First, what is the point of the paper?

Diagnostic tests are frequently used in medical care. They often have no direct impact on health but it is hoped that the information from them will improve the clinical management of a patient.

In health economics cost-effectiveness analysis and cost-utility analysis are the main ways we decide whether money should be spent on a health technology.

Evidence on diagnostics is often limited, so we need to use decision modelling to extrapolate outcomes.

It's very common to estimate the lifetime costs and quality-adjusted life years (QALYs) resulting from implementing a diagnostic test by using a decision tree and estimating those quantities conditional on whether the patient truly has the disease and whether the test correctly diagnoses.

This gives us terms True Positive (TP), False Positive (FP), True Negative (TN) and False Negative (FN).

There is a lot of literature on evaluating the accuracy (clinical validity) of diagnostic tests...

...but the methods for economic evaluation of diagnostic tests have not really changed dramatically for decades.

One of the first common mistakes I draw attention to in the paper is how to handle using diagnostic tests in combination.

Many analysts (myself included once upon a time) make the mistake of assuming the sensitivity and specificity of Test 2 will be unaffected by whether Test 1 has been used to filter the population in some way.