Has Carbohydrate-Restricted Eating Been Studied in Type 1 Diabetes?

Sweden has lots of blondes

Sweden has lots of blondes

Yes, there are few published scientific reports. Let’s take a close look at one today. (See the references below for more.)

In the introduction to the study at hand, the authors note:

The estimation of the amount of carbohydrates in a meal has an error rate of 50%. The insulin absorption may vary by up to 30%. It is therefor virtually impossible to match carbohydrates and insulin which leads to unpredictable blood glucose levels after meals. By reducing the carbohydrates and insulin doses the size of the blood glucose fluctuations can be minimized. The risk of hypoglycemia is therefore minimized as well. Around-the-clock euglycemia [normal blood sugar] was seen with 40 g carbohydrates in a group of people with type 1 diabetes [reference #2 below].

The immediate resulting stable, near-normal blood glucose levels allow individuals to predict after-meal glucose levels with great accuracy.

For individuals with type 1 diabetes one year audit/evaluation of group education in this regimen has shown that the short-time lowering of mean hemoglobin A1c by 1 percentage unit and the reduction in mean rate of symptomatic hypoglycemia by 82% was maintained [reference #3].


There is no evidence for the use of the widely recommended high-carbohydrate, low-fat diet in type 1 diabetes.

Study Set-Up

Swedish investigators educated study participants on carbohydrate-restricted eating from 2004 to 2006 [reference #1]. They recently audited their medical records for results accumulated over four years. At the outset, participants were given 24 hours of instruction over four weeks. My sense is that they all attended the same diabetes clinic. The subjects’ mean age was 52 years and they had diabetes for an average of 24 years. Seven had gastroparesis. Fourteen used insulin pumps. Of the 48 study subjects, 31 were women, 17 were men. The diet regimen restricted carbohydrates to a maximum of 75 grams a day, mainly by reducing starchy food.


As measured three months after starting the diet, HDL-cholesterol rose and triglycerides fell to a clinically significant degree (p<0.05). Average weight fell by 2.7 kg (5.9 lb); average baseline weight was 77.6 kg (171 lb). Hemoglobin A1c fell from 7.6 to 6.3% (Mono-S method).

As measured one year after start, meal-time insulin (rapid-acting, I assume) fell from 23 to 13 units per day. Long-acting insulin was little changed at around 19 units daily.

By two years into the study, half the participants had stopped adhering to the diet. The remainder were adherent (13 folks) or partly adherent (10). We don’t know what the non-adherents were eating.

Four years out, the adherent group had hemoglobin A1c of 6.0%, and the partly adherents were at 6.9% (p<0.001 for both). The non-adherent group had returned to their baseline HgbA1c (7.5%). Remember, at baseline the average HgbA1c for the group was 7.6%.

The authors don’t say how many participants were still adherent after four years. From Figure 2, adherence seems to have been assessed at 60 months: 8 of the 13 adherent folks were still adherent, and 5 of the 10 partly adherent were still in the game. So, of 48 initial subjects, only 13 were still low-carbing after five years later. By five years out, half of all subjects seem to have been lost to follow-up. So the drop-out rate for low-carbers isn’t as bad as it looks at first blush.


The authors write:

An educational program involving a low-carbohydrate diet and correspondingly reduced insulin doses for informed individuals with type 1 diabetes gives acceptable adherence after 4 years. One in two people attending the education achieves a long-term significant HbA1c reduction.

They estimate that this low-carb diet “may be an option for 10-20% of the patients with type 1 diabetes.” Only 17% of their current diabetes clinic population is interested in this low-carb diet. They didn’t discuss why patients abandon the diet or aren’t interested in the first place. Use your imagination.

Major carbohydrate restriction in type 1 diabetics significantly improves blood sugar control (decreases HgbA1c), lowers insulin requirements, and improves cardiovascular disease risk factors (increases HDL cholesterol and lowers triglycerides).

Paleo diets vary in total carbohydrate grams and percentage of calories derived from carbohydrate. Paleo diets tend to be lower in carb than usual Western diets, with 30% of total calories from carbohydrate probably a good rough estimate. The typical American eats 250 to 300 grams of carbohydrate daily, or about 50% of total calories. In the study at hand, the daily carb gram goal was 75, which would be 15% of calories for someone eating 2,000 cals/day.

Low-carb eating wasn’t very appealing to Swedes in the mid-2000s. I wonder if it’s more popular now with the popularity of LCHF dieting (low-carb, high-fat) in the general population there.

Steve Parker, M.D.


1.  Nielson, J.V., Gando, C., Joensson, E., and Paulsson, C. Low carbohydrate diet in type 1 diabetes, long-term improvement and adherence: A clinical audit. Diabetology & Metabolic Syndrome, 2012, 4:23. http://www.dmsjournal.com/content/4/1/23

2.  O’Neill, D.F., Westman, E.C., and Bernstein, R.K. The effects of a low-carbohydrate regimen on glycemic control and serum lipids in diabetes mellitus. Metabolic Syndrome and Related Disorders, 2003, 1(4): 291-298.

3.  Nielsen, J.V., Jönsson, E. and Ivarsson, I. A low carbohydrate diet in type 1 diabetes: clinical experience – A brief report. Upsala Journal of Medical Sciences, 2005, 110(3): 267-273.

3 responses to “Has Carbohydrate-Restricted Eating Been Studied in Type 1 Diabetes?

  1. Thank you for your valuable blog and thanks for reviewing this study!

    One thing though: 2.7 kgs = 5.9 pounds.

  2. Pingback: Paleosphere News Roundup – 04 November 2013 | Must Love Paleo