Central government rejects WHO’s estimate of India's COVID deaths

The Indian government has rejected outright the WHO estimate that COVID-19 caused 4.7 million additional deaths in the country in 2020-21 and questioned the methodology used to arrive at it. While other countries have tended to dispute WHO’s absolute numbers of excess deaths, India has denied that they exist at all.

At a cremation ground in New Delhi on April 26, 2021, family members carrying the body of a relative who succumbed to COVID-19.

ON May 5, the World Health Organisation (WHO) released its much-awaited report on excess mortality due to the COVID-19 pandemic. According to it, globally, there were an estimated 14.9 million deaths associated with the disease, directly and indirectly, between January 1, 2020, and December 31, 2021, as against the little over 5.4 million of total deaths officially recorded by countries and reported to WHO (Fig. 1). The WHO’s latest estimates supersede the draft estimates it published in May 2021. Excess mortality is the difference between the number of deaths that have occurred and the number that would be expected in the absence of the pandemic based on data from previous years.

As explained in WHO’s press release, the number includes deaths caused directly by the COVID-19 disease and those caused indirectly because of the pandemic’s impact on health systems and society. The estimated excess mortality figures are also influenced by deaths averted during the pandemic owing to restrictions imposed on mobility, including quarantines and complete lockdowns as in India, and the consequent lower rates of events such as road accidents and occupational injuries, which could result in negative excess mortality at some point of time during the 24-month period.

Explaining the rationale behind its approach of estimating excess mortality rather than direct counting of registered deaths, WHO said in its release: “Some deaths that are attributable to COVID-19 have not been certified as such because tests had not been conducted prior to death. There have also been variations in the death certification rules countries have applied in the presence of comorbidities and COVID-19.”

Both these factors are likely to have played a role in whether deaths in India were registered as being due to COVID-19, and thus contravened the guidelines of the Indian Council of Medical Research for the registration of COVID deaths. As reported by the media, the first factor has been true in several parts of India. The second factor too has been at play, and deaths have been registered as something other than COVID-19 when the patient had some comorbidity, often deliberately, to show government in a good light and in control of the situation.

But, more significantly, as Prabhat Jha—an epidemiologist at the University of Toronto who has reviewed the WHO work—pointed out in an editorial in The Lancet , in India eight million of “all-cause” deaths lacked medical certification of the cause. “In light of the challenges posed by using reported COVID-19 data, excess mortality is considered a more objective and comparable measure,” WHO said. Excess mortality figures are thus a proxy for COVID-19-related deaths.

Of the estimated 14.9 million COVID-19-related death toll worldwide (lower estimate 13.3 million and upper 16.6 million), India alone accounted for nearly one-third of them at 4.7 million (range 3.3 million to 6.5 million), the highest figure for any country, as against its official figure of about 481,000 deaths in 2020 and 2021 (Fig. 2). This implies that India undercounted its COVID death numbers by about a factor of 10. The current official mortality figure (as of May 10) is 524,103.

Fatality undercounting has occurred in all countries. In terms of absolute excess mortality numbers, India is followed by Russia and Indonesia. According to WHO, even in pre-pandemic times, six out of 10 deaths went unregistered globally. Missing deaths are only to be expected during a pandemic, especially one caused by an unknown virus.

Given the globally inferred infection fatality rate of this virus and the level of infection that was prevalent in the country as evidenced by three serosurveys—the third serosurvey suggested that testing had captured only 3.5 per cent of the infections—it was obvious that official death figures just did not gel with the above two features of disease spread in the country. Now WHO has put a figure on the undercounting of these deaths. In India’s case what stands out is the high undercounting factor of 10. India is second only to Egypt, which has undercounted its COVID deaths by a factor of nearly 11. The undercounting factors for Russia and Indonesia are 4+ and 5 respectively. According to the report, India saw about 0.8 million excess deaths (17 per cent of the total of 4.7 million) following the first wave, between January 1 and December, 2020, and over 3.9 million excess deaths in 2021. About 2.4 million of this (over 61 per cent) occurred between May and June 2021 following the peak of the killer second wave caused by the Delta variant.

Yet, the Indian government has rejected WHO’s numbers outright and questioned the methodology it used to arrive at them, which included predictions based on modelling. Out of the 194 member states, India is the only country that has objected to the figures and dismissed the entire exercise as unsound. In fact, as Jha pointed out in a live interview he gave to theWire.in on May 10, the key difference between the responses of other countries and India has been that, while other countries have tended to dispute WHO’s absolute numbers of excess deaths, India has denied that such excess deaths exist at all.

The WHO said that its Technical Advisory Group for COVID-19 Mortality Assessment—which comprised leading demographers, epidemiologists, data and social scientists, and statisticians from a range of backgrounds and geographies—developed the methodology used for estimating excess mortality. It relies on a statistical model derived using information from countries with adequate data; the model is then used to generate estimates for countries with little or no data available. Countries of the former kind are categorised as Tier I and those of the latter kind as Tier II. That is, on the basis of the data of Tier I countries, a statistical model is built and used to arrive at estimates for Tier II countries. According to WHO’s analyses, mortality data were not available for 85 countries, 41 of which are in Africa.

In a sharply worded rebuttal issued on the day the WHO report was released, the Union Ministry of Health and Family Welfare said: “India has been consistently objecting to the methodology adopted by WHO to project excess mortality estimates based on mathematical models. Despite India’s objection to the process, methodology and outcome of this modelling exercise, WHO has released the excess mortality estimates without adequately addressing India’s concerns.”

Drawing attention to the Civil Registration System (CRS) report released just two days earlier, the Health Ministry said in its rebuttal: “The CRS data of 2020 published by Registrar General of India (RGI) on 3rd May 2022 clearly reveals that the narrative sought to be created based on various modelling estimates of India’s Covid-19 deaths being many times the reported figure is totally removed from reality. ... It is a known fact that modelling, more often than not, can lead to overestimation and on [a] few occasions, these estimates may stretch to the limits of absurdity.”

The government went on to assert: “The data released by the [CRS] report-2020... was shared with WHO for preparation of excess mortality report. Despite communicating this data to WHO for supporting their publication, WHO for reasons best known to them conveniently chose to ignore the available data submitted by India and published the excess mortality estimates for which the methodology, source of data, and the outcomes has been consistently questioned by India.”

It is, of course, patently absurd to expect that information contained in a 140-page report and shared with the world body just two days before the release of a global report covering 194 countries—which, in fact, was ready to be released in April—could be incorporated just like that without scrutiny and analysis. More pertinently, India shared data only of 2020 and not of the entire two-year period for which WHO performed the study. However, in keeping with its stated policy of updating the report when fresh data become available, WHO has stated that it had not fully examined the new data India provided. It also said that it might add a disclaimer to the report highlighting the ongoing dialogue with India.

Not Tier II

One of India’s key objections to WHO’s methodology has been that India was placed in Tier II (which equates it to countries like war-torn Iraq) when it has a robust system of registering births and deaths and has evolved a sound definition for a COVID-19 associated death. It has questioned WHO’s recourse to modelling when India systematically registers deaths and their causes from the lowest administrative unit of a village to the highest in urban centres.

Steve MacFeely, director of Data and Analytics at WHO, however, stated in response that the classification represented both the quality of data in a country and its willingness to share them. Countries may have quality data, and it is their prerogative to make them available but they would be classified accordingly. In response to an article in The New York Times , the Health Ministry said in a release: “India has asserted if the model is accurate and reliable, it should be authenticated by running it for all Tier I countries.” According to Jha, WHO had indeed done this for Italy, which, according to him, has an extremely good death registration system. The WHO exercise found that the methodology adopted for Tier II countries reproduced Italy’s excess death data fairly accurately.

(The robust system of registration that India has referred to is the CRS under which State-level births and deaths get registered on a monthly basis and are compiled to form a national-level CRS report that is released annually. Since births/deaths are recorded under the CRS only if citizens provide the relevant information to the system, not all of them get recorded. Registration is, therefore, particularly weak in rural areas. CRS data, therefore, give a lower bound to the “all-cause mortality” data over a given period. It is only the Sample Registration Survey (SRS), a door-to-door survey, that provides the true figure of births and deaths in the country.)

The WHO’s exercise of estimating excess mortality, according to reports, began in February 2021. According to WHO, it conducted a country consultation with member states in October-November 2021 to review the draft estimates, data sources and methods. Countries were also asked to provide advice on primary data sources that might not have been previously reported or used and share inputs or additional feedback. Until May 3, 2022, India failed to share any primary data with WHO. India has apparently not shared its mortality data, as registered by the CRS and the SRS, with WHO since 2007, Jha said to theWire.in. According to an unnamed WHO official, as quoted by IndiaSpend, WHO had explained its methodology to Indian officials multiple times since December 2021 but to no avail. MacFeely told the media in April: “We know that India is not happy, and the final publication will mention that. But for all countries, we need to put these numbers out now.”

Last year, when it became apparent that COVID-19 deaths were being undercounted, researchers and a few media publications began to publish monthly death data from either the State-level CRS portals (which record all deaths in a district for every month) or from mortality information accessed through the Right to Information (RTI) Act. Data for as many 18 States, accounting for nearly 80 per cent of the country’s population, thus became available to researchers. In the absence of any official communication in this regard from the government, WHO statistically extrapolated the data of 18 States using modelling techniques to obtain mortality data for the entire country.

The government also questioned WHO’s use of CRS data obtained indirectly from media publications and through the RTI as the basis for its estimates even though these are from government sources. The last SRS data available are of 2019 and the CRS data, are of 2020. The irony is that the CRS data not only corroborate WHO’s estimates but, being a lower bound, imply that WHO estimates are actually conservative.

Consistent with WHO

There have been many independent published studies and (non-previewed) e-prints on web repositories on the excess deaths in India. These are all consistent with WHO’s estimate. These research groups have used data from the CRS, different surveys and other government data sources such as the Health Management Information System, and the range within which their findings fall is three to six million, very similar to the bounds set by WHO for its excess mortality data for India.

Even though the CRS report itself does not make the extraordinary and unbelievable claim, as the Health Ministry did in its May 5 rebuttal to WHO, that death registrations in the country were at the level of 99.9 per cent in 2020, the Ministry has said that the figure was “internally shared”. In fact, the report itself cites the problems different States faced in registration of births and deaths due to the pandemic.

For instance, Telangana reported that COVID-19 was one of the major causes of under-reporting. Further, it pointed to the lack of staff in municipalities and panchayats and the lack of travelling allowance for district-level statistical personnel. Similarly, Tripura raised issues relating to online registration and shortage of staff and machinery and network issues at every level in the State. Uttar Pradesh, too, highlighted problems faced by the general public in reporting due to the pandemic. In such a scenario, to claim 99.9 per cent registration of deaths in 2020 is “pure fantasy”, as Jha pointed out to theWire.in.

But, more significantly, according to the report of the fifth round of the National Family Health Survey (NFHS-5), which was also released on May 5, the average percentage of deaths registered with the CRS in the three years preceding the survey period of 2019-21 was only 70.8.

These were normal years. And if the national average level of death registration was only about 71 per cent (rural: 66 per cent; urban: 83 per cent) in normal years, it is bound to have dropped significantly during the pandemic. In light of this, even the government figures for 2018 and 2019 (at 84.6 per cent and 92 per cent respectively) would seem fictitious. Even if one takes the CRS death figure of 8.1 million (see the table) for 2020 to arrive at excess deaths during the year, one must subtract it from the average of at least the previous two years (about 7.3 million). Subtracting it from the 2019 figure alone is “cherry-picking” and being disingenuous, as Jha pointed out in his theWire.in interview. And if one does the mathematics correctly, one gets excess deaths during 2020 to be 0.8 million, as compared to WHO’s figure of 0.83 million, and not 0.47 million as the government statement has claimed. Voila !

If one actually applied the historic corrective factor of about 71 per cent death registration as per the NFHS-5 report (actually it would have been less due to the pandemic), the figure of excess deaths in 2020 jumps to about 1.2 million. And, given the extended lockdowns and other COVID-related restrictions across the country in 2020, it stands to reason that a large fraction of these excess deaths would be attributable to the pandemic. These estimates (albeit crude) should be compared with the official figure of 148,738 for COVID-19 deaths in 2020 (up to December 31, 2021), which already implies a factor of about 5×. The government’s own figure of excess deaths in 2020 (of 0.47 million) is a factor 3× the official COVID-19 death toll in 2020.

So, as Jha pointed out in the interview, the CRS release corroborates the WHO numbers at least for 2020. “If the Indian government wanted to get to the truth of what is going on, then release the CRS data by week so you can look exactly during the peak weeks to determine if there is an excess,” Jha said. While the annual CRS data for 2021 is yet to be released, the government could have easily shared the monthly State-level data with WHO if it was not happy with the organisation using data from “websites and media sources”. But it has not. The government owes the world this information so that the scientific community can understand the virus and the disease better and be prepared for the next pandemic. As Jha wrote in his editorial in The Lancet : “Counting the global COVID-19 dead will help the living.”