The health and family welfare ministry issued a statement on May 5 disputing the World Health Organisation’s estimate of Covid deaths in India. The WHO had said that India lost 47.4 lakh people to Covid in 2020 and 2021, several times the Indian government’s tally of 5.24 lakh. The ministry claimed that the WHO estimate was “totally removed from reality” and way out of agreement with the “extremely robust” Civil Registration System, or CRS, which is used to record vital statistics such as births and deaths in the country.

But the latest CRS report, for 2020, released two days before the WHO came out with its estimate, itself casts questions about the health ministry’s statement. It claims that India recorded 99.99% of all deaths in 2020, totalling 81.15 lakh. In 2019, the CRS had registered only about 92% of all deaths, totalling 76.41 lakh. How was this percentage calculated? “The level of registration is defined as the percentage of registered births and deaths to the births and deaths estimated through SRS,” the 2019 report explains, referring to the Sample Registration System.

The level of registration, or LOR, equals the number of events registered during the year divided by the number of estimated events for the year multiplied by 100.

The denominator – the “number of estimated events for the year” – is taken from a baseline survey called the Sample Registration System. The last SRS survey, done in 2019 and covering 9,000 rural and urban units, estimated 83.01 lakh deaths.

Using the above formula, we get 76.41 lakh/83.01 lakh x 100 = 92%.

The 2020 CRS report, in contrast, is silent on the level of registration as the SRS report for that year hasn’t been released yet. Still, the health ministry’s statement pegs the number of expected deaths for 2020 at 81.20 lakh, just marginally more than 81.15 lakh actual registrations, returning a rate of 99.99%.

Newslaundry contacted health ministry officials asking how they arrived at these numbers but didn’t receive a response. An unnamed official from the Registrar General of India, however, speculated to the Print that the 99.99% figure could have been calculated using the 2019 SRS report death rate and the projected 2020 population.

The latest National Family Health Survey notes that only 71% of the deaths in India were registered in 2019-21, meaning that three out 10 deaths weren’t fed into the government system. The National Family Health Survey, which provides vital data on health, education and other such indicators, covered 6.36 lakh households in 2019-21.

“The 99% doesn’t look very sound,” said PC Mohan, chairperson of the Kerala Statistical Commission. “First, the share of non-institutional deaths in the total registered deaths is around 70%. For example, in far-off areas it may be difficult for people to get deaths registered. Second, every death registered by families may not be fed into the government system due to technological issues.”

Indeed, the CRS report acknowledges that the death registration systems of many states are hobbled by the lack of adequate staff. Another problem is variations across states. In 2019, as many as 19 states and union territories reported 100% registration while the seven big states of Uttar Pradesh, Madhya Pradesh, Bihar, Jharkhand, Chhattisgarh, Assam, Jammu and Kashmir were below the national level of 92%.

Sayeed Unisa, who heads the Department of Biostatistics and Epidemiology at the International Institute for Population Sciences, Mumbai, said the National Family Health Survey’s rate of death registrations could be an undercount “due to lack of knowledge of registration” while the CRS figures “seem an overcount”. “Some say the registration level could be around 80-85%. In any projection, we cannot say with certainty that it is 100% correct,” she added. “That is why scientists give a lower limit and an upper limit for their projections.”

If we take a mid-range between the health survey’s 71% and the health ministry’s 99.99%, we get a rate of 85%. Which means that India may have witnessed 95.47 lakh deaths, from all causes, in 2020, about 14 lakh more than the government’s tally for a pandemic-hit year.

Prabhat Jha, a Canadian epidemiologist, told the Wire that India sees about one crore total deaths every year according to the WHO, far higher than what CRS reports acknowledge. In 2020, for example, the WHO estimated 8.3 lakh Covid deaths as against the government figure of 1.49 lakh.

The health ministry’s May 5 statement strongly criticised the WHO methodology of calculating the deaths. Yet, its silence on how the Indian government arrived at the 99.99% death registration level raises questions. And we will have to wait for the SRS report for 2020 to find answers.