I’d visited about a dozen hospitals across southern India by the time I reached the medical college at Alappuzha, a city snuggled between Kerala’s famous backwaters and the glinting Arabian Sea. The visits had become routine: Collect data from Brilliance, our phototherapy lamp for treating babies with jaundice; interview doctors and nurses; glean the hospital’s diagnosis and treatment methods.
At the overtaxed Medical College in Alappuzha, India, doctors double up babies under phototherapy lamps.
Alappuzha demanded to be different. The neonatal intensive care unit (NICU) there was at least twice as busy as any I’d seen. In fact, several phototherapy lamps pulled double duty, with two softly sleeping infants under each. The doctor said they treat multiple babies simultaneously about 75 percent of the time.
Is Alappuzha an outlier? Is the doctor’s offhand estimate anywhere near correct? Is there some variable, easily observed from an office in San Francisco, that could have predicted the medical college’s extreme busyness?
These are not merely academic questions. D-Rev spends a lot of time and money measuring the impact of Brilliance. Some questions are basic: How many units have we sold? Where are they in use? Others are far more complicated: How many babies have we treated? How many deaths have we prevented? The answers aren’t just nice-to-haves, destined for brochures. They are numbers that can motivate employees, sway donors, and influence the next generation of products.
Ours is a data-centric age—even wordy professionals like journalists and English historians have nuzzled up to numbers—and there has never been a greater need to quantify our work. This is also a techno-centric age, with an emphasis on low-cost, scaleable solutions.
International non-profits, which often work in places more challenging than, say, Palo Alto, are developing impact-assessment methods that, at least philosophically, wouldn’t be out of place at Facebook HQ. Charity:Water, with an investment from Google, is building durable sensors to wirelessly report when wells run dry or malfunction. The Public Lab has since 2010 used weather balloons and off-the-shelf cameras to cheaply make maps that convey on-the-ground conditions during events like the Gulf oil spill. Farming insurer Kilimo Salama receives data from weather stations in Kenya to automatically determine when conditions are bad enough to warrant a payout.
But no technology is able to capture the near-infinite variability of real life. In Erode, a modest city in Tamil Nadu, I spoke to a doctor whose half-year-old Brilliance unit had barely been used; his practice is new, he explained, and most of his expectant mothers haven’t given birth yet. At a public hospital in the pilgrimage town of Thrissur, doctors treat jaundiced infants for two or three times longer than elsewhere. The hospital’s lab, they explained, is unreliable; doctors purposefully over-treat babies because they have no sure way to confirm when bilirubin levels (the toxin that causes jaundice) reach safe thresholds.
But no one can visit every hospital or field site. The big, unanswered question is how to account for the randomness of life in the headline numbers that indicate whether we’re moving in the right direction. D-Rev thinks about this a lot, and collaborates with academics, doctors, consultants, and others to get it as right as possible.
In military or scientific terms, hospital visits like the ones I’ve done might be called “ground-truthing”: Going out into the field to ensure assumptions conform to reality. It’s not sexy stuff. It doesn’t involve wireless chips or high-altitude cameras, but conversations, long train rides, language barriers, and other challenges. It doesn’t yield crisp, structured data, but complexity and snarls of meaning. The difficulty makes it all the more important and is why D-Rev invests in it: The best way to get clean, meaningful results is to understand the messy world underlying it.