88 - Stats Revision Flashcards
How can sensitivity/specificity be ascertained?
Compare performed test to ‘gold standard’ test.
Positive predictive value
Of all positive tests, what proportion are truly positive
Negative predictive value
Of all negative tests, what proportion are truly negative
Measures of the accuracy of a test
Sensitivity and specificity (inherent to a test and are constant)
What are PPV and NPV dependent on?
Sensitivity and specificity.
Prevalence of outcome.
What can make PPV appear really low, even though specificity can be high?
If a disease is rare enough. The number of false positives can overwhelm the number of true positives, as the number of negative cases is so much higher than positive cases.
What does PPV positively correlate with?
Underlying prevalence of disease
What does the utility of a diagnostic test rely on?
The underlying prevalence of disease (because is related to PPV)
When do you order diagnostic tests?
When there is a strong clinical suspicion of disease.
Because test relies on underlying prevalence of disease.
When are diagnostic tests not used?
To screen for diseases.
Who are targeted for screening tests?
Those with high risk of outcome of interest.
EG: Mammography on women over 50
Why wouldn’t you use a diagnostic test to screen for disease?
Because the underlying prevalence of a disease that is screened for with a diagnostic test is low. This means that the PPV will be low, and the chance of a true positive result is overwhelmed by the chance of a false positive result.
What is biostatistics concerned with, w/r/t samples?
The extent to which the study (sample) results reflect the ‘truth’