USMLE Forums banner
1 - 3 of 3 Posts

·
Registered
Joined
·
4,342 Posts
Discussion Starter · #1 ·
Hey guys,

I found a question about positive likelihood ratio and I thought it's the same as positive predictive value but then I found out it's not.

Can anybody explain for me the difference and how to calculate the likelihood ratio briefly?

Thanks in advance,
 

·
Registered
Joined
·
3,316 Posts
The positive predictive value is not the same as positive likelihood ratio though they are closely related.
The positive predictive value is the probability the disease is actually present when you have a positive result and it's calculated (a/a+c).
The positive likelihood ratio (LR+) reflects the degree to which a positive test result increases the likelihood that the disease in question is present.
It's calculated by divding true positive rates over false positive rates or in other words sensitivity divided by 1-specificity.
LR- on the other hand is the rate of true negatives divided by false negatives (specificity divided by 1-sensitivity).
Nowadays LR+ is being used to derive the positive predictive value from it in a somewhat complex calculation steps. Negative predicitve value is derived from LR-.
LR+/LR- are more seen in European journals than in US medical journals and it's extremely unlikely to be seen in USMLE exam and you should not worry about it. However, you may be asked to calculate the positive predictive value.
 

·
Registered
Joined
·
133 Posts
The positive predictive value is not the same as positive likelihood ratio though they are closely related.
The positive predictive value is the probability the disease is actually present when you have a positive result and it's calculated (a/a+c).
The positive likelihood ratio (LR+) reflects the degree to which a positive test result increases the likelihood that the disease in question is present.
It's calculated by divding true positive rates over false positive rates or in other words sensitivity divided by 1-specificity.
LR- on the other hand is the rate of true negatives divided by false negatives (specificity divided by 1-sensitivity).
Nowadays LR+ is being used to derive the positive predictive value from it in a somewhat complex calculation steps. Negative predicitve value is derived from LR-.
LR+/LR- are more seen in European journals than in US medical journals and it's extremely unlikely to be seen in USMLE exam and you should not worry about it. However, you may be asked to calculate the positive predictive value.
Hey Sabio, I think the formula you quoted for [LR-] has an error,

[LR-] = (1-Sensitivity)/Specificity​
 
1 - 3 of 3 Posts
Top