Comparing sensitivity and specificity of two diagnostic tests

[#1]

Hi,

I have two test done on the same indeviduals, hence dependent, which I calculated the sensitivity and specifity related to the "golden standard". I would like to compare them to test if the difference is significant. I believe the test is mcnamar test. How to perform this in JMP?

TIA

[#2]

The test you want is used with categorical variables, not continuous (measurement) variables. Is that correct?

See Help > Statistics Index:

So make sure that both variables use the nominal or ordinal modeling type. Select Analyze > Fit Y by X. Select the test data column and click Y. Select the standard data column and click X. Click OK. Click the red triangle at the top and select the test.

[#3]

Thank you very much for the reply.

I calculated sensitivity and specificity for each of two diagnostic tests, separately, against the same golden standard.

I'm still unclear on how to compare the two diagnostic tests and show that one is significantly better than the other, and that the better results obtained in the sensitivity and specificity are not random.

So, just to make sure I got this right, you are suggesting to test the difference between the two diagnostic tests (regardless of the golden standard) and if the McNemar's Test results that the tests are significantly different this proofs that the diagnostic test with better sensitivity and specificity is the better choice.

Correct?

[#4]

No, I did not suggest a method for testing the difference between two diagnostic tests to a gold standard test. I answered your question about how to perform McNemar's test using JMP.

I believe that there are standards, protocols, and guidance documents for method comparison, for example, by FDA, Clinical Laboratory Standards Institute, and the United States Pharmacopeia. You should refer to them first..