Crowdsourcing death-data: An international health measurement solution?

A new study published online in The Lancet perfectly illustrates the challenges and opportunities presented by what might seem like a simple health priority: Measuring and recording cause of death. By analyzing health records on 122,000 randomly selected deaths in India in 2001, researchers from the University of Toronto find that one disease, malaria, may be killing 13 times more people in India than international health officials currently estimate. The World Health Organization statistics say that 15,000 Indians die from the mosquito-borne disease each year, while the Canadian researchers say the figure is closer to 205,000. The upper and lower bounds of their confidence interval were 125,000 and 277,000 deaths. Malaria accounted for 3.6% of all deaths in the studied population, with each Indian having about a 1.8% chance of dying from the disease before age 70.

Among the reasons for this: 86% of malaria deaths occur outside of hospitals that are tied into national reporting systems (fully 90% of malaria deaths were in rural settings). Further, the researchers fear that other death estimates related to other conditions may also be equally skewed.

In terms of opportunities for innovation, this study raises an interesting issue. Most death rates related to international health are statistically weak guesses. Since autopsies or other definitive findings on cause of death are exceedingly rare outside of the developed world, in many cases deaths go unattributed, unrecorded, or even unrecognized. This means that statistics on rates of death caused by, for example HIV/AIDS, and any related decisions on allocating funding or setting international health priorities are essentially based on educated guess work.

Are there opportunities here to harness information technologies such as smart phones as a means for crowdsourcing death data? It sounds bizarre or macabre, but innovative organizations such as InSTEDD are building expertise on how to get lay-populations with low literacy in culturally diverse, rural settings to report on potential health disasters. Could similar expertise help provide "digital-autopsies," or a questionnaire -based mobile app that allows a relative to provide enough data for a trained reviewer to make an informed declaration of what the most likely cause of death was. The Lancet study used trained reviewers who assessed cause of death by careful review of health records, suggesting that this approach might provide better estimates than are currently available.

Any such system would create a deluge of data that would be costly and time consuming to analyze. But from the perspectives of both compassion and managing costs, shouldn't we do a better job recording what is killing people?

UPDATE: Hear DataDyne's Joel Selanikio talk to the BBC about the role that digital technologies could play. Select the October 26 podcast on Malaria Surveillance in India at http://www.bbc.co.uk/podcasts/series/digitalp

Photo credit: The author, via Flickr