Talking to The Dead
A shocking and little known fact: only 3% of individuals that die worldwide have an official cause of death certificate. More specifically, in India, a country of over a billion people, 80% of deaths take place outside the healthcare system.
Indian-Canadian epidemiologist Prabhat Jha saw this as a serious problem and decided to fix it. In 1998 he initiated a revolutionary project, the Million Death Study, where he planned to create a cause of death index which would effectively verify data on how these millions of people had passed.
An ambitious project by any measure, it has covered the length and breadth of India, with many volunteers and health care officials carrying out verbal autopsies in 2.3 million homes, on 14 million people. This is a project which has and will change global health policy for years to come.
Prabhat worked his way through the ranks of many eminent institutions such as the World Bank and the World Health Organization, ironically turning a lot of their own trusted and tried methods on their head. Prabhat highlights of the importance of counting the deceased, but also reminds us that they are more than just a statistic and deserve to be remembered.
I want to start by running by a couple of stats by you about India’s healthcare system, where you mainly operate. 1.2 percent of India’s GDP is spent on health, compared to around 5 or 6 percent for most countries and roughly 70 percent of the population still live in rural areas without easy access to hospitals and clinics.
Could you perhaps give us an insight into why India still lags behind much of the world in terms of development?
Well I think India is exhibiting a combination of good and bad news. The good news is that there’s been remarkable improvements in premature mortality. In 1970 most Indians wouldn’t expect to reach age seventy and the probability of death in childhood was at around 20 percent. Those numbers have now improved substantially. Overall probability of death is down to about 6 percent in childhood. So there’s been an extraordinary decline in child mortality.
But there are still big gaps. for instance, girls in central India, those poor states, have five times the pneumonia mortality rates than boys in South India and similarly for diarrhoea, the number of deaths are about four times greater in Central India that in west India.
India is an amazing laboratory of understanding major patterns of diseases and their risk factors. And it’s not the case that Indians are so genetically different. So if genes aren’t explaining this extraordinary variation then what is? Well we’ve identified some existing factors as being particularly important, and for adult health the top of the list is smoking.
Smoking causes about 1 million deaths in India per year and that number is probably rising as they switch from the local cheap cigarettes, the Bidi, to the western manufactured cigarette. But some other factors we simply don’t know and one of the reasons for our research is to point out that there are intermediate predictors of risk and other mechanisms that await discovery.
Your work is primarily centred around your project the Million Deaths Study and the verbal autopsy approach, can you outline what it’s about?
So most deaths in India occur in rural areas and the body is cremated or buried without any certification. In the absence of that certification you need an alternative system. What the Indian government has been doing since 1971 is, after every census they split India into a million small areas and randomly select a few thousand of those and then follow births and deaths in the areas for ten years. That’s called the sample registration system. What we’ve done is trained the 800 government staff who do that field work to collect a cause of death by filling out simple forms.
They write a half page narrative in the local languages, simply noting the key symptoms. Then we convert those to electronic records. We have an online system that has 400 physicians from all over the world. The physicians basically have an inbox, they get a record and they code that record and add their key words to it. Each record is coded by two doctors independently. If those two doctors disagree about a diagnosis then a senior doctor adjudicates. That system works very well because doctors aren’t afraid of much but they are afraid of having other doctors check their work.
Since 1997 we’ve covered 2.3 million homes and 14 million people. Eventually we’ll get more than a million deaths recorded of which 800,000 will have a detailed verbal autopsy.
Can you tell us about some of the major findings and surprises from this study?
Sure, our research has been very central to having the Indian government take tobacco more seriously and raising taxes on cigarettes in response to our evidence on tobacco hazards. We’ve shown that in 2005 India only had about 100,000 HIV deaths, a quarter of what was predicted. As a result of this Indian spending on HIV went down appropriately.
We also pointed out that the number of malaria deaths were far greater than previously predicted and that’s led to some states having a little more spending on malaria, although in my view not enough. So the role of the Million Death Study is very much to give data that informs the priorities. It gives a path for the Indian government to know if they’re making progress or not.
Another example is our estimate in 2005 that there were 50,000 snake bite deaths. This was the same as was estimated worldwide. So there’s a dramatic global underestimation. But we also pointed out that snake bite deaths occur just in a handful of areas and if you align the anti-venom availability to those areas you might be able to reduce snake bite deaths considerably. So all causes of death are possible to study if you have a really simple system like the Million Deaths Study
"There’s about 56 million deaths worldwide per year and most of those are in lower income countries...without medical attention. So what can we do? I think the solution is the Million Deaths Study."
It obviously seems like we can learn from this study worldwide. I know this is very hypothetical but if you were to consult the world how would you do that? As a global health expert, how would you apply the knowledge gained from developing nations and make structural changes on a global scale?
You’ve hit upon a fundamental problem. We don’t know how to count the dead worldwide. The solution from the west, which is that everyone dies in hospital, will take a hundred years to achieve. It will happen eventually. In China that transition from deaths at home to in hospital is already well underway.
But there’s about 56 million deaths worldwide per year and most of those are in lower income countries and most occur at home without medical attention. So what can we do? I think the solution is the Million Deaths Study. My diagnosis for the world’s problems on mortality is very simple. Collect a random sample of all the deaths in every major population, enumerated through verbal autopsy.
To really know what’s going on you need to study the dead and report the causes. And once you’ve filled in that really simple stuff there’s all sorts of things you can add to it. For example when you interview a household about a death, you can also find out if they availed the services. Did they use the local ARV treatment for AIDS? Did they use the TB clinic? Then you can ask, what is the relationship of air pollution to those deaths? We’ve just done this in India.
"To really know what’s going on you need to study the dead."
Do you think it’s possible to utilize the technology we have available to make even further developments?
Absolutely. Our application runs on a cell phone so you can collect all the information I just mentioned very easily. And hopefully we can marry that with another development China has already done very well which is a national ID number. If you have a number for every citizen then you can make these links across the data and really understand a lot about what’s happening. This will make it easier to identify places where different types of treatment should be made more available.
So the twenty-first century technologies enable these ideas to go further. But the core tool is a simple one. It involves a respectful conversation with a household member who has had someone die and that human conversation is really at the heart of the information you get.
We have to remember that human conversation, particularly with someone who’s had a loved one die can be extremely informative. And there’s a little bit of catharsis that happens there. When I was a medical student I went back to the village where my grandmother still lived and I asked her to tell me how my grandfather had died. From her description I could tell that he’d probably had a massive stroke. I was able to convey that to my mother and grandmother and it brought them a bit of closure.
Only later when I started doing the epidemiological work did I look back and realize that listening to my grandmother was something that had worked, and that it might work on a larger level.
Each of these deaths is a story. It’s that quote from Stalin, “A single death is a tragedy; a million deaths is a statistic.” So we’re capturing a million individual tragedies. The magic of the system is the conversation.
"We’ve shown that in 2005 India only had about 100,000 HIV deaths, a quarter of what was predicted. As a result of this Indian spending on HIV went down appropriately."
I’m really fascinated by some of the numbers you’ve uncovered. You’ve said you like to think of yourself as an epidemiological terrorist, blowing up assumptions.
Let’s take the 15,000 malaria deaths estimated by the World Health Organization versus the 200,000 malaria deaths you noted, or your 100,000 AIDS related deaths versus their 400,000. In light of some of these incredible findings what kind of response did you have from the World Health Organization?
With the malaria numbers there was a push back from the World Health Organization, I think largely because they had already made the statement of having reduced malaria deaths and so this was kind of an inconvenient truth. Now what has happened is after the initial pushback and the disagreement, the Indian government did convene a task force to try to look at the estimates and they also came to the conclusion that the WHO approach wasn’t working. And if you think about it it’s a very logical problem. Malaria isn’t just a disease you can treat, you can cure it.
Now they may well ask how we know it was malaria that killed these people rather than something else and that’s a fair criticism. When I first got the numbers from malaria I thought we must have something wrong here, that this must be some other disease. But the more we looked the more we were able to show that it was very likely malaria.
The cases occurred where malaria is often reported, it had a seasonality pattern, the other common diseases like dengue and meningitis didn’t occur in those areas. The WHO is a reasonable organization and now our findings have in fact spurred their interest in doing more to measure the causes of deaths. And they’re actually trying to get other countries to adopt these kind of systems.
What have been the greatest obstacles or most frustrating elements in the work you’ve done?
The main frustrating challenge is really the lack of funding for sustained efforts like this. We were doing research over a ten year time period and research funding rarely goes beyond five years. We were very fortunate that we got grants that let us move this forward.
Anything really important in global health is never a matter of doing things in the next year or two. You have to think in terms of a decade. In the last ten years there have been extraordinary improvements worldwide but funders need to realize it takes time.
Not three years, not five years but a decade. And that’s where you see progress. And we’ve seen that here. India gave the world the “zero” and that was a huge contribution to mathematics. Here India is giving the world a simple way to count and the idea is that other countries will use these technologies much more widely.
Header image courtesy of Finn O’Hara – finnohara.com