Evidence Flashcards
Likelihood ratio?
Probability that you’d see some evidence if your hypothesis were true
___________________________
Probability that you’d see that exact same evidence if your hypothesis were false
What is a likelihood ratio?
A number repenting the diagnosis usefulness of a test
LR = 1 is useless
LR > 1 increasing probability (the higher the better)
LR < 1 decreases probability (the lower the better)
LR + vs LR - ?
LR + = the finding was present (not that it necessarily increases probability)
LR - = the finding was absent, not that it necessarily decreases probability
Where can you find LR values?
- diagnostic literature
- calculate them
What to avoid when applying likelihood ratios?
Double-counting evidence:
- do not apply more than one correlated finding
- when choosing between multiple options, pick the one that provides the best evidence
How to calculate likelihood ratios?
LR = sensitivity / (1 - specificity)
LR - (finding is absent):
LR - = 1 - sensitivity / specificity
What is sensitivity & specificity?
Sensitivity: in patients who have the disease, the probability the test will be positive
Specificity: in patients who don’t have the disease, the probability that your test will be negative
What is the least reliable method (but better than guessing) for likelihood ratios?
imagine 2 rooms:
- room 1 has 100 people who have the condition
- room 2 has 100 people who do not have the condition
- if you asked everyone in either room how many people had the finding you were considering, how many people in each room will raise there hand?
How is applying evidence intuitive and unintuitive?
Intuitive: the more certain you initially are, the harder it should be to change your mind
Unintuitive: The more uncertain you, the less evidence you need to change your mind
What is a rough calculation of low pre test probabilities and moderate LRs?
post test probability = low pretest probability x moderate LR
ex. pretest = 2%, LR =5
post test prob = (2)(5) = 10
What is positive predictive value (PPV) & a negative predicative value (NPV)
PPV: Probability that disease is present given that a test was positive
NPV: Probability that a disease is absent given the test was negative
What is “pathognomonic findings” ?
Findings that, if present, strongly increases the probability of a condition
-High LR + value
“pathos” = disease
“Gnomon” = indicator
What is “Sine qua non findings”?
“without which it could no be”
Findings which, if absent, strongly decreases the probability of a condition
-Low LR - (close to 0)
Evidence steps in practice (5)?
- recognize the need for more evidence
- choose a test (question, physical exam, lab test)
- Perform the test correctly
- Interpret the results correctly
- Repeat steps 1-4 until you cross a threshold
Sources of errors in practice with evidence (4)?
- Fail to recognize the need for more evidence (premature closure)
- Poorly chosen or missed test (question, physical exam, lab)
- Incorrectly performed test (esp physical exam)
- Incorrect interpretation (lack of knowledge, bias)