That “Low-Fat Beats Low-Carb” Study

I’d have to dig through my Outlook archives to say for sure (and I won’t), but this one may have set the new record for the number of Did you see this?! emails I received.

If you follow the health news (and if you haven’t been on a retreat in the wilderness or otherwise deprived of the internet for the past week), you already know a new study declared that low-fat beats low-carb for weight loss … once and for all, end of story, final word, move along folks, there’s nothing else to see. Let’s look at some media treatments of the news.

From a BBC article titled Low-fat diets ‘better than cutting carbs’ for weight loss:

Cutting fat from your diet leads to more fat loss than reducing carbohydrates, a US health study shows.

Scientists intensely analysed people on controlled diets by inspecting every morsel of food, minute of exercise and breath taken. Both diets, analysed by the National Institutes of Health, led to fat loss when calories were cut, but people lost more when they reduced fat intake.

From a Washington Post article titled Scientists (sort of) settle debate on low-carb vs. low-fat diets:

Seeking to settle the debate, scientists from the National Institutes of Health set up a very detailed and somewhat unusual experiment.

They checked 19 obese adults (who were roughly the same weight and had the same body-mass index) into an inpatient unit at the NIH clinical center, for two-week increments.

For the first five days of each visit, the volunteers were given a baseline diet of 2,740 calories that was 50 percent carbohydrate, 35 percent fat and 15 percent protein. This wasn’t very different from what they were eating before. But for the following six days, they were given either a low-fat diet or a low-carb diet, each having 30 percent fewer calories. Each participant was also asked to exercise one hour a day on the treadmill.

After analyzing everything from how much carbon dioxide and nitrogen they were releasing to their hormone and metabolite levels, the researchers concluded that the calorie-per-calorie, low-fat diets beat out low-carb diets.

My favorite headline was from the Los Angeles Times: For fat loss, low-fat diets beat low-carb diets handily, new research finds.

Low-fat won handily? Must’ve been real butt-whippin’ demonstrated in those results.

It is a central dogma of the low-carb lifestyle: that while avoiding carbohydrates will force the human body into fat-burning mode, any diet that fails to suppress insulin will trap body fat in place and thwart a dieter’s hope of shifting to a leaner, healthier body type.

But researchers from the National Institutes of Health have found that the hallowed creed of Atkins acolytes doesn’t hold up in the metabolic lab, where dieters can’t cheat and respiratory quotients don’t lie.

So it was the Atkins diet that got a butt-whippin’ by low-fat. I repeat: The Atkins Diet. I don’t know about you, but I would take that to mean the diet prescribed by Dr. Atkins.

And how long did the diets last? Let’s check the LA Times again:

As the 19 subjects recruited for the current study dieted their way through four weeks of low-carb and low-fat regimens, Hall and his colleagues conducted brain scans and other tests to glean how diets with differing nutrient compositions affected their mood, motivation and sense of satisfaction.

My goodness … they dieted their way through four weeks of low-carb and low-fat regimens, according to the LA Times. I take it that means the subjects were on diets lasting four weeks.  That ought to be long enough for real differences to emerge.

But wait a second … I seem to recall the Washington Post describing the diets a bit differently …

But for the following six days, they were given either a low-fat diet or a low-carb diet, each having 30 percent fewer calories.

Hmmm, we seem to have conflicting stories here.  Four weeks vs. six days on each diet.  Perhaps we should check the study itself – which I did. After reading it, I suspect we have a case of “let’s design a study to produce the results we want.” In fact, I can’t help but imagine the conversation:

“Okay, Jenkins, grab your laptop and step into my office. We need to design a good, solid, scientific study to settle this low-fat versus low-carb issue once and for all.”

“Excellent, sir. You mean in a metabolic ward and everything?”

“Exactly. Let’s start with the low-fat portion.”

“Well, sir, the usual definition of a low-fat diet is less than 30 percent of total calories, so I suppose we should—”

“Don’t be ridiculous, Jenkins. If we’re going low-fat, let’s really go low-fat!”

“Ahh, I see. Something like the very-low-fat diet Dr. Ornish pushes. Okay, 10 percent of total calories, then.”

“Damnit, Jenkins, you’re not listening! I said really low-fat! Let’s go down to, say, 7.7 percent of total calories from fat.”

“So a diet nobody would ever follow voluntarily in real life for any length of time, then?”

“Correct. Now, for the low-carb side of things …”

“That’s easy, sir. The Atkins books recommend starting at 30 grams of carbohydrate per day, so—”

“Good grief, man, we can’t put human beings on such an extreme diet!”

“But—”

“So we’ll go with 140 grams per day, including, say, 37 grams of sugar. That should be a fair comparison for our purposes.”

“But that’s twice as many carbohydrates per day as the Atkins diet recommends even in the maintenance stage, must less when starting a—”

“Well, it’s complicated, Jenkins, so let me explain it this way: shut up.”

Actually, the explanation isn’t particularly complicated. Here’s a quote from the full study:

Given the composition of the baseline diet, it was not possible to design an isocaloric very low-carbohydrate diet without also adding fat or protein. We decided against such an approach due to the difficulty in attributing any observed effects of the diet to the reduction in carbohydrate as opposed to the addition of fat or protein.

In other words, they didn’t want to add or remove protein from either diet, and they didn’t want to add fat to the low-carb diet or carbohydrates to the low-fat diet. They wanted to compare restricting carbs to restricting fat with no other changes, period.

Okay, fine. But in that case, the “low-carb” diet is nothing like the low-carb diet recommended by the Atkins diet books, or by any doctors who promote low-carb diets. So the accurate conclusion and/or headline should be something like Extreme low-fat diet produces more fat loss than a sort-of, kind-of, almost-low-carb diet … at least when the diets last six days.

Yup, six days. It was the Washington Post’s description of the duration that was accurate. The L.A. Times got it wrong. Based on those six days, the researchers then describe in their paper how computer models predict substantially more fat loss for the low-fat group if both diets lasted six months.

Uh-huh. I rank that up there with Al Gore claiming his computer models can accurately predict the climate in 2050 … even though those models didn’t accurately predict the previous 10 years. I’m a programmer, so trust me on this: computer-simulation models tell you what you tell them to tell you. The only way we’ll actually know how these diets perform over six months is to keep people on them for six months.

And we’d also want more than 19 people involved. I just wrote a post last week demonstrating how random chance alone can create “significant” differences in small study groups. Nineteen people, diets that lasted a whopping six days … I wouldn’t bet on those results being reproduced with large groups over a long time.

But about those results … the Los Angeles Times assured us the low-fat diet beat the (ahem) “low-carb” diet handily. So what were the big differences in outcomes?

Well, people on the low-fat diet lost (on average) 1.296 pounds of body fat. People on the (ahem) “low-carb” diet lost (again, on average) 1.166 pounds of body fat. The difference was therefore just a shade over one-tenth of one pound. If you don’t believe random chance can produce that trivial of a difference in a study group of 19 people put on diets lasting six days, I suggest you take a class in statistics.

But wait … did I say 19 people? Well, that’s not quite true. According to the paper, 19 people were enrolled in the study – 10 men and nine women. The study had a crossover design, meaning everyone goes on one diet, then goes back to normal eating for a couple of weeks, then goes on the other diet. They’re randomly assigned to do one diet or the other first.

But the results table shows n=19 for the (ahem) “low-carb” diet and n=17 for the low-fat diet. That means two of the subjects didn’t complete the low-fat diet. So I can’t help but wonder why the researchers didn’t simply toss the results for those two people from the study altogether. Why calculate their results on the (ahem) “low-carb” diet into the average if they didn’t finish the other diet? I thought the goal here was a head-to-head comparison of the same people on different diets.

I also can’t help but wonder why, given the small group, the researchers didn’t just show us the full results for everyone. In studies with hundreds of subjects, sure, you pretty much have to present group-average results to make sense of the numbers. But for the 17 people who completed both diets, heck, just show us everyone’s results and stick the averages at the bottom of the table.  If some individuals lost a lot more weight on low-fat vs. low-carb or vice versa, that would be worth knowing.  It would also be worth knowing if one or two outliers skewed the averages for the groups.

Well, apparently that did happen.  I found this in the paper:

The data were analyzed using a repeated-measures mixed model controlling for sex and order effects and are presented as least-squares mean ± SEM. The p values refer to the diet effects and were not corrected for multiple comparisons. One female subject had changes in DXA % body fat data that were not physiological and were clear outliers, so these data were excluded from the analyses.

Uh … okay.  I’d sure like to see those individual results, though.

All those complaints aside, there were some interesting results in the study tables (again, keeping in mind the small groups and short durations). During the six-day diets, triglycerides dropped by 17.5 points in the (ahem) “low-carb” group, and by 4.3 points in the low-fat group. For total cholesterol, the drop was 8.47 points in the (ahem) “low-carb” group and 19.1 points in the low-fat group. HDL dropped by 2.67 points in the (ahem) “low-carb” group and by 7.27 points in the low-fat group.

So if low triglycerides and high HDL are indicators of heart health (and if these results are actually meaningful), I’m sticking with a lower-carb diet … but one with more fat, thank you very much, because I want my HDL to go up, not down.

The results I found most interesting were for glucose and insulin. In the (ahem) “low-carb” group, glucose dropped by an average of 2.69 points … but in the low-fat group, glucose dropped by 7.1 points. So it’s clearly possible to reduce glucose levels with a very low-fat diet, despite the high carb intake, if calories are restricted enough.

This study has been hyped by the anti-Taubes brigades as a refutation of the insulin hypothesis, but the tables show very little difference in insulin levels. The (ahem) “low-carb” group showed a drop in fasting insulin of 2.76 points, while the low-fat group showed a drop of 2.04 points. Nonetheless, here’s how the researchers described the difference:

The experimental reduced-energy diets resulted in substantial differences in insulin secretion despite being isocaloric.

Hmmm … I’m thinking there’s a reason they chose the word “substantial” instead of “significant.” Let’s check the tables again …. Yup, the p value (RC versus RF) for the change in fasting insulin is .48. The threshold for “statistically significant” is .05 or below. So the difference here wasn’t even close to significant — in a study some people are waving around as proof that insulin levels aren’t a factor in the ability to lose body fat … perhaps because they read what the researchers wrote in their conclusions instead of checking the study tables.

And by the way, the p value (RC versus RF) for change in body fat was .78 — so unless I’m misinterpreting the meaning of p value (RC versus RF), we would interpret that as “statistically, the odds of this difference being due entirely to chance are 78 percent.”

Within the obvious limitations, the study does show that restricting calories can produce a drop in insulin even when the overall carb count stays the same. So it’s not as simple as carb intake = fasting insulin level. Total energy intake figures into it as well.

That being said, it would be very, very interesting to see what the differences in insulin levels (among other results) were if 1) the study ran much longer, 2) there were more than 17 people who completed both diets, and 3) the “low-carb” diet was actually low-carb and didn’t include 37 grams per day of sugar.

I believe the less-hype, more-substance reporting on the study was in an article that appeared in Forbes magazine online:

But a well-controlled new study finds that – at least in the lab – low-fat might be slightly better for weight loss over the long term. That does not mean that we should all revert to the low-fat insanity of the ’80s and ’90s. Rather, the more valuable take-home message might be that rejecting carbs may not be so necessary for long-term weight loss as many of us believe, and that a nutrient-balanced diet is probably the smarter strategy in the long term.

And frankly, whatever kind of diet is most doable for an individual is probably the one to be on. If it’s easier to stick to low-carb than low-fat, then by all means do it. But a balanced diet is still king.

Bingo. It’s certainly possible to lose weight on a high-carb, very-low-fat diet. It’s possible to lose weight on any diet if you restrict calories enough. I tried a Pritikin-style diet (10% of calories from fat) twice, and lost a bit of weight both times – and then I had to quit both times because I was miserable, hungry all the time, and eventually felt too lethargic and depressed to continue. Meals were an exercise in monkish discipline, choking down tasteless food and trying to convince myself I was fine with it.

Now I’m not miserable, not hungry all the time, and never depressed … which means when people hype a study like this as “proof” that a low-fat diet is better than an (ahem) “Atkins” diet, I can enjoy a hearty laugh.