Why India needs random Covid tests

Covid-testing strategies have been focused on individuals with “high-probability” of infections and, obviously, only such high probability cases have been identified so far. There is little chance of identifying somebody who is not “high-risk” with this targeted strategy.

Yet, all the data we have thus far garnered about the pandemic indicates that it throws up massive numbers of asymptomatic cases. These cases will not be picked up in India without random sampling across the general population. Given there are over 23,000 known positive cases already, community transmission has surely happened. But we have no idea of the true infection rate.

Some news reports out of the UK and the US suggest the death toll from the pandemic is far higher than the official statistics, even in nations with good data-reporting systems. An American newspaper analysed mortality data from 11 countries (mostly from the First World) and compared reported deaths in the last month to the historical averages. Their conclusion:  At least 36,000 excess deaths have occurred across the US, France, Italy, Spain and Turkey, compared to the historical mortality rate.

A British newspaper did a similar analysis for the UK and concluded that the pandemic  claimed over twice the official number of victims. Even China, which is not known for reliability of data, has officially doubled the number of victims in Wuhan.

There is no easy way to do similar analysis for India. Mortality data is not easily available, disaggregated into causes and locations. What we do know is that only about 22 per cent of registered deaths are medically certified, and only about 77 per cent of deaths are officially registered, according to the Registrar General of India. That is, roughly 16 per cent of all deaths are medically certified. Even in certified cases, the causes may be obscured by certification that just says death from “heart failure” without listing underlying causes.

India has 750-odd districts and they average 37-38 deaths per day, or about 28,000 deaths daily across India. In the last month or so, deaths from road and rail accidents, which claim close to 200,000 victims annually would have drastically reduced. So will deaths from crime. Doing an analysis that takes these numbers into account and adjusts for these omissions will be daunting, even if somebody diligently collects data from death registers at every crematorium and graveyard.

A few deaths here and there from Covid could easily slip under the radar and not be noticed at all. Given the sheer scale of India’s demographics, such unrecognised mortalities could easily add up to tens of thousands, or lakhs.

This scenario of unrecognised Covid-related deaths in obscure rural locales is not unlikely.  Armies of migrant labourers have walked back from their urban workplaces to rural homes. They may have taken the disease with them. Young asymptomatic labourers could plausibly infect entire families in rural locations where the healthcare system doesn’t exist and social distancing is a joke. Elderly people dying in such locales would not be registered as Covid cases, if they are registered at all.

Try this maths problem. An infection rate of 0.1 per cent (one in a thousand infected) would add up to 1.3 million cases across India’s 1.3 billion population. India is currently testing 200 persons per million only in targeted high-risk populations. Let’s assume that it starts randomly testing 200 per million in the general population, (where there would be about 1,000 cases per million). What is the chance that even one infection case would be picked up in 200 tests?  It’s less than 20 per cent! There is an 81 per cent chance that all those infected persons— every single one— would go undetected.

Testing only high-risk cases leaves us in the dark about the general rate of infection. Randomly testing general populations with only samples of 200 per million would still leave us in the dark. Our strategy of lockdowns and flattening the curve depends on extremely unreliable data. We need to test orders of magnitudes more, and we need to do this ASAP. Somewhere along the line, we also need to improve mortality data reporting. That is a must.