Covid-19: What you need to know today
How many dead are there, really? That’s actually the first line of a column I wrote on May 2, putting forth a model for calculating a range for the number of deaths India has seen from Covid-19, but more on that anon. The question is worth repeating because of recent work by data journalist Rukmini S in Dainik Bhaskar and Scroll on excess deaths per million using death data from the states of Madhya Pradesh, Andhra Pradesh, and Tamil Nadu. This data isn’t publicly available, and not all states share it (in fact, most do not). According to the data, Madhya Pradesh under-reported deaths between January 1 and May 31 by a factor of 42, Andhra Pradesh by a factor of 34, and Tamil Nadu by a factor of eight. The number has been arrived at by comparing historical and current data.
These death numbers have been put together by the states themselves, and usually, when such data is revised, it is only upwards. Sure, death reporting has always been patchy across India, although it has improved over the years – but it is unlikely to have suddenly seen a radical improvement in a year that actually made such reporting even more difficult. Anecdotal evidence suggests that some states keen to show their ability in managing the pandemic have under-reported deaths by simply ascribing them to other causes – pneumonia and typhoid are both beneficiaries, I am told – an approach that involves the use of all-cause deaths to calculate deaths has emerged, around the world, as the best way to capture the actual toll of the pandemic. It’s an approach that has been used to good effect by data investigators in other countries. This is actually different from the reconciliation of death numbers carried out in recent weeks by states such as Maharashtra and Bihar, which, as the term suggests, is more of an exercise in factoring in backlogs, updating records, and (in Bihar’s case) accounting for deaths outside the public hospital system.
The reports on the excess deaths in Madhya Pradesh, Andhra Pradesh, and Tamil Nadu, new research, and the fact that health systems in at least some parts of the country were completely (or partially) overwhelmed during the second wave together suggest that it makes sense to revisit some of the assumptions of the May 2 column.
In that column, I accounted for the fact that the number of actual infections could be 10-15 times the reported number. But recent research by scientists at Public Health England has shown that the Delta variant of the Sars-cov-2 virus that was first sequenced in India (B.1.617.2 is how it is otherwise known) is around 64% more infective than the original Wuhan virus. And a study by Delhi’s Institute of Genomics and Integrative Biology shows that Delhi’s second wave (in April and May) was caused by the Delta variant. It’s entirely possible – we do not sequence enough viral genomes to say for sure – that all of India’s was.
That could well mean that the actual infections, at least in those two months, may have been even higher – perhaps (and this is a straight mathematical extrapolation that may not be accurate) 16-24 times the reported number.
Finally, there’s enough reporting to show that India’s Covid-19 fatalities in April and May increased on account of inadequate access to hospitals, medicines, or oxygen. The deaths were also inflated by the emergence of mucor. In the May 2 column, I assumed a range for the infection fatality rate (proportion of those dying to those infected), 0.115% to 0.23% (the upper bound was based on work carried out by Stanford’s John PA Ioannidis across 51 regions, and by Middlesex University’s Murad Banaji, for Mumbai; the lower bound was to factor in advances in Covid-19 treatment since the start of the pandemic). It is entirely possible that in April and May, the infection fatality rate in India may have been higher. How high is anybody’s guess, but the math of its impact is clear – a higher infection fatality rate on an elevated infected base (on account of a more infectious variant) is certain to cause a spike in deaths.
Over the weekend, the Indian government responded to an article in The Economist (without naming the publication) that said India’s reported death toll was “a small fraction of the real figure” saying there was no way there were “five to seven times excess deaths”. As Rukmini S’s analysis shows, at least in a few states, and for a few months, it was that and more. We need a nationwide comparison of deaths from all causes in 2020 and thus far in 2021 with the historical average.
Research has shown that the Delta variant of the Sars-cov-2 virus that was first sequenced in India is around 64% more infective than the original Wuhan virus. And a study by Delhi’s Institute of Genomics and Integrative Biology shows that Delhi’s second wave (in April and May) was caused by the Delta variant.
According to data, Madhya Pradesh under-reported deaths between January 1 and May 31 by a factor of 42, Andhra Pradesh by a factor of 34, and Tamil Nadu by a factor of eight.
Scan the code to read about a model to calculate the range of death numbers