Title:

The Ideal Diet for Humans to Sustainably Feed 10 Billion Healthy People – Review, Meta-Analyses, and Policies for Change

Author:

Dr. Galit Goldfarb PhD 

Affiliation:

Dept. of Nutrition. OUS University, The Royal Academy of Economics and Technology, Zürich, Switzerland.

Disclaimers: I, Galit Goldfarb do hereby solemnly declare that this article submitted to your journal is my original piece of work and is not under consideration for publication elsewhere.

Source(s) of support: None

Word count manuscript excluding abstract and supplementary data. 4569 Words 

Word count abstract. 300 Words

Conflict of Interest declaration. No conflict of interest

Number of Figures. 14

Number of Tables. 7

Abstract

INTRODUCTION:

As of now, no study has combined research from different sciences to determine the most suitable diet for humans. This issue is urgent due to the predicted population growth, the deterioration of human health with age and associated costs, and the effect on the environment. 

METHODS:

A literature review determined whether an optimal diet for humans exists and what such a diet is, followed by six meta-analyses. The standard criteria for conducting meta-analyses of observational studies were followed. A review of literature reporting Hazard Ratios with a 95% confidence interval for red meat intake, dairy intake, plant-based diet, fiber intake, and serum IGF-1 levels were extracted to calculate effect sizes.

RESULTS:

Results calculated using NCSS software show that high meat consumption increases mortality probability by 18% on average, plant-based and high-fiber diets decrease mortality by 15% and 20% respectively (p < .001). Plant-based diets decreased diabetes risk by 27%, and dairy consumption (measured by increased IGF-1 levels) increased cancer probability by 48% (p < 0.01). On the other hand, a vegetarian or Mediterranean diet was not found to decrease heart disease probability. A vegetarian diet can be healthy or not, depending on the foods consumed. A Mediterranean diet with high quantities of meat and dairy products will not produce the health effects desired. The main limitations of the study were that observational studies were heterogeneous and limited by potential confounders. 

DISCUSSION:

The literature and meta-analyses point to an optimal diet for humans that has followed our species from the beginnings of mankind, allowing our species to become the most developed on Earth. The optimal diet is a whole food, high fiber, low-fat, 90+% plant-based diet.

To ensure people’s nutritional needs are met healthily and sustainably, governmental dietary interventions are necessary.

Introduction

Research shows that 95% of the global population is living in poor health. (1-3) The obesity epidemic has led to higher incidences of non-communicable diseases (NCDs) including heart disease, type 2 diabetes, and specific cancers. (4-9)

Currently, individuals are confused as to what diet is ideal for promoting health and longevity.

Individuals need to have a set standard for healthy food based on human needs, to enable them to improve health and require less healthcare in the future.

An extensive literature review combining data from different scientific fields helped determine underlying factors relating to the ideal diet for humans. A look at human evolutionary research was necessary for a thorough approach to this subject. 

The literature review points to a major change within the hominin species occurring about 5 million years ago with the introduction of bipedalism. 

This change came to hominin advantage about 2 million years ago, following a global change in weather conditions when the earth entered the Ice Age.

The North African tropical rain forest began receding and the savanna grasslands expanded. Bipedalism gave hominin’s the ability to leave the rainforest where food resources were dwindling as a result of the ice age, for the growing savannas. (10)  Skull fossils show that hominins living in the rainforest had a brain size of 320-380 cm³ in volume. (11-13)

Upon moving to the savanna grasslands, hominins needed to change dietary practices to survive. Dental microwear and stable isotope analysis show evidence of C4 resources, mainly underground storage organs (USOs), as a significant factor in hominin diets on the savanna. (14–22)

The next hominin fossils demonstrate a steady process of brain growth. 

The dry savanna habitat was also rich in sedge and grass grains, rich in carbohydrates, perfect to nutritionally support hominin brain growth. (20-28)

A larger brain requires more energy to fuel. It is a metabolically expensive organ and therefore requires a stable, high-energy, nutrient-dense food source to support its growth under natural selection. (16, 20, 29)

Fossils of the next ancestor of the genus Homo, Homo ergaster, who lived exclusively on the savanna, show a brain 50% larger than their predecessor. (30, 31) The larger more advanced brain allowed Homo ergaster to thrive in the hostile savanna environment. Homo ergaster had a jaw and tooth size closely resembling that of modern humans, and fossils show that their digestive tract grew smaller.

Due to the vast amount of animals on the savanna, it would seem obvious to suggest that the change in diet was towards a meat-based diet, however, to be dependent on meat as a stable food source on the savanna, hominins would have needed to hunt big game that roamed the savanna. To hunt successfully, hominins would have needed to master the skills of hunting. Research suggests that hunting skills have a very long learning curve. It is suggested that this took about one million years to master. (32-34)

Evidence also suggests that big game roaming the savanna was a poor food source for hominins. Their meat was lean, with almost no fat, and the protein levels were too high to consume in abundance, rendering them an inefficient source of calories.

Dietary protein, when consumed in excess, becomes toxic to the human body. Humans can metabolize about 250 grams of protein per day; exceeding this level produces toxic waste that the body has difficulty eliminating. Furthermore, some of the protein consumed is required for cellular growth and repair and would not be available for energy. In fact, 250g of protein provides only 1000 calories, not nearly enough to sustain a modern-day sedentary adult, let alone an active hunter-gatherer on the harsh African savanna. It is well documented that those who consumed excessive amounts of lean protein, without sufficient fat, developed a condition called “mal de caribou,” (35) whereby ammonia builds up in the blood. If this diet persists, the person will suffer from diarrhea and mineral losses and will eventually die. (36-39)

Research shows that modern hunter-gatherer groups such as the Hadza and San, with modern sized brains, fail to catch meat on 97% of their hunts and share the meat mainly with their co-hunters. Furthermore, hunting is usually practiced when staple foods are available in abundance. (40-46) It is suggested that hunting occurred primarily for rituals and shows of courage rather than solely for dietary necessity. (46-48) 

In 1993, more than 1,000 researchers participated in a study titled The Lost Crops of Africa, examining Africa’s ancient crops. The report comes to a clear conclusion: “Grass seeds have sustained humans throughout time.” (49-50)

In 1984, anthropologists found the nearly complete skeleton of a Homo ergaster child assumed to be 1.7 million years old. The skeleton referred to as Nariokotome Boy, (51-52) had a brain size of 880 cm³ in volume. In the fossil remains we see evidence from the shape of the ribcage, that at this stage, hominin guts shrunk to the size of a modern human gut. This shrinkage of gut size allowed the available energy to be directed to feed the growing brain.

The brain is an incredibly energy-expensive organ using 22% of human basal metabolic energy. To accommodate the increase in brain size, humanity’s forebears needed a good, high-quality, readily available food source. (53)

Such food sources grew readily all year round on the savanna and included legumes, grains, plants with USOs, seeds, and fruits. (54-76)

The switch to starchy foods, very different from the dietary habits in the rainforests, allowed early humans to thrive on the savanna with shorter guts, which required less energy to maintain. The excess energy became available for brain tissue growth (77) because hominin survival depended on intelligence.

Enzyme inhibitors in plants stop enzymatic reactions. As a result, enzyme inhibitors can have an anti-nutritional effect. (77-79)

Cooking deactivates these anti-nutrients. But when cooking was still unavailable, anti-nutrients are easily deactivated by simple soaking in water (79-80, 61), causing enzyme inhibitors to stop functioning. Also, an adapted gut microbiota helps the breakdown of such starches. (79) Furthermore, young grass grains do not have enzyme inhibitors. 

The probable low bio-accessibility of nutrients from legumes and grains is less relevant due to the wealth of legumes, grains, and tubers available on the savanna. (79-83)

By contrast, USOs have a physical defense mechanism by being located underground or covered in a thick outer layer. USO-bearing plants are edible in raw form. (74) 

Fossil dental calculus, accepted as a significant pool of dietary data, shows that Neanderthals 400,000 years ago ate a wide variety of plant matter, especially USOs and grass seeds. Neither geographic region, species, nor known stone tool technology significantly impacted the number of plant species consumed. (84) Fossilized Neanderthal feces were also found to have large amounts of plant matter. (85) Other research has also revealed that plant foods had a leading role in early human diets. (84, 86-91) In later fossils, evidence of damaged grass seeds in dental calculus is a sign of cooking. (84, 92-94)

The control of fire by hominins began sometime between 400,000-700,000 years ago. At this time, another major brain growth spurt in hominin fossils is observed (from 800 cm³ to 1,100 cm³ in volume). (86, 95-100) Fire enabled a reduction in the need for chewing and detoxification of anti-nutrients, making more energy and nutrients available for the brain. (101-102)

Approximately 195,000 (±100,000) years ago, the first evidence of Homo sapiens (modern humans) appeared and replaced other Homo species in Africa. Omo, the oldest fossil remains of modern humans, show that they had the same anatomical build as we have today. (103, 104)

12,000 years ago, agriculture began at the geographic corridor through which humans left Africa. The first foods chosen to be grown through agriculture were grains, emphasizing their significance for humans during the hunting-gathering period.

Grains and legumes were easily domesticated from their wild ancestors because they required very little genetic change to domesticate. (105)

Animals were domesticated 6000 years ago only in a few areas on earth, principally in Asia and Europe. Animals were rarely domesticated in America. Animals were not domesticated in tropical Africa or Australia. (106-108) This lack of domestication is probably due to the abundance and variety of grains, legumes, and USOs found in these places, reducing the need to domesticate animals for food.

The domestication of animals for food resulted in increased meat consumption beyond previous consumption patterns of hunter-gatherers who ate meat sparsely when available. (109-112)

Human brain size decreased after the dawn of agriculture. (112-113)

When agriculture was introduced, life expectancy dropped to 20 years because human eating habits changed dramatically. (112, 114) Only certain crops were grown, exposing farmers to much risk and leading people to suffer from severe nutrient deficiencies, shortening lifespans. (115-117, 105, 112)

After the industrial revolution, life expectancy increased, but life’s quality did not necessarily follow suit. (118-124)

The awe for meat and processed grain consumption increased with the industrial revolution when food processing became popular for storage and transport. Grains that could originally supply a wealth of nutrients were stripped of their healthy bran and germ layers, becoming nutritionally deplete. This move to processed grains caused large populations to develop nutrient deficiencies, including protein deficiency, which led to the discovery of a disease named Kwashiorkor.  Kwashiorkor was caused by consumption of nutrient and protein depleted dried grains that were ground into flour for children as food. Kwashiorkor sometimes healed with animal protein consumption. (125, 126)

However, nutrient and protein shortage was never a problem for humans until grain processing began. (127-130)

A whole-food, mostly plant-based diet, from varied plant sources, as ancient humans consumed, easily incorporates all of the nine essential amino acids for humans.

Protein content in human breast milk is lowest in comparison with other lactating mammals and is higher in carbohydrates and mono- and polyunsaturated fats. (131, 132)

This suggests that although protein is necessary for human health, consuming large quantities of protein carries a considerable price on human health.

In recent decades, the world has seen a rise in NCDs such as heart disease, cancer, and diabetes. Seven out of every ten deaths are due to NCDs. (133)

Nowadays, domesticated animals are rich in fat. The fatty deposits among muscle fibers soften the cooked meat and improve its flavor. 

The fat composition of domesticated meat has also changed over time. Previously, animal fat had equal amounts of omega-6 and omega-3 fatty acids (FAs). Nowadays, animal fat is rich in inflammatory omega-6 FAs and low in health-promoting omega-3 FAs because of the intensive rearing methods. (134, 135) The levels of omega-3 FAs in meats 100 years ago were 170 mg/100 g of meat. Now they are 20 mg/100 g of meat. (134, 136) 

Animal milk is also less suitable for human consumption. Milking animals only began around 6000 years ago. (137)

Most of the current world population (75%) is lactose intolerant, leading to side effects such as mineral losses, diarrhea, cramping, bloating, and gas. 

Although calcium is found at high levels in dairy foods, for calcium to be efficiently absorbed, a relatively equal amount of magnesium must be present in the diet. Cow’s milk contains only traces of magnesium. Therefore, it is common for only about 25% of dairy calcium from milk to be absorbed. The remaining unabsorbed 75% may end up deposited around the body, leading to atherosclerosis, gout, and kidney stones.  

Methods

A will to bring to the table valid results that prove the literature review formed the reasoning behind the following meta-analyses since meta-analysis is considered by many to be the platinum standard of evidence.

A look at mortality statistics for people following different diets as well as a look at different dietary patterns and the most common diseases in the world today (cancer, heart disease, and diabetes, were examined through a collection of studies performed in the last decade only. The results were calculated using NCSS software. Furthermore, other meta-analyses that were performed in this time period were used as a comparison between results received with the meta-analysis results received in this research.

Search Strategy

Six different meta-analyses combining the results of multiple studies were performed to support the argument that this ideal diet for humans can also prove optimal in our day and age. Each meta-analysis had its own H1. The probability factor (p=value) was calculated to see if the null hypothesis was rejected or not. If the probability was small (less than 5% or less than 1 in 20 chance of being wrong), then the null hypothesis was rejected, and I could safely conclude that there was a connection between the independent explanatory variable and the dependent variable.

The standard criteria for conducting and reporting meta-analyses of observational studies were followed (Stroup et al., 2000). Studies were identified through a systematic review of the literature (through December 2018) by using the PubMed database (http://www.ncbi.nlm.nih.gov/pubmed) and Google Scholar. The search terms used for all studies included “meat,” “beef,” “pork,” “veal,” “lamb,” “steak,” “hamburger,” “ham,” “bacon,” or “sausage,” “IGF-1,” “dairy,” “milk,” “Plant-based,” “vegan,” “vegetarian,” “fiber,” “Mediterranean diet,” in combination with “mortality,” “death,” “heart disease,” “diabetes,” and “cancer.” In addition, the reference lists of relevant publications were also searched for more studies.

Study Selection

Prospective studies performed only in the last ten years (as required by OUS university) that reported relative risks and Hazard Ratios with 95% confidence intervals for the associations of unprocessed red meat, processed meat, total red meat consumption, dairy, high fiber, plant-based, vegan, vegetarian, with all-cause mortality, diabetes, cancer, and heart disease were included. Furthermore, meta-analyses that were performed in the last ten years but also included studies performed beforehand were added to each meta-analysis performed for every subject so that the aggregated HR derived from the analysis could be compared, and thus, the results would have a more holistic nature.

Data Extraction

From each publication, the first author’s last name, year of publication, study location, sex, age, sample size (total number of participants and number of deaths or disease), relative risks, and Hazard Ratios with a 95% confidence interval for each category of red meat intake, dairy intake, plant-based diet, fiber intake, serum IGF-1 levels, and covariates adjusted for in the analysis were extracted.

Data Analysis and Statistical Methods

The meta-analysis’s primary goal was to compute the aggregative effect of specific food group consumption on mortality/disease, taking into consideration standalone and heterogeneous results. To examine the aggregate effect size, individual studies were gathered estimating mortality of people consuming a specific food group (e.g., red meat/high-fiber/plant-based) compared with people who do not consume this specific food group. Each individual study calculated Hazard Ratio (HR), the event rate corresponding to the conditions (dead/alive; disease/no disease) described by two levels of an explanatory variable (red meat vs. no red meat; vegetarian vs. non-vegetarian; high fiber vs. low fiber).

The meta-analysis’s main function was to estimate effects in the population by combining the effect sizes from a variety of studies. Specifically, the estimate is a weighted mean of the effect sizes. The ‘weight’ that is used is usually a value reflecting the sampling accuracy of the effect size, which is typically a function of sample size. The meta-analysis’s final goal was to determine the aggregative effect size beyond all effects that were gathered, its significance, 95% confidence interval, and possible moderators for the results (variables that could explain non-random variance between effects).

For the final effect, the HR as an effect-size estimate; a confidence interval (lower limit [LL] and upper limit [UL]); Q statistic; and its p-value were reported. Q, a chi-square statistic, reflects variability among effect estimates due to true heterogeneity rather than sampling error. The null hypothesis is that all studies used to calculate each effect shared the same effect size. Under the null hypothesis, Q should follow a central chi-square distribution with degrees of freedom equal to k ? 1. When the p-value is less than .05, the null hypothesis is rejected, and it can be concluded that there is true variance in the studies’ common effect size (Borenstein, Hedges, Higgins, & Rothstein, 2009).

The random-effects model, which assumes that variance between effects is basically random, was applied, and therefore any variance was not attributed to specific moderators.  Forest plots, which depict the effects on a single figure, and the aggregated effect and confidence interval were produced.

Six separate meta-analyses were performed:

  1. Meat consumption and mortality
  2. Plant-based nutrition and mortality
  3. High fiber diet and mortality
  4. Plant-based nutrition and diabetes 
  5. vegetarian/Mediterranean diet and heart disease
  6. IGF-1 in dairy products and cancer

The results were calculated using NCSS software, which has a statistical package for calculating aggregated hazard ratio (HR). The input included individual HR from each study, the variance for each HR (calculated as SE2), and SE (a standard error that was calculated manually using CI with the following calculation:

SE= HR-lower border of CI/2

Outputs included the following figures:

1. A forest plot that shows individual HR and its CI, in addition to aggregated HR and CI. 

2. A radial plot which shows the study bias for aggregated HR according to heterogeneity.

The results are as follows:

4.1. Meat Consumption and Mortality 

To assess the aggregative Hazard Ratio (HR) of meat consumption and mortality, two individual studies were used, yielding three effect sizes (see Table 4.1).

Table 4.1: Individual studies evaluating HR between meat consumption and mortality

Study Total sample  Total Death cases Follow-up (years) HR HR 95% CI – Lower HR 95% CI – Upper Weight in meta-analysis
  1. Pan et al., 2012
121,342 23,926 28 1.13 1.07 1.2 36.11
  1. Sinha et al., 2009 (men)
322,263 47,976 10 1.22 1.16 1.29 36.11
  1. Sinha et al., 2009 (women)
223,390 23,276 10 1.20 1.12 1.3 27.79
Total 666,995 95178 1.18
Meta-analyses
  1. Wang et al., 2015
1,493,646 150,328 1.15 1.11 1.18
  1. Larsson & Orsini, 2013
1,320,980 135,601 1.10 0.98 1.22

Note: HR was calculated for meat consumption vs. non-meat consumption. 

Results of meta-analysis (n = 666,995 xxx) showed that the aggregated effect size between meat consumption and mortality is HR = 1.18 (HR S.E. = 0.03, 95% CI [1.12,1.24]). This result is significant ?2 (DF=2) = 3737.16, p < .001 and means that people consuming meat at any time point during the study period were 18% more likely to die than people that were not consuming meat, and we are 95% confident that people consuming meat are between 12% and 24% more likely to die at any given age than people not consuming meat.

No heterogeneity was found between studies, meaning there are no potential moderators that could bias this effect, Q (2) = 3.80, p = 0.09. Hence, differences between individual studies are not significant and considered homogeneous (see Figure 2, all studies are within the CI borders).

To conclude, consuming red meat significantly increases death probability by about 20% on average in comparison with not consuming red meat. This effect size is larger compared with the effect size received by Wang et al. (2015) (HR=1.15). It is also important to note that in a similar meta-analysis conducted by Larsson & Orsini (2013), no significant, consistent effect was found between meat consumption and mortality.

Figure 4.1: Forest plot of HR between meat consumption and mortality 

pastedGraphic.png

Here we can see that the confidence intervals for studies overlap.

Figure 4.2: Radial plot of HR between meat consumption and mortalitypastedGraphic_1.png

4.2. Plant-Based Nutrition and Mortality 

To assess the aggregative Hazard Ratio (HR) of plant-based nutrition and mortality, four individual studies were used, yielding four effect sizes (see Table 5.2).

Table 4.2: Individual studies evaluating HR between plant-based nutrition and mortality

Study Total sample  Total Death cases Follow-up (years) HR HR 95% CI – Lower HR 95% CI – Upper Weight in meta-analysis
  1. Orlich et al., 2013
96,469 2,570 5 0.88 0.80 0.97 25.09
  1. Key et al., 1999
76,172 8,330 10 0.76 0.62 0.94 18.05
  1. Fraser, 1999
34,192 6 0.80 0.74 0.87 27.37
  1. Kim, Caulfield, & Rebholz, 2018
11,879 2,228 6 0.95 0.91 0.98 29.47

Note: HR was calculated for plant-based nutrition vs. non-plant-based nutrition. 

Results of meta-analysis (n= 218,712 people), showed that the aggregated effect size between plant-based nutrition and mortality is HR = 0.85 (HR S.E. = 0.04, 95% CI [0.77,0.94]). This result is significant ?2 (4) = 3497.7, p < .001 and means that people consuming a plant-based diet at any time point during the study periods were 15% less likely to die than people that were not consuming a plant-based diet, and we are 95% confident that the true value is lying between 6%-23%. (we are 95% sure that people consuming a plant-based diet are between 6% and 23% less likely to die at any period of time than people not consuming a plant-based diet).

Significant heterogeneity was found between studies, meaning that the studies included in this analysis were different by several methodological aspects, which could bias the aggregated HR effect, Q (3) = 20.70, p < 0.01. Differences between individual studies are significant and considered heterogeneous (see Figure 5.4, the result of Kim, Caulfield, & Rebholz, 2018 exceeds CI borders), in this study among Seventh-day Adventists, vegetarians were healthier than non-vegetarians, but this cannot be ascribed only to the absence of meat.

To conclude, plant-based nutrition significantly decreases death probability by about 15% on average, in comparison with non-plant-based nutrition.

Figure 4.3: Forrest plot of HR between plant-based nutrition and mortality 

pastedGraphic_2.png

Here we can see that the confidence intervals for studies overlap.

Figure 4.4: Radial plot of HR between plant-based nutrition and mortality 

pastedGraphic_3.png

4.3. High Fiber Diet and Mortality 

Six individual studies were used to assess the aggregative Hazard Ratio (HR) of a high fiber diet and mortality, yielding six effect sizes. In addition, two meta-analyses were gathered in order to compare aggregated HR derived from our analysis. These meta-analyses were documented in order to compare results between independent meta-analyses and published meta-analyses (see Table 4.3).

Table 4.3: Individual studies evaluating HR between fiber diet and mortality

Study Total sample  Total Death cases Follow-up (years) HR HR 95% CI – Lower HR 95% CI – Upper Weight in meta-analysis
  1. Park et al., 2011
567,169 31,456 9 0.78 0.73 0.82 22.14
  1. Chan & Lee, 2016
15,740 3164 6 0.87 0.79 0.97 17.54
  1. Dominguez et al., 2018
19,703 323 10.1 0.91 0.84 0.99 19.06
  1. Huang et al., 2015
367,442 46,067 14 0.78 0.76 0.80 26.07
  1. Buil-Cosiales, et al., 2014
7,216 425 8.7 0.63 0.46 0.86 7.90
  1. Xu et al., 2014
1110 300 10 0.66 0.48 0.91 7.29
Meta-analyses
  1. Kim & Je, 2016
1,409,014 45,078 0.77 0.71 0.84
  1. Yang et al., 2015
982,411 67,260 0.84 0.80 0.87

Note: HR was calculated for fiber diet vs. non-fiber diet. 

Results of meta-analysis (n= 978,380) showed that the aggregated effect size between fiber diet and mortality is HR = 0.80 (HR S.E. = 0.03, 95% CI [0.74,0.86]). This result is significant ?2 (5) = 8155.70, p < .001. and means that people consuming a high fiber diet at any time point during the study period were 20% less likely to die than people that were not consuming a high fiber diet, and we are 95% confident that the true value is lying between 14%-26%. (we are 95% sure that people not consuming meat are between 14% and 26% less likely to die at a given age than people consuming a high fiber diet). 

Significant heterogeneity was found between studies, meaning that the studies included in this analysis are different by several methodological aspects, which could bias the aggregated effect, Q (5) = 21.44, p < 0.01. Hence, differences between individual studies are significant and considered heterogeneous (see Figure 4.6, result of 3. Dominguez et al., 2018 exceeds CI borders).

To conclude, a high fiber diet significantly decreases death probability by about 20% on average, in comparison with a non-fiber diet. This effect size is in line with the effect size received by Yang et al. (2015) (HR=0.84) and by Kim & Je (2016) (HR=0.77). 

Figure 4.5: Forest plot of HR between fiber diet and mortality 

pastedGraphic_4.png

Here we can see that the confidence intervals for studies overlap.

Figure 4.6: Radial plot of HR between fiber diet and mortality 

pastedGraphic_5.png

4.4. Plant-Based Nutrition and Diabetes 

Three individual studies were used to assess the aggregative effect size of plant-based nutrition and diabetes, yielding three effect sizes. These effects were based on random control trial designs in which individuals in a diet group were compared to individuals in a control group. These designs yielded an effect size of the difference between means. In addition, a single meta-analysis was found which computed the Hazard Ratio between plant-based nutrition and diabetes (see Table 4.4). 

Table 4.4: Individual studies evaluating Effect size between plant-based nutrition and diabetes

Study Cases in Vegan Diet  Cases in non-Vegan Diet Cohen’s d  95% CI – Lower  95% CI – Upper Weight in meta-analysis
  1. BARNARD et al., 2006
21/49 13/50 -0.32 -0.57 -0.07
  1. Kahleova et al., 2018
38 37 -1.0 -1.2 -0.8
  1. Lee et al., 2016
46

-0.5 SD 0.8

47

-0.2 SD 0.7

-0.40 -0.65 -0.15
Meta-analyses Total sample  Total Diabetes cases HR HR 95% CI – Lower HR 95% CI – Upper
  1. Tonstad et al., 2009
60,903 3,430 0.51 0.40 0.66

Results of meta-analysis (n=133) showed that the aggregated effect size between plant-based nutrition and diabetes is Cohen’s d = -0.17 (S.E. = 0.06, 95% CI [-0.30,-0.03]). When translated to HR = 0.73 (95% CI [0.58, 0.94]. This result is significant ?2 (2) = 6.24, p < .05 and means that people consuming a plant-based diet at the end of the trial (dietary change to PBD) showed 27% improvement in their diabetic status.

No heterogeneity was found between studies, meaning there are no potential moderators that could bias this effect, Q (2) = 1.06, p = 0.50. Hence, differences between individual studies are not significant and considered homogeneous (see Figure 5.8, all studies are within the CI borders).

To conclude, individuals who keep plant-based have decreased risk of diabetes in comparison with individuals who do not keep this type of diet. This result is stronger when translated to HR = 0.73, in comparison with an effect size of 0.51 (Tonstad et al., 2009).

Figure 4.7: Forest plot of HR between plant-based nutrition and diabetes

pastedGraphic_6.png

Here we can see that the confidence intervals for studies overlap.

Figure 4.8: Radial plot of HR between plant-based nutrition and diabetes

pastedGraphic_7.png

4.5. Vegetarian or Mediterranean Diet and Heart Disease

Two individual studies were used to assess the aggregative Hazard Ratio (HR) of Healthy diet and heart disease, yielding two effect sizes. In addition, a single meta-analysis examining this effect was found (see Table 4.5).

Table 4.5: Individual studies evaluating HR between a vegetarian or Mediterranean diet and heart disease

Study Total sample  Total heart disease cases Follow-up (years) HR HR 95% CI – Lower HR 95% CI – Upper Weight in meta-analysis
  1. Li et al., 2013
4098 1133 0.73 0.51 1.04

36.54

  1. Stewart et al., 2016
15,482 2885 3.7 0.94 0.89 0.99 64.36
Meta-analyses
  1. Kwok et al., 2014
183,321 0.84 0.74 0.96

Note: HR was calculated for vegetarian or Mediterranean diet vs. non-healthy diet. 

Results of meta-analysis (n=19,580) showed that the aggregated effect size between vegetarian or Mediterranean diet and heart disease is HR = 0.86 (HR S.E. = 0.10, 95% CI [0.67,1.06]). This result is not significant ?2 (1) = 25.10, p = .413

These effects were homogenous between the two studies, meaning, studies included in this analysis were not different by methodological aspects, which could bias the aggregated effect, Q (1) = 3.34, p = 0.07. 

To conclude, a vegetarian or Mediterranean diet was not found to decrease the probability of heart disease. Although non-significant, the effect size found in this meta-analysis is similar to the effect size received by Kwok et al. (2014) (HR=0.84). They come to the conclusion that there is a modest cardiovascular benefit but no clear reduction in overall mortality associated with a vegetarian diet.

A vegetarian diet can be healthy or not, depending on the foods consumed in this diet. A vegetarian diet rich in processed foods or a Mediterranean diet with high quantities of meat and dairy products will not produce the desired health effects. Furthermore, only two studies were found that met all the criteria for inclusion.

Figure 4.9: Forest plot of HR between a vegetarian or Mediterranean diet and heart disease 

pastedGraphic_8.png

Here we can see that the confidence intervals for studies overlap.

Figure 4.10: Radial plot of HR between a vegetarian or Mediterranean diet and heart disease pastedGraphic_9.png

4.6. IGF-1 and Cancer

Insulin-like growth factor 1 (IGF-1) is a protein produced in the liver, encoded by the IGF-1 gene, which stimulates growth in cells throughout the body. Protein intake increases IGF-1 levels in humans under age 65, independent of total calorie consumption.

IGF-1 has a role in regulating lifespan and in assisting growth hormone in its anabolic function. It plays several roles in human physiology, including tissue growth and development, especially at a young age, where it promotes growth in children and ensures that they grow tall (Laron, 2001). IGF-1 is also found in breast milk. 

Research shows that IGF-1 continues to have anabolic effects as the person gets older, where increased levels of IGF-1 seem to have several adverse effects on health as people reach adulthood and age. 

Studies have implicated IGF-1 with a few forms of cancer, including colon, pancreas, endometrium, prostate, breast, lung, and colorectal cancer (Renehan, 2000; Yu, 2000; Annekatrin, 2002; Renehan, 2004; Allen 2007; Roddam, 2008; Key, 2010; Rinaldi, 2010; Clayton 2011; Yang & Yee, 2012; Christopoulos, 2015) as IGF-1 exerts strong mitogenic actions and triggers a signaling cascade leading to increased proliferation and differentiation of cells and has an anti-apoptotic effect.  Certain drug companies are working on medications that reduce the level of IGF-1 as a means to protect from cancer (Pollak, 2008).  However, there is no definitive association between IGF-1 and cancer in the Japanese population (Mikami, 2009; Suzuki, 2009). This may be because IGF-1, which can also be attained through the diet, is not found in foods regularly consumed as part of the Japanese diet. When examining dairy products and the Japanese population, we will see the same results as with the rest of the population (Akter, 2013)

Epidemiological evidence shows that dairy food consumption significantly increases circulating IGF-1 levels, and dairy consumption after the weaning period maintains high IGF-1 signaling (Hoppe, 2006; Rogers, 2006; Rich-Edwards, 2007; Crowe, 2009; Qin, 2009; Major, 2010).  A study showed that when insulin-like growth factor-1 is taken in through the diet, further to the added exogenous dose of IGF-1 in the body, there is also increased stimulation of IGF-1 production in the body (Allen, 2002), which promotes certain cancer proliferation.

To assess the aggregative Hazard Ratio (HR) of diary products (IGF-1) and cancer, thirteen individual studies were used, yielding thirteen effect sizes (Hankinson, 1998; Yu, 1999; Renehan, 2000; Annekatrin, 2002; Spitz 2002; Renehan, 2004; Allen, 2007; Roddam, 2008; Gunter, 2009; Mikami, 2009; Endogenous Hormones and Breast Cancer Collaborative Group & Key et al., 2010; major, 2010; Rinaldi, 2010). In addition, a single meta-analysis examining this effect was found (Shi, 2004) (see Table 4.6).

Table 4.6: Individual studies evaluating HR between IGF-1 and cancer

Study Control  Cancer cases HR HR 95% CI – Lower HR 95% CI – Upper Weight in meta-analysis
  1. Endogenous Hormones, T. E., & Breast Cancer Collaborative Group, 2010
9,428 4,790 1.28 1.14 1.44 17.28
  1. Gunter et al., 2009
841 810 1.46 1.00 2.13 8.09
  1. Annekatrin et al., 2002
263 132 4.97 1.22 20.2 8.72
  1. Renehan et al., 2000
293 52 3.05 2.04 4.57 0.21
  1. Renehan et al., 2004
7137 3609 1.49 1.14 1.95 2.51
  1. Rinaldi et al., 2010
1121 1121 1.43 1.13 1.93 10.72
  1. Roddam et al., 2009
5200 3700 1.38 1.16 1.60 12.21
  1. Allen et al., 2007
630 630 1.39 1.02 1.89 14.77
  1. Mikami et al., 2009
302 101 1.01 0.49 2.10 10.2
  1. Spitz et al., 2002
297 297 2.21 0.35 12.84 6.98
  1. Major et al., 2010
559 74 1.61 1.28 2.02 5.13
  1. Hankinson et al., 1998
620 397 2.33 1.06 5.16 3.23
  1. Yu et al., 1999
218 204 2.06 1.19 3.56 17.28
Meta-analyses
  1. Shi et al., 2004
6030-1017 30-397 1.05 0.94 1.17

Note: HR was calculated for cancer vs. non-cancer. 

Results of meta-analysis (n=26,909) of these studies showed that the aggregated effect size between IGF-1 and cancer is HR = 1.48 (HR S.E. = 0.09, 95% CI [1.31,1.65]). This result is significant ?2 (12) = 914.23, p < .001 and means that people consuming high IGF-1 products (dairy products) at any time point during the study period were 48% more likely to be diagnosed with cancer than people that were not consuming a high IGF-1 diet, and we are 95% confident that the true value is lying between 31%-65%. (we are 95% sure that people consuming dairy are between 31% and 65% more likely to be diagnosed with cancer than people consuming a low/no dairy diet).

Significant heterogeneity was found between studies, meaning that the studies included in this analysis are different by several methodological aspects, which could bias the aggregated effect, Q (12) = 25.67, p < 0.01. Hence, differences between individual studies are significant and considered heterogeneous (see Figure 5.9, the result of Annekatrin et al., 2002 exceeds CI boarders). 

To conclude, high IGF-1 levels were found to increase the probability of cancer diagnosis by about 48% compared to patients with low IGF-1 levels. This finding was larger compared to the effect size derived from the meta-analysis of Shi et al., 2004 (HR=1.05). This suggests that reduced dairy product consumption will lead to improved health in the long term.

Figure 4.11: Forest plot of HR between IGF-1 and cancer

pastedGraphic_10.png

Here we can see that the confidence intervals for studies overlap.

Figure 4.12: Radial plot of HR between IGF-1 and cancer

pastedGraphic_11.png

Results:

The results from these meta-analyses presented, which involved 6,172,423 people, of which 326,274 became sick or died during the studies, aimed to assess aggregate effect sizes of several nutrition types with both mortality and diseases. 

To conclude, all meta-analyses conducted to assess mortality showed highly significant results. Specifically, meat consumption increased mortality probability by 18% on average, a plant-based diet and a high fiber diet decreased mortality by 15% and 20%, respectively. 

Also, dietary dairy consumption (as measured by IGF-1) was found to increase the probability of cancer by about 48%, while plant-based nutrition reduced diabetes by about 27%. 

No significant effects were indicated for meat consumption or vegetarian or Mediterranean diet on diabetes or heart disease. 

Table 4.7: Aggregate Effect Sizes for Meta-Analyses 

Aggregate Effect Size 95% CI – Lower 95% CI – Upper Significant
Mortality 
Meat consumption and mortality 1.18 1.12 1.24 Yes
Plant-based nutrition and mortality 0.85 0.77 0.95 Yes
High fiber diet and mortality 0.80 0.74 0.86 Yes
Disease
Vegetarian/Mediterranean diet and heart disease 0.86 0.67 1.06 No
IGF1 and cancer 1.48 1.31 1.65 Yes
Plant-based nutrition and diabetes 0.73 0.58 0.94 Yes

 The possible limitations of these meta-analyses should be taken into consideration.  Although the combination of results from different studies will increase statistical power in detecting significant associations because of the increased sample size, this often results in heterogeneity. Heterogeneity is expected as the studies took place in different geographic locations, used different dietary assessment methods, and included participants in different gender and age groups. In general, there was significant heterogeneity in many of the meta-analyses, as can be seen in the redial plots.

Publication bias is another concern. The statistical tests did not suggest the presence of publication bias in these meta-analyses. However, some may have had limited statistical power due to the sometimes low number of studies. On the other hand, very large numbers of participants do reduce this bias.

Completely ruling out the possibility of residual confounding or a temporal bias cannot be done, but if the associations found are real. It is safe to say that a whole food plant-based diet can reduce the risk for common diseases and increase longevity.

Summary

The evidence from this research component suggests that the most suitable diet for human consumption for health and longevity is a natural, whole-food, high-fiber, and 90+% plant-based diet, with small amounts of lean meat. This diet leads to health for our species, reducing the need for healthcare costs, especially for the growing elderly population.

To produce change and have more of the population follow this line of feeding, governments and individuals need practical methods and health policies that improve health with a smaller carbon footprint.  

The consumption of animal products influences not only personal health but also environmental health. The current situation shows that the rearing of livestock and meat consumption on a commodity-basis accounts for the highest greenhouse gas (GHG) emissions, respectively, supplying 41 percent and 20 percent of the sector’s overall GHG outputs. (192) Rearing of livestock is also the single greatest anthropogenic source of methane, a GHG about 25 times more powerful than CO2 (from raising cattle for food) and nitrous oxide emissions (from fertilizer and manure usage), two very potent GHGs.

The rearing of livestock is responsible for approximately 37% of anthropogenic methane emissions and approximately 65% of human nitrous oxide emissions globally. (193) 

Furthermore, animal agriculture is a notable contributor to global warming due to the quantities of fossil fuels used, together with deforestation. Worldwide, fossil fuel energy is responsible for 40% of human GHG emissions, which does not include deforestation at about 18%, and animal agriculture at 18%. Most of the deforestation is done to rear animals and expand pastures, and arable land is used to grow crops for feeding livestock. Thus, of the 91-97% human-induced GHG emissions, 60% is due to animal agriculture.

Manufacturing beef demands significantly higher resources. Beef production needs 28 times more land, six times more fertilizer, and 11 times more water than chicken or pork production. Furthermore, producing beef releases four times more GHGs than the same amount of pork on a calorie basis and five times more than poultry. (194) 

The consumption of plant-based foods produces very low GHG emissions. (195) It is more economical to grow crops for food than to grow crops for animal feed, necessitating muscle mass and bone tissue build-up.

Overall, the goal of agriculture and governments must be to build a sustainable future and support the population’s health. This means finding solutions that will continue to meet human food and energy requirements in cheap, safe, and high-quality ways even for a growing human population, while leaving little or no negative effects on our planet, along with disease control and caring for animal welfare, in a way that is profitable for the farmer.

Policies for Change

To ensure that every person on the planet would be able to meet their nutritional needs in the future, we would need to (1) build stable national relationships between different countries for consistent import and export of agricultural goods for food security, (2) establish domestic and global policies including meat and dairy taxes to be implemented to ensure that a price is paid for the destruction of the earth and its resources, and (3) make agricultural policies and trade rules compatible with global food security and sustainability.

We also need to improve dietary habits. The number of men and women existing in sub-optimum health is a dramatic 95% of the global population, and this number is estimated to grow in the upcoming years. (196, 197, 198) 

If there is no change in prevailing dietary habit trends, GHG emissions in 2050 connected with food systems will rise by 51% compared with current levels.

If the global population followed a 90% plant-based diet, with meat and animal products forming 10% of the diet, GHG emissions would decrease by 55%. These statistics show a clear and simple way to effect change. Reducing meat and dairy products from the diet to twice a week and changing the common diet composition to mostly vegetables, nuts, fruits, whole grains, nuts, and legumes, can immensely influence global GHG emissions, slow deforestation, and prevent many diseases.

Governments and civil society are profoundly unwilling to intrude into people’s diets and tell them what to eat. But we will soon understand that this is the only way to go. 

Although reducing overall meat and dairy product consumption will help, the type of meat chosen also has an immense effect on our planet and the future of food. Some animal products are more sustainably produced than others. A way of comparing species is to look at how efficiently an animal converts feed into biomass for human nutrition. 

One needs only 1.1 kg of pellet food to get 1 kg of salmon meat. (199, 200, 201)

Chickens also use feed efficiently, where 1.7 kg of feed produces 1 kg of chicken because they are grown quickly and slaughtered at a young age. 

By comparison, 2.9 kg of feed is needed to make 1 kg of pig meat, and 6.8 kg of feed is needed to make 1 kg of cow meat.

A reduction in animal consumption would have major effects on life on Earth:

  • Of the available 12 billion acres of agricultural land available on earth, 68% is used for livestock. (202) Some of this land could be restored for grasslands and forests to help capture carbon, further reducing carbon emissions, or be diverted to growing plants for human consumption.
  • People previously involved in the livestock industry (about one-seventh of the global population) would require help making the shift towards a different career, whether in plant-based agriculture, in reforestation, in the biofuel industry from the byproducts of crops now used as food for livestock, or in caring for the animals (re)introduced into the wild or into sanctuaries and zoos.
  • About one-third of the planet’s land is arid to semi-arid rangeland, only able to support livestock agriculture. In these areas, land could be used to house the growing African population, for vertical farming facilities, for growing native trees found to be of medicinal value (e.g., moringa or shea), and for growing livestock for wool for populations such as the Mongols and Berbers, who would otherwise lose their cultural identity, causing them to settle permanently in cities or towns. Solar farms could also be located on this land, providing sustainable sources of energy to local communities. 

Apart from the myriad reasons to lower meat consumption, meat has an important role in tradition and cultural identity. Giving up meat has an impact on the culture of many societies, so governments and people have personally failed to reduce meat consumption.

This indifference can be combated by increasing the price of meat so that farmers can raise fewer animals and earn the same. This shift in production would make meat take the form of a treat rather than staple food, as it is today.

Governments can subsidize fresh vegetables and fruits, making them more affordable and more widely accessible to all populations, instead of subsidizing meat and dairy products. 

The environmental issue may also be solved with meat coming from the use of technology, such as lab-grown meat and fish. (203, 204, 205)

The current problem is that newly rich societies are increasing their demand for animal products. As people’s incomes increase, they start buying more dairy, poultry and meat, and fish.

Therefore, to improve people’s eating habits, a whole food, mostly plant-based diet, should be encouraged and taught in schools, including medical school. Just as teaching first aid is common practice all over the world, the same should be done with regard to plant-based food choices. 

Discussion

As we see, food choices are very dependent on prices and can therefore be influenced by prices.

Difficulties also arise in making healthy food choices in food swamps, where affordable, fresh, and healthy foods are accessible, but there is an overabundance of energy-dense, low-nutrient foods. Here, unhealthy food choices are much easier to make than healthy food choices due to their cheaper prices. This is where education is critical.

Sugar taxes

Since sodas are not a necessary element of a wholesome diet, soda is a welcome candidate for taxation. We see that if a tax of about 20% is introduced, it has a serious effect on consumers’ buying behavior producing many health benefits.

In Mexico, a sugar tax on soft drinks has been successful due to the fact that the funds were spent providing free drinking water in schools. (206, 207, 208)

Meat and dairy taxes

The price of animal products must match their real cost to society, including their carbon footprint. A meat tax puts a specific price on the harm they cause the environment. Currently, there are no consequences for raising livestock, even though these industries are proven to be detrimental to health and the environment. There must be a financial deterrent such as taxes, fines, or penalties to discourage their production and usage. To date, no economic incentives are in place for industries or individuals to move away from the generation and consumption of animal goods. Taxing meat and dairy products will put economic pressure on people and these industries to make changes.

This tax money can then be given back to meat farmers as government support.

People will still buy meat, but more as a treat rather than as a staple. 

Meat tax will lead to a major reduction in GHG emissions and preserve over 500,000 lives per year through healthier diets.

If we add a 35% meat and dairy tax and encourage sellers to sell meat and dairy products at 50% higher prices, people will make different choices.

Food Stamps:

For 40 years, the food stamps program has been a very significant domestic hunger safety net that helps provide economic well-being, access to proper nutrition, food security and accessibility, and a reduction of child poverty and money for food spending that benefits those most in need. Food stamps are also good for the economy. (209)

There is one major drawback of such programs, namely the foods they include.

People can use food stamps to purchase any food item for human consumption, including candy, soft drinks, ice-cream, crackers, cookies, and cakes. (210) 

Food stamp policy should change to allow the purchase of only natural foods, without options for foods that are the leading causes of illness and chronic diseases. 

Food stamps can thus help guide populations using them towards making healthy food choices by default. In this way, the government will target the poorest, most needy families first, which will lead the way for agriculture to follow suit.

Subsidies 

In the current US food pyramid, The USDA suggests that the meat, poultry, fish, eggs, and legumes together should comprise 10% of our diet and that dairy products should comprise 23% of our diet. When putting these foods together and reducing legumes’ intake, the pyramid suggests we should be consuming about 30% of our calories from animal-based products. (211, 212) However, the meat and dairy industries get 74% of USDA subsidies. According to the USDA food pyramid, vegetables and fruits are to form 38% of total calories. However, these industries get under 3% of the subsidies. (213)

Grains receive 13% of subsidies, with most going to feed livestock; sugar, oil, starch, and alcohol, 11%; nuts and legumes, 1.9%; and fruits and vegetables, 0.4% of subsidies.

To improve health, subsidies should become more similar in percentages to the USDA’s tiers in the food pyramid. 

Table 5.2: The distribution of foods in the current USDA food pyramid. (2pastedGraphic_12.png

The Recommended Food Pyramid Based on This Research

According to this research into the ideal diet for humans, the optimal food pyramid would reflect the following breakdown:

      • grain consumption (27%), recommending that all grains should be consumed as whole grains;
      • vegetables (26%), highlighting a variety of dark green vegetables, as well as root vegetables;
      • legumes (20%), peas, lentils and beans, and their spreads;
      • fruits (15%), emphasizing variety and deemphasizing fruit juices;
      • meat (7%), emphasizing lean meats such as fish and chicken;
      • oils, nuts, seeds (5%), recommending nuts and seeds and their pastes as sandwich toppings.
      • honey, emphasizing whole natural honey (0-0.5%)

Table 5.3: The distribution of foods in the recommended food pyramidpastedGraphic_13.png

To conclude

A whole-food, high-fiber, plant-based diet consisting mainly of whole grass grains, legumes, USOs, nuts, seeds, and fruits, with reduced quantities of meat and dairy products, has been statistically proven to not only prevent most modern-day diseases but may also reverse them while supporting the growing population healthily and sustainably.

Individuals and governments should aim to use this knowledge through policies to feed the growing global population healthily and sustainably without causing further environmental destruction.

In developed countries, interventions can include increased income-earning opportunities, changes in the food pyramid, meat, dairy, and sugar taxes, change in food stamp guidelines, subsidies for farmers, support for excess food sharing, supermarket availability for all communities, and school feeding and education programs in all countries. In developing countries, this effort could include local markets for food producers, improved infrastructure, secured purchasing power through governmental prevention of price fluctuations, securing land ownership, easier access to credit, knowledge-sharing through demonstration farms and websites, and legal structures supporting private investors.