Choosing Tests

The limits of my language mean the limits of my world.
-Ludwig Wittgenstein, 1918.

Throughout the pandemic of 2020, the vocabulary of laboratory medicine has been used indiscriminately and imprecisely, resulting in muddled communication and poor decisions.  Today we will discuss basic terms used to evaluate laboratory tests so that you can address these issues confidently and intelligently.

Reduced to its simplest possible terms, tests are measured by how good positive and negative results correlate to the presence or absence of the condition being tested.  Tests are either positive or negative, and patients either have the condition or not.  True positive results match a positive patient condition; false positive results are positive even though the patient does not have the condition.  Similarly, true negative results match a negative patient condition, and false negative results are negative even when the patient has the condition.  We can visualize these terms with a simple matrix:

In the case of COVID-19, laboratories test for the presence of SARS-CoV-2 in patients who may or may not be infected by the virus.  Using this example, we can rewrite the matrix:

Adding the number of true positives and false negatives, you get the number of infected patients.  Similarly, the number of uninfected patients is the false positives plus true negatives.  Sensitivity, the measure of the test’s ability to detect infection, is the number of true positives divided by the number of infected patients: TP/(TP+FN).  Sensitivity is low when there are many false negatives, but it gets close to 100% when false negatives are rare.  A highly sensitive test system minimizes false negatives; when the test result is negative, you can believe it is true.

Specificity is the measure of the test’s ability to detect nothing but infection, is the number of true negatives divided by the number of uninfected patients TN/(TN+FP).  Specificity is low when there are many false positives, but it gets close to 100% when false positives are rare.  A highly specific test system minimizes false positives; when the test result is positive, you can believe it is true.  Obviously, good test systems aspire to be both highly sensitive and specific, but like so many things in life, having both at once is often impossible.  Trade-offs are inevitable.  So how do we decide which is more important?  We will discuss that next time.