Dietary patterns

The Global Burden of Disease Study 2017 estimated the intake of major foods and nutrients in 195 countries and the share of disease in adults aged 25 and over attributable to each dietary risk. It attributed 11 million deaths (95% uncertainty interval 10 to 12 million) and 255 million disability-adjusted life years (a measure combining years of life lost to early death and years lived with illness) to diet in 2017. The three leading dietary risks were high sodium intake (3 million deaths), low whole-grain intake (3 million deaths) and low fruit intake (2 million deaths). The authors noted that dietary data came from mixed sources and were missing for some countries, which widens the uncertainty of the estimates [1].

A 2022 modelling study used meta-analyses and Global Burden of Disease data to estimate how sustained dietary change would alter life expectancy. Changing from a typical Western diet to an optimised diet, with more whole grains, legumes, fish, fruit, vegetables and nuts and less red meat, processed meat, sugary drinks and refined grains, was estimated to add 10.7 years of life expectancy for US women and 13.0 years for US men if begun at age 20, 8.0 and 8.8 years if begun at 60, and 3.4 years at 80. The largest single gains came from eating more legumes, whole grains and nuts and less red and processed meat [8]. These figures are population estimates that depend on the model's assumptions, not measured outcomes in individuals.

The Mediterranean diet is modelled on traditional eating in southern Europe: plentiful vegetables, fruit, legumes, whole grains and nuts, olive oil as the main fat, moderate fish and poultry, and little red or processed meat. The PREDIMED trial in Spain assigned 7,447 adults aged 55 to 80 (57% women), all at high cardiovascular risk but without cardiovascular disease, to one of three diets: a Mediterranean diet supplemented with extra-virgin olive oil, a Mediterranean diet supplemented with mixed nuts, or a control diet consisting of advice to reduce dietary fat. Participants received quarterly education sessions and, depending on their group, free olive oil, nuts or small non-food gifts [2].

After a median of 4.8 years, a major cardiovascular event (heart attack, stroke or cardiovascular death) had occurred in 3.8% of the olive oil group, 3.4% of the nut group and 4.4% of the control group. The hazard ratios (the risk in one group relative to another; 0.69 means 31% lower) were 0.69 (95% CI 0.53 to 0.91) for the olive oil diet and 0.72 (95% CI 0.54 to 0.95) for the nut diet. The original 2013 report was withdrawn after protocol deviations were found, including household members enrolled without randomisation at some sites. The 2018 reanalysis did not rely on the assumption that every participant had been randomly assigned, and results were similar after 1,588 participants with known or suspected deviations were excluded [2]. Associations between diet and cognitive decline are described under dementia risk factors.

Within dietary patterns, several individual foods and nutrients have been studied in large prospective studies or trials. Dietary fiber has the broadest evidence: systematic reviews of 185 prospective studies and 58 clinical trials found 15 to 30% lower all-cause and cardiovascular mortality, and lower rates of coronary heart disease, stroke, type 2 diabetes and colorectal cancer, in the highest compared with the lowest fiber consumers, with the greatest reductions at 25 to 29 grams a day [9]. Fiber is described in detail under dietary fiber. In 76,464 women and 42,498 men followed for up to 30 years, eating nuts was associated with lower mortality in a graded way, from 7% lower for less than once a week to 20% lower for seven or more times a week, with fewer deaths from cancer, heart disease and respiratory disease [4]. A review of fish intake estimated that one to two servings a week, especially of oily fish rich in the omega-3 fats EPA and DHA, reduced coronary death by 36% (95% CI 20 to 50%) and total mortality by 17%, and judged that these benefits outweighed the risks from contaminants such as methylmercury [10].

Salt reduction has trial evidence on hard outcomes. In the Salt Substitute and Stroke Study, 600 villages in rural China were randomly assigned so that 20,995 people with previous stroke, or aged 60 or over with high blood pressure, used either a salt substitute (75% sodium chloride and 25% potassium chloride) or regular salt. Over 4.74 years, the substitute group had lower rates of stroke (rate ratio 0.86), major cardiovascular events (0.87) and death (0.88), without a significant increase in serious high-potassium events [11]. Sugar-sweetened drinks show the opposite association: among 37,716 men and 80,647 women, drinking two or more servings a day was associated with 21% higher total mortality and 31% higher cardiovascular mortality than drinking less than one a month [12].

The International Agency for Research on Cancer classified processed meat as carcinogenic to humans and red meat as probably carcinogenic to humans in 2015, mainly on the basis of colorectal cancer [13]. A meta-analysis of prospective studies found colorectal cancer risk 18% higher per 50 g a day of processed meat and 17% higher per 100 g a day of fresh red meat [6]. Isolated nutrient supplements have not reproduced the associations seen with foods. The US Preventive Services Task Force concluded in 2022 that the harms of beta-carotene supplements outweigh their benefits, that vitamin E supplements give no net benefit for preventing cardiovascular disease or cancer, and that evidence on multivitamins is insufficient to judge [14].

Full deep dive: Foods and nutrients.

Vegetarian diets exclude meat and fish; lacto-ovo vegetarian diets include dairy and eggs, pescatarian (pesco-vegetarian) diets include fish, and vegan diets exclude all animal foods. A meta-analysis (a study that pools the results of earlier studies) of 86 cross-sectional and 10 prospective studies found that vegetarians and vegans had lower body mass index, total and LDL cholesterol and blood glucose than omnivores. In the prospective studies, vegetarians had 25% lower incidence or mortality from ischaemic heart disease (relative risk 0.75) and 8% lower total cancer incidence, but no significant difference in all-cause mortality, total cardiovascular disease or cancer mortality [5].

The two largest studies of vegetarian groups reached different conclusions about lifespan. In the Adventist Health Study 2, which followed 73,308 Seventh-day Adventists in North America for a mean of 5.79 years, vegetarians as a whole had 12% lower all-cause mortality than non-vegetarians; the lowest mortality was in pesco-vegetarians (19% lower), and the associations were stronger in men than in women [15]. In two pooled UK studies of 60,310 people, which recruited health-conscious meat eaters as the comparison group, there was no significant difference in all-cause mortality between regular meat eaters, low meat eaters, fish eaters and vegetarians [16]. The US study of plant-based diet quality showed that plant-based diets can raise or lower coronary heart disease risk depending on whether they are built on whole or refined foods [3].

Diets that exclude animal foods carry specific nutrient risks. Vitamin B12 occurs naturally only in animal foods; a review of 18 studies found B12 deficiency in 62% of vegetarian pregnant women, 25 to almost 86% of children, 21 to 41% of adolescents and 11 to 90% of older adults, with higher rates in vegans than in other vegetarians [17]. In the EPIC-Oxford study of 54,898 people followed for an average of 17.6 years, vegans had 43% higher risk of total fractures and 2.31 times the risk of hip fracture of meat eaters, equal to 14.9 more hip fractures per 1,000 people over 10 years; the associations weakened slightly but remained after adjustment for calcium and protein intake [7].

Omnivorous diets vary widely in quality. In individual-level data from six US studies of 29,682 adults followed for a median of 19 years, each additional two servings a week of processed meat was associated with 7% higher risk of cardiovascular disease, unprocessed red meat with 3% higher risk and poultry with 4% higher risk, while fish showed no association [18].

Full deep dive: Vegan, vegetarian, omnivore and carnivore diets.

Low-carbohydrate diets replace carbohydrate with fat, protein or both. In the Atherosclerosis Risk in Communities study, 15,428 US adults aged 45 to 64 were followed for a median of 25 years, and the relationship between carbohydrate intake and mortality was U-shaped, with the lowest risk at 50 to 55% of energy from carbohydrate. In a meta-analysis combining this study with seven others (432,179 participants), both low intake (under 40% of energy) and high intake (over 70%) were associated with higher mortality than moderate intake. The source of the replacement mattered: replacing carbohydrate with animal fat or protein was associated with 18% higher mortality, and replacing it with plant fat or protein with 18% lower mortality [19].

The carnivore diet consists of animal foods only and excludes most or all plant foods, and with them nearly all dietary fiber. No prospective study has followed its health outcomes. The largest description is a 2020 social media survey of 2,029 self-identified followers, who had eaten the diet for a median of 14 months and reported high satisfaction and few adverse symptoms; among those who reported blood tests, median LDL cholesterol was markedly raised at 172 mg/dL [20]. The survey was self-selected and had no follow-up for disease or death.

Full deep dive: Vegan, vegetarian, omnivore and carnivore diets.

Most evidence on dietary patterns is observational and depends on self-reported food frequency questionnaires or dietary records, which measure habitual intake imprecisely. People who follow a given diet often differ from others in smoking, alcohol use, physical activity, income and health consciousness, and these differences can be only partly adjusted for. This is one explanation for the contrast between the Adventist study, whose participants share a health-focused religious lifestyle, and the UK studies, which deliberately recruited health-conscious meat eaters [15, 16]. Randomised trials with hard outcomes are rare, and the main one, PREDIMED, required reanalysis after randomisation problems [2]. Modelled life expectancy gains rest on the assumption that the associations in observational studies are causal [8]. Related conditions are covered under cardiovascular disease, colon health, body weight and metabolic health and type 2 diabetes.