability above which diagnosis is sufficiently likely to warrant treatment
What is the effect of a greater adverse effect of treatment on the Treatment Threshold?
What is the effect if the next test the patient has to be subjected to is more invasive than the current test to be performed on the Treatment Threshold?
Prevents more people from being subjected to the next more invasive test
Define: Test/Diagnostic Threshhold
Probability below which a diagnosis warrants no further consideration
What is the effect of a serious missed diagnosis on the Test/Diagnostic Threshold?
Missing diagnosis would be fatal, lowering it would ensure more people to take further tests to confirm
What is the effect if the next test to be done is riskier than the current test on the Test/Diagnostic Threshold?
This is done to ensure less people take the riskier next test if they don't reach the high enough diagnosis warrant in the first place
Probability of Person w/ the Disease Testing Positive / Probability of a Person w/o the Disease Testing Negative
LR > 1
Increased disease probability
LR < 1
Decreased disease probability
Diagnostic tests with LRs close to 1 have practical significance
Goal is create a significant increase in the post test probability as compared to pre test probability
LRs > 10 and < 0.1
Large and often conclusive
LRs 5-10 and 0.1-0.2
LRs 1-2 and 0.5-1
Small and Rarely Important
LRs 5-10 and 0.2-0.5
Small but Sometimes Important
Definition: Gold Standard
Accepted definition of disease and can dictate presence or absence of it
Systemic Lupus Erythematosus
CT Scan + 2D Echo
Probability of a positive test among diseased
Sensitive tests are more important for generalists or specialists?
To not miss people with the disease
Probability of a test being negative among non-diseased people
A positive test is more helpful for Specific or Sensitive tests?
Specific tests are more important for generalists or specialists?
To help confirm diagnosis
Proportion of people correctly diagnosed by tests as either having the disease or not having it
(+) Predictive Value
(-) Predictive Value
( + ): Probability that patient has the disease given a + test result
( - ): Probability that patient does not have the disease given a - test result
Define: Kappa Statistics
Measures agreement beyond chance
K = 1 or 100%
K > 0.80
0.60 < K <= 0.80
0.40 < K <= 0.60
0 < K <= 0.40