Examine publishes rigorous, unbiased analysis of the latest and most important nutrition and supplementation studies each month, available to all Examine Members. Click here to learn more or log in.

Quick Navigation

Issue #52 (February 2019)

From the Editor

Volume 1

In NERD #41, volume 2, we covered the DIETFITS trial which examined how certain genotypes or insulin production capabilities impact weight loss on healthy low-carb or low-fat diets. The main outcome was that neither mattered for weight loss, which gives another point to the theory that the best diet is the one you can stick to. In our coverage of DIETFITS, we mentioned that secondary analyses were planned to see what else the data speaks to. One of those analyses was published last month[1].

As a quick refresher, the participants of DIETFITS were generally healthy and had body mass indices ranging from 28 to 40. The participants were then put on either healthy low-fat or low-carb diets for twelve months, with the primary outcome of the study being weight loss. As mentioned above, the researchers didn’t find any evidence that certain genes or insulin secretion influenced weight loss on low-fat versus low-carb diets.

As an aside, this is interesting, since the study covered in NERD #51, volume 1’s NERD Mini found that insulin levels actually did affect weight loss. There are a few reasons I can think of that could account for the difference between that study and DIETFITS. I’ll state three. The first is that the populations were different; DIETFITS looked at generally healthy people, whereas the study covered in the NERD Mini found an effect of insulin levels only in people with prediabetes. Also, the insulin measurements between the two studies differed: DIETFITS used insulin levels 30 minutes after an oral glucose tolerance test, whereas the other study used fasting insulin levels. Finally, the study covered in the NERD Mini was a secondary analysis, whereas DIETFITS was designed to directly test the question of how insulin mediates different diets’ effect on weight loss. Secondary analyses are more prone to error, so the NERD Mini study could be a false positive.

Speaking of secondary analyses, the latest DIETFITS secondary analysis didn’t touch the weight loss question — that will probably be a subject for future secondary analyses. Instead, the paper published last month explored how dietary saturated fat intake in the context of a low-carb diet affected the DIETFITS participants’ lipid levels. In other words, they attempted to answer the question: in the context of a healthy low-carb diet, how does saturated fat intake impact lipids? In general, the participants increased the percentage of saturated fats they took in over the course of the study relative to their baseline diet. However, the total grams of saturated fat didn’t really change that much, since the participants cut the amount of food they took in.

Within this context, the researchers found that triglyceride levels creeped up with the saturated fat intake percent, but when they adjusted for carbohydrate intake, the correlation disappeared, suggesting that carbohydrates were what actually led to the bump in triglycerides. This jibes with a meta-analysis we cover in this volume of the NERD examining low-carb diets’ effect on blood lipids. Read on for the nitty-gritty.

Gregory Lopez, MA, PharmD
Editor-in-chief, Nutrition Examination Research Digest


Volume 2

While taking a dive into the literature to look for articles to review for the NERD, I came across an interesting opinion piece concerning the Prevención con Dieta Mediterránea (PREDIMED) trial. I’ve previously opined about PREDIMED in these pages myself a couple of times, but the article offered some unique insights that I hadn’t considered previously.

PREDIMED was a major randomized controlled trial that aimed to compare a Mediterranean-style diet supplemented with either extra virgin olive oil or nuts to a control diet. It’s a pretty big deal for a few reasons. One was that it was large, involving over 7,000 people, which allows for more precise effect measurements. Also, it measured actual clinical endpoints as opposed to surrogate markers, like cholesterol levels. Finally, it measured these endpoints over almost a five-year period, which is pretty long term as far as RCTs in nutrition go.

PREDIMED’s primary outcome was a combination of heart attacks, stroke, and cardiovascular-related deaths. The trial was stopped early after a planned interim analysis (i.e., an analysis of the results to date done before the trial’s official end) found that people following the Mediterranean-style diets saw their risks for this combined outcome drop by around 30% compared to a control diet. Early stopping is usually done when extraordinary benefit or harm is found in one of a clinical trial’s arms since if one intervention is clearly helping or hurting people compared to another, it would be unethical to continue the trial. Early stopping also saves resources; if there’s a clear answer to the main question you’re exploring, why continue? In order to avoid bias, it’s important that early stopping is preplanned, and statistical corrections are made. PREDIMED did lay these criteria out ahead of time, so nothing was amiss in terms of the early stopping.

At least, nothing was amiss as long as the study was properly randomized. As I mentioned in this space previously, the original PREDIMED paper was retracted last year due to problems with its randomization process and was reanalyzed. The authors of the opinion piece mention that this means that PREDIMED was not really a randomized trial, which is a big strike against it. Furthermore, the reanalysis didn’t satisfy the rules for early stopping, meaning that the trial should have probably continued. In fact, PREDIMED has continued with follow-up data, but no analysis of how the three groups compared in terms of the primary endpoint has been released yet.

These are all pretty big issues, but the one that stood out the most to me was the authors’ discussion of the effect on the secondary literature surrounding PREDIMED. They note that over 200 secondary PREDIMED articles were published before the retraction, and it’s far from clear how the retraction and republication affects these papers’ conclusions. Furthermore, they document how some key data is inconsistent between the secondary publications, suggesting that there could be deeper problems concerning PREDIMED.

While it’s always a good idea to lower your confidence in conclusions that stem from secondary analyses, it looks like it’s especially important for conclusions from secondary analyses of the PREDIMED trial.

Gregory Lopez, MA, PharmD
Editor-in-chief, Nutrition Examination Research Digest

See other articles in Issue #52 (February 2019) of Study Deep Dives.