On the ground floor of a 22-story tower at the University of Washington in Seattle, stands a 25,000 CPU supercomputer. All that data-crunching strength was flexed to calculate the most recent iteration of a report called the Global Burden of Disease (GBD). Considered one of the largest data science projects in the history of global health, the GBD is revolutionizing the way people “see” the world around them.
The GBD puts a country’s health burden into a global context,” says Dr. David Blazes, a specialist in epidemiology and surveillance at the Bill & Melinda Gates Foundation. “It truly is a global look at health burden, rather than just taking into account one disease, in one country, in one age group.”
The health world is not short on measurements and statistics. But the GBD represents a quantum leap in how health data is calculated and reported.
“We’ve gathered over 100,000 data inputs from 195 countries, covering more than 350 different causes of health loss,” says Stephen Lim, Senior Director of Science and Engineering at UW’s Institute for Health Metrics and Evaluation (IHME). “Each time we run the GBD, we’re producing somewhere on the order of 38 billion unique numbers.”
In producing these numbers, the GBD must account for all the ways people measure health. Consider, for example, the different ways for just testing diabetes.
“They may do a fasting glucose test,” Lim says. “Or, do an oral glucose tolerance test, giving you a certain amount of glucose and measuring the results of you ingesting that. What we have to do is build a set of statistical methods to try to deal with those different ways of collecting data.”
Making sense of all those statistical methods is where the real work begins for the IHME team. And one of their biggest challenges has been finding ways to compare seemingly incongruous health conditions.
“Traditionally, it’s been hard to compare conditions like ischemic heart disease and depression,” Lim says. “One primarily causes premature death (ischemic heart disease), while the other is primarily disabling, (depression). We found the best way to compare them was to create time-based measurements of health loss. If you die at age 5, the stream of life lost is much larger than if you die at age 80.”
The Global Burden of Disease uses time-based health metrics such as DALY (disability-adjusted life year), YLL (years lost due to premature mortality), YLD (years lived with disability). The methodology is proving so informative, the nation of Brazil is embracing it as a baseline for their own health surveillance programs.
“For a long time in Brazil, we monitored 42 key health indicators,” says Dr. Fatima Marinho, head of surveillance at Brazil’s ministry of health from 2003 to 2019. Marinho now teaches at the University of Sao Paulo. “But the old indicators were reinforcing some fragmentation of health problems and interventions at the local level. By adopting the GBD’s measurements, we’ve been able to integrate our own surveillance system.”
When Marinho started working at Brazil’s ministry of health, then-President Lula De Silva launched sweeping healthcare reforms. She joined the newly-formed secretariat of health and surveillance, helping reform its system for gathering epidemiological intelligence. When an IHME researcher contacted her to gather local health data for the GBD report, Marinho was curious enough to attend a two-week training course about the GBD. She returned to Brazil feeling certain it was a model that could be used to reform Brazil’s health surveillance system.
“We’ve been expanding our use of the GBD in waves,” Marinho says. “We started by creating knowledge at the university level. Then, when the students graduate, they go on to advise local health authorities. Now that knowledge has reached the state level where the GBD is used as the basis for their annual reports.”
The shift has triggered something of a renaissance in academic analysis of health surveillance in Brazil. The GBD’s results are the subject of 38 academic papers, including IHME’s own in-depth report published in The Lancet.
“It helps us to see the progress we’re making in Brazil,” Marinho says. “It shows that we’ve reduced the gaps of inequality between rich and poor states. We expected the possibility of this improvement, but we didn’t have a way to measure it before the GBD.”
The GBD’s great strength is focusing the optics of global health to create clarity at the subnational level. In Brazil, that takes shape in several forms. A gun control law passed in 2008 had been credited with driving down the national rate of interpersonal violence. However, the GBD showed those gains were isolated to urban areas like Sao Paolo, while interpersonal violence was increasing in rural areas.
When the Zika epidemic struck, in 2015 and 2016, the GBD empowered Marinho’s department to advocate for new infant health policies. And a spike in the number of children with diarrhea in one specific area led to the discovery of bad water quality.
“Brazil is a big territory,” Marinho says. “The GBD helps the ministry of health build bridges with local experts and make national cases for changing healthcare policies.”
Meanwhile, back in Seattle, the IHME team continues to roll out reports synthesizing the results of the GBD: a global increase in stroke victims
, the global burden of typhoid fever
, even an analysis of how well nations are meeting sustainable development goals
By putting global health into a worldwide context, the Global Burden of Disease is revealing subnational health trends that show health ministers where to take action.