Oxford Reviews Many COVID-19 Antibody Tests, And Finds None of Them are Accurate

Getting your Trinity Audio player ready...

COVID-19 testing has mostly been based on reverse transcription polymerase chain reaction (RT-PCR), in which the viral genetic material in an upper respiratory tract sample is amplified to detectable levels. Antibody tests, on the other hand, detect the presence of immunoglobulin produced by B cells to target Sars-CoV-2. Unlike RT-PCR, antibody tests can detect if a person has previously been infected, and do not require specialist training to perform.

Why did the UK need 100,000 tests a day?. (2020). Retrieved 6 May 2020, from https://www.bbc.com/news/health-51943612

Many hoped that antibody test kits would provide much needed improved access to testing, and enable those tested to confirm whether they have previously had COVID-19, potentially allowing them to return to work and a semblance of normal life.

Unfortunately, as these testing kits have become available, their reliability has left something to be desired. A paper authored by the National COVID Testing Scientific Advisory Panel in Oxford investigated how well 9 devices considered by the UK government compare with RT-PCR:

Of the RT-PCR-confirmed cases, the proportion that were identified as positive by the antibody testing devices ranged from 55% to 70% (the remainder being false negatives).
Of those confirmed negative by RT-PCR, the proportion that were identified as negative by the antibody testing devices ranged from 95% to 100% (the remainder being false positives).

In other words, a person who tests positive can be fairly confident that they are or have been infected, but a person who tests negative still cannot rule out having had the disease.

The researchers’ conclusion: that the currently available devices are not suitable for testing on an individual patient basis. For diagnosis of symptomatic patients, a test would need to identify at least 90% of true positives to be acceptable. On the other hand, were tests to be used as a basis to allow previous cases to cease social distancing, anyone falsely identified as positive would be put at risk. Consequently, the rate of identification of negative cases needs to be at least 98%.

The good news to come from the study was that the lab based antibody test proved significantly more accurate than any of the point-of-care tests, and could work as a starting point to understand what proportion of the population have been infected.