L11 - EXAMINATION DIAGNOSIS ACCURACY Flashcards
What is a cross-sectional study in the context of diagnostic accuracy?
Blind and independent comparison of the test under study versus the diagnostic reference standard in a population of consecutive patients.
What is the index test?
Diagnostic test or procedure under investigation.
What is the reference test?
Referred to as gold standard, the best available method or procedure used to confirm presence or absence of disease.
What does a true positive (TP) indicate in a 2x2 table?
Both index & reference test identify condition.
What does a false positive (FP) indicate in a 2x2 table?
Index test identifies condition, but reference test does not.
What does a false negative (FN) indicate in a 2x2 table?
Index test fails to identify condition that reference test confirms.
What does a true negative (TN) indicate in a 2x2 table?
Both tests confirm absence of condition.
How is sensitivity defined?
Probability of positive test in patients with disease.
What is the formula for sensitivity?
SN = TP / (TP + FN)
How is specificity defined?
Probability of negative test in patients without disease.
What is the formula for specificity?
SP = TN / (TN + FP)
What does a higher value indicate in sensitivity and specificity?
Higher = better.
What do sensitivity and specificity fail to provide?
Exact probability of presence.
What does the acronym SNnOUT stand for?
If test has high sensitivity & is negative, pathology can be ruled out.
What does the acronym SPpIN stand for?
If test has high specificity & is positive, pathology can be ruled in.
Which tests are typically conducted before those with high specificity?
Tests that have high sensitivity.
What is the positive predictive value (PPV)?
Probability that person with positive test has disease.
What is the formula for positive predictive value (PPV)?
PPV = TP / (TP + FP)
What is the negative predictive value (NPV)?
Probability that person with negative test has disease.
What is the formula for negative predictive value (NPV)?
NPV = TN / (TN + FN)
What do predictive values depend on?
Prevalence of pathology.
What is the purpose of likelihood ratios?
Determine probability of having a condition after a positive or negative test result.
What does a high positive likelihood ratio (LR+) indicate?
Better test performance.
What is the formula for positive likelihood ratio (LR+)?
LR+ = SN / (1 – SP)
What does a low negative likelihood ratio (LR-) indicate?
Better test performance.
What is the formula for negative likelihood ratio (LR-)?
LR- = (1 – SN) / SP
What are the best diagnostic test accuracy values?
Likelihood ratios.
What do ROC curves help identify?
Best cut-off for test performance.
What does the area under the curve (AUC) represent?
Overall measure of test accuracy.
What is QUADAS-2?
Gold standard for assessing risk of bias of studies.
What are the components assessed by QUADAS-2?
- Patient selection
- Index test
- Reference standard
- Flow & timing
True or False: Standard meta-analysis is suitable for diagnostic accuracy.
False.
What does bivariate meta-analysis provide?
Pooled estimates for sensitivity & specificity.
