Looking AHEAD to a day when Intent to Treat is a thing of the past ...

... it's not a competition, the compliant/successful are analyzed separately, and while we're at it, perhaps we can do away with this notion that any single outcome or cluster of outcomes is somehow more important because we've designated them as "primary" ahead of time.

Such is where I find my thoughts wandering as the Look AHEAD diabetes trial passes in and out of my head as I absorb the discussions of it here and about the web.  

{This rambles, sorry!}

I understand the scientific basis behind intent to treat (ITT), I just think it mostly destroys the meaningful/actionable information us "consumers" (including doctors wishing to provide the best advice possible based on such information) might hope to get from many of these studies.   I think back on my questions when in 2009 I had plateaued out and was eating pretty low carb (induction levels)  most of the time with the occasional "cheat" (which was never a binge).  I wanted to know what the long term healthfulness of such a diet was in maintenance, which in my case involved substantial losses but still remaining excess weight.  I have never found the answer in the science because such a trial does not exist.  Which is why the next best thing is to look at those who have been eating such a diet for long periods and at least attempt to gauge from their results, and ... well ... you know what I think there by now.  But back to ITT and related issues, and let's start with one of those first.

These diet trials always seem to be stacked up as competitions between various means to get to an end.     With the possible exception that such contests might sway both dieters and those doling out diet advice towards trying more effective approaches, the contests are pretty meaningless in the larger picture.   Despite boasts that LC diets are easier to stick to for longer periods, the various studies are wholly inconclusive on these points.  In the end, it matters to me that *I* can stick to a certain diet or lifestyle or whatever euphemism you want to use, and not how many out of 30 subjects in some randomized study were able to stick to their's.  BUT, and here's the biggie, I'm VERY interested in how the 25 subjects who did stick to Diet XYZ fared.  I might not see the same results, but if 25 people stuck to XYZ, lost weight and had improvements in various metabolic markers, and those persisted for years in those who maintained the weight loss, that's what I want to know.  And if those who adhered to the more controversial LMNOP diet lost weight but had worsening metabolic profiles after several years despite maintaining their weight losses, that is important to me too.  In the end, most of these studies belong more in the behavioral journals than the nutrition/biochem/physiology ones.

I get very tired of the Destruct ELMM Street movement out there who seem to think if they repeat it enough it will magically become true ... ahem that pool noodle wielding hustler Gatewing ... that caloie accounting methods are failed approaches to weight loss and the answer lies in the his IONOsphere.

This is simply not true.  Imagine if you will, a "perfect" blood pressure lowering medication is developed.  It is inexpensive and has no known side effects, but you had to take it religiously every 8 hours.  You conduct a trial on this drug vs. placebo, but after the first month, only 10% of the study group has gotten into the groove of taking their pill on time, and 40% have dropped out because their compliance was too poor to maintain sufficient BP control to satisfy the ethics board.  What you're left with are data dominated by 50% of the original group that is complying only well enough to stay in the study that swamps out the performance of this drug in the 10%.  Personally, if I needed BP meds, something like this would not be my option, but I've known enough people taking these who don't like the side effects that might just be motivated enough to set an alarm for every 8 hours, pop their pill and then go about their merry way.  This med works beautifully for them and their blood pressure is flawlessly controlled.  This is the same thing with ELMM -- because deliberate efforts to alter ones diet and/or activity level require work in many cases.  This is why some people do well on Jenny-style plans and the low carb community seems dominated by those (me included) who have difficulty with the sort of structure required for long term implementation of a standard CRD.  To say that ELMM doesn't work because not everyone can stick to it long enough to see results or because long term adherence and permanent habit changes required for maintenance are difficult is just wrong-headed.

Now it is fair to say that ELMM may be bad advice to give to the masses because it is so difficult to implement, just as it would be fair to say that my wonder BP drug might not be highly recommended for the same reason.  But the efficacy of the treatment remains valid.

I realize nothing I write here will ever change a darned thing about how studies are analyzed, but a girl can dream.  And this goes as much for the various medications I may one day have to weigh the risk/benefits of.  For those actually taking some medication as prescribed, what are the outcomes?  Yes, I am concerned about the incidence of side effects, especially ones so severe as to force discontinuation of the med.  But at that point that data point is useless, unless you really think it's worthwhile studying the effects of writing a prescription for a medication and syndrome management, and nothing else in between.  This, sadly, seems to all-to-often be how clinical data is treated.   And this is exactly what's concluded when ITT analyses muddy the waters.  You'll have 30 who start out in a year long study, perhaps 28 of whom "complete the study".  At 93% that's a rather impressive retention rate.  But all that means is that they showed up for and completed all meetings and assessments.  If I'm assigned to the XYZ Diet, I can follow it for a week and so long as I show up for everything, my data gets put in the ITT mix just like the person who takes to XYZ like a duck to water and follows it to the letter for the entire year.  My results are ultimately meaningless as to the efficacy/safety/healthfulness of XYZ except for the fact that I'd be a data point against the practicality of such a plan based on ease of compliance.  To make matters worse, if I'd dropped out at 6 months, my 6 month data would often be carried forward to the end of the study and included in all subsequent time points in ITT.  

Rarely is the compliance bar for "adherence" set very high.  When one looks at the dietary intake data from Shai, for example, we are to believe that in the second six months of year one, the low carbers ate the same or even a little less and gained back some lost weight.  In the Ebbeling study where all food was provided, daily diaries were apparently kept, but we're not made privy to any accounting of some sort of percent (by number of participants) or degree of compliance.  Often compliance is so poor as to render the study results rather moot.  Such was the case in a diabetes study I blogged on back in May (here) pitting LF v. LC.  The table at right shows the targets in blue and the actual intakes.  Whatever the results there, that study did not end up comparing the diets it purported to be testing because very few actually followed the assigned diet.  They did do a separate analysis here for those who complied best, but this was still lacking (see my original post for a more detailed discussion on that).  

It may well not be practical to present, or the numbers so small as to be inconsequential, but I always want to know about the fate of the subjects assigned to XYZ who complied and completed.  Then it's sort of up to me to decide if XYZ is a protocol I could comply with for any length of time or if I want to chance the negative outcomes the dutiful XYZers encountered along with their impressive results in other areas.  But when all I can see is that 30 XYZers lost an average of 5 lbs in a year -- a number hopelessly influenced by dropouts carried forward and those who didn't eat the XYZ diet after all -- there seems to be little value in the outcomes.  I, personally, cannot consider long term VLC diets anything but "do at your own risk" until I see some hard data -- including organ fat scans -- on people actually eating that way in maintenance for several years ... and don't give me the Inuit!

And so here we are again, for the third time in as many days, back to the Look AHEAD diabetes trial. Here we have all the "bad things" listed in the title and opening sentence of this post in one tidy decade long package of a study.  The study purpose was quite narrowly defined -- it was to pit usual diabetes education against "intense" lifestyle (diet and exercise) intervention to see if it reduced the CVD-related outcomes.   It's not all that far from testing whether giving advice improves CVD risk!  Further, the CVD outcomes were severe "events" -- heart attacks, strokes and/or CVD related death.  To put this in perspective, my father-in-law is in his 80's and has had heart disease since I've known him for 20 years now.  I'm not quite sure how long he's been on at least one BP lowering med, but likely at least a decade more than I've known him.  He's had a couple of angioplasties, triple bypass, has some new/returning blockages currently and had a pacemaker put in a couple three years back.  Yet he's never had a heart attack or a stroke and he's still around now 20 years after he enrolled in my study (e.g. came with the hubby package).  He's not a diabetic, but his doctors have always been on him for his weight, which has increased 10% or so as well.  As Stephan Guyenet mentioned in comments here, this study was halted because they weren't seeing enough "events" to distinguish between the two groups.  Which is good news for those considering signing up for such trials -- it probably doesn't hurt to have some degree of accountability/attention/etc. on a regular basis.   

The "dud" here is that they defined the parameters so narrowly -- pitting A v. B and measuring outcome X and looking for differences between groups.  Classic ITT with single minded focus on the primary outcome.   It's set up as a contest -- intensive lifestyle intervention vs. status quo advice -- which is better?   Since Phase I defined goals for the intervention, the ITT analysis of all patients assigned to this group just muddies the water.  At the very least you have two major confounders here:

• Both groups used various pharmaceutical interventions 
• Some members of both groups both lost or gained weight in amounts in line with the weight goals of Phase I of the ILI 

This in the context of there appearing to be very little in the way of documenting/reporting compliance in any meaningful way.   

Some have pointed out in comments that perhaps this is just one more depressing study on how difficult it is to lose and maintain significant amounts of weight.  There are some other more encouraging outcomes here.  First, thanks to Stephan's comments, we learn from Dr. Sharma that the reason the study was halted in terms of the intervention was because:  

"... given a remarkably low incidence of ‘hard’ endpoints in both the interventions and control groups, it became evident the the study would stand little chance of demonstrating superiority of the lifestyle intervention in terms of preventing cardiovascular complications." 

Perhaps they needed to enroll sicker patients to begin with, but as my own n=1 example of my father-in-law demonstrates, standard medical care has done a good job of preventing "hard" endpoints, if not preventing all progression of his heart disease.  But the guy is in his 80's, fully in control of his faculties, mobile and self-sufficient.  Perhaps what this study shows is that some of the newer medications out there, and/or better diagnostics for self-monitoring of blood glucose, insulin pumps, various insulin formulations, etc.etc. are improving the quality of life for diabetics.  I know there are the self righteous out there who consider the bulk of those "burdening" the rest of us with their medical costs as just that -- burdens.  I know there are the conspiratorial out there as well, who think the government, medical establishments, pharma and food manufacturers have proactively acted to make lifestyle intervention next-to-impossible while hooking us on expensive medications.    The irony here is that folks heralding this study as the death of ELMM lifestyle intervention, and even speculating that somehow the researchers are hiding a bodycount of ELLMers who died of other causes from the harm of the intervention, are falling right into the trap set by the conspiracy if there is such a thing.  

I absolutely do not get why the low carb and paleo communities are SO antagonistic towards other ways of eating.  Period.  I will never forget the time, early on at Jimmy's forum, when I wondered out loud if I should try a WW or NS style approach to lose more weight since LC had stopped working for me.  Oh my the nastiness!!  "Good luck with that!"  and basically "don't come crying back to us when you gain all your weight back".  Paleo is literally littered with former vegans and vegetarians who look down their noses at practitioners of that lifestyle -- or gawd forbid someone goes back to eating that way (Don & Tracy).  Honestly, don't admit you ever eat grains, you are an inferior being for not "duh" seeing the light that Wheat Belly hath shone on the evil that is the seed of a grass.  If you want a really good lashing for eating anything grain, dairy or legume, have a gander at Paleoista.com.   Replace your peanut butter with, I'm not making this up, coconut oil spread on raw veggies, and eschew hummus because garbanzos aren't paleo.   I probably don't need to even bring up the bashing that goes on of folks like Melissa M. and Go Kaleo which begins with the fact that neither follows the Primal Directive (TM).  

But the exact nature of the intervention aside, what I would like to know were I diabetic is whether or not weight loss is something necessarily health improving.   What I hear in the medical world is that "losing as little as 5% of one's body weight can reduce the risks".  This comes from studies on recently diagnosed diabetics and/or studies on pre-diabetics and/or groups at risk (family history) who may or may not have early signs of diabetic progression such as IR, IGT, etc.  But forestalled progression or reversal of recently detected diabetes is different than reversing longstanding disease.

I do not think there is much of any doubt that for a large proportion of those predisposed to diabetes, it is obesity that ultimately "causes" the disease to manifest.   If we can swallow the argument that the human genome hasn't changed in the 10K or so years of the Neolithic, surely we can agree that the human genome hasn't changed much in the past half century.  Classic T2 has tracked right along with obesity, and as obesity hits younger populations, we see the disease develop earlier in life.  It is not even in question that Americans eat more.  But what do you tell the 55 year old who is 50 lbs overweight 5 years into his/her diabetes diagnosis?   I think this WAS the goal of Look AHEAD.   

So I hope they'll look at the data another way because I believe there may yet be some valuable information in there.  It may not be one intervention vs. another and whether or not you'll drop dead of a heart attack, but how about this.  Lump it all together and break it out into quartiles or quintiles of weight loss percents.  Who cares how they got there, let's look and see if just the weight changes things.  If losing 10% by whatever means (and keeping it off) makes a difference, that might be a motivating factor.  I realize that if the result is still "no difference",  the tendency may be to dissuade diabetics from making lifestyle changes in favor of pharmaceutical interventions.  Here is where that judgment comes in.  Be honest, most reading this will think less of the person opting for pharmaceutical intervention.   How many times do we need to hear "if only they'd just eat a low carb diet" ... from people who have been doing so and become more and more insulin resistant and, in my opinion, unhealthy (by their own admissions of maladies).  

Personally, the rapid reversals of diabetes in the early days following GBP and that crash diet study will always occupy a box seat in my mind when thinking about these things.  This is not to say that everyone's diabetes can be reversed, or that everyone should consider GBP when all else fails or even before all else fails.  But it serves as a reminder that even the weakest beta cells amongst us are rather more resilient than we've been led to believe, and that diabetes need not be a progression.  Oddly enough, though I believe there are other health-damaging implications of doing this, more dramatic yo-yo dieting may be better than half-hearted modest attempts at losing weight.  Periodically draining the cellular swamp sufficiently so as to restore pancreatic function buys time even if weight is gained back -- provided the draining has been sufficient so as to restore proper sequestration of excesses where they belong -- the fat tissue.  

What's most demoralizing to me here is that in the interests of personal dietary agenda, many are too eager to throw very reasonable lifestyle changes out the window as useless, and worse yet, potentially harmful.  (Please, let's not encourage the Jonathan Bailors out there with claims that one can be doing more harm than good, and yes, even Paul Jaminet seems to be insinuating that if the exercise involved walking one's dog at a brisk pace in the evening this could even be harmful???!!!!).    I will note that most of those I'm seeing doing so generally have no clue what it is to be "simply" obese ...  And I will also note that those suggesting other dietary intervention is superior have NO data to back it up with, only rank speculation by many who still believe carbohydrate consumption causes diabetes.