Compared to a mildly carbohydrate-restricted American Diabetes Association diet, a very-low-carbohydrate ketogenic diet was more effective at controlling type 2 diabetes and prediabetes, according to University of California San Francisco researchers.
Thirty-four overweight and obese type 2 diabetics (30) and pre diabetics (4) were randomly assigned to one of the two diets:
- MCCR: American Diabetes Association-compliant medium-carbohydrate, low-fat, calorie-resticted carb-counting diet. The goals were about 165 grams of net carbs daily, counting carbohydrates, an effort to lose weight by eating 500 calories/day less than needed for maintenance, and 45–50% of total calories from carbohydrate. Protein gram intake was to remain same as baseline. (Note that most Americans eat 250–300 grams of carb daily.)
- LCK: A very-low-carbohydrate, high-fat, non-calorie-restricted diet aiming for nutritional ketosis. It was Atkins-style, under 50 grams of net carbs daily (suggested range of 20–50 g). Carbs were mostly from non-starchy low-glycemic-index vegetables. Protein gram intake was to remain same as baseline.
Baseline participant characteristics:
- average weight 100 kg (220 lb)
- 25 of 34 were women
- average age 60
- none were on insulin; a quarter were on no diabetes drugs at all
- most were obese and had high blood pressure
- average hemoglobin A1c was about 6.8%
- seven out of 10 were white
Participants followed their diets for three months and attended 13 two-hour weekly classes. Very few dropped out of the study.
Average hemoglobin dropped 0.6% in the LCK group compared to no change in the MCCR cohort.
A hemoglobin A1c drop of 0.5% or greater is considered clinically significant. Nine in the LCK group achieved this, compared to four in the MCCR.
The LCK group lost an average of 5.5 kg (12 lb) compared to 2.6 kg (6 lb) in the MCCR. The difference was not statistically significant, but close (p = 0.09)
44% in the LCK group were able to stop one or more diabetes drugs, compared to only 11 % in the other group
31% in the LCK cohort were able to drop their sulfonylurea, compared to only 5% in the MCCR group.
By food recall surveys, both groups reported lower total daily caloric intake compared to baseline. The low-carbers ended up with 58% of total calories being from fat, a number achieved by reducing carbohydrates and total calories and keeping protein the same. They didn’t seem to increase their total fat gram intake;
The low-carbers apparently reduced daily carbs to an average of 58 grams (the goal was 20-50 grams).
There were no differences between both groups in terms of C-reactive protein (CRP), lipids, insulin levels, or insulin resistance (HOMA2-IR). Both groups reduced their CRP, a measure of inflammation.
LCK dieters apparently didn’t suffer at all from the “induction flu” seen with many ketogenic diets. They reported less heartburn, less aches and pains, but more constipation.
Hypoglycemia was not a problem.
If I recall correctly, the MCCR group’s baseline carb grams were around 225 g.
Very-low-carb diets help control type 2 diabetes, help with weight loss, and reduce the need for diabetes drugs. An absolute drop of 0.6% in hemoglobin A1c doesn’t sound like much, translating to blood sugars lower by only 15–20 mg/dl (0.8–1 mmol/l). But remember the comparator diet in this study was already mildy to moderately carbohydrate-restricted. At least half of the type 2 diabetics I meet still tell my they don’t watch their carb intake, which usually means they’re eating around 250–300 grams a day. If they cut down to 58 grams, they most likely will see more than a 0.6% drop in hemoglobin A1c after switching to a very-low-carb diet.
This is a small study, so it may not be reproducible in larger clinical trials and other patient populations. Results are consistent with several other similar studies I’ve seen, however.
Reference: Saslow, Laura, et al (including Stephen Phinney). A Randomized Pilot Trial of a Moderate Carbohydrate Diet Compared to a Very Low Carbohydrate Diet in Overweight or Obese Individuals with Type 2 Diabetes Mellitus or Prediabetes. PLoS One. 2014; 9(4): e91027. Published online Apr 9, 2014. doi: 10.1371/journal.pone.0091027 PMCID: PMC3981696
PS: When I use “average” above, “mean” is often a more accurate word, but I don’t want to have to explain the differences at this time.
PPS: Carbsane Evelyn analyzed this study in greater detail that I did and came to different conclusions. Worth a read if you have an extra 15 minutes.