What Bill Gates doesn’t know: real-time data in global health

Joel Selanikio ON 26 February 2013

You may have seen this recent tweet from Hans Rosling — the noted physician, global development data specialist, and data visualization guru:


On one level, of course Dr. Rosling is right that it IS very bad news to find out that there has been no reduction in child mortality in Gabon in 10 years. What I find just as tragic, though, is the unstated assumption that it is normal for child mortality in Gabon to be measured authoritatively only every 10 years (in the DHS survey that he was referring to).

Every ten years??? No wonder they don’t seem to be making any progress! It’s hard to imagine making progress in any activity when you only measure things every ten years.1

And all of global health typically reacts in the same way: most global health institutions and practitioners grew up in the days before the internet and the mobile phone had really expanded into Africa and the rest of the developing world, and they’ve gotten used to extremely infrequent data. In the “old days” — up to the last decade — the only possible way to get data was on paper: it was normal, and understandable, to gather data only infrequently.

The problem is that times and the technology infrastructure have changed — but global health hasn’t changed with it. Despite 10 years of explosive growth of mobile telephony even in rural Africa — and our Magpi mobile data system (formerly “EpiSurveyor”) freely available to all and used by more than 15,000 worldwide — we are still collecting almost all global health data on paper. And in some cases only every ten years.

Really, this has changed from a situation where we were limited by lack of technology and limited infrastructure to a situation where we are more limited by the inability of global health leadership to understand and take advantage of the new reality.

Happily, a few organizations have moved to the leading edge and are using these very cost-effective technologies right now, and they show us a way forward.

The IRC (International Rescue Committee —, for example, is a long-time Magpi user and gives us a great example of how things might be different: how we can achieve real-time data on child mortality, for example, cost-effectively and at scale.

IRC uses the SMS data capture feature of Magpi in Sierra Leone. They have rural community health workers utilizing incredibly cheap mobile phones (typically costing $10 or so — and that is not a misprint) to send child birth and death counts for a population of about 10,000 from their villages on a weekly basis (I highlighted this in a recent TEDx talk).

Yes, weekly. Not every 10 years, or every 10 months, or even every 10 weeks. Every single week.

That is effectively a real-time mortality monitoring system in just the kind of rural environment found in Gabon, yet IRC manages to supplement the very extensive (and incredibly useful) data collection in the DHS survey at that frequency for a total technology cost of less than $6000 a year.

For the whole program. And that is not just the software cost:

  • The servers are operated by DataDyne and so that cost is included in the $6000 cost.
  • The phones are $10 phones.
  • No programmers are required, so that cost is eliminated.
  • No tech consultants are required, so that cost is eliminated.
  • No outside trainers were required — since the system is designed to require little or no training — so only IRC local or international staff already working in Sierra Leone were needed.
  • The tech support is provided by DataDyne as part of the IRC’s paid subscription to Magpi, so that is also included in the $6000 cost — although it should be said that IRC has required almost no technical support at all, and no one from DataDyne has even visited the project.

To expand such a system to give an excellent weekly picture of child mortality for the entire country of Sierra Leone would only cost about $100,000 annually — roughly the annual salary and benefit cost for a single international development consultant.

Given that Gabon has about one-third of the population of Sierra Leone, the cost for implementing a country-wide system for weekly mortality monitoring would be correspondingly less. But even at $100,000 it’s a bargain: we’re talking about a country-wide real-time mortality system that would truly help the health system to understand what’s happening on the ground, and to save more lives.

And one can easily imagine how much better those with such data can plan and implement their programs — designed to decrease child and maternal mortality.

Bill Gates recently wrote about the importance of better tools for measuring data for international development, education, and other purposes in the Wall Street Journal, but in the section of his article on global health the mobile data projects he discusses are all pilots and still-to-be tested technologies, which typically spend most of their budgets on programmers and consultants. That is a model that is simply too expensive to scale. Magpi is utilizing the cloud and mobile to deliver scaleable, real-time data for health and other purposes at incredibly low-cost, right now.

It is ironic that Mr. Gates’ vision of inexpensive and scaled real-time measurement in health, education, and other areas is already being implemented by organizations like IRC (and our other users like Abt and JSI and Camfed and WHO and many others), but Mr. Gates seems to be unaware of it.

I’d love to hear suggestions for how we can change that!



1 Of course, progress may have been made in those ten years: the mortality rate may have fallen, then risen again — or the reverse. We just cannot say: a ten year interval just does not give enough information.