‘May 2021 is as horrific as November 1918, during the influenza pandemic’
The Age of Pandemics author Chinmay Tumbe, who studied death registrations in five states, found they exceed by at least 10 times the official toll from the second wave of Covid-19. Tumbe, a professor at the Indian Institute of Management Ahmedabad, spoke to Sunetra Choudhury about his study amid reports of underreporting of Covid-19 deaths. Edited excerpts:
What did you find when you looked at the data from Karnataka, Gujarat, Andhra Pradesh, Tamil Nadu, and Madhya Pradesh?
If you just look at four states...Madhya Pradesh, Andhra Pradesh, Karnataka, and Tamil Nadu...from January 1 to roughly May 31, (they) cumulatively and collectively recorded over half a million excess deaths. More than 500,000 excess deaths as compared to...46,000 deaths, which were reported of Covid. Now, that is a large difference if you multiply 46,000 to 10, you get 460,000. So, this is a factor of more than 10. Now, not all excess deaths need to be of Covid... but if you plot a monthly or annual distribution chart of deaths in India for the last 50 years, you see very little variation from month to month in general in non-pandemic times. One of the things that pandemics do is precisely increase death substantially. That is why pandemics are so dangerous. So, if you look at any state government chart, you will see very little variation across months, even across years. Obviously, as the population is growing, fewer deaths will grow which poses some trend line.
Now when you get numbers from Madhya Pradesh or Andhra Pradesh, where you see more deaths in May. ...May 2021 is as horrific as November 1918, hundred years back during the influenza pandemic...because I’ve looked at those numbers as well...that’s precisely the kind of spikes that you’ve seen in the data when you see for the last 20 years...months are reporting the same death rates, and then suddenly you see three or five times more.
So, it’s clearly the signature of the pandemic... these 500,000 excess deaths...are not accidents. They’re definitely related to the pandemic. It could be, of course, that some of these deaths are of people who got a heart attack and could not get to the hospital in time because there’s a lockdown or the hospitals were full of other patients. So, there’s obviously some diversion of resources that have claimed deaths. But there’s no doubt that the bulk of this is Covid, and especially because of massive deaths in rural areas on a per capita basis.
Andhra Pradesh and Madhya Pradesh are much more rural than Karnataka and Tamil Nadu. You’re seeing a much higher excess deaths per capita ratio. What this also tells you is because we are counting Covid dead only. ...a lot of people who simply didn’t have testing facilities can never be counted as a Covid death, even if they actually got Covid. So, this is an important point to keep in mind and that is why excess deaths is a very powerful metric...it’s a very horrifying statistic because these four states comprise 300 million people. That’s about the size of the US.
And you’re seeing deaths which are about the same as were reported for the US, which is about half a million deaths. So, all this while people say, you know, India’s death numbers are not very high, and you can discount it because on a per capita basis they are very low. Unfortunately, now what we’re finding is that on a per capita basis, India ranks very badly for the data that is available.
The thing that strikes immediately is that 10 times figure. Now I know that you’ve just studied four or five states. But is it, if we extrapolate, that means that at the moment 370,000 death figures could be actually 10 times more?
I think you have to break down wave one and wave two. So, what we should be doing is ideally look at 2020 and 2021 very differently. So that 370,000 figure you said is basically about, let us say, 210,000 for this year. Right. And 150,000 or 170,000 deaths as reported for the first year. Now there is some work that suggests that the underreporting factors in the first year were very low. You know, because these underreporting factors themselves vary from state to state. ...It will also vary from wave to wave. So, I don’t think we should be applying this factor of 10 to 370,000.
Could you apply it to, say 200,000 for wave two? I’m focusing right now on wave two because that is where really the pandemic hit us very hard. And so, if we just focus on wave two, how plausible is 10 times as an all-India factor?
We still have data from Uttar Pradesh, Rajasthan, Chhattisgarh, Jharkhand, and so many states to come, where the underreporting factors are, in my view, definitely going to be substantially more than what we’re seeing for Tamil Nadu and Karnataka. So, while obviously, one has to wait for the eventual data, clearly, this is going to be now a lower bound for an all-India factor and that is why it’s very horrific. So, I wouldn’t multiply 10 with 370,000 of the overall deaths so far. But if you want to multiply 10 with, say, 180 or 200,000 deaths of wave two, because my analysis is only for wave two, that is 2021. That’s the kind of number you see.
So, whereas earlier, I was placing a lower bound estimate of 1,000,000 excess deaths based on just the stuff I’ve done on Gujarat, which is only 5% of India. Now we have more data for a longer period, so I’m revising that lower bound from 1 to 1.5 million.
So how are you calculating this? Just to be sure, because, you know, when the Economist came out with its data and said the other factor was five to seven times the government suggested data, the government immediately pushed back saying, no, it’s wrong. Would you like to explain your particular study, which has looked at particular states, what is the data that you are going by yet?
The data comes from the civil registration system. This is where the deaths are registered. Think of it as when somebody dies, you get a death certificate issued. So, all those death certificates are counted. And so, we have a process called death registration. Obviously, this varies from state to state. In some states, virtually every death is registered while in some states, it is less. But the fact is that this is one of the most scientific ways to calculate deaths to give some sense of how much the pandemic really hit. I mean, you can’t just think that nobody died of Covid in India if you supposedly did not have a testing mechanism. Think of the 1918 pandemic. They didn’t have any testing mechanism. They in fact used more common sense and said if a person died of fever in October-November 1918, it must be a very high chance that it was influenza which we are not doing.
You’re saying you have to test positive to be counted as Covid dead. And then there’s an ICMR protocol on how you deal with comorbidities, which is one classically many states are using for underreporting and so on.
But my point is that a lot of India does not have access to testing facilities, especially in rural India. So, if you go with a strict approach to ICMR protocol, we are missing a huge part of the deaths. We basically have to assume that if deaths are going up by five times in May, it means that most of it is likely to be Covid, and it’s quite likely that none of those people actually tested for it because the testing facilities were not there in the first place.
So, this data that we are using comes from death certificates and I would say, it is probably one of the most scientific methods to do.
There is another method... the intercensal growth rate method, which is using census data. And so, you might ask, but look, the bodies floating in the Ganga, they never left a death certificate. Even those deaths could be estimated to a certain extent. When the next census is conducted, we can find the shortfall in certain age cohorts.
If you compare 2001 to the next census, you can find out how many people are missing in villages and so on. And that is why that is an additional method to calculate deaths. So, for example, I did this for the 1918 pandemic, the first estimate was six million, and this was based on something like fever, which was actually a better way to understand the severity than now. But that six million figure was revised again by various researchers and now the number is 20 million.
So, the underreporting factor at that time was three, which is lower than what it is now because people use a very simple idea of fever, just a fever. Today, we have a very stringent norm to be classified as Covid dead; you have to be tested positive then you should have died because of Covid. So, it’s a very stringent claim. I think the government needs to just ensure that all the statistics which researchers have been asking the government to release for more than one and a half years now, should be open.
If the government does want the scientific analysis of this, they just have to make all the registration statistics public, which is already there online. They just have to literally sign a document saying just go and make it public, and then it’s up to researchers to make the decisions.
Tamil Nadu did not really have many excess deaths and it wasn’t really that much of a difference. Right? Does that show that perhaps they had a better way of counting their deaths and ascribing it to Covid than perhaps how the other states were doing it?
Yeah, I think the difference in these four states is that if you look at Andhra Pradesh and Madhya Pradesh, they have an urbanisation rate of 30%. If you look at Karnataka and Tamil Nadu, they have an urbanisation rate of 50%.
Now, here is one speculation and one theory as to the numbers we’re seeing, the more urban the state, the lower the underreporting factor you’re likely to see. This is because there are going to be better testing facilities. People are going to be saved more because there are more hospitals, so that’s a simple reason.
In wave one, the virus did not go to rural areas. Mumbai was a big problem last year. Thus, we can characterise wave one as predominantly urban-centric. But wave two has moved to rural areas where there are simply no testing facilities. And so that is why what we have observed, and that is why researchers have been trying for the last year asking to make the registration statistics visible. What is visible today are the Covid numbers that are being tested.
But that’s a huge urban bias towards places that have testing facilities. So, imagine a state which has no testing available, but that doesn’t mean that nobody has died of Covid.
Compared to the rest of the world, where do we stand in the underreporting?
Oh, very bad. In wave two very, very bad, again. As I said, you have to separate out waves one and two, but there have been estimates for the US, for the richer countries, which, by the way, have been publishing these registration statistics on a very regular basis. Those studies are putting the underreporting factors from 1.5 to 2. That’s the kind of range. There’s a study that says that in wave one in India, the underreporting factor could be 2.2. In Mumbai, there are studies for 2020 where many places have a factor excess of 1.6. So that’s the kind of underreporting you would get in, you know, in normal times, in urban areas with good medical facilities, good testing facilities. These underreporting factors, however, are still very large. Just for these four states, if you’re getting a factor of 11 cumulatively, that’s one of the highest in the world, I would imagine.
And explain to people who are just saying, you know, why does it matter so much? I mean, people are dying. They’re going to die anyway. Why is it important to know exactly how it stands or know the exact number of people who died? What would you say to them?
There are two reasons. One is, if we know the places which have been hit very hard, that’s precisely the places where we need to invest more if at all there is a third wave coming. Because now what we need to do is a plan for the next wave. So, we need to firstly get a sense of which are the districts of India which were caught up and that’s what these numbers are going to tell us, not the official numbers. The official numbers tell you districts, you know, with virtually zero deaths, but huge excess mortality. So, one it helps you in better planning for the next wave or two. Second thing is that, If we make these numbers publicly available regularly, we can actually start identifying when the wave is going to pick up because then we have an addition. We just don’t have the numbers. We have just this excess death number to see if it’s going up in precisely those places that don’t have the facilities. We can then start diverting resources and give a simple example, we can use the official Covid numbers for that oxygen allocation.
Now, just think of it as saying that actually more oxygen was technically required in Madhya Pradesh, it has huge relevance. This is not just an academic matter. It’s a matter which can save lives and that’s why we need to publish these statistics as soon as possible.
Apart from more transparency of publishing in all the deaths that happen, is there anything else that we need for better covid management?
I think, broadly put, there are maybe two or three things that you can do better than what we are doing. One is in terms of investing in tracking new mutations. I think why wave two was so different was primarily because of this strain and we do not track the mutations very well. This is also a lesson from past pandemics. So, investment in understanding the strain varieties that are in the public. Increasing testing capacity, even more, would be the second while the third is to ramp up the facilities. Hopefully, there’s no third wave, but if it is, you know, you don’t want to be in the same situation that we faced in the second wave. What people are now saying is, you know, the first wave, nobody expected and now people are saying again in the second wave that nobody expected. Then, I don’t know which way people are going to expect. The idea should be to now ramp up as much as possible, before a third wave if at all it materializes.