A majority of deaths in developing countries occur at home, not in the hospital. Therefore, millions of causes of death each year are not recorded with medical death certificates or reported in any statistical manner worldwide.

This is problematic because accurate mortality data can be very telling of the health of a society. Accurate mortality statistics can reveal health trends in various demographics of developing nations, and therefore provide a basis for the government or NGOs to implement changes in healthcare or public health education policies.

Like many other developing nations, India currently has poor data regarding the cause of death of most of its citizens. About 75 percent of the 9.5 million annual deaths in India occur at home, and are not recorded.

Organizations like the World Health Organization (WHO) often base their estimates of mortality in India on “hospital data, but in many developing countries most people die outside hospitals.”

In order to obtain a more accurate picture of how people are dying in India, Professor Prabhat Jha, an epidemiologist from the University of Toronto, began a project called the Million Death Study (MDS) for the Center for Global Health Research.

Jha told Record that “not knowing how people died in developing countries was a big [public health] constraint. We didn’t have a good sense of how many deaths occur, so we wanted to improve measurement.”

The MDS is meant to give data on premature mortality, which Jha defines as deaths that occur before old age (70 years of age in the developing world) from avoidable causes.

Accidental deaths are not included in avoidable mortality, but deaths from cancer and heart disease are considered avoidable, as they can be the product of environment or risk factors, says Jha.

The MDS looks at 8,000 randomly selected areas in India. Each area has about 150 houses. Jha says that this size gives a good sample of all the conditions. MDS data is not limited to hospitals; the study also collects information from homes and therefore, the MDS data represents a larger demographic than WHO data.

The MDS uses a technique called the “verbal autopsy” to obtain information, using an on-the-ground approach to collect mortality data.

The Indian government employs non-medical personnel who knock on doors across the nation to inquire about recent deaths in the home. If there has been a death and the bereaved are willing to discuss it, field workers fill out a checklist and compile a narrative about the deceased person’s life, focusing on their health practices.

All the work is conducted in the region’s local language, helping ease communication among participants.The researchers feel that on-the-ground data collection should be done by people without a medical background, because they may be more objective while listening to people’s stories.

Each narrative is then sent to two Indian doctors who diagnose the deceased person’s cause of death in a double-blind fashion – the two doctors do not know each other. This method allows both doctors to make an individual judgement and allows for a verification step when the two diagnoses are cross-checked against each other.

If the diagnoses do not match, a senior physician decides. Then the diagnoses are sent to Jha’s team to be coded and analyzed, and for statistics to be compiled.

Implications of the research

When the MDS was starting, Jha and his colleagues thought they would only be monitoring tobacco related deaths. The project became much bigger than they anticipated as it produced data on many other health issues, such as maternal and child health, alcohol, malaria, suicide, and cardiovascular disease.

Nature News reported that the MDS has also been able to uncover that the “top killers of Indians aged 30-69 are vascular disease (heart disease and stroke); chronic respiratory conditions; tuberculosis; and cancer.”

Jha says the only group in India the MDS does not reach are homeless people. He says that there are not enough homeless people in India to have that be a limitation on the holistic national picture the MDS is trying to create.

Doubts regarding the verbal autopsy technique

The MDS results have raised controversy in the public health community. Statistics that were previously held to be the standard representation of diseases in India are being questioned by the MDS results.

Results from Jha’s research “suggests there are 13 times more malaria deaths in India than the World Health Organization (WHO) estimates,” the BBC reported in 2010.

WHO argues that the verbal autopsy method over-represents the amount of malaria deaths because “verbal autopsy can be poor at differentiating malaria from other diseases that cause fever symptoms, which include septicemia, viral encephalitis and pneumonia,” Nature News reports.

In 2012, the WHO published their currently accepted instrument for verbal autopsies. On their website, they say that the verbal autopsy paper tool that physicians use is just as accurate as computer software in reaching the correct diagnosis.

Still other scientists say that the verbal autopsy method does not provide accurate results because it is too low-tech, and that computer models would represent the situation better.

Instead of diagnosing on a case by case basis, as the verbal autopsy method does, these scientists argue that individual diagnoses are not the right building blocks to compile aggregate data. “For most public health purposes, primary interest lies not in the cause of any individual death in the community but rather the aggregate proportion of community deaths that fall into each category,” writes Dr.Gary King, a statistician at Harvard University Population Health Metrics journal.

Looking ahead

The US National Institutes of Health, the Canadian Institute of Health Research and the Li Ka Shing Knowledge Institute have supported the MDS program. Jha and his colleagues began the MDS in 1997. While the last phase of the MDS was completed in 2013, the project is far from over. Data for over one million deaths has been collected and about half of that has been coded.

Jha foresees the next four or five years to be spent on coding and analysis of the data. Jha plans to extend his research from the state to district levels in India. He also thinks it would be valuable to “expand this approach to other settings around the world.”