Do excess deaths in four States suggest COVID-19 mortality has crossed one million in India?

Based on modelling studies, death underreporting in the first wave was estimated to of two-five times the official count. Underreporting is predicted to be massive in the second wave. Now, excess deaths in the order of five times and more as recorded in the Civil Registration System in four large States of Madhya Pradesh, Andhra Pradesh, Tamil Nadu and Karnataka suggest that total COVID-19 death tally might have well crossed one million as against official count of 0.38 million.

If the official count as on May 15 was over 0.27 million deaths in India since the pandemic began, based on modelling, The Economist had estimated around one million COVID-19 deaths as on May 15.

Against a daily tally of over 4,000 deaths in May this year, The Economist estimated between 6,000 and 31,000 excess deaths per day. The Institute for Health Metrics and Evaluation (IHME) estimated that India will have one million COVID-19 deaths by August 1. 

Underreporting based on modelling studies

Based on modelling, a few Indian researchers outside India too had predicted that unrecorded COVID-19 deaths since the beginning of the pandemic might be at least five times greater, taking the total COVID-19 deaths in India to over one million. “We have estimated death underreporting by a factor of two-five in the first wave. Now with the surge, the reporting infrastructure has probably eclipsed dramatically. So I expect the underreporting of deaths to be massive right now,” Dr. Bhramar Mukherjee, Professor of Epidemiology and biostatistics at the University of Michigan told me on April 24.

“I think that underreporting varies a lot from State to State, and possibly also over time. Even within States, it is probably different in urban and rural areas. Many of the estimates (including my own median estimate of five times) are based on 2020 data. It’s too early to know whether underreporting has been higher during this wave, but as more data comes in, this will become clearer,” Dr. Murad Banaji who has been modelling COVID-19 and a senior lecturer in Mathematics at Middlesex University, London says in an email. “It may be that the speed of the surge meant that fewer of the deceased were tested, fewer received hospital care, and more died at home. All of these factors would tend to increase underreporting.”

Death registration reveals excess deaths

While the official cumulative COVID-19 death toll stands at over 0.385 million (3.85 lakh) as on June 18, the actual count might far exceed the one-million mark if the models are to be believed. Besides modelling studies, all-cause mortality numbers from India’s Civil Registration System (CRS) from Madhya Pradesh, Andhra Pradesh, Tamil Nadu, Karnataka and Assam do indicate that COVID-19 deaths in the country might be much greater than the official figures.

Based on data from the Civil Registration System, from January to May this year Madhya Pradesh reported 42 times excess deaths while Andhra Pradesh reported 34 times excess deaths. In comparison, Tamil Nadu reported 6.2 times more deaths from April last year to May 2021, and Karnataka reported five times the reported deaths in 2021.

The excess deaths in the four large States have exceeded 0.5 million in the first five months of 2021 as against around 46,000 reported COVID-19 deaths in these States this year. “Not all these ‘excess deaths’ need to be COVID-19 but they are pandemic-related, in the sense that we would not be seeing them if the pandemic was not around,” says Chinmay Tumbe, Assistant Professor at the Indian Institute of Management (IIM) Ahmedabad and author of the book The age of pandemics. “These four States comprise 21% of India and collectively are showing an under-reporting factor above 10. I would estimate the lower bound for excess deaths for all-India for the first five months of 2021 at 1.5 million.”

Dr. Banaji is a little more cautious while interpreting excess deaths based on Civil registration data but does agree that it serves as a pointer to huge underreporting of COVID-19 deaths. “Civil registration data is not always easy to interpret, and we have to remember that civil registration is not complete in every State. Bearing this in mind, and bearing in mind that we can’t assume that the excess deaths are all from COVID-19, the all-cause mortality data coming through strongly suggests – as suspected – that underreporting of COVID-19 deaths is a major and widespread problem,” he says.

Evidence from crematoriums

Additional, indirect evidence of the shocking rise in mortality during the second wave came from far in excess cremations seen in Gujarat, Madhya Pradesh and Uttar Pradesh, and a huge number of bodies floating in Ganges and buried on the banks of the river in Uttar Pradesh. “The speed of the surge means that the rise in mortality became very clear whether we look at obituaries, cremations, news reports from villages, or registration data,” Dr. Banaji says.

“The excess cremations and bodies floating in the river may provide some empirical evidence for underreporting of deaths, but any conclusive statement cannot be made without reviewing linked medical certificates and/or verbal autopsies,” cautions Dr. Tanmay Mahapatra, epidemiologist and Public Health expert in an email.

Despite ICMR guidelines on appropriate recording of COVID-19 deaths, many States simply did not record deaths as due to COVID-19 in the absence of positive test results, leading to many COVID-19 deaths being missed. According to the guidelines, suspected COVID-19 deaths which are not confirmed through testing should be reported as such. The same applies to comorbidities. States such as Gujarat wrongly claimed to be following ICMR guidelines when deaths of patients with comorbidities were not counted.

Undercounting due to absence of positive tests 

Besides testing capabilities being stretched beyond capacity during the second wave, Dr. Mahapatra points out the role of variants in returning negative results even when the patient has COVID-19 symptoms. “There was some concern that newer variants are somewhat altering the sensitivity of the tests,” he says.

“If we applied our current protocols to the 1918 influenza pandemic, we would find that no one died of influenza in India even though the estimates are now placed at 20 million,” says Dr. Tumbe. “The reason: No one was ‘tested’ for influenza back then and the government had the common sense to note that if you died of a fever in late 1918, there was a high chance that it was influenza. We need to acknowledge that we have missed out on a large number of deaths due to the insistence that only those tested positive can be considered COVID-19 dead.”

If Maharashtra added 1,328 deaths and Delhi reconciled 437 deaths after an audit in June 2020, Chennai added 444 deaths to COVID-19 toll in July last year based on the recommendations of a Death Reconciliation Committee. Bihar recently added nearly 4,000 deaths to its tally after auditing COVID-19 deaths. But these were restricted to private hospitals and clinics. But the auditing did not include rural areas and deaths that had happened outside the healthcare facilities. “It is unlikely that a reconciliation like the one which occurred would capture these deaths. As civil registration in Bihar is also weak, to capture the scale of mortality in the State, surveying (as in Jharkhand) will be needed,” Dr. Banaji says.

Death reconciliation

Jharkhand became the first State to conduct a door-to-door counting of deaths during April-May 2021; it covered three-fourths of the total population. The State added nearly 25,500 more COVID-19 deaths, which is 43% more than the reported data. Verbal autopsy fills a critical gap in measuring mortality from COVID-19 for deaths that occur outside of healthcare settings and in rural areas. “In such scenarios where there is limited cause of death data, verbal autopsies have been recognised as a scientifically valid method of estimating cause-specific mortality,” says Dr. Mahapatra.

Emphasising the importance of verbal autopsy Dr. Banaji says, “Jharkhand has set a good example by doing this survey. Trying to accurately measure mortality over the course of the pandemic is crucial to understanding its impact. Verbal autopsying can be a valuable element in such surveying.”

Underreporting even in cities

Though Mumbai and Delhi witnessed a huge number of cases in both waves, one can expect underreporting in cities to be on a lower scale than in rural areas but cannot be ruled out. “We know that underreporting in cities can also be high. The impacts on marginalised communities (homeless, migrant workers, and those in slums) may not be captured in death registration data. So, even in cities, it’s crucial to have civil registration data, and ideally mortality surveying to assess the pandemic’s impact,” says Dr. Banaji.

For instance, excess deaths that went unrecorded were five times in Bengaluru since January this year and 10 times in Hyderabad during the period April 2020 to May this year. Dr. Banaji says: “It is hard to know why this factor is so high, and whether we’ll see similarly high underreporting from other cities during this wave when they release mortality data. It is possible that the speed and scale of the surge meant that more people died at home without being tested. It is also possible that a large number of deaths were incorrectly classified.”

“These examples show that while we might, in general, expect reporting from cities to be better than from rural areas, we cannot always assume that this will be the case,” Dr. Banaji adds.

But it must be borne in mind that deaths not reported within 21 days are considered as being missed and counted as excess deaths. In most cases, cities have access to medical certification and have the data within the public system to know what the actual number of excess deaths is. So it should not take much time to analyse the data. It is important to note that transparency and timeliness in death reporting are integral for the success of health emergencies and in ensuring better preparedness.

Published in The Hindu on June 19, 2021