The Adverse Effects of Ketogenic Diets on Body Composition: A Review of the Evidence

What is covered in this review:

• Ketogenic diets are often promoted to improve body composition but were found to have a negative effect compared to a minimally processed, very-low-fat, plant-based diet in an inpatient feeding study.
• A meta-analysis of 15 crossover, inpatient feeding studies presented here found that isocaloric substitution of fat with carbohydrate resulted in a 21.6 g/day greater reduction in fat loss (Figure 1).
• Greater fat loss occurred during the lower fat diets even when there was a rapid and sustained decrease in insulin secretion during carbohydrate restriction.
• Inpatient feeding studies and a meta-analysis of 21 randomized controlled experiments demonstrate that minimally processed and fiber-rich sources of carbohydrate also lead to lower energy intake, explaining the discrepancies in studies examining ad libitum intake.
• Large “lifelong” Mendelian randomization studies demonstrate a causal and permanent effect of substituting carbohydrate with dietary fat and greater body fat.
• Clinical and genetic evidence indicates that improved psychological well-being partly mediates the relative effect of dietary carbohydrate on reducing body fat.
• Inpatient feeding studies demonstrate that whole-food plant-based diets reduce body fat compared to minimally processed omnivorous diets, including ketogenic and Mediterranean diets.
• Inpatient feeding studies and a meta-analysis of 13 randomized controlled trials demonstrate that ketogenic diets adversely affect skeletal muscle mass.
• A meta-analysis of 25 controlled trials established that insulin reduces muscle protein breakdown and increases net protein balance, helping explain the beneficial effect of dietary carbohydrate.
• A meta-analysis of 6 controlled trials presented here found that compared to low-fat diets, ketogenic diets reduce total testosterone by a magnitude that would result in hypogonadism in the average man over the age of 26 (Figure 2).
• A meta-analysis of 18 controlled trials found that ketogenic diets adversely affect measures of strength and endurance.
• A meta-analysis of 4 cross-sectional studies presented here found that adherence to a vegan diet is associated with higher maximal aerobic capacity (V̇O2max), considered a gold standard for measuring cardiorespiratory fitness (Figure 3).
• Evidence from >300 randomized controlled trials demonstrate that plant-derived antioxidants improve skeletal muscle mass and recovery, V̇O2max, physical performance, exercise tolerance, endothelial function, and blood pressure.
• Mendelian randomization studies found that higher Apolipoprotein B (ApoB) and LDL are causally associated with skeletal muscle loss and frailty, demonstrating the importance of cardiovascular health on skeletal muscle mass and function.
• Individuals with a ketogenic diet-induced mean LDL of 254 mg/dL were found to have a 4-fold increase in the annual rate of atherosclerotic plaque volume progression relative to individuals with otherwise similar health status who follow a Western diet. 
• Evidence from >1,000 controlled dietary experiments has established that plant-based diets and the associated nutrients reduce atherogenic blood lipids, including ApoB, the primary lipid determinant of atherosclerotic cardiovascular disease.
• Controlled feeding trials demonstrate that animal-based, ketogenic diets significantly increase ApoB, predicting >5-fold increased odds of major coronary events for a lifetime of cumulative exposure.
• A meta-analysis of 10 randomized controlled trials presented here found that reducing dietary saturated fat and cholesterol reduces the odds of cardiovascular disease by 50% per 1 mmol/L reduction in total cholesterol for 5 years of cumulative exposure (Figure 4).
• A meta-analysis of 5 randomized controlled trials presented here found that compared to low-fat diets, ketogenic diets rich in saturated fat impair flow-mediated dilation (artery dilation in response to increased blood flow) by a magnitude at least comparable to that observed for cigarette smoking (Figure 5).
• Hunter-gatherer Inuit developed a genetic mutation that made them resistant to ketosis and consumption of algae helped to prevent Vitamin C deficiency which did occur.
• Hunter-gatherer Inuit generally had lower V̇O2max than active plant-based populations with very high intakes of carbohydrate, and often, very low intakes of protein (Table 1).
• Autopsy and other clinical evidence consistently demonstrate that hunter-gatherer Inuit had a high prevalence of atherosclerotic cardiovascular disease relative to plant-based populations with very high intakes of carbohydrate, including the Okinawans (Table 2).
• Nomadic pastoralist populations that subsisted chiefly on large quantities of grass-fed meat and raw milk had a high prevalence of obesity, atherosclerotic cardiovascular disease, rheumatoid arthritis, gout, habitual constipation, and erectile dysfunction.

 

Ketogenic, carnivore, and other low-carbohydrate diets have become very popular over the years and are often promoted to improve body composition.¹ However, these diets are typically rich in animal protein and saturated fat and have raised concerns due to their adverse effects on blood lipids.² Moreover, these diets were found to have unfavorable effects on body composition compared to a minimally processed, very-low-fat, plant-based diet under inpatient feeding conditions.³ Thus, the effects of ketogenic diets on body composition likely importantly depend on the quality of the comparison diet. This review will explore possible explanations for the apparent discrepancies between studies and the underlying mechanisms by which ketogenic diets may adversely affect body composition.

The overwhelming body of evidence has established that an energy deficit in a manner consistent with the energy balance model is the primary underlying cause by which carbohydrate-restricted diets reduce body fat, refuting the main propositions of the carbohydrate-insulin model of obesity. As reviewed by Hall 2022, body fat storage can occur in the absence of dietary carbohydrate or increases in insulin levels, and satiety can improve in the presence of a higher carbohydrate diet that increases postprandial insulin.⁴ More importantly, Hall 2017 demonstrated in a meta-analysis of 20 controlled isocaloric feeding trials that differed only in carbohydrate and fat intake that lower fat diets resulted in a 16.4 g/day greater reduction in fat loss.⁵ In these studies, total energy ranged from 1% to 83% from carbohydrate, and 4% to 84% from fat. Only one study found a significantly greater rate of fat loss with a lower carbohydrate diet.⁶ This was one of only three studies with a parallel design, the remaining having a crossover design. Thus, this outlier may be explained by inter-individual variability in response introduced by the parallel design rather than a benefit of lower carbohydrate intake. The inclusion of studies of free-living populations may also be a cause for concern, as compliance has often been found to be poor even when all food is provided.⁷

To address the issues of both inter-individual variability and protocol adherence, the data from Hall 2017 was reanalyzed here to determine the rate of body fat loss based on the estimates from inpatient feeding studies with a crossover design. Based on 15 crossover, inpatient feeding studies, lower fat diets resulted in a 21.6 g/day greater reduction in fat loss compared to lower carbohydrate diets (Figure 1). After excluding one study⁸ in which subjects spent up to 50% of the period outside the metabolic ward, lower fat diets resulted in a 20.7 g/day greater reduction in fat loss. Importantly, a greater rate of fat loss occurred during the lower fat diets in all 15 studies, and statistically significantly so in 13.

Figure 1. Meta-analysis of 15 crossover, inpatient feeding studies of isocaloric substitution of carbohydrate (CHO) with fat but constant dietary protein and mean difference in body fat (g/day).

 

Multiple studies in this meta-analysis found a rapid and sustained decrease in insulin secretion during carbohydrate restriction, fully satisfying the experimental conditions required to test the traditional carbohydrate-insulin model. Indeed, despite a decrease in insulin secretion during carbohydrate restriction, one study found a 39.3 g/day greater reduction in fat loss when consuming 71% compared to 29% of energy from carbohydrate, predicting a 3 kg greater reduction in body fat after 6 months.⁹ Similarly, despite a decrease in insulin secretion during 4 weeks of a ketogenic diet, a study partially funded by the Nutrition Sciences Initiative (NuSi), which promotes ketogenic diets found a 15.5 g/day greater reduction in fat loss when consuming 50% compared to 5% of energy from carbohydrate.¹⁰ Interestingly, both studies found a greater reduction in total body weight during carbohydrate restriction due to greater reductions in fat-free mass, demonstrating that total weight loss can be a poor determinant of fat loss during carbohydrate restriction.

The discrepancies between studies comparing the effects of diets differing in carbohydrate quantity on body fat at ad libitum is likely explained by the quality of carbohydrate. A randomized, crossover, inpatient feeding study found that a diet rich in ultra-processed foods led to a higher intake of total energy and energy from both carbohydrate and fat compared to an ad libitum, minimally processed diet naturally richer in fiber, despite meals being matched for energy density, macronutrients, sugar, sodium, and fiber.¹¹ This resulted in a body fat increase of 0.4 kg during the ultra-processed diet and a decrease of 0.3 kg during the unprocessed diet after 2 weeks. Given that ultra-processed foods now contribute to more than 50% of energy of the diets in Western populations,^{12 13} and fast approaching these levels in Asian populations,^{14 15} studies on free-living populations will inevitably be complicated by a high intake of such foods, especially as these are often more affordable.¹⁶ However, even modest improvements in carbohydrate quality may contribute to reduced energy intake. A meta-analysis of 21 randomized controlled experiments found that even supplemental soluble fiber results in reduced energy intake.¹⁷ The importance of carbohydrate quality in terms of the extent of processing was also demonstrated in a randomized, crossover, inpatient feeding study that found that an ad libitum, minimally processed, very-low-fat, plant-based diet with a high glycemic load led to a lower energy intake in all participants and reduced body fat by 35 g/day despite no differences in reported palatability and higher postprandial insulin levels compared to an ad libitum, minimally processed, animal-based, ketogenic diet.³ Thus, evidence from inpatient feeding studies support neither the proposed metabolic advantage nor an enhanced satiety effect of a ketogenic diet.

In addition to the problem of carbohydrate quality and non-compliance, studies on free-living populations are also complicated by inaccuracies in participant reporting. It is well documented that when prescribed low-fat diets, free-living study participants consistently under-report energy intake and over-report adherence.¹⁸ Thus, when prescribed low-fat diets, reductions in blood cholesterol achieved relative to reported intake are often smaller than that predicted by inpatient feeding studies, despite significant weight loss.¹⁹ Interestingly, a Cochrane review involving 57,079 participants from 37 randomized controlled trials that lasted at least six months and were not intended to induce weight loss found that a lower fat intake resulted in lower body weight, BMI, waist circumference, and body fat percentage compared to usual or fat modified diets and that greater weight loss occurred even when there were no differences in energy intake reported between groups.²⁰ Nevertheless, outpatient studies cannot guarantee the accuracy of reported energy intake and therefore cannot fully determine whether or how much weight loss was independent of energy intake.

As it is not feasible to conduct very large and long-term inpatient feeding studies, while not free from certain biases,²¹ another approach that has been argued to address the problem of non-compliance is through the use of Mendelian randomization (MR).²² A MR study with adiposity measurements from 322,154 participants found that a genetically predicted lower intake of carbohydrate and a higher intake of fat as a percentage of total energy is causally associated with a higher BMI and waist circumference.²³ These findings were corroborated by another MR study with adiposity measurements from 776,787 participants that found that a higher total energy-adjusted relative carbohydrate intake is also causally associated with a lower hip circumference, waist-to-hip ratio, and body fat percentage.²⁴ The lifelong nature of MR studies implies that the relative effect of dietary fat on greater body fat is permanent.

A large MR study confirmed the findings from epidemiological studies that depression is a cause of metabolic syndrome and higher waist circumference, indicating that the effects of diet on psychological well-being may partly mediate the effects of diet on adiposity.²⁵ In addition to 71% reduced odds of hypertension, a MR study found that based on psychological well-being measures from 1,796,145 participants, a higher total energy-adjusted relative carbohydrate intake is causally associated with lower levels of depressive symptoms, major depressive disorder, and neuroticism, and higher levels of positive affect and life satisfaction.²⁴ Another MR study found that 74.6% of the effect of carbohydrate on reducing depression was independent of a lower BMI.²⁶ However, these effects may importantly depend on the quality of carbohydrate. A meta-analysis of 18 epidemiological studies found that each 5 g/day increment in dietary fiber intake was associated with 5% lower odds of depression.²⁷ Another meta-analysis of 11 randomized controlled trials found that flavonoids improved symptoms of anxiety and depression.²⁸ Consistent with these findings, a randomized controlled trial found that compared to a fiber-rich, low-fat diet, a calorie-matched, minimally processed, animal-based, ketogenic diet increased anger, confusion, depression, and the total mood disturbance score after one year, despite similar weight loss in both groups.²⁹ A meta-analysis of 8 randomized controlled trials also found that compared to lower-fat diets of non-specific quality, diets that restricted carbohydrate to less than 26% of energy increase anxiety.³⁰

Conversely, the benefits of a fiber-rich plant-based diet on weight loss are well documented. A review of 51 randomized controlled trials of at least 6 weeks found that plant-based diets reduced BMI, body weight, and waist circumference compared to usual diet.³¹ A randomized, crossover, inpatient feeding study with no intention to induce weight loss found that a minimally processed, very-low-fat, vegan diet reduced body fat by 0.48 kg after 2 weeks compared to a minimally processed, animal-based, ketogenic diet.³ Another randomized, inpatient feeding study found that a low-fat, macrobiotic vegan diet reduced BMI (1 kg/m²), body weight (2.7 kg), and waist (1.9 cm) and hip (1.6 cm) circumference after 3 weeks compared to a calorie-matched, but higher protein, healthy Mediterranean diet.³² Similarly, a recent crossover, partial feeding trial with no intention to induce weight loss found that among identical twins, a healthy vegan diet reduced body weight by 1.9 kg after 4 weeks compared to a healthy omnivorous diet.³³ These findings are further corroborated by a meta-analysis of 10 randomized, placebo-controlled trials without energy-restriction protocols that found that soluble fiber meaningfully reduces BMI, body weight, and body fat percentage.³⁴

Ketogenic diets have also raised concerns due to possible adverse effects on skeletal muscle. A randomized, crossover, inpatient feeding study found that a minimally processed, very-low-fat, plant-based diet better preserved skeletal muscle than a minimally processed, animal-based, ketogenic diet despite a 135 kcal/day lower protein intake.³ The finding of a state of negative nitrogen balance only during the ketogenic diet is consistent with numerous other inpatient feeding studies demonstrating that higher carbohydrate diets improve nitrogen balance compared to ketogenic diets.^{35 36 37 38 39} A meta-analysis of 13 randomized controlled trials of up to 12 weeks also found that ketogenic diets resulted in greater fat-free mass loss compared to usual diet and that these effects were not mitigated by the addition of resistance training.⁴⁰ Moreover, multiple placebo-controlled experiments have demonstrated that dietary carbohydrate reduces muscle protein breakdown.^{41 42} This benefit of carbohydrate may be mediated by insulin as indicated by a meta-analysis of 25 controlled experiments which established that insulin infusion reduces muscle protein breakdown and increases net protein balance.⁴³

A recent meta-analysis found that in the absence of significant weight loss, low-carbohydrate diets reduce total testosterone (TT), a risk factor for loss of muscle mass and strength.^{44 45} However, this study did not specifically examine ketogenic diets (≤10% carbohydrate). For this review, the data was reanalyzed here to examine this effect compared to low-fat diets (≤30% fat). Based on 6 controlled trials, ketogenic diets reduced TT by 3.75 nmol/L compared to low-fat diets (Figure 2). This magnitude of TT reduction would result in hypogonadism (TT of <10.4 nmol/L) in the average man over the age of 26.^{46 47}

Figure 2. Effect of ketogenic diets on total testosterone compared to low-fat diets in a meta-analysis of 6 randomized and non-randomized controlled trials.

 

A meta-analysis of 18 randomized and non-randomized controlled trials found that in addition to a greater reduction in fat-free mass, ketogenic diets impair cycling time-trial performance and strength compared to usual diet in athletes and trained adults.⁴⁸ While reductions in TT and muscle mass may explain impaired strength, impaired cycling performance may be mediated by a reduction in maximal aerobic capacity (V̇O2max).⁴⁹ A randomized, crossover, controlled feeding trial found that an animal-based, ketogenic diet decreased absolute V̇O2max by 7.7% and cycling time to exhaustion in all participants, and increased muscle fatigue, and heart rate during exercise after 4 weeks in healthy women compared to a calorie-matched, low-fat diet based on Nordic nutrition recommendations.⁵⁰ Findings from other studies were less consistent, but few were randomized and typically did not consider carbohydrate quality. However, one study found that a ketogenic diet resulted in a 6.6% decrease in absolute V̇O2max compared to baseline after 6 weeks in highly trained men.⁵¹ Another study found that a ketogenic diet resulted in a 10.4% decrease in absolute V̇O2peak compared to baseline after 4 weeks in female athletes.⁵² In a study with a disclosed conflict of interest, the authors reported no adverse effects of a ketogenic diet on V̇O2max compared to a low-fat diet in endurance athletes but only reported estimates relative to body mass.⁵³ However, absolute V̇O2max, as calculable from the body mass estimates, increased by about 7.6% in the low-fat group and decreased by 0.4% in the ketogenic group after 12 weeks. V̇O2max is considered a gold standard for measuring cardiorespiratory fitness and has been implicated as a good predictor of muscle quality.⁵⁴

On the opposite end of the dietary spectrum, two cross-sectional studies found that vegans had higher relative V̇O2max than their omnivorous counterparts.^{55 56} Similarly, a cross-sectional study on endurance athletes found a higher relative V̇O2max among vegetarians, of which more than 50% were vegans.⁵⁷ However, one study found no significant difference between vegans and omnivores, possibly due to an insufficient sample size and the vegan diet group being older.⁵⁸ Absolute V̇O2max was significantly or non-significantly higher in 3 of the 4 studies. In a meta-analysis of 4 cross-sectional studies presented here, adherence to a vegan diet was significantly and positively associated with V̇O2max (mL/kg/min) compared to an omnivorous diet, despite the vegan diet group being older (Figure 3).

Figure 3 Cardiorespiratory fitness measured by V̇O2max (mL/kg/min) and age of vegans compared to omnivores in a meta-analysis of 4 cross-sectional studies.

 

The possible benefits of a plant-based diet are likely in part mediated by a higher intake of antioxidant phytochemicals. A meta-analysis of 11 randomized, placebo-controlled trials found that quercetin, a polyphenolic flavonoid had a small but statistically significant effect on increasing V̇O2max.⁵⁹ A meta-analysis of 20 randomized, placebo-controlled trials found that flavonoid supplementation increases appendicular lean mass and 6-minute walk distance in middle-aged and older adults.⁶⁰ Evidence from 48 randomized, placebo-controlled trials found that flavonoids also improve physical performance, exercise tolerance, and recovery of muscular strength, and reduce muscle soreness.^{61 62} Similarly, a meta-analysis of 12 randomized, placebo-controlled trials found that Vitamin C attenuates the oxidative stress and inflammatory response to exercise.⁶³ Evidence from a further 250 randomized controlled trials also established that soluble fiber and plant-derived antioxidants improve endothelial function and blood pressure, effects that may improve physical performance by reducing the work of the heart during exercise.^{64 65 66 67 68 69 70 71}

Consistent with these findings, prospective cohort studies with up to 1.2 million person-years of follow-up have found that a high intake of healthful plant foods was associated with a lower risk of frailty, including a lower risk of fatigue and a reduced loss of strength and aerobic capacity in older adults.^{72 73 74 75} Epidemiological studies have similarly found that high levels of circulating plant-derived antioxidants, which are surrogate markers of healthful plant food intake are associated with increased skeletal muscle mass, strength, and physical performance.^{76 77 78 79 80 81} There are more than 25,000 known bioactive food constituents, of which the vast majority are antioxidants derived from plants.⁸² Given the established benefits of the oxidative stress-reducing and arterial blood flow-promoting properties of many antioxidants it is reasonable to expect that many of these less well-studied antioxidants found almost exclusively in minimally processed plant foods may have combined additive or synergistic effects on improving muscle quality and physical performance, explaining evidence of a benefit of healthful plant foods.

A recent study found that individuals with a ketogenic diet-induced mean LDL of 254 mg/dL, but otherwise near optimal risk factors experienced a 4-fold increase in the annual rate of atherosclerotic plaque volume progression than observed in mostly healthy individuals with a mean LDL of 112 mg/dL who mainly consume a Western diet.^{83 84} As it is ultimately the cardiovascular system that delivers nutrients and oxygen to every cell in the body, the effects of diet on body composition and physical performance will likely importantly depend on their effects on cardiovascular health.⁸⁵ Indeed, a Mendelian randomization study with muscle atrophy measurements from more than 454,000 participants found that higher Apolipoprotein B (ApoB) and LDL are causally associated with measures of skeletal muscle loss, including lower appendicular lean mass, whole body fat-free mass, and trunk fat-free mass.⁸⁶ MR studies have also found that higher LDL and systolic blood pressure are causally associated with frailty.^{87 88}

As it is well established that cumulative exposure to ApoB-containing lipoproteins and higher blood pressure causes the progression of atherosclerosis throughout life, these findings implicate atherosclerotic arterial obstruction as a probable cause of loss of skeletal muscle mass and function.^{89 90} Thus, while the initial rapid reduction in skeletal muscle observed during carbohydrate restriction may attenuate over time, or be partly mitigated by supplements,⁹¹ the effects of animal-based ketogenic diets on increasing ApoB^{2 3} and blood pressure^{3 24 64} is likely to cause long-term cumulative adverse effects on skeletal muscle. These findings also underscore the limitations of commonly used measures of protein quality, such as the Digestible Indispensable Amino Acid Score (DIAAS), in that these only measure the absorption of amino acids in the digestive tract and not the effects that different foods have on the cardiovascular system, responsible for transporting these nutrients absorbed from the digestive tract to the muscles. Indeed, despite a lower average DIAAS, a study on 85,871 older women with repeated dietary assessments throughout 22 years of follow-up found that substituting 5% of energy from animal protein with plant protein was associated with a 38% reduced risk of frailty, and that plant protein intake was associated with a reduced loss of strength and aerobic capacity, independent of diet quality.⁹² Thus, when protein intake is adequate, measures of protein digestibility may be of secondary importance to improving cardiovascular health in promoting very long-term skeletal muscle mass and function.

Evidence from over 1,000 controlled dietary experiments has unequivocally established that plant-based diets and the associated nutrients reduce atherogenic blood lipids.^{3 32 33 93 94 95 96 97 98 99 100} Specifically, the substitution of animal protein with plant protein,^{101 102} reducing the intake of dietary cholesterol,^{103 104} and increasing the intake of soluble fiber¹⁰⁵ and plant sterols¹⁰⁶ each reduce ApoB by up to 5 to 10 mg/dL, while the substitution of each 1% of energy from saturated fat with carbohydrate, monounsaturated fat, and polyunsaturated fat linearly reduces ApoB by 4 to 10 mg/dL.¹⁰⁷ ApoB, which represents the total number of circulating atherogenic lipoprotein particles has been established by evidence from randomized controlled trials and MR studies as the primary lipid determinant of atherosclerotic cardiovascular disease.^{108 109 110 111 112}

In addition to increasing systolic and diastolic blood pressure, postprandial triglycerides, uric acid, CRP, and heart rate, a randomized, crossover, inpatient feeding study found that a vegetable and nut rich, but otherwise minimally processed, animal-based, ketogenic diet increased ApoB by 20 mg/dL compared to a minimally processed, very-low-fat, vegan diet.³ Similarly, a randomized, crossover, controlled feeding trial found that an animal-based, ketogenic diet increased ApoB by 52 mg/dL compared to a calorie-matched diet based on Nordic nutrition recommendations.² Based on evidence from randomized controlled trials and MR studies, this would predict a 94% increased risk of major adverse cardiovascular events for 4.7 years of cumulative exposure,¹⁰⁸ and 414% greater odds of major coronary events for a lifetime of exposure.⁹⁰ Based on evidence from 104 controlled feeding experiments,¹⁰⁷ the substitution of equal parts of energy from carbohydrate, monounsaturated fat, and polyunsaturated fat with an additional 3% of energy from saturated fat would increase ApoB by a further 22 mg/dL, predicting a 10-fold increased odds of major coronary events for a lifetime of cumulative exposure.

A meta-analysis of prospective cohort studies commissioned by the World Health Organization found that substituting dietary saturated fat with either slowly digestible carbohydrates, plant-derived monounsaturated fat, or polyunsaturated fat was associated with a decreased risk of coronary heart disease.¹¹³ More importantly, a Cochrane review of randomized controlled trials found that reducing dietary saturated fat reduced the risk of cardiovascular disease directly proportional to the absolute reduction in total cholesterol.¹¹⁴ However, estimates of risk were not provided per unit reduction in total cholesterol. Therefore, the data was reanalyzed here to determine this effect using the adjusted odds ratio based on methods described elsewhere.¹¹⁵ In a meta-analysis of 10 randomized controlled trials, reducing dietary saturated fat reduced the odds of cardiovascular disease by 50% per 1 mmol/L reduction in total cholesterol for 5 years of cumulative exposure (Figure 4).

A reduction in dietary cholesterol may explain up to 20% of the reduction in total cholesterol found in the saturated fat reduction trials, and therefore likely contributed to the reduction in cardiovascular disease.¹¹⁶ Importantly, even if it is assumed that almost all of the reduction in total cholesterol observed in these dietary trials was from LDL, this would still indicate a two-fold greater reduction in cardiovascular events per unit lower LDL than observed in statin trials with a similar median follow-up.^{117 118}

Figure 4. Effect of reducing dietary saturated fat and odds of cardiovascular disease per 1 mmol/L reduction in total cholesterol in a meta-analysis of 10 randomized controlled trials involving 53,390 participants and 4,528 cases. RR=0.64 CI, 0.44-0.91 (data not shown). Differences in cardiovascular events and total cholesterol were derived from Analysis 1.35 and Analysis 3.1, respectively, from Hooper 2020.

 

Ketogenic diets have also raised concerns due to possible adverse effects on arterial blood flow. Flow-mediated dilation (FMD), which measures the dilation of an artery in response to increased blood flow is considered a gold standard for measuring vascular endothelial function, with each 1% reduction in FMD observationally associated with a 14% increased risk of future cardiovascular events.¹¹⁹ Lower FMD is also associated with lower V̇O2max and increased physical frailty.^{120 121} A meta-analysis of 90 controlled dietary experiments found that a single high fat meal reduced postprandial FMD, and that meals with greater than 80% fat reduced FMD by 2.77% after 3 hours.¹²² Separately, a randomized, crossover trial found that a very-low-carbohydrate Atkins diet reduced fasting FMD compared to a very-low-fat Ornish diet after 4 weeks and that the reduction was associated with saturated fat intake.¹²³ Moreover, a randomized, crossover, controlled feeding trial found that substituting the equivalent energy from a combination of refined and unrefined sources of carbohydrate, monounsaturated fat, or polyunsaturated fat with 50 g/day butter reduced fasting FMD by 5.41% after 3 weeks.¹²⁴ This is an effect that would predict a 2-fold increased risk of future cardiovascular events.

An earlier meta-analysis of randomized controlled trials of at least 3 weeks found that low-carbohydrate high-fat (LCHF) diets reduced FMD compared to low-fat diets.¹²⁵ For this review an updated meta-analysis was carried out to examine this effect for ketogenic diets that did not limit saturated fat intake. Change-from-baseline standard deviation was calculated using the formula described in the Cochrane Handbook and assumed a correlation coefficient of 0.50.¹²⁶ Based on 5 randomized controlled trials, ketogenic diets rich in saturated fat significantly reduced FMD by 2.61% compared to low-fat diets (Figure 5).^{127 128 129 130 131} In a subgroup analysis in which all food was provided to maximize compliance, ketogenic diets reduced FMD by 3.65%, a reduction that is at least comparable to that observed for cigarette smoking.¹³²

Figure 5. Effect of ketogenic diets high in saturated fat on flow-mediated dilation (%) compared to low-fat diets in a meta-analysis of 5 randomized controlled trials.

 

While there has been some interest in comparing evidence of health and physical performance across populations, it is important to recognize that not even Arctic populations sustained ketosis for extended periods. The hunter-gatherer Inuit were observed to be very resistant to physical ketosis, with ketones typically only detected during starvation, and even then, only in small amounts.¹³³ While the exact reasons are not clear, it has been hypothesized that the deleterious effects of permanent ketosis may have driven the rise of the prevalent genetic mutation in the Inuit that decreases ketogenesis.¹³⁴ Also contrary to popular belief, marine algae was observed to have contributed to about 50% of Vitamin C intake and likely played a critical role in preventing scurvy in the Greenland Inuit.¹³⁵ The high Vitamin C content of algae was not realized until 1933 and was therefore unappreciated by earlier explorers.¹³³ Nevertheless, Vitamin C deficiency and its symptoms have been observed in the hunter-gatherer Inuit, as has riboflavin deficiency.^{135 136}

Interestingly, while the relative V̇O2max of some but not other hunter-gatherer Inuit populations was observed to be moderately high,^{137 138 139 140} these levels were either similar to or lower than that observed in other populations consuming carbohydrate-rich, predominantly plant-based diets, such as the highly active traditional Tarahumara^{141 142} and the horticulture-based Papua New Guinean highland populations.^{140 143 144 145} This difference is particularly apparent when accounting for the linear decrease in V̇O2max with increasing altitude (Table 1).¹⁴⁶ The physical performance and the very lean, but muscular stature of the New Guinean highland populations are intriguing given that they consume among the lowest observed intakes of total and animal protein of any population, and who similar to the traditional Inuit, had high exposure to dwelling smoke, but also tobacco smoke.^{143 144} While the estimated intake as a percentage of total energy for the highland regions described in Table 1 was <1-1.3% and 4.4-6.5% for animal and total protein, and 84-94% for carbohydrate,^{143 144} more extreme intakes had been commonly observed in other regions.¹⁴⁷ Importantly, the physical fitness of both the traditional Tarahumara and New Guinean highland populations remained high in middle age, and total cholesterol levels, blood pressure, and body fat remained low throughout life.^{141 142 145} Moreover, many individuals in these populations demonstrated exceptionally high levels of physical fitness, with one Tarahumara man found to have a V̇O2max of 81 mL/kg/min at 2,300 m, a near, if not world record level at high altitude.¹²⁹  

Also contrary to popular belief, evidence based on clinical examination has consistently demonstrated a high prevalence of atherosclerosis among the Inuit despite a high intake of marine fat (Table 2).^{133 136 148 149 150 151} Bertelsen 1940 stated based on decades of clinical practice and reviewing reports of medical officers in Greenland dating back more than 185 years that “…arteriosclerosis and degeneration of the myocardium are quite common conditions among the Inuit, in particular considering the low mean age of the population.”¹⁴⁸ Høygaard 1941 stated based on X-rays and clinical examinations of the hunter-gatherer Inuit of Greenland, estimated to consume 3% of energy from carbohydrate that “Arteriosclerosis was relatively frequent and occurred at an early age”.¹³³ Similarly, Brown 1951 stated of the Inuit, including the Igloolik hunter-gatherers of northern Canada “We have found well-marked general arteriosclerosis and also coronary heart disease proved by electrocardiogram and, in one case, by post mortem.”¹³⁶ Consistent with these observations, examinations of mummified remains of hunter-gatherer Inuit indicate that atherosclerosis was likely common even among pre-European contact populations, and occurred as early as the second decade of life.^{149 150 151} This greatly contrasts the observed near absence of atherosclerotic disease in the earlier autopsy findings of the predominantly plant-based Papua New Guinean^{145 152 153 154 155} and Okinawan^{156 157} populations, who around the time were observed to consume up to 94%¹⁴³ and 85%¹⁵⁸ of energy from carbohydrate, respectively (Table 2).   

Table 1. Cardiorespiratory fitness measured by V̇O2max (mL/kg/min) in Inuit, Tarahumara, and Papua New Guinean men. *Assumes that V̇O2max (mL/kg/min) is reduced linearly by 5% per 1,000 m in individuals who live and train at high altitude as described by Wehrlin 2016.

Table 2. Clinical evidence of the prevalence of atherosclerotic cardiovascular disease in the Inuit, Papua New Guineans, and Okinawans.

 

In the 1920s, John Boyd Orr, a spokesperson for British Milk Marketing brought the Maasai, a nomadic group that inhabit Kenya and Tanzania, to attention by associating a diet rich in meat and milk to apparent good health. Orr, however, failed to recognize that historically, many Maasai communities were forced to fall back to plant-based diets during the frequent periods of significant livestock loss, and the inevitable influence this would have had on future health.¹⁵⁹ Nevertheless, Orr conceded that 80% of the Maasai reported rheumatoid arthritis and habitual constipation.¹⁵⁹ History repeated itself when research supported in part by the livestock industry revisited the Maasai in the 1960s and failed to sufficiently recognize the nutritional impact of the near-famine state sustained for 4-6 months during the dry season each year, let alone the severe drought and flooding in the Maasailand in the prior decade that resulted in the loss of up to 80% of livestock.^{160 161} Interestingly, an autopsy study of 50 confirmed Maasai men funded by the meat industry found that the extent of atherosclerosis was at least comparable to that observed in U.S. men.¹⁶²

Perhaps more relevant to the sedentary populations with unlimited access to food that ketogenic diets are marketed towards than the observations on highly active, semi-starving populations, may be those on the nomadic pastoralists of Central Asia that subsisted chiefly on large quantities of grass-fed meat and raw milk. In the 1960s the prevalence of coronary heart disease among the nomadic pastoralists of Xinjiang described as subsisting chiefly on sources of animal fat was up to more than 30 times higher than populations in other parts of China where saturated animal fat was scarcely consumed.¹⁶³ A century ago, Maxime Hans Kuczynski examined a population of nomadic pastoralists from the Eurasian Steppe, and attributed a “pure meat-milk-diet” to a high prevalence of obesity, stroke, premature extensive atherosclerosis, peripheral artery disease, kidney disease, gout, habitual constipation, and erectile dysfunction.^{164 165} Kuczynski did not observe such health issues in their neighboring counterparts who consumed carbohydrate-rich, predominantly plant-based diets and retained sexual function into their seventies. These findings mirror the classical observations of Hippocrates who in 400 B.C.E described the nomadic Scythian as subsisting chiefly on meat, milk, and cheese and having high rates of obesity, rheumatoid arthritis, gout, and erectile dysfunction.¹⁶⁶

The effects of ketogenic diets can only be elucidated by the quality of the comparison diet. By deliberately or carelessly overlooking the importance of the quality of the comparison diet, advocates of ketogenic diets have likely only contributed to the public’s confusion surrounding diet and health. Evidence indicates that calorie-for-calorie, dietary carbohydrate has a very modest, but long-term effect on reducing body fat relative to dietary fat and that high-quality sources of carbohydrate lead to lower energy intake. Furthermore, ketogenic diets likely cause the loss of skeletal muscle mass and function, in part due to lower levels of testosterone and insulin, and these adverse effects may be exacerbated when these diets are animal-based due to adverse effects on cardiovascular health and being deficient in antioxidants.