Insulin For Diabetes on a Low Carb or Ketogenic Diet 

This article covers the insulin preparations used for type 1 and type 2 diabetes with particular reference to their use with a low carb or ketogenic diet.

But first, I would like to give you some information about insulin, its discovery in 1921, and the evolution of insulin manufacture that has been essential as a life-saving therapy for those with insulin-requiring diabetes. In animals and humans, insulin is made in the beta-cells of the pancreas starting with a preproinsulin, one long chain of amino acids. When the signal (high blood glucose or post-meal amino acids) to secrete insulin is received, a short signaling section is cleaved off forming proinsulin. Then a short middle section is cleaved off forming insulin which is composed of an A chain and a B chain of amino acids chemically bound together. The A chain has 21 amino acids and the B chain has 30 amino acids. More info here. The initial insulin preparations were made by purifying the pancreas of cows called bovine insulin or from pigs called porcine insulin. These preparations were quite effective to treat T1DM since they only differ by a few amino acids from human insulin. However, they did result in autoimmune or allergic reactions in a small percentage of patients and thus were no longer effective for those patients. Fortunately technology has improved and in 1986, the first human insulin was manufactured using recombinant DNA technology from bacteria,  Escherichia coli or yeast, Saccharomyces cerevisiae. This avoided the allergic reactions to animal insulin preparations as well as other advantages. More info here.

Regular and Rapid-Acting Insulin Preparations

One problem with regular human insulin as a meal-time insulin is that its onset of action (0.5-1 hr), peak action (2.5-5 hrs), and duration of action (8-12 hrs) is longer than the rate of absorption of nutrients from the gastrointestinal tract such that blood glucose (BG) rises significantly after a meal and may not normalize before the next meal. This is especially true when one is eating a high-carbohydrate meal. Regular human insulin may work quite well for those who follow a low carb or ketogenic diet. This is the insulin Dr. Richard Bernstein recommends for his patients who follow a 30-gram/day low carb diet. The regular human insulins on the market currently are Humulin Regular and Novulin Regular. However, since the vast majority of insulin-requiring diabetes patients are not aware of the low carb diabetes diet, the pharmaceutical industry devised rapid-acting insulin analogs to address this imbalance between the rate of insulin action and rate of nutrient absorption. Insulin analogs are insulin molecules with a few changes in the amino acid sequence that affects its rate of absorption from the site of subcutaneous injection into the blood stream. It also affects the potency of the insulin somewhat, with insulin analogs being more potent than regular human insulin. More potent means it takes a smaller dose to get the same BG lowering effect. These insulin analogs are also made with recombinant DNA technology from bacteria or yeast. The three rapid-acting insulin analogs on the market currently are lispro (Humalog), aspart (Novolog), and glulisine (Apidra). 

Chemically, insulin lispro is Lys(B28), Pro(B29) human insulin analog, created when the amino acids at positions 28 and 29 on the insulin B-chain are reversed.

These insulin analogs have a more rapid onset of action and a duration of action of at most 5 hrs. For example the insulin analog lispro (Humalog) which I use, begins working in 15-30 mins, peaks at 0.5-1.5 hrs, and has a duration of action of about 2-5 hrs. Each of the pharmacokinetic profiles of the insulin analogs are slightly different and thus may be more effective for some persons compared to other analogs. See Table 1 below. There are individuals, like myself, for whom insulin analogs work quite well on a low carb ketogenic diet. Each insulin will work somewhat differently in each person. Thus, it is best to try several and see which works best for you. For example, I use Humalog which I take immediately before I start eating if my BG is normal or high (> 70 mg/dl). But if my BG is low (< 70 mg/dl), I wait until I have finished eating the meal to take the Humalog. I then wait 2.5 to 3 hrs after injecting to measure my BG again. I have determined that after this time, my BG remains stable (excluding exercise or other similar factors). In contrast I found that when the dose of regular insulin was underestimated, the BG was elevated for much longer compared to rapid-acting insulin due to the longer duration of action of regular insulin. With regular insulin you risk hypoglycemia if you use a correction dose a few hours after  a meal due to stacking of the additional dose onto the previous dose. As with any meal-time insulin, you must determine its duration of action by measuring BG repeatedly after the meal to see when the BG levels off. In the case of regular insulin, if the BG continues to slowly decrease after the meal up to the next meal, you can't use a correction dose, but instead have to adjust the next meal-time dose to factor in correcting the BG. A note about eating away from home: I have had several hypoglycemic episodes at restaurants naively thinking the food was coming soon. Since then, I never injection insulin until all the food I plan to eat is in front of me. Not eating dessert helps with this too (typically ordered after a meal).

Takiya L, Dougherty T. Pharmacist’s guide to insulin prepara- tions: a comprehensive review. Last accessed August 2012.

Basal Insulin Preparations

In order to extend the duration of action of animal insulins initially, and later human insulins, zinc ions and/or highly basic proteins such as protamines were added to insulin to produce a basal insulin for persons with insulin-requiring diabetes. The first of these insulins was NPH (neutral protamine Hagedorn) insulin which is now classified as an intermediate-acting insulin. It has an onset of 1-1.5 hrs, a peak at 6-14 hrs, and a duration of action of 16-24 hrs. Problems with NPH insulin when attempting to it use it as a basal insulin include variable absorption with notable inter- and intra-individual variation, duration of action of less than 24 hrs in many individuals, and most importantly a distinct peak of action. Thus, many of those taking NPH insulin had to take it twice a day and the peak of action causes low blood sugars (hypoglycemia). Taking NPH with breakfast and dinner is a common regimen. Because of the peak of NPH at 6-14 hrs, taking NPH at dinner (7 PM) can result in hypoglycemia anytime between 1 AM and 9 AM while sleeping. Hypoglycemia occurring at night (nocturnal hypoglycemia) is the most dangerous time to have hypoglycemia due to the fact that one is less aware of the symptoms. This has resulted in death. Unfortunately, 4-10% of persons with T1DM die from hypoglycemia mostly occurring while sleeping. Despite these disadvantages, NPH is still available on the market. These reasons I believe are that it is much less expensive than insulin analogs and the convenience of being able to mix it with short- or rapid-acting insulins which reduces the number of injections needed. I understand the cost issue, but the convenience of fewer injections does not balance the danger of hypoglycemic episodes that result from NPH in my opinion. Another disadvantage of NPH and short- or rapid-acting insulin mixtures is that the relative dose ratio is fixed which eliminates the flexibility of being able to estimate the best dose of each insulin. If your finances can afford a basal insulin analog (see below), that would be a better choice over NPH or NPH insulin mixtures. If your finances require you to use NPH, purchase the NPH and short- or rapid-acting insulins separately, decide on each dose separately, and then both can be drawn up in the same syringe at injected together.

After NPH was on the market, the basal insulins, Lente and Ultralente, were developed. I used both of these, finding Ultralente worked better due to less of a peak and longer duration of action. However, after the basal insulin analog, Lantus, was developed in 2000, it was so much better than Lente and Ultralente that the demand for them dropped and they were both removed from the market.

The next step in basal insulin production was using the recombinant DNA technology to make basal insulin analogs. This has really made a valuable contribution to the safety of insulin therapy because the basal insulin analogs do not have a peak of action which dramatically reduces the risk of hypoglycemia especially at night. The basal insulin analogs currently on the market include glargine (Lantus), detemir (Levemir), degludec U200 (Tresiba), and glargine U300 (Toujeo).

Glargine (Lantus)

Glargine (Lantus) was the first basal insulin analog to be developed in 2000. Recombinant DNA technology was used to produce the insulin molecules with substitution of asparagine at position 21 on the A chain (A21) with glycine (GlyA21) and addition of two arginine amino acids at positions 31 and 32 on the B chain (ArgB31ArgB32). See Figure below. In addition, phenol and zinc are added. All of these alterations result in an insulin that is slowly and evenly released from the site of subcutaneous injection to prolong its duration of action to about 24 hrs. This helped to reduce the risk of hypoglycemia compared to previous insulin preparations (NPH, Lente, Ultralente).

Concern was raised by an initial epidemiological study that glargine (Lantus) increased the risk of cancer primarily in patients with type 2 diabetes (T2DM). These patients are already at increased risk of cancer due to insulin resistance and hyperinsulinemia. Adding any exogenous insulin to these patients may increase that risk. However, epidemiological studies cannot determine any cause and effect relationship. Subsequently, the ORIGIN (Outcome Reduction with Initial Glargine INtervention) study of 12,537 people with early T2DM or pre-diabetes, the longest randomized controlled trial of insulin therapy extending over a median period of 6.2 years, found no increase seen in the incidence of all cancers combined, any organ-specific cancer (including breast, lung, colon, prostate, and melanoma), or cancer in the glargine group compared with the standard care group. I have taken Lantus since 2005 with good success, thus I may have some bias in favor of Lantus. I also tried detemir (Levimer) which was not as effective as Lantus because I needed a higher dose and had to take it twice daily and it was much more expensive.  I have not tried degludec U200 (Tresiba) which has a duration of action of > 25 hrs. Degludec U200 (Tresiba) is also much more expensive than glargine (Lantus). Finally, glargine is the first insulin preparation to be available in a generic form presumably at an even lower cost than Lantus. That said, each person requiring insulin for control of diabetes should try as many available meal-time and basal insulin options as needed to obtain the best BG profile i.e. BG as close to normal (96 mg/dl) as is safely (minimal hypoglycemia) possible or as directed by your physician. 

Humalog Mix 75/25 or 50/50

Humalog Mix is another mixed insulin that comes in two ratios: 75/25 and 50/50. That means 75% is protamine lispro and 25% is lispro. Adding protamine slows the absorption of the lispro insulin from the subcutaneous injection site making it an intermediate insulin. Mixing the two together allows for convenience of fewer injections, but loss of control over the amount of each insulin given separately, i.e. changing the dose changes the amount of both lispro and protamine lispro. This insulin is only slightly more expensive than insulin lispro, and might be an option for those with limited financial resources. An advantage over NPH is that Humalog Mix does not have a delayed peak of action. I have no experience using this insulin in persons following a low carb diet or ketogenic diet. It is entirely possible that given the much smaller doses of rapid-acting insulin needed on a low carb diet, that the protamine lispro activity would not be sufficient to keep BG controlled overnight.

Humalog 75/25 Pharmacokinetics. Note: No delayed peak of action.
Comparison of Pharmacodynamics of Humalog Mix Preparations with other insulins. From

Insulin Use With A Low Carb or Ketogenic Diet

A low carb or ketogenic diet addresses both the carbohydrate intolerance of T1DM and T2DM and the insulin resistance of T2DM and double diabetes (insulin resistance in those with T1DM). However when starting a low carb or ketogenic diet for insulin-requiring diabetes, it is very important that you discuss this with your physician because it will require a significant reduction in meal-time insulin dose and often a smaller reduction in basal insulin dose. For example, in April 2007 before I started exercising regularly and while on a relatively high-carb balanced diet, I took on average of 24.9 IU/day of Humalog and 32.3 IU/day of Lantus at a body weight of 153 lb. or total daily insulin dose of 57.3 IU/day or 0.82 IU/kg/day to achieve an average BG of 123 mg/dl. The percentage of meal-time insulin was 43% of the total daily dose. Contrast this with May of 2018 while exercising daily (improves insulin sensitivity and thus less insulin) and eating a whole-food ketogenic diet (requires less insulin), I took on average of 6.9 IU/day of Humalog and 21.2 IU/day of Lantus at a body weight of 165 lb. or a total daily insulin dose of 28.1 IU/day or 0.37 IU/kg/day to achieve an average BG of 100 mg/dl. The percentage of meal-time insulin was 24% of the total daily dose. Thus, the percentage of meal-time insulin was almost half and the total daily insulin dose was less than half after changing my lifestyle to regular exercise on a ketogenic whole-food diet. Another way to view the change is that 62% of the reduction in total daily insulin dose was due to the reduction in meal-time insulin. The basal insulin reduction accounted for 38% of the reduction in total daily insulin dose. Keep in mind that these doses were 11 years apart, with and without exercise, and at a different body weight in addition to the change in diet. The point is that this significant reduction in total daily insulin dose with the reduction in meal-time insulin being the majority of the reduction, is typical of using a low carb ketogenic diet for diabetes. This reduction in insulin dosage has the definite advantage of reducing the risk of hypoglycemia. In addition to the reduction insulin dose as a result of the low carb ketogenic diet and exercise, the average BG improved to normal levels and the BG variability as measured by standard deviation of the BG improved from 54 mg/dl to 38 mg/dl. Average daily BG in persons without diabetes is 96 ± 12 mg/dl (mean ± standard deviation (SD)) and coefficient of variation is 13% which is the weighted mean from two studies of continuous glucose monitoring in healthy subjects (see references here and here). This is the average BG that I am seeking because I can be fairly confident that I will not develop long-term complications of diabetes by doing so. However, your BG target should be determined in discussion with your physician because the caveat is that this be done safely with minimal hypoglycemia.

Using Insulin For Type 1 Diabetes

These are general guidelines for using insulin for T1DM. These guidelines are not intended as medical advice and should be discussed with your physician prior to using. Your physician should discuss with you what your target BG range and average should be. That value should be as close to normal as is safely possible. Here "safely" means with minimal hypoglycemia and "as close to normal" is the target BG that you and your physician decide upon. Based on this decision, your physician formulates your insulin regimen. There are different ways to achieve the desired results. My guidelines below may or may not apply to you. They are generally applicable to those seeking a BG close to 96 mg/dl with minimal hypoglycemia in persons with T1DM who have reached an agreement with their physician to adjust each dose based on the BG reading, the previous and intended exercise, the meal to be eaten, as well as the BG response to previous insulin doses over the past 1-3 days.

Regarding meal-time insulin dose,

  • Always measure BG before taking insulin.
  • Keeping the quantity and type of low carb foods consistent at each meal from day to day helps to keep the response to insulin consistent as well. If you can eat the same meal every day for breakfast, lunch, and dinner, that is best. This is very important if you desire predictable BG results. That does not mean all three meals need to be the same. It means all the breakfast meals are the same, all the lunch meals are the same, and all the dinners are the same. I do not follow this exactly. All of my lunches and dinners are the same. However, I do rotate the breakfast protein source between beef, liverwurst, and salmon, but the quantity is the same. I eat the exact same amount and types of nuts and seeds at each breakfast, lunch, and dinner. The quantity of non-starchy cooked vegetables is fairly accurately estimated by the amount of space covered on the plate. You can make changes to any of your meals, but that may require an adjustment to your insulin dose so it is better to stick with that change for some time to achieve more predictable results. 
  • Each meal-time insulin dose is carefully decided upon based on the BG reading, the previous and anticipated exercise, the actual response to previous insulin doses over the past 1-3 days. By keeping your exercise routine consistent from day to day, the variation in insulin sensitivity can be minimized. This is also a very important part of a strategy to achieve predictable BG results. Exercising one day, but not another, or doing twice as much exercise one day versus another, or exercising in the morning one day, but in the evening the next, can certainly explain unpredictable BG results. Similarly, eating entirely different meals day-to-day will result in more variable BG results even if the dietary carbohydrate is restricted. The insulin dose is also adjusted using the pre-meal BG reading compared to previous days at the same meal. For example, If the past three morning the pre-breakfast BG, post-breakfast BG, and Humalog dose respectively were Day 1) 70 mg/dl, 86 mg/dl, 2.5 IU, Day 2) 90 mg/dl, 69 mg/dl, 3 IU, Day 3) 120 mg/dl, 82 mg/dl, 4 IU. Now today, pre-breakfast BG is 100 mg/dl and the Humalog dose needs to be determined. If we choose 4 IU, we might expect the post-breakfast BG to be 62 mg/dl based on Day 3's result. If we choose 3 IU, we might expect the post-breakfast BG to be 79 mg/dl based on Day 2's result. If we choose 2.5 IU, we might expect the post-breakfast BG to be 116 mg/dl based on Day 1's result. All of these are calculated based on the difference between the pre- and post-meal BG. In other words, each meal-time insulin dose contributes to processing the food eaten and to adjusting the post-meal BG reading to your goal target range. These calculations assume that because the meals are constant, that differences in the insulin dose given is based on the change in BG. Based on these considerations, I would choose a Humalog dose of 3 IU in the hopes of getting a post-breakfast BG of 79 mg/dl. Often times, I am not quite sure of the best dose because of variation in my response to insulin in the previous 3 days. In these cases, I prefer to err on the side of giving less insulin rather than risk hypoglycemia. I know I can always give another correction dose of Humalog later. For example, use 0.25-0.5 IU less than your unsure estimate. But the main point is, when in doubt about the correct dose, use a little less to avoid hypoglycemia.
  • Always measure BG after a meal at the time you have determined represents the duration of action of your particular insulin. You can determine this by measuring your BG repeatedly after a meal until it no longer changes significantly. When it stops changing, that is your duration of action. If your post-meal BG is < 70 mg/dl or another threshold set by your physician, treat with glucose tablet(s) (4 grams each). If your post-meal BG is above your target, you can take a correction dose of insulin to bring it down to range. Again, you have to consider any planned exercise as that may affect how much insulin to take, if any. For example, if your BG typically falls during the planned exercise, you would might not want to take insulin if your BG is just slightly above target. If it is > 300 mg/dl, you should correct it with insulin and recheck the BG before exercising. Exercising with a very high BG can precipitate diabetic ketoacidosis (DKA).

Regarding basal insulin dose,

  • Always measure your fasting BG because this is the value used to decide the proper basal insulin dose. You are aiming for the target BG value agreed upon by you and your physician. If your fasting BG is below 70 mg/dl, you should reduce the next basal insulin dose by 10-20% (discuss with your physician). If your fasting BG is above your target, and has been on the previous 2 or 3 mornings, check with your physician about increasing the dose by 5-10%. Once a dose change is made for a high fasting BG, you should stick with that change for 3-4 days for Lantus or Levemir and once a week for Tresiba before changing it again. This is due to the long half-life of basal insulins. If you increase the dose every day, after several days, you will become hypoglycemic. Specifically for T1DM, the basal insulin analogs have been shown to reduce the frequency of hypoglycemia. I think this justifies their increased cost.
  • Basal Insulin can be given twice a day. This applies especially to detemir (Levemir), but also to Lantus. It unlikely that degludec U200 (Tresiba) would need to be given twice daily. The reason to give it twice daily, would be if it is being given in the morning and the fasting BG is high and is not coming down with an increase in dose. This could be because the dose is not lasting 24 hrs. The duration of action is largely, but not entirely, unrelated to the dose given. In this example, taking the basal dose at dinner or bedtime might correct the problem which is why it is usually recommended to be given at these times. When given at dinner-time, if the basal insulin is running out just a few hours before the next dose, the meal-time insulin dose from lunch may be able to compensate for the waining basal insulin activity. You should ask your physician about when to take your basal insulin and whether or not twice daily basal is appropriate.

Using Insulin For Type 2 Diabetes

A low carb or ketogenic diet addresses both the carbohydrate intolerance and the insulin resistance of T2DM. For those with T2DM taking insulin who are contemplating starting a low carb ketogenic diet, be advised that when starting a low carb or ketogenic diet, it is very important that you discuss this with your physician because it will require a significant reduction in meal-time insulin dose and often a smaller reduction in basal insulin dose. However, the amount of reduction in insulin dose that will be necessary depends on how much insulin your pancreas is able to make. Most persons with T2DM have the ability to make insulin because their pancreatic beta-cells are working, but they are simply unable to produce enough insulin to both process a high-carb diet and overcome the insulin resistance that has developed over the years from excess sugar, carbohydrate, total caloric intake and body fat (if present). In other words, removing the high-carb intake will often quickly allow the beta-cells to produce enough insulin to appropriately normalize BG. So insulin doses usually need to be reduced dramatically (50 - 100%) when starting a low carb ketogenic diet. This is why you need to work with your physician, otherwise, you could end up with severe hypoglycemia. Sulfonylurea and meglitinide medications work by stimulating beta-cell insulin secretion without regard to the prevailing BG level and can also result in hypoglycemia when starting a low carb ketogenic diet if they are not appropriately reduced or discontinued. In many cases, insulin, sulfonylurea, and meglitinide medications can be discontinued either at the start of or over time on a low carb ketogenic diet. This is really good news because this will help to reverse the insulin resistance that lead to T2DM in the first place. It also makes losing excess body fat easier.

A smaller subset of persons with T2DM will not be able to completely stop insulin due loss of beta-cells, either in number or in their ability to make insulin. This is no reason to despair because the main purpose of a low carb ketogenic diet, exercise, and lifestyle changes is to reverse insulin resistance. It is insulin resistance and hyperinsulinemia that leads to the other chronic diseases that are associated with T2DM. Appropriate amounts of insulin to control BG, even if it is injected, is not a problem.

For persons with T2DM who are not taking insulin but have been recommended to start it, I would suggest that you discuss with your physician about getting started on a low carb ketogenic diet ASAP. A low carb ketogenic diet may greatly improve or normalize your BG so that insulin is not needed. If you do decide to start a low carb ketogenic diet, it is important to discuss with your physician about reducing or stopping any medications in these two classes: sulfonylureas and meglitinides. Please read above to understand why.

In my view, taking insulin for T2DM should be accompanied with maximal use of lifestyle changes including a low carb ketogenic diet and exercise. The following guidelines for using insulin for T2DM assume that you are maximizing the lifestyle benefits discussed on this website.

Persons with T2DM starting on insulin usually start with basal insulin to normalize the fasting BG. Sometimes this, combined with lifestyle changes and other diabetes medications, is all that is needed. This is the reason for starting with basal insulin first before adding meal-time insulin.

Regarding basal insulin for T2DM,

  • Insulin is usually reserved in those with T2DM who have failed to achieve the target BG you have set with your physician by lifestyle changes, metformin and two additional diabetic medications. Usually these persons will have a morning fasting BG > 130 mg/dl (7.2 mmol/l).
  • The purpose of using basal insulin for T2DM is to normalize the morning fasting BG. One of the basal insulins discussed above is started at 10 IU/day or 0.1-0.2 IU/kg/day, and increased by about 2-4 IU or 10-15% every 3-4 days or once per week depending on which basal insulin is being used. If you have a hypoglycemic episode decrease the basal insulin dose by 4 IU or 10-20% and work with your physician to determine the cause and to help you adjust your dose.
  • Once the fasting morning BG is normalized, if the 2-hour post-meal BG values are consistently > 140 mg/dl (7.8 mmol/l) and the average BG or HbA1c are not at target, then consider adding a meal-time rapid-acting insulin to one or more meals.

Regarding meal-time rapid-acting insulin for T2DM,

  • The initial dose of rapid-acting meal-time insulin is usually 4 IU, 0.1 IU/kg, or 10% of the basal insulin dose. Increase by 1-2 IU or 10-15% if the post-meal BG is > 140 mg/dl once or twice weekly, until the post-meal BG is normalized. 
  • Please see the guidelines above for meal-time rapid-acting insulin for T1DM for additional tips on adjusting insulin.

My Books Are Available in Downloadable PDF or in Print on Amazon.

 Click to buy Type 1 Diabetes Book.                              Click to buy Type 2 Diabetes Book.

Both books are available as a PDF ebook and as a print book from Amazon. Note: Clicking on the links above will take you to the website of Ellen Davis, MS, my coauthor. Her website is

Personal Consultation

If you would like to consult with me about managing your diabetes or other metabolic conditions discussed on this website, you may send me an email to discuss your needs, receive a patient history form to complete, and schedule a Skype meeting. After we have scheduled a meeting, you can use the payment button below. There is an email form on the Contact page.


Donate Here

If you like the information provided and have benefited from it, please support my efforts by purchasing a book or making a voluntary donation below.

Receive My Newsletter Below

Comment Box is loading comments...