Variations in Test Accuracy

Test Accuracy

The accuracy of any diagnostic test is based on two metrics: Sensitivity and Specificity. Sensitivity denotes the ability of a test to correctly identify a true positive case among the population of positive subjects while Specificity denotes the ability of a test to identify a true negative case among the population of negative subjects.

Let’s use an example to illustrate what these two numbers mean:

XYZ Test Corp’s COVID-19 Antigen Test (which tests to identify those with an active COVID-19 infection) has a sensitivity of 95% and a specificity of 90%. Within a sample group of 1,100 people to be tested, 1,000 are known to be negative (uninfected) and 100 are known to be positive (infected). The test’s 95% sensitivity indicates that among the positive (infected) population of 100 people, the test would correctly identify 95% of them or 95 people as being positive (TRUE POSITIVE). Conversely, this test would miss 5 people that actually have the disease, incorrectly identifying them as negative (FALSE NEGATIVE). Similarly, of the 1,000 people that are known to be negative, the test’s 90% specificity indicates that the test would correctly identify 90% or 900 people as being negative (TRUE NEGATIVE). The remaining 100 people would be incorrectly identified as positive (FALSE POSITIVE).

Now, since failing to correctly identify (and as a result, failing to quickly quarantine) an infected and presumably contagious person can potentially have a very harmful effect on the community, false negatives are of particular concern for this type of test and should be guarded against. In contrast, while the false positive scenario is an issue, it is relatively innocuous compared to the false negative case and will likely just mean these subjects will have to undergo additional testing in order to correct the false positive identification. However, the impact to the community is minimal.

So what are the implications for a COVID-19 testing program? First of all, we should recognize that no test is perfect. That is certainly true of EVERY SINGLE COVID-19 test that is currently on the market today or that will come to market in the future regardless of technology or cost. Tests that have gained the FDA Emergency Use Authorization designation have sensitivity and specificity values that range from as low as 75% to near 100%. Therefore, one of the key elements of a comprehensive testing strategy is to recognize the limitations of the tests being deployed and to mitigate the potential impact of inaccurate results. That could mean potentially repeating tests for groups that have the largest potential negative impact of an incorrect result (false negatives in the example above), conducting tests more frequently to mitigate against variability, or conducting additional validation testing using a different testing methodology.

A second issue to keep in mind is that actual real-world accuracy results can vary significantly from the accuracy values specified by the manufacturer due to variability in test execution. Two factors at work in such variability may be:

1. the manual nature of some tests, and;
2. variability in specimen collection technique.

For the former, tests that require multiple manual processing steps in a lab are potentially susceptible to processing errors which would impact the accuracy. An example of the latter is variability in swabbing technique for certain COVID diagnostic tests that require a nasal swab. If the test specimen is not collected at the right location in the nasal passage, the test could give an incorrect result. Given this concern, a testing program with simple repeatable specimen collection requirements and a higher degree of automated test processing can improve the chances of an accurate and repeatable test result.

When viewed from this perspective, Inspire Diagnostics’ point-of-care tests lend themselves to more accurate and repeatable results due to the simplicity of specimen collection and the fully automated nature of the test.