Diagnostic Accuracy

Sensitivity, Specificity, & Likelihood Ratios

diagnostic test accuracy

• “test is testing what it is designed to test”

• ability to accurately detect presence or absence of a condition

criterion validity

how accurately a test measures the outcome it was designed to measure

ROC

receiver-operating characteristic

diagnostic accuracy

the proportion of all correct tests in a sample

sensitivity

• the test’s ability to obtain a positive test when the target condition is really present

• a test’s ability to correctly identify pts w/ a condition

• the true positive rate

specificity

• the test’s ability to obtain a negative test when the condition is really absent

• a test’s ability to correctly identify those pts w/o the condition

• the true negative rate

sensitive test

based on identifying those w/ the condition

Sn-N-OUT

Sn = sensitive test

N = negative test result

OUT = rule out condition

SP-P-IN

SP = specific test

P = positive test result

IN = rule in condition

Specific test

based on identifying those w/o the condition

pretest probability

estimate of how likely a diagnosis is before we conduct diagnostic testing

differential diagnosis

list of conditions the pt might have

Baye’s theorom

• the initial probability of an event based on prior knowledge of conditions related to that event

• puts a quantitative value & estimates how much we should change our belief & how confident we should be

probability theory

revise probabilities when given additional info

pretest probability

• after subjective questioning

• what we think after initial exam & hx, before any formal testing

posttest probability

• after special tests

• what we think after diagnostic special test

LR

likelihood ratios

likelihood ratios

determines how likely a person has the diagnosis after the test is done

LR+

how many times more likely a positive test will be seen in those w/ the disorder than in those w/o the disorder

LR-

how many time more likely a negative test will be seen in those w/ the disorder

a good test will have a ___ LR+ and a ____ LR-

• high

• low

Fagan’s Nomogram

this tool uses pretest probability & likelihood ratios to determine the posttest probability