When Christopher Murray was 10 years old, he and his family packed a few suitcases and a portable generator and left their home in Golden Valley, Minn., for a flight to England. From there they traveled by car and ferry to Spain, to Morocco and finally through the Sahara Desert to the village of Diffa in rural Niger. For the next year the family of five, led by a physician father and microbiologist mother, set up and then administered the local hospital. As the young Murray tended the hospital pharmacy and ran errands, he could not help noticing that the citizens of Niger came down with all kinds of odd diseases that nobody had back in Minnesota. He recalls wondering, “Why are people in some places so much sicker than in other places?”
The boy also grew frustrated. He and his family had worked hard helping the inhabitants of Diffa. But at the end of the year “I had a sense that when we left, things weren't that much different,” Murray says. “It planted the question: How do you make bigger, longer-lasting changes?”
That question drove Murray, over the next four decades, to push doctors and health ministers to pay more attention to the big picture, the long-term trends that determine why so many individuals die so young from preventable causes. He focused his career on setting up a system that could supply a crucial ingredient that is too often lacking in global health policy: reliable information.
The picture of what ails our species is often blurry, Murray realized, because politicians withhold or fudge embarrassing numbers. And it can be tricky to compare statistics from one country to the next. But there are ways around these problems.
With the help of others, Murray set out to build a new kind of instrument that could bridge the gaps in data and reveal the true state of the world's health, including what needs to be done to reduce the suffering of entire populations for generations to come.
The widespread use of the microscope triggered a revolution in human health in the 19th and 20th centuries by allowing scientists to focus on the smallest actors in human health—the identification of germs, which in turn led to improvements in sanitation, the development of antibiotics and the creation of vaccines. Murray's new instrument would be the opposite of the microscope. It would clarify useful details about illness at its largest scales: in countries, across continents and throughout the human species as a whole. You might call it a macroscope.
A Universal Language
Murray made the first stab at developing such an instrument at the World Bank when he and a few others assembled a landmark report in 1993 surveying the burden of human disease worldwide. In 2007 he founded the Institute for Health Metrics and Evaluation (IHME) at the University of Washington in Seattle and started recruiting a global network of collaborators to generate a much more sophisticated and comprehensive macroscope.
Like so many 21st-century innovations, the macroscope is made of software and big data. In one end go gigabytes of health statistics, collected from every corner of the planet and vetted by a multinational team of more than 1,000 scientists. Out the other end come interactive graphics and tables that illuminate with the kind of detail and accuracy never before possible, virtually all the different ways in which people fall ill, get injured or die around the world—from heart attacks to donkey bites. In the middle is supercomputer code that uses clever statistical mathematics to fix biases, identify and throw out unreliable data, and make intelligent estimates for the many parts of the globe where good numbers simply do not exist.
After 10 years and tens of millions of dollars in development, the system is now producing snapshots of the health of Homo sapiens with nearly the regularity of an annual physical. The IHME released its first expanded set of statistics in 2012 and another in 2014. An update, reflecting data collected in 2015, is due to appear in September, with annual checkups to follow. With each successive release, the snapshots—not only of the present but also those of the past—get more accurate, detailed and complete.
The reports, collectively known as the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD, for short), have spotlighted surprising trends in ailments as diverse as air pollution, kidney disease and breast cancer—and in countries both poor and rich. Some of the findings have also ignited controversy because they contradict those published by the World Health Organization (WHO) and other branches of the United Nations.
Like many others in the developing world, however, Agnes Binagwaho, health minister of Rwanda, is among the macroscope's fans. “It's not a simple new tool—it's a revolution,” she said at its official unveiling in 2013. “We are going to have a universal language for scientists, a universal language for [health] policy…, and our people are going to have better lives.”
Beginning in a Barn
The impetus for the macroscope arguably originated in the early 1990s with Dean Jamison, an economist then at the World Bank who was assembling an in-depth report about the interplay of health and economics around the globe. Jamison commissioned Murray to survey the economic toll exacted by disease and injury in 1990. Murray was later joined in the effort by Alan Lopez, an epidemiologist then at the WHO.
“Most of our work was done in Chris's barn in Maine,” says Lopez, who now works on the GBD at the University of Melbourne in Australia. “We hunted mercilessly for all available data on about 120 diseases and 10 risk factors. We were working 20 hours a day, totally obsessed.”
Compared with the latest GBD, which will cover 317 causes of death and disability in every country with a population greater than 50,000 people, the World Bank report, which came out in 1993, was rudimentary. It turned out to be enormously influential nonetheless. Microsoft co-founder Bill Gates cites it as one reason that he and his wife, Melinda Gates, decided to deploy most of their philanthropy—more than $37 billion so far—into efforts to curtail infectious disease [see box below]. And in 1998 Gro Harlem Brundtland, who was then director general of the WHO, hired Murray to build a similar system for her agency.
Click or tap to enlarge
Murray and his colleagues brazenly ranked the comparative performance of the health care systems in various nations. The rankings provoked a firestorm of criticism, both from academic epidemiologists and from government officials unhappy with the rank assigned to their nations (the U.S. was 37th; Russia was 130th). The WHO, which answers to its member nations, has never published such rankings since. “It became this extraordinary political battle,” Murray says. The fight was so distracting that his project to measure the global burden of disease foundered, and he left the WHO to go to Harvard University in 2003. The hard lesson, Murray says, was that “it is crucial that we firewall this work against government influence.”
He started looking for private money to launch an academic institute that would be free from political meddling. On a trip to Seattle, Murray met with Bill Gates, who says he agreed to the proposal almost immediately. In 2007 the Bill & Melinda Gates Foundation donated $105 million to launch the IHME. Work on a macroscope was soon under way.
It is indeed a messy business, accounting for human suffering across a disorganized world. But Murray found other researchers who shared his and Gates's confidence that the flaws in statistics available from governments, advocacy groups, the WHO and other U.N. agencies could be fixed—and that better numbers would save lives in the long term. Mismeasurement, bias and missing data: these were the demons they needed to battle.
Much of the raw data fed into the system comes from health ministries, aid organizations or the scientific literature, but it first goes through rigorous quality control. “Whenever we get a new data set, our first question is, ‘What's wrong with it?’” Lopez says. “We scrub it for garbage coding—deaths from ‘an act of God,’ for example—and use scientific methods to reallocate those to a well-defined list of causes.” This process helps to control for certain outliers, such as France's seemingly low death rate from heart disease despite a high prevalence of risk factors. It turns out that French physicians tend to mark down some other coexisting cause when people die of heart attacks. “That cultural practice accounts for about half of the so-called French paradox,” says Theo Vos, one of the lead scientists at the IHME.
Hundreds of scientists from across the globe who have expertise in each disease and region also adjust data sets to account for variations in how diseases are defined. By putting all the results on the same footing, Lopez says, “we can compare cancer in Hungary with cancer in El Salvador or South Africa or anywhere else.”
Then there is political influence. “It's increasingly hard for governments to dictate numbers to the WHO and the U.N.,” Murray says, “but there are subtle effects. UNAIDS puts out estimates of HIV/AIDS prevalence each year, for example, but China and some other countries do not like them. So they do not publish these numbers for those countries”—83, as of 2015.
The problem of missing data is the thorniest. In many of the sickest parts of the world, comprehensive health statistics simply do not exist. The GBD team fills the gaps in two ways. First, the investigators hoover up surveys performed by a small army of workers moving village to village—in some places, house to house—to gather information from handwritten records and, where necessary, interviews with people about sickness and death in their families. The Million Death Study, under way in India, is conducting such work on a grand scale; its preliminary results in 2010 suggested that malaria takes at least 10 times more lives in India than had been reported by the WHO, which relied mainly on hospital records and thus missed many of the deaths that occurred at home.
The second way to fill in missing data is to extrapolate it from various common patterns that researchers have observed for specific diseases, injuries and risk factors. Malaria is typically worse during or just after the wet season, for example; cancer rates are higher among the elderly; and HIV is more prevalent in countries that border on nations where many people are HIV-positive. Such correlations allow one to use statistical indicators that are well measured in one part of the world to make reasonable estimates about what the health numbers should be in another part where data are incomplete.
“We have a database of 200 of these [well-measured variables], ranging from latitude, population density and rainfall to how many cigarettes people smoke and how many pigs are consumed,” Vos says. The system generates myriad combinations of these variables, plugs them into a large number of mathematical models of differing form, and then tests to see which combination produces the most consistently accurate predictions for each disease.
Click or tap to enlarge
This approach, known as ensemble modeling, has been widely used in weather forecasting, finance and insurance, among other areas. But it falls “outside the comfort zone” of many epidemiologists, Vos says, in part because it requires enormous computational horsepower.
The GBD now tracks more than 1,000 health indicators for 188 countries covering 25 years, which are in turn double-checked against 20 to 40 statistical models. The team also runs each model 1,000 times to wiggle all the data points within their range of plausible values, a technique that allows the researchers to bracket their estimates with error bars indicating the degree of uncertainty. The computations are run on the IHME's supercomputer, where 12,000 high-performance processing cores churn away at the math for four days to complete a single snapshot for the planet—a kind of report card on the health of the human species.
“Just the fact that somebody is trying to publish these numbers and put error bars on them focuses the discussion,” Gates says. “Now that the IHME has created a central repository, you no longer have to read hundreds of articles and try to assemble the big picture yourself. People can argue over specific numbers, but if the right process is in place, the error bars will either get bigger or some study will be done, and the state of knowledge will get better.”
The initial release of GBD numbers in 2012 sent ripples even through countries that pride themselves on their health information systems. Officials in the U.K. were alarmed, for example, to see that the health of Britons has been lagging behind that of their European neighbors. “The risk-factor analyses led them to change their priorities and put more emphasis on diet,” Murray says. At least 33 countries, including China, Brazil, Germany and Russia, are now undertaking similar studies of their own populations to improve the quality and detail of their health statistics—data that will flow into the global macroscope.
Some of the new pictures of human health produced by the GBD system have stirred up debate because they contradict numbers long accepted as authoritative. A GBD analysis of HIV prevalence published in 2014, for example, suggested that the UNAIDS estimates for 2005 to 2012 were too high by 17 to 19 percent, a difference of about 6.6 million infections and 635,000 deaths. If correct, the low numbers would raise the question of whether certain prevention and treatment strategies are working better than expected and, if so, whether they might be used more widely.
In another controversial finding, the IHME estimated that in 2013 about a third of those killed by malaria were adults. It has long been thought that whereas the mosquito-borne illness sickens many older people, malaria fatalities are largely limited to children. “Ninety percent of the field thinks [Murray is] wrong,” Gates says. But within the next few years better data gathering should settle the dispute.
The GBD results also contain heartening revelations. They report, for instance, that the burden of death and disability from diarrheal diseases—which are caused primarily by unsanitary water—dropped by 70 percent from 1990 to 2013. Indeed, dirty water has dwindled to become so much less of a threat than other risks that Murray and his colleagues suggest that it may be wise to redirect aid in some regions from water projects to, say, preventing road injuries. Traffic accidents are on the rise in part because children who used to die from waterborne illnesses now survive to become pedestrians, bicyclists and teen drivers.
By the same token, the system also highlights blind spots in the global surveillance system. “It drives us crazy that we don't really know where typhoid and cholera are,” says Gates, who now takes GBD charts with him to meetings with officials and aid workers. Because recording of these diseases is so spotty, the IHME reckons that typhoid may account for as few as six million years of healthy life lost worldwide—or as many as 18.3 million years. Giant uncertainties similarly surround the impacts of whooping cough, measles, and hepatitis A and C.
Murray remains confident that the view through the macroscope will sharpen with time. Leaders could then shift their focus from the numbers of people sick and dying from each disease to the trend from year to year. That is what happened in macroeconomics, he notes. Individuals no longer pay much attention to median income, total number of jobs or the dollar value of the GDP; it is the rate of change that matters more. We stop asking, “Can we do better?” Murray says. The question becomes, “How can we improve faster?”
Q&A: Bill Gates on Global Health
Bill and Melinda Gates have been among the principal funders of many global health initiatives, including efforts to reduce the burden of HIV and tuberculosis and to eradicate polio and malaria. The Bill & Melinda Gates Foundation also paid to create the Institute for Health Metrics and Evaluation (IHME) and has bankrolled its ongoing work to measure the occurrence of various maladies and their causes around the world, a project known as the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD, for short). In an April 2014 interview with Scientific American's contributing editor W. Wayt Gibbs, Bill Gates reflected on the origins of these efforts and the progress that has been made to date.* An edited excerpt follows.
Scientific American: You were an early backer of Christopher Murray and his push to create an independent organization, the IHME, to pull together rigorous statistics on human health worldwide largely independently of the World Health Organization [WHO]. How did you two meet and decide on this course?
Bill Gates: I met Chris in 2001 when he was working for the WHO and was doing the first-ever ranking of national health systems. Some countries were pushing back because they didn't like how he ranked things. This idea that somebody should try to pull together the best understanding of health, particularly for poor countries, was an attractive one. So we gave money to the University of Washington to create IHME.
SA: But the WHO and other U.N. agencies collect and publish lots of health statistics on countries around the world. Why is it necessary to have a whole separate effort to do that?
BG: I love the WHO, and Margaret Chan [the WHO director general] has done a lot of great things. But it is a U.N. agency, and that creates certain complexities. When Chris was doing country rankings inside the WHO, he found out that both funding and their inability to take controversial positions were limiting. Ranking their customers ended up being tough for them to do.
SA: What's your high-level impression of the collective health of our species? Are we healthier than we were 20 years ago?
BG: We are so much healthier now than in the past. It's one of those mind-blowing good news stories. Because it is good news, and there is no villain, it doesn't get the attention it deserves. But if you look at countries like Vietnam, Cambodia, Sri Lanka, Rwanda, Ghana—basically all communicable diseases are going down. The only infectious disease we work on that is now going up is dengue.
Yes, the burden of noncommunicable diseases is increasing in developing countries. We do have a diabetes epidemic. We do have an explosion in medical costs. On a global basis, though, what's happened in the past 20 years is incredibly dramatic.
Editor's Note (7/27/16): This sentence from the print article was edited after it was posted online to add the interview date.