India crossed the 3-million Covid-19 case mark on Saturday. Data released on Saturday morning showed that India conducted over a million tests on Friday. This column recommended on July 7 that India set that target for itself with a deadline of August 15. The only country to have done more, albeit in a short burst, is China, where, on one day in May, Chinese health officials tested around 1.5 million people in Wuhan (where it all began). In a two-week period, they tested almost all 11 million of the city’s residents. China clearly has more testing capacity. A June 24 report by Reuters, citing an official at the country’s National Health Commission said the country was capable of conducting 3.78 million nucleic acid (molecular) tests for Covid-19 in a day, up from 1.26 million in early March.

In absolute terms, India is third in terms of testing (according to data from worldometers.info on Saturday morning). The country has thus far carried out almost 34.5 million tests, behind China’s 90 million and the US’s 75 million. In terms of tests per million, the country (it has tested around 25,000 per million of its population) lags many others. It is ninth among the 10 countries that have seen the most number of cases, and 16th among the 20 countries with the most cases. Russia tops in terms of tests per million (around 232,000), followed by the US (225,547) and the UK (223,403). Still, India’s achievement, given its size and the low testing capacity it started with — see front page — is very significant. Just to put the number in context, if India does as many tests as it did Friday on every day between now and the end of September, even accounting for the drop off in testing over weekends, it would have tested around 5.5% of its population. In fact, that should be the next target.

Also read | Covid-19: A million and a manifesto

India’s overall number hides extremes (like averages do). At one end, Delhi has tested 70,300 people per million of its population. At the other, Madhya Pradesh has tested 13,788. Many Indian states are testing above the 140-per-million daily benchmark recommended by the World Health Organization, but this is a woefully inadequate number. For instance, it would require a state with a population of 100 million to carry out a mere 14,000 tests a day. Starting from zero, it would take the state 286 days to cover 4% of its population.

But how much should a state test? What tests should it use? And, given all the talk of pooled testing as a way to rapidly test a large number of people, should India start doing that? These are the three most important questions India’s health ministry, states, and the Indian Council of Medical Research have to answer — and the answers may well decide the trajectory of the pandemic in India from now to the end of the year.

The first question is simply answered. Universally, epidemiologists recommend that 10% of the population be tested. This writer’s own opinion (which has been stated before) is that states with a population in excess of 50 million should test 4-5% of their population; a population between 20 million and 50 million, 5-7%; and a population below 20 million 7-10%. If the positivity rates in these states plateau and then start to fall with increased testing, and reaches 5-7%, it means they are testing adequately. Otherwise, they aren’t, and they should just test more, even if they have already met the benchmark listed above.

The second question, too, has been answered before in this column (but no one seems to be listening). The best tests to use for diagnosis are molecular tests such as the RT-PCR one. Antigen tests, which accounted for roughly around 30% of the tests on Friday, are much faster (effectively an hour compared to 3-4 days for molecular tests) but inaccurate. They should be used when time is a constraint — in a containment zone where cases are peaking for instance, and infected people have to be quickly identified and isolated; or at a public event which needs to be held; even in airports and railway stations (every traveller should be administered one). I have previously suggested using two different antigen tests on a sample (or using the same test twice) to reduce the inaccuracy, and testing authorities should consider doing this. When time isn’t a constraint, testers should only use molecular tests.

Also read | Covid-19: What we still don’t know

As for the third question, pooled testing would appear to be out for India. Positivity rates are still high, and antibody surveys (where they have been conducted) indicate that the proportion of those infected is high. Pooled tests work only when a fraction of those being tested is infected.

One target has been met — now India should focus on the next.