The Great American Diet Experiment
Shaping the U.S. Food Supply
“If you don’t know where you are going, you might wind up someplace else.” — Yogi Berra
The U.S. dietary guidelines shape America’s food supply in three ways:
- They establish standards of care for medical guidance.
- They influence consumer demand.
- They encourage manipulation of food products.
Regardless of how many people actively follow the guidelines, the guidelines shape food production for the American population.
The food produced is the food available and the food available plays a significant role in determining the health of the nation.
Observations on the Guidelines
There are two documents which provided the foundation for the 2015 U.S. dietary guidelines:
- The 2013 Report from the American College of Cardiology (ACC) and American Heart Association (AHA) Lifestyle Work Group.
- The 2015 Report from the Dietary Guidelines Advisory Committee (DGAC).
The 2015 dietary guidelines continue to support two food supply shaping goals which have been in place for the past thirty-five years:
- Increase the amount of grain consumed by the American population.
- Change the fat profile consumed by the American population by encouraging the replacement of saturated fat with polyunsaturated fat.
The primary driver behind both goals is an attempt to reduce the risk of cardiovascular disease in the United States. Ironically, the dietary guidelines have had the opposite effect.
Both shaping goals increase the amount of linoleic acid, an omega-6 polyunsaturated fat, in the food supply.
Food shaping efforts have been effective. Quality saturated fat is hard to find, omega-6 polyunsaturated fat is hard to avoid.
The unintended consequences of additional consumption of linoleic acid has been increased rates of cardiovascular disease, liver disease and obesity.
The dietary guidelines were reverse engineered by specialists in the study of disease.
Three factors caused the 2013 ACC/AHA Lifestyle Work Group and the 2015 DGAC to overestimate the benefits of their recommendations and significantly underestimate the risk:
- An outdated model of cardiovascular disease.
- Terminology which has lost it’s meaning.
- Tools which are being stretched beyond their capabilities.
The Nutrition Scientists in both the 2013 ACC / AHA Lifestyle Work Group and the 2015 DGAC are intelligent people from some of our most prestigious institutions. Given their positions and backgrounds, a review of the two reports provides two surprising observations:
- The U.S. Government Dietary Guidelines are based on a remarkably narrow viewpoint.
- The U.S. Government Dietary Guidelines are based on a remarkably thin slice of evidence.
The narrow viewpoint and thin slice of evidence are used to justify a dietary intervention for the U.S. population (increased grain consumption and an alternative fat profile).
The intervention is being prescribed based on a predicted rather than measured benefit.
The risk associated with the intervention is not reported.
An Outdated Model
The model of cardiovascular disease which the 2013 Lifestyle Work Group and the 2015 DGAC followed is over forty years old and based on an assumption rather than fact.
The assumption can best be summarized by a statement made by Professor Alice Lichtenstein in an October 2014 letter to the New York Times:
“The conclusions of an extensive review, released by the American College of Cardiology and the American Heart Association last fall, found strong evidence for a link between saturated fat (butter, cheese, fatty meat) and heart disease.”
Professor Lichenstein was a member of the 2013 Lifestyle Workgroup and the Vice Chair for the 2015 DGAC. Her comment is interesting because the Work Group did not find a link, it started with the assumption a link existed.
The 2013 ACC / AHA Lifestyle Workgroup did a systematic review of clinical trials and statistical analysis of clinical trials which linked changes in diet with changes in blood lipids (fat and cholesterol in the blood stream).
The Lifestyle Work Group Report which Professor Lichtenstein references does not provide any evidence of a link between saturated fat and cardiovascular disease (CVD). The preamble of the report lists the workgroup’s focus:
“The Work Group focused on CVD [cardiovascular disease] risk factors to provide a free-standing Lifestyle document and to inform the Blood Cholesterol guideline and the hypertension panel. It also recognized that RCTs [Randomized Controlled Trials] examining the effects on hard outcomes (myocardial infarction, stroke, heart failure, and CVD related death) are difficult if not impossible to conduct for a number of reasons (e.g., long-term adherence to dietary changes).”
During the review, blood cholesterol was used as a surrogate endpoint for cardiovascular disease. A substitute endpoint was used because cardiovascular disease is a chronic condition which develops over years or decades, much longer than any controlled clinical trial.
Because links between cause and effect for food are so difficult to determine, the Work Group focused on links between diet and blood lipids.
The 2013 Lifestyle Workgroup operated on an assumption that lowering LDL-C (total low density lipoproteins) is the same as lowering the risk of cardiovascular disease.
LDL-C is a poor surrogate for cardiovascular disease.
According to a 2010 study on biomarkers and surrogate endpoints conducted by the Institute of Medicine:
“Because cardiovascular disease is a multifactorial chronic disease, a single risk factor for the disease (e.g., LDL-C) cannot fully account for all the variability that leads to a particular outcome” (pg. 167)
The Work Group operated with an oversimplified model of cardiovascular disease. The oversimplification created a misperception of both the cause of cardiovascular disease and the cure.
The 2013 Lifestyle Work Group accepted the link between saturated fat and cardiovascular disease as fact and didn’t look for evidence supporting it.
The 2015 DGAC looked for evidence supporting a link between saturated fat and cardiovascular disease and didn’t find any. The 2015 DGAC found no studies or trials which linked saturated fat consumption with cardiovascular disease.
“recent meta-analyses of prospective observational studies did not find a significant association between higher saturated fat intake and risk of CVD in large populations.” (DGAC Report, pg. 451).
The Credit Suisse Research Institute also released a 2015 report on dietary fat. The Credit Suisse Research team also searched for studies linking saturated fat and cardiovascular disease and found:
“One of the biggest myths in nutrition is that saturated fat intake above a certain level — say 10% based on most dietary guidelines — significantly increases your risk of heart attack. This conclusion that has held for almost half a century is inconsistent with the wealth of epidemiological data or scientific evidence in the form of clinical randomized trials. Plenty of research funding has been earmarked to study and back this hypothesis, yet we cannot find a single research paper written in the last ten years that supports this conclusion. On the contrary, we can find at least 20 studies that dismiss this hypothesis.” (pg. 34)
In the past few months prominent Nutrition Scientists from Harvard (source), Tufts (source) and the Academy of Dieticians (source) have all backed away from attacks on saturated fat. They have instead shifted their attention to glycemic index and promoting polyunsaturated fats.
Terminology. Words Have Meaning.
An attempt to simply nutrition has created a significant amount of confusion regarding fat and cholesterol.
Chemically complex lipoproteins have been neatly divided into “good” cholesterol (HDL) and “bad” cholesterol (LDL), despite the fact neither LDL or HDL are cholesterol and both are critical to health and well being.
Because saturated fat was incorrectly believed to cause cardiovascular disease it was labled “bad”.
Because polyunsaturated fat was “essential” and it was not saturated, Nutrition Scientists labeled it “good”.
Biochemists have a dramatically different view of polyunsaturated fatty acids than the ACC/AHA Lifestyle Work Group and the DGAC, especially when it comes to linoleic acid.
Saturation is a measure of chemical stability in the presence of oxygen.
Saturated fat is stable in the presence of oxygen. Monounsaturated fat is less stable than saturated fat. Polyunsaturated fat is less stable than monounsaturated fat.
Replacing saturated fat with polyunsaturated fat replaces a fuel which is chemically stable with a fuel that is chemically unstable.
Research performed by biochemists specializing in the study of lipid metabolism (how the body processes fat) indicate a very low biological tolerance level for linoleic acid (3% of calories). (Source)(Source).
“Essential” is different than “good” and certainly does not mean “more is better.”
The words used to describe fat in Nutrition Science have largely lost their original meanings.
Stretching the Tools
Nutrition Science is dominated by epidemiology (the study of the distribution and control of disease). Epidemiologists rely heavily on mathematical modeling.
There are three problems with using mathematical modeling to determine dietary interventions:
Relevance. The models show association strength between input and output variables but provide no information on whether the associations are relevant.
Accuracy. The models are only as accurate as the measurements plugged into the computers.
Complexity. Biological systems are complex and non-linear.
Determining relevance is a significant challenge for statistical analysis. Associations identified by statistical analysis can have a confidence level of 95% or higher without having any relevance to the problem being investigated.
As a simple example, in Florida, increases in the number of shark bites each year are strongly associated with an increase in ice cream consumption, but the association is irrelevant. Both shark bites and ice cream consumption increase in warmer months but eating less ice cream doesn’t offer any protection from shark bites.
Fivethirtyeight is a company with a proven track record of using statistical analysis to separate signals from noise. In a well written article they provided examples of how easy it is to establish irrelevant links between variables when talking about food.
Nutrition Science borrows heavily from the terms and methods used in Evidence Based Medicine.
In Evidence Based Medicine, the lowest level of the evidence pyramid is expert opinion. From expert opinion, clinical trials are designed to determine risk and benefit. Meta-Analyses, statistical analysis of vast pools of combined data from multiple trials, sit at the top of the evidence ranking pyramid.
For surgery and pharmaceuticals, measurements of the input variables in clinical trials are simple. A patient took the medicine or they didn’t. A patient had the surgery or they didn’t.
Fat consumption is much harder to measure than prescription drugs or surgery. Fat is a complex chemical compound which has been oversimplified in many studies. “Saturated fat” in many studies could be mean cake or could mean steak. Even if steak and cake are measured separately, what the cow ate before it became steak will influence the proportions of saturated, monounsaturated and polyunsaturated fat the steak contains and the study participants consume.
Adding to the difficulty of linking food and health in clinical feeding trials is the open ended reality of what’s being observed.
For trials on procedures like surgery or interventions like prescription pharmaceuticals, the researchers have clear start and stop points.
For feeding trials, the study subjects were eating before they started the trial and will continue to eat after the trial.
For a complex chronic disease like cardiovascular disease, the majority of the impact is happening at scale (molecular) and time periods (decades) which are exceedingly difficult to measure. Researchers rely heavily on rough approximations and discreet snapshots.
With trials of medical procedures and prescription medicine, pooling the results of multiple trials helps to weed out inaccuracies.
With nutritional studies where the input variables are rough approximations, pooling results amplifies error.
The results of a meta-analysis of food trials can vary dramatically based on which trials the authors of a study choose to include in their pool. The results of meta-analyses of feeding trials depend more on the choice of data than anything else.
Nutrition Science effectively flips the evidence ranking pyramid used in Evidence Based Medicine upside down.
Two meta-analyses which pool the results of multiple feeding trials are enough to award a “strong and consistent evidence” classification, even though the conclusions of the studies are based more on the opinions of the researchers and their selection of data than biological facts.
Replacing nutrient A with nutrient B is different than just consuming less of nutrient A and more of nutrient B.
The DGAC report makes a claim that:
“For every 1 percent of energy intake from SFA replaced with PUFA, incidence of CHD is reduced by 2 to 3 percent.” (DGAC, pg.452)
The claim stretches the tools being used by Nutrition Scientists far beyond what they are capable of measuring.
The DGAC claim is a prediction based on statistical analysis, it is not a dietary intervention which has been tested and measured.
The DGAC claim implies a direct cause and effect, linear relationship between a shift in fat profile and risk of cardiovascular disease.
Changes in nutrient levels are not linear. Each nutrient has two different thresholds. There is a threshold between not enough and enough and another threshold between enough and too much.
A 5% increase in linoleic acid intake from 1% of calories to 6% of calories will have a much different effect than an 5% increase in linoleic acid from 10% of calories to 15% of calories.
Remarkably Narrow Viewpoint
“For, quarreling, each to his view they cling. Such folk see only one side of a thing.” (Source)
Nutrition Scientists from the Lifestyle Work Group and DGAC discount the warnings of biochemists because they have not been able to measure lipid perioxidation (the degenerative impact of oxygen on polyunsaturated fats) during feeding trials on humans.
Nutrition Scientists haven’t seen signs of lipid perioxidation in studies, so their assumption is it isn’t something to be concerned with.
The same can be said about information from many other sources outside of the small circle of Nutrition Scientists shaping U.S. dietary guidelines. If it isn’t observed during feeding trials, it isn’t factored into the discussion, even if it is a possible health hazard.
The Credit Suisse researchers took a much broader look at the information available on dietary fat, cardiovascular disease and obesity and came away with a dramatically different picture of fat than the Lifestyle Workgroup and the DGAC.
The Credit Suisse Researchers reached this conclusion on saturated fat:
“The conclusion of this report is simple. Natural unprocessed fats are healthy and key to the evolution of a society that focuses on developing healthy individuals, not just on treating those who are sick. Natural foods high in monounsaturated and saturated fats are one of the preferred sources of energy for our bodies to use and store.” (CSRI, pg. 3)
And this conclusion on omega-6 fats (linoleic acid):
“A proper review of the so called “fat paradoxes” (France, Israel and Japan) suggests that saturated fats are actually healthy and omega-6 fats, at current levels of consumption in the developed world, are not necessarily so.” (CSRI, pg.4)
Remarkably Thin Slice of Evidence
“The Committee encourages the food industry to continue to reformulate and make changes to improve the nutrition profile of certain foods. Examples of such actions include lowering sodium and added sugars content, achieving better saturated fat to polyunsaturated fat ratio… The Committee also encourages the food industry to market these improved products to consumers.” (DGAC, Part B, Chapter 2, page 9).
The manipulation of the food supply recommended by the 2015 DGAC is based on an assumption, a predicted benefit and a remarkably thin slice of evidence.
The evidence cited by the 2015 DGAC as rationale for replacing saturated fat with polyunsaturated fat consisted of three trials (DASH, DASH-S and DELTA) and five studies (Mensink et al., Mensink et al., Mozaffarian et al., Farvid et al. and Jakobsen et al.).
In two of the three trials, the effect of changes in saturated fat could not be isolated:
“Of note, in the DASH trials, the effect of saturated fat on LDL-C could not be isolated because macronutrients and other nutrients such as dietary cholesterol were not held constant.” (ACC/AHA Full Report, pg.30)
Two of the five studies the DGAC cited were also cited by the 2013 Lifesyle Work Group. Both studies were conducted by the same researchers and were predictions, based on mathematical models, which estimated changes in LDL from changes in dietary fat:
“We used two meta-analyses from the same authors published 11 years apart in which they used the same inclusion/exclusion criteria and generated predictive equations to estimate changes in plasma lipids when substituting dietary fat types with carbohydrates or other fat types.” (ACC/AHA Full Report, pg. 31)
The three studies the DGAC added to the report from the 2013 Lifestyle Work Group all started with an incorrect assumption (saturated fat causes cardiovascular disease) and were focused on identifying a replacement for saturated fat:
“Reduced saturated fat (SFA) consumption is recommended to reduce coronary heart disease (CHD), but there is an absence of strong supporting evidence from randomized controlled trials (RCTs) of clinical CHD events and few guidelines focus on any specific replacement nutrient.” (Mozaffarian et al.)
“Greater intakes of trans-fat and saturated fat compared with polyunsaturated fatty acids (PUFAs) are associated with increased risk of CHD. Therefore, substitution of PUFAs for saturated fatty acids (SFAs) has been recommended to reduce CHD risk.” (Farvid et al.)
“Saturated fatty acid (SFA) intake increases plasma LDL-cholesterol concentrations; therefore, intake should be reduced to prevent coronary heart disease (CHD). Lower habitual intakes of SFAs, however, require substitution of other macronutrients to maintain energy balance.” (Jakobsen et al.)
“Nutrition science, which after all only got started less than two hundred years ago, is today approximately where surgery was in the year 1650 — very promising, and very interesting to watch, but are you ready to let them operate on you?” — Michael Pollan
“Doubt is not a pleasant condition, but certainty is absurd” — Voltaire
“We thought we nailed it.” — Babara Millen, Chair, DGAC (Source)
Both the 2013 ACC / AHA Lifestyle Work Group and the 2015 DGAC were focused on preventing cardiovascular disease.
Unfortunately both groups started from an incorrect assumption: saturated fat causes cardiovascular disease.
From their start point, the goal of both groups was to find a replacement for saturated fat.
Both the 2013 ACC / AHA Lifestyle Work Group and the 2015 prescribe a dietary intervention to replace natural sources of saturated fats with polyunsaturated fats, primarily vegetable oils at concentrations which do not occur in nature.
The unintended consequences of the 2015 dietary guidelines will be increased rates of cardiovascular disease, liver disease and obesity.
From the viewpoint of both the 2013 Lifestyle Work Group and the 2015 DGAC there is risk associated with consuming saturated fat and a limit to the amount of saturated fat which should be consumed (5%-6% of calories in the ACC/AHA report, 10% of calories in the DGAC Report).
There is no limit provided in either report for polyunsaturated fat and no mention of risk for overconsumption of polyunsaturated fat.
Every nutrient has an upper limit and risk associated with excess consumption.
Both the 2013 ACC/AHA Lifestyle Workgroup and the 2015 DGAC take an academic approach, seeking to defend a hypothesis, rather than a scientific approach which seeks to identify inconsistencies in the available information and find the best answer.
The one sided approach which paints the picture of a simple solution with no risk is dangerous.
There are two sides to the nutrition debate. Nutrition Science is on one side, biochemistry is on the other.
The American experiment of replacing saturated fat with polyunsaturated fat is not going well.
Median saturated fat consumption is within 1% of the DGAC target but none of the benefits predicted by Nutrition Science are being realized. (DGAC, Part D, Ch. 6, pg. 11 ).
At an estimated mean intake of 6.3% of calories (DGAC App E2.2), linoleic acid (PFA 18:2) consumption is twice the safe upper limit recommended by lipid metabolism researchers (source) (and given the difficulty of measuring actual consumption of linoleic acid, the estimates are most likely low). The impacts to health of excess linoleic acid predicted by biochemistry are being felt by the American population.
The news is not all bad. There is much to be optimistic about. The United States has a wealth of untapped nutrition expertise. Americans are learning more about nutrition everyday.
Nutrition Science will continue to improve and has a role in the development of dietary guidelines but it can not be allowed to dominate the guidelines while maintaining a scientifically curious mix of hostility and contempt for dissenting viewpoints.
We can improve the health of the American population by shaping the food supply.
The next step is to improve the quality of the discussion on nutrition by extending it well beyond the narrow confines of Nutrition Science.
A broader look at linoleic acid and it’s impact on cardiovascular health, liver health and obesity is a good place to start.