Testing rate no check on Covid

K Srinath Reddy
President, Public Health Foundation of India

Covid-19 is an RNA virus which enters and spreads through the respiratory system — from nose, sinuses and throat to the airways and lungs. It uses the genetic material of the human body to replicate itself. During the period of active replication, the viral antigen can be detected by an RT-PCR test performed on a sample collected from the throat or nose by a swab, or, rarely, from the deeper airways by bronchial lavage.

Virtually, every test, for any health condition, can yield false negatives (missing real cases) or false positives (diagnosing cases incorrectly). RT-PCR is considered to have very high specificity (hardly any false positives) but relatively lower sensitivity because of variations in the sample collection techniques. Since the antigen will not be detected if we test too late, the test must be performed early in the symptomatic phase. However, even the specificity for detecting an active virus came into question in a recent study from South Korea, where over 200 persons first tested positive for the virus, later tested negative during follow-up, and again puzzlingly tested positive. Was it reinfection, reactivation or was one of positive tests a false positive? After examining all available evidence, scientists concluded that the second positive test was false possibly because it detected ‘dead virus particles’. Similar concerns were raised in China. However, RT-PCR remains the mainstay of viral diagnosis.

Antibody tests measure two immunoglobulin antibodies that are produced by humans who are infected. The IgM antibody makes an early appearance and exit, between eight and 28 days, while the IgG antibody rises after the 10th day, peaks at four weeks and stays on for some months. Presently, it is unclear whether the antibodies confer protection against reinfection and, if so, at what levels, and for how long. Immunological defence may also need cell mediated immunity, which is presently not tested. Recognising these uncertainties, the WHO has declared that the antibody test is not an ‘immunity passport’.

Can the antibody test provide clear proof of recent infection with the Covid virus? If we test too early in the infection, the test will be negative. Also, the antibodies can cross-react with other corona viruses and are not highly specific. When we apply the test on a mass scale, the false positivity rate gets inflated. This is due to ‘conditional probability’ of the Bayes theorem.

According to the theorem, the post-test probability of a diagnostic test being true is a product of both the test result and the pre-test probability, and not indicated by the test result alone. An example will clarify the concept. An exercise ECG test for detecting coronary heart diseases is frequently used in cardiology. If the test is positive in a 60-year-old male diabetic who smokes, has high BP and complains of chest pain on exertion, the post-test probability of a ‘true positive’ is very high. If the test is positive in a 25-year-old woman with anaemia and atypical chest pain, the post-test probability of a ‘false positive’ test is very high.

In Covid-19, an asymptomatic person without history of exposure to a patient has a low pre-test probability. So, the chances of a false positive test result are high. If a person has contact history, the pre-test probability goes up. If there are clinical symptoms of fever and respiratory illness, the probability goes up further. These stepwise additions to the probability of disease give a higher pre-test probability before an antigen or antibody test is performed. A positive result then provides the best estimate of post-test probability to diagnose Covid infection with confidence. Unfortunately, asymptomatic persons in the general population will yield many false positive results.

American statistician Nate Silver, renowned for his accurate predictions of baseball games and US elections, writes, ‘Information becomes knowledge only when placed in context. Without it, we have no way to differentiate the signal from noise and our search for truth will be swamped by false positives.’ The large number of ‘positive’ cases found by antibody tests, among asymptomatic volunteers in the US, become suspect because of this reason. We can still use random sampling of the population of each district to provide a comparative estimate of virus exposure among districts, though we cannot confidently pin the label of past virus infection or immunity on anyone who tests positive.

The chorus of calls for ‘more testing to win this war’ will have to contend with the reality that there is no correlation whatsoever between the number of tests per million population and the number of deaths per million. This holds true, whether we look at all countries or at geographically clustered groups of countries. Many countries which have had low testing rates have had far fewer deaths than countries with much higher testing rates.

Antigen testing can be helpful for directing specific drugs, if such life-saving drugs emerge from ongoing clinical trials on hospitalised patients. Such symptomatic persons are already covered by the Indian testing protocols. The clamour for screening asymptomatic workers returning to offices or factories is misplaced. A negative antigen test today is no guarantee against acquiring infection and transmitting during the pre-symptomatic phase a few days later.

Testing should continue, with judicious choice of test type, person and timing. Beyond testing, there is much to gain from behavioural interventions like physical distancing, masks and hand hygiene. With uncertainty about whether a stranger we pass by on the road or an office colleague is carrying the virus or not, this is the best means of protecting ourselves. In cricket, a good bowler does not depend on one stock ball nor does a good batsman have a single stroke in his repertoire. So also with testing.

Views are personal