Continuous glucose monitor startups still have to prove their worth

I have rarely, if ever, thought about my blood sugar. I think about sugar on occasion — the dentist demands it, and I get a headache when I eat an entire bag of Haribo dinosaur gummies.

But a new spate of startups wants me, and everyone, to start thinking about our blood sugar all the time. Over the past few years, companies like Levels, January, and Nutrisense started selling programs that promise to help people start managing their blood sugar levels. They sell devices called continuous glucose monitors: small sensors that embed a tiny needle in someone’s body to track the way their blood sugar (their glucose) levels rise and fall.

Continuous glucose monitors (CGMs) are usually used by people with Type 1 diabetes, a chronic condition where the body doesn’t produce the insulin needed to break down blood sugar. For them, keeping track of the amount of sugar in their blood is vital so they can give themselves insulin.

These companies say that people without diabetes should also track their blood sugar levels. They point to research showing that most Americans are metabolically unhealthy, a designation that accounts for blood pressure, blood sugar, waist size, cholesterol, and triglycerides. Keeping tabs on blood sugar, and taking steps to bring it down, can improve metabolic health and help people lose weight.

“There are some basics that, if everyone followed, the country would be healthier,” says Lauren Kelley-Chew, the head of clinical product at glucose monitor company Levels. “Specifically, trying to limit high blood sugar spikes and trying to keep blood sugar within a relatively healthy range.”

These applications are new, and there still isn’t clear evidence that wearing a CGM can help people lose weight or fix metabolic problems. Many experts on obesity and metabolic disorders (like diabetes) aren’t convinced we know enough yet to recommend them. But that’s the pitch — wear a monitor, figure out which foods spike your blood sugar, and avoid them.

I don’t have any medical conditions that impact my blood sugar or my insulin levels. When I eat something that changes my blood sugar, my body regulates it on its own. My friend John, though, lives with Type 1 diabetes. He was diagnosed with diabetes as a small kid and, for most of his life, has monitored his blood sugar with finger sticks around six to eight times per day. However, he’s used a continuous glucose monitor on and off since around 2011, giving him a constant stream of blood sugar data.

John and I wanted to compare what it was like to use the monitor as someone without diabetes, like me, and someone living with Type 1 diabetes, like him — the medical use of the devices versus the wellness, startup approach. So I ordered a kit from Levels, and for the same week in October, we kept track of the information we were getting from our continuous glucose monitors and the things we did in response. Here’s how it went.

Morning

John: I woke up this morning to a blood sugar of 176 milligrams per deciliter (mg/dL), sloping gently upward. Ordinarily not a cause for celebration, in this case, it represented a victory. For the past couple months, I have been adjusting everything — from what I eat at night to my insulin doses — to try to counteract the “dawn phenomenon,” which for me, basically means that my blood sugar tends to climb as I go to sleep, crest in the middle of the night, and then come back down in the early morning. Depending on how steep the rise is, this can mean that I have high blood sugar most of the night and wake up in the morning to the frustrating realization that it’ll be hard to keep my blood sugar “in range” (which, for me, is between 70 and 180 mg/dL) for the target amount of the day — which my doctor has set at 70 percent. Being “in range” for that amount of time is the goal because spending too much time with high or low blood sugar can increase the risk of other health issues, like kidney problems or strokes.

Last night, I ate a big dinner of Indian takeout, setting the stage for what I worried would be a classic night of persistent highs. Indeed, when I woke up around midnight, I was at 272. I decided to add insulin above my usual dose. For a couple hours, it worked — until I started to spike again around 2:30AM. Luckily, I saw it when I woke up to use the bathroom and threw another dose of insulin far above my usual rate at it. For one night, at least, I brought the dawn phenomenon to heel.

Nicole: My alarm went off, and I rolled over to grab my phone. Usually, the first apps I open in the morning are iMessage and Gmail. Today, though, I tapped into the Dexcom and Levels apps. According to the app, my glucose levels were 89 mg/dL.

---

For someone without diabetes, fasting glucose — or blood sugar after not eating all night — is usually between 70 and 100. Someone with prediabetes or diabetes would have higher fasting glucose: between 100 and 126 for prediabetes and over 126 for diabetes.

During the day, most people without diabetes have blood sugar levels between 70 and 140 outside of meals. Things rise with food and fall during the hours after eating. People with diabetes try to stay in a similar range, between 70 and 180, for as much of the day as possible.

But despite those general guidelines, there’s still a lot researchers don’t know about what blood sugar levels, ranges, and responses actually look like, particularly after eating. People without diabetes can have a wide range of blood sugar responses to the same foods. Some people have large swings in glucose levels, while others stay more steady.

Lunchtime

John: I often get low around lunchtime if I haven’t eaten much beforehand, which was the case today. Typically, I expect to spike after lunch on days like this, but it doesn’t always work that way. Today, around 11:45AM, I ate a high-fat, moderate-carb lunch and took a normal dose of insulin for it. By 12:30PM, I had fallen to 52 mg/dL, interrupting the work call I was on with the “urgent low” alarm that Dexcom offers no way to disable. Fat slows down the speed that your body processes carbohydrates, which means that a high-fat meal sometimes “kicks in” and raises blood sugar significantly later. Today, it resulted in a reading hovering around 200 for much of the afternoon. Taking a walk and a couple of correction doses did little to move the needle.

Nicole: I ate half of a bagel and some scrambled eggs for lunch, and the Levels app informed me that a bagel could have a “large” impact on my glucose. Sure enough, my glucose spikes up to 176. The Levels app tells me that “physical activity like a walk can blunt your meal-related spike.” I’m in the middle of work, so I do not go for a walk. But I keep my eye on the app, and after a few hours, my blood sugar levels start to fall again.

---

The main message in the Levels app is to keep your blood sugar steady. The app rewards you for staying “stable” and drops in warnings when things start to spike. Each food logged gets a score for how much it spiked blood sugar. The goal is to have the glucose changes with food be “rolling hills,” rather than steep mountains.

With that goal in mind, the app feeds you insights when you log some foods. It told me bagels were “glucose spikers.” Another day, it told me I should avoid oatmeal because it’s a refined carbohydrate. “Try chia pudding instead,” the app said.

Working out

John: Today is an ambitious day: my girlfriend and I are planning a 25-ish-mile bike ride to and from a pizza place in the suburbs of Chicago. Long periods of exercise are a time when the CGM’s constant stream of data and the granular level on which it can track the rate of change really come in handy. I’m planning to run myself a little high to avoid having to stop for a low — but not too high. For now, I’m around 180, right where I want to be. I’m suspending insulin delivery for about an hour — that will increase my blood sugar, but it has a delayed effect.

• 11AM: We’re underway! After holding steady for a few minutes, I’m starting to fall rapidly into the 140s. I eat a couple of glucose tablets.
• 11:30AM: Now I’m spiking the other way, back over 200. I decide to hold things steady for now and wait for the exercise to do its work.
• 12:30PM: I’m falling again, so I eat a couple more glucose tablets. I hold steady around 140 for a bit.
• 1:30PM: I’m falling again, this time into the 120s, but good news — we’ve made it to the pizza place! When our food comes, I take a sharply reduced rate of insulin in anticipation of the bike ride back.
• 3PM: The reduced rate worked a little too well — we’ve been back on the road for about an hour, and I’m up around 200 again. Since we still have a while to go, I don’t take any insulin.
• 4:30PM: We’re back home and I’m hovering very steadily in the 150s. A whole day of biking without a single low! That’s an accomplishment in my book.

Nicole: My glucose levels are 94, and I head out to the gym. I spend about 45 minutes in the weight room. Things stay fairly steady throughout my workout, and when I leave, I’m at 88. I get home and grab some protein powder (#gains). It’s a citrus lemonade flavor — the only kind I can stomach. There is only one gram of sugar per serving, so I’m not expecting it to do much to my blood sugar levels. And I’m right: over the next hour, I don’t see much of a change.

I’m starting to get hungry, so I grab some plain Greek yogurt and mix in a spoonful of blackberry jam. When I log the meal, the Levels app sends me a pop-up that warns that I should avoid any types of added sugar. “Stick to whole, unrefined foods,” it advises. Then, I get a suggestion — to find foods without added sugar, the app says I should “choose quality trusted brands like Primal Kitchen.” But I don’t see much change in my glucose levels, even with the forbidden added sugar.

---

Exercise can lower blood sugar because the activity helps cells more effectively pull glucose out of the bloodstream. Levels encourages exercise; every time there’s a blood sugar spike, it nudges a user to go for a walk. “There are a few different things that can help most of us when it comes to glycemic control,” Mike DiDonato, the head of member success at Levels,” told me. “One of them is just the power of being active: going for a walk, doing some air squats, whatever is good for you.”

The app also nudges people to make food swaps (like rice for cauliflower rice) and suggests recipes. It also suggests brands, like the screen directing me toward Primal Kitchen. DiDonato says the nudges aren’t paid advertisements — they’re just companies Levels employees like.

Eating pizza

John: I am sitting at 110 but in something of a race against time: I’ve been on a downward slope and need to eat soon to avoid getting low. I’m carrying a bunch of pizzas to a work event. I could use a slice already.

• 12:25PM: I’ve dipped down to 66. I ate two slices of pizza a few minutes ago, but my body doesn’t know it yet. (And the CGM, which lags a few minutes behind, really doesn’t know it yet.)
• 1:10PM: I’m back up to 138. I took a dose of insulin for the pizza later than I otherwise would to allow my body time to recover from the low, so we’ll see what happens when it kicks in.
• 1:50PM: Low again, back down to 68. I eat some fruit gummies.
• 2:45PM: I rose pretty quickly after the gummies and seem to have peaked at 164. Not bad!
• 4PM: Just kidding! Inexplicably, I spiked again over the past hour and am now holding steady at 250. Pizza is notorious for being hard to predict.

Nicole: I’m planning to meet a friend for dinner at a pizza place. It’s a nice night, so I decide to walk there, which takes me about 40 minutes. I order a glass of red wine, and we get a pizza to share with pepperoni, shallots, and peppers. I’m expecting a huge glucose spike. But… nothing happened? I have only a small rise in glucose, and it only goes up to 130. Maybe the walk to the restaurant kept things flat. But I’m thrilled to learn that, according to Levels, wine and pizza is a great choice for dinner.

---

Comparing a cookie

John: Nicole and I decide to both eat 18g of carbs at the same time. She eats a homemade cookie; I eat some cookies from my workplace’s vending machine. I take the normal dose of insulin.

I’m getting low, which is frustrating and seemingly unpredictable. I need to bike to my next meeting soon, so I eat the rest of my cookies, don’t take any more insulin, and wait around for my blood sugar to rebound. I’m a few minutes late.

Nicole: I eat one of the chocolate chip cookies I made the other day (from out of the freezer, the superior way to eat homemade cookies). I feed the recipe into an online nutritional calculator, and it tells me each serving has 18 carbs. The Levels app chides me for my choice. “Processed food is harmful to the body,” it warns. But the cookie causes only a “gentle rise” in glucose.

---

It’s counterintuitive that a food that has the nutritional value of a cookie or pizza — lots of carbs, sugar — wouldn’t cause spikes in blood sugar. Both the cookie and the pizza had lower and gentler blood sugar spikes than the homemade lentil and eggplant situation I also ate during my week of using Levels. That meal, full of veggies and fiber, gave me my biggest glucose spikes all week.

DiDonato says context is important. “Isolated events rarely tell the whole story,” he says. Eating something with fat or protein before eating a cookie might curb a spike. Alcohol can actually decrease blood sugar levels, he says. Blood sugar might act differently at different times of the day. “One of the things we’ve always tried to do is provide context,” DiDonato says.

Photo by Amelia Holowaty Krales / The Verge

Glucose monitoring for people without diabetes is likely more than just a passing trend. Tech companies like Apple and Fitbit are interested in blood sugar, and groups are looking to develop noninvasive ways to measure glucose — which could allow them to be part of smartwatches or other wearable devices. It’s important, then, to scrutinize groups marketing glucose monitors to everyone right now to figure out what that type of information can actually do for people.

After a week of using the continuous glucose monitor and the Levels app, I learned a few things about my body and my blood sugar levels. Bagels spike my blood sugar. So do lentils. Eating a meal after going to the gym caused a lower spike than eating that same meal without going to the gym first.

Overall, my blood sugar stayed within the range of what research shows people without diabetes generally experience: spending the vast majority of the time with glucose levels between 70 and 140, with occasional excursions outside of that range that my body quickly corrects.

Still, armed with that information from the Levels program, I could adjust my dietary and exercise habits to try and keep my blood sugar as low as possible. I could walk before meals and avoid simple carbs. Kelley-Chew from Levels says that type of adjustment would make me healthier in the long run. “What we currently believe is optimal is for blood sugar to stay, in general, below 110,” she says. “And also, for any given meal, to try to limit the increase in blood sugar to about 30 points or less.”

The Levels team bases its recommendations in part on a study showing that glucose levels stay around 110 to 120 in young, healthy adults after meals. Based on that research, it’s “a reasonable goal to strive for,” Levels said in a blog post. The company says fasting glucose levels should be 72 to 85, citing research that shows people with higher fasting glucose levels (even if they’re under 100) have a higher risk of dying of heart disease or developing Type 2 diabetes. Kelley-Chew also says people using Levels report losing weight, having more energy, and seeing improvements in their mood.

But other researchers say that it’s too soon to say if adjusting blood sugar will lead to any health improvements. For someone who doesn’t have diabetes, keeping track of blood sugar response could be a useful educational tool, but it’s not clear if it’ll have major health impacts.

“In terms of overall health, if you’re keeping your blood sugar at 110, I’m not sure you’re going to be affecting long-term outcomes or healthcare costs or quality of health,” says Nicole Ehrhardt, an endocrinologist specializing in diabetes care at the University of Washington.

There also isn’t evidence outside of anecdotal experiences that managing blood sugar spikes could help people lose weight, which is part of the marketing pitch for companies like Levels, Nutrisense, and Signos.

“The idea that a lower blood sugar and not having spikes leads to weight loss has not been seen,” says Mitchell Roslin, a bariatric surgeon at Lenox Hill Hospital in New York City. “We have no idea what normal looks like or what a curve that promotes weight loss is.”

For its part, the Levels team says it recognizes that research is still limited. Still, it says that diets high in sugar can disrupt metabolic health and increase the risk of chronic health conditions. “While aiming for relatively stable blood sugar is not by any means the only answer to optimal health, working toward it very likely has benefits,” Kelley-Chew said in a statement.

If studies end up showing benefits, making changes to keep glucose levels steady using a continuous glucose monitor is difficult. People at risk for diabetes or who have Type 2 diabetes who use continuous glucose monitors tend to only see small improvements in their glucose levels, Ehrhardt says. There’s still limited research to show if wearing these devices leads to behavioral changes for people with prediabetes or Type 2 diabetes that could improve their blood sugar.

Photo by Amelia Holowaty Krales / The Verge

Even for John, the continuous glucose monitor isn’t a panacea for managing blood sugar levels.

The promise of the CGM is that more data can lead to more informed decisions and ultimately narrow the bounds of that unpredictability. But I have a sense of constant unpredictability with Type 1 diabetes that the steady stream of data still hasn’t, at least for me, done much to mitigate.

In theory, blood sugar is determined by a few manageable components — insulin, exercise, food, time — yet to me, it still often feels hard to know what these inputs even are, much less predict how they will interact with one another. I sometimes have a nagging feeling that I could be doing more — that if I just paid closer attention or changed my insulin delivery settings more often or standardized the times at which I eat and exercise and sleep more exactly, I could lower the amount of variability. To an extent, it’s probably true, but it’s frustrating that the more technology there is, the more time I seem to spend thinking about diabetes, not less.

Continuous glucose monitors are also expensive. The Levels program is $199 for each month’s supply of monitors and an additional $199 for an annual membership to the program. People without diabetes who can afford it can pay to use these tools. But at the same time, many people with diabetes don’t have access to them. The costs aren’t quite so high outside of a program like Levels, but for many, they’re still out of reach. “The first thing in my mind is make it accessible to those that are at the highest risk for disease or have the disease already,” Ehrhardt says.

Outside of those criticisms, though, companies like Levels are helping build a body of knowledge around glucose trends in people without diabetes — which researchers still don’t know much about. They’re running studies with users: Levels, for example, is aiming to enroll 50,000 people in a study that will track blood sugar patterns in people without diabetes through their day-to-day lives. “We expect that we will dramatically increase our knowledge about blood sugar patterns, blood sugar, baselines, and just glycemic control and metabolic health in general,” Kelley-Chew says.

It’s still early days — which even people promoting the devices recognize. “We all need to be humble here and understand the fact that research is emerging,” says Kelley-Chew.

Correction November 21st, 1PM ET: This story originally attributed a quote to Josh Crist, a spokesperson, which should have been attributed to Lauren Kelley-Chew, head of clinical product at Levels.