The Carnivore Code: Unlocking the Secrets to Optimal Health

Overall score

Scientific accuracy

Reference accuracy

Healthfulness

How hard would it be to apply the book's advice? Very difficult

Book published in 2020

Published by Mariner Books

First Edition, Kindle

Review posted February 5, 2022

Primary reviewer: Stephan Guyenet

Peer reviewer: Mario Kratz

Table of contents

1. Introduction
2. Scientific accuracy
3. Reference accuracy
4. Healthfulness
5. Most unusual claim
6. Other
7. Conclusion
8. Updates

Claim 1

Humans evolved as carnivores, only occasionally eating plants as “fallback foods” to avoid starvation.

Supporting quote(s) and page number(s)

Page 10: “At this point in the book, I want to share with you my Carnivore Code Hypothesis: I believe that throughout our evolution, our ancestors have hunted animals preferentially, and have eaten plant foods only during times of scarcity or starvation.” (original emphasis)

Page 10: “If we couldn’t find animal foods, we might have eaten plants as a backup plan, but they do not appear to have composed any significant part of our ancestors’ diets.”

Page 151-2: “A carnivore diet is written into our book of life as the most fundamental diet that humans can thrive on. Plants have always been “fallback” foods and can be quite triggering for some people…”

Criterion 1.1. How well is the claim supported by current evidence?

1 out of 4

This claim received a score of 1, meaning that it’s poorly supported by current evidence. It’s mostly unknown how much plant vs. animal food was eaten by our distant ancestors over the last few million years, and the limited information that’s available suggests that most diets probably included plant foods. Studies of current and historical hunter-gatherer cultures suggest that they eat a variety of plant foods, usually as a major part of the diet, and that strict carnivory is uncommon.

The first clear evidence of animal food consumption by our distant ancestors is bones with cut marks from tools that date to around 2.5 million years ago, possibly left by Homo habilis. Homo erectus emerged about 2 million years ago, spread to many parts of the world, and survived until about 100,000 years ago. Animal bones at archeological sites reveal that Homo erectus was a successful hunter. Homo erectus was survived by Homo neanderthalensis (Neanderthals) and Homo sapiens, both also successful hunters.

Yet, since plant foods don’t preserve as well as animal bones, this doesn’t tell us how much animal food they ate relative to plant food. Scratch patterns on ancient teeth provide a bit more information. A 2012 review of dental evidence on the diet of early Homo states: “The most obvious conclusion we can draw from a review of the fossil evidence for diet in African early Homo is that there is not much of it, and what we do have is not very compelling.” The paper reviews evidence suggesting that early Homo probably ate a diverse diet, i.e. both plant and animal foods. Surface erosion of the teeth of Homo habilis suggests they may have regularly eaten acidic fruit.

A 2018 article by a leading researcher on ancient diets states that Homo habilis and Homo erectus both ate broad diets, but the diet of erectus appears to have been broader than the earlier habilis. It concludes that “Foodprints teach us that early hominin diets varied over time and space and that we mostly likely evolved to be flexible eaters, driven by ever changing climates, habitats and food availability.”

Analysis of the oldest identified feces from some of our ancestors– Neanderthals living about 50,000 years ago in what is now Spain– suggests that they “predominantly consumed meat… but also had significant plant intake”. Analysis of tooth plaque from Neanderthals in Europe suggests that they ate a wide variety of plant foods including cooked grains and legumes, although teeth from some sites suggest a mostly animal-based diet while others suggest a mostly plant-based diet. Neanderthals in Asia also ate both animals and plants. The overall picture is that Neanderthals ate animals and plants, but the proportion of plant vs. animal foods in the Neanderthal diet may have varied.

Stable isotopes in ancient bone collagen are another way to study what our ancient ancestors may have eaten, but most of the research comes from Neanderthals and Homo sapiens in ice age Europe, while our ancestors in other times and places (representing most of human evolution) have gotten less attention. These stable isotope studies suggest that Neanderthals and Homo sapiens in ice age Europe ate a heavily animal-based diet. A recent analysis reports that the methods used in these stable isotope studies may have underestimated plant food intake and provides evidence that Neanderthals at one site ate plant foods regularly, possibly as a major part of the diet. This would be consistent with the data from ancient tooth plaque and feces, which together provide evidence for an omnivorous dietary pattern.

It’s important to keep in mind that Neanderthals only contributed a small amount of genetic material to modern humans of European and Asian origin, and no genetic material to modern humans of African origin. In other words, Neanderthals are (some of) our ancestors, but barely. Similarly, only a minority of modern European DNA can be traced back to the Homo sapiens hunter-gatherers who lived in ice age Europe. So it’s not clear how relevant these ancient menus are to modern people of European descent, and they don’t seem very relevant at all to people of non-European descent. The rest of the ancestry of Europeans comes from farmers who migrated northward into Europe, and herders who migrated westward from the Eurasian steppe, both after the ice age.

Another line of evidence is the diets of current and historical hunter-gatherers, who live somewhat similarly to how our distant ancestors did. Loren Cordain and colleagues estimated the proportion of plant and animal foods for 229 hunter-gatherer cultures around the world. All of these cultures ate animal foods, but more than 95% have been documented to eat plants. Cultures with very low plant food intake were typically located in the arctic where plant foods are scarce, suggesting that humans eat plant foods when they are available. The Carnivore Code argues that current and historical hunter-gatherers eat less animal food than our distant ancestors did because they live in marginal habitats, but two studies of this idea did not support it.

The Carnivore Code argues that “fruit might be available [to hunter-gatherers] a few times a year for a very short period of time” (p. 9), implying that fruit was a minor and seasonal component of the diet. However, detailed studies of the diets of two African hunter-gatherer groups, the Hadza and !Kung, suggest that fruit is a major part of the diet for most or all of the year (The !Kung San, Lee 1979). This is because the areas where they live, and where our human ancestors did most of their evolving, do not experience the same type of seasonality that temperate regions experience. The diet of both cultures is omnivorous but mostly based on plant foods.

We aren’t experts on ancient diets (and neither is Paul Saladino) and we recognize that there is still a lot of uncertainty that may be resolved through future research and debate. Yet the bulk of the currently available evidence doesn’t point in the direction of our distant ancestors consistently being carnivores and only relying on plants as occasional “fallback foods”. The evidence suggests that most ancient human diets were probably omnivorous. The picture of ancient human diets is more uncertain and complex than painted in The Carnivore Code.

Criterion 1.2. Are the references cited in the book to support the claim convincing?

2 out of 4

The book’s references received a score of 2, meaning that they are variable in quality (including some high quality) and only weakly support the book’s claim. The references establish that our distant ancestors ate animal foods, but not that they avoided plant foods.

Chapter 1 contains most of the references related to this claim. We focus on key references related to central pieces of the argument. The Carnivore Code cites references suggesting that our probable ancestors used tools to butcher animals around 2.5 million years ago. This is true but it doesn’t establish that they avoided plant foods.

The strongest evidence supporting the book’s claim comes from stable isotope studies, but the book only directly cites one such study. This study reports that the diet of early Homo “was based more on meat products” than Australopithecus africanus and Paranthropus robustus, which are more distantly related to us. The book also cites a review paper from 2007 that reviews some of the isotope studies and concludes “these results alone would indicate that even very early hominids consumed a considerable proportion of meat in their diet.” This is consistent with other evidence but it doesn’t establish that they avoided plant foods.

The Carnivore Code points out that salivary amylase, a starch-digesting enzyme that most humans have extra gene copies of, duplicated after the split between Homo sapiens, Neanderthals, and another branch of Homo, the Denisovans. This is taken as evidence that Homo sapiens didn’t eat much starch at the time they diverged from Neanderthals and Denisovans about 600,000 years ago, bolstering the case for carnivory.

The book cites a study that establishes that Neanderthals and Denisovans didn’t have duplications of the salivary amylase gene, supporting the idea that this duplication occurred after Homo sapiens split from those lineages. The book argues that this means starch intake was low before the split, and implies that this is evidence for carnivory. However, three facts undermine this argument. First, salivary amylase only plays a minor role in digesting starch, while other digestive enzymes are more important. Second and relatedly, salivary amylase gene duplications are not required to digest starch-rich foods, since people without these duplications can apparently digest starch just fine. This means that ancient humans could have consumed starch-rich foods prior to salivary amylase gene duplication. Third, not all plant foods are rich in starch (e.g., leafy greens or fruit), so even if ancient humans ate little starch, that doesn’t necessarily imply a carnivorous diet.

The book cites references suggesting that humans have adaptations that help us hunt, like shoulders adapted for throwing and feet adapted for distance running. We think this is logical, and it’s consistent with other evidence that our ancestors began hunting and eating animals more regularly sometime around 2.5 million years ago. But it doesn’t establish that they avoided plant foods.

Criterion 1.3. How well does the strength of the claim line up with the strength of the evidence?

1 out of 4

The claim received a score of 1, meaning that the strength of the claim is substantially overstated. The Carnivore Code states confidently that we evolved as carnivores, but there is a lot of uncertainty about the diets of our distant ancestors. The current scientific understanding of ancient diets relies on very spotty evidence, and the evidence that’s available doesn’t align well with the claim in The Carnivore Code.

Overall (average) score for claim 1

1.3 out of 4

Claim 2

Dietary fiber is not helpful for constipation.

Supporting quote(s) and page number(s)

Page 135: “Examination of the literature regarding fiber and constipation quickly illustrates that plant fiber does not lead to better outcomes in patients with this condition.”

Page 136: “Yes, you read that correctly, research has shown that kicking fiber to the curb can lead to complete resolution of constipation.”

Page 137: “How anyone can claim that plant fiber benefits constipation is beyond me… I would go even further to suggest that plant fiber is worsening constipation in many individuals.”

Criterion 1.1. How well is the claim supported by current evidence?

2 out of 4

This claim received a score of 2, meaning that it’s weakly supported by current evidence. Some evidence supports the claim that dietary fiber is not helpful for constipation, but the weight of evidence weakly suggests that it’s helpful, at least for adults. One trial emphasized in The Carnivore Code reports that removing nearly all plant fiber from the diet completely resolved constipation in adults living in Singapore, but while the finding is intriguing, it has serious limitations and to our knowledge it has not yet been replicated.

In support of the idea that fiber is not beneficial, The Carnivore Code cites a 2012 meta-analysis of randomized controlled trials in which people with constipation were given fiber as a treatment. It reports that fiber increases stool frequency but not the likelihood of successful treatment. However, the meta-analysis includes very few trials. For treatment success in particular, it only includes two trials, which is usually considered too few studies to do a meta-analysis. The most we can conclude from this is that in two trials, fiber didn’t improve treatment outcomes in people with constipation.

We conducted a quick search for other meta-analyses. A second meta-analysis published in 2016 includes more randomized controlled trials. Across four trials, supplementing fiber increased the likelihood of successful constipation treatment by 71%. Fiber also improved stool frequency and consistency. Effects were consistent across different types of fiber, and high-dose fiber supplementation was more effective than low-dose.

A third meta-analysis in children didn’t find evidence that fiber improves treatment success for constipation.

We haven’t deeply evaluated these meta-analyses, but it seems that there is at least some evidence that fiber is beneficial for constipation. The 2016 meta-analysis showing benefits of fiber seems more convincing than the 2012 meta-analysis, in part because it includes a larger number of trials, but we would have to examine them more closely to come to stronger conclusions. It is true that not all trials or meta-analyses show benefits for constipation, so the evidence seems inconsistent.

The Carnivore Code repeatedly emphasizesone study in particular, in which 63 adults in Singapore with constipation were instructed to eat a very-low-fiber diet for 6 months. They were asked to eat white rice and white bread rather than whole grain, and minimize fruit and vegetable intake. The paper refers to it as a “no-fiber diet”, but even assuming perfect compliance it would still have included some plant fiber, given that participants were not instructed to eliminate all fruit and vegetables, and even white rice and white bread contain a small amount of fiber.

The paper reports that at 6 months, 41 patients were still following the very-low-fiber diet, while 16 were on a reduced-fiber diet, and 6 were back to their previous diet. Among those who remained on the very-low-fiber diet, no one experienced any recorded symptom of constipation at 6 months. Among those on the reduced-fiber diet, 75% were still constipated. Among the 6 people who went back to their original diet, 100% were still constipated.

This finding does seem remarkable, but we don’t think it provides convincing evidence that a low-fiber diet helps with constipation, for the following reasons:

1. This wasn’t a randomized trial comparing people assigned to high-fiber vs. low-fiber diets (which is the gold standard for making cause-and-effect conclusions). The people who discontinued the low-fiber diet and returned to their previous diet did so by choice rather than random assignment. The paper analyzes them as if they were different groups that can be compared to one another, but they were all the same group at the beginning and the paper is comparing people who adhered to the diet vs. people who stopped adhering at various times. This could introduce a type of self-selection bias that can make interventions look more effective than they really are. In this way, the study is more like an observational study than a controlled trial.
2. Only 65% of people remained on the very-low-fiber diet at 6 months. This is a high dropout rate, which can introduce attrition bias. For example, it could be the case that some of the people who discontinued the very-low-fiber diet did so because it didn’t help or worsened their constipation, and only the people who did well on the diet stuck with it. This would make the diet look better than it really was.
3. There were only six people that returned to their previous diet. This isn’t enough people to come to reliable conclusions about differences between diets. Sixteen people were supposedly eating an intermediate amount of fiber by the end of the study, but it’s not clear how much fiber they were eating or for how long.
4. Regression toward the mean could explain some of the improvement in the very-low-fiber group. When people are selected for a study based on symptoms, some of them tend to improve spontaneously over time (people often see a doctor and get recruited into studies when their symptoms are at their worst). So even with no intervention, they might be expected to improve over 6 months. Randomized controlled trials avoid this pitfall by comparing the intervention group to a randomly assigned control group that’s also followed over time, but that wasn’t done here.
5. The paper doesn’t report how much fiber people actually ate, making the results hard to interpret.
6. To our knowledge, no one has tried to repeat this experiment in a more rigorous way to see if it holds up. It could be true, but given that it’s one study with a surprising finding, we’d like to see it confirmed.

This study is thought-provoking but it’s nothing to hang your hat on. It seems worth following up on this finding with additional research, but until then, its conclusions remain highly uncertain.

We also think it’s worth mentioning that when researchers study constipation in rats, the way they give them constipation is by feeding them a low-fiber diet.

Criterion 1.2. Are the references cited in the book to support the claim convincing?

2 out of 4

The references in support of the book’s claim that fiber is not helpful for constipation received a score of 2, meaning that they’re only weakly convincing but most of them are consistent with the book’s claim. The references for this claim appear in chapter 9.

The Carnivore Code cites three studies to support its claim that fiber isn’t helpful for constipation. The first is a 2012 meta-analysis of randomized controlled trials that reports that supplementing fiber isn’t effective for treating constipation in adults. We discuss this meta-analysis in more detail above, but the bottom line is that it only represents a meager two trials for the most important outcomes, and the result wasn’t replicated by a 2016 meta-analysis that included more trials. Still, it shows that at least some studies haven’t found fiber supplements to be effective for constipation.

The second study cited by The Carnivore Code is a randomized controlled trial that compares two different high-fiber diet interventions in children with constipation. Both diets increased fiber intake, but one increased it slightly more than the other at 3 months. This difference was no longer significant by 6 months. Not surprisingly, there was no difference in constipation outcomes between groups. However, at one year, 60-70% of parents in both high-fiber groups reported that the diet had improved their child’s constipation. This study has serious problems, one of which is very high dropout rates. It doesn’t support the book’s claim.

The third study cited by The Carnivore Code is the very-low-fiber intervention study discussed above. It supports the book’s argument, although it has a number of problems that make it a weak piece of evidence.

Criterion 1.3. How well does the strength of the claim line up with the strength of the evidence?

1 out of 4

The claim received a score of 1, meaning that the book presents the evidence as substantially stronger than it is. As reviewed above, the evidence is more uncertain and less supportive than portrayed in The Carnivore Code.

Overall (average) score for claim 2

1.7 out of 4

Claim 3

High LDL cholesterol on a carnivore diet is not harmful.

Supporting quote(s) and page number(s)

Page 182: “I’m sad to say that we’ve been led astray with regard to the true character of LDL, which is much more of a superhero than a supervillain.”

Page 183: “Furthermore, there are many studies suggesting that higher levels of LDL are protective as we age…”

Page 185: “Doesn’t it seem a bit incongruous that nature would have designed something that is so valuable but also damages our arteries and causes atherosclerosis? How can LDL be both protective and harmful? This doesn’t seem to make any sense? The answer is that LDL itself is not harmful, but in certain situations, it can be involved in the process of responding to injury and inflammation– making it look like it’s a bad actor when it’s merely present at the scene of the crime.”

Page 189: “If we are insulin resistant, higher levels of LDL may very well contribute to plaque formation and progression, but if we are insulin sensitive, higher levels of LDL are not associated with increased atherosclerosis and are likely protective.”

Page 192: “Those of us with good insulin sensitivity are essentially a different breed, and there are many striking stories of plaque regression among insulin-sensitive individuals with ‘elevated LDL’ eating carnivore or ketogenic diets.”

Page 192: “Want to live a long time? Eat in a manner that allows for insulin sensitivity, decreases inflammation, and leads to a robust amount of valuable LDL particles. Carnivore diet, anyone?”

Criterion 1.1. How well is the claim supported by current evidence?

0 out of 4

This claim received a score of 0, meaning that the evidence opposes the claim overall. We think this claim is not merely undermined by the evidence, but also dangerous. Given the very strong evidence that high LDL cholesterol promotes cardiovascular disease in people eating typical diets, and the lack of evidence that people on a carnivore diet are exempt from this, we think it would be very unwise to ignore or even celebrate high LDL as the book recommends.

LDL is one of the particles that carries cholesterol and fats in the bloodstream. It causes plaque to form in arteries that leads to cardiovascular events like heart attacks and strokes. The higher the concentration of LDL particles in the bloodstream, the higher the risk. The amount of cholesterol carried by these particles (LDL cholesterol) is also a good predictor of risk over long periods of time.

The idea that higher LDL increases the risk of cardiovascular events is strongly supported by several independent lines of evidence:

1. Higher lifetime exposure to LDL cholesterol correlates with higher cardiovascular risk.
2. Lowering LDL via multiple drug classes or diet reduces the risk of cardiovascular events. The more it’s lowered, the lower the risk of events.
3. People with genes that raise or lower lifetime LDL exposure experience higher or lower risk of cardiovascular events, respectively. These genes influence LDL via multiple biological pathways but have a remarkably consistent impact on risk. These studies show that lifetime LDL exposure is a powerful determinant of cardiovascular risk.

There’s not much information on the impact of a carnivore diet on LDL cholesterol. Anecdotally, many people on a carnivore diet appear to eat large amounts of red meat rich in long-chain saturated fat and low in polyunsaturated fat, suggesting that it would probably increase LDL. This is consistent with the findings of a survey of 2,029 people who are currently eating a carnivore diet, which reports that in a subset of people who measured it before and after, LDL cholesterol averaged 126 mg/dL before the diet and 172 mg/dL while on the diet. This is a large increase, and the paper describes LDL cholesterol on the carnivore diet as “markedly elevated”. However, this has to be interpreted cautiously because it’s a survey, not a controlled study.

In this context, we think it’s relevant that the book’s author Paul Saladino has publicly shared that he has extremely high LDL cholesterol (533 mg/dL) and LDL particle number (3,283 nmol/L). LDL cholesterol above 190 mg/dL is considered “very high”. Saladino states that a coronary artery calcium scan revealed a score of zero, meaning that he doesn’t have the type of advanced coronary disease that type of scan detects (calcified plaque). Given that coronary artery disease usually takes decades to accumulate and he’s only been on the diet for 3-4 years, this is hardly reassuring.

Does high LDL cholesterol have a different impact on people eating a carnivore diet than on the general public? This hasn’t been directly studied. Cardiovascular risk is determined by multiple factors, and although LDL is a big one, it’s not the only factor. For any given level of LDL, people who have healthier blood pressure, better blood sugar regulation, a healthier level of body fat, don’t smoke, and have higher physical fitness will be at lower risk than people without those advantages.

If the carnivore diet improves some of these other risk factors, it could mitigate the risk caused by high LDL. It’s even possible for LDL to increase but overall risk to decline. Yet that’s not what the book argues. What it argues is that LDL itself becomes irrelevant as a risk factor, or even beneficial, on a carnivore diet. Even when it’s extremely high. We see no reason to believe the well-established harmful effect of LDL cholesterol on cardiovascular health wouldn’t apply to people eating a carnivore diet, and The Carnivore Code presents no convincing evidence that it doesn’t. See a more detailed discussion of the book’s arguments and citations in the next section.

This brings us to a recent study conducted by epidemiologist Deirdre Tobias, ScD. She sought to answer the question of whether high LDL cholesterol predicts a higher overall risk of dying in people who are metabolically healthy, using a long-term (20+ year) study of about 19,000 women. Even among metabolically healthy women with high “good cholesterol” (HDL >50 mg/dL), low triglycerides (<100 mg/dL), not overweight (BMI <25), not smoking, normal blood pressure, and healthy blood sugar (HbA1c <6.5), high LDL predicted a higher overall risk of dying relative to normal LDL (Tobias summarized her findings on Twitter here). The impact of LDL was remarkably consistent across different combinations of other risk factors.

It’s also notable that metabolically healthy people in the highest LDL category, who had a higher risk of dying than metabolically healthy people with lower LDL, had an average LDL cholesterol similar to people eating a carnivore diet in the survey study discussed previously (169 vs. 172 mg/dL).

These findings suggest that regardless of metabolic health, high LDL still increases risk. This study hasn’t been peer reviewed yet so it should be interpreted with caution, but it also has notable strengths: analyses were all designed and publicly registered in advance, it was conducted by an experienced epidemiologist, and the data set is large, detailed, and represents 20+ years of follow-up. In addition, it’s an extension of previous work that has been peer reviewed.

We also think it’s worth mentioning studies of 500-year-old Inuit mummies. Inuit people live in the arctic and traditionally eat a nearly carnivorous diet, although much of it is seafood. A few ancient individuals have been found, preserved by the cold, and their arteries have been studied. The results show that these people suffered from atherosclerosis (plaque in the arteries) that was surprisingly severe for their age. There are possible explanations for this other than diet, such as indoor smoke exposure, but the nearly-carnivorous diet certainly didn’t prevent them from developing cardiovascular disease.

We think it would be very unwise for people eating this diet to assume they’re exempt from the harmful effects of high LDL. LDL is easy to measure, and high LDL can be treated effectively using diet and/or medication.

Criterion 1.2. Are the references cited in the book to support the claim convincing?

1 out of 4

The book’s references in support of this claim received a score of 1, meaning that they generally don’t support the book’s claim that high LDL on a carnivore diet is not harmful. We focus on the references in The Carnivore Code related to the connection between LDL, cardiovascular disease, and all-cause mortality. They are in chapter 11.

On page 183, The Carnivore Code states that high LDL is not a risk factor for cardiovascular or all-cause mortality in the elderly, and in fact higher LDL is “protective” in this age group. It cites fourteen studies to support this argument (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14).

Only three of these studies measured LDL cholesterol, so The Carnivore Code’s conclusions about LDL are somewhat misleading. Instead, most of them report total cholesterol, which is a mix of “bad cholesterol” (LDL, VLDL) and “good cholesterol” (HDL). These studies are consistent with the commonly-accepted observation that high total cholesterol isn’t a strong predictor of death risk in older people, and can sometimes even predict lower risk.

Three of the cited studies report on LDL, or measures closely related to it (LDL cholesterol, ApoB, and non-HDL cholesterol). Of those, one reports that higher LDL is correlated with more heart attacks in women but not men, and two report no association between LDL and cardiovascular or all-cause mortality.

We think these studies do establish that measuring LDL isn’t a very informative predictor of risk in the elderly, although they don’t suggest it’s “protective”. However, they are observational studies, and as The Carnivore Code points out in other contexts, observational studies have limitations. As stated in one review paper, these observational studies are “unavoidably vulnerable to confounding, reverse causation, and other forms of bias”.

How do we get around this problem? One approach is to use Mendelian randomization, which uses the random distribution of gene variants in a population to measure the causal impacts of LDL. Some genes cause people to have higher lifetime LDL, others lower, and we can see how those genes correlate with risk in older people. Fortunately, Mendelian randomization has been applied to this question. The conclusion: “a genetic predisposition to high LDL-C levels contributes to mortality throughout life, including in the oldest old, and a beneficial LDL genetic risk profile is associated with familial longevity.” In other words, if you want to live a long time, it’s probably best to try to avoid high LDL cholesterol.

The Carnivore Code also states that “epidemiology studies of women, Canadian and Russian men, Maoris, and Asians do not show associations between total cholesterol and or LDL levels and incidence of heart disease or all-cause mortality” (p. 187). It cites five references to support this (1, 2, 3, 4, 5). The argument is this: how could LDL be a major cause of heart disease if the correlation only shows up in some types of people, but not others?

Of these five citations, only two are directly relevant to the relationship between LDL and heart disease or mortality. In the first, a “J-shaped association” is reported between LDL cholesterol concentrations and heart disease mortality risk. This means that risk was highest at both low and high LDL, which doesn’t support the book’s argument that high LDL cholesterol is not harmful. The second is an interesting series of articles that deal primarily with the observation that total and LDL cholesterol do not predict risk of cardiovascular disease and all-cause mortality in Japan.

The observations in Japan are certainly interesting, and consistent with the book’s claim. But does this lack of correlation in observational studies really mean high LDL doesn’t damage arteries in Japan? A cleaner way to answer this question is to examine randomized controlled trials of LDL-lowering drugs. These drugs lower the risk of coronary heart disease in Japanese people too, suggesting that LDL works the same way in Japan as elsewhere.

The Carnivore Code states that “epidemiology studies of… Asians do not show associations between total cholesterol and or LDL levels and incidence of heart disease or all-cause mortality” (p. 187), but this sweeping claim doesn’t appear to be true. A meta-analysis that pooled 29 studies from the Asia Pacific region reports that higher total cholesterol is associated with higher cardiovascular disease mortality. So it seems that the correlation does exist in Asia, just not in Japan specifically.

Hundreds of studies have been published on the relationship between total cholesterol, LDL cholesterol, cardiovascular disease, and all-cause mortality. Citing a small number of studies or contexts that don’t confirm the hypothesis isn’t a strong argument that the overall idea is wrong. If this were true, it would be nearly impossible to prove anything in the biological sciences. What matters is the weight of evidence.

As previously discussed, there are several strong lines of evidence supporting the idea that high LDL increases cardiovascular risk. Together, these arguments are far more convincing than a small number of unsupportive studies selected from a large body of mostly-supportive literature. This paper provides an overview of the evidence.

A key piece of evidence presented by The Carnivore Code is a publication from the Framingham Study, including two graphs of its data on pages 187 and 189. The graphs show that there is a correlation between LDL cholesterol and risk of coronary artery disease, but that at any given level of LDL, people with higher HDL (“good cholesterol”) are at lower risk. In the graph on page 189, it is apparent that even among people with the highest HDL (85 mg/dL), going from an LDL of 100 to an LDL of 220 approximately doubles risk.

The Carnivore Code uses these graphs to argue that LDL only matters in people who have low HDL, which is correlated with insulin resistance. In fact, the graphs illustrate that LDL impacts risk at all levels of HDL. It’s widely accepted that LDL is not the only driver of cardiovascular disease and that at any given level of LDL, differences in other risk factors can alter risk (that’s what makes these independent risk factors). Obviously, a person who has high LDL cholesterol and smokes cigarettes is at higher risk than a person who has high LDL and doesn’t smoke. Similarly, as illustrated in the graphs, if two people have high LDL cholesterol, the one with lower HDL cholesterol will tend to be at higher risk. This isn’t an argument against a role for LDL, it just shows that cardiovascular risk is determined by multiple factors independently. Independent in this sense simply means that each contributes to increased risk, regardless of other risk factors.

The next argument in The Carnivore Code is that the “stickiness” of the LDL particle and the artery wall determine how much LDL gets stuck and causes disease, and that in people with high insulin sensitivity, LDL doesn’t stick in the artery wall so high LDL can be safely ignored (and might even be beneficial). The book cites seven references to support its argument that “during the states of insulin resistance and inflammation, both the LDL particle and the intimal space get coated in ‘molecular velcro’ and become more sticky” (p. 191, 1, 2, 3, 4, 5, 6, 7).

These citations are interesting and are consistent with the idea that insulin resistance accelerates the development of cardiovascular disease, but they don’t support the book’s claim that high LDL cholesterol on a carnivore diet is not harmful. Insulin resistance is a risk factor for cardiovascular disease, and these studies provide one mechanism to explain that, but none of them report that LDL becomes irrelevant or even beneficial in a person who is insulin-sensitive. We think Deirdre Tobias’s study we discussed earlier is relevant here. Even in people who are metabolically healthy, high LDL cholesterol correlates with a higher risk of dying.

Furthermore, it’s not clear that the carnivore diet even increases insulin sensitivity. A study using the gold-standard method for measuring insulin sensitivity reported that insulin sensitivity decreases on a very-low-carbohydrate diet. This often comes as a surprise to people since less accurate but more common measures like HOMA-IR suggest the opposite (due to a decline in fasting insulin, which doesn’t necessarily track with insulin sensitivity in the context of carbohydrate restriction). We aren’t aware of evidence that the carnivore diet increases insulin sensitivity, and The Carnivore Code doesn’t provide any.

In summary, the references cited in The Carnivore Code to argue that high LDL on the carnivore diet isn’t harmful are valid studies and sometimes thought-provoking, but they generally don’t support the argument.

Criterion 1.3. How well does the strength of the claim line up with the strength of the evidence?

0 out of 4

The claim received a score of 0, meaning that the claim is greatly overstated.  As described previously, there is strong evidence that high LDL promotes cardiovascular disease, and we haven’t seen substantial evidence that people eating a carnivore diet would be exempt from this.

Overall (average) score for claim 3

0.3 out of 4

Overall (average) score for scientific accuracy

1.1 out of 4

Reference 1

Reference

Chapter 13, reference 15.  Mahley et al. J Molecular Med 94:739. 2016.

Associated quote(s) and page number(s)

Page 268: “Adequate cholesterol is essential for proper membrane function and fluidity, but cholesterol in the blood doesn’t cross the blood-brain barrier, so the brain must make its own. ApoE is the bridge by which this cholesterol is transported to neurons, and in people with the ApoE4 gene variant, this transfer appears to happen more slowly under conditions of insulin resistance.”

Criterion 2.1. Does the reference support the claim?

2 out of 4

This reference received a score of 2, indicating that it offers weak support to the claim. It’s a review paper that discusses the cardiovascular and brain impacts of ApoE, which is involved in the risk of Alzheimer’s disease. The paper does state that ApoE is involved in cholesterol transport to neurons when they are injured, but it says nothing about this cholesterol transport process being impaired in people with ApoE4. It also doesn’t mention insulin resistance. Instead, it says the ApoE4 protein contributes to neurological disease because of the toxicity of the ApoE4 protein itself and of fragments derived from it.

Reference 2

Reference

Chapter 14, reference 45.  McClellan and DuBois. J Biological Chem 87:651. 1930.

Associated quote(s) and page number(s)

Page 298-299: The Carnivore Code describes an experiment in which two people ate a carnivore diet for approximately a year in 1928 and 1929. The book describes the diet as containing “muscle, liver, kidney, brain, marrow, salt, and fat”, and includes a quote from the paper stating that the two men did well on the diet.

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. The paper is generally as described in The Carnivore Code. We could nitpick a little bit because the paper does say that one of the two men had digestive problems initially. We don’t think this is important enough to dock points. Not mentioned in The Carnivore Code but consistent with its arguments: one of the men had an improvement in blood pressure, and the other experienced a resolution of his gingivitis (gum disease).

Reference 3

Reference

Chapter 6, reference 17. Gonzalez et al. Am J Surg Pathol 15:586. 1991.

Associated quote(s) and page number(s)

Page 83: “Deposition of oxalate crystals in breast tissue has also been documented in multiple studies and is associated with precancerous lesions known as lobular carcinoma in-situ, or LCIS.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. It’s a paper reporting that in biopsies of breast tissue, calcium oxalate crystals are associated with lobular carcinoma in situ, a pre-cancerous lesion. This was only observed in 3 of 100 women, but was also associated with other types of pre-cancer in 6 others. The paper cites other papers in which calcium oxalate crystals have been associated with lobular carcinoma in situ.

Reference 4

Reference

Chapter 4, reference 28.  de Figueiredo et al. Recent Patents on Endocrine, Metabolic & Immune Drug Discovery 9:24. 2015.

Associated quote(s) and page number(s)

Page 46: “Many of the laurels thrown toward sulforaphane come from researchers touting its potential as a cancer chemoprotective agent and claiming that it acts as an antioxidant.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference received a score of 4, indicating that it offers strong support for the claim. The reference discusses the benefits of sulforaphane as an anti-cancer agent and antioxidant, as stated in The Carnivore Code.

Reference 5

Reference

Chapter 13, reference 8. White Oak Pastures beef reduces atmospheric carbon.

Associated quote(s) and page number(s)

Page 277-8: “Analyses of farms like White Oak Pastures in Georgia has demonstrated that regenerative agriculture practices result in a carbon negative ecosystem. On farms like this one and others employing similar grazing practices, more carbon is sequestered into the soil than is released into the atmosphere. This results in a net decrease in greenhouse gas emissions as grass-fed cattle are raised on the land.”

Criterion 2.1. Does the reference support the claim?

3 out of 4

This reference received a score of 3, indicating that it offers moderate support for the claim. The citation is to a blog post on the website of a farm called White Oak Pastures that uses a pasture-based farming system called “regenerative agriculture”, which involves rotational grazing and applying generous amounts of compost to the soil. The blog post links to a lifecycle analysis conducted by a company called Quantis suggesting that beef produced under this farming system actually fights global warming because the greenhouse gases emitted are more than offset by carbon sequestered in the soil.

A later study of the same farm, which includes one of the authors of the Quantis analysis, paints a more complex picture (we note that this study was published after The Carnivore Code). The issue is that White Oak Pastures doesn’t just produce beef, and cows aren’t the only way carbon gets into the soil. The farm is an integrated production system for beef, pork, lamb, goat, rabbits, and poultry (laying hens, guinea fowl, turkeys, ducks, and geese). Carbon sequestered in the soil is the net result of the activity of all these animals, plus the addition of large amounts of compost to the soil for the first three years.

The study reports that for the production system as a whole, net greenhouse gas emissions are 66% lower than conventional meat production, due to carbon sequestration in the soil. In other words, net greenhouse gas production is much lower than regular meat, but not negative. The regenerative agriculture system also caused “dramatic improvement” in soil health relative to the starting point, which was depleted agricultural land. However, the system requires 2.5X more land per unit production, compared with conventional meat.

The follow-up study doesn’t focus on beef specifically since White Oak Pastures produces several types of meat as an integrated system and the beef can’t really be separated out. The paper does say “if we were to attribute the soil C sequestration across the chronosequence to only cattle, MSPR beef produced in this system would be a net sink of −4.4 kg CO2-e kg CW−1 annually”. In other words, if all soil carbon sequestration is attributed to the cattle, the beef alone is carbon-negative. However, since soil carbon levels are due to compost application and the activity of all the farm’s animals, attributing it all to the cattle would be misleading.

The claim in The Carnivore Code is broad, referring to “farms like this one and others employing similar grazing practices”, while the citation refers to one specific farm. We did a quick literature search to see if the study it cites is representative of other farms using similar practices. All studies we identified report that carbon sequestration greatly offsets greenhouse gas emissions from cows, but not necessarily enough to make production carbon-negative (1, 2, 3).

Furthermore, it’s not clear that carbon sequestration continues indefinitely. One study reports that soil carbon levels increased through year 6 of regenerative grazing practices, after which they plateaued. Interpreting the data for White Oak Pastures is more complex, but data from the study suggest that soil carbon does appear to level off at some point after 10 years. This is consistent with research reporting that soils have a limited capacity to store carbon, and when soil carbon hits that limit they don’t continue to store more of it. This implies that regenerative agriculture may only be offsetting greenhouse gas emissions temporarily, after which soil carbon levels reach equilibrium and meat production once again results in high net greenhouse gas emissions. Nevertheless, benefits for soil health are expected to persist indefinitely.

Overall, we think the citation offers substantial support to the book’s claim, but the reality of the situation isn’t quite as rosy as the book suggests.

Reference 6

Reference

Chapter 13, reference 3. Koliaki et al. Int J Peptides 2010:1. 2010.

Associated quote(s) and page number(s)

Page 260: “Losing weight is about creating a caloric deficit, and the easiest way to do this is to eat satiety-promoting foods. Of these, good quality animal fats are king.”

Criterion 2.1. Does the reference support the claim?

0 out of 4

This reference received a score of 0, indicating that it undermines the claim.

Ghrelin is a gut hormone that plays a role in creating hunger around meals. The citation is to a review paper about the impact of macronutrients on ghrelin levels. It concludes that ghrelin is suppressed by protein and carbohydrate, while fat “exhibits rather weak and insufficient ghrelin-suppressing capacity”.

The citation doesn’t discuss animal fat, and notes that “the potential impact of varying fatty acid composition (saturated, monounsaturated and polyunsaturated fat) on postprandial ghrelin response has been only scarcely investigated”. It goes on to review one study that concluded that “increasing saturated fat consumption had no deleterious effects on fasting and postprandial plasma ghrelin concentrations”.

This might be taken to contradict the book’s claim, but we don’t think examining one specific gut hormone in isolation is very informative for understanding the impact of a food on satiety. However, the paper discusses satiety outcomes, and again fat doesn’t fare very well:

• “…fat consumption has also displayed a diminished capacity to induce satiety.”
• “The investigators conclude that increased fat intake might promote obesity not only through its high caloric content and adverse metabolic effects, but also through its failure to suppress postprandial hunger.”

Reference 7

Reference

Chapter 9, reference 26. Foster et al. Biol Trace Element Res 149:135. 2012.

Associated quote(s) and page number(s)

Page 144: “In another epidemiological study of healthy and diabetic women, there was a strong correlation between fiber intake and lower blood levels of zinc, a mineral crucial for proper hormonal balance and the functioning of hundreds of enzymes in the human body.”

Criterion 2.1. Does the reference support the claim?

0 out of 4

This reference received a score of 0, indicating that it undermines the claim. The paper reports no correlation between fiber intake and blood zinc level: “No significant correlations were found between plasma zinc and the intake of dietary zinc or any other individual nutrient. Regression analysis indicated that plasma zinc is not significantly predicted by age, BMI, health status, the presence or absence of ZnT8 mRNA expression, or any of the zinc bioavailability ratios.”

Given that plasma zinc is an informative measure of zinc status, this study suggests that fiber intake is not correlated with zinc status in this sample of Australian women.

The study does report that overall zinc bioavailability from the diet may be on the low side in this sample. However, the participants’ zinc status appeared to be fine, probably because zinc bioavailability in the diet was sufficient: “The estimated average zinc absorption of the participants in the present study (30% of 12 mg/day) may be considered sufficient to meet total endogenous zinc losses.”

Reference 8

Reference

Chapter 7, reference 29. Gundry. Circulation 137:238. 2018.

Associated quote(s) and page number(s)

Page 104: “Dr. Gundry has published an impressive case series describing significant improvements in a large group of patients with autoimmune disease who were treated with a very low-lectin diet that involved the removal of all grains, beans, legumes, peanuts, cashews, nightshades, squashes, and dairy products. …I agree with him that the elimination of foods with high-lectin content was likely at the root of the reversal of these patients’ autoimmune issues.”

Criterion 2.1. Does the reference support the claim?

2 out of 4

This reference received a score of 2, indicating that it offers weak support for the claim.  There are two claims being made here, one trivial and one substantial. The trivial claim is that the citation is as The Carnivore Code describes it. The more substantial claim is that this citation provides evidence that reducing lectin intake improves/reverses autoimmune disease.

The citation in question is an abstract for a poster presentation that was given at a scientific meeting. It reports that in a consecutive series of 102 patients with diverse autoimmune diseases who were advised to eat Gundry’s diet, 95 “achieved complete resolution of autoimmune markers and inflammatory markers within 9 months”. The diet is intended to eliminate the intake of lectins that Gundry believes are harmful, and it also includes probiotics, prebiotics, and polyphenols intended to improve gut health. For the most part, the study is accurately described in The Carnivore Code, although it appears that the patients were only asked to eliminate A1 dairy rather than all dairy.

This is a poster abstract, not a peer-reviewed scientific publication, so it hasn’t been through the critical expert review that science typically goes through. Furthermore, the abstract was posted in 2018 and we have not been able to find a peer-reviewed paper resulting from this work as of January 2022. In fact, we have been unable to find any peer-reviewed paper published by Gundry on this topic as of January 2022.

The findings reported in this abstract are certainly extraordinary. If they’re true, they would represent a radical advance in the understanding and treatment of autoimmune disease. The idea that dietary lectins could affect gut permeability, and ultimately autoimmune disease risk, is intriguing and certainly not out of the question. However, we don’t think this study lends strong support to the claim, for the following reasons:

• Clinical case series like this are a valid form of evidence but they’re much less convincing than gold-standard study designs like randomized controlled trials. Case series are at much higher risk of bias for several reasons, which is why they’re usually viewed as a way to generate hypotheses rather than test them. Usually, when extraordinary preliminary findings like this are followed up with more rigorous study designs, the results aren’t as impressive and they often disappear completely.
• The findings aren’t peer-reviewed.
• The findings aren’t published in a journal article where we can see the details.
• The findings haven’t resulted in a peer-reviewed scientific paper since they were posted in 2018, which is unusual and raises major questions.
• Gundry has no peer-reviewed scientific studies on this topic that we were able to find, nor did we find peer-reviewed studies from other researchers with similar findings.

Reference 9

Reference

Chapter 9, reference 5. Dukowicz et al. Gastroenterology & Hepatology 3:112. 2007.

Associated quote(s) and page number(s)

Page 137: “At its root, [small intestinal bacterial overgrowth, SIBO] appears to be a problem with gut motility. Normally, peristaltic waves pass through the length of our small bowel and sweep down towards the colon, preventing overgrowth of bacteria in the upper portions of our digestive tract. These waves are known as the migrating motor complex and occur every 45-180 minutes between meals. In patients with SIBO, the migrating motor complex appears to be hypoactive, allowing the populations of bacteria in the colon to move up into the small intestine, leading to an imbalance and a loss of diversity there.”

Criterion 2.1. Does the reference support the claim?

3 out of 4

This reference received a score of 3, indicating that it offers moderate support for the claim. The reference is a review paper on SIBO. Although the paper does present gut motility problems as an important contributor to SIBO, it presents low stomach acid secretion as equally important (“The two processes that most commonly predispose to bacterial overgrowth are diminished gastric acid secretion and small intestine dysmotility”). It also lists immune system problems and anatomical abnormalities as other possible contributing factors. The Carnivore Code argues that problems with gut motility are the root cause of SIBO, but the paper paints a more complex picture.

In addition, The Carnivore Code says gut peristaltic waves “occur every 45-180 minutes between meals”, but the paper says they occur every 90-120 minutes.

Reference 10

Reference

Chapter 11, reference 131. Mavropoulos et al. Nutrition & Metabolism 2:35. 2005.

Associated quote(s) and page number(s)

Page 205: “Science has also recently begun to support the things we already knew to be true, and studies have shown that high-fat ketogenic diets containing plenty of saturated fat reverse diabetes and insulin resistance. They also lead to weight loss and improvements in inflammatory markers, as well as a reduction in hypertension, dementia, polycystic ovarian syndrome, and a host of other conditions.”

Criterion 2.1. Does the reference support the claim?

2 out of 4

This reference received a score of 2, indicating that it offers weak support for the claim. For the purposes of this citation, the relevant part of the quote in The Carnivore Code is arguing that “high-fat ketogenic diets containing plenty of saturated fat” improve polycystic ovarian syndrome (PCOS). The broader context of the passage is that The Carnivore Code is arguing that saturated fat is not unhealthy.

The citation is a pilot study testing the impact of a low-carbohydrate ketogenic diet on signs and symptoms of PCOS. Eleven women with overweight or obesity were asked to eat a low-carbohydrate ketogenic diet containing 20 grams of carbohydrate a day or less, for 24 weeks (this is very low). There is no measurement or discussion in the paper of how much saturated fat the diet contained, and it included items that are both low and high in saturated fat (diet advice was based on the book Dr. Atkins’ New Diet Revolution). The research team took measurements every two weeks.

Of the 11 who started the trial, only 5 completed it. Among the five completers, there were improvements in signs and symptoms of PCOS, including hormonal mediators of the condition. They also lost weight.

The citation does suggest that a ketogenic diet may be helpful for PCOS, although it was a very small study with a 55% dropout rate and no control group so it must be taken with a large grain of salt. It was accurately described by the authors as a “pilot study”, in other words, good for generating hypotheses but not a rigorous test of the idea.

Importantly, the study doesn’t provide support for the book’s argument about saturated fat (and also doesn’t refute it), which was the main argument being made in the passage. We don’t know how much saturated fat the participants ate, and we don’t know whether outcomes would have been better or worse with a different type of fat.

Overall (average) score for reference accuracy

2.4 out of 4

Summary of the health-related intervention promoted in the book

The Carnivore Code recommends a diet based on meat, including various organs, and eggs. Chapter 12 describes five tiers of strictness, from a “carnivore-ish diet” that includes the “least toxic” plant foods like low-sugar berries, to one that is entirely carnivorous. It presents the fully carnivorous Tier 5 diet as the optimal one. All diets include water as the main beverage, and 6-10 grams of salt per day, but exclude spices.

Condition targeted by the book, if applicable

The book’s most emphatic claim is that the diet will help the reader “kick as much butt as possible”, “kick more butt”, “kick a whole lot of butt”, etc. We interpret this to mean the diet is intended to make people feel better and perform better.

The book states or implies that the diet is effective for a wide variety of health conditions, including obesity, diabetes, depression, and autoimmune disease:

• Page xxii: “There are now thousands of people with experiences similar to mine demonstrating improvement and resolution of a variety of diseases such as ulcerative colitis, Crohn’s, lupus, thyroid disease, psoriasis, multiple sclerosis, rheumatoid arthritis, and psychiatric illnesses like depression, bipolar, and anxiety. This is in addition to the thousands of others who have used the carnivore diet to lose weight, reverse diabetes and insulin resistance, or to improve libido and mental performance.”
• Page 103: “One of the more positive effects experienced by those on a carnivore diet is weight loss. There are now thousands of stories of people easily losing weight when plants are eliminated from their diet and high-quality animal foods become the focus.”
• Page 116: “Remember when I said red meat might just be the best anti-depressant? I wasn’t joking!”
• Page 214: “If you have autoimmune disease, significant gut issues, or evidence of inflammation, I would recommend starting with at least a Tier 2 carnivore diet or, preferably, Tier 3.”

Apparent target audience of the book

The book appears to be written for a broad audience who may want to feel or perform better, lose weight, address a wide variety of existing health conditions, or prevent a wide variety of health conditions.

Criterion 3.1. Is the intervention likely to improve the target condition?

2 out of 4

The intervention received a score of 2, meaning that there is very little evidence supporting a beneficial effect of the diet but it may be helpful for some of the conditions it discusses. On one hand, this book is hard to score because it claims to help with just about everything. On the other hand, it’s easy to score because there is virtually no direct evidence for any of its health claims.

Although there’s virtually no direct evidence, we think there are some common health conditions that a carnivore, or “carnivore-ish”, diet is likely to improve. The most obvious is obesity. There is no compelling direct evidence on the impact of the carnivore diet on body weight in people who have obesity, but the diet has properties that are expected to cause weight loss:

• It’s high in protein.
• It’s very low in carbohydrate.
• It’s relatively monotonous (low variety).
• It eliminates tasty, calorie-dense processed foods that many people have a hard time controlling their intake of.

Based on these properties, we would expect a carnivore diet to be more effective for weight loss than most other diets for people who can stick with it, but this has yet to be proven.

We also think the diet is probably helpful for managing type 2 diabetes, both because it probably causes fat loss, and because it’s very low in carbohydrate. Although it wasn’t a randomized controlled trial, the Virta Health study was a high-quality non-randomized controlled trial that reported that a very-low-carbohydrate (ketogenic) diet plus intensive support is helpful for managing type 2 diabetes.

Regarding the other claims in the book, the evidence is nearly all anecdotal so it’s highly uncertain whether the diet is effective. The strongest single piece of evidence we’ve found is a survey-based study of 2,029 people eating a carnivore diet. In this survey:

• 69% reported improvement in a variety of chronic diseases
• 95% reported improvement in overall health
• 85% reported improvement in mental clarity
• 78% reported improvement in strength
• 89% reported improvement in energy

At face value, this looks extremely impressive, and supportive of the book’s claim that the diet will heal diseases and help us “kick more butt”. However, it’s hard to interpret anecdotal figures like these. This survey was conducted on people who were part of carnivore social media groups such as the World Carnivore Tribe Facebook group. These are people who are enthusiastic about the diet and feel that it has helped them. People who tried the diet, didn’t benefit or were harmed, stopped eating it, and aren’t part of these social circles wouldn’t be represented in these data. The results could also be biased by social pressure and the respondents’ desire to paint their chosen diet in a positive light. This is why interpreting anecdotal evidence is hard.

It’s easy to find anecdotal evidence from people who did poorly on the carnivore diet and as a result, didn’t stick with it. The book’s author himself, Paul Saladino, no longer eats a fully carnivorous diet. His diet is still mostly meat but it now includes a substantial amount of fruit, honey, and white rice, and his stated reason is that he feels better with the addition of these foods. It seems that at a minimum, a purely carnivorous diet doesn’t work for everyone.

Overall, we believe there are people who have benefited from the carnivore diet, but we don’t know how representative they are of the typical person who might be interested in trying the diet. We think the diet is likely to be helpful for obesity and type 2 diabetes, but beyond that we really don’t know. We think the anecdotal evidence for improvements in health and well-being are intriguing and could be worth additional study.

Criterion 3.2. Is the intervention likely to improve general health in the target audience?

1 out of 4

The intervention received a score of 1, mostly reflecting the fact that its long-term effects on general health are unknown, but it has a few predicted benefits and risks.

The evidence on the impact of the carnivore diet on long-term health is very slim, limited almost entirely to anecdotal reports like those collected in this survey-based study and a smattering of published clinical cases. Above, we argue that the diet probably has some effectiveness in treating obesity and type 2 diabetes.

One condition we’re concerned about is cardiovascular disease, which is the most common cause of death globally. The typical composition of the carnivore diet is expected to increase LDL cholesterol, which is a key driver of plaque buildup in the arteries. This is consistent with the data in the survey-based study, which reports average LDL cholesterol values of 126 mg/dL before the diet and 172 mg/dL while on the diet. These two values are considered “near or above optimal” and “high”.

The Carnivore Code argues that high LDL on a carnivore diet is not harmful, and may even be beneficial. We evaluated this claim in the scientific accuracy section and found it poorly supported. LDL is an important contributor to cardiovascular disease risk in people eating typical diets, and there is no substantial evidence that people on a carnivore diet would be exempt from this.

That said, LDL isn’t the only determinant of cardiovascular risk, and it isn’t the only cardiovascular risk factor potentially impacted by the carnivore diet. To the extent that the diet causes body fat loss and metabolic improvements, it may not be as harmful as implied by changes in LDL cholesterol. Depending on the balance of changes that occur in various risk factors, it could even reduce cardiovascular risk in some people. We don’t really know what the overall impact would be on average, and in fact no one knows because rigorous studies haven’t been done yet. However, we think that people who experience high LDL cholesterol on the diet should be concerned and seek medical advice.

It’s also worth pointing out that the carnivore diet described in the book is high in sodium (salt). In addition to the roughly 800 mg of sodium contained in the foods themselves (according to our nutrient analysis below), The Carnivore Code recommends eating 6-10 grams of added salt per day (page 217). This totals about 3,100-4,700 mg of sodium per day, which is higher than the USDA Dietary Guidelines recommended intake of 2,300 mg/day or less. The high sodium content and low potassium content of this diet is a potential risk factor for high blood pressure and cardiovascular events. We note that some people have argued that sodium needs are higher on very-low-carbohydrate diets, and this is consistent with the fact that insulin plays a role in sodium handling in the kidney. We’re uncertain about whether the higher salt intake is a problem in this context.

As far as the impact of the diet on other health conditions, we don’t think existing evidence is very informative. We think the anecdotal evidence for improvements in health and well-being are intriguing and could be worth additional study.

Criterion 3.3. Does the diet portion of the intervention promote an adequate nutrient intake for general health in the target audience?

0 out of 4

The diet received a score of 0, indicating that it is substantially nutritionally inadequate (moderately insufficient levels of three or more relevant nutrients).

The Carnivore Code presents its purely carnivorous nose-to-tail “Tier 5” diet as the optimal version. We began by putting the one-day sample meal plan on page 243 into the Nutrium.com meal planner and analyzing how well it meets the recommended daily allowance (RDA) for nutrients. Highlights:

• The suggested meal plan contains 4,320 kilocalories, which is much more than the average person eats per day. Realistic calorie intake for the average adult is around two-thirds of this.
• The meal plan is 77% fat, 21% protein, and 1.5% carbohydrate.
• It’s adequate in most vitamins and minerals.
• It’s inadequate in magnesium (60% of RDA), manganese (78%), potassium (70%), vitamin C (60%), vitamin E (41%), and vitamin K (16%).
• It’s inadequate in vitamin D, but so is almost any diet. Vitamin D must be mostly obtained from sunlight, fortified foods, and/or supplements.
• Its adequacy in calcium depends on eating bone meal powder (or powdered eggshell), which we are skeptical that most people will do. Without these supplements, calcium intake is 29% of RDA.

We note that if we cut back all portions by one-third, making total calorie intake more realistic for the average adult, nutrient inadequacies would be even greater than this.

One could debate the accuracy of the RDA for certain nutrients, and we note that the RDA is set higher than the average person’s needs (it’s intended to meet the needs of 97.5% of people). Most people could probably eat this diet without developing obvious symptoms of nutrient deficiency. However, we still think the Tier 5 carnivore diet doesn’t look stellar in the nutrient adequacy department.

Other tiers have better or worse nutrient adequacy, depending on which foods are selected. A Tier 1 diet including fruit and squash, for example, would probably fare better in some respects. A Tier 2 diet (muscle meat, salt, and water only) would fare the worst, although we note that The Carnivore Code acknowledges the nutrient inadequacy of this tier (page 227).

Overall (average) score for healthfulness

1 out of 4