Caution in interpreting covid19 test results

THE EDITOR: According to recent statements from the Minister of Health, more widespread testing for SARS-CoV-2, the virus that causes covid19, will be initiated very soon at various health centres across the country.

While it is evident that ramped-up testing is a precondition to any “exit strategy” from the current “lockdown,” a note of caution is necessary regarding how test results are interpreted.

All tests have some degree of error; there are two dimensions to this, one related to the sensitivity of the test, the other related to its specificity. A sensitivity of 98 per cent means that in 100 samples with the virus, it will be picked up and reported as positive 98 times. A specificity of 90 per cent means that nine times out of ten, a sample without the virus will be correctly reported as a negative test.

What is not intuitive to most people is that if the incidence of the virus in the population (or in the number of samples) is small, say five per cent, then the numbers above will mean that a positive test implies that there is only a 35 per cent chance that the person or sample actually has the virus.

To see how this works, assume that you take 1,000 test samples. A five per cent incidence means that you would expect 50 of these to have the virus, and 950 not to have it.

Of the 50 with the virus, the test will pick up 49 of them and report as positive. So far, so good. But of the 950 without the virus, the test will report 95 of them as positive (the ten per cent false-positive). So you will end up with 144 samples reporting positive, when only 50 of them actually were positive – hence 65 per cent of them will be erroneous.

The sensitivity, specificity and incidence percentages above are illustrative but not unreasonable. The implication is that extreme caution will be needed before taking action related to any person testing positive in such a scenario. For example, if such a person is forcibly quarantined by the State, 65 per cent of the time this will be a completely unwarranted action.

As a footnote, about ten per cent of the samples submitted to Carpha by TT are coming back as positive.

While the actual sensitivity and specificity of the PCR tests being done there are not publicly disclosed, it would be valuable to understand the implications – whether the true incidence in the samples is low (eg there are lots of people with flu-like symptoms who are not infected with covid19; or lots of precautionary tests of non-symptomatic people are being submitted), or there are other issues such as the quality of the swabbing being done etc.

KRISHENDATH MAHARAJ

adjunct lecturer

Arthur Lok Jack

Global School of Business