In the last year, we have seen the consensus against cholesterol shatter and crumble before our eyes. The consensus against saturated fat has remained strong, but in the last week we have seen it begin to soften. Here’s my take.

The Cholesterol Consensus Crumbles

The Scientific Report of the 2015 Dietary Guidelines Committee removed its traditional recommendation to limit dietary cholesterol to 300 mg/d and stated that “cholesterol is not a nutrient of concern for overconsumption”. It did also state that the types of “dietary patterns” that reduce cardiovascular risk are low in cholesterol, but it didn’t argue that those dietary patterns are effective because they are lower in cholesterol. Others argued that this was too general and failed to appreciate the evidence that cholesterol raises cardiovascular risk in “hyper-responders” (those whose blood cholesterol rises in response to dietary cholesterol) and diabetics, and could increase the risk of diabetes.

The eighth edition of the actual Dietary Guidelines was more ambivalent than the report of the advisory committee: “The Key Recommendation from the 2010 Dietary Guidelines to limit consumption of dietary cholesterol to 300 mg per day is not included in the 2015 edition, but this does not suggest that dietary cholesterol is no longer important to consider when building healthy eating patterns. As recommended by the IOM, individuals should eat as little dietary cholesterol as possible while consuming a healthy eating pattern.”

Both reports made reference to low-cholesterol eating patterns, but they did so in different ways. In the report of the advisory committee, this was relegated to the section on eating patterns and was considered secondary to the types of whole foods included in the diet. The advisory committee didn’t even mention this in their section on cholesterol. By contrast, the Dietary Guidelines place this front and center in their discussion of cholesterol. Although the Dietary Guidelines acknowledge that “adequate evidence is not available for a quantitative limit for dietary cholesterol,” they use the dietary pattern argument to recommend purposefully reducing cholesterol intake to “as little as possible,” which is obviously less than the previous cap of 300 mg/d.

Although the Dietary Guidelines have not come to a dietary pattern recommendation that is substantially different from the pattern they would recommend if the traditional limit of 300 mg/d cholesterol stayed, it seems obvious that the final result is an amalgamation of different interests and viewpoints. There is no consensus that cholesterol is no problem, but the consensus that dietary cholesterol is a problem is no more. That consensus has shattered, crumbled, fallen by the wayside.

Nevertheless, both the report of the advisory committee and the actual guidelines remained steadfast that saturated fats should be replaced with polyunsaturated fats to reduce blood cholesterol levels and thereby reduce the risk of cardiovascular disease.

The Ramsden Paper: The Stance Against Saturated Fat Softens

On April 12, things began to rumble and shake. Christopher Ramsden, a medical investigator with the NIH and an adjunct professor with the University of North Carolina Chapel Hill, published an analysis of previously unpublished data from the Minnesota Coronary Survey. This study was one of the most important clinical trials of a dietary intervention in history. Although it is limited by its short duration, it was one of only two double-blind, randomized controlled trials testing the the effect of replacing saturated fat with polyunsaturated fat on cardiovascular disease. It was the only such trial that included women, and with just shy of 10,000 subjects it was by far the largest.

The trial itself has not been completely ignored. The investigators published a manuscript from it in 1989 that suggested there was a slight increase in total mortality in the vegetable oil group but that the difference was not statistically significant.

But this was only a small portion of the data. Ramsden and his colleagues uncovered loads of original documents that shed light on other questions, especially these:

• What did total mortality look like when separated by age?
• What did total mortality look like when separated by the change in serum cholesterol?
• What was the difference in autopsy-confirmed heart attacks?

Media articles have tended to focus on the first two questions. The Ramsden analysis found that people were more likely to die the more that vegetable oil successfully reduced their cholesterol. This was largely driven by subjects older than 65: in this group, a drop of 30 mg/dL or more of serum cholesterol was associated with a 35% greater risk of death. Ramsden and his colleagues openly admitted that this is actually an observational finding nested within a randomized controlled trial, because no one was randomized to have a smaller or larger drop in cholesterol levels. As such, our confidence that it represents a cause-and-effect relationship is lower.

In my opinion, the data on autopsy-confirmed heart attacks is far more interesting: only 22% of autopsies from the saturated fat group showed evidence of a heart attack, whereas 41% of autopsies in the vegetable oil group showed the same evidence. This is a difference between the randomly allocated groups and can be inferred as a likely cause-and-effect phenomenon. Thus, the Ramsden analysis provides new and convincing evidence that replacing saturated fat with polyunsaturated fat nearly doubled the incidence of heart attacks.

There is one caveat here: 295 autopsies were performed, but so far Ramsden and his colleagues were only able to recover half of these. Recovery of additional autopsies could change the estimate, but there is no reason to think that the partial recovery would introduce a bias in assessing the difference in heart attack incidence between groups.

Walter Willett and Frank Hu Respond

Major figures in the public health establishment went on the defensive. In a blog for the Well section of the New York Times, Anahad O’Connor quoted Walter Willett as dismissing the findings as “’irrelevant to current dietary recommendations’ that emphasize replacing saturated fat with polyunsaturated fat.” This is unsurprising: a few years ago Willett referred to an obesity study as “a pile of rubbish” and “dangerous” because the findings could “confuse the public and doctors, and undermine public policies to curb rising obesity rates.” Surely the Ramsden paper could confuse a public that has been told for decades to replace saturated fat with polyunsaturated fat to prevent heart disease and could undermine public policy aimed at convincing the public to make that change, so this is just the type of paper we should expect Willett to dismiss as irrelevant.

Frank Hu said, in O’Connor’s paraphrasing, “the Minnesota trial was not long enough to show the cardiovascular benefits of consuming vegetable oil because the patients on average were followed for only about 15 months” and “pointed to a major 2010 meta-analysis that found that people had fewer heart attacks when they increased their intake of vegetable oils and other polyunsaturated fats over at least four years.”

Hu is right that a major limitation of the Minnesota trial is its short duration. But the longest such trial ever conducted, and the only other double-blinded trial making the single substitution of polyunsaturated fat for saturated fat, was the LA Veterans Administration Hospital Study. That study, which I have discussed previously (e.g. here) showed that vegetable oils increased the risk of cancer primarily after five years and that this increased non-cardiovascular mortality primarily after seven years. There was no difference in total mortality between groups, but the trial suggested that, as time goes on, the picture looks dimmer and dimmer for vegetable oil, not better and better.

The 2010 Meta-Analysis Was Deeply Flawed

The meta-analysis that Hu cited was also deeply flawed. It included the Finnish Mental Hospitals Study as a randomized controlled trial, but this was anything but. The study had one hospital serve a diet based on saturated fat for a few years and a second hospital serve a diet based on polyunsaturated fat for a few years. Then, they switched the hospitals. The point of allocation was the hospital, not the people in the hospital, and there was one hospital in each group. Obviously, no statistical analysis could be done with one hospital per group, so the statistics were carried out on the people in the hospitals. But the people in the hospitals were not randomly allocated to anything. The Finnish Mental Hospitals Study was observational in nature and it is indefensible to consider it a randomized controlled trial. Ramsden, moreover, previously made a convincing argument that all of the excess cardiovascular death that has been attributed to saturated fat in that trial can actually be attributed to the fact that one of the hospitals during one of the time points was liberally administering a cardiotoxic antipsychotic medication.

The 2010 meta-analysis also included a number of multi-factorial trials. The most important one — both because of its size and because of its strong results — was the Oslo Diet-Heart Study. Oslo’s intervention went way beyond replacing saturated fats with vegetable oils: the intervention also restricted margarine, increased the intake of fruits, vegetables, and fish, and distributed free sardines canned in cod liver oil. How can we attribute the results to the vegetable oil when so many other changes were made?

The 2010 meta-analysis, finally, failed to include two landmark trials — the Rose 1965 corn oil trial and the Sydney Diet-Heart Study — that found vegetable oils to increase the risk of heart disease.

The Ramsden paper selected the five trials that were randomized and made only one substitution of replacing saturated fats with vegetable oils. The pooled data suggested heart disease mortality was 13% higher with vegetable oil and total mortality was 7% higher with vegetable oil, but neither of these findings were statistically significant.

When looking at the pooled data, vegetable oil looks like it has barely any effect. When taking a nuanced and detailed view of individual trials, vegetable oils look dangerous.

The Money Quote: The Ramsden Group on Humility

Here is the money quote from the Ramsden paper:

The Big Picture:

The molecules that we eat every day as foods act as substrates, which enter into and regulate numerous highly leveraged biochemical pathways. Thus, although the story of the traditional diet-heart hypothesis did not unfold as predicted, the foods that we eat likely play critical roles in the pathogenesis of many diseases. Given the complexity of biological systems and limitations of our research methods, however, current understanding of the biochemical and clinical effects of foods is rudimentary. The history of the traditional diet-heart hypothesis suggests that nutrition research could be improved by not overemphasizing intermediate biomarkers; cautious interpretation of non-randomized studies; and ensuring timely and complete publication of all randomized controlled trials. Given the limitations of current evidence, the best approach might be one of humility, highlighting limitations of current knowledge and setting a high bar for advising intakes beyond what can be provided by natural diets.

I couldn’t have said it better myself. Traditional diets obtained the bulk of their calories either from carbohydrate or from a mix of saturated and monounsaturated fats with a small amount of polyunsaturated fats. We should only be replacing traditional fats with modern vegetable oils if we have strong and compelling evidence that this will advance our health, and that level of evidence just doesn’t exist.

If you liked this post, also check out its associated podcast episode.