Question: What dietary factors contribute to non-communicable chronic diseases (NCDs) among populations transitioning from their original to westernized diets?

Findings: Our systemic literature review examined four populations that transitioned from their original to a more westernized diet and lifestyle. We also reviewed seven additional populations that underwent a similar transition. We identified a strong association between NCDs and increased sugar and refined carbohydrate consumption, and weaker associations with increased total calories with reduced physical activity. Neither fat nor saturated fat intake were associated with risk of developing NCDs in any of the populations.

Meaning: Increased consumption of sugar and refined carbohydrates were strongly associated with the development of NCDs in all four populations. Increased calories and decreased physical activity were less strongly correlated although both of these measures are imprecisely defined and not quantified in any of these group. Neither fat nor saturated fat intake were associated with NCD risk in any population.

US obesity prevalence increased 313% from 1960 (13.4%) to 2018 (42%) (1), increasing morbidity and mortality, and costing >$150 billion yearly in US direct health care and lost productivity. As obesity rose worldwide, a parallel but delayed rise in T2D, hypertension, heart disease, stroke, cancer, Alzheimer's disease, and gout usually followed. Debate persists on the roles of energy balance, total fat, SFAs, sugar and refined carbohydrates, total calories or inactivity as primary cause(s) of obesity. Medical orthodoxy holds that positive energy balance (calories consumed exceed those expended) causes obesity, while others argue that sugar/refined carbohydrate-induced metabolic alterations drive obesity and NCDs via metabolic syndrome (2–9). They are not mutually exclusive. Medications (e.g., insulin, antipsychotic, antiseizure), medical and psychiatric disorders, stress, sleep deprivation, and other factors influence weight, confounding attempts to definitively link factors such as caloric or sugar consumption and weight. Further, quantifying caloric consumption and activity levels in populations over time is unreliable. Together, these limitations make it difficult to identify causal relationships.

Nutritional causes of disease have focused on heart disease and obesity and less often, on cancer. Keys' Diet-Heart hypothesis (10) posited that dietary saturated fatty acids (SFAs) increased heart disease risk and was supported by the Seven Countries Study and subsequent research (11, 12). Recent Cochrane (11) and American Heart Association Presidential Advisory reviews (13) of randomized controlled trials (RCTs) conclude that replacing SFAs with polyunsaturated fatty acids (PUFAs) in vegetable oils or unrefined carbohydrates - but not with refined carbohydrates and sugars - lowers low-density lipoproteins (LDL) and reduces the risk of all cardiovascular disease (CVD) events by ~14–30% (11, 13). However, no benefits were seen in non-fatal myocardial infarction, coronary heart disease (CHD) deaths, stroke, cancer diagnoses or deaths, or in overall mortality (11). Other expert groups argue that the evidence fails to support any adverse health effects of eating whole-fat dairy or unprocessed meats (14). Further, large prospective epidemiological population studies of single populations, some with >6-fold differences in SFA consumption, and RCTs found little or no relationship between SFA consumption and weight, cholesterol levels or cardiovascular disease (CVD) risk (15–26). Finally, some studies found increased rates of CVD, cancer, or death when PUFAs were substituted for SFAs or in individuals with low cholesterol levels (27, 28).

In the 1970s and 1980s, health agencies extended the recommendation to reduce dietary fat and SFAs from older men at high-risk for CVD to all individuals over age 2 years. No evidence supported this broader recommendation. The low SFA recommendation was generalized to reduce cancer risk and promoted by the National Cancer Institute (29), National Academy of Sciences (29), American Cancer Society (30), and Surgeon General (31, 32). However, population-based studies correlated low cholesterol with higher cancer rates (30, 32, 33), and clinical trials revealed that cholesterol lowering diets or drugs led to increased cancer rates (34–36). Populations consuming larger amounts of animal fat such as the Inuit and Plains Native Americans had very low rates of cancer (37, 38). Since obesity was associated with increased risk of multiple cancers, fat or SFA were linked to cancer indirectly because fats are energy dense. Yet, only populations consuming high fat diets with abundant sugar or other refined carbohydrates had high rates of cancer and other NCDs. Large prospective studies and RCTs failed to support the link between dietary fat or SFAs and cancer (11, 34, 35), and health agencies no longer support fat/SFA reduction to reduce cancer risk (39).

The dietary causes of obesity, heart disease, and NCDs have been studied with 1) animal models of genetic obesity, brain lesions, and overfeeding of calories or selected macro- or micro-nutrients, and drugs; 2) human volunteer studies of variable duration with overfeeding, underfeeding, increased exercise, different macronutrient diets, and hormonal and lipid changes associated with dietary changes; 3) retrospective and prospective observational studies on diet and health outcomes within and between populations; and 4) randomized controlled trials. However, NCDs may be best understood as evolutionary mismatch disorders. Evolutionary mismatch refers to advantageous traits that evolved in one environment, but became maladaptive after a rapid environmental change (40). NCDs are rare or absent among hunter-gatherer, nomadic, foraging-gardening, and agricultural-dependent populations with diverse genetics, diets and lifestyles (40). Some groups consumed primarily plant-based carbohydrates, others primarily animal fat and protein. NCDs were rare in all. These diets were minimally processed, lacking refined sugar, white flour, white rice, vegetable oils, chemical preservatives or meat from animals bred for high fat content or fed corn. When non-western populations transition to western diets and lifestyles, NCDs follow (41–47). Americans consume a typical western diet: high total calories and mostly processed foods including sugars, white flour and rice, meats (chicken, pork, beef), and dairy; with limited vegetables and fruit (48–50). Our domesticated plants and animals were bred for increased sugar [e.g., grapes (51) and apples (52)] and intramuscular fat (e.g., Aberdeen angus cow) while decreasing fiber (e.g., fruits). Westernization reduced physical activity as machines replaced manual labor and self-powered transport; while for others, a fitness revolution increased physical activity (41). In contrast to RCTs, transitioning populations offer the opportunity to observe natural changes in population health that occur with an increase in western diet and lifestyle over extended periods.

Most cultures have moved toward a more industrialized western diet, with more sugar, refined flour and rice, and processed oils and meats. Greater availability of refined carbohydrates reflected lower cost, longer shelf-life, ready-to-eat, and possibly more addictive; fueling increased consumption that drove poor health outcomes (42–44). Although this western transition occurred in hundreds of populations, few studies systematically quantified changes in diet, physical activity, stress and other risk factors with health outcomes. Two longitudinal studies included in our analysis are the Tokelau Island Migrant Study and the Seven Countries Study, but both had limited data on diet and health outcomes and did not prospectively assess changes in total calories, unrefined vs. refined carbohydrates, physical activity, medication use, and many other factors.

Studying populations transitioning toward western diets may inform pathogenic factors underlying some NCDs. What dietary or lifestyle changes are critical - total calories, reduced physical activity, sugar and refined carbohydrates, processed foods, mental stress, reduced sleep, medications, or other factors? Given the many limitations, even for rigorous prospective studies, we systematically reviewed the health outcomes of four extensively studied, diverse populations before and after a nutrition transition. We also reviewed data from seven additional populations to assess potential bias in our selection. Our goal was to identify common factors associated with NCDs in these populations.

Our systematic review of the dietary and lifestyle transitions of the Yemenite Jews, Tokelauns, Tanushimaru and Maasai revealed overlapping but distinct health outcomes that do not implicate total fat or SFA in NCD pathogenesis. Our data support sugar and refined carbohydrates in causing obesity, diabetes and other NCDs. Increased calories and decreased physical activity were also present in several groups. Our analysis of seven other groups (Pima and Navajo Native Americans, Aboriginal Australians, Hadza, Inuit, Natal Indians and Zulus) undergoing a diet and lifestyle transition all support that the common element associated with rising obesity and diabetes were sugar and other refined carbohydrates (Table 2). The exception was the Hadza who largely returned to their traditional diet and lifestyle and who did not develop high rates of NCDs. Data from these 11 groups were inadequate to assess the potential role of increased calories, decreased physical activity, stress, sleep disorders, medications and other factors. Although fats and SFAs were linked to obesity, diabetes, and CHD in the United States (US) and elsewhere (128), using international comparisons, rising obesity and diabetes occurred during periods when US fat and SFA consumption declined and sugar consumption increased (129, 130).

Several groups replaced SFAs from animal (Maasai, Inuit and Navajo) or plants (Tokelau) with refined carbohydrates, and often with MUFAs and PUFAs, but experienced increased obesity, diabetes and other NCDs. For Yemenite Jews, consumption of animal fat was stable while vegetable oil doubled and sugar increased nine-fold, and calories increased by 10% (50). The pre-transition Tanushimaru diet was mainly unrefined carbohydrates. Their post-transition diet was higher in fat but heart disease and cancer rates were stable while stroke rates declined. Among the 11 groups, sugar and refined carbohydrates increased markedly in all with large increases in obesity, diabetes, and NCDs. This concept is ancient. Excessive sugar and carbohydrate consumption were considered causes of obesity and diabetes in India (500–1000 BCE), and other 20^th century groups that transitioned from traditional diets to diets rich in sugar and refined carbohydrates: Asian Indians (131, 132), underprivileged Americans (133), and Icelanders (134), observed by hundreds of missionary physicians across the world (115, 135, 136).

The potential adverse health outcomes of SFAs were indirectly supported in 1961 by the Framingham Heart Study (FHS) finding that high cholesterol levels were a risk factor CHD, since SFAs raised total cholesterol in feeding studies (137). However, the prospective FHS diet study followed 1000 subjects and found no relationship between dietary fat nor SFA consumption and obesity, cholesterol levels, or CHD. This study was suppressed from peer-review publication (15). Decades later, the FHS director admitted in an editorial, that ‘the more saturated fat one ate, the more calories one ate, the lower the person's serum cholesterol…(and they) weighed the least (100, 109, 133, 138–145).”

Excess calories may have contributed to obesity and NCDs in many [e.g., Bengali Brahmans (131), Yemenite Jews (146), Tokelauns (138), Maasai (113), Pima (147)] but not all groups, as some were impoverished, consuming low calorie diets with obese and diabetic adults [e.g., Natal Indians (1600–2000 calories/day), (148, 149) Africa, (150) Trinidad, (151) Jamaica, (152) Sioux on reservations (137)] while their children were malnourished. This dual burden – obesity in mothers and underweight malnourished children – is present worldwide, affecting wealthy and poor nations (153). Some impoverished obese adults held physically demanding jobs and consumed low calorie diets with little access to healthcare and limited resources. Dual burden populations challenge the energy balance hypothesis of obesity, which implies that mothers overeat and underfeed their children (8). Unlike adults, underfed children cannot lower their metabolic rate until they have lost more than 30% of their weight (154). Adults reduce metabolic rate up to one-third to compensate for semi-starvation (128, 155–158). Mothers consuming low calorie diets rich in refined carbohydrates develop elevated insulin levels, insulin resistance, obesity and metabolic syndrome over decades (130, 159, 160).

Additional challenges to the energy balance model of obesity come from metabolic and hormonal studies (8, 9), chidren in foraging populations with higher physical activity but similar total energy expenditure as children from industrialized communities (139), prospective population studies (161, 162), overfeeding studies (140, 148). The energy balance hypothesis was not supported by the dietary-weight observations in Framingham and randomized dietary trials (161, 162). Even in prospective studies, caloric estimates are imprecise and caloric changes inconsistently correlate with weight changes. In overfeeding studies, healthy subjects with stable activity levels consumed a ~90% fat diet of 5,975 calories daily for five weeks with stable or lower weight (140, 148). Moderate to high calorie low-carbohydrate diets can reduce weight and diabetic symptoms (6, 148), sometimes yielding greater weight loss than lower calorie high carbohydrate diets (141). Discrepancies between calorie consumption and weight may reflect inaccurate calorie counts, differences in physical activity, refutation of the energy balance hypothesis, or some combination of these. In the Women's Health Initiative (WHI), 20,000 women consumed a “heart-healthy” diet with 114 fewer daily calories than 29,000 women on a conventional American diet. After 7.5 years, their weights were nearly identical; energy balance predicts that 328,000 excess calories should translate to more than 80 pounds of fat, but neither BMI nor waist circumference differed between the groups. The serial NHANES data on caloric consumption and weight, as well as the FHS diet study, failed to support energy balance.

Since obesity, diabetes and cardiovascular disease were rare among early-mid 20^th century Japanese consuming very high carbohydrate/low fat diets, Joslin, Himsworth and others considered fats pathogenic and all carbohydrates non-pathogenic. They assumed that all carbohydrates were “equivalent” (142–144, 163) since starch is digested to simple sugars. The safety of a high carbohydrate, high sucrose diet for diabetics was promoted by the American Diabetes Association up to 2008 (145). Early 20^th century epidemiological data strongly linked sugar consumption with obesity and diabetes within and between populations (164), and experimental and clinical data showed that low carbohydrate diets reversed diabetic symptoms in animals and humans (165). In rhesus monkeys, in six months, a high sucrose diet produced insulin resistance, central obesity, dyslipidemia, and some developed type 2 diabetes (166). Remarkably, Joslin's first diabetes textbook (1916) used a single case as an exemplar of diabetes etiology. This 49 year old overweight man had a strong family history of diabetes and ate large amounts of candy. After gallstone surgery, friends ‘sent him much candy, which he ate' and within a month, was diabetic (156). Joslin reported (1946) that tooth destruction often preceded diabetes onset, yoking sugar, tooth decay and diabetes (157). The first report of the now epidemic nonalcoholic fatty liver disease was a 41-year old obese man who drank >20 Coca-Cola bottles daily (158). Disentangling high caloric and high sugar intake in obesity and diabetes is a recurrent challenge, exemplified by the Pima (147, 167). Duration of exposure to high sugar diets may be critical. Yemenites had been in Israel far longer than the Tokelauans were in New Zealand, consistent with Campbell's incubation period of >20 years for diabetes to emerge in populations consuming >70 pounds of sugar yearly (51). By contrast, the Pima had obesity more than 50 years before diabetes rapidly increased, although data on obesity from 1900 to 1970 is sparse. Among the four systematically reviewed populations, the two with the largest increases in sugar consumption – Tokelau (650%) and Yemenites (600%) - had the highest increase in diabetes (~3- and ~40-fold, respectively). Similar patterns were seen in the Aboriginal Australians, Pima, Navajo, Natal Indians, Zulu speakers, and Inuit.

Different metabolic and health effects of sugars and refined vs. unrefined carbohydrates emerged over the 20^th century and are critical to understand NCD pathogenesis. The glucose in starch and the fructose/glucose in sucrose have different metabolic fates and different effects on hormones, lipogenesis and inflammation. Refined and unrefined carbohydrates have different glycemic indices, temporal effects on insulin release, nutritional elements (e.g., more fiber, protein and minerals in unrefined), and health effects (e.g., appendicitis and constipation with refined). Populations consuming >80% unrefined carbohydrates had very low rates of obesity or NCDs (135). Studies in experimental animals and human populations consuming diets rich in sugar and refined carbohydrates developed the most prevalent American hyperlipidemia (elevated VLDL and triglycerides and low HDL) (168, 169), metabolic syndrome (hyperinsulinemia, insulin resistance, abdominal obesity, hypertension, hyperglycemia, non-alcoholic fatty liver disease) (4), T2D and other NCDs (159, 166, 170, 171). Together, these findings support that sugar and refined carbohydrates contribute to or cause NCDs.

The shift toward a more industrialized diet began in Europe and North America and spread worldwide (153, 160, 172, 173). Key drivers of this shift include lower cost, higher calorie density, convenience, urbanization and other demographic shifts (e.g., working mothers and pressure for ready-to-eat meals), influences of the food industry advertise highly profitable/low quality foods (e.g., soft drinks, sugary cereals), and the addictive nature of these highly processed foods (42, 173, 174). Affluent countries and affluent individuals in poor countries often have greater access to more expensive, healthier minimally processed foods such as fruit, vegetables, nuts, and fish.

Our study has limitations. Foremost was variability in dietary surveys (pre- to post-transition and across groups), assessment across seasons, and among different population subgroups and gender, and degree of detail regarding macronutrient components (e.g., SFA, MUFA, PUFA; unrefined vs. refined carbohydrates). Studies used different criteria and methods for how they measured the diet and physical activity between these populations, making it difficult to draw concrete conclusions. Diets also varied between individuals which studies attempted to correct for in various ways, making the data less standardized. Health outcomes data was often limited for the pre-transition period. Recall bias and filling-in are additional confounds prevalent in all studies. For example, the Yemenite diet was recalled after a decade or more. Obese individuals may underestimate calorie intake during recall studies than lean individuals (175). Post-transition diets changed over time; for example, decreased fiber, increasing fructose in fruits and fat in corn fed animals. Game meat has different SFA/MUFA/PUFA profiles than store-bought meat. Information on animal parts (e.g., offal vs. muscle) consumption was rarely specified. Few studies accounted for seasonal food availability, which influenced macronutrient composition and calorie intake. The pre- to post-transition interval varied among groups. The detail, rigor and validation of the assessments varied widely, although these populations were among the most rigorously studied. Further, carefully designed prospective dietary surveys have limited validity. Other factors, including physical activity, cigarette smoking, stress, sleep, comorbid psychiatric disease, and medications with positive (e.g., antihypertensive) or negative (e.g., antipsychotics causing metabolic syndrome) were infrequently and informally assessed. Rarely did a study include all of these factors together in a standardized way. Thus, the reduced stroke rate in Tanushimaru associated with higher fat intake may be due to the increased use of anti-hypertensive medications. Importantly, some groups underwent a physical transition (Yemenite, Tokelau, Maasai) while others did not (Tanushimaru). Different environmental exposures were not accounted for in these studies. We only systematically studied four among many populations that underwent a nutrition and lifestyle transition. However, among the six other groups that had a dietary and lifestyle transition (Pima, Inuit, Rural Chinese, Natal Indians, Aboriginal Australians), all had negative health outcomes and increased sugar consumption. Finally, many studies only assessed men or specific age ranges, excluding most of the population. Therefore, data from these transitional populations cannot definitively determine the role of dietary factors in health outcomes.

Four geographically and culturally diverse populations with different diets and lifestyles underwent a nutrition and lifestyle transition. The development of obesity, T2D and other NCDs was associated with a marked increase in sugar and refined carbohydrates and a more modest increase total caloric intake. The dietary transition of three populations we studied included a significant reduction in SFAs and in two, total fat consumption. All three developed NCDs despite the reductions in SFAs and fats. The Tanushimaru, our only population without increased sugar and refined carbohydrate consumption, had stable caloric intake with increased fat consumption: their health improved, with reduced breast cancer and stroke (possibly due to antihypertensive medications). These findings were supported by data from seven additional populations. In the ten populations with high rates of post-transition NCDs, increased sugar and refined carbohydrates were the common factor. Increased calories and decreased activity was present in many but not all, and may contribute to NCDs. Neither SFA nor total fat consumption were associated with NCDs. Although many other potential contributory factors to healthy outcomes were not adequately assessed (e.g., stress, medications, sleep), similar observations among scores of other populations who underwent a nutrition and lifestyle transition implicate sugar and refined carbohydrates as key dietary factors.

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

MP, JD, MD, JL, CG, SW, JL, and DF contributed to data analysis, writing and reviewing the manuscript. TR contributed to data and reviewing the manuscript. OD contributed to study design, data analysis, writing and reviewing the manuscript.

This work was supported by the Finding A Cure for Epilepsy and Seizures (FACES) organization affiliated with NYU Langone Health and the Comprehensive Epilepsy Center. The foundation had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

OD Research funding, Novartis, PTC Therapeutics, Marinus Pharmaceuticals, GW Pharmaceuticals; Zogenix; Equity interest, Rettco, Pairnomix, Tilray, Egg Rock Holdings; California Cannabis Enterprises. DF receives salary support for consulting and clinical trial related activities performed on behalf of The Epilepsy Study Consortium, a non-profit organization, receives no personal income for these activities. NYU receives a fixed amount from the Epilepsy Study Consortium toward DF's salary. Within the past two years, The Epilepsy Study Consortium received payments for research services performed by DF from: Axcella, Biogen, Cerevel, Crossject, Engage Pharmaceuticals, Eisai, Pfizer, SK Life Science, Xenon, and Zynerba. He has also served as a paid consultant for Eisai and Neurelis Pharmaceuticals. He has received travel support from Medtronics, Eisai and the Epilepsy Foundation. He receives research support from the CDC, NINDS, Epilepsy Foundation, Empatica, Epitel, UCB, Inc. and Neuropace unrelated to this study. He serves on the scientific advisory board for Receptor Life Sciences. He holds equity interests in Neuroview Technology and Receptor Life Sciences. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.