Wearing a Continuous Glucose Monitor for 3 months

Does a CGM help you lose weight? Probably not. But it's fun if you're a nerd!

tl;dr:

• A CGM probably won’t help you lose weight, unless you’re unknowingly binging on carbs

• If you’re diabetic or pre-diabetic ABSOLUTELY get a CGM

• The technology is incredible and you will learn very quickly which foods cause glucose spikes

• They’re prescription-only but you can just ask your doctor for one or get a prescription online, e.g. from AgelessRX.

• Cost is about $70-200/month depending on source/discounts. Shop around!

• If you take ANYTHING away from this post: Walking even 500 steps drastically lowers your blood glucose! Starbucks syrup drinks are sugar bombs!

Author’s blood glucose over the last couple of months

Why wear a Continuous Glucose Monitor if you’re not diabetic?

You don’t need to. I just thought it might be fun because I’m a nerd.

I also hoped that it would maybe give me an indirect insight into my insulin levels, i.e. I might see my blood glucose drop after eating a high-protein, but low-carb meal because the protein would spike insulin, which would, among the protein, also clear glucose from my blood.

The short answer is that I basically haven’t been able to see this correlation at all with most sources of protein in moderation, and even only a “maybe” with eating a ton of fatty meat.

The only moderation “maybe” is cheese and yogurt. Now, dairy is known to be very insulinogenic. If you look up the Insulin Index (a list of foods ranked by how much insulin secretion they cause) yogurt (115) is ranked higher than the baseline white bread (100), with sour cream (98) not far behind and milk (90) higher than donuts (75). Most beef and fatty cheese seems to rank around 45-50.

But even then, eating a pretty big amount of cheese or yogurt doesn’t seem to push down blood glucose much if at all. Sometimes I think I can see a signal in the natural glucose fluctuation, but it might well be a coincidence.

Eating a ton of BBQ meat did seem to push my glucose significantly lower than before, and stayed there for a day. More on this later. But I was eating 3-4lbs of BBQ meat here, so quite a high amount. That’s pushing 350-500g of protein.

Absolutely, positively get a CGM if you’re diabetic or pre-diabetic

It will be well worth the money. You’ll get almost instant feedback on everything you eat. It’s bananas how much sugar is in some common things that you probably consume without thinking much of it. It’s only $70-200/mo even without insurance. How much would you pay not to become, or stop being, type 2 diabetic?

Some people recommend that you count calories for a few weeks just to get an intuition for what’s in things. I did it a few times years ago, and I can confirm that I’m pretty decent at eye balling macros since then.

Wearing a CGM for a few weeks only costs you a few hundred dollars, and you’ll have an intuition about how much sugar is in things for a lifetime.

Basic Anatomy of my Blood Glucose

3 months of blood glucose. You can see a daily pattern and a longer-term pattern.

If we look at this macro picture you can see that my glucose spikes daily (hint: Starbucks) and has a daily low (this is at night). My average seems to slowly glide down over time while on ex150 and rapidly go back up when I am on my regular “dirty keto” diet even just a few days, although this diet doesn’t contain much sugar. It then takes weeks for the average to slide back down.

It seems that being on a mostly-fat (ad-libitum) diet for weeks at a time slowly lowers your average blood glucose. At one point in January I spent most of the day below 70mg/dL, which the CGM considers hypoglycemic and thinks you’re about to die. Yet on a ketogenic diet this seems absolutely fine - I never even had low energy then.

Being on a slightly higher carb and significantly higher protein diet quickly sent the average glucose levels up. Maybe there’s some relation between medium-term average glucose levels and fat loss? I’m not sure I can tell from this data.

Ex150choctruffle, which includes a 100g dark chocolate bar every day, keeps the average significantly higher than regular ex150. This is despite a bar of 85% chocolate only containing about 15g of non-fiber carbohydrates. I suspect it has to do with how slow that sugar gets digested with all the fiber or something, because 15g of sugar should just be a blip, as it is with 15g of liquid sugar (lactose) in my daily Starbucks drinks.

Unintentional Oral Glucose Tolerance Test

Now let’s talk about my all-time high. A few weeks into wearing a CGM, after ordering a flat white at Starbucks, I accidentally received a Frappuccino. Probably just got misheard by a busy cashier. The Frappucino is a blended iced drink with a syrup base, and instead of sending it back I thought I’d just see what the CGM says.

It sent my blood glucose to an all-time-high of over 180mg/dL. For comparison, in 3 months I never even touched 150 at any other time. In fact, there are only 2 times I ever (barely) broke 140.

The Day of The Frappucino Incident

About 2 hours after drinking the Frappucino, my blood sugar was back down to below 100 or even 90mg/dL, depending on when you count (because I didn’t chug the whole cup in one go, of course, but sipped it over 20-30 minutes). My glucose tolerance seems fine after 7+ years on keto and 6 months on a crazy high fat diet.

So a single Starbucks Frappucino skyrocketed my blood glucose by almost 100 points, 40pts higher than my next-highest record. Pretty crazy. I looked the macros up on the Starbucks website later and the Frappucino has between 65-90g of sugar, depending on which one you get and which size. An oral glucose tolerance test contains 75g of glucose. Now I think table sugar is 50/50 glucose/fructose, and those don’t metabolize the same. But I think it’s fair to say that every time you drink a syrup-based drink from Starbucks, you’re giving yourself an oral glucose tolerance test.

Lattes, flat whites, cappuccinos are all way lower in sugar. I get one of these espresso-based drinks with foamed milk pretty much every day, and you can see them as the regular daily spikes in the graph. They contain around 10-20g of sugar and will raise my blood glucose by about 30mg/dL very rapidly, coming down just as fast. A lot of it has to do with how fast/slowly I drink them.

My normal meals on ex150 raise blood sugar so little you can’t even see them at the macro scale.

Day-to-day blood glucose

Three days of blood glucose. Glucose drops very low during sleep, goes up during the waking process, and typically peaks with the ~15g lactose Flat White at Starbucks.

Here we see 3 days of glucose in detail. These are just zoomed in from the macro graph above.

During sleep, glucose is pretty low. During the waking process it rises, known as the “dawn phenomenon” when rising glucagon dumps glucose into the bloodstream, waking you up. Copious amounts of coffee with heavy cream (the orange stripes) don’t do much to change it, often times blood sugar will even go down while I drink coffee. Each of the orange stripes typically represents 2-3 coffees with heavy cream.

My ex150 meal of 150g beef with some vegetables and sauce will spike glucose but not by much. The main offender seems to be the sauce, especially tomato sauce. It’s visible but doesn’t really budge glucose much, maybe going from 74 to 84 or so.

The big spike every day is when I get an espresso based drink with lots of foamed milk at Starbucks, like a latte or flat white. These contain 10-20g of sugar and will easily spike glucose by 30mg/dL, sometimes more. But it comes down just as quickly after that.

You can also tell by the graph that eating 200mg of whipped cream doesn’t do much of anything to blood glucose. No surprise there - there’s only a tiny bit of sugar in heavy cream. It makes less of a difference than the tomato sauce I put in my meal.

Overall no surprise. Sugar consumption causes spikes in blood sugar. Don’t need a CGM for that. Still interesting to see how low and stable a (low-protein) ketogenic diet will keep your blood glucose.

Technical CGM stuff

This might be specific to the Freestyle Libre (2 or 3), the CGM I’m using.

Which CGM to get?

When I started out I only had access to the Libre 2. It’s pretty great but the 3 is a massive improvement. The Libre 2 is the size of a thick silver dollar whereas the 3 is the size of a thin penny. The 2 requires you to tap your phone to it to collect the data. It only saves 8 hours worth of data and I would get a gap every night if I didn’t tap last thing before going to sleep and first thing in the morning. The Libre 3 uses bluetooth to constantly stream the data and doesn’t have this problem. It also reports in 5 minute intervals instead of 15, probably because it can just offload all the data via bluetooth instead of having to store it for 8h.

The other big CGM out there is the Dexcom, and there are various other ones that seem to be more niche. I don’t have experience with any of these because I’ve only used the Freestyle Libre system. I’m told the Dexcom is very similar in capabilities.

Analyzing the data

On the first or second day of using the CGM I realized that the phone app and the graphs from their web app weren’t going to cut it.

The apps are designed for diabetics, mainly Type 1, and their main purpose seems to be to warn them when they get into hypoglycemia or hyperglycemia. As long as you’re not way outside of the acceptable glucose window, they don’t show you much.

Since I’m not diabetic, and spent almost the entirety of the 3 months in the acceptable window, I wanted more detail.

The CGM saves a measurement every few minutes (every 15 for the Libre 2 and every 5 for the Libre 3). That’s 96 or 288 measurements per day. Neither the phone nor the web app let me look at these and analyze them in a meaningful way, generating only the tiniest graphs in which I couldn’t zoom in.

But luckily you can download CSV files from the Freestyle Libre website and so it was easy enough to whip up a script that will parse the data and display it in graphs, the same way I analyze my weight data. In fact, as you might’ve noticed, I plot them in the same graph :)

You can also just put the CSV files into Excel or Google Sheets or any other spreadsheet and visualize them there. That should give you way more granularity.

Accuracy

I cross-compared with a finger prick glucose meter a couple times, and usually it was within 10mg/dL.

Heat seemingly causes the CGM to phantom-spike

At first I was very surprised to see that one of the biggest rises in my daily glucose level was caused by hot showers or baths. At first I speculated that the heat caused some sort of metabolic activity.. maybe that’s why cold baths or sauna are healthy?

But then I read on the internet that CGMs often just falsely show elevated glucose when they get hot. I double-checked with the finger prick method and indeed, the spike seemed not real.

Here are some examples after taking hot showers:

• 80 stick vs. 97+ on CGM

• 71 stick vs 90+ on CGM

• 74 stick vs. 100+ on CGM

Since the CGM is usually within 10mg/dL of the finger prick method, this leads me to believe that the heat really causes phantom spikes.

Cold disables the CGM

Hilariously, the sensor can get too cold to read glucose even when it's not THAT cold outside. This happened to me in slightly below 50F temperatures. Not exactly freezing.

Even short walks are unreasonably effective at lowering blood glucose

Maybe the #1 surprise to me in these 3 months was how insanely powerful walks are.

It basically doesn’t matter how short. My routine for a while was to take a walk after finishing work. I’d walk to the nearby grocery store, buy an energy drink, and drink it on my way to the local park. I’d do one lap of the park and then walk home, sometimes sitting down to text with a friend or have a call along the way. The whole routine would typically take 45-60 minutes.

When wearing the CGM I realized that my blood glucose dropped dramatically in the first few minutes of the walk, before I had even reached the grocery store that marked the beginning of the routine!

So I tested it. Just walking 1 block, which consisted of around 500 steps, brought my glucose down from 135mg/dL to 84mg/dL one time. I don’t know what it is about walking, but a large benefit seems to be just to get the body going.

All those people recommending you take a brief walk after a meal are probably onto something. It’s really quite remarkable.

No need for a 45 minute walk either. It seems that just walking up and down the block already gives you a massive advantage in terms of blood glucose normalization.

I say normalization because my glucose wouldn’t drop below the 80s even on longer walks. If I had very low glucose before, say in the 70s, the walk would actually quickly bring the glucose up to a mid-80s level. It seems that’s where my body wants to be for low-level physical activity, and it’ll bring glucose up or down depending.

Other random findings

Bad sleep makes me worse at handling sugar

A few nights I slept really badly. The next day, the usual Starbucks drink I got seemed to spike my blood glucose by a whole 20mg/dL more than usual.

Lower chronic stress seems to lower blood glucose

Being stressed all week vs. not seemed to make about 20mg/dL difference.

Lying down lowers blood glucose

When I’d lie down for prolonged periods of time, e.g. to read, my blood glucose would drop. Maybe just the body saving energy when it doesn’t require any to keep the body upright? Sometimes this would cause the CGM to freak out, thinking I’d gone hypoglycemic.

Ordering your Latte as a breve (made w/ half & half) reduces the glucose load

You can order pretty much any espresso drink at Starbucks and other places as “breve,” which just means they replace the milk with half & half. That’ll add more fat and reduce the amount of sugar. This seemed to make a 10-20mg/dL difference in my daily glucose spikes. It costs a bit extra though.

Digestion time drastically alters the shape of glucose load

Liquid sugar (lactose) intake - sharp spike, rapid fall afterwards.

My regular ex150 fare is pretty low in glycemic load, with the exception of the lactose in the espresso drinks I get at Starbucks, and some tomato sauce. So most of the little sugar I do get is in liquid form. This is easily visible in the sharp and dramatic spikes in blood glucose that come down just as rapidly.

But on ex150choctruffle, which involved adding 100g of 85% chocolate or similar to my 200ml of heavy cream each day, I noticed that the blood glucose pattern looked very different. The digestion was so slow that glucose didn’t even begin spiking until about an hour after I ate!

Glucose eaten mixed with fiber and fat - blood glucose doesn’t even begin to rise until 1h later, and doesn’t come back down until about 4h later.

This example shows me eating a bowl of chocolate ganache, which is the mixture that chocolate truffle is made of. I took a walk soon after dinner and my blood glucose temporarily went down. But then it started rising by nearly 40mg/dL from 8:30 to 10pm, peaking about 2.5h after I finished eating. It didn’t come back all the way until 11pm, or about 3.5h after.

Remember, an amount of sugar about 5x higher spiked to over 180mg/dL and came all the way back down to 90mg/dL in ~2h in The Frappucino Incident!

I suspect mixing the fiber and fat with the glucose blunted the response, making it a much slower and longer plateau instead of the sharp spike.

What’s interesting: I think many people never go much longer than 3.5h between meals except when they sleep. If you eat breakfast at 9am, lunch at noon, snack at 3pm and eat dinner at 7pm, your body probably hasn’t had a chance to reset your blood glucose even once throughout the day. Unless you were doing keto and only ingesting liquid glucose, of course. But who does that?

Vegetables differ wildly in glycemic load, but all are low

• 1 bowl of fajita vegetable mix spiked glucose a bit

• 1 bowl of okra had practically no impact

• 1 bowl of green beans had practically no impact

• 1 bowl of spinach had no impact at all

I didn’t try any really starchy vegetables.

The medium-term average blood glucose is interesting

Despite eating the same exact meals day after day, my blood glucose kept slowly creeping down over weeks on ex150.

Check out my all-time macro view once again. This time I want to highlight in green the areas where I’ve done ex150 or ex150deli, which was almost exactly the same.

My average blood glucose on January 20th, when I stopped ex150 for 5 days, was below 70mg/dL, which is considered hypoglycemic by my CGM (!). It’s not that crazy if you’re in ketosis though.

As expected, eating more protein and some carbs during the 5 days between ex150 experiments raised my blood sugar considerably and the average was almost 100. But then I went back on ex150 and it never even fell back to an average of 80 for several weeks!

What’s interesting here is that, on ex150deli, my average glucose actually started going up again. ex150deli replaced my usual ground beef with deli meats, including turkey, chicken, roast beef, and ham. I still lost weight, but I do find it interesting that my average glucose started sliding up during those 2 weeks. Suspects? There is SOME sugar in many deli meats (though the amounts are very low). I might’ve eaten more protein, although I tried protein-matching to the ground beef. Or maybe the PUFAs from chicken/pork? The meat was very lean, though. Not sure.

This is very interesting to me. There seems to be some longer-term mechanism. Even eating the same exact meals day after day, week after week, the average or baseline glucose levels can vary drastically.

What could I have possibly eaten during the 5 days off-diet that kept my blood glucose baseline elevated for several weeks?

Things with very little impact on blood glucose

• Cream in coffee. I can seemingly do 3-5 of these in a row and I can’t even tell from the glucose chart.

• Butter

• 150g beef + vegetables (w/o sauce): it's probably mostly the sauce in my normal ex150 meals that causes spikes

• Jocko energy drinks sweetened w/ monk fruit (even 5 in ~45 minutes)

Super Good Energy

Good.

Extremely useful and legit science experiment last: I was very fond of Jocko Go energy drinks for a while. They are sweetened with monk fruit, which is advertised as having no impact on blood glucose. I wanted to test this.

Careful: Do Not Try At Home!
I am a caffeine professional and performed this experiment on a closed circuit.

So, naturally, I purchased all the Jocko Go energy drinks in stock at the local store (5, because I had already bought 1 earlier that day) and downed them all within 45 minutes or so. You can see in the graph that even the combined load barely had any impact on my blood glucose, if any. The small rise visible isn’t more than the natural fluctuation in blood sugar throughout the day.

And I was extremely productive after drinking 6 energy drinks. It adds up to about 600mg of caffeine. Plus the coffee, of course.

Ad-libitum BBQ meat did have a significant impact on my blood glucose

I ended ex150choctruffle by eating ad-libitum BBQ. Mostly brisket, but also ribs and hot links. No sauces, no sides. I took some of the brisket home, and ate more the following day. So you could say it was 2 days of ad libitum fatty meat.

Average daily blood glucose. Note dip and then rise after BBQ.

The red line is the ad-lib BBQ meat. Before that, ex150choctruffle. My average blood glucose drops quite a bit after indulging on the meat, and stays down for almost a day. Then it climbs back up and goes higher than it was before! This despite only eating ad-lib meat for 2 days.

Maybe the steep drop after eating a ton of meat is a sign of lots of insulin being released after eating too much protein, which forced glucose down? But what is causing the high average glucose lasting until about a week after I stopped eating ad-lib meat? Gluconeogenesis? Hm.

Going Forward: Still got a CGM

Although the shiny-new-gadget effect wore off after a month or two, I still like having the CGM. Especially with the Libre 3 now, where I don’t even have to tap it once every few hours to collect the data. I can ignore it for days and it’ll just sit there and record. The size of a thin penny, it just sits on my arm and I never notice it. Even when showering I forget it’s there.

Being able to download the data and quickly dive into any period and look for correlations is great.

And my last refill cost me only $72 - for 2 CGMs or 28 days’ worth! At that price it’s significantly cheaper than my Starbucks habit, so it seems like a no-brainer.