Translate AMICOR contents if you like

Friday, June 08, 2018

Global Burden of Diseases

In 1997 The Lancet published “Mortality by cause for eight regions of the world: Global Burden of Disease Study”. This Global Burden of Disease (GBD) study was the first in a series of four articles that ushered in a new era in descriptive epidemiology, and launched the ascendancy of the GBD in the then nascent field of global health. The four landmark papers gave the GBD study and its authors, Christopher Murray and Alan Lopez, scientific credibility and exposure in the scientific community.
The first GBD study, describing the epidemiology of death worldwide, was important because of its findings and its discernible historical and scientific trajectory. Working for WHO in Geneva, Lopez had been specialising in vital statistics and defining causes of death for over a decade. Meanwhile, Murray, then based at Harvard's Department of Population and International Health, was investigating the levels, patterns, and causes of adult mortality in the developing world, at a time when, understandably, the world's attention was focused on child survival. The intellectual roots of this project were to be found in earlier collaborative work for the World Bank's World Development Report 1993: Investing in Health. Put together by a team led by Dean Jamison, this influential report used the disability-adjusted life-year (DALY) to measure the burden of disease. DALYs are calculated by combining the years of life lost (YLL) from premature mortality with the years of life lived with disability (YLD), weighted according to severity grading. For the report's analysis, Jamison turned to Murray and Lopez. Their work, quantifying the impact of the combined loss of life from premature death together with loss of healthy life from disability, formed an appendix to the report, entitled The global burden of disease, 1990.
 Opens large image

Christopher Murray and Alan Lopez at the launch of 2010 GBD in December, 2012
Courtesy of the Institute for Health Metrics and Evaluation
The 1997 Lancet papers represented a melding of intellectual approaches: for Lopez it formed a quantum leap in the evolution of a science that he and others had pioneered 15 years earlier, whereas for Murray, the analysis went beyond a body count and enabled him to uncover the causes of premature death—information that is critical for the strategic planning of health-care systems. Lopez and Murray were following in a long tradition. The collection of vital statistics relating to population size, growth, and health had been a preoccupation of rulers and governments since the beginning of recorded history. It has only been in the past 200 years, however, that there has been much of a change in the pattern of mortality. For example, an individual born at any time before the middle of the 18th century had less than a 50% chance of surviving long enough to produce any children. But since the 1950s in high-income countries, life expectancy has improved more than in the entire previous span of human history. As a consequence, to adapt the thinking of the epidemiologist and one of the founders of medical statistics William Farr, how people live and how—of what causes, and at what ages—they die are among the most important questions that can be considered.
As Murray and Lopez stated in their 1997 GBD paper, “Reliable information on causes of death is essential to the development of national and international health policies for prevention and control of disease and injury.” Yet while over the previous 30 years medical services had expanded and statistical techniques had advanced, the accuracy of death certification was low in many developing countries and even in some high-income countries. In fact, Murray and Lopez found that medically certified information was available for less than 30% of the estimated 50·5 million deaths that occurred each year worldwide. Fortunately, a mass of other useful information had been collected, including “sample-registration for India and China, and small-scale population-study data sources”. Collectively, these data made it possible to piece together a realistic picture of the distribution of mortality over large parts of the world, proving the molecular biologist John Cairns' 1997 aphorism that “whenever statistics are available, it is folly not to use them”. Working together towards their shared goal of objectively measuring the health of the entire world, Murray and Lopez recognised that to add to the canon of accepted knowledge, their estimates needed to be internally consistent and plausible. As such, registration data “were corrected for miscoding, and Lorenz-curve analysis was used to estimate cause-of-death patterns in areas without registration”.
Unlike the sophisticated GBD research of today, with its many collaborators and use of Bayesian algorithms that bring biological knowledge into statistical analysis, the 1997 paper was a fairly low-budget affair. With only basic computational power available and a reliance on the use of spreadsheets, interpreting the study's findings was, as Murray recalled, to prove “as much an art as it was straight science”. Bringing the diffuse datasets together for eight regions of the world could be achieved only through a painstaking analysis of the evidence and the application of skilled understandings of mathematics and demography. So while interpretation and inference played a part, the study's judgments were most clearly guided by elements of the science of statistics. Much of the groundwork, the “gathering, vacuuming, cleaning, torturing, and interpretation of data”, was done by Lopez. As Lopez freely admits, “I didn't have the vision Chris had, and the GBD would have never have happened if I led it; and…I didn't think it could be done. But I realised that ‘this guy [Murray] has got a vision, and I'm going to go with it’.” But the vision would require more than one person to bring it into existence and Lopez's role in bringing things together was to prove invaluable.
Together, Lopez and Murray selected eight geographic regions as the basis of the study's datasets: established market economies; the former socialist economies of Europe; Latin America and the Caribbean; China; India; the Middle Eastern crescent; other Asia and islands; and sub-Saharan Africa. Having analysed the available data as thoroughly as possible, the study's description of the world as it was in terms of mortality revealed some dramatic findings. In 1990, 98% of all deaths in children younger than 15 years were in the developing world. The probability of death between birth and 15 years ranged from 22·0% in sub-Saharan Africa to 1·1% in the established market economies. Probabilities of death between 15 and 60 years ranged from 7·2% for women in established market economies to 39·1% for men in sub-Saharan Africa, while injuries accounted for 10% of worldwide mortality, a statistic that the researchers noted was “often ignored”. As well as revealing striking patterns of mortality, the study also raised compelling epidemiological questions: why were suicide rates among women in China and South India so high? Why were women in India two to three times more likely to die from a burn, whereas in all other regions combined, men were more likely to die from burns? Why was suicide so common?
The GBD reported that five of the leading killers were communicable, perinatal, and nutritional disorders largely affecting children. A major finding of the study was the importance of non-communicable diseases in global and regional patterns of death: these were revealed to be major public heath challenges in all of the designated regions, with ischaemic heart disease being the leading cause of death in 1990 (6·3 million deaths). This finding ran counter to the accepting thinking at the time that so-called “diseases of affluence” must be higher in better-off populations, yet as Murray explains, “the big surprise was that once you had removed the effects of age structure, the risks of death in the developing world for non-communicable causes were actually higher than in the developed world”.
It would take another 10 years before Murray and Lopez were able to take their analysis down to the country level, but the 1997 papers set out a methodological template that was to establish the GBD as the accepted metric for the worldwide study of health: a formative moment in descriptive epidemiology. The quartet of papers attracted much attention and no little criticism within the wider scientific community. But any criticism was moderated by a broad recognition that seeking a reliable measure of disease, suffering, and death was a laudable aim. Indeed, the study was an affirmation of the researchers' wishes to, in Lopez's words, “change the world” and to develop a measurement framework that could not only improve but also change global health thinking by presenting a reasonably scientific view of the comparative order of risks. Thus for Murray and Lopez the main audience for the papers was a community of actors: academics, health professionals, politicians, and policy makers who could be persuaded by solid science to expedite health improvement policies for the world's poor. Although the 1997 paper contained no randomised evidence, its findings not only established a new kind of measurement metric that could be applied globally, but also influenced the cadence and direction of medical research. Pharmaceutical companies were understandably interested in what the big health problems of the future were going to be and as a consequence where they should be investing their R&D resources. This interest had not been anticipated by Murray and Lopez, but they recognised that the world had an appetite for knowing about future health needs and that GBD data could direct companies towards trial-able hypotheses.
Rarely has an 8-page article announced the inauguration of something as influential as the GBD. The sheer ubiquity and reliance upon GBD-generated data in the subsequent medical literature stands as testimony to its intellectual coherence and importance for the strategic planning of health systems. But back in 1997, Murray and Lopez suspected that it was touch and go whether The Lancet would publish their research as at the time they were fairly unknown, the innovative approach that they propounded was, as Murray describes, “totally out of left field”, and the journal had probably never published four articles from a single study before. With the support of the journal's wide reach and authority however, the GBD papers formed a milestone in the dissemination of knowledge and showed how science could engage with the global health community. Moreover, the first GBD study marked the beginning of a close relationship between The Lancet and the GBD based on a shared humanitarian ideal that global health measurement should inform policy making.

Further reading

  1. Cairns, J. Matters of life and death: perspectives of public health, molecular biology, cancer, and the prospects for the human race. Princeton University PressPrinceton, NJ1997
  2. Keating, C. Smoking kills: the revolutionary life of Richard Doll. Signal BooksOxford2009
  3. Lopez, AD and Hull, TH. A note on estimating the cause of death structure in high mortality populations. Pop Bull United Nat19831466–70
  4. Murray, CLJ and Lopez, AD. Mortality by cause for eight regions of the world: Global Burden of Disease Study. Lancet19973491269–1296
  5. Murray, CJL, Yang, G, and Qiao, X. Adult mortality: levels, patterns and causes. in: RGA Feachem, T Kjellstrom, CJL Murray, M Over, MA Phillips (Eds.) The health of adults in the developing worldOxford University PressNew York199223–112
  6. Smith, JN. Epic measures: one doctor seven billion patients. Harper CollinsNew York2003
  7. Williams, A. Calculating the global burden of disease: time for a strategic appraisal. Health Econ199981–8
  8. World Bank. World development report 1993: investing in health. Oxford University Press/World BankNew York1993

No comments: