In an email interview with indianexpress.com, Professor Paul Kattuman, an economist and applied econometrician at Judge Business School, who worked on developing the India Covid-19 tracker, tells us that patterns in the data that suggest cases are now increasing super-exponentially and the brunt of the current wave fuelled by Omicron, which is more transmissible and has better immune-escape mechanisms, will fall on the older and the immunocompromised. He, however, says that once the Omicron wave has swept the world, Covid-19 could well become endemic and operate at a level that more or less standard public health measures can cope with.

Edited excerpts from the interview:

Tell us a little bit about the work that went into building the India Covid-19 tracker? What is your association with the project?

My research is centred on applying statistical methods to analyse data in order to draw out valid inferences.

Early in 2020 when the UK government geared up to address the threat of the pandemic, Public Health England (now called the UK Health Security Agency) in the East of England region reached out to the University for help with the analysis of voluminous pandemic data. This generated a number of collaborative efforts.

A small team assembled at Judge Business School by my colleague Stefan Scholtes, Professor of Health Management, began work on forecasting trajectories of a variety of pandemic related variables—new cases, hospital admissions, ventilator demand and deaths. The objective was to give reliable forewarning so that the health system could be operationally prepared for the immediate future. The urgency and importance of this mission led me to focus on new and more reliable time series forecasting models.

Constant engagement with Public Health England and National Health Service helped us develop and finetune these methods and make them fit for operational purposes. We collaborated with the National Institute of Economic and Social Research in London to start off a UK wide tracker in February 2021.

In August 2020, I had started applying these methods to provide weekly forecasts for Kerala, working with Rajeev Sadanandan, CEO of Health Systems Transformation Platform, who had been appointed advisor on Covid-19 to the Chief Minister of Kerala.

The usefulness of the state level forecasts led us to extend our work, in May 2021, to all Indian states and union territories. The proven value of the national tracker led to requests for more detailed weekly forecasts by Punjab and Tamil Nadu. At the end of October 2021, the national tracker and state-level forecasts moved to an occasional series.

What is the kind of data you use for your modelling and the methodology that helps you make the forecasts?

The data used to generate forecasts are all publicly available. Both for the national tracker and for district-level forecasts for Kerala, Punjab and Tamil Nadu, I used data sourced from the Covid-19-India API at https://api.covid19india.org/. This excellent voluntary collective closed its operations after 19 months of intense effort in October 2021. Since then, I have relied on Johns Hopkins University of Medicine coronavirus resource center for data on India.

In terms of method, we use a new time series forecasting model developed by my colleague Prof. Andrew Harvey and myself. The model is based on classical statistical methods and extracts the time-varying trend of the pandemic variable of interest, stripping out the day-of-the-week effects and purely random variation.

The model belongs to the family of structural time series models. It is capable of taking note of variation in the trajectory over time, due to mitigation measures and changes in social behaviour, for example. The model makes minimal assumptions in using the time series data to estimate its parameters. The forecasts produced are more accurate over shorter horizons. Work on a number of useful extensions is ongoing.

Our model stands in contrast to SEIR-type models which are commonly used in epidemiology and are more useful for determining the impacts of alternative policies using simulations. The two model types complement each other.

(Source: India Covid-19 tracker)

In a recent interview to Bloomberg, you said, “It is likely that India will see a period of explosive growth in daily cases and that the intense growth phase will be relatively short.” Could you please explain how you made this prediction?

The trend in the filtered growth rate of daily cases (with day-of-the-week effects and purely random variation stripped out) is revealing. On December 25, the filtered value of the daily growth rate of India-wide new cases was marginally negative at -0.4%. This rose to 0.6% on December 26, to 2.4% on December 27, and to 5% on December 29. The increasing growth rate indicates super-exponential growth of daily cases.

This pattern of a short period of explosive growth in new cases is a common experience with Omicron, internationally. In the UK, for example, the first Omicron case was identified in the third week of November. The filtered trend in the daily growth rate of new cases in England remained close to zero at the beginning of December. After an initially gradual rise, daily growth rate in infection shot up sharply over just five days, from about 1.5% on December 13, and to 7.5 % on December 18.

The growth rate has since then declined significantly, having run though the larger part of the susceptible population (that had been reduced by an intense booster vaccination programme). However, the growth rate remains positive as of date. Daily cases are expected to continue to peak in the first or second week of January 2022.

The Cambridge India tracker had correctly predicted the peak of the second wave in India in May. What are your predictions for a third wave? Also, how long is the third wave likely to last in India and when can it peak?

At this point in time, when cases are growing in a super-exponential fashion, the time series model cannot reliably indicate when exactly the peak will occur or how large the number will be at the peak. The environment in which the disease progresses is changing over time, not only due to mitigation measures but also social behavioural changes. We will be keeping a close look on the trajectory and hope to be able to indicate the approaching peak as soon as the signal is clear.

(Source: India Covid-19 tracker)

There has been a lot of discussion and debates on how intense the third wave will be. Many scientists have said it can be milder than the devastating second wave that wreaked havoc across the country. According to your model, roughly how many daily cases, hospitalisations and deaths could we be looking at during the third wave?

One of the lamentable limitations to do with Covid-19 data for India is the lack of reliable data series on hospitalisation and deaths, as far as I know. So, model-based predictions of hospitalisations and deaths cannot be generated.

While Omicron is now known to be much milder in causing serious illness than Delta, it is also known to be much more transmissible and more capable of immune-escape. Experience in the UK suggests that these opposing forces play out, after a lag of time, to generate some noteworthy increases in hospitalisation of those in older age groups. While the young—by far the larger majority in India—should be able to recover readily, those who are immune-compromised are likely to be more severely affected.

According to emerging data patterns, which states according to you are at risk of seeing a surge immediately? Is it possible that a third wave may involve localised surges and impact some states more than the other? Or are we going to see a more or less widespread impact across the country as was the second during the second wave?

Looking at the experience of Indian states, on December 24 only six states saw daily growth rates exceeding 5%. By December 26, this had grown to 11 states, and by the 29th, to 14 states.

At present there is no reason to believe that any state will wholly escape the Omicron wave, unless totally isolated. But then again, these extremely high growth rates cannot persist for very long. Of course, daily cases will continue to rise till the growth rate itself declines to zero.

Is Omicron likely to replace Delta as the dominant variant? And will it be fair to say that Omicron will fuel the third wave?

Reliable and comprehensive data on variants do not seem to be available. From the UK experience, Omicron will play by far the decisive role in the current wave.

The big question everyone seems to be asking now is if Covid-19 could enter a stage of endemicity later in 2022. Or is that contingent on the role of vaccine coverage, how long jabs offer protection and their effectiveness, and the emergence of new variants?

As I mentioned earlier, cases are growing in a super-exponential fashion now, and the model cannot reliably indicate when exactly the peak will occur. We hope to keep a close eye on the trajectory and hope to indicate the approaching peak as soon as a clear signal emerges.

Epidemiologists and virologists are keenly studying the hypothesis (based on a now well-known South African study of small sample size) that Omicron infection boosts immunity against the more virulent Delta variant.

If that is the case, acknowledging the imponderable of further variants, once the Omicron wave has swept through the world, Covid-19 could well become endemic and operate at a level that standard public health measures should be able to cope with. But the brunt of a sweeping Omicron wave will fall heavily upon those of us who are older or immunocompromised.