Combining two different test results NPV and PPV (in medical diagnostic testing)

#1
Thanks for your time in answering my question!

I have 2 different lab tests, both of which are tested on urine to determine if a patient has an infection. I would would like to combine the two tests to help the provider have a higher likelihood of treating the patient appropriately. Here are the statistical results of the two tests:


Test #1: Urine Dipstick
Test #2: Catalase Test

1) +Leukocyte and +Nitrite: 79% chance the patient has an infection (PPV)
2) -Leukocyte and -Nitrite: 85% chance the patient doesn't have an infection (NPV)

3) +Leukocyte and -Nitrite: 67% chance the patient has an infection (PPV)
4) -Leukocyte and +Nitrite: 74% chance the patient has an infection (PPV)

While the catalase test has a 33% PPV and 97% NPV.

Here is how I envision us using it:
Scenario #1: Treat the patient without the use of the catalase test, because it is really only good for "ruling out the disease and not for ruling it in"
Scenario #2: Do not treat the patient, because they already have a high likelihood of not having an infection

Scenario #3-4: Use the catalase test to determine when not to treat the patient (if the catalase test is negative).
--With these scenarios, if the catalase test is negative, then we would not treat, but if it was positive, would this change the chance that they have the disease? Essentially can I combine a PPV of 67% and 33% or 74% and 33% to come up with a different %?

Thank you so much for your help!
-Jeremy
 

Attachments

Last edited by a moderator:

hlsmith

Not a robit
#2
Are you missing signs on Nitrite for "1) and 2)"?

You need to be more explicit on what is the dipstick and catalyst test results above. Also, keep in mind the two test combined may not have an additive effect if they are both getting the same cases right. Also, there is the scenario where you take negative or positives of one test and then apply second test to those groups and see how they function given the stratification.
 
Last edited:
#4
What do I need to be more explicit about in the test results?

It seems as though your answer would be that I can’t independently combine these results and I would need to do a study myself and determine the new statistical values from that.

Correct?