An analysis of the Carbsnap data shows that people tend to overestimate the carb count of meats. We see this first by looking at the average error patients make on meat pictures, shown to the right.
Ideally, the mean error across all images should round to 0 — a balanced spread of underestimates and overestimates. Our data instead shows that, on average, people estimated meat pictures as containing roughly 2.5 grams more carbs than they actually had. A 2.5g deviation might look small, but it's an average across many evaluations of many images.
To illustrate this, here are specific meat pictures and how patients and dietitians rated them:
There is a clear disconnect in these figures: patients generally estimate carbohydrate counts much higher than dietitians for the same meats. Be mindful these are general trends — while individual patients may not overestimate carbohydrates, our data reveals that many do.
Please make sure your patients understand the relatively small amount of carbohydrates in most meats.
Box plots
For the first meat figure above, at least 50% of dietitians picked a range between 0 and 5 grams, while at least 50% of patients picked 12 to 40 grams (median 20). A more thorough guide to reading box plots is here.
Keep using Carbsnap
These insights are only as good as the data we get from Carbsnap usage — keep using the app, and tell anyone who could benefit from sharpening their carb-counting skills. Practice your carb counting →