Christopher Murray led a massive analysis of the “health loss” caused by diseases and injuries.
CREDIT: J. COHEN/SCIENCE
SEATTLE, WASHINGTON—If you had stumbled into Christopher Murray's office in October without knowing who he is or what he does, the cryptic notations written in six shades of felt pen on the whiteboards on his walls would have told a tale as intriguing and revealing as cave paintings. The formulas, graphs, and arrows suggest an ambitious attempt to decipher something exceedingly complex. These are some of the words and symbols scattered about: 187 countries, health, disease, $, mortality, partnership, methods, and—in bright purple and all uppercase letters—UNCERTAINTY.
Murray heads the Institute for Health Metrics and Evaluation (IHME), a branch of the University of Washington (UW) that contends it has created the most detailed and authoritative report ever on the state of the world's health. The so-called Global Burden of Disease (GBD) 2010 study will appear on 15 December in the largest issue of The Lancet ever published, and Murray hopes it will have a major impact on how policymakers, donors, and researchers allocate resources to help people lead healthier, longer lives.
The effort, largely bankrolled by the Bill & Melinda Gates Foundation, is “a huge, ambitious, and highly disciplined attempt to describe the totality of death and illness in every part of the world,” says global health veteran Richard Feachem of the University of California, San Francisco (UCSF), who chairs an independent scientific oversight group for IHME. “There's nothing else like it or even approaching it.”
GBD 2010 consists of eight papers, 194 pages in total, that examine the epidemiology and loss of health caused by 291 diseases and types of injuries in 187 countries and a whopping 1160 of their lasting effects. It analyzes changes in disability and death from 1990 to 2010; using new computer models based on complex statistics, it ranks the major causes of mortality and morbidity in 20 age groups in 21 regions of the world and identifies 67 underlying risk factors. As Murray's whiteboard telegraphed, the studies give uncertainty intervals for the estimates as well, bringing scientific rigor to a field that often relies on squishy data.
But another type of uncertainty surrounds the project: How much credence will it have with fellow scientists and policymakers? Many have questions about how IHME arrived at its results and how they fit with similar efforts by the World Health Organization (WHO), until now the main source of global health data. IHME caused an uproar in February when it gave a sneak peak of GBD 2010 with a paper in The Lancet that tallied nearly twice as many malaria deaths as WHO did (Science, 15 June, p. 1372). Other numbers may well be equally contentious.
Passions run high about these fights in part because the money spent on research and control measures for any disease is determined largely by the perceived suffering that it causes. Advocacy groups and researchers alike try to trot out evidence that “their” affliction is a major global problem.
In IHME's case, the debates are intensified by some scientists' frustration about what they say is an arrogant attitude and a lack of transparency at the institute. Murray, widely admired for his intellect and abundant enthusiasm and energy, has come under criticism for his domineering style. “There are issues with methods, results, and personalities,” says Dean Jamison, a UW health economist who quit IHME 2 years ago and acknowledges that his views are “clouded by my general lack of perfectly good and cordial relations with Chris Murray.”
This much is certain, however: GBD 2010 demands serious attention. Even its sharpest critics can't ignore it.
Murray's efforts to take stock of humanity's health go back 2 decades to when the World Bank published a watershed report called World Development Report 1993: Investing in Health, prepared by a team that Jamison led. Murray, who has a Ph.D. in international health economics and a medical degree, wrote an appendix that introduced the GBD concept to a wider audience, together with WHO epidemiologist Alan Lopez, who is now at the University of Queensland in Brisbane, Australia.
Until then, the relative importance of diseases had simply been assessed by the number of deaths they caused, which was fairly easy to track. Murray and Lopez wanted to “quantify the full loss of healthy life” and take into account nonfatal conditions such as paralysis, depression, and blindness. They devised a metric called the disability-adjusted life year (DALY), which combined the years of life lost because of a fatal disease or injury with the years of life lived with disability. Controversial at first, DALYs revealed startling patterns. According to the 1993 report, for example, neuropsychiatric diseases caused a higher burden worldwide than cancer.
In 1998, Murray moved from Harvard University to WHO's Geneva headquarters to head the Global Programme on Evidence for Health Policy, which created the organization's first burden of disease unit, led by Lopez. GBD reports soon became a mainstay of WHO. Murray returned to Harvard in 2003 hoping to form his own institute, but promised funding fell through; he came to Seattle in 2007 with a $105 million commitment from the Gates Foundation, which believed that all global health funders would benefit from better metrics to evaluate the impact of investments. UW contributed another $20 million. IHME's staff, now numbering nearly 100, built up a vast network of collaborators that included WHO: The new papers in The Lancet have 486 co-authors from 302 institutions.
The papers look at everything from DALYs to risk factors, causes of death, illness, and impairment, and how to weight the severity of nonfatal illnesses; their tables, maps, bar graphs, and charts reveal a multitude of intriguing patterns. Although mortality in children under age 5 has plummeted between 1990 and 2010, for example, more people now suffer from mental disorders and back pain. HIV/AIDS jumped from the 35th leading cause of death in 1990 to the sixth in 2010. Noninfectious diseases such as heart disease account for increasing amounts of “health loss.” Several infectious diseases, including diarrhea and malaria, are on the decline.
Some of the findings are perplexing. Tuberculosis mortality, for example, has dropped steeply, but new cases have not. In 2010, road injury accounted for 10.7% of deaths in males in the reproductive age bracket, but only 0.5% in females. Lower back pain ranks immediately below HIV/AIDS in DALYs.
Geographic differences jump out as well. Mortality in people of reproductive age changed little in Russia between 1970 and 2010, but skyrocketed in southern African (because of HIV/AIDS) and dropped in upper-income countries. Self-inflicted harm, including suicide, ranks as the 13th most common cause of life-years lost worldwide but is rare in sub-Saharan Africa. Alcohol disorders have had a devastating impact in the former Soviet Union and parts of Latin America, where people drink more and liquor tends to be of lower quality.
Epidemiologist Peter Piot, who runs the London School of Hygiene & Tropical Medicine (LSHTM), says the absolute figures interest him less than the changes over time. “I don't care—and I don't think many people care other than disease advocates—whether 1.5 or 1.6 million die from a disease,” says Piot, who serves on IHME's board. “What's important is what direction the world is going in and what's happening in my region.”
Murray says that, after the fight over malaria, he doesn't anticipate much debate about other high-profile diseases, such as tuberculosis and HIV/AIDS. “The smaller diseases, those communities get more riled up,” he says. “If our numbers are smaller, it's going to hurt their bid for funding, so they get very restive. You'll have a million of those types of conversations.”
They're already beginning. Peter Hotez, a pediatrician at the Baylor College of Medicine in Houston, Texas, who specializes in neglected tropical diseases, is a co-author on the GBD 2010 paper about DALYs. But he thinks the paper's estimates for schistosomiasis and Chagas—which he cares greatly about—are too low. Jamison says that IHME didn't properly factor in stillbirth in its calculations of under-5 mortality—“a conceptual hole of some magnitude.”
Sandy Cairncross, a public health engineer at LSHTM who specializes in water and sanitation and who served on one of many expert groups for GBD, says that unsafe water and poor sanitation should have ranked much higher in risk factors. His concerns are so serious that he co-authored a commentary in this week's issue of The Lancet questioning whether policymakers should even use GBD 2010's rankings of risk in their decisions.
Cairncross says that IHME dismissed much of the literature he selected that showed the important health benefit of delivering water to houses through pipes. “They only accepted one study in the world that got over their bar of scientific rigor,” Cairncross says. “And that particular study apparently showed no significant effect on house connections, unlike most others that showed [disease] reductions of about 50%.”
Black box step
Cairncross and several other critics say a fundamental problem with IHME's conclusions is that researchers used complex statistical models and computer analyses—what he calls a “black box step”—that baffle outsiders. The GBD 2010 paper on years lived with disability gives a flavor: “To address these challenges, we have developed a Bayesian meta-regression method, DisMod-MR, which estimates a generalized negative binomial model for all epidemiological data.”
UCSF's Feachem says this “analytical sophistication” presents real challenges, but he contends that it's required because the jigsaw puzzle is so complicated. “By the nature of the beast, it will be very hard to get it to the point where the average epidemiologist with the average mathematical skills will be able to seriously reanalyze and arrive at different conclusions,” he says.
UW's Jamison says his former employer would mollify many critics if it embraced the transparency it espouses. “There's a lack of access to data,” Jamison insists. “Their results can't be honestly checked and we don't have a capacity to interpret the underlying numbers.”
The complaint is part of a bigger gripe about IHME's headstrong ways—and what some assert is Murray's overcertainty about debatable issues—that has also frayed its ties with WHO. Initially, WHO envisioned working with IHME in a tight collaboration and even adopting its estimates. “We stepped into it because we thought it was a joint exercise,” says Ties Boerma, director of WHO's health statistics and informatics, “but it became more of an IHME exercise.”
A “briefing note” written by a WHO assistant director general last winter told WHO staffers that it would “not be appropriate” to be co-authors to the GBD 2010 papers or for WHO's logo to appear on IHME publications. According to the memo, obtained by Science, WHO developed serious concerns about the numbers in GBD 2010 after IHME researchers presented them to WHO staff members in September 2011. Based on those data, the memo says, big discrepancies between the GBD 2010 papers and WHO estimates were to be expected not just for malaria but also for child and maternal mortality, deaths due to neglected tropical diseases, vaccine-preventable diseases (including measles), cancers, and tobacco.
IHME subsequently adjusted its deaths for HIV/AIDS, Boerma notes, but in a commentary in this week's issue of The Lancet, WHO Director-General Margaret Chan says GBD 2010's estimates still “differ substantially from analyses by WHO and other UN entities.” Boerma notes that one major discrepancy is that GBD 2010 estimates the total number of deaths annually at 52 million, WHO at 56 million. More differences may come to light as the published reports receive closer scrutiny.
IHME alienated several other erstwhile contributors along the way, and an external evaluation completed in November concluded that the institute “is not consistent in when and to whom it shares methods, data sources, [and] authorship and this is perceived as not being transparent.” The report said that “IHME is viewed as a competitor vs. collaborator by many researchers in the health metrics field.” Murray has gone so far as to suggest that WHO get out of the business of assessing GBD. “Bureaucracies don't do statistical innovation. Researchers do.” But Boerma says that WHO will continue putting together its own GBD.
In a commentary in The Lancet package, Murray, Lopez, and other key IHME staff members say it's “reasonable and to be expected” that some contributors in an enterprise this large would disagree and choose not to be co-authors. But Murray challenges the accusation that IHME has not shared data and methodology. “The core tenet throughout this collaboration has been that an open and voluntary process would provide for rigorous debate to ensure the best possible results,” he says.
Hotez of Baylor says he has “a lot of sympathy” for Murray and his team. “It's incredibly complicated to bring all those investigators together,” he says. And in the end, policymakers should keep the findings in perspective, Hotez adds. “It's one of several metrics that should be used when trying to control disease and exploring policy,” he says, noting that it doesn't factor in economic costs of diseases, existing tools to combat them, or health system capabilities.
No fudged consensus
IHME intends to release another ocean of data in January, when it will report even more granular analyses of country-by-country information. It will also make a new interactive database publicly available that Murray says will lead people to explore questions that his team never imagined.
As debates about those data inevitably kick in, WHO plans to hold a meeting in February that will gather IHME scientists with experts from WHO and elsewhere to discuss how GBD 2010 reached its conclusions and how it differs from other estimates. Feachem says those discussions are exactly what's needed. “The last thing we want is fudged consensus,” he says. “Some of these disagreements are healthy because they force tough questions. And that's how science works. In time we'll find a better outcome.”