In most developing countries, including India, the quality of data on the cause of death is very poor. This is because most people die in rural areas and at home, a medically certified death in a hospital is rare, and medical certificates more often than not are filled incorrectly.

Dr Prabhat Jha, founding director of the Centre for Global Health Research in Toronto, is one of the world’s leading experts on mortality in India. He is a professor of disease control and an expert advisor to the Bill and Melinda Gates Foundation as well as various governments.

The Million Death Study, which he led in collaboration with the Registrar General of India, uncovered new insights about death in India and changed our understanding of malaria, HIV/AIDS and suicide, among other causes of death.

In this interview, Dr Jha speaks to Rukmini S about Covid-19-related mortality trends in India and discusses probable strategies that can be adopted. It is part of a nightly podcast The Moving Curve on the coronavirus pandemic in India.

Given what we know about deaths in India – that most happen at home, that about 70% are registered, and then about 22% of those are medically certified – is there reason to suspect that our officially reported Covid-19 mortality is an underestimate?
Yes. In India, most deaths occur in the rural areas – about 80% of deaths occur in rural areas, and most deaths do not have a medical certificate at the time of death. What that does mean is any major cause of death in the country – whether routine like, say, cardiovascular deaths, or an epidemic such as Covid-19 – will be missed because simply the reporting systems are not complete. So it does suggest that the true extent of Covid-19 mortality in India remains unknown.

However, it is also reasonable to think that in some of the urban areas where the death reporting is more complete – Mumbai, for example – that the emerging data from India points to rather an interesting phenomenon that the growth in the epidemic has been slower than it was in high-income countries. But it’s still going up, and it might be the case that it’s early on in the epidemic still and that in subsequent weeks, it will accelerate.

So we do know some things about mortality in India. We don’t know the absolute levels of Covid-19 mortality deaths, but in some areas, we know the trends of mortality, with sufficient sense of the data to know that there is something different going on.

You have found that in cases where medical certification does happen, there seems to be a tendency among doctors to confuse the proximate cause of death with the underlying factors. Given the high overlap between comorbidities and Covid-19 mortality, is there reason to think this issue might get worse with Covid-19?
Yes. That’s a common problem with death certification around the world, which is that unless there is some quality control on what a physician writes on a death certificate, much of the time they will write things that don’t make much sense. And the most common problem is they confuse the mode of death with the cause of death. So, for example, they will put cardiac arrest – well, we all die of cardiac arrest. If you get hit by a bus, you die because blood is leaking out and then your heart stops. This problem has been remedied in high-income countries and in some urban areas in India by making sure that the underlying cause of death is not used except where it is supposed to be.

Is this going to be worse in the time of Covid-19? Well, it might be but then what you would see mostly is people dying of respiratory diseases or “respiratory failure”. That’s an argument for getting good death statistics or verbal autopsies such as done in the Million Death Study.

So one of the things that I have certainly recommended is that the Registrar General get the Sample Registration System deaths quickly reviewed for the last half of 2019 and the first few months of 2020 to try to understand: was there an increase in death, for example, from people dying of pneumonia or ill-defined pneumonia. Or, in fact, has there been no increase or, surprisingly we found out that some of the data from Mumbai and Ahmedabad is that their daily death counts are lower. That’s a puzzle and my belief is that mostly that’s driven by out-migration of people, particularly young men in those cities.

India hasn’t conducted a verbal autopsy-based Sample Registration System survey since 2013, though.
I believe that the 2015 data is now with the Registrar General. I think they should release that immediately but also prioritise collecting data for the current year. Another way would be to use mobile phone data to see how many young men from say, Mumbai, have migrated back in March and do a sample survey in the urban areas – Mumbai – but also in the top districts these young men have returned to.

Another strategy is that every district of India has coroners and doctors who are supposed to do autopsies for basically police investigation cases. For example, men that are alcoholics sometimes die in the streets. So a very simple strategy is that because they are there in every district of the country, swab the dead person’s throat and nose because they undergo an autopsy anyway, and from that you would get some prevalence in the unrelated dead – people who died from traffic injuries and so on and so forth – to see if they also had Covid-19.

Interestingly in South Africa, a similar strategy was done to look at HIV, and showed that something like 20% of the young men and women dying in road traffic injuries, completely unrelated to HIV, had in fact HIV infection. That was a very powerful study because it helped convince a then-very sceptical president that HIV was not a foreign problem in South Africa, but a domestic problem. So similarly, here I think getting some sense of how prevalent Covid-19 [is] in the dead unrelated to Covid-19 disease. There are a few strategies in the absence of complete registration and I think these would be some simple things to do.

And the last one, which is an obvious one, is that in the urban areas, they do have a Register of Deaths that counts the registered deaths. Now the registered deaths don’t give you much detail but reporting those weekly and comparing them – let’s say, what happened in each of the weeks of March versus what happened in March 2019 or 2018 – can give you a sense to say, well is there an increase in the overall deaths?

We are clearly seeing a huge disruption in health services across the country but particularly in rural areas, for everything from immunisation to cancer treatment. For the next few months, there is the issue of trying to accurately identify Covid-19 mortality. But after that, and for the next few years, how can we estimate the impact of this disruption on mortality?
You have to think about the epidemic in context. The dramatic gets attention, but the routine deaths from this disruption might well be more important. In Sierra Leone, the Ebola outbreak probably killed more children from lack of treatment for malaria than it did directly from the disease. So disruption of services is actually quite important and that will apply particularly to treatment of children or adults, particularly hospital treatment. We know the treatment saves lives if it is available. And if this is disrupted there will be an impact on mortality, an unfavourable impact.

Rukmini S is an independent data journalist based on Chennai. Her nightly mini-podcast is on Medium, Spotify and Apple Podcasts.