There is a study, headed by Barnard (of PCRM infamy), purported to show the benefits of a low fat vegan diet for the management of type two diabetes in humans. This is, obviously, counter intuitive. So let's have a look.
The first thing to say is that it's extremely difficult to extract any hard data from the study. Many of the results are expressed as derived from "intention to treat" analysis. To appreciate quite how difficult this makes any sort of deconstruction, you have to read Dr Eades on this subject. Much of the HbA1c data are presented as the value before any medication change occurred. A change could be up or down (both occurred) and could have occurred after one week or after 73 weeks on the diet. The HbA1c value from just before the change in medication gets carried forward throughout the rest of the graph of HbA1c that features in the discussion, see later.
But let's take the study at face value. I'll begin with the top section of Table 2.
Now you may be forgiven for skipping down to the carbohydrate intake line where the massive increase from 47.7% to 66.3% is noted. This would be a mistake. First line to note is the SPONTANEOUS drop in caloric intake noted in the vegan group. 1798kcal/d to 1366kcal/d. This compares with the ENFORCED caloric restriction on the ADA diet, which did not alter the macronutrient ratio in any way from the pre study diet. This is where Stephan found the study interesting. I have a couple of concerns about the explanation of the vegan diet providing limited reward which I would need to see addressed before going along these lines:
No one sells a diabeso-genic diet for rats based on starch. There are two main techniques for making rats fat and glucose intolerant. The first is sucrose, the second is Crisco. You can also do it with lard, but that's a rat oddity and I doubt it applies to humans, as far as I can see.
The vegans reduced their caloric intake percentage of trans fats from 2.3% of calories to 1.1% of calories. The ADA dieters didn't. p < 0.0001.
If you want to look at either total fructose or HFCS or table sugar intake comparison between the two groups I'm afraid you will have to ask Barnard for this information, it's not in the paper. But I think it's safe to assume the vegan diet included the advice to consume minimal sucrose/HFCS and the ADA, with it's role in promoting diabetes to maintain medication sales, had some. Possibly quite a lot. Who knows? Well of course, as I said, Barnard knows; but he's not saying. Let's just assume p < 0.0001 again.
I would want these two factors to be controlled for before I would become interested in looking at food reward effects of the relative diets, important though this might be.
You may have noticed that the concept of components described as "a percentage of calories" has already sneaked in.
So next we have to look at that % of calories from carbohydrate expressed as a number of grams per day of carbohydrate, especially on that vegan "high carbohydrate" intervention.
Pre study the vegans were eating 47.7% of 1798kcal as carbohydrate, ie 858kcal or 191g/d. By week 74 of the study they were eating 906kcal/d or 201g/d of carbohydrate (66.3% of 1366kcal).
So the total carbohydrate intake increased by 10g/d. Did you notice that on initial browsing of Table 2? Tee hee.
Two other factors have to go in. Fibre, included in total carbohydrate, increased from 10.8g/d to 21.7g. Most of this was insoluble fibre, ie it literally became vegan cr*p. You know, wipe, flush... Let's see, 201g-11g=190g! Wow!!! I couldn't have asked for the arithmetic to work out that neatly, just a happy coincidence. Made me laugh anyway.
So, is this a high carbohyrdate diet, compared to the SAD? No. NO.
Weight loss. For the later months of the study subjects were weight stable. During this phase the vegans and ADA dieters deteriorate at remarkably similar rates. During the initial few months, when both groups demonstrated a very small improvement in HbA1c, there will have been marked weight loss. We don't know what the calorie intake was during this phase. If it was less than 1366kcal per day with a similar % of calories from carbohydrate this might actually have been a carbohydrate restricted diet!
If we briefly scan the HbA1c graph
it is pretty obvious that the ADA dieters stopped losing weight at 11 weeks and the vegans stopped losing weight at 22wks. End of weight loss is the time at which you would expect HbA1c to stop dropping and metabolic deterioration to set in for both arms of the study. Please note that we have no idea how many patients at a given time point are providing true HbA1c values or how many data points include "carried forward" values before a medication "change". Over 74 weeks 17 vegans actually reduced their meds by an unspecified amount and 10 increased them, again by an unspecified amount. So 17 out of 49 subjects provided "hypothetical" HbA1c values for some unspecified amount of the study period. Go figure. Must have been written by a low fat vegan!
I would be willing to bet that the 17 vegans reducing their meds were all "carried forward" from the first 22 weeks and the 10 increasing their meds were in the >22 week period... You don't generally increase meds in type 2 diabetics during weight loss.
You can't lose weight for ever.
Once weight loss stops and human adipose tissue consumption ceases we are back to starch for diabetics and progressive HbA1c deterioration.
What would you expect?
It is quite clear that healthy humans can consume a very wide range of macro nutrient ratios. If you are glucose intolerant you cannot. Are you glucose tolerant? Are you sure? Russian roulette?
In summary: Any person with diabetes who considers that dropping their HbA1c from 8.05% to 7.71% (intention to treat basis) by following 74 weeks of low fat veganism will help them, is deluded. If they think it is going to get them out of the queue for dialysis, they're wrong. I suppose that weighing 4.4kg less in the dialysis room might give you a better self image, but this benefit might be wasted if you are blind by that stage...
A couple of add-ons:
Dr Haimoto in Japan produced this result by simply restricting carbohydrate to 137g/d of which 14g was fibre. The study was not intended to cause weight loss and it didn't, what there was was non significant. So this improvement in HbA1c is WITHOUT weight loss. Neat hey? Better end-study HbA1c values would need lower total carbohydrate intake.
Of course Dr Haimoto gives us all of the individual HbA1c plots and of course the individual weight change data. No intention to treat here, just numbers, including the data from the non compliers.
Then we shouldn't forget the Nielsen and Joensson study in Sweden
These people were asked to lose weight (not spontaneous). Note that maximum fall in HbA1c was at the time of max weight loss which occurred at the time of maximum compliance, here are the individual weight loss plots. No intention to treat fudge. Carbohydrate limit was recommended to be at 80-90g/d but some carb creep was noted... Despite this, note the downward trend in HbA1c over the years. Sorry there was no room to extend the graph to the full 44 months.
Oh, and for chopping off the top of the last graph!