Case Example 1 – Unselected ED Population:
John Doe presents to the ED and is found to have a positive troponin. What are the chances he has an acute MI given no other information? To start, assume the prevalence of AMI among the 131 million patients presenting to the ED annually is 0.3% (430,000). Also assume our troponin assay has a sensitivity of 95% and a specificity of 80%. To start, here is a 2×2 table constructed using the above values.
Since we know the formula for PPV, we can plug in our numbers and we get a PPV of 1.54%.
PPV = TP / (TP+FP)
PPV = 408/26522 = 0.0154 = 1.54%
With a positive predictive value of 1.54%, a strategy of testing all patients who walk into the ED would provide us with little to no useful information!
Case Example 2 – Selected ED Population:
Now suppose John Doe presents to the ED with symptoms suggestive of an MI and an elevated troponin. Now what is our PPV using the same assay? You are told that the prevalence of MI in a population of 22,600 patients with symptoms suggestive of MI is 15.3%. Start with another 2×2 table.
Again, plugging in for PPV yields:
PPV = TP / (TP+FP)
PPV = 3285/7114 = 0.462 = 46.2%
Now that we are only testing patients with suspected MI, a positive result is more meaningful. Here our patient has a full 46% of having an MI given his elevated troponin in the setting of his symptoms.