#40 Metformin For Patients With Type 1 Diabetes & May 2017 Update on My T1D Management

I would expect that the majority of those with diabetes, including type 1 diabetes (T1DM), have at least heard of the medication, metformin. It is primarily used to lower blood glucose (BG) in those with type 2 diabetes (T2DM), but does so by addressing insulin resistance, the primary defect in T2DM. However in recent years, it has been used “off-label” for persons with insulin resistance that leads to other conditions including pre-diabetes, severe obesity, polycystic ovarian syndrome (PCOS), cancer, and T1DM.

Its use in T1DM has been targeted to those with signs of insulin resistance which include hypertension, elevated serum triglycerides (>150 mg/dl), reduced HDL-C (<40 mg/dl in males or <50 mg/dl in females), overweight/obesity or elevated waist-to-height ratio (>0.5) (which includes about 25% of those with T1DM). Although any of these signs can be associated with insulin resistance, the more signs one has, the higher the risk of having insulin resistance. The term “double diabetes” has been used to describe those with T1DM who are also insulin resistant. Insulin resistance as has been reviewed in detail in my blog post #22 primarily in the context of prediabetes and type 2 diabetes. But you may ask, “How does a person with T1DM become insulin resistant?” You probably won’t hear very many provide an explanation for it, but here is mine. I think it is a combination of a diet high in refined (processed) carbohydrates and sugar with the use of exogenous insulin to cover those carbohydrates. Exogenous insulin is far from physiologic and there are many hours of the day or night where there is insufficient insulin to suppress hepatic (liver) glucose production leading to elevated BG (glucotoxicity): one of the proposed mechanisms of insulin resistance in T1DM. Additionally, there are hours of the day or night where insulin levels are excessive which can also contribute to insulin resistance. But most importantly the combination of taking lots of dietary carbohydrate, to which those with T1DM are intolerant, with lots of exogenous insulin, directly leads to insulin resistance just as it would in a person with prediabetes or T2DM (in their case the insulin would be endogenous in response to the dietary carbohydrate).

The use of metformin for cancer is still in the animal research arena primarily, but originates from the observation in population studies that patients with T2DM taking metformin was associated with reduced incidence and mortality rates of cancer compared to those not taking metformin.

As far as drugs go, metformin is particularly useful in addressing the problem of insulin resistance. Metformin improves insulin resistance by decreasing hepatic glucose production and intestinal glucose absorption and improves insulin sensitivity by increasing peripheral glucose uptake and utilization. There have been numerous small studies of using metformin in patients with T1DM. A systematic review from 2010 found that “Metformin was associated with reductions in: (1) insulin-dose requirement (5.7–10.1 U/day in six of seven studies); (2) HbA1c (0.6–0.9% in four of seven studies); (3) weight (1.7–6.0 kg in three of six studies); and (4) total cholesterol (0.3–0.41 mmol/l in three of seven studies). Metformin was well tolerated, albeit with a trend towards increased hypoglycaemia.” A meta-analysis of eight randomized-controlled trials published in 2015 came to similar conclusions: “Metformin was associated with a reduction in daily insulin dosage, body weight, total cholesterol level, low-density lipoprotein level, and high-density lipoprotein level but an increase in risk of gastrointestinal adverse effects compared with placebo treatment in T1DM patients. No significant difference was found between the metformin group and the placebo group in HbA1c level, fasting plasma glucose level, or triglycerides level. No significant difference was found between the metformin group and the placebo group in the risk of severe hypoglycemia or diabetic ketoacidosis.” Note: I had access only to the abstract so I can’t give you any more details from this meta-analysis.

I hope it goes without saying that insulin resistance is expressed in those with a genetic tendency who are also exposed to an adverse environment, in this case, dietary carbohydrate consumption, particularly refined (processed) carbohydrates and sugars, in excess of one’s tolerance. This may also require excess calorie consumption as well. However, there is good evidence to suggest that refined (processed) carbohydrates and sugars stimulate appetite and fat deposition (via the signal of increased insulin). Thus, there is a link between refined (processed) carbohydrate and sugar consumption and excess calorie consumption. It has also long been known that physical activity affects insulin resistance. Other things being equal, sedentary behavior increases the likelihood of insulin resistance whereas regular physical activity decreases it. Thus, the first-line treatment for insulin resistance should be carbohydrate restriction, shedding of excess body fat (if needed), and regular physical activity before initiating any drug therapy.

Personally, I have been contemplating trying metformin for my T1DM for many years, but have not done so because I have never had any signs of insulin resistance (see above). More recently however, I have decided to give it a try primarily for the purpose of seeing if metformin might further reduce my insulin requirement which could, in turn, further reduce the frequency of hypoglycemia. Since my insulin requirements do fluctuate significantly over time, I estimate that it could take many months (up to 6 months possibly) to understand whether or not metformin has reduced my insulin requirement. I might be surprised sooner if the effect is more dramatic than I anticipate of course. However, if I am already insulin sensitive, metformin may not improve it much further. So on June 1, I will start metformin at 250-500 mg twice daily with meals and possibly eventually increase to a maximum of 1,000 mg twice daily to see if there is any reduction in insulin doses or hypoglycemia. Needless to say, anyone contemplating a similar experiment with metformin should discuss that with their own physician.

Glycemic Management Results for May 2017

May 2017 glycemic results were rather typical despite travel and a back strain that eliminated exercise for 6 days this month. This also changed my exercise to a less insulin-sensitizing activity (walking instead of weightlifting) for several more days which in turn affects BG and insulin doses. My total daily insulin dose ranged from a high of 42 IU/day to a low of 24 IU/day. It still remains baffling how my insulin dose can vary so much in the span of one month. These changes in insulin doses are in response to both hypoglycemia (leading to a reduction in insulin doses) and hyperglycemia (leading to increases in insulin doses), but I had more hypoglycemia this month resulting in insulin dose reductions for most of the month. Fortunately, none of the hypoglycemic episodes were symptomatic. I also checked blood ketones twice this month, both values happened to be the same at 1.2 mM beta-hydroxybutyrate. I did not measure breath ketones this month.

Below are my mean blood glucose (BG) values, mean insulin doses, and BG frequency distribution for April 2017 compared to previous time periods.

Post 40 Means Table

As presented in blog post #15 exogenous insulin cannot mimic normal insulin secretion, so persons with type 1 diabetes (T1DM) should not expect to have truly normal BG values. They just need to be low enough to prevent long-term complications and not so low as to cause unpleasant hypoglycemic symptoms, brain damage, seizure, injury, coma, or death. I have set my target BG range at 61-110 mg/dl because values in this range are not likely to lead to harm or complications of T1DM. Your target BG range should be determined with your physician because one size does not fit all. Normal BG is 96 ± 12 mg/dl (mean ± standard deviation (SD)) and coefficient of variation is 13% which is the weighted mean from these two studies (here and here) of continuous glucose monitoring in healthy subjects. The standard deviation and coefficient of variation are measures of BG variability which I believe are important in T1DM. However, be advised that clinical outcomes in type 1 diabetes (i.e. microvascular and macrovascular complications) have only been documented to correlate with measures of mean BG, particularly HbA1c. This does not mean BG variability is not important, but it just has not been documented to correlate with outcomes and complications of T1DM. Achieving a normal standard deviation or coefficient of variation in T1DM would be difficult, if not impossible, with current exogenous insulin therapy (injected or pumped). I hope that adding a continuous glucose monitor (CGM) to my therapeutic regimen will improve my BG variability and thus the standard deviation and coefficient of variation. I plan to get the FreeStyle Libre CGM as soon as it becomes available in the U.S. Monitoring the standard deviation and/or coefficient of variation and finding ways to improve them to the best of one’s ability is desirable in my opinion. Following a low carbohydrate ketogenic diet is one such method of reducing BG variability, mean BG, insulin doses, and hypoglycemia. A ketogenic diet may also provide an alternate/additional brain fuel in the form of ketones to protect the brain when BG does go low. The alternative energy that ketones supply to the brain may prevent or blunt the sympathoadrenal response to hypoglycemia which in turn reduces or eliminates the symptoms of and harm from hypoglycemia. This hypothesis needs to be tested before it can be stated as fact. Having BG close to normal most of the time (some of which are hypoglycemic) also minimizes the symptoms of mild hypoglycemia and potentially the harm from hypoglycemia as well due to lack of activation of the sympathetic nervous system and adrenal gland responses to hypoglycemia i.e. sympathoadrenal-induced fatal cardiac arrhythmia, see here.

Below are my BG readings along with exercise type and time for May 2017.

Post 40 Blood Glucose and Exercise Graph

The table below shows the BG variability results for current and previous time periods. The percentiles (10th, 25th, 75th, 90th) on the right show the spread of the BG readings about the median. The interquartile range, the difference between the 75th and 25th percentiles, is a measure of BG variability. In the middle of the table are the %Time in three BG ranges: %Time BG < 61 mg/dl and the mean BG during that time, then %Time BG 61-110 mg/dl and the mean BG during that time, and %Time BG > 110 mg/dl and the mean BG during that time. The other measures of BG variability were defined and explained in blog post #10.

Post 40 Variability Table

The actual daily insulin doses and daily insulin dose totals are shown in the graphs below. I had to take multiple extra rapid-acting insulin doses to correct hyperglycemia and the breakfast and dinner rapid-acting insulin doses increased during the first week of the month. Then BG began decreasing with more hypoglycemia require progress reductions in rapid-acting insulin doses. I made small changes in my basal insulin doses based on the fasting BG results as usual. I still find it interesting that my insulin doses vary so much over time for reasons that I largely do not understand. Again, this is due to the very nature of exogenous insulin therapy and the effect of my exercise on insulin sensitivity.

Post 40 Insulin Dose Graph

I am omitting my Ketonix breath acetone results this month since I did not take any measurements in May.

In June, I will continue olympic weightlifting most days while trying to avoid injury and overtraining (by adjusting the load (intensity times repetitions) up or down) and do aerobic exercise (swimming, rowing, walking, or cycling at low intensity for ≈ 0.5 – 2 hours) the remainder of the days.

My Thoughts About Management of Type 1 Diabetes With A Ketogenic Diet

My goal of glycemic management in T1DM with a ketogenic diet is to keep BG as close to normal i.e. 96 ± 12 mg/dl (mean ± SD) as is safely possible (i.e. avoiding hypoglycemia) to avoid diabetic complications, a reduction in lifespan, and unpleasant symptoms of as well as injury and death from hypoglycemia. For me, a well-formulated whole-food nutrient-dense ketogenic diet (see blog post #9 for more details), daily exercise, frequent BG measurements, and lower insulin-analog doses (Humalog/Lantus) have improved my glycemic control, hypoglycemic reactions, and quality of life. My current version of ketogenic diet has changed slightly since I last wrote about it in detail in blog post #9. Since that post, I have eliminated dairy and decreased my fat intake to further improve my body composition so as to be able to compete in masters olympic weightlifting in the 77 kg weight class without having to think about when and how much I eat before weigh-in.

My current diet looks like this.

What I Cook & Eat

•Beef, grass-fed, including meat (85% lean), heart, liver, and kidney (liverwurst)

•Fish, mainly wild Alaskan salmon

•Canadian bacon (uncured)

•Lamb occasionally

•Chicken & Turkey occasionally

•Eggs (from chicken)

•Non-starchy vegetables (about 5% carbohydrate content by weight) including Cabbage (Red, Green, Napa), Kale, Collard Greens, Spinach, Bell Peppers, Raw Carrots, Leeks, Onions, Brussels sprouts, Home-made Sauerkraut from Red Cabbage, Bok-Choy, Broccoli, Cauliflower, Yellow Squash, Zucchini, Cucumber, Lettuce (Iceberg & Romaine), and some others.

•Fruit – Avocado, Tomatoes, Olives, Strawberries, Blueberries, Blackberries, lemon juice on fish and salads

•Nuts & Seeds – Pepitas, Macadamia, Brazil, Pecan, Walnut, Pistachio, Cashew.

•Note: I developed an intolerance to milk prior to my diagnosis of T1D. I did try heavy whipping cream after starting my KLCHF diet, but am also intolerant of it. I do tolerate butter, but wanted to decrease my fat intake, so eliminated all dairy including cheese and yogurt.

What I Drink

Water (filtered by reverse osmosis), Unsweetened Tea & Coffee

What I Don’t Eat

•Grains – Wheat, Corn, Rice, Oats (there are many more) or anything made from them, which is too numerous to list here. Gluten is a protein present in a number of grains (all varieties of wheat including spelt, kamut, and triticale as well as barley and rye.) which can cause a number of medical problems for a significant portion of the population with gluten sensitivity or celiac disease. In my case, I avoid them due to their carbohydrate content.

•Starchy vegetables – potatoes, sweet potatoes, yams, most root vegetables (turnip root okay), peas

•Legumes – peas, beans, lentils, peanuts, soybeans

•High sugar fruits – includes most fruits except berries, see above.

•Sugar and the fifty other names used to disguise sugar.

•Vegetable Oils (really seed oils) – Canola, Corn, Soybean, Peanut, Sunflower, Safflower, Cottonseed, Grape seed, Margarine & Butter substitutes, Shortening.

•All Processed Food-like Substances i.e., most of what is in the grocery store.

•I avoid restaurants except when traveling, and then order fish or steak with plain steamed non-starchy vegetables (no gravy or sauces that typically contain sugar, cornstarch, or flour) or salad.

•Refined, but healthy, Fats – I have eliminated refined fats from my diet including butter, coconut & olive oils.

What I Don’t Drink

•Colas (both sweetened and unsweetened).

•Fruit Juice except small amounts of lemon juice.

•Alcohol (can cause hyperglycemia or hypoglycemia in persons with diabetes).

•No artificial sweeteners, don’t need or like them.

A large part of my fat intake comes from nuts & seeds which hypothetically could result in potential adverse consequences from omega-6 polyunsaturated fats in the opinion of some low carb advocates. However, I have yet to see any studies that show harm from eating nuts & seeds. As a reference, the average fat breakdown of the seven nuts & seeds that I eat daily is 33% polyunsaturated, 52% monounsaturated, and 15% saturated fat. However, when my entire diet is analyzed, 26% of my fat intake is from polyunsaturates, 56% is from monounsaturates, and 18% is from saturated fats. When my diet is broken down by macronutrients, I consume 170 grams of fat (or 68% of my total daily calories), 70 grams of carbohydrate, 30 grams of which is dietary fiber (or 12% of my total daily calories), and 110 grams of protein (or 20% of my total daily calories). In calories, it totals to 2,250 kcal/day.

My exercise regimen and its resulting varying insulin sensitivity and hormonal changes actually makes glycemic management more difficult i.e. challenging, but I enjoy exercise and feel it has other health and lifespan-extending benefits. Hopefully, my BG values and variability as well as my lower insulin doses that result from my ketogenic diet and exercise are close enough to optimal to avoid any reduction in lifespan, diabetic complications, and harm from hypoglycemia. Only time will tell.

Till next time …


  1. Alex Romayev

    Hi Keith,

    I was wondering if you had more information or just had an opinion on the MGM trial that’s about to go into phase II clinical trials (http://www.faustmanlab.org)? It seems that the vaccine is able to restore at least some beta cell function and (my thinking) in combination with a ketogenic diet, might just be enough to significantly reduce if not completely avoid the need for injections.



    • Keith Runyan, MD

      Sorry Alex, I do not have any additional knowledge about the trial. I am just glad research is being conducted to find a way to reverse or mitigate type 1 diabetes.


  2. ÉRIC

    Hi, Keith Runyan.
    I have recently found some studies (pubmed) which report euglycemic ketoacidosis.
    Wouldn’t we expect a higher risk of this kind of ketoacidosis with a low carb diet?



    • Keith Runyan, MD

      I read the articles you referenced. I will refer to them as 1. through 4.
      Your question is would a low carbohydrate diet increase the risk of euglycemic diabetic ketoacidosis. If we limit the discussion to patients with insulin-requiring T1DM or T2DM who follow a well-formulated low carbohydrate ketogenic diet and maintain normal blood glucose (70-100 mg/dl) most of the time and do not develop an illness e.g. infection, heart attack, or stroke and do not have nausea or vomiting that leads them to reduce all food intake and therefore reduce insulin intake, then I would say it is very unlikely, but not impossible. However, it is so unlikely that it should not be a reason to not utilize a well-formulated low carbohydrate ketogenic diet for the management of insulin-requiring T1DM or T2DM.
      Article 1. as you point out not related to the topic we are addressing.
      Article 2. This article references to case reports of patients presenting with illnesses, not the topic of your question. Case 1 had a Bartholin’s gland abscess and Case 2 had acute pancreatitis. Case 1 had a blood glucose of 191 mg/dl which I do not consider to be euglycemic. Neither patient was following a low carb diet.
      Article 3. This is a case series from 1973. They defined euglycemia as a blood glucose (BG) than 250 mg/dl which meets the criteria for diabetic ketoacidosis (DKA) without needing to add the term euglycemic. There is no mention of any of these patients following a low carbohydrate ketogenic diet.
      Article 4. Again the authors define euglycemia as <250 mg/dl glucose which does not fit the profile of a person following a well-formulated low carbohydrate ketogenic diet to maintain a normal blood glucose (70-100 mg/dl) most of the time. The authors give their list of reasons why euglycemic diabetic ketoacidosis might develop, but do not list a low carbohydrate diet as a possible cause.

      In summary, none of the articles presented provide any useful information regarding your question. My opinion again is that it is very unlikely, but not impossible, that a well-formulated low carbohydrate ketogenic diet designed to maintain normal blood glucose (70-100 mg/dl) most of the time in the absence of an illness would increase the risk of euglycemic (blood glucose (70-100 mg/dl)) diabetic ketoacidosis in persons with insulin-requiring T1DM or T2DM. When BG is elevated above normal (70-100 mg/dl) in persons with insulin-requiring T1DM or T2DM, this by definition represents an insulin-deficient state. An insulin-deficient state is required to develop DKA. This is why I take extra insulin when BG is elevated except perhaps if I am about start some physical activity that usually decreases BG. Thus, having abnormally high BG increases the risk of DKA and leads to longterm diabetic complications: two reasons to avoid hyperglycemia. I hope that answers your question.


  3. ÉRIC

    Thank you again for your time.

    I was told by my doc here in Spain that the LCHF would CLEARLY increase the probability of this kind of ketoacidosis. He couldn’t explain why (nor he defined this ketoacidosis), but I found it quite worrying.

    In the second article I just took the “He admitted to abstaining from any solid food intake on the previous day”statement (similar in patient 1) as a weak relation to the LCHF diet. It is obviously not the same, but as my doctor was so sure I just wanted to discard.

    Your last paragraph is very interesting. I wish I would have had an answer with this precision from my doctor.


  4. ÉRIC

    Hi again, Keith Runyan.

    I have been thinking for a while and I would be pleased if you gave me your view of another topic also related to DKA.

    I practice endurance sports (running and MTB), and when I followed my doctor’s advice (eating 20g-30g of CH each 30 minutes, although in the end I finished eating 15g every 45-60 min trying to avoid hyperglycemia) I found that my BG reached high levels (among 160 and 220 mg/dl) for, maybe, 40 minutes. It almost never went under 140 mg/dl because of the CH intake frequency. Of course, my performance was poor.

    The question is: isn’t there more likelihood to develop DKA in this situation? I mean: doing sport with high BG. I asked my doctor several times, and he answered that no way, because that situation is normal when we do sport.

    Thank you once more.


    • Keith Runyan, MD

      I have covered this topic in several blog posts and in my book, The Ketogenic Diet for Type 1 Diabetes also available on Amazon in print. If exercise is intense, blood glucose (BG) can increase normally due to the hormones released in response to intense exercise. By intense, I mean sprinting and all-out efforts. Otherwise, the increase in BG provides no advantage, may interfere with performance, and if it occurs often may lead to diabetic complications. Whether that could lead to DKA is doubtful, but possible I suppose. I have never seen or read about a case of DKA in this situation.

      However, hyperglycemia with exercise is not necessary to avoid hypoglycemia. That is where a ketogenic diet is an advantage. Being fat-adapted allows dietary or body fat to be used as a source of energy for exercise more effectively than a person who is carb-adapted. Hypoglycemia can occur during exercise on a ketogenic diet and glucose tabs should be available during exercise and blood or interstitial glucose levels should ideally be monitored during exercise to avoid hypoglycemia. This approach will minimize hyperglycemia and the amount of glucose supplementation needed. I used to do triathlon training and about 90% of my workouts did not require any glucose supplementation despite starting the workout with a normal or near-normal BG.


    • Brian Lucido

      Hi Eric – I do a ton of endurance and ultra endurance exercise, so I wanted to chime in. First, everything Keith says above reflects my experience (i.e. intense efforts raise blood glucose). During moderate exercise, blood glucose can remain stable for a while, or it can go down quickly. If I’m going solely on basal (i.e. haven’t had a Novolog bolus in 3 hours), then BG will remain stable for about 2-3 hours at moderate effort – without ANY food. By moderate effort, I mean 50-70% of maximum heart rate… After that time, the BG will start to drop. Once it starts dropping, I’ll need to eat carbs on a regular basis for the rest of the ride. I assume that the drop occurs due to a depletion of glycogen in the liver/muscle. That same moderate effort, however, with even 1U of Novolog onboard will cause a rapid drop during moderate exercise. In that scenario, your doctor’s advice is good. Obviously, everyone reacts differently, but I’ve read a lot of stuff online suggesting that this is generally the way things work in most individuals. If you can get your hands on one, a Continuous Glucose Monitor can be very helpful in ascertaining caloric needs during exercise. One other thing: when you do take on your carbs during exercise, it can be good to just eat a little at a time. A lot of the “easy to carry” energy bars pack some 20g – 30g of carb. Rather than eat that all at once, I’ll take a small bite, ingesting the bar over the course of an hour in about 4 increments. If you’re like me, and running a BG of 85… and you ingest 30g fast carb all at once, your BG will quickly increase by 150mg/dL – taking you to 235 mg/dL. Even though you will likely be able to clear that with continued exercise, you’re not doing your body any favors by being that high during exercise. Above 180mg/dL, your kidneys are going to try and get rid of the excess glucose – taking water with them. It’s hard enough to stay hydrated during exercise as it is! Hope that helps.


  5. ÉRIC

    Thank you both for the answer (I have also found the book on amazon.es) and for the advices. I use the Freestyle, even if while riding it usually ends to be quite inexact (for example, 190 mg/dl on Freestyle, 120 mg/dl BG).

    I have already experienced that BG is very steady for 3h (up to 85% VO2max). In my case, I may use 1U Novolog for breakfast (until today I use basal insuline exceptionally) and it’s not dropping faster. So I think I should try again 5 or 6 hour routes following your guidelines. Anyway, I will try to improve my fat-adaptation looking for some of the advantages Dr. Keith Runyan underlines.

    Thanks again.

    PD: fantastic bike routes, Brian.


    • Keith Runyan, MD

      Eric, thanks for that comment. As a person waiting for FDA approval of the Abbott Freestyle Libre insterstial glucose monitoring system in the U.S., I would be interested in hearing more about the accuracy of the system compared to blood glucose readings and whether you find the device useful despite its inaccuracies.


      • ÉRIC

        Hi, Keith Runyan.

        Of course. I have been using it for 2 years (from it’s first approval in Europe). I would remark this:

        1) Price: 60 euros in Europe for each sensor. 60 euros for the reader. You also have the app.

        2) Each sensor lasts 14 days. There is no way to extend this life.

        3) VERY easy to put on. I have used about 50 sensors until now and I didn’t have any problem with the application on my skin. It gets stuck reliably to your skin. 14 days later, when you remove it, you will find how stuck it still is.

        4) Although it may be tightly adhered, bumping into doors is possible. And then you will lose it. You have to be careful and put it right on the back of the upper arm.

        5) When swimming, it may be detached. That is a problem (I have lost 2 of them this way). I use a compression bandage to protect the sensor. And you now have specific products for the same purpose (freestylesticker.de).

        6) The sensors don’t use bluetooth but NFC. So if you want to check your interstitial glucose with your Freestyle reader you will have to “touch” the sensor with your phone or the reader. That means you can’t see any information on your phone (or your reader) unless you approach it to the sensor.

        7) Accuracy:
        7.1. In case the error rate exceeds 30% CONSISTENTLY (thats a bit subjective, I know, but to my experience you need more 5 or more wrong numbers to convince Abbott), Abbott will replace your sensor for free. Until now I have had 5 or 6 units replaced. You have no more than 4 days to report this problem. If you take more time, you won’t be able to use warranty.
        7.2. Compared to the BG data, the error rate is usually over 10%. They have improved it a lot since the beginning, though. And, of course, normally the higher the rate of change on your BG, the higher the error rate of the Freestyle. I can send you a sample of my data (5 references per day in the last two years), if you need it.
        7.3. The error rate variation from one sensor to another can be high. So you can have 1 sensor which offers EXCELLENT accuracy, but the next can have a 20% error rate.
        7.4. This is very important: if a sensor is wrong, you will notice it from the beginning. The first 12h it will show a GREAT error rate. But that is the time it needs to get calibrated. After that time, values should be much better. If they are still wrong for 2 days, then they have to replace it.
        7.5. Accuracy normally improves as you use it. Normally, not always. BUT the last 2, maybe 3 days, it will offer underestimated values (Freestyle < BG). I would say this happens always like this.
        7.6. When your BG changes very slowly (LCHF diet), the sensor may throw strange values: maybe 10 mg variation in 1 or 2 minutes, even if your BG is stable. When my diet was high carb, as changes on BG where stronger, I think it was less misleading.
        7.7. The Freestyle tends to overstate low BG, and, on the other hand, underestimate high BG. That happens always (I don’t remember any sensor which worked the other way). I think Abbott does it to avoid hypos.
        7.8. Freestyle has maybe a 10-20 minute (not more) delay compared to BG changes.
        7.9. The tendency of glycaemic variation shown by Freestyle is good. No mistakes even if the sensor is wrong.
        7.10. GREAT problem (exception to 7.9.): my last tree sensors have always shown very wrong numbers and also tendency when I practise sport. Abbott has accepted warranty for two of them, but I’m still wearing one of them because I have not received the substitute yet. When I say wrong numbers, I mean wrong numbers: 190 Free VS 120 BG, 160 Free VS 107 BG, 140 Fee VS 97 BG… And I also shows a going up tendency while my BG is falling down very slowly. After maybe 2 hours the tendency gets corrected, but you have a permanent difference of 50 or 60 mg. That is a problem because you have to check constantly your BG… so the Freesyle is useless when this happens.
        I have to underline two things:
        a) With my last 3 sensors this has happened every workout, morning or evening, with or without Metformin, with or without insuline. Maybe it is not wrong, and interstitial glucose increases while BG goes down. I have no idea. When I phoned Abbott the girl I spoke to told me that she had also experienced that strange variation.
        b) What happened with my sensor before my LCHF diet? Did it show growths in interstitial glucose in spite of BG going down tendency? Well, I suppose I will never know. In morning workouts it always showed a growth in interstitial glucose, but not as much as it shows now. Not as much, definetly. But then I didn’t compare the values to the BG. That’s the problem. So how high was the error rate? No idea. At that time I always thought it was a sort of rebound after an intense fall caused by insuline dose (2u by that time) and Metformin. I started comparing them now, when I have change my diet, because I was very astonished (and worried thinking of DKA) when I saw levels rise so fast and high. Now my insuline dose is 1u (Novolog): there is no big drop at all and just when I start riding it goes up for 1 hour or 1h 15 minutes. Always. It can rise 60 mg/dl; sometimes even more.
        In evening workouts I just don’t have information. The problem is that with my high carb diet I ate 30 to 40g of carbohydrate just before the workout. So, then, the interstitial glucose increased. Why? I assumed glucose was the reason. That made sense to me.

        8) Usefulness:
        8.1. It has (yet) no alarm function in case of hypo or hyperglycemia. In the US, Ambrosiasys will introduce this function with another device you put on the Freestyle sensor (that’s what they say, at least). In Spain the university of my village has developed a product to put on the sensor (it is like a bracelet) that allows you to program alarms on your phone. It uses bluetooth. I have it and works fine, but, of course, it depends on the sensor accuracy. Tonight I heard the alarm: I was 57 on Freestyle, but my BG was 95 mg/dl! (I’m waiting for Abbott to replace this sensor, so we shouldn’t take this into account.)
        8.2. I’m not very happy. That is my overall impression. I will still use it for 3 reasons:
        a) Dexcom is too expensive for me.
        b) It is fine for my daylife because it is not usual to find wrong sensors (and, as I said before, if the sensor is wrong you will notice it easily from the beginning). The trend is a very good information even if is not as accurate as I would like.
        c) I hope the problem while practising sport is exceptional or gets soon solved. If not, I will definetly stop using it and make an effort with Dexcom. So, I need more time to make my decision.

        Let me know if you need more information.


      • Keith Runyan, MD

        Eric thanks for that review of Abbott’s Freestyle Libre continuous glucose monitoring system. The few others have given me feedback, some very positive, some largely negative. I guess I will eventually try it after it gets approval and goes on sale in the U.S., but will take its results with a grain of salt at first until I can assess its accuracy. I have heard similar negative statements about Dexcom’s accuracy as well, just FYI. Maybe biting the bullet and taking more BG measurements is the answer until interstitial glucose monitoring becomes more accurate.


  6. ÉRIC

    Of course, I believe everyone that can afford it should try Freestyle. I live with less uncertainty. The only problem is accuracy of some sensors and accuracy while exercising (very important for me).

    Uf, thank you a lot for the information about Dexcom’s accuracy. I thought it was much (MUCH) better. Let’s hope Abbott improves, then. I have clear in the future I will still use these devices (one or another), even though the problems I have reported.

    Until that happens, no doubt: more BG measurements when practising sport.


    • Keith Runyan, MD

      Thanks, Eric. My experience has been similar to information in the link you sent that the Abbott Freestyle Lite meter and strips are very accurate. I buy my strips in the U.K. to reduce the cost compared to that in the U.S.


      • Éric

        Sorry, I sent half of my original message. I wanted to compare Free’s accuracy and Lite’s accuracy.

        Yesterday I was a bit concerned again about accuracy: at night Freestyle Libre always showed around 50 mg, but in fact my BG was more than 80 mg. The sensor was still in the half of its life and I had to throw it.

        In the morning I phoned Abbott and they answered it was still in an acceptable error rate, despite those differences. It was in zone B of Clarke’s error grid. And then they explained that If the sensor is in first two zones (A and B), then Abbott doesn’t replace them unless you want to. But in case you want to, they will replace maximum 3 sensors (per year or in the whole life? I forgot to ask this). On the other hand, Abbott replaces the sensor automatically if its measurements are in C, D or E zones.

        Here they show the grid:


        And the study:


        The thing is that they remark Dexcom G5 has similar (but worse) MARD than Freestyle!

        Two more references:


        That was all. Sorry for the first message.


      • Keith Runyan, MD

        Thanks Eric. I looked at all the references. These studies need to be done to justify their value in the marketplace. However, I think each person needs to decide for themselves whether the device adds value to their own glycemic management. The inaccuracies of the CGMs at least result in more frequent capillary blood glucose measurements. I would be wary to take insulin based on the CGM results unless my experience was much better than the average results in the published studies. I think dealing with a company to replace inaccurate sensors on a regular basis would frustrate me. Thanks for sharing your experience. Keith.