What is India's 'True' COVID Toll? 3-8 Times, Says Murad Banaji

5 min read

"I think the worse case scenario of 4.2 million deaths is on the pessimistic side, but because our own data collection has been so poor, I won't rule it out entirely," says Dr Murad Banaji.

A recent The New York Times article about India's 'true' COVID toll set up four scenarios, based on multiple factors like India's own national Sero Surveys, international data and predictive modelling. According to the article the best-case scenario is 600,000 deaths and the worst-case scenario is 4.2 million deaths.

The piece has since been called 'baseless' by the Central government in a strong rejoinder, and some epidemiologists in India have also pointed out flaws with the methodology.

FIT speaks with Dr Murad Banaji, a mathematician from Middlesex University in the UK, who has been mapping India's undercounting of COVID cases and deaths.


How do researchers like yourself come up with predictive models when faced with data that is not often 'true'?

Dr Murad Banaji: When it comes to estimating the scale of cases and death undercounting, there isn't any one method we look at. The sensible thing to do is to look at where you have data with higher integrity, for e.g., if you have cities that release excess all-cause mortality data, or if it is obtained by news organisations as it was done in Gujarat, what you can try to do is ask based on this data, what is the fatality rate that is occurring in these parts of the country. Based on that, if you have similar fatality rates occurring across the country, then what number of deaths would you expect?

In a sense you try and use those places that have good data, and not extrapolate to directly, for e.g. if undercounting was by a factor of 2 in this area, so it would be the same elsewhere, that's the wrong approach. What you say is, what is the fatality rate from COVID from what we can tell, and if that was the case across the country, what number of deaths can we expect. When you do that, you can still get a very wide range of answers, but at least you get some sense of the scale at which deaths might be undercounted.


What does your scale indicate? At what scale is India undercounting the COVID deaths?

Dr Murad Banaji: Based on 2020 data, the scale lies between 3-8 times the reported toll. This is my broad prediction, and it has a big range there. If you were to press me on the actual number, I would say it would be 5 times the total deaths reported. This is based on using available data from Mumbai and internationally.

When it comes to this wave, we don't have enough real-time data so we don't have enough to work with. But I would suggest that things are at least as bad this year as they were last year. Testing may have improved and reached some rural areas, but you also have a surge at such a large scale that it is very likely deaths are being missed as well.

Given that a lot of deaths happened outside the healthcare system, given that our healthcare systems were overwhelmed and testing had reached its limit, would you say the deaths in this surge are more than last year?

Dr Murad Banaji: In some places like Mumbai, the testing was much higher during this wave than last year, on the other hand, when it comes to rural Bihar or UP, I don't know if testing has improved enough or if testing teams are reaching these villages. But from news reports, what's coming out is that you have an epidemic, the disease is spreading through a cluster of villages, you have a large number of deaths, and subsequently, the official teams are reaching these villages doing some testing, providing care, so quite often the testing and medical attention is following the epidemic. And from these reports it's very clear that very few of those who died were actually tested before they died, so the chances of them making it to any official figures is very slim.

So what you do have in this wave is a lot of unavoidable deaths outside the health system, you have disease spreading in areas that will likely not make official figures. On top of that, you have obfuscation like we have seen in Gujarat, where deaths are occurring in people who have probably tested positive, but haven't made it to official figures because they are being classified as something else, for eg death by comorbidities.


Recently the Jharkhand Chief Minister Hemant Soren has said that his state wants to do a death audit. Given political will, how can states like Jharkhand go about rectifying the data?

Dr Murad Banaji: When you have a chief minister who says that we actually, genuinely want to rectify the data, that's excellent. First, you would need to do a survey that covers rural and urban areas. And similar to other mortality surveys, you ask people if a death occurred in the family and the circumstances of these deaths, and based on these surveys you piece together the extant of the mortality. Mortality surveying is a crucial thing that the state governments can do if they are genuinely keen to understand the scale of the pandemic.

In light of the NYT article, there is concern when figures like 4.2 million deaths are mentioned. I want to ask you what is the purpose of these predictions or models? Do they just spread panic or do they help in policy planning?

Dr Murad Banaji: The very upper limits they have in the NYT article of around 4 million deaths are on the pessimistic side, so for me that really is a worse-case scenario. I am not sure where they come from, but I also won't say that it's impossible, because the data we have is so poor at the moment, that it would be foolish to rule out that you have that many deaths. But I do think it is on the pessimistic side.

Why would we want to give our honest best guess on the number of deaths? Because it does two things, it helps you understand what might happen next, and plan for it, whether it is through direct modelling in the mathematical sense or through narrative modelling on possible scenarios which could unfold. The other aspect is that if people are modelling and making estimates, it spurs governments towards greater data transparency and greater data collection.

(At The Quint, we are answerable only to our audience. Play an active role in shaping our journalism by becoming a member. Because the truth is worth it.)

Read Latest News and Breaking News at The Quint, browse for more from fit and coronavirus

Topics:  Covid Death Toll 

Speaking truth to power requires allies like you.
Become a Member
3 months
12 months
12 months
Check Member Benefits
Read More