Quantcast

Same conclusion drawn from specificty and NPV?

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

Lothric

Full Member
2+ Year Member
Joined
Aug 30, 2016
Messages
234
Reaction score
21

Members don't see this ad.
Hi,

Can the same conclusion be drawn from specificty and negative predictive value?

For example, a high specificty or high negative predictive value --> rules IN disease?


And vice versa for PPV and sensitivity?
 

TheIllusionist

Full Member
7+ Year Member
Joined
Aug 9, 2013
Messages
584
Reaction score
1,162
No because PPV can vary with prevalance. Low prevalence will wreck your PPV

Sensitivity and specificity are resistant to that.
 

Lothric

Full Member
2+ Year Member
Joined
Aug 30, 2016
Messages
234
Reaction score
21
No because PPV can vary with prevalance. Low prevalence will wreck your PPV

Sensitivity and specificity are resistant to that.
Sure, but why does FA say that BNP has good negative predictive value? Couldn't they just have said that the specificity is high for BNP?
 

aldol16

Full Member
5+ Year Member
Joined
Nov 1, 2015
Messages
5,424
Reaction score
4,213
Sure, but why does FA say that BNP has good negative predictive value? Couldn't they just have said that the specificity is high for BNP?

NPV and specific are two entirely different concepts. You should draw a 2x2 table to see the difference. NPV is useful clinically because in clinical settings, you always start with a pre-test probability (which is informed by prevalence). You do the test and the NPV of that test will govern how much the scales tip in favor of or against your diagnosis. Specificity doesn't have that property. Specificity just measures how many people without the disease don't have a specific marker. So say you have a specificity of 100%, i.e. healthy people don't have elevations in marker X. You think your patient has disease X so you do a test for marker X. It comes back negative. How much does the scale move? You don't know. You don't know because it could be true that even though 100% of healthy people don't have the marker, only 1% of the diseased people have it. So doing the test didn't really tell you much.

That's why a high specificity helps to rule in disease. If your test is positive, that means that you likely have the disease because high specificity means that healthy people are not likely to have elevations in that marker.
 
  • Like
Reactions: 1 user

Lawpy

38% Full Member
7+ Year Member
SDN Ambassador
Joined
Jun 17, 2014
Messages
60,428
Reaction score
152,078
NPV and specific are two entirely different concepts. You should draw a 2x2 table to see the difference. NPV is useful clinically because in clinical settings, you always start with a pre-test probability (which is informed by prevalence). You do the test and the NPV of that test will govern how much the scales tip in favor of or against your diagnosis. Specificity doesn't have that property. Specificity just measures how many people without the disease don't have a specific marker. So say you have a specificity of 100%, i.e. healthy people don't have elevations in marker X. You think your patient has disease X so you do a test for marker X. It comes back negative. How much does the scale move? You don't know. You don't know because it could be true that even though 100% of healthy people don't have the marker, only 1% of the diseased people have it. So doing the test didn't really tell you much.

That's why a high specificity helps to rule in disease. If your test is positive, that means that you likely have the disease because high specificity means that healthy people are not likely to have elevations in that marker.

Good stuff :thumbup:
 
  • Like
Reactions: 1 user

YCP Pre-Med

Full Member
2+ Year Member
Joined
Oct 18, 2015
Messages
76
Reaction score
91
I wish I could like this post 50 times. Thank you.


NPV and specific are two entirely different concepts. You should draw a 2x2 table to see the difference. NPV is useful clinically because in clinical settings, you always start with a pre-test probability (which is informed by prevalence). You do the test and the NPV of that test will govern how much the scales tip in favor of or against your diagnosis. Specificity doesn't have that property. Specificity just measures how many people without the disease don't have a specific marker. So say you have a specificity of 100%, i.e. healthy people don't have elevations in marker X. You think your patient has disease X so you do a test for marker X. It comes back negative. How much does the scale move? You don't know. You don't know because it could be true that even though 100% of healthy people don't have the marker, only 1% of the diseased people have it. So doing the test didn't really tell you much.

That's why a high specificity helps to rule in disease. If your test is positive, that means that you likely have the disease because high specificity means that healthy people are not likely to have elevations in that marker.
 
  • Like
Reactions: 1 users
Top