Skip to main content

A woman stands in front of a free medicine counter at a government hospital in Shahbad village in Baran district in the northwestern state of Rajasthan January 31, 2012.Adnan Abidi/Reuters

When Prabhat Jha makes his pitch to public-health officials in developing countries, they can find it counterintuitive, or even downright nutty: The Toronto epidemiologist argues that one of the best expenditures they can make on the well-being of their people is to spend a bit of their limited money on the ones who have already died.

Dr. Jha spearheads an 11-year-old initiative known as the Million Death Study – an effort to figure out how one million Indians, mostly poor, mostly rural, died – and to use that information to evaluate and possibly redirect health policy. Its findings have proved so valuable that it is now expanding across the developing world.

Globally, 60 million people die each year – 50 million of them in the developing countries – and yet a cause of death is known for only 10 million of these; the biggest gaps are in Asia and Africa. In India, for example, most people die at home, in rural areas, and are cremated or buried within a day or two. Less than half of deaths are registered at all, and those that are simply record the deceased's name and whether the death was violent.

It is important for individual families to know a cause of death, of course, but also for public policy: Given the absence of data, much of that policy in India and other countries has been, up until now, formulated based on estimates from United Nations agencies or research by donors about what kills people. And sometimes, they're just plain wrong, leading to health-care budgets all out of whack with how people actually die.

In the Million Death Study, causes of death have so far been identified for 350,000 people – and dramatically challenged the conventional wisdom on how Indians die, showing a far greater role, for example, for smoking, snake bite and suicide, and a far less significant one than was assumed for AIDS. Budgets are being rejigged accordingly.

"The dead matter because of the living – they help you figure out what actually people are dying from and those are the things that help you figure out what can be done," said Dr. Jha, who heads the Centre for Global Health Research at St. Michael's Hospital and the University of Toronto. He called the lack of accurate information on why people die a "critical constraint" to health spending, particularly for massive initiatives like the Millennium Development Goals programs on child and maternal mortality.

Shailaja Chandra, who until recently served as the senior civil servant in the Indian Health Ministry, says the study is having a critical impact on health policy. "If the village head man writes that a person has died of 'cold fever' in Hindi, it does not help people to get medicine which can actually cure malaria," she said. But hard data does.

Dr. Jha has recruited partners and is marshalling funds to expand the study to 10 million deaths, with data collection under way in Bangladesh and South Africa, and starting soon in Egypt.

The study began in 2001, when Dr. Jha, then with the World Health Organization, partnered with Jayant Banthia – then India's census commissioner and registrar general; the census produces remarkably vast and comprehensive data on this subject. The team trained 800 surveyors, who normally collected standard data on births and deaths, to also conduct a "verbal autopsy" and it then dispatched them to 6,671 districts chosen at random across the country.

It works like this: The surveyors visit each house and ask about anyone who died in the past six months. If the family volunteers a clear history of the illness and death, they record it, including details gleaned using a list of 12 questions about symptoms. "Even for those folks who don't know what happened to their loved one, we ask them 'yes or no' for each symptom; for any positive ones, the field staff probe more – then read it back – and when you do that the household can piece together what happened," Dr. Jha explained.

Data are recorded on sturdy, $200 Indian-made laptops and matched to GPS co-ordinates, then e-mailed to one of 250 trained doctors, scattered across India, who are paid 40 rupees (75 cents) per death; they use the symptoms and compare them to guidelines to establish a cause of death.

While this might sound a bit sketchy, Dr. Jha says it has proved surprisingly accurate – better even than he had hoped – in testing where results were compared against cancer registries and other known causes of death. It is less effective with the deaths of the elderly, whose cause of death may be opaque (and described in the community simply as "old age") but works well on younger adults and children. It works because death is a "big event" about which people remember many details, and because people are willing to talk, he said.

"There is magic in it," he said. "Most households have received little or no medical attention, even from the public system, and they have had a big loss, especially when the person died before old age, so they find it quite cathartic to talk to someone about it, and they like to be heard," he said. "We try to train the surveyors to be good listeners, with empathy, like a good doctor who takes your history at the hospital in Canada. The families appreciate the attention, even though we tell them, 'You won't benefit, but the community will, from understanding how people die.' "

Using data from the first 250,000 deaths catalogued, Dr. Jha and colleagues showed that smoking was killing one million Indians each year, far more than commonly thought; the evidence prompted government to accelerate the use of warning labels on tobacco products even as the powerful industry lobby was trying to kill smoking-control legislation. Next the study found that 100,000 people were dying of AIDS each year, only a quarter of what the United Nations was reporting for India – and this has led to a rethink on spending on HIV treatment, which is comparatively expensive. The study also found 50,000 snakebite deaths in India a year, which was the World Health Organization's worldwide total.

The latest evidence concerns malaria: Dr. Jha says the verbal post-mortems show that far more adults are dying of malaria than the WHO estimates for India. Malaria is, of course, a curable illness, and the data suggest that the Indian government needs to make a new push to ensure treatment is available in the rural areas most affected. "If you talk to clinicians in states like Orissa where malaria is common, it's no surprise to them, but at the central government, it's been a shock," he said.

Ms. Chandra noted that the million-death findings have helped counter the widely held belief that the problems of the urban population (such as smoking and heart attacks) did not affect the rural poor. They also help governments make policy with less risk of influence by lobbying from pharmaceutical companies, which buy media space to generate disproportionate hype about a particular disease.

The verbal autopsy has its origins in Dr. Jha's own life: His grandfather died in a village in India in 1975. His mother, then 30, had immigrated to Canada. Anguished at being so far away when her father died, she later came back to India and meticulously interviewed everyone who had been involved in her father's death. When, years later as a medical student, Dr. Jha asked her about it, she gave him an account that allowed him to identify the cause of death as a stroke. This left him wondering why that technique could not be used widely in India and beyond.

Today, the survey costs between $2 and $5 per household to conduct, but Dr. Jha aims to lower the costs to below $1 a household using software to do the diagnoses. The Million Death Study has been funded by the Canadian International Development Agency, the Canadian Institute for Health Research, the Bill and Melinda Gates Foundation and the United States National Institutes of Health.

Despite the huge impact of the study's findings, Dr. Jha is modest about his radical idea. "It's a good rule in epidemiology that if you don't look, you don't know – and no one really looked before."

Interact with The Globe