Mathematical models project a peak of 800,000 daily new infections over the coming weeks, double the current levels. Through May, at least 100,000 will die. New hotspots could be in Uttarakhand, Odisha, West Bengal. Data-driven policy, missing until now, could help flatten the curve
Mumbai: India is poised to be home to the world’s biggest Covid-19 crisis since the pandemic began in March 2020, with daily new cases projected to soar to 800,000 new infections daily, double the current levels, according to expert estimates and projections based on mathematical models.
Those models predict that the month of May 2021 could record the worst ever death toll independent India has witnessed in decades, with the coronavirus projected to claim up to 4,500 lives every day by May-end, up from 3,521 on 2 May 2021.
At least 100,000 people will die in the next month, a “tragedy” because vaccines will be widely available in a few months, said Bhramar Mukherjee, PhD, professor of biostatistics and epidemiology at the University of Michigan, whose study group predicted a peak in the number of reported new cases at 800,000 to a million per day in mid-May.
Having crossed the threshold of 300,000 daily cases on 21 April, India is already recording more new infections every day than any other country since the pandemic began. There may be thousands more unrecorded because testing has plateaued around 1.8-2 million from 1.1 million in early April. Experts said India may be undercounting cases by a factor of between 5 and 20.
The United States is the only other country to have recorded more than 300,000 new infections in a day, its largest single-day spike coming on 2 January, at 3,00,310 new cases. On 3 May, India recorded 355,769 new infections and 3,439 deaths.
Prompted by reports of the ceaseless cremations; desperate social media pleas for oxygen, ICU beds and ventilators; and a vaccine policy that neglects the vast majority of Indians, the White House’s chief medical adviser Anthony Fauci called it a “terrible, tragic situation”.
The unfolding oxygen crisis in Delhi has seized global attention, but experts say the new hotspots could be cities in Assam, West Bengal, Karnataka, Andhra Pradesh, Telangana, Odisha and Uttarakhand, all states with a reproduction number or R-number of more than 1.5—which indicates a higher transmission rate—and inadequate daily testing:
On 2 May, this transmission rate, the number of people every patient will infect, is 1.94 for Assam, a state that has been unable to scale up testing during April. On 1 April, Assam reported 58 new Covid-19 cases. A month later, on 1 May, the state reported 3,453 new cases. By 1 June, according to Mukherjee, the small state could see its cumulative case count grow to 965,000.
Uttarakhand, which recorded a rise in daily new cases from 500 to 5,493 between 1 April and 1 May, halved its daily tests between 26 April and 1 May. The state’s R-value is 1.61 as of 2 May. The University of Michigan model predicts 1.1 million cases in the state by 1 June.
West Bengal, with a test positivity rate of 31% and R-value of 1.62 as of 2 May, has also not been able to scale up daily tests that have hovered in the 50,000-56,000 range for the last two weeks. On 1 April, West Bengal reported 1,274 new cases, which rose to 17,512 new cases on 1 May. Mukherjee’s model predicts 3.2 million cases in the state by 1 June.
As international aid began to pour in, the tsunami of new cases in India will remain under worldwide observation through May. Recent studies by groups of biostatisticians, epidemiologists and health economists converged in their projection that health infrastructure and administration will have to brace for impact as the ‘peak’, or the maximum number of daily new infections, will occur around the second or third week of May.
India Driving Up Global Numbers
India’s unfolding calamity is driving up worldwide pandemic numbers.
A weekly epidemiological update by the World Health Organisation (WHO) for 19-27 April said “India accounts for the vast majority of cases” from the rising graphs in South-East Asia, and 38% of global cases reported during that week. The highest weekly numbers were reported by India, United States, Brazil, Turkey and France, but all the others showed a decline from the previous week. India’s weekly numbers were up by 52%.
The Institute for Health Metrics and Evaluation (IHME), an independent global health research centre at the University of Washington, estimated that the peak of new daily infections, including those not tested, occurred on 23 April, at more than 14 million cases. Daily deaths, considered the best indicator of the progression of the pandemic, are projected as crossing 13,000 on 15 May—the IHME estimated the lag between infection and death to be 17 to 21 days.
This study projected that by the end of July, India would have recorded a total of more than 950,000 Covid-19 deaths. Universal use of masks would cut total deaths at 880,000, according to the IHME. They estimated current daily fatalities at nearly 8,000, as opposed to the reported mortality data of 3,521 for 30 April.
The government of India itself believes that the ongoing surge will peak by mid-May, when the daily count of new cases could reach 5 lakh, according to a presentation made by Niti Aayog member V K Paul at a meeting of prime minister Narendra Modi with chief ministers. According to Paul’s presentation, the surge may subside by June-July.
All these projections are based on mathematical models that help analyse a range of factors to create effective intervention strategies. Qualitatively, the different projections agree that the worst is to be witnessed in mid-May, but also that waiting to verify whether evolving data validates the projections is dangerous.
“What are we waiting for?” tweeted the University of Michigan’s Mukherjee on 28 April, wondering why more stringent lockdowns were not in place yet. Her first warning about a sharp rise in cases came in February.
Data Deficits And Data Denials
The models converge in one more important aspect—they agree that official Indian statistics on the coronavirus pandemic vastly underreport the truth.
The University of Michigan’s Mukherjee built a mathematical model with her COV-IND-19 study group of researchers at the beginning of the pandemic. According to her projections, India will have totalled 50.3 million cumulative cases by the end of May 2021.
She estimates that the total death tally could be two to five times more than official statistics, and the total number of infections could be 15 times higher.
The IHME’s estimate of underreporting is even higher. Their analysis of seroprevalence pegged the infection detection rate at below 5%. “This means that the number of cases that are being detected needs to be multiplied by 20 or more to get the number of infections that are occurring in India,” IHME professor Christopher J L Murray said in a weekly video update posted on 23 April.
More infections were happening in India currently than the global total from the first week of April, “extraordinarily large” numbers in Murray’s words.
Underreporting is not the inability to detect cases, a problem across countries, Mukherjee told Article 14 in an email interview. “This must be decoupled from the lack of reporting of symptomatic cases due to data suppression or lack of testing,” she said. Her estimates also cross reference past sero-surveys and other models.
Health economist Rijo M John, PhD, a consultant to the WHO and guest faculty at the Indian Institute of Management-Kozhikode, said India may never record these projected numbers, even if the models are accurate, for testing will not be adequately expanded.
Testing for the SARS-CoV-2 virus has plateaued in India despite the exponential growth in the number of infections. Through April, as daily infections more than quadrupled from about 80,000 a day to nearly 400,000 a day, tests rose from 1.1 million to a little over 1.9 million a day, slowing on Sundays and holidays, dipping to a little more than 1.4 million on 25 April, 1.8 million on 1 May and 1.5 million on 2 May.
John said he expected India to conduct 2 million to 2.5 million tests a day by the next three to four weeks, during the predicted peak. “I think we would have a manufactured peak sometime around then,” he said, at about 500,000 new cases a day. “That doesn’t mean the models are incorrect—they may be very right that we have more than a million cases on a daily basis. But our reported data may not reflect that,” he told Article 14.
Several experts have expressed concerns over the dressing of data by state agencies, especially when the numbers of the dying visibly exceed official statistics in multiple cities across India, a trend that endangers policy response.
Virologist Shahid Jameel, PhD, director of the Trivedi School of Biosciences at Ashoka University, said information from the ground suggested widespread “fudging of data”, confirmed by an Article 14 check of actual and official death tolls in six Indian cities.
Jameel cited the example of a friend who has returned from Zurich to care for his mother, hospitalised with Covid-19 in one of the larger cities of Uttar Pradesh. When he was trying to get tests done, the representative of the pathology laboratory told him he was allowed to test only 10 people a day. “Denial of data doesn’t help anyone because all your projections will be wrong,” Jameel said.
Models, their projections and their implications on policy, can only be as good as available data, said experts.
Unpacking The Data And Projections
India’s cumulative number of Covid cases surpassed those in Brazil on 11 April, and is now second only to the US, which recorded 15% fewer infections in the week ending 27 April than the previous week.
Since 26 April, India is also recording more daily deaths from Covid-19 than any other country currently, surpassing Brazil. On 1 May, India recorded 3,685 deaths—the highest single-day mortality globally since the start of the pandemic. India’s daily mortalities have more than trebled in two weeks. The previous worst single day spike in deaths was recorded in the US on January 13, with more than 3,400 deaths.
According to OurWorldInData.org, the actual death toll is likely to be higher than the number of confirmed deaths on account of limited testing and problems in the attribution of the cause of death. In sheer numbers, India’s body count will be staggering, even assuming a lower rate of mortality of 1.1% of all infections that union health minister Harsh Vardhan recently alluded to.
Immunologist and faculty member at Pune’s Indian Institute of Science Education and Research (IISER) Satyajit Rath, PhD, said the lower mortality rate must also be seen against the backdrop of under-counting of cases and deaths. “…we should also remember that our current case load is approximately 225 cases per million people per day, while the highest reported so far, in early January 2021 in the UK, was over a thousand cases per million people per day,” he told Article 14.
The data, and at least some of the projections, are also highly granular, with significant regional disparities and contradictions with the on-ground experience .
According to Mukherjee’s forecast, cases in Maharashtra will continue to rise sharply to reach 7 million cases by the end of May. The state’s current cumulative cases stands at 4.6 million, with a curve that is appearing to flatten out.
In fact, Maharashtra’s contribution to the total number of active cases in the country has been on a steady decline. At the start of the current surge, Maharashtra was reporting 60 to 65% of all active cases in the country. That has now reduced to 23%, or less.
Among the states whose contribution to the total number of active cases in the country is growing are Uttar Pradesh and Karnataka, with about 10% each. Kerala’s contribution to India’s active cases is rapidly increasing too, and approaching 10%.
Why Some States Report More Cases
These regional case variations will have to be seen in the context of regional disparities in the testing curve—more testing is simply the cause for more cases in some states, and not in others.
John cited the example of Delhi, which has seen a rising graph for new cases despite a reduction in testing by 27%. On 11 April, Delhi conducted its highest-ever number of tests in a day at 114,000, including rapid-antigen tests and RT-PCR tests. Since then, testing has shrunk consistently to a total of 81,829 tests on 29 April, expectedly sending test positivity rate (TPR) skyrocketing to nearly 32%. On 26 April, the state tested less than 58,000 samples.
“This is a very grave situation in Delhi,” said John. “They are missing out on reporting a lot of cases due to lack of testing.”
In Telangana, similarly, testing has declined alongside more cases being detected, leading to a higher TPR that has risen from 7.5% on 25 April to nearly 10% five days later.
Kerala, on the other hand, is showing a rising trend of cases as well as tests. In the span of a month, Kerala’s TPR grew from less than 5% to 25%.
But experts are more concerned about states with a low testing-high infections trajectory like Delhi’s. Lower testing indicates a lost opportunity to identify and isolate patients, raising the risk for the spread of infection.
These wide state-level discrepancies in testing also question the idea of following the trajectory of the first and second waves to identify a geographical priority route for vaccination. At a recent webinar on India’s second wave, Gagandeep Kang, FRS, one of India’s most eminent vaccinologists and professor of microbiology at the Wellcome Trust Research Laboratory, Christian Medical College, Vellore, said this was akin to “looking for cases under the lamppost”; states with better infrastructure have been able to considerably scale up testing.
Another problem with the testing data is the growing discrepancy between the total number reported by the Indian Council of Medical Research (ICMR) and the aggregate of states’ reported tests combined. This discrepancy has been growing, and states are now reporting an average of nearly 4 lakh cases daily in excess of ICMR’s testing numbers. “That opens up a Pandora’s box,” said John, who has been tracking publicly available data related to the pandemic since March 2020. “It can’t be that ICMR is deliberately under-reporting actual tests done, there is no incentive for them to do so.”
So, are the states inflating the numbers? John believed this could be the case, for states may inflate testing to artificially achieve lower TPR, a particularly key measure now that the Centre has issued guidelines asking for strict restrictions in districts where TPR is greater than 10%.
But with the ICMR not publishing its testing data in a state-wise format, there is no way to say which states’ testing numbers show the widest mismatch with data from the former.
Across specialisations, experts agree that the lack of transparency in data has stymied their efforts for insightful analysis or accurate predictions or better models. “Once the experts fail, the eventual result is that we fail to control the pandemic,” said John.
Identifying Good Warnings, Bad Signs
These projections come at a time when India’s capital city’s best-equipped public hospitals are overwhelmed, while from rural India reports emerge of febrile patients unable to get tested.
The sobering forecast is no bolt from the blue. Warning signs and express messages of caution to the Government of India were made as early as November last year when the Parliamentary standing committee on health and family welfare submitted to the Rajya Sabha chairman and Lok Sabha speaker its report on the outbreak and its management, specifically referring to precautions around the “super spreading series of festive events”.
In February 2021, experts tracking the data had begun to notice the curve inching back upward. The effective nationwide R-number or reproduction-number was above one.
On 28 February, Mukherjee tweeted about the uptick in cases, calling for more data on proposed vaccinations in order to better inform her projections. “Double masking, avoid large gatherings, sanitization, get vaccine when our turn comes,” she wrote, and added, “Policymakers: Scale up vaccine roll-out.”
February and March also saw new strains of the virus driving an exponential spread of infection in different pockets of the country, straining medical infrastructure. Jameel said it appeared that the clinical progression of the disease may not be very different, but as the strains are more infectious, entire families took ill very quickly.
“The numbers are overwhelming the hospital system, and whatever you may be seeing in terms of mortalities or people not getting well as quickly as earlier may be a result of that,” he said.
Jameel told Article 14 that if data and current trends are reported incorrectly, no model can be accurate, or effective. And because human behaviour is unpredictable, there will be too many unknowns to make reasonable projections for an infectious disease’s trajectory beyond a few weeks, after which it would require to be updated.
Modelling and data also appear to have not informed more location-specific policy, where they could have proven useful. Mukherjee said Kolkata, Bengaluru, specific cities in Uttar Pradesh and Bihar have high case counts as well as high R-values, something the government should be paying attention to. Assam and Odisha have a higher R-Number, but a smaller case load.
“We need stringent lockdowns in the next set of anticipated hotspots instead of waiting like we did in Mumbai and Delhi,” said Mukherjee. “That is why data-driven agility can help policymakers.”
Apart from the models, New Delhi appears to have missed other warnings.
The Warnings From Manaus
In January and February, Manaus, the capital of the state of Amazonas in Brazil, suffered a severe second outbreak. Manaus witnessed much the same scenes that New Delhi’s residents are enduring. Hospitals ran out of oxygen, cemeteries ran out of space, entire families including those of health-workers got infected, patients died at the gates of hospitals as they waited for admission.
In September 2020, a preprint of a non-peer reviewed study suggested that the people of Manaus had achieved herd immunity, with an estimated sero-prevalence of up to 66%. Serological surveys assess the percentage of a population with antibodies from exposure to the virus.
As the second surge tapered off, later studies questioned the concept of pursuing herd immunity through natural infections—Manaus had detected the P.1 variant or the UK variant, and reinfections were also by now a clear reality.
It was a lesson for the world. The riverside city of about 2 million inhabitants had weathered a severe first outbreak in 2020 and that scale did not pre-empt a second surge.
Despite all the warnings, in March, union health minister Harsh Vardhan said it was the “endgame” for Covid-1i in India, while preparations were afoot for election rallies and the Mahakumbh.
“Hubris brought us to this sorry pass of not having made adequate provision in time for likely outbreak levels,” Rath, the immunologist from IISER, told Article 14. “We need pro-active governance based on sober caution.”
From Crisis To Course Correction
It remains a mystery why the second wave hit at precisely this moment, but the experts are unanimous that it was not an unexpected surge.
As cases in all states surge, with the exception of Maharashtra and Chhattisgarh where the rise in numbers has slowed, it is difficult to pinpoint whether the five election states, which witnessed huge public rallies, could have contributed towards the nationwide second wave.
Jameel said people most likely dropped their guard when they saw their leaders declare that the worst is over. In January, the prime minister addressed the World Economic Forum's Davos Dialogue via video conferencing and said India not only solved its problems but was helping the world fight the pandemic.
“People see their leaders lead rallies of lakhs of people without a mask. You lead rallies in the day and promote masks in the evening. What sort of message is this?” Jameel asked.
John agreed that Covid-inappropriate behaviour may have increased exponentially because of statements by political leaders, in addition to the five state elections, the Kumbh and various religious festivals across the country.
The government of India pinned its hopes on a vaccine, and on high seropositivity found in surveys—or a sizeable percentage of the population found to have been exposed to the virus already. This would ‘help in bending the curve’, India’s Principal Scientific Advisor K Vijayraghavan said in an interview. These seropositivity conclusions, when presented as city-wide averages, were misleading, he said.
Rath, who was involved with the seropositivity study in Pune city, agrees. The level of exposure to the virus was in fact vastly different in diverse parts of the city, making the average percentage a poor rendition of ground level reality.
“The limitations of this and other sero-surveys in India have been exactly the same as the limitations of all other pandemic-related Indian studies; too small in scale, and too limited in their time span,” Rath said, adding that sero-surveys have tended to make too-broad generalisations. Studying many more smaller localities, and studying them repeatedly over time, could have yielded more granular and useful data on seropositivity.
For John, the most urgent course correction is to increase testing even if this throws up larger numbers of infections. He said he remembers that the peak of the first wave stopped short of reaching 1 lakh new infections daily, because suddenly, “magically”, testing slowed down.
“It was our testing curve that was flattened,” he said, adding that India can confidently infer that a peak has been crossed only when daily cases fall despite increased testing.
With various models agreeing that India has a few weeks to go before the curve flattens once again, policy-makers are left with revised projections and not many options for action.
John is in favour of identifying micro-containment zones and sealing these with tough restrictions, but this would require district-level, granular data to be studied, data that has not been provided by the government.
Mukherjee acknowledged that locking down was a hard decision. “There is no win here,” she said. Taking this route early means people “do not see the tsunami” and think the lockdown was unnecessary. But balancing lives and livelihoods is necessary, particularly for a country like India, she said. “If you are not alive, livelihood is not an issue, if I may put it that bluntly.”
Mukherjee said given the resource shortages, a national lockdown may be unavoidable if there is no turnaround in coming days. “Lockdown, and vaccinate, to get through this surge like a sandstorm,” she said. “Burying your head until it passes through.”
(Kavitha Iyer is an independent journalist based in Mumbai.)