What science really says about saturated fat

Please join me in giving a very warm welcome to Athlegan’s first expert guest writer – Christine Crumbley!

Christine is a postdoctoral associate, weightlifter, and vegan – brain and brawn both. :)

Following up on my article on carbs vs. fat, Dr Crumbley will dive deep and elaborate on the dangers of certain fats, as well as bust some popular Internet myths. Get your notebook out and tag along!

It’s a heavy read – but well worth the time – so if you’re short on time, check out the TL;DR.

Once upon a time, I took a course in animal development. I remember being so fascinated by all of the mechanisms that are involved in development. That sparked my interest in transcription and signaling, and ultimately deepened my interest in the connection between lifestyle exposures and disease.

A particular area of interest is the connection between diet and disease; therefore, I chose to study nuclear receptors.

This protein superfamily has many members that bind to dietary lipids (fats) and regulate many important physiological processes including metabolism. Given that I study lipid-binding proteins, I am always intrigued when new nutrition advice is published in the literature and the lay press.

The lay press tends to poorly report the data, particularly the nutrition advice around fat intake. A popular example of this is the 2010 meta-analysis by Siri-Tarino et al ¹ claiming that saturated fats are not associated with heart disease.

In the same edition of American Journal of Clinical Nutrition, Siri-Tarino et al point the finger at refined carbohydrates for contributing to abnormal blood lipids that promote heart disease ².

Siri-Tarino et al acknowledge in the second, but less popular, paper that data does exist to support replacement of saturated fat with unsaturated fats to reduce heart disease risk. They also mention the size of LDL particles as an area of concern, claiming refined carbohydrate intake is associated with small LDL particles. Together, these papers are frequently cited as “proof” that higher fat diets, particularly high saturated fat diets, are fine for health.

Before we start dissecting this topic, we should review a few points.

What are saturated fats?

Saturated fats have all the bonds are single bonds and every remaining position is occupied by hydrogen, hence the molecules are saturated with hydrogen atoms ³.

For comparison, an unsaturated fat has at least one double bond within the molecule. This means an unsaturated fat has at least one position where a hydrogen atom cannot be attached.

A monounsaturated fat has one double bond, whereas polyunsaturated fats have at least two double bonds. Saturated fats are typically solid at room temperature, eg butter, animal fats, and coconut oil. Unsaturated fats are typically liquid at room temperature, eg oils made from various plants.

How much saturated fat should we eat?

In the US, the NIH carries out surveys regularly to understand the American diet, which is representative of the Western diet in developed countries. The data from 2007-2010 shows that each population group examined had more than 50% of its members consuming too much saturated fat ⁴.

It is worth noting that the restriction on saturated fat intake was not removed in the 2015 dietary guidelines and remains at <10% of calories ⁵.

What are refined carbohydrates?

Refined carbohydrates are processed foods where the fibrous part of the plant is stripped away, leaving mostly simple sugars. Examples include soda, baked goods, sugar, sweetened cereals, etc.

However, Siri-Tarino et al point the finger at refined carbohydrates without acknowledging that many foods containing refined carbohydrates are also high fat, eg donuts, cookies, and similar foods.

Replacement of saturated fat-rich foods, eg meats, with these foods would not likely provide as much health benefit as replacing saturated fat-rich foods with vegetables. Replacement of saturated fats with carbohydrates should include complex carbohydrates, such as whole grains and fibrous vegetables, as the current recommendations are to keep added sugars to <10% of total calories ⁵.

Many studies have shown whole grains, vegetables, and fruits are associated with positive health outcomes ^{6 7 8 9 10}, so plant foods should be used to replace foods high in saturated fats.

What are the data to suggest we should eat less saturated fat?

An excellent rebuttal to Siri-Tarino et al is available in the same edition of the ACJN. Stamler, who has an extensive publication record on the connections of diet and disease, outlines many concerns with the paper and its conclusions, including potential confounders (variables that can skew the conclusions when they are not controlled for during experimental design), the lack of distinction between “total” heart disease events and fatal events, effects of saturated fats on cholesterol, the connections between cholesterol and heart disease, the effects of other macronutrients on various metabolic parameters, and total calorie intake ¹¹.

For potential confounders, Stamler identifies three studies used in the meta-analysis adjust for serum or dietary lipids. If dietary lipids affect serum lipids and serum lipids affect heart disease risk, then adjusting for these variables obscures these relationships and can affect the conclusions drawn from such data.

Many of the studies also rely on self-reported dietary data, which can be unreliable as humans forget or omit various aspects of their food intake, and this can bias the results towards finding no association between the variables of interest (in this case, dietary fat intakes and health outcomes). This is called regression dilution bias or attenuation.

In the criticism by Katan et al, the use of unvalidated methods and 1-day dietary recalls is identified as a major weakness in the Siri-Tarino meta-analysis, and approximately half of the studies included in the meta-analysis used these weak data collection methods ¹².

Katan et al note that randomized controlled trials do show effects of diet on serum cholesterol, but these effects are not found in observational studies using weak methods of dietary assessment. They state, “Such imprecision in the assessment of disease determinants systemically reduces the strength of association of determinants with the disease.”

Over-adjustment is the main point of the criticism by Scarborough et al ¹³. Scarborough et al state that serum cholesterol is part of the casual chain between consuming saturated fat and developing coronary heart disease (CHD) and cardiovascular disease (CVD); therefore, adjusting for serum cholesterol will affect the analysis and conclusions.

They identified seven of the sixteen studies in the CHD meta-analysis and four of the eight studies in the stroke analysis as being adjusted for serum cholesterol, and these studies represent “nearly half of all CHD and CVD events” used in the Siri-Tarino et al meta-analysis. They suggest that using non-adjusted data would provide “different and more informative” results.

Siri-Tarino et al did not mention this as a limitation of their study.

Similarly, a criticism on the McDougall website points that ~40% of the weight of the Siri-Tarino study is derived from six adjusted studies ¹⁴.

Stamler re-analyzed the data provided in the meta-analysis and his results contrasted those of Siri-Tarino et al. Stamler’s relative risks are 1.32 for studies listing fatal events, 0.99 for studies listing total events, and 1.09 for all of the studies combined. A value of 1 is neutral and a value above 1 indicates increased risk. Stamler’s relative risk when all studies were examined was slightly higher than Siri-Tarino et al’s relative risk of 1.07. The risk for fatal events is striking, although the overall risk of fatal and non-fatal events is small.

When the six studies identified in the McDougall criticism were removed from the dataset, the remaining studies showed a statistically significant 30% increased risk of death from coronary heart disease.

When only the six adjusted studies were examined, there was still a statistically significant 18% increased risk of death. (The McDougall criticism also discusses a meta-analysis about saturated fat by Chowdhury et al from 2014 ¹⁵, which has limitations similar to the Siri-Tarino meta-analysis covered here.)

Stamler commented that cholesterol was largely ignored. The effects of saturated fats on cholesterol levels were not addressed, as key papers in this research area were not cited.

Keys et al and Hegsted et al developed equations to describe this relationship; they showed a direct relationship between saturated fat intake and cholesterol, whereas an inverse relationship was observed between polyunsaturated fats and cholesterol.

Our understanding of the role of dietary cholesterol in heart disease is derived from experiments showing that cholesterol intake is the major dietary change needed to induce atherogenesis. This has been observed consistently since the early 1900’s in a variety of animal species, including non-human primates.

To add to this point about cholesterol, an older review from Hopkins examines the relationship between baseline cholesterol intake and additional cholesterol intake during experimental studies ¹⁶. (Hopkins’ research focuses on the relationships between genetic variants and cholesterol-related disorders.)

The Hopkins review mentions the long-standing observation that increasing dietary cholesterol has diminishing effects, meaning the relationship does not appear to be linear between dietary and serum cholesterol levels.

Hopkins re-analyzed the cholesterol feeding studies in humans available at that time and incorporated the baseline dietary cholesterol intake, as well as other variables. In his analysis, 76 studies with completely controlled diets were used for a non-linear regression, and the studies were weighted by the sample size.

• When baseline dietary cholesterol is very low (like it would be for vegans), the response to increasing dietary cholesterol is a very drastic change in serum cholesterol.

• When baseline dietary cholesterol is increased, the changes in serum cholesterol are smaller.

The curves have a hyperbolic shape, meaning they rise and then plateau, and it is that plateau where eating more cholesterol does not affect serum cholesterol.

• For people who consume nearly zero cholesterol, consumption of 300 mg cholesterol (the previous recommended upper limit) raises serum cholesterol by 15 mg/dL. The increase in serum cholesterol becomes more dramatic at higher intake levels, eg ~35 mg/dL change when 1000 mg cholesterol is added to the diet.

• For people already consuming 300 mg cholesterol per day, the addition of 1000 mg cholesterol increases serum cholesterol levels by ~10 mg/dL.

This may sound small, but is worth considering for someone with borderline cholesterol levels or a family history of heart disease.

For reference, 300 mg cholesterol is found in the following portion sizes of various foods: 4 pieces of fried chicken, 1.2 Sausage Egg McMuffin sandwiches from McDonalds, or 1.6 Double Whopper with cheese sandwiches from Burger King.

For the argument that these are fast foods, 2 chicken breasts (~350 g), 2 “medium” eggs (~80 g), 16 oz steak, 3.5 oz medium shrimp (~100 g of the 41-50 count size), and 1 lobster tail also contain 300 mg cholesterol ¹⁷.

Other reviews show dietary cholesterol raise LDL (“bad”) cholesterol more than HDL (“good”) cholesterol ¹⁸.

Bloggers have commented that this is not a concern, because only large, “fluffy” LDL increases on a high-fat diet. While it is possible that small, dense LDL and oxidized LDL are potentially reasons for concern in their own right ^{19 20 21}, this does not mean increases in total LDL should be ignored.

Brown and Goldstein published an excellent review about the relationship of LDL cholesterol and heart disease ²². Studies examining diet, genetic variants, and drugs targeting various aspects of cholesterol biosynthesis have concluded the same thing:

“LDL lowering can prevent heart attacks, especially when the lowering is begun before atherosclerotic plaques have developed.”

The LDL levels in adult animals from a variety of species, including primates, are lower than 60 mg/dL, and newborn humans have levels <50 mg/dL. Brown and Goldstein note that LDL >100 mg/dL are “frequent only in humans and other animals that have consumed the typical Western diet, which is high in fat and cholesterol.” (Brown and Goldstein won the Nobel Prize in Physiology or Medicine in 1985 for their work on cholesterol metabolism, and they still work in this area in their lab at UT-Southwestern in Dallas, TX.)

How should we change our diets to reduce saturated fat intakes?

Siri-Tarino et al express concern about carbohydrates affecting metabolic parameters if people use carbohydrates to replace saturated fats in both papers published in the March 2010 edition of ACJN ^{1 2}.

The meta-analysis mentions more data is needed before making recommendations of which macronutrients to consume in place of saturated fats. Their second paper claims replacing saturated fat with “a higher carbohydrate intake, particularly refined carbohydrate, can exacerbate the atherogenic dyslipidemia associated with insulin resistance and obesity that includes increased triglycerides, small LDL particles, and reduced HDL cholesterol…

There are few epidemiological or clinical trial data to support a benefit of replacing saturated fat with carbohydrate” ².

Their claim is that consuming carbohydrates, particularly refined carbohydrates, will worsen blood lipid profiles and promote metabolic diseases (heart disease, type 2 diabetes, and obesity). Stamler comments that international epidemiological data do not support the carbohydrate-induced dyslipidemia hypothesis.

Data from various countries with high carbohydrate intake, mainly from complex carbohydrates (foods rich in fiber or starches), do not show evidence of metabolic syndrome, such as lower HDL, high total cholesterol or LDL cholesterol, or high triglycerides.

Dietary interventions such as the DASH/Omniheart trials demonstrate benefit from consuming complex carbohydrates as the replacement food for saturated fats.

Stamler identifies a calorie surplus that promotes overweight/obesity as the “main driver” of diet-induced changes in lipid profiles and diabetes. Weight loss, even modest weight loss, using various diets with different nutrient composition improves these conditions.

Given the emphasis on refined carbohydrates by Siri-Tarino and the positive data on complex carbohydrate intake, it would be wise to limit your candy and cookie intake and instead focus on whole grains, starches, vegetables, and fruits.

Stamler concludes with the similarities between various heart-healthy diets, including DASH, OmniHeart, Mediterranean, and East Asian dietary patterns.

”[These diets] emphasize vegetables, fruit, whole grains, legumes/seeds/nuts, fat-free/low- fat dairy products, fish/shellfish, lean poultry, egg whites, seed oils in moderation, alcohol (if desired) in moderation, and portion size/calorie controlled and deemphasize red and processed meats, cheeses, ice cream, egg yolks, cookies/pastries/pies/cakes/other sweets/sweetened beverages, snacks, and salt/commercial foods with added salt. Estimated nutrient composition of this fare is as follows: total fat ~20-25% of kcal, SFA 6-7%, MUFA 7-9%, PUFA 7-9%, cholesterol <100 mg/1000 kcal, total protein 18-25%, vegetable protein 9-12%, carbohydrate 55-60% (mostly complex), fiber 30-35 g/d, 50-65 mmol Na/d (2900-3770 mg NaCl), mineral/vitamin intake high.”

Katan et al mention that the replacement of saturated fat with polyunsaturated fat (PUFA) has been shown to lower LDL cholesterol, lower the LDL to HDL cholesterol ratio, and reduce risk of CHD in prospective cohort studies and randomized trials. Katan et al assert, “The benefit of replacing saturated by polyunsaturated fat has been proven beyond reasonable doubt.”

Siri-Tarino et al fail to identify this in their meta-analysis, just as the association between saturated fat and CHD is not found. Siri-Tarino attempted to estimate the effect of this replacement by selecting five studies that listed intakes of carbohydrate, protein, and fats, but these studies did not list polyunsaturated fat intake.

Katan et al state, “It requires a leap of faith to assume that outcome of such a calculation truly represents what happens when saturated fat is replaced by polyunsaturated fat.”

Other reviews and meta-analyses on replacing saturated fat with polyunsaturated fats demonstrate reductions in heart disease risk (see below) and in mortality ²³. For example, several Cochrane Reviews have focused on dietary advice to reduce heart disease risk and on saturated fat intakes.

Cochrane Reviews are very comprehensive and use standardized tools for analysis. They include lay summaries for the public and provide direction to public health organizations and physicians.

A 2012 Cochrane Review suggested that at least two years of dietary fat modification reduces risk of cardiovascular events, notably for people with heart disease risk factors or established heart disease ²⁴. This dietary fat modification involves “replacing some saturated (animal) fats with plant oils and unsaturated spreads.”

The review notes no clear health benefit to reducing fat intake and increasing starchy food consumption. In reading the text of this review, I could not find a definition for starchy foods and there were several places where the power of various analyses was noted as “limited.”

A 2013 Cochrane Review noted that receiving any dietary advice increased fruit and vegetable consumption by 1.18 servings per day, increased fiber by 6.5 g/day, decreased fat intake by 4.48%, and decreased saturated fat intake by 2.39% ²⁵. The authors concluded dietary advice has an “effect in bringing about modest beneficial changes in diet and cardiovascular risk factors” when interventions lasted for a year, but long-term outcomes are unknown.

These changes are extremely modest, and I suspect that is why the health outcomes are so subtle.

Looking at the nutrition information for various fruits and vegetables, a fiber intake of 6.5 g corresponds to a half-cup of cooked chickpeas, a cup of cooked leafy greens, a cup of berries, a cup of cooked quinoa, or a cup of cooked wheat pasta ²⁶. The CDC reviewed data from 2013 and found half of American adults ate <1 cup fruit and <1.5 cups vegetables daily ²⁷, therefore adding 1 serving through dietary counseling in studies is still far below the recommended intakes (at least 5 servings of produce per day).

A 2015 Cochrane review found no clear effects on cardiovascular or all-cause mortality with reduced saturated fat intake, but there was a 17% decrease in combined cardiovascular events (eg, heart attacks, strokes) ²⁸.

Breaking the data into smaller groups (subgroups) for analysis found that replacing saturated fats with polyunsaturated fats resulted in a 27% decrease in cardiovascular events. Several subgroups had more obvious reductions in cardiovascular events: studies with higher baseline intakes of saturated fat, studies where saturated fat intake was reduced much more in the intervention groups compared to control groups, and studies with greater decreases in total cholesterol levels.

Studies that reduced saturated fat, regardless of replacement nutrient (carbohydrate, other fats, or protein), suggested decreases in total and LDL cholesterol, but no effects were observed on HDL cholesterol, triglycerides, or ratios between these items (in the few studies that reported ratios). This addresses the concerns of Siri-Tarino et al that replacement with carbohydrates would produce negative changes in HDL cholesterol and triglycerides.

The Cochrane review notes that behavior-based interventions, such as changing one’s diet, can have widely variable rates of compliance. To address this, “subgrouping and meta-regression suggested that greater reductions in saturated fats were associated with greater reductions in the risk of cardiovascular disease events. This suggestion of a dose response strengthens our belief that there is a true effect of reducing saturated fat on CVD events.”

The Cochrane review addresses the Siri-Tarino meta-analysis by citing the criticisms already shared here and by using the available data on total and LDL cholesterol in the studies included in the Cochrane review. The authors confirm the link between serum lipids and heart disease by stating:

“Despite limited data there was a clear suggestion from meta-regression that there was a greater reduction of risk of cardiovascular events in studies with greater total serum cholesterol reduction, supporting the central role of serum lipids in the link between dietary saturated fats and cardiovascular events.”

This refutes the Siri-Tarino meta-analysis and its conclusions.

The Cochrane review estimates ~5% reduction in deaths from cardiovascular disease and ~3% reduction in all deaths when dietary saturated fats are reduced, although confidence intervals are wide.

The conclusions about carbohydrates in the Cochrane reviews, particularly the 2012 review, are rather curious. A 2004 paper in the Archives of Internal Medicine found a 27% reduction in coronary mortality risk as fiber intake increased in 10 g increments ²⁹. The authors noted that cereal and fruit fiber had an inverse relationship with risk, but associations could not be found with vegetable fiber.

The >27 g/day fiber intake, energy-adjusted, group was listed as the high fiber group. By the current dietary guidelines, this is a high fiber diet.

A vegan diet full of whole plant foods easily provides much more fiber than this, and there appears to be no upper limit for fiber from a safety perspective. Too much fiber may cause some GI distress, but this is not considered dangerous to health.

A similar outcome of reduced heart disease risk with higher whole grain consumption was found in a 2008 paper ³⁰, as well as reduced overall mortality with whole grain consumption in a 2015 paper ³¹. In addition to examining unsaturated fats, Li et al found a 9% lower risk of CHD when whole grains were used to replace saturated fats, specifically 5% of energy intake from saturated fats ³².

Interestingly, Siri-Tarino and coauthors have published other papers suggesting that polyunsaturated fats and certain types of carbohydrates should be used to replace saturated fat in the diet.

In the same edition as the controversial meta-analysis, Siri-Tarino et al also acknowledge the success of the DASH diet and the Lyon diet heart study (a Mediterranean style diet that also included alpha-linolenic acid, an omega-3 fat found in nuts and seeds) and the potential of saturated fats to promote inflammation, affect vascular function, and to increase LDL, particularly in genetically sensitive individuals ². Later in 2010, minimally processed grains were added to the list of acceptable substitutions ³³.

When mentioning carbohydrates, they again emphasize refined carbohydrates and added sugars. Siri-Tarino et al recommend consumption of unsaturated fats, but not refined carbohydrates, and also mention dietary cholesterol as a nutrient of concern because cholesterol down-regulates LDL Receptor activity (therefore raising LDL levels).

It is difficult in practice to separate dietary cholesterol from saturated fat, as they occur together in animal foods.

Part of Siri-Tarino’s concerns about refined carbohydrate consumption and dyslipidemia may be attributable to an excess of calories and not the refined carbohydrate foods per se, as Stamler called calorie surplus the “main driver” of metabolic syndrome.

In 2011, Siri-Tarino notes added sugars, but not high glycemic carbohydrates, decrease HDL cholesterol ³⁴. In researching the glycemic index, moderate glycemic index (GI value 56-69) carbohydrates include potatoes, white rice, and ripe sweet fruits (eg, cantaloupe, watermelon, and pineapple) ³⁵. I assume this is what Siri-Tarino means by high glycemic carbs, instead of the processed sugar-rich or baked goods typically associated with high GI values (>70).

Complex carbohydrates, starches, and the Mediterranean diet that emphasizes whole grains were acknowledged for their beneficial effects on CHD risk.

A 2015 review makes similar statements, as well as expanding on the potential of low-fat or fermented dairy foods and vegetarian eating patterns to improve CVD risk and mortality ³⁶.

A 2016 publication from Siri-Tarino and coauthors agrees with the 2015 Cochrane review that having higher starting LDL cholesterol levels will produce a greater response when saturated fats are restricted from the diet and polyunsaturated fats beneficially modulate cholesterol levels ³⁷.

Consumption of added sugars, particularly sugar-sweetened beverages, at 10-25% of calorie intake or >100 g fructose per day were specifically named as concerning carbohydrate sources in the diet.

Although obesity gets only a brief mention in many of Siri-Tarino’s publications, it is worth noting that the atherogenic dyslipidemia mentioned in the meta-analysis can be resolved by weight loss. Siri-Tarino has investigated this, and shown that small, dense LDL is reduced in moderately overweight men when they lost weight ³⁸. There might be some genetic variability involved in how bodies process different macronutrients (namely, fats and carbohydrates) for energy and produce small, dense LDL particles ^{39 40}.

Maintaining a healthy body weight and body composition will influence many aspects of health and daily quality of life.

It would appear over the years since the controversial meta-analysis, Siri-Tarino and coauthors have shifted focus towards mostly whole food carbohydrate sources as appropriate substitutions for saturated fat, instead of implying all carbohydrates should be limited.

The TL, DR Version

Internet bloggers love the saturated fat meta-analysis done by Siri-Tarino et al ^{1 2} that claims higher saturated fat intakes do not affect heart disease. This meta-analysis is frequently used to support high fat, low carb diets.

Several criticisms of this meta-analysis have been published ^{11 12 13 14}.

Large systematic reviews and meta-analyses, such as those done by the Cochrane Review team, have found positive heart-health benefits to reducing saturated fat intake and increasing unsaturated fat intakes ^{23 24 25 26 27 28}.

Siri-Tarino et al have focused their concern on the replacement of saturated fat with processed carbohydrates and added sugars, instead of unsaturated fats or whole food sources of carbohydrates.

Later work by Siri-Tarino and co-authors suggests that minimally processed grains can be used to replace saturated fat ^{36 37}, as studies using heart-healthy diets support carbohydrate intakes from whole grains, starches, fruits, and vegetables ^{6 7 8 9 29 30 31 32}.

Tell your story

Do you have something you’d want to share with your fellow vegan athletes? We’re always scouting for good stuff and we’d love to hear what you have to say!

Send an email to [email protected] and let’s talk!