Likelihood Ratios vs Predictive Values: USMLE* Biostatistics-5 

no, it would tell you that it's 2x more likely to have the disease. For a test to be useful, a negative likely hood ratio (LR) has to be less than 1. Here are some examples: A negative LR of 1 = a person who tests negative has the same chance of having the disease as someone who rests positive. A negative LR of 0.5 means that a person who tests negative is 1/2 as likely to have the disease or 2x less likely (1/0.5)

itekish Thanks so much.

It's been two years since you posted this... Super not timely, but... I think what the question 1 is asking for is what the positive predictive value is. (What is the probability of the patient having the disease when tested positive?) All the values are given in percentage, so for the ease of argument, I assume the population is 100,000 as usual. The disease prevalence is given as 0.31. So 31,000 people actually have the disease. (sum of left vertical column in our usual 2x2 table.) Those who are free of the disease are 69,000. So the actual number of people who test positive and actually have the disease = 31,000 x 0.75 (given sensitivity) = 23,250 = true positives. The number of people who do not have the disease but test positive (false positive) = 69,000 x (1-specificity) = 69,000 x 0.19 = 13110. PPV = TP/(TP + FP) = 23,250/(23,250+13110) = 0.639. So I got 64%.