How many excess deaths in India were because of Covid-19? It’s difficult to say

Andhra Pradesh, Bihar, Kerala, Madhya Pradesh and Tamil Nadu saw over 4,60,000 excess deaths in the first five months of 2021, compared to a typical year. Yet the official Covid-19 death toll for these states in this period, which covers the peak second-wave months of the pandemic in India, accounts for just 6% of these excess deaths. How many of the remaining 94% are “missed” Covid-19 deaths will be central to the debate over how well or poorly India handled the pandemic, but will not be easy to disentangle.

By the time India’s first wave had receded, there was already reason to be sceptical about the numbers. For one, multiple states had admitted that they were using a stringent definition of a Covid-19 death. Only deaths of people who had tested positive for Covid-19, and had died soon after in hospital with a typical progression of disease, were being counted as Covid-19 deaths.

This despite the fact that India’s official Indian Council of Medical Research guidelines included a World Health Organization code for recording suspected Covid-19 deaths: deaths of people with the symptoms, but who had not necessarily tested positive prior to death. Through the last year, no state has included a suspected Covid-19 death in its published data. “Including such deaths also would have helped in understanding the disease better and in taking appropriate clinical and public health actions,” Prashant Mathur, director of the National Centre for Disease Informatics and Research who authored these guidelines told IndiaSpend.

Then there was the fact that India’s case fatality rate – the ratio of officially reported deaths to officially reported Covid-19 cases – was (and remains) much lower than in most of the developed world, for reasons that were not evident. “I think the question to be asked then was whether India was exceptional, or whether there was something exceptional going on with the data, and I think that we are finding that it was the latter,” Murad Banaji, a mathematician and lecturer at Middlesex University in the UK who has been studying and writing on India’s numbers, told IndiaSpend.

During India’s second wave, as hospitals began to fill up and turn patients away and Twitter turned into a national helpline for desperate relatives looking for beds, and as local reporters from across the country began to report crematoria and burial grounds running at far more than full capacity, louder questions began to be raised over India’s official Covid-19 count, and better ways to estimate the true toll from Covid-19.

Over the past month, journalists from across the country have accessed Civil Registration System data from cities, districts and states to estimate what monthly mortality in recent non-pandemic years looked like versus mortality during the pandemic. The difference between the two produced excess mortality estimates for pandemic months, a system that has also been used in the United Kingdom, South Africa and Peru to gain a better understanding of missed Covid-19 deaths.

We analysed Civil Registration System data for only five states – Andhra Pradesh, Bihar, Kerala, Madhya Pradesh and Tamil Nadu – for which month-wise Civil Registration System data are available. A comparison of Civil Registration System deaths from March 2020 to May 2021 compared to the equivalent period in 2019 provided excess mortality figures during both Covid-19 waves.

The ratio of pandemic mortality to usual mortality is the total Civil Registration System deaths from March 2020 to May 2021 divided by 2019 data for an equivalent time period. We used 2019 as the year of comparison as it had the highest recent mortality for all the states considered.

The data show that both the absolute excess mortality and its value proportionate to that region’s officially recorded Covid-19 deaths vary significantly, with Madhya Pradesh reporting the highest excess deaths of any state for which data are thus far available and Andhra Pradesh reporting the highest numbers proportionate to its usual mortality. Kerala reports both the lowest excess mortality and the lowest proportionate numbers.

Rough assessments

One crude way to estimate the extent of Covid-19 undercounting is to assume all of these excess deaths are Covid-19 deaths, and compare the excess mortality with officially reported data for the same periods to come up with “undercount factors”.

Indeed, such an approach is not without precedent: for instance, during the “Spanish Flu” epidemic in 1918. “In 1918 for instance, the death rate increased five times during November, the peak month of the flu pandemic,” Chinmay Tumbe, historian and assistant professor at the Indian Institute of Management, Ahmedabad, told IndiaSpend. “This was never seen in previous Novembers. You get such large spikes in death numbers only in abnormal times.”

Tumbe has recently authored a book, The Age of Pandemics: (1817-1920): How they shaped India and the World including on the Spanish Flu, particularly on how past pandemics played out in India.

No foolproof method

Given the reported difficulties in accessing healthcare during the pandemic’s second wave in particular in India, there might be a reason to exercise some caution in attributing all excess deaths to Covid-19.

“There is really no way of attributing what percentage of excess deaths are really due to Covid-19 in an emphatic, foolproof way,” Bhramar Mukherjee, chair of biostatistics and professor of epidemiology and global public health at the School of Public Health, University of Michigan, US, told IndiaSpend.

In the United States, where official Covid-19 data can largely be trusted, officially recorded Covid-19 deaths account for 72% of the excess mortality between March 2020 and January 2021, Mukherjee said, pointing to a recent paper as an example. Even so, the door is left open to acknowledging that some of the remaining deaths could be missed Covid-19 deaths. The paper ends by saying: “Excess deaths not attributed to Covid-19 could reflect either immediate or delayed mortality from undocumented Covid-19 infection, or non-Covid-19 deaths secondary to the pandemic, such as from delayed care or behavioural health crises.”

“For India, I believe a smaller portion of excess deaths can be explained by reported Covid-19 deaths (perhaps 10%-20% instead of 72% as in the US), thus increasing our probability for believing there were many undocumented Covid-19 deaths. But we cannot identify the different buckets of direct or indirect impact, or what can be attributed to deliberate/intentional data suppression,” Mukherjee said.

Mathur agrees that attributing the cause of death from excess mortality alone is difficult. “Without proper cause of death data, it is difficult to say how many of these were Covid-19 deaths,” Mathur said. “We know for example that the treatments of other serious and chronic conditions including cancer were impacted. There could be both excess Covid-19 deaths as a result of patients not being tested especially in rural areas, as well as an increase in deaths from other serious health conditions.”

The only state for which cause of death data are available is Tamil Nadu, for which The Hindu accessed Civil Registration System data from 2015 onwards. These data show that deaths from pneumonia, diabetes and unclassified causes increased substantially in the first five months of 2021, pointing to the possibility of misclassified Covid-19 deaths, an official in the state’s health department who asked not to be quoted, admitted.

Knowing the true extent of Covid-19 mortality matters in planning for the future and from a broader public health perspective. But ultimately, estimating the toll from Covid-19 should not distract from the human tragedy of the pandemic, said Tumbe.

“The larger point is that all excess deaths in a pandemic should matter,” said Tumbe. “The lower the percentage of Covid-19 deaths in excess deaths, the worse it reflects on the government of the day because the question arises of why did so many die in that case.”

This article first appeared on IndiaSpend, a data-driven and public-interest journalism non-profit.