Ki recently hosted a webinar to report on the results of the wasting rally to the data contributors whose data was included in the analysis. This was an important step toward creating a complete feedback loop linking all the ki partners.

Ki’s wasting rally, conducted by an analysis team at Berkeley in coordination with members of the foundation’s nutrition team, combined a total of 37 studies with data on more than 120,000 children. Our understanding of wasting has always been limited by the fact that since it’s relatively rare, sample sizes tend to be very small. Given the size of ki’s carefully curated “super data set,” however, the team was able to conduct a more robust analysis. In particular, they were able to look at incidence, or the rate of occurrence of new episodes of wasting, instead of prevalence, the proportion of episodes in the total population at any given time.

The shift from prevalence to incidence yielded many new insights, one of which was especially surprising and important: it turns out that episodes of wasting are most common from the ages of 1-6 months, whereas prevalence is highest at 12 months or older (some of the children whose wasting episodes begin in their very first months never recover, which explains this seeming contradiction). This finding points the way toward significant changes in how we think about preventing and treating wasting. At present, the guidelines are targeted toward older children, and this analysis may provide the evidence needed to start work to adjust the guidelines to better fit the reality of the situation.

This single result could save and improve lives and change the research agenda in the nutrition field. At the same time, the webinar itself, with a long period for discussion between data analysts and data contributors, is part of a new process that facilitates ongoing knowledge exchange that will improve the way data science and research science are blended to solve problems. We look forward to hosting more webinars as we get more results back from our analyses.