Sensitivity & Specificity Analysis

[gemmaspencer]

Hello,

I have carried out sensitivity and specificity analysis on two neuropsychological tests to determine their utility in detecting impairment (mild cognitive impairment/Alzheimer's disease) relative to controls. The tests show opposite profiles: one has a sensitivity of 83.3% and specificity of 57.1% and the other has a sensitivity of 61.1% and a specificity of 91.5%.

Although the first test has quite a low specificity, am I right to report that this test is more balanced? (difference of 26.2% compared to 30.4% for the second test)

Thank you.

 

[hlsmith]

But do you want balanced? Is it more important to have more false positives or false negatives or a balance? Many people initially think that balance is good, but if I am screening for MCI would I want to have a test that says more may have MCI or have it under-reported MCI? Its clinical function should guide one's preference. Also, it wouldn't hurt for you to slap confidence intervals on these estimates to see how much they may vary with repeated sampling from the target population.

Thanks.