The intent of this analysis was to evaluate the healthfulness of several popular diets and the OmniHeart diets; to do this, we evaluated each diet's consistency with national dietary guidelines. We found that all diets met the AMDR for protein; the Atkins, Mediterranean, Omni-Unsat, and South Beach diets exceeded the AMDR for total fat (40−62%); and the Atkins, South Beach, Mediterranean (9−38%), and Ornish (75%) diets fell outside the AMDR for carbohydrate. Each diet varied with respect to its compliance with national guidelines from the AHA, ACS, and ADA.
Whether or not a diet that falls outside the AMDR for a given macronutrient is necessarily “unhealthy” depends on the food choices that cause a diet to exceed the macronutrient target. A diet that is 30% fat, 50% carbohydrate, and 20% protein might meet the AMDR, but could still contain 15% saturated fat, carbohydrates from refined grains, and high amounts of processed meat (ie, a typical “Western” dietary pattern). However, the Omni-Unsat, Mediterranean, and South Beach diets exceeded the upper AMDR for total fat, but were high-fiber diets low in saturated fats that included ample unsaturated fats, whole grains, fruit, and vegetables. In contrast, the very-low-fat content of the Ornish diet did not meet the minimum fat requirements of national organizations, yet provided healthful sources of protein (eg, legumes and nuts), ample fruit and vegetables, and low amounts of saturated fat. These exceptions indicate that compliance with the AMDR should not be the only criterion by which to assess a diet.
The possible health effects of popular diets, though numerous in theory, have not been tested in controlled feeding studies with hard endpoints (eg, CHD or cancer), likely because large randomized controlled trials of diet are subject to ethical, financial, logistical, and methodologic constraints. Therefore, a discussion of the impact of various macronutrient combinations on chronic disease risk must draw on literature exploring the relation between macronutrients and biologic intermediates (eg, LDL cholesterol). To provide some context for our evaluation of the diets, we discuss and comment on the potential impact of these diets on disease risk in light of the best available evidence for outcomes; and, in the absence of data on hard endpoints, we discuss the effects of the diets and their components on intermediates such as blood lipids, blood pressure, and insulin sensitivity.
The Omni-Protein, South Beach, and Zone diets were higher-protein diets generally consistent with national recommendations for intakes of saturated fat, fiber, and cholesterol; however, several aspects of the higher-protein, higher-fat Atkins approach raise concerns for general health. Unlike the healthful higher-protein diets mentioned above, the Atkins diet severely restricts the intake of fruit, vegetables, and whole grains, which results in a lower fiber intake than all other higher-protein diet. Randomized trials suggest the actual fiber intake with the Atkins diet is likely 10−15 g/d, and participants frequently report constipation while following this diet (8
). This low fiber intake may increase the risk of diverticular diseases (29
). A meta-analysis of controlled trials reported that Atkins-type diets raise LDL-cholesterol concentrations slightly, even in the setting of weight loss (30
), and, the liberal intake of red and processed meats advocated by this diet may increase the risk of colon, rectum (31
), and prostate (34
) cancers. For these reasons, the high-protein food sources associated with an Atkins diet are not prudent for overall health.
The IOM, AHA, ADA, and ACS emphasize balancing physical activity with energy intake to achieve and maintain a healthy body weight (1
). Although we lack long-term controlled studies with large samples comparing diets with different macronutrient contents designed for weight loss, 6 randomized controlled trials conducted between 2003 and 2007 found Atkins-type diets superior to high-carbohydrate, low-fat diets for weight loss at 6 mo (7
) in overweight individuals with or without components of the metabolic syndrome. In these trials, Atkins diets also preserved HDL cholesterol (7
) and reduced triacylglycerols (27
). When follow-up in 4 of these cohorts was extended to 1 y, however, the Atkins weight-loss advantage did not persist, and improvements in serum lipids were attenuated (7
Studies of high-protein diets have consistently reported high dietary adherence early in the trials, when contact between participants and the research team was frequent, and reduced compliance at later time points, when participants were not as closely monitored. Although this “real-world” condition improves generalizability, it likely resulted in smaller than expected contrasts between macronutrient profiles at the end of follow-up (8
). Attrition rates in 6-mo trials were high in Atkins and in low-fat arms (), averaging 8−45% for Atkins and 6−47% for low fat (7
); for 12-mo trials, the dropout rate ranged from 12% to 48% in Atkins and from 23% to 43% in low fat (7
). The proportion of participants that cited a dislike of the diet as the main reason for discontinuation was similar for both diets (8
). Notably, in the one large trial that collected a detailed symptom diary, those in the Atkins trial arm reported more episodes of constipation, headache, halitosis, muscle cramps, diarrhea, and general weakness than did those in the low-fat arm (27
), which may limit the general applicability of such a diet in clinical practice.
FIGURE 4 Comparison between attrition rates in studies of the Atkins diets and in low-fat diets. Although early studies compared Atkins diets only to low-fat diets, later studies included other popular diets (7, 8, 27, 28, 35–40). For studies that included (more ...)
Evidence is emerging that higher-protein diets, even in the absence of weight loss, may be beneficial for health. A large prospective cohort study with 20 y of follow-up found low-carbohydrate, higher protein diets compared with low-fat diets are not associated with increased risk of CHD in women, and diets rich in vegetable protein and fat were found to be protective against CHD (41
). Vegetable protein, but not animal protein, was found to have an inverse relation with blood pressure in the INTERMAP study (42
), consistent with the blood pressure–lowering effect of the Omni-Protein diet. Compared with a typical diet, all Omni diets lowered LDL cholesterol, but the Omni-Protein diet showed a greater reduction than did the Omni-Carb diet, and the triacylglycerol-lowering effect was stronger for protein than for unsaturated fat, which suggested that a specific beneficial effect of protein on lipid risk factors (21
). In addition to these benefits, higher-protein diets may offer better protection against the progression of insulin resistance to type 2 diabetes (43
), by reducing circulating free fatty acid concentrations, than higher-fat, low-carbohydrate diets (44
Whether a higher protein intake (26−35% of energy) is un-healthful remains controversial. The upper limit of the AMDR was set to mitigate against osteoporosis, kidney stones, and renal failure that might accompany chronically high protein intakes (45
). Two small 12-wk intervention trials of high-protein diets in healthy adults showed either an adverse effect (46
) or no effect (47
) on markers of bone turnover or risk of nephrolithiasis, and other trials suggested beneficial effects of a high protein intake on bone health (48
). Observational studies in the general population did not find a relation between a high animal protein content in the diet and the development of renal stones (50
) and suggest that high-protein diets may only be detrimental in those with existing renal insufficiency (52
). This may be of concern to individuals with hypertension or diabetes, because they may have impaired renal function. Further research is needed to better understand the potential adverse effects of chronic high protein intakes.
Although the Omni-Unsat diet exceeded the IOM's target for total fat, it remained within recommended intakes of carbohydrate, saturated fat, cholesterol, and fiber. Furthermore, this diet was rich in whole grains, fruit, and vegetables and thus might be expected to improve CVD risk factors, as it indeed did, and to possibly provide protection against several cancers and type 2 diabetes. Other higher-fat diets, such as the South Beach and the Mediterranean diets, had a slightly higher total fat content than did the Omni-Unsat diet (40−45% of energy), and all of these diets exceeded the AMDR upper limit for total fat, but, similarly to the Omni-Unsat diet, were low in saturated fat and cholesterol because of the inclusion of oils and nuts rich in MUFAs. The South Beach and Mediterranean diets also encourage the consumption of whole-grains, which enhances the potential health benefit. In contrast, the Atkins diet provided considerably more fat (62%) than did the other higher-fat diets, provided excessive amounts of saturated fat (23%) and cholesterol (731 mg/d), and provided carbohydrate and fiber in amounts below the RDA. The different fatty acid profiles of these diets are also important determinants of their healthfulness and likely are of much greater importance than the total fat content.
Despite their claim that insufficient data were available to determine a defined total fat intake at which there is a risk of chronic disease (25
), the IOM arrived at an upper limit of 35% for total fat intake largely on the basis of 2 theoretical concerns—the risk of excessive energy intake and weight gain associated with very-high-fat diets and the correlation of saturated fat intake with total fat intake (1
). These lines of reasoning are controversial (53
With respect to the first concern, although fat is the most energy-dense nutrient, in 2 small randomized trials of weight loss, Mediterranean-style diets that exceeded 35% of energy from fat, were high in MUFAs, low in saturated fat, and rich in fruit, vegetables, legumes, whole grains, and fish, resulted in greater weight loss and more improvement in cardiovascular disease (CVD) risk factors than did low-fat, high-carbohydrate diets (11
). Furthermore, in the classic Seven Countries Study, the population of Crete had the lowest rates of CVD and cancer, despite a 43% total fat intake. The Cretan diet was low in saturated fat (<8%), high in MUFAs (27%) (55
), and rich in fruit, nuts, legumes, and meat of grazing animals, which contained n–3 fats.
The Women's Health Initiative found that a 29%-fat diet was not superior to a 38%-fat diet for long-term weight maintenance in postmenopausal women (56
), nor did the lower-fat diet significantly reduce the risk of CHD, stroke, CVD (57
), or several cancers (58
) A major criticism of the study is that the targeted fat reduction of 20% was not met, which calls into question the sustainability of a low-fat approach to reduce the risk of heart disease and cancers.
Second, a diet high in total fat that is low in saturated fat and does not lead to weight gain (or promotes weight loss in the setting of a caloric deficit) does not appear to contribute to the development of insulin resistance, type 2 diabetes, or certain forms of cancer. Although early epidemiologic studies have linked high-fat diets to insulin resistance (61
) and cancer, these correlations were confounded by obesity (63
), a lack of physical activity (64
), and other dietary factors. MUFAs do not appear to increase the risk of type 2 diabetes (65
), and, when substituted for high-glycemic index foods, high-MUFA foods improve the postprandial glucose and insulin response (66
). An inverse association between MUFA (67
) or olive oil (68
) intake and breast cancer has been shown in several studies; others showed no increased risk with increasing MUFA consumption (69
). Similar results were found for diets high in MUFA-rich vegetable oils and prostate cancer (70
). These findings suggest that the benefits of a high MUFA intake as part of an Omni-Unsat diet or a Mediterranean dietary pattern may outweigh concerns related to a higher total fat intake.
The Omni-Carb diet conformed to the IOM's AMDR and national guidelines in all respects, providing 54% of energy from carbohydrate and acceptable amounts of saturated fat, cholesterol, and fiber. However, replacement of 10% of the energy from carbohydrate with either protein or unsaturated fat reduced blood pressure, triglycerides, and non-HDL cholesterol in the OmniHeart trial (21
). The AMDR for carbohydrate took into consideration the adverse effects on triglycerides and HDL cholesterol seen when higher-carbohydrate, low-fat diets are adopted by a sedentary population (72
). An emerging concern is that low-fat diets may not be effective for weight management (56
). However, these effects may be avoided if low-glycemic-index carbohydrates are the predominant foods in the diet (1
) and individuals adopt an active lifestyle.
The Ornish diet exceeded the upper limit of the AMDR for carbohydrates (75%); however, it was high in fiber and provided generous amounts of low-glycemic-index whole grains, such as brown rice, buckwheat, and whole-grain flour. These are choices consistent with AHA guidelines. The most impressive support for the health benefits of this diet come from the Lifestyle Heart Trial (12
), which randomly assigned patients with angio-graphically documented CHD to either usual care or an intensive lifestyle change that included a very-low-fat, vegetarian diet; smoking cessation; and an exercise and meditation program. Patients in the treatment arm (n
= 22) were highly compliant with the lifestyle changes, lost 20 lb (≈9 kg), and reduced LDL cholesterol by 40% after 1 y (73
). These patients had a significant regression in atherosclerosis and had less than half as many cardiac events than did the usual care group at 5 y (12
). However, because diet was not the only lifestyle factor modified, it is difficult to ascribe these benefits solely to diet. Subsequent trials found a less impressive reduction in risk factors when larger groups of less motivated individuals free of CHD adopted the same diet principles without intensive lifestyle modification (8
). Furthermore, very-low-fat diets may result in suboptimal concentrations of essential long-chain polyunsaturated fatty acids if food products are not carefully chosen (74
). In addition, such a high-carbohydrate, low-fat diet may have both health-promoting (eg, reductions in LDL cholesterol and blood pressure) and possibly harmful (eg, reductions in HDL cholesterol and increases in triglycerides) effects.
Implications for practice
Although the optimal macronutrient profile of the diet remains unknown, data are emerging that compliance, rather than a specific macronutrient distribution, may be the most important determinant of whether or not a diet achieves its intended effect (eg, weight loss and chronic disease risk reduction) (8
). To maximize adherence to any diet plan, it is imperative to consider an individual's food preferences, culture, lifestyle, and religious beliefs. Advocating a single dietary pattern to reduce chronic disease risk for a diverse population is unlikely to be an effective approach to promote health.
To this aim, the OmniHeart trial provided 2 alternatives to the conventional low-fat, high-carbohydrate DASH approach to maintaining health and reducing CVD risk. The OmniHeart trial showed that a modest redistribution in macronutrient composition from carbohydrate to protein or unsaturated fat can reduce the CVD risk according to the Framingham criteria by 16% to 21% (21
). Of great interest is that these modifications in CVD risk were achieved in overweight subjects who did not lose weight. Both Omni-Protein and Omni-Unsat diets may be more sustainable than a high-carbohydrate diet and appear to have health advantages over extreme diets, such as the Atkins (which may increase the risk of CVD, cancer, and diabetes) and the Ornish (which may lower HDL cholesterol, increase triglycerides, and have inadequate polyunsaturated fatty acids) diets.
All 3 OmniHeart diets provided adequate amounts of macronutrients. In practice, the Zone, South Beach, and Mediterranean diets are all generally acceptable options that might be helpful for individuals looking for similar benefits for heart disease, cancer, and diabetes risk. These diets replace carbohydrate with protein and/or fat, are similar in concept to the OmniHeart trial, and may also have several salutary effects on chronic disease risk (9
Implications for research and policy
This article raises several important issues that are pertinent to research and policy recommendations on macronutrient intake. First, although the IOM recommends a wide range of macronutrient intakes, termed the AMDR, it is likely that some portion of this range is optimal. For instance, in the OmniHeart trial, a carbohydrate intake of 48% of energy was associated with a greater improvement in CVD risk factors than was a carbohydrate intake of 58% of energy, although both values fell within the AMDR for carbohydrate (21
). Second, the AMDR limits should be reconsidered. The percentage of energy from fat tested in the Omni-Unsat arm (37%) exceeds the AMDR upper limit for fat of 35%. Third, considerable research is needed to understand the optimal sources and types of macronutrients. For example, types (eg, fructose) and food sources (eg, low-glycemic-index foods) of carbohydrate may have an even greater impact on health than the absolute amount of carbohydrate. Fourth, isocaloric feeding studies with 3 or more arms are an especially powerful type of study design to test the effects of diets with different macronutrient profiles.
As an isocaloric study, OmniHeart tested the effects of the diets without the confounding effects of weight loss. Also, the 3-group design made it possible to draw inferences about the macronutrient responsible for the observed effects. For instance, the Omni-Protein diet lowered blood pressure compared with the Omni-Carb diet (21
). Without a third arm, it would be difficult to attribute the blood pressure reduction to lower carbohydrate or increased protein. However, the concurrent finding that the Omni-Unsat diet relative to the Omni-Carb diet also lowered blood pressure provides a parsimonious explanation—lower carbohydrate, rather than increased protein or unsaturated fat, reduced blood pressure. The public health implications of these findings warrant further attention.