Within weeks of the outbreak beginning in early March, mathematical and epidemiological models predicted that India was on track for a high case load given its large population and dense cities.

“The first million has happened in four months, the next million will happen before the end of next month if we give in,” says Bhramar Mukherjee, the head of the Biostatistics department at University of Michigan in an interview to HT.

Mukherjee and academics from two other American universities were part of an India study group that first predicted India’s high caseload will reach to the levels it has now. Edited excerpts:

Your initial projections included assumptions about how the outbreak would pan out in India and these now appear to have come true. Can you explain to us how you arrived at the results?

Several things that led to the accuracy of our projections:

(1) We updated our projections daily and our model got calibrated with more data on the ground each day. By May, it was robust and ran like a well-oiled machine. Initially it had larger uncertainty. I still do not trust long term projections (beyond two months) very much.

(2)We made assumptions about the reproduction number when India comes out of lockdown under two scenarios of cautious return and moderate return. For forecasting we set R (reproduction number) at 1.2 and 1.5 respectively. Coincidentally the national R has stayed between 1.2-1.3 for the last one month. That also explains the agreement.

(3)We quickly realised national data was masking state level trends so started to predict which states should be alert through state-level models. Our metrics dashboard is a very cohesive public health summary.

What could authorities and the general public in India have done to have undercut the predictions you and your team made?

First, centralised isolation. It is hard to self-isolate in Indian households. This can prevent household transmission and cluster transmission in high density neighbourhoods and large households. For populous countries like India, this is critical. Second, India must ramp up testing and carry out symptom tracking in absence of testing.

There should be an integrative leadership that coordinates between different sectors of healthcare, public health, social support and economy. The measures must earn trust from the public around data and information.

India should have data-driven policy making at a granular level. It is data deficient in the health sector and that is hurting us.

And lastly, there should be a balance between alarmism and denial. We need to accept that this is a long haul. We cannot surrender to destiny. We are all tired but we have to keep at it.

What part of the current data would you particularly like to draw attention to in order to slow the next million cases and what should India focus on?

It is very crucial that the results of population-based large sero-surveys are released soon. Existing data should be used to:

(a) Flag states/metros at nascent stages of growth and impose strong measures including punctuated and modulated lockdown. This cannot be done when infections are already leading to large case-counts, as these come with a lag. What you see as cases today happened as infections two weeks ago.

(b) Scale up Covid-testing and treatment facilities. The healthcare system is fragile. Everyone cannot fall sick at the same time and still have a bed, oxygen. Anticipate emerging peaks, deploy resources in a data-driven way.

(c) Before imposing lockdown, quantify what a two week lockdown will do for you, how much can you ramp up testing and Covid treatment? How many infections can you delay? Lockdown is not the cure or a default strategy. We need to re-lock only when necessary. A long term effective control of the virus will automatically revive the economy.

With 400,000 active cases, we will never be able to afford a long enough lockdown to get the cases to 100s.

While the million cases prediction has come true, the government has said that it has reduced fatalities and slowed the growth of the epidemic. Would you agree with the assessment?

The first million happened in 4 months, I expect India to hit 2 million by the end of August if things are going as they are, even if Delhi, Maharashtra and Tamil Nadu bring it under control as they seem to be closely inching toward.

The low case fatality rate is a positive factor in favour of India but in India, a very large percentage of deaths are not medically reported. Even if they are, cause of death is often misclassified. Case fatality rate is a ratio of the number of deaths divided by the number of cases. Neither do I believe the numerator, nor the denominator, so I would not rely on the ratio to rescue me from this long and painful crisis.