Covid-19 in India — Whom to test?

Disclaimer: Data used as available till April 6, 2020. Below are largely opinions not backed by any formal medical background. Continued from Covid-19 in India — What Constrains our testing

ICMR has now allowed Rapid antibody tests to scale up testing. As we head towards the end of the 21 day lockdown period, hotspots are being identified for exhaustive testing [1]. Demarcating these hotspots and exhaustive testing in them becomes a priority. While we are expecting 5 million antibody tests and 3.4 ready PCR tests [2], who to test becomes important.

Israel has already experimented with pooling samples [3] in groups of 32 and 64, indicating that the individual sample dilution doesn’t affect the test efficacy [4]. A simple mathematical model, assuming a 3% infection rate amongst tested samples, indicates an ideal pool size of 6 — leading to 0.33 tests per person, or 3 people tested for every test*.

Pooling has been done with PCR tests — it still needs to be tried with rapid tests. If rapid tests are demonstrated to work with pooling, they can become the community level mass testing mechanism — followed by individual PCR tests of the positive pools.

Pooling allows 5 million rapid / 3.4 million PCR tests to test upto 10–15 million people. Hotspots remain a priority, here’s a further thought experiment on who must we test next if (and we must) exceed our testing capacity.

The choice is between the more at risk (older or with chronic conditions — who have been seen to have higher fatality rates) or the more asymptomatic (often young — who carry on spreading the diseases)?

The highest fatality rates have been found in the higher age group (say 65+ years of age) and having cardiovascular diseases, blood pressure, diabetes or any chronic respiratory illness [5].

Covid-19 fatality rates (source: Vital Surveillances: The Epidemiological Characteristics of an Outbreak of 2019 Novel Coronavirus Diseases (COVID-19) — China, Feb 2020)

5.3% of India population is above 65 yrs of age [6], prevalence of hypertension and diabetes is 25.3% & 7.5% respectively [7] and common flu may have infected at least 15% of Indians in the last 6 months (American numbers [8], but infection rates in India should be comparable).

On this basis, here are potentially high risk segment sizes:

  1. > 65 years of age with symptoms of flu and having comorbidity — ~ 2.6 million

(We have assumed independence of age, flu and chronic conditions. The segment is likely to be higher, since their dependent probability is actually higher)

Our testing so far has largely been symptomatic — and the above segments have already been allowed for testing. Now that our testing capacity is up, a more proactive outreach and communication to this segment — through mass media and clear action items on how to get tested will help cover more of them.

We haven’t done any asymptomatic testing (except in hot spots). The starting point of an asymptomatic carrier is likely to be international travellers who missed thermal screening or quarantines. Given restrictions on international travel in place, anyone infected is well past the time for antibody generation and thus suited for rapid antibody testing. Perhaps fetch the list of international travellers from Feb & Mar 2020 (estimated to be of the order of 2–3 million [9]) that were asymptomatic on landing, sort in the priority of then highly infected countries (ie China, Italy, Spain, Iran, then others) and even days spent outside — and have them take the Rapid test. If someone does test positive, have him now added to the contact tracing list to contact trace around. This search for imported cases goes hand in hand with contact tracing.

And this testing could be incentivised by using the government’s own Aarogya setu app. Mandate that people install the app, and their status on this app be necessary for them to step out (even those outside the quarantine / containment zones) post the lockdown period. People log in with their phone number, verify themselves with a government ID — and the system flags them as being in hotspot, quarantined or needing testing basis certain criteria — symptomatic, asymptomatic but with international travel, or contact traced of infected or needing testing. Only if their status is cleared are they allowed to venture out.

And the above still doesn’t work for those without smartphones or any documents. Obviously, it’s a can of worms with privacy issues — so it must be done right.

* Assuming infection rate of a person being tested as X%, possibility of at least 1 positive in sample of size N is (1- (1-X) ^ N)), in which case all the N people are to be tested. Hence average tests per person = 1/N + 1 — (1-X)^N


  1. Centre’s Aggressive Containment Plan To Control Runaway COVID-19 Spread

Arvind ( is a Computer engineer (IIT, MIT, Google) turned technology/impact entrepreneur.