How death by numbers promotes global health

Myanmar relies on its midwives to record causes of death, but like many countries the data is chronically incomplete. Now, with help from technology the midwives are doing accurate autopsies as well as delivering babies.

Andrew Trounson, University of Melbourne

Published 17 July 2017

In rural Myanmar, the local midwife is at your side at the start and end of your life.

Not only is she responsible for delivering babies and registering them, she is also responsible for registering deaths and cause of death. And it’s all done the old-fashioned way, using scribbled notes that are sent back to the bureaucrats on bicycles or by mail.

It hardly makes for a foolproof system producing accurate records, especially for deaths. Midwives are understandably more expert at looking after the newly born rather than assessing the causes of the newly dead. As a result, over half of the deaths recorded in the country have no doctor-certified cause. That leaves health services and policy makers with no accurate idea of what is killing people, and no reliable data on which to prioritise healthcare spending and prevention.

But a basic bit of technology – a computer tablet - and an algorithm that can accurately diagnose the cause of death without the expertise of a doctor, is now revolutionising the collection of death data in Myanmar.

Senior midwife Tin Thein needs to attend to her malaria and primary health patients by motorbike. Myanmar’s midwives are now collecting accurate data on causes of death using questionnaires and computer software thanks to the Bloomberg Data for Health Initiative and researchers from the University of Melbourne and University of Washington. Picture: Department of Foreign Affairs and Trade/Flickr

“From a public health perspective, the information we need the most is information on how people have died. But in rural Myanmar, collecting this data is entirely up to village midwives, and for them it is understandably just another burden behind safely delivering babies,” says University of Melbourne public health expert Associate Professor Deirdre McLaughlin, who is deputy director of the $120 million Bloomberg Data for Health Initiative (D4H) to boost global health data.

There are about 450,000 deaths recorded each year in Myanmar, but of these about 80 per cent occur outside urban areas and outside of medical facilities. It means the burden of information gathering really is on the midwives, but each year 250,000 deaths are simply registered with no information on what caused or contributed to them.

Based on the limited data available from these records, malaria could either be the single leading cause of death in Myanmar, or possibly the 17th biggest killer. Similarly heart disease may be Myanmar’s second biggest killer but it may also be its 20th biggest, while asthma may be its 6th biggest killer but it may also be its 15th. We simply don’t know.

But in an innovative solution to the problem, the University of Melbourne, as the lead technical institution in D4H, is now training midwives to gather so-called ‘verbal autopsies’ on computer tablets. The information is then fed into a laptop where pattern recognition software automatically generates a probable cause of death, all without the need for a doctor.

A verbal autopsy is when, instead of relying on having a doctor to examine a body, a cause of death is determined from the victim’s symptoms and lifestyle as gathered from interviews with relatives. However, while verbal autopsies need less expertise to gather, the results still needed to be interpreted by a doctor, taking them away from their critical health care duties.

To bypass the need for a doctor, researchers from the University of Melbourne and the University of Washington developed software that allows the midwives to load the data they gather from household interviews directly into a program called ‘SmartVA’ that then generates a probable cause of death. SmartVA has been shown to be 13 per cent more accurate in diagnosing the cause of death than doctor evaluation of the same verbal autopsy questionnaires in the field.

The researchers also stripped down the length of the questionnaire from around 50 minutes to 20 minutes to make it easier for the midwives to collect, with only a small reduction in overall accuracy.

“Suddenly in Myanmar, where midwives had been collecting and shipping around pieces of paper, they are now doing 20 minute interviews and then feeding the data back into laptops to get reliable cause of death data in real time for rural populations for whom no such information existed before,” says D4H Director and University of Melbourne Laureate Professor Alan Lopez.

Midwives in Myanmar being trained on how to collect data on causes of death in their communities and then load it into software that interprets it to produce accurate autopsies. Picture: Supplied

A leading international population health expert, in 2016 Professor Lopez was rated one of the world’s top ten most influential scientists for his work on improving global health data to guide policy action.

So far D4H has rolled out the SmartVA system to 14 townships or districts in Myanmar, covering 2.2 million people. They plan to expand it to a further 34 townships by 2019.

“We have just started a similar pilot project using SmartVA in Sri Lanka and the government there is delighted with the results,” says Professor Lopez, who was one of the leaders in the development of SmartVA. “From virtually no useable data on who dies of what, they will soon have reliable information on causes of death in rural Sri Lanka.”

And information like this can’t come fast enough. Globally 40 million people die every year with no reliable documentation on what killed them, mostly in low and middle-income countries with antiquated and piecemeal bureaucracies.

“Without data, health planning is done in a vacuum, it is done on the basis of ideology, or there is no planning done at all,” says Professor Lopez.

Professor Lopez has championed the need for better data ever since he and colleague Dr Christopher Murray launched the Global Burden of Disease (GBD) Study in 1993. Twenty five years later, the GBD is now published annually in the British medical journal, the Lancet, and remains the most comprehensive effort worldwide to measure the world’s health problems, involving over 2,300 researchers in more than 150 countries.

Launched in 2015 as a four-year project largely funded by businessman and former New York mayor Michael Bloomberg’s Bloomberg Philanthropies, D4H is complementing the work of the GBD by improving the quality of health data across 20 participating countries and cities, including data on health risks like tobacco use and nutrition. D4H has a target to extend documented cause of death to 2 million people a year, and extend high quality birth and death certificate systems to cover 250 million people.

The Australian government is contributing $24 million.

School children on an excursion in Galle, Sri Lanka. The country is one of the participants in the Bloomberg Data for Health Initiative. Picture: Shutterstock

According to Professor Lopez, D4H is about 40 per cent of the way to achieving its goals with some countries ahead of others. He says Myanmar is a stand out performer, driven by the enthusiastic support of the government there. “Myanmar is twice as far along compared to where we thought it would be at this time.”

Professor Lopez says a key part of the project is working with governments and “cajoling” them into reforms. Often that means showing ministers and bureaucrats the evidence to shock them into action.

For example, in 2009 Professor Lopez led research in Thailand, funded by the Welcome Trust, that found deaths from stroke and heart disease in the country were actually three times higher than the government’s own figures. Deaths from HIV/AIDS were five times higher, and even deaths from road accidents were double what the government thought.

Professor Alan Lopez is leading the effort to convince countries to improve their health data collection systems. Picture: Supplied

In the Solomon Islands D4H has been rolling out training to improve the medical certification of deaths in hospitals and, in April, the country was able for the first time to report its leading cause of death. It proved to be heart disease followed by diabetes. “I think that this turned out to be quite a surprise for many people, including the government”, says Professor Lopez.

“We are trying to achieve bold changes in these countries and to break through decades of inertia with their data collection systems,” he says.

“We have to find innovative, compelling and cost-effective ways to show countries the evidence about who is dying of what and preach the benefits of better data for planning. But most importantly we need to convince them that getting better data is achievable, the technology exists and that we at the University of Melbourne are here to help in the long term.

“That is the true benefit of the Bloomberg funding. It means we can keep going back to governments to provide the technical assistance we promised, to keep making the case, establishing trust, and encouraging them to make sustainable improvements.

“Timely and accurate data on leading causes of premature death in countries is critical in allowing governments and civil society alike to agitate for essential health reforms and better tackle emerging health challenges.”

Banner Image: Shutterstock

Find out more about research in this faculty

Medicine, Dentistry and Health

Content Card Slider

Content Card Slider

Subscribe for your weekly email digest

By subscribing, you agree to our

Acknowledgement of country

We acknowledge Aboriginal and Torres Strait Islander people as the Traditional Owners of the unceded lands on which we work, learn and live. We pay respect to Elders past, present and future, and acknowledge the importance of Indigenous knowledge in the Academy.

Read about our Indigenous priorities
Phone: 13 MELB (13 6352) | International: +61 3 9035 5511The University of Melbourne ABN: 84 002 705 224CRICOS Provider Code: 00116K (visa information)