Glucose sensors

How to use a CGM to test your meals

By GyroscopeUpdated November 15, 20267 sources

A sensor is at its best as a meal tester. It can’t tell you much from one number, but it can show you, in your own body, that a cup and a half of rice makes a bigger hill than half a cup, or that the same dinner with a 10-minute walk after it comes back down sooner. The trick is to change one thing at a time and compare you with you.

Why your friend’s food rules don’t fit you

Online, people swap lists of foods that “spike” them. Bananas, oatmeal, sushi. The lists rarely agree, and there’s a good reason. In a study of 800 people who ate 46,898 meals while wearing sensors, the same meals made very different hills in different people.

A sensor is useful for your own list. It’s just less useful for anyone else’s, and it won’t tell you which foods are healthy, only which raise your sugar.

How to run the experiment

Keep the rest steady: the same time of day, similar sleep, no big workout right before. Compare each meal with its own repeat, since a breakfast hill and a dinner hill differ for reasons that have nothing to do with the food.

How to judge a hill

• How high. The peak is the easiest number to see, and the noisiest.
• How long. Minutes above your usual level tell you more than the peak. In our model, built from fingertip tests in healthy adults, most hills come back near where they started within two to three hours.
• How it ends. Some hills dip below where they started before they settle. A real low is under 70 mg/dL with symptoms, so check one with a finger stick.

In trials, eating the carbs last lowered the peak by more than 40% in 15 adults with prediabetes, and 10-minute walks after meals made the rise 12% smaller than one daily walk, 22% after dinner, in 41 adults with type 2 diabetes. Expect your own to land somewhere around there, and maybe somewhere else.

Why the sensor exaggerates the hills

Sensors run high, and more after some foods than others. In 15 healthy adults, a popular sensor overstated the time above 140 mg/dL about 4-fold against fingertip blood, and about 2-fold once each person’s fasting offset was taken out. It also put a fruit smoothie at a GI of 69 when fingertip blood gave 53. And glucose takes 5 to 6 minutes to reach the fluid the sensor measures, so the peak on your phone comes a little after the real one.

Treat the size of a single hill loosely, and the difference between two versions of the same meal as the real result. Repeat a test on another day before you trust it, since one meal can come out differently for reasons you can’t see.

Where to start

If you want ideas for what to test, the 100 meals that raise blood sugar most and the 100 that keep it steady rank common meals by their carbs and GI. What happens when you eat walks through pasta, rice, soda, fruit and a sandwich stop by stop, and the simulator shows what each switch should do before you try it on yourself.