Tidepool Releases First Analysis from Big Data Donation Project

By Amelia Dmowska, Maeve Serino, and Adam Brown

What does donated diabetes data tell us about time-in-range across different age groups?

A few months after launching its Big Data Donation Project, software nonprofit Tidepool has analyzed data from 354 people with type 1 diabetes on continuous glucose monitoring (CGM). The Big Data Donation Project lets people anonymously donate their CGM data to Tidepool, who then partners with researchers and device makers to use the donated data to support new diabetes innovations and research.

Donated CGM data can also help provide benchmarks – e.g., how much time do people in different age groups spend in range (70-180 mg/dl), in hypoglycemia (less than 70 mg/dl), and in hyperglycemia (over 180 mg/dl)? The graphs and charts in Tidepool’s blog post break this down by different age groups, with some very interesting findings. For instance:

Time-in-range (70-180) came in at around 70% in the entire group, with the highest rate in 65-85-year-olds (83%!) and the lowest rate in 1-5-year-olds (57%).

Average CGM-measured glucose levels were highest in children, peaking at an average of 174 mg/dl in 1-5 year-olds, and lowest in older groups, with an average of 132 mg/dl in 65-85-year-olds.
Hypoglycemia (less than 70 mg/dl) was highest in 30-39 year-olds, who spent approximately 6% of the day with low glucose levels (nearly 90 minutes!). On average, most groups spent roughly 3%-4% of the day with low glucose levels.
Hyperglycemia (over 180 mg/dl) was highest in younger populations, with 1-5 year-olds spending an average of 40% of the day above target. Hyperglycemia was lowest in 65-85-year-olds, who spent just 14% of the day above 180 mg/dl.

It is critical to note, however, that the donated data comes from a self-selecting group of people with access to CGM and a willingness to donate their own data, so the data could be biased. Indeed, the average glucose levels reported in Tidepool’s data set are much lower than those reported in the T1D Exchange Registry – roughly 1-2 A1c points lower than their age-group average. For more representative data, it would require a wider variety of people to join the Big Data Donation Project. To learn more about how you can help with this initiative by donating your own data anonymously, click here – it requires clicking just one button in Tidepool’s web-based app.

Still, Tidepool’s data provides insight into diabetes management in the real world. Data like this, especially when it is more representative, might even help guide targets for outcomes beyond A1c. For instance, while time-in-range, hypoglycemia, and hyperglycemia CGM thresholds are now agreed on (70-180, less than 70, and over 180 mg/dl), there is still no consensus on how much time-in-range should be the goal. Real-world CGM data might help refine and personalize future recommendations for these measures. (Read more about reaching consensus for measuring outcomes like time-in-range here.)

Tidepool believes that people with diabetes own their data, so anyone who donates has full control over how their data are shared.