LC v. LF Diet Comparison Study Shows Calories Determine Weight Loss

This time in a somewhat heterogeneous (mixed) population of Type 2 diabetics.  In type 2 diabetes, randomisation to advice to follow a low-carbohydrate diet transiently improves glycaemic control compared with advice to follow a low-fat diet producing a similar weight loss

From the paper:

In the low-fat group, the mean age was 62.7±11 years, there were 13 men and 18 women and the mean duration of known diabetes was 8.8±6.2 years.Corresponding figures for the low-carbohydrate group were 61.2±9.5 years, 14 men and 16 women and a duration of known diabetes of 9.8±5.5 years. Age, sex distribution and known duration of diabetes did not differ between the groups (all p>0.5).

This was a two year study with assessments at 6, 12 and 24 months along with an additional assessment of intake at 3 months.  These time frames match up with Shai.   Unlike Shai, the dietary prescriptions for both LF and LC were calorie restricted -- 1600 cal/day women, 1800 cal/day men and for LC there was no induction/phasing in more carbs.  Average intake was assessed with 3 days of journaling including 1 weekend day.  The oral medication usage at baseline was quite a mixed bag but similar between groups.  For LF/LC the numbers were:  diet only - 2/2,  oral hypoglycemics - 13/15, oral hypo + insulin - 11/10.  The breakdown of statin use by number of patients wasn't reported, though average dosage was provided in the summary table.  

For most of the participants, the intervention was comprised of some initial dietary instruction and monthly group meetings.  Dietary instruction was provided at baseline, 2, 6, and 12 months by different physicians.  A dietitian provided the first week's menu, recipes, and was available to ask questions at group meetings but individual counseling was not provided for all.  For some reason, during the first year, the dietitian had individual consultations with 4 LC'ers and 3 LF'ers.  Retention was good as only 3LF/4LC were lost for failure to attend meetings.    However at 24 months, the dietary record submission was quite different between groups with only 4 LF'ers (out of 28 completers) not submitting records while 10 LC'ers (out of 26 completers!) did not submit.

So, here are the results.  (Intent to treat analysis includes all 61 participants at all time points although 3LF and 4LC did not complete the study).  At 6 months, the weight loss was the same:   LFD = 4.0 ± 4.1 kg, LC = 4.3 ± 3.6 kg.  The plot for all time points is shown at right.  The weight change was statistically similar at all time points.  ● = LC, ■ = LC

Here's the major significant finding some low carb advocates are crowing over. This is change in HbA1c from baseline. First, I don't think this is anything to write home about for low carb. Yes, at 6 months it had dropped significantly, but it was back to baseline at two years. But let's look at that data with some absolute numbers shall we? I cleaned up the portion of Table 1 related to HbA1c (removed p-value columns). If you look at the full text, there is only one p-value for differences at any time point that even comes close to a level of significance. That is the HbA1c in mmol/mol (which is what is plotted in the figure) for the LC diet at 6 months compared to baseline. And if one looks at the HbA1c more commonly measured, we're talking no real impact whatsoever! It seems to me that highlighting this in the manner in which it was done shows a bit of a bias here.

Remember, these are T2's that are mostly receiving diabetic meds. I don't see the point in taking the time to include the entire med breakdown here, I'll put 0, 6, 12, 24 numbers in ( ). If you check, the only meds that appear to differ between the diets are insulin (LF: 39,38,38,36 ; LC: 42,30,33,35) and glibenclamide (LF: 0.4,0.3,0.3,0.3 ; LC: 1.1,0.5,0.5,0.1). I'm thinking that the mix of what oral meds were taken etc. is why the glibenclamide is not mentioned, probably one person taken off meds? In any case, only the difference in insulin dose is mentioned by the authors.  This is important, IMO, if one views insulin as a drug rather than a supplement for an underactive pancreas much like thyroid meds are viewed.  It should be no surprise that insulin needs would be reduced slightly according to the glycemic load of one's diet.  How much insulin any diabetic needs is not a measure of health or management of disease unless you're at war with a hormone critical to sustain life.

The only other thing mentioned in the study discussion was HDL.  Here are HDL levels for 0, 6, 12 and 24 months in mmol/l  ± SD:

LF: 1.09 ± 0.29, 1.10 ± 0.30, 1.17 ± 0.24, 1.20 ± 0.32 (42 → 46 mg/dL, 4 pt increase)

LC: 1.13 ± 0.33, 1.25 ± 0.47, 1.24 ± 0.38, 1.36 ± 0.44 (44 → 53 mg/dL , 9 pt increase)

Eh ... Given the variability of cholesterol determinations, while the LC group made it over 50, this is hardly a compelling case for LC over LF. When you look at the "compliant completers" (we are not given numbers here per group, but we do know that at 24 months this at most included 24 LF'ers and 16 LC'ers).

So ... what can we learn from this study?   First let's take a look at the diets themselves.  From the paper:

%F / %C / %P:       Low Fat:  30/ 55-60 /10-15            Low Carb: 50/ 20 /30

I find it odd that the macro percents were fixed for LC but varied for LF.  And at first glance that protein differential popped out at me!  I mean, the LC group would have consumed 2-3X the protein of the LF group and I couldn't help but think of the LoBAG studies and the insulin sensitizing and HbA1c improving qualities of protein.  (See here).   I also found this study being touted by a low carb advocate in the context of high fat diet wins.  By LCHF standards, this is not a high fat diet.  A piddly 50% fat?  Heck, that's almost SAD-like.  Usually studies like this are criticized for not using an appropriate low carb diet, but when the headlines sound favorable, what's a little deviation amongst friends?   

In any case, we are provided intake data for compliant completers. That is those who attended meetings, and whose submitted 3-day food logs (for which they were provided scales/measuring devices) at the 24 month point demonstrated adherence to calorie targets (1600F/1800M).  The weight losses are even more dismal!  Only 3.1 kg on average for the LF, 3.6 kg for the LC group.  The HbA1c and HDL is shown at right.  Not much different than the whole group.  For the HbA1c thing, the values are a total wash between groups from the 1 to 2 year mark and if you want to to hang your hat on HDL, have at it!

So now, I ask, what can we learn -- if anything -- from this study?  Well, for starters, we've got the usual result that over the long term it's all a wash.  But this one is a little different to me.  Because at least the target caloric intake was counseled to be the same.  Voila!  And there went the usual differences in short term weight loss out the window!  It wasn't because the LC'ers didn't eat enough fat and calories, although to this day if you go on a low carb forum and ask for help as to why you've been LC for two years and only lost like 10 lbs, you'll probably be asked if you're eating enough should you dare mention counting/tracking calories!  

No ... this study shows calories count, macros, not so much.

Sadly, we don't get the intake breakdown for just the compliant completers, but this is the breakdown for those who submitted 3-day dietary records at each time point.  I note that even fewer reported at 12 months, and I remind you that at 24 months, only 4 LF'ers didn't report while 10 LC'ers failed to report.

I find the progressions for both groups odd.  So they were more compliant 3-6 months, slacked off a bit at 12 months then clamped down big time at two years?   Not really buying that!  So they the LF and LC groups were consuming ~1450 and ~1250 cal/day respectively at the two year mark despite counsel to consume an average of 1700 cal/day?  

The LF group never exceeds the 50% mark for carbs, and protein is roughly 20% despite prescribed at 10-15%.  Meanwhile, the LC group never makes it to 25% protein despite the 30% target, and they also miss their carb target by quite a bit consuming ~25% carb at 6 mo and 1 yr, and 30% at 2 years!  And fat?  well, the LC group were higher fat consumers at baseline!  And look at the low fatties, they barely altered fat consumption, though I suppose the LC group can write home about eating "a lot" more fat for the first year.  The 2 year fat consumption is only a smidge more than baseline.   Let's put some grams to this, shall we?  At 6 months:     LF:  1553 cal,  190g carb,  50g fat,  82g protein    and   LC:  1384 cal,   87g carb,   75g fat,  83g protein.  So, actually absolute protein was similar and this just goes to show that if anything, metabolic "tie" weight-loss-wise goes to LF as they had same weight loss consuming over 150 cal/day more cals.  And at 2 years?  Well:  LF:  1459 cal,  171g carb, 50g fat, 73g protein    and    LC:  1251 cal, 97g carb, 61g fat, 75g protein.  It's no wonder such minor differences were observed.

Do we learn anything from this study?  Something about reality I suppose.  That if you're over 60, somewhere in the neighborhood of 1500 calories per day is likely maintenance for you, because the nominal weight loss indicates the participants were actually consuming close-to-maintenance levels.  Either the LC'ers have to consume roughly 200 calories less per day than LF'ies, or they are more prone to lying on food logs (grin).   It's calories and a caloric deficit that elicits weight loss, not insulin.  But that won't stop LC advocates from misrepresenting this study as a shining star for the cause.  More on that last point in a follow up.  Oh ... and the Swedish LCHF trend may be overstated ;-)