clinical decision making Flashcards
2 types of probabilities in decision analysis
decision - square
chance- circle
outcome- Triangle
perceived value of outcome to patient
standard gamble - time with illness vs risk of death/cure
time trade off- time in illness vs time in perfect health with treatment
visual analogue- scale 0 death, 100 health
Qaly calculated by TTO
4 years of perfect state vs 10 years of illness - TTO 0.4
Incremental cost effectiveness ratio
ICER–measure of the change in cost with change in a unit of effectiveness (QALY, saved admission)
ICER = (C1 – C2) / (E1 – E2) note that this is the formula for the slope of a line
* Steep positive slope = expensive intervention
* Relatively flat slope = inexpensive intervention
* Negative slope = cost-saving intervention
Decision tree
Linear
Recursive, bidirectional (recursive)–Markov
Models for decision tree
Deterministic. (variable states determined by parameters set in the model
Stochastic-variable states are determined by probability or random distributions
Monte Carlo simulation
Sensitivity = True Positive Rate
P(test + | disease +) = A / (A+C)
* Conditional probability: “given the patient has disease, what is
the likelihood the test will be positive?”
1-sensitivity = False Negative Rate
Very sensitive tests have low false negative rate
* Therefore, sensitive tests are good for ruling out disease (you
can trust a negative result)
* Therefore, sensitive tests are good screening tests
SeNsitivity = rule OUT
“SNOUT”
Positive Predictive Value
P(disease + | test +) = A / (A+B)
* “given a positive test, what is the likelihood
of disease”
Note that PPV depends on likelihood
of disease in population
Negative Predictive Value
- P(disease - | test -) = D/(C+D)
- “given a negative test, what is the
likelihood of no disease”
Note that NPV depends on likelihood
of disease in population
