I thought of just updating the last post, but this is long enough on its own to warrant a Part II of sorts. If you haven't read it yet, may I suggest: LC v. LF Diet Comparison Study Shows Calories Determine Weight Loss. What distinguished this study from many other comparison studies, was that there was an attempt at keeping calories constant between the diets. Usually LC is ad libitum, which, especially the first time one follows a low carb diet, tends to substantially spontaneously reduce caloric intake more than what is generally prescribed for a "responsible" calorie restricted diet (CRD). This was also a longer term study and also compared two different fixed prescribed diets rather than the usual Atkins-style induction/progression formula. Presuming compliance, then, the study controlled for calories and sought to look at the effects of just the macronutrient composition of the diet.
In my opinion, the rather negligible differences for both groups after a year and continuing through two years show that neither dietary intervention was particularly effective, resulting in under 10 lbs weight loss and only transient improvements in HbA1c as measured by one method only. So why was that? Well, the prescribed caloric intake was fixed at 1600 cal/day for women and 1800 cal/day for the men. Right there this part of the study design pretty much doomed it to failure. While men generally have higher caloric needs than women, total daily energy expenditure varies so widely between individuals that clearly some women have higher needs than some men, and some men and women have higher needs than others of their gender.
When I read the study, I thought that the 1600/1800 cal/day was a bit high but didn't pay much attention to the kJ's: 1600 cal = 6,694 kJ , 1800 cal = 7,531 kJ. I had looked at the caloric breakdown of the "diets" in the last post, but for some reason neglected to look at the breakdown of the baseline diets. So now I look at the baseline intakes -- based on 3 days logging for which the subjects were provided scales/measuring devices and included one weekend day.
LF group = 7569 kJ = 1809 cal LC group = 7071 kJ = 1690 cal
Hmmmmm .... Perhaps the reason conventional advice "fails" for weight loss is because a caloric deficit is not even prescribed in many cases?? Ahh ... but maybe this wasn't really intended to be a weight loss study, rather a diabetes management one. I think this excerpt from the intro indicates otherwise:
Weight loss in obesity generally leads to improvement in cardiovascular risk factors and glycaemic control. However, few randomised studies have specifically targeted type 2 diabetes to compare the effect of different diets in this respect.
How is it nobody thought to alter the study protocol from the get-go based on the fact that the calorie prescriptions were so darned close to maintenance calories for the participants? I've said this many times here, that caloric needs estimated by formulas, rules of thumb, and whatnot are notoriously erroneous, and almost universally overestimated. What I find even more strange was that this current study design was based on results of a pilot study!
The size of the study was based on an earlier 6 month pilot study of 28 participants with type 2 diabetes who were randomised to the same diets as in the study presented in this paper. Twenty individuals completed the pilot study and both diet groups achieved similar weight reductions, while HbA1c levels tended to be lowered in the low-carbohydrate group only, without taking change in medication into account (low carbohydrate, p = 0.068; low-fat group, p = 0.8). Based on these results, the study sample was increased to at least 30 individuals in each group in the present study.
In my opinion, when calorie restriction is part of the intervention, efforts should be made to ensure the calorie deficit is as uniform as possible. Two ways I've seen this implemented in better studies are:
- 1-2 week lead-in period where usual intake is logged (or even better tightly monitored) to determine weight-stable intake
- Measuring REE and assigning a caloric intake on a %REE basis or estimating TDEE from REE and assigning a fixed deficit.
This study did neither, it just assigned an arbitrary calorie level based on gender, and somehow didn't notice in either the pilot study or in the baseline dietary logs that these prescriptions were unlikely to produce weight loss. So this study gives us a clue perhaps as to why "traditionally recommended [diet] for the treatment of type 2 diabetes in Sweden, with 30 E% from fat (less than 10 E% from saturated fat), 55–60 E% from carbohydrate and 10–15 E% from protein" apparently fails patients.
I've augmented the intake data from the last post to include a column at right for the target prescription. I used 1700 avg. cals as the target intake. Note: The n's are for those who completed the 3-day logs at each time point, at 24 months of the 14 non-reporters, 10 were LC and 4 LF. I noticed a few things when I gave this table a second look. First, they probably lost nominal weight because they didn't consume the target calories, although the authors acknowledge that underreporting could be at play here.
Next just scan across each of the macro lines for the LF diet -- they remain essentially unchanged! IOW, the "prescribed" LFD is quite similar to how these people were eating to begin with. Now one interpretation of this is that such a diet is what made these people fat and diabetic. EDIT: But another way to look at this is that the LF "intervention" was more of a control than an intervention. The LF group fell far short of the goal to reduce protein and increase carbs, and supposedly were already eating a low fat diet (end EDIT). I note also that the LC group baseline macros are more in line with the SAD (I have no idea what the SSwedishD really breaks down to) surveys. If that group had been assigned LF, then there would at least have been a nominal reduction in fat. But I would also like to quibble with what constitutes a low fat diet. Despite what certain science journalists would have you believe, the diet (to use an example of one population now at high risk for obesity and diabetes) eaten by those "sprightly" Pima was anything but LCHF: "We estimated that the traditional Pima diet, although seasonably variable, was ∼ 70–80% carbohydrate, 8–12% fat, and 12–18% protein." So what the Swedes, and most people for that matter, consider a low fat diet these days, is not really so.
Did the LC group really eat a "high fat" diet? Nope! They did initially up their fat intake considerably from baseline (39% to 49%), however combined with decreased consumption this resulted in an average of only 2g fat/day more! And by 24 months it was only 5% of calories more, and ... da, da, daaaah! ... at 61g almost a tablespoon of butter's worth less than the 73g/day at baseline.
The more I look at this data, the bigger cluster you-know-what it all seems to shake out to be. Neither group altered their protein consumption much to comply with the prescribed diet guidelines. The LF group in particular actually increased protein slightly when it should have been cut, and while the LC group increased it a bit more, they fall far short of the 30% target. All-in-all, compliance seems extremely poor here, yet I see no real mention of that in the discussion. They did a separate "completers" analysis for those who complied with the target fat percentages defined as <35% for LF (n=20) and >45% for LC (n=12). Check out those n's! Only roughly 1/3rd of the "high fat" group actually ate the prescribed diet. The rest, as they say, actually ate a lower fat (in absolute amounts) diet.
Given as medications were continued, and in some cases (statins) started during the course of the study, compliance with prescribed diet was poor resulting in differences in the interventions being far less stark than the abstract would indicate, this study seems pretty useless in providing meaningful comparisons. The main outcome touted in the title -- the glucose levels -- is, as the title says "transient".
In comments here and elsewhere the usual "it's not a proper LCHF diet" and "imagine how much better the outcome would be with keto diet" responses have been made. That's all fine and good, but then this study should not be touted as how a high fat diet -- in the 70+% range touted by most advocates -- is effective in treatment of diabetes. It may well be an alternative that produces nominally better short term results than a not-very-LF diet. And both were equally relative failures for weight loss and long term diabetes management. To the "what if it was proper keto?" I suggest: Diabetes "Crash" Cures: VLCal vs. VLCarb as it discusses the Westman study referenced by this paper. The Israeli Shai et.al. study that Taubes loves to do a mangled analysis of is also cited:
Interestingly, in a randomised 2 year study from Israel that achieved good compliance, a high-fat diet was shown to induce better weight reduction and improved blood lipid levels than a traditional LFD in obese participants, while the subgroup of patients with diabetes who were randomised to the high-fat diet exhibited the largest reduction in HbA1c levels.
I'm not sure I'd describe Shai as demonstrating good compliance, and again, I get rather tired of the HF v. LF comparisons gleaned from that data when there was a third dietary intervention that goes ignored: the highest carb Mediterranean diet! I've discussed this here: Gary Taube$, Shai-ster (yep, $ sign and play on words and all). The LF diet in Shai was similar to this current study -- aim of 30% -- but calories were a bit lower (1500F/1800M). The LC (HF) diet was Atkins induction then climbing the rungs aiming for 120g/day max. In the end (2yrs), the "high fat" dieters in Shai reported eating slightly fewer fat grams than at baseline, as their fat intake as a percent only increased from 32% to 39% (and carbs were only reduced from 51% to 40%). Yes, the "low carbers" did have the largest reduction in HbA1c: "Among the participants with diabetes, the proportion of glycated hemoglobin at 24 months decreased by 0.4±1.3% in the low-fat group, 0.5±1.1% in the Mediterranean-diet group, and 0.9±0.8% in the low-carbohydrate group." While the reduction in the LC group was the only statistically significant change vs. baseline, it was not reported whether the differences between groups were significant. Given the small numbers of subjects in this subset and the large SD's, I doubt there was a statistically significant difference. In any case, I wonder why the authors ignored the other outcomes in Shai for the diabetics, blogged on here: