Verbal autopsy and global mortality statistics: if not now, then when?
- PDF / 181,474 Bytes
- 2 Pages / 595.276 x 793.701 pts Page_size
- 51 Downloads / 194 Views
COMMENTARY
Open Access
Verbal autopsy and global mortality statistics: if not now, then when? Philip W Setel
More than a decade ago, the World Health Organization pointed out the degree to which deficits in the production, sharing, and use of critical health information hampered evidence-based health development in countries with the poorest health status [1]. This “information paradox” in global health came to refer to the predicament in which countries with the greatest need for timely, accurate, and comprehensive health information - including on causes of death at the population level have had the least access to it [2,3]. Since that time, some (but not enough) improvements in information systems and technologies have begun to fill voids in our knowledge of population health [4,5]. Throughout this period, however, comparatively little attention has been paid to advancing the science and practice of direct measurement of mortality and its causes - particularly among adults [6-11]. In 2000, the state of knowledge on verbal autopsy (VA, a term that covers the design and application of postmortem caregiver interviews, procedures for assigning one or more probable causes of death, and the aggregation and tabulation of population-level mortality statistics based on this data source) centered on a small group of demographers and epidemiologists, many of whom ran intervention trials in various demographic surveillance sites. Almost the entire community of scholarship was on a first-name basis; we could easily gather in a medium-sized conference room, and any of our students or colleagues could become an expert on the VA literature with a week or two of focused reading. Throughout this period, those who remained dedicated to maximizing the potential of VA made steady progress. Yet throughout, a deep and sometimes reflexive scepticism remained that VA could ever really deliver the goods as a reliable measurement tool. The persistent shortcomings in cause of death data, and reluctance to widely embrace VA outside of demographic surveillance Correspondence: [email protected] Measurement Learning and Evaluation for Global Health, Bill & Melinda Gates Foundation, PO Box 23350, Seattle, WA 98102, USA
sites, have forced the global health community to make do with sources of limited coverage and dubious quality and consistency, applying increasingly complex statistical analyses to “correct” for all manner of bias and nonsampling error. The papers in this issue of Population Health Metrics go far in addressing central questions about how much VA can contribute to our measurement of health and health impact. How close to truth can VA ever get? How good is “good enough” for decision-making? Is our putative “gold standard” of medically certified deaths all that robust to begin with - in industrialized or lowerincome countries? Can we make the production of VA data better, faster, and cheaper? What alternatives to demographic surveillance systems exist to permit the collection of mortality data from large, represe
Data Loading...
