research paper for vegetarian

• Research article
• Open access
• Published: 23 November 2020

Vegetarian and vegan diets and risks of total and site-specific fractures: results from the prospective EPIC-Oxford study

• Tammy Y. N. Tong   ORCID: orcid.org/0000-0002-0284-8959 1 ,
• Paul N. Appleby 1 ,
• Miranda E. G. Armstrong 2 ,
• Georgina K. Fensom 1 ,
• Anika Knuppel 1 ,
• Keren Papier 1 ,
• Aurora Perez-Cornago 1 ,
• Ruth C. Travis 1 &
• Timothy J. Key 1  

BMC Medicine volume  18 , Article number:  353 ( 2020 ) Cite this article

196k Accesses

85 Citations

1825 Altmetric

Metrics details

There is limited prospective evidence on possible differences in fracture risks between vegetarians, vegans, and non-vegetarians. We aimed to study this in a prospective cohort with a large proportion of non-meat eaters.

In EPIC-Oxford, dietary information was collected at baseline (1993–2001) and at follow-up (≈ 2010). Participants were categorised into four diet groups at both time points (with 29,380 meat eaters, 8037 fish eaters, 15,499 vegetarians, and 1982 vegans at baseline in analyses of total fractures). Outcomes were identified through linkage to hospital records or death certificates until mid-2016. Using multivariable Cox regression, we estimated the risks of total ( n  = 3941) and site-specific fractures (arm, n  = 566; wrist, n  = 889; hip, n  = 945; leg, n  = 366; ankle, n  = 520; other main sites, i.e. clavicle, rib, and vertebra, n  = 467) by diet group over an average of 17.6 years of follow-up.

Compared with meat eaters and after adjustment for socio-economic factors, lifestyle confounders, and body mass index (BMI), the risks of hip fracture were higher in fish eaters (hazard ratio 1.26; 95% CI 1.02–1.54), vegetarians (1.25; 1.04–1.50), and vegans (2.31; 1.66–3.22), equivalent to rate differences of 2.9 (0.6–5.7), 2.9 (0.9–5.2), and 14.9 (7.9–24.5) more cases for every 1000 people over 10 years, respectively. The vegans also had higher risks of total (1.43; 1.20–1.70), leg (2.05; 1.23–3.41), and other main site fractures (1.59; 1.02–2.50) than meat eaters. Overall, the significant associations appeared to be stronger without adjustment for BMI and were slightly attenuated but remained significant with additional adjustment for dietary calcium and/or total protein. No significant differences were observed in risks of wrist or ankle fractures by diet group with or without BMI adjustment, nor for arm fractures after BMI adjustment.

Conclusions

Non-meat eaters, especially vegans, had higher risks of either total or some site-specific fractures, particularly hip fractures. This is the first prospective study of diet group with both total and multiple specific fracture sites in vegetarians and vegans, and the findings suggest that bone health in vegans requires further research.

Peer Review reports

Fractures in adulthood and older ages are a common occurrence which pose a significant burden to health systems worldwide [ 1 ]. Previous epidemiological studies have shown that vegetarians had lower bone mineral density (BMD) than non-vegetarians [ 2 , 3 ], but the associations of vegetarian diets with fracture risks are unclear. Potential risk differences are plausible however, owing to differences in several dietary factors, such as the substantially lower intakes of calcium in vegans [ 4 , 5 ], lower intakes of dietary protein in both vegetarians and vegans [ 6 , 7 ], and the lower body mass index (BMI) of non-meat eaters [ 2 , 8 ].

Prior studies have linked both calcium and protein intakes to bone health, but their relationships with fracture risks are nuanced. For calcium, although previous meta-analyses have found that calcium supplements are effective in producing small increases in BMD [ 9 ], it is less clear whether this degree of improvement would be sufficient to reduce fracture risks [ 10 ]. However, a recent meta-analysis of randomised trials showed that combined vitamin D and calcium supplementation, but not vitamin D supplementation alone, was effective in fracture prevention, therefore supporting the importance of calcium [ 11 ]. For protein, while older studies suggested that high protein intake might lead to higher calcium excretion and therefore weaker bones [ 12 ], more recent evidence has suggested a positive association between protein and bone health, although this might not translate to differences in fracture risk [ 13 ]. In addition, BMI is also an important factor for fracture risk [ 14 ], and a recent study suggested that the lower BMD observed in US vegetarians might be largely explained by their lower BMI and waist circumference [ 15 ]. However, the directions of association between BMI and fracture risk differ across fracture sites, and low BMI has been associated with a higher risk of hip fracture but lower risk of ankle fracture [ 14 ].

The largest study to date on vegetarian diet group and fracture risks came from previous analyses in EPIC-Oxford on around 30,000 participants, and reported that vegans, but not vegetarians, had higher risks of total fractures, although this analysis had a short follow-up (5 years) and relied on self-reported outcome data [ 16 ]. The only two other studies on the topic included a small number of participants and did not report on site-specific fractures [ 17 , 18 ]. Hence, the possible differences in fracture risks by vegetarian diet groups are still unclear, and it is not known whether the risks might differ by fracture sites.

Therefore, the aim of this study was to examine the risks of total and site-specific fractures in a prospective cohort with close to 18 years of average follow-up, including a large proportion of non-meat eaters, and with outcome data based on record linkage.

Study population

EPIC-Oxford is a prospective cohort study which recruited approximately 65,000 men and women across the UK between 1993 and 2001, via either general practices or by postal questionnaire. Details of the recruitment process and eligibility criteria for inclusion in the analyses can be found in Additional File  1 : Supplementary methods [ 4 , 19 ] and in the participant flow chart (Additional File 1 : Fig. S1). The study has approval by a Multicentre Research Ethics Committee (Scotland A Research Ethics Committee). All participants provided written informed consent.

Classification of diet group

At recruitment, participants completed a questionnaire which asked about diet, socio-demographic characteristics, lifestyle, and medical history. A follow-up questionnaire which asked similar questions was sent to participants in 2010. Based on the responses to both questionnaires (if the participant completed the follow-up questionnaire), the participants were categorised into meat eaters, fish eaters (did not eat meat but ate fish), vegetarians (did not eat meat or fish, but ate one or both of dairy or eggs), and vegans (participants who did not eat meat, fish, dairy, and eggs) at both time points. Further details on the questionnaires, classification of diet group including agreement of diet group at baseline and follow-up, and data collection of other baseline characteristics can be found in Additional File 1 : Supplementary methods [ 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 ].

Outcome assessment

Participants were followed up for health outcomes via record linkage to National Health Service records until 31 March 2016 in England, 31 May 2016 in Wales, and 31 October 2016 in Scotland. The outcomes of interest were the first recorded hospital admission (inpatient admissions in England, inpatient admissions and day cases in Wales and Scotland) or death from total and site-specific fractures, including fractures of the arm (i.e. humerus, radius, and ulna), wrist, hip, leg (i.e. femur [excluding neck of femur], patella, tibia, and fibula), ankle, and other main sites (i.e. clavicle, rib, or vertebra), identified by the relevant 9th or 10th revisions of the World Health Organization’s International Classification of Diseases (ICD-9/ICD-10) codes (Additional File 1 : Table S1). For total fractures, incidence was defined as the first recorded occurrence of any diagnosis of any fracture; for site-specific fractures, incidence was defined as the first recorded occurrence of any fracture at that particular site, without censoring for previous fractures at other sites. Fractures at the clavicle, rib, and vertebra were examined as one composite outcome due to the small number of cases at these sites, but the three sites were examined separately in secondary analyses.

Statistical analyses

Baseline characteristics and food and nutrient intakes of the EPIC-Oxford participants were summarised by diet group. Cox proportional hazards regression models were used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations between the four diet groups (meat eaters, fish eaters, vegetarians, vegans) and total and each site-specific fracture of interest, using meat eaters as the reference group. The underlying time variable was the age at recruitment to the age at diagnosis, death, or administrative censoring, whichever occurred first. For participants who completed both the baseline and follow-up questionnaires, diet group and relevant time-varying covariates (smoking and alcohol consumption, BMI, dietary calcium or protein) were updated at follow-up; otherwise, the baseline dietary or covariate information was carried forward.

All analyses were stratified by sex, method of recruitment, and region of residence, and adjusted for year of recruitment, ethnicity, Townsend deprivation index [ 25 ], education level, physical activity [ 26 ], smoking, alcohol consumption, dietary supplement use, height, and in women menopausal status, hormone replacement therapy use, and parity. We tested models with and without adjustment for BMI. Details on the categorisation of covariates can be found in the Supplementary methods. The proportional hazards assumption was assessed on the basis of Schoenfeld residuals and was not violated for the variables of interest in the adjusted model for any of the outcomes. Subsequently, we estimated absolute rate differences based on the BMI adjusted model, using a previously reported method [ 28 ].

To evaluate the influence of dietary calcium and protein on the associations, we included models further adjusting for either dietary calcium or dietary protein intake, and simultaneously adjusting for both variables. Additional analyses were also performed limited to people with sufficient dietary calcium (≥ 700 mg/day) or dietary protein intake (≥ 0.75 g of protein per day/kg body weight) in accordance with UK dietary guidelines [ 29 , 30 ].

As sensitivity analyses, we repeated the analyses (with adjustment for BMI) further adjusting for energy intake, excluding the first 5 years of follow-up, excluding participants with prior diseases (baseline history of diabetes, heart disease, stroke, or cancer), excluding participants who were receiving long-term treatment for any illness, and with multiple imputation for missing covariates [ 31 ]. Heterogeneity of results by age at recruitment (below and above age 50), sex, menopausal status, physical activity level (inactive/low and moderate/high activity), and BMI (below and above 22.5 kg/m 2 ) was assessed for total and hip fractures, which had the largest numbers of cases. Cut-offs of age and BMI were chosen to ensure a reasonable distribution of number of cases in categories across all diet groups, based on analyses of total fractures.

All analyses were performed using Stata version 15.1 (StataCorp, TX, USA), and 2-sided p values < 0.05 were considered significant. The forest plot was generated using R (R Foundation for Statistical Computing, Vienna, Austria).

The study population included a minimum of 54,898 participants (in analyses for total fractures), of whom 30,391 had repeated measures of diet 14 years later (details in Additional File 1 : Fig. S1). Baseline characteristics in the overall cohort are tabulated by the four diet groups in Table  1 , and separately for men and women in Additional File 1 : Table S2. Other dietary and nutrient intakes are tabulated by the four diet groups, separately for men and women in Additional File 1 : Table S3. A summary description of the baseline and dietary characteristics can be found in Additional File 1 : Supplementary results.

Over an average of 17.6 years of follow-up, we observed 3941 cases of total fractures (including 12 first reported at death; 943,934 person-years), 566 arm fractures (1 at death; 967,829 person-years), 889 wrist fractures (965,127 person-years), 945 hip fractures (1 at death; 967,599 person-years), 366 leg fractures (1 at death; 968,985 person-years), 520 ankle fractures (967,399 person-years), and 467 other main site fractures (968,921 person-years). The results of longitudinal associations between diet group and total and site-specific fractures are shown in Fig.  1 and Table  2 . Absolute rate differences (AD) in the outcomes by diet group based on the BMI adjusted model are shown in Table  3 .

Risks of total and site-specific fractures by diet group in EPIC-Oxford. Estimates also shown in Table  2 as model 2. All analyses were stratified by sex, method of recruitment (general practice or postal), and region (7 categories), and adjusted for year of recruitment (per year from ≤ 1994 to ≥ 1999), ethnicity (white, other, unknown), Townsend deprivation index (quartiles, unknown), education level (no qualifications, basic secondary (e.g. O level), higher secondary (e.g. A level), degree, unknown), physical activity (inactive, low activity, moderately active, very active, unknown), smoking (never, former, light, heavy, unknown), alcohol consumption (< 1 g, 1–7 g, 8–15 g, 16+ g/day), dietary supplement use (no, yes, unknown), height (5 cm categories from < 155 to ≥ 185 cm, unknown), body mass index (< 18.5, 18.5–19.9, 20–22.4, 22.5–24.9, 25–27.4, 27.5–29.9, 30–32.4, ≥ 32.5 kg/m 2 , unknown), and in women menopausal status (premenopausal, perimenopausal, postmenopausal, unknown), hormone replacement therapy use (never, ever, unknown), and parity (none, 1–2, ≥ 3, unknown). Other main site fractures are defined as fractures of the clavicle, rib, or vertebra

Compared with meat eaters, vegetarians (HR 1.11; 95% CI 1.02, 1.21) and vegans (1.50; 1.26, 1.78) had higher risks of total fractures after adjustment for confounders (Table  2 model 1). The associations attenuated with additional adjustment of BMI (vegetarians—1.09; 1.00, 1.19; vegans—1.43; 1.20, 1.70), but remained clearly significant in vegans (Table  2 model 2, Fig.  1 ). The equivalent rate differences were 4.1 (0.8, 7.6) more cases in vegetarians and 19.4 (9.6, 30.9) more cases in vegans for every 1000 people over 10 years. The associations were attenuated further but remained significant in vegans with additional adjustment for dietary calcium (1.31; 1.10, 1.57, Table  2 model 3), total dietary protein (1.39; 1.16, 1.67, Table  2 model 4), or both dietary factors simultaneously (1.30; 1.08, 1.56, Table  2 model 5).

For site-specific fractures (Fig.  1 and Table  2 ), the largest magnitudes in risk difference by diet group were observed for hip fractures. After adjustment for BMI, the risks were higher in fish eaters (HR 1.26; 1.02, 1.54, or AD 2.9; 0.6, 5.7), vegetarians (HR 1.25; 1.04, 1.50, or AD 2.9; 0.9, 5.2), and vegans (HR 2.31; 1.66, 3.22, or AD 14.9; 7.9, 24.5) than meat eaters. Similar to the findings for total fractures, the associations appeared stronger before BMI adjustment and attenuated but remained strongly significant in vegans after further adjustment for both calcium and protein.

For the other sites, after adjustment for BMI, the vegans had a higher risk of leg fractures (2.05; 1.23, 3.41) and other main site fractures (clavicle, rib, vertebra, 1.59; 1.02, 2.50) than the meat eaters (Fig.  1 and Table  2 ). When the other main site fractures were examined separately, a significantly higher risk was observed in the vegans for vertebral fracture (2.42; 2.31, 4.48), but not for the other two sites (Additional File 1 : Table S4). No significant differences in risks between diet groups were observed for arm, wrist, or ankle fracture, after adjustment for BMI (Fig.  1 and Table  2 ), although a higher risk of arm fractures was observed in both vegetarians (1.28; 1.03, 1.60) and vegans (1.67; 1.07, 2.61) in the multivariable model before BMI adjustment (Table  2 model 1).

Results from secondary analyses are reported in more detail in the Supplementary results. Overall, results were consistent when the analyses were restricted to participants with sufficient intakes of calcium and protein (Table  4 ), and also in other secondary analyses, including with further adjustment for energy intake, excluding the first 5 years of follow-up, excluding participants with prior diseases or receiving long-term treatment for any illness, or with multiple imputation for missing covariates (Additional File 1 : Table S5).

In stratified analyses of total (Table  5 ) and hip fractures (Additional File 1 : Table S6), a significantly higher risk of both total and hip fractures was only observed in vegetarians over age 50 at recruitment, although vegans had higher risks in both age groups, and a significant p for interaction was only observed for total fractures. For both types of fractures, the significant associations in vegans appeared stronger in women, particularly those who were postmenopausal, and participants with low physical activity and lower BMI, possibly partly due to the larger number of participants in most of these subgroups, but a higher risk of hip fracture was only observed in the fish eaters and vegetarians in the higher BMI category. Because the numbers of cases in these subgroup analyses were often very small, it is likely that we did not have sufficient power to identify possible differences.

Summary of findings

Overall, vegans in this study had higher risks of total and some site-specific fractures (hip, leg, vertebra) than meat eaters. The strongest associations were observed for hip fractures, for which fish eaters, vegetarians, and vegans all had higher risks. These risk differences might be partially explained by the lower average BMI, and lower average intakes of calcium and protein in the non-meat eaters. However, because the differences remained, especially in vegans, after accounting for these factors, other unaccounted for factors may be important.

Comparison with previous studies

Few previous studies have examined the associations of vegetarian diets with fracture risk. In previous EPIC-Oxford analyses of self-reported fractures with short follow-up, vegans, but not fish eaters or vegetarians, were reported to have 30% (HR 1.30; 1.02, 1.66) higher risks of total fractures, but in contrast to the current findings, the association attenuated completely when restricted to participants who reported consuming at least 525 mg/day of calcium [ 16 ]. This apparent inconsistency might be explained by several differences between the current and previous analysis; while the current analysis included close to 4000 hospital-admitted cases over more than 17 years of average follow-up on around 55,000 participants, the previous study included under 2000 self-reported fracture cases over 5 years of follow-up on around 35,000 participants. Given the difference in case ascertainment method, the current analysis is less prone to reporting error and is not susceptible to selective drop-out. It is also possible that there was insufficient power to detect a difference after stratifying by calcium intake status in the previous analysis, which also did not examine site-specific fractures.

The only other studies which reported on risks of fractures by diet groups were one small prospective study in Vietnam of 210 women (105 vegans) which found no significant difference in fracture incidence (10 cases in total) between vegans and omnivores over 2 years [ 17 ], and one prospective study in India which reported a higher crude rate of stress fractures (604 cases in total) among 2131 vegetarian than 6439 non-vegetarian army recruits [ 18 ]. Separately, previous findings from the Adventist Health Study 2, which has a large proportion of vegetarians, showed that participants who ate meat more than three times a week had lower risks of hip fractures (HR 0.60; 0.41, 0.87) than participants who ate meat less than once a week [ 32 ], while combined analyses of peri- and postmenopausal women from Adventist Health Study 1 and 2 found that participants who ate meat more than four times a week had lower risks of wrist fractures (HR 0.44; 0.23, 0.84) than participants who never ate meat [ 33 ], but these results cannot be used to infer risks in fish eaters, vegetarians, or vegans as separate diet groups.

Interpretation of results and implications

The higher observed risks of fractures in non-meat eaters were usually stronger before BMI adjustment, which suggests that the risk differences were likely partially due to differences in BMI. Vegetarians and vegans generally have lower BMI than meat eaters [ 2 , 8 ], and previous studies have reported an inverse association between BMI and some fractures, particularly hip fractures, possibly due to reasons including the cushioning against impact force during a fall, enhanced oestrogen production with increased adiposity, or stronger bones from increased weight-bearing [ 14 , 34 ]. However, a positive association between BMI and fracture risk has been observed for some other sites, including ankle fractures, possibly as a result of higher torques from twisting of the ankle in people with higher BMI [ 14 ]. No significant differences in the risks of ankle fractures by diet group were observed in our study, but the point estimates were directionally consistent with a lower risk in all non-meat eaters before BMI adjustment, and the results might reflect a counterbalance between a protective effect from lower BMI but higher risk due to lower intakes of nutrients related to bone health in the non-meat eaters.

In our stratified analyses, there is limited evidence of heterogeneity in fracture risk by BMI categories. Although a statistically significant higher risk of total and hip fractures was only observed in vegans in the lower BMI category (< 22.5 kg/m 2 ), our interpretation is limited by the small numbers of cases in each stratum in these analyses, especially because of the strong correlation between diet group and BMI, which results in very few vegans in the higher BMI category, and vice versa comparatively small numbers of meat eaters with a low BMI. In addition to BMI, previous studies have reported that muscle strength is an important risk factor which is protective against fall risk and subsequently fractures in older adults [ 35 ]. A previous study in the UK found lower lean mass and grip strength in vegetarians and vegans compared to meat eaters [ 2 ]; therefore, the possible influences of muscle strength and fall risk in addition to bone health on fracture risk in vegetarian and vegan populations should be further investigated. Fractures at some sites, especially at the hip, may also be more related to osteoporosis than fractures at some other sites, which might be more likely to be the result of violent impacts in accidents [ 36 , 37 ]. We were unable to differentiate fragility and traumatic fractures in this study, since data were not available on the causes of the fractures.

In this study and previous studies, vegans had substantially lower intakes of calcium than other diet groups since they do not consume dairy, a major source of dietary calcium [ 4 , 5 ], while both vegetarians and vegans had lower protein intakes on average [ 6 , 7 ]. In the human body, 99% of calcium is present in bones and teeth in the form of hydroxyapatite, which in cases of calcium deficiency gets resorbed to maintain the metabolic calcium balance, and thus, osteoporosis could occur if the calcium was not restored [ 38 , 39 , 40 ]. A recent meta-analysis reported that increasing calcium intake from either dietary sources or supplements resulted in small increases in BMD [ 9 ], but the evidence on fracture risk has been less consistent. Previous analyses in EPIC-Oxford found a higher risk of self-reported fractures in women, but not men, with calcium intakes below 525 mg/day compared with over 1200 mg/day [ 41 ]. A recent meta-analysis of both randomised trials and prospective studies concluded that there was no evidence of an association between calcium intake from diet and fracture risk, but a possible weak protective association between calcium supplement use and some fractures [ 10 ]. More recently however, a separate meta-analysis showed a protective effect against fractures of combined vitamin D and calcium supplements, but not vitamin D supplements alone [ 11 ].

For protein, some older studies suggested that excessive protein intake would lead to an increased metabolic acid load, subsequently buffered by bone resorption and calciuria, and thus poorer bone health [ 12 , 42 ]. However, more recent experimental evidence has shown that high protein intake also increases intestinal calcium absorption [ 43 ], and stimulates the production of insulin-like growth factor (IGF)-I [ 44 ], which in turn is associated with better bone health [ 45 , 46 ]. Two meta-analyses, which included different studies, both reported a possible protective effect of higher protein intake on lumbar spine BMD [ 13 , 47 ]; several epidemiological studies have reported inverse associations between protein intake and fracture risks [ 48 , 49 , 50 ], though a recent meta-analysis found no significant association between protein intake and osteoporotic fractures [ 51 ].

The higher risks of fractures especially in the vegans remained significant after adjustment for dietary calcium and protein, which suggests that these factors may at most only partly explain the differences in fracture risks by diet group, and other factors may also contribute. However, estimation of intakes of these nutrients by questionnaires has substantial error, and we were only able to account for differences in dietary calcium but not differences in calcium supplement use, since data on the latter were not available. A detailed analysis of the associations of specific foods, such as meat or dairy, with fracture risk is beyond the scope of the current study, but should be explored in further studies. Future research should also focus on possible effects of other nutrients or biological markers on fracture risks, for example circulating vitamin D, vitamin B 12 , or IGF-I, which may vary by degree of animal-sourced food intake [ 52 , 53 , 54 ]. The value of incorporating habitual dietary habits in addition to established parameters for predicting fracture risks in clinical settings should also be further explored.

Strengths and limitations

The strengths of this study were that it included a large number of non-meat eaters with a long follow-up, and studied both total and site-specific fractures, after accounting for a range of confounders. We updated diet group and relevant confounders where possible, to account for changes over the period of follow-up. There was little evidence of reverse causality, as results were similar after excluding the first 5 years of follow-up. The outcome data were ascertained based on hospital records, which reduced misreporting and selective loss to follow-up, although a possible limitation of this approach was that less serious fractures that did not require hospitalisation would not have been captured.

Of other limitations, while we excluded known cases of fractures before baseline based on hospital records, this may not be a complete exclusion, since no questions on previous diagnosis of fractures (prior to the earliest available hospital data) or osteoporosis were asked at baseline, and no data on the use of anti-osteoporosis medication were available. Repeat measures of diet were not available in all participants, and the exact date of dietary change during follow-up was also not recorded, but considering the good agreement of diet group in participants who did provide a repeat measure, and the fact that a dietary change may only influence fracture risk after a period of time, we do not expect substantial misclassification. As with all observational studies, residual confounding from both dietary and non-dietary factors may be present; for example, the role of calcium might have been underestimated due to measurement error. As the study predominantly includes white European participants, generalisability to other populations or ethnicities may be limited, which could be important considering previously observed differences in BMD [ 2 , 55 ] and fracture risks [ 56 ] by ethnicity. We also observed only a small number of cases in many subgroup analyses, and thus, it is likely we had insufficient power to reliably assess whether there might be any heterogeneity by these subgroups including age, sex, menopausal status, or BMI; additional data are therefore needed to confirm or refute possible differences. In particular, because the EPIC-Oxford cohort consists predominantly of women (77%), further work should be conducted in cohorts with a larger proportion of men to explore heterogeneity by sex and to derive reliable sex-specific estimates.

Overall, we found that compared with meat eaters, vegans had higher risks of total, hip, leg, and vertebral fractures, while fish eaters and vegetarians had higher risk of hip fractures. These risk differences were likely partly due to their lower BMI, and possibly to lower intakes of calcium and protein. More studies are needed especially from non-European and contemporary populations to examine the generalisability of our findings and to explore possible heterogeneity by factors including age, sex, menopausal status, and BMI. Future work might benefit from examining possible biological pathways by investigating serum levels of vitamin D, vitamin B 12 , or IGF-1, or in assessing the possible roles of other nutrients that are abundant in animal-sourced foods.

Availability of data and materials

The data access policy for the EPIC-Oxford study is available via the study website ( www.epic-oxford.org/data-access-sharing-and-collaboration/ ).

Abbreviations

Bone mineral density

• Body mass index

European Prospective Investigation into Cancer and Nutrition

International Classification of Diseases

Insulin-like growth factor-1

Odén A, McCloskey EV, Kanis JA, Harvey NC, Johansson H. Burden of high fracture probability worldwide: secular increases 2010–2040. Osteoporos Int. 2015;26:2243–8.

Google Scholar  

Tong TY, Key TJ, Sobiecki JG, Bradbury KE. Anthropometric and physiologic characteristics in white and British Indian vegetarians and nonvegetarians in the UK Biobank. Am J Clin Nutr. 2018;107:909–20. https://doi.org/10.1093/ajcn/nqy042 .

Article   Google Scholar  

Ho-Pham LT, Nguyen ND, Nguyen TV. Effect of vegetarian diets on bone mineral density: a Bayesian meta-analysis. Am J Clin Nutr. 2009;90:943–50. https://doi.org/10.3945/ajcn.2009.27521 .

Article   CAS   Google Scholar  

Davey GK, Spencer EA, Appleby PN, Allen NE, Knox KH, Key TJ. EPIC-Oxford: lifestyle characteristics and nutrient intakes in a cohort of 33 883 meat-eaters and 31 546 non meat-eaters in the UK. Public Health Nutr. 2003;6:259–68. https://doi.org/10.1079/PHN2002430 .

Sobiecki JG, Appleby PN, Bradbury KE, Key TJ. High compliance with dietary recommendations in a cohort of meat eaters, fish eaters, vegetarians, and vegans: results from the European Prospective Investigation into Cancer and Nutrition-Oxford study. Nutr Res. 2016;36:464–77. https://doi.org/10.1016/j.nutres.2015.12.016 .

Papier K, Tong TY, Appleby PN, Bradbury KE, Fensom GK, Knuppel A, et al. Comparison of major protein-source foods and other food groups in meat-eaters and non-meat-eaters in the EPIC-Oxford cohort. Nutrients. 2019;11:1–18. https://doi.org/10.3390/nu11040824 .

Bradbury KE, Tong TYN, Key TJ. Dietary intake of high-protein foods and other major foods in meat-eaters, poultry-eaters, fish-eaters, vegetarians, and vegans in UK Biobank. Nutrients. 2017;9:1317. https://doi.org/10.3390/nu9121317 .

Spencer EA, Appleby PN, Davey GK, Key TJ. Diet and body mass index in 38 000 EPIC-Oxford meat-eaters, fish-eaters, vegetarians and vegans. Int J Obes. 2003;27:728–34. https://doi.org/10.1038/sj.ijo.0802300 .

Tai V, Leung W, Grey A, Reid IR, Bolland MJ. Calcium intake and bone mineral density: systematic review and meta-analysis. BMJ. 2015;351:h4183. https://doi.org/10.1136/bmj.h4183 .

Bolland MJ, Leung W, Tai V, Bastin S, Gamble GD, Grey A, et al. Calcium intake and risk of fracture: systematic review. BMJ. 2015;351:h4580. https://doi.org/10.1136/bmj.h4580 .

Yao P, Bennett D, Mafham M, Lin X, Chen Z, Armitage J, et al. Vitamin D and calcium for the prevention of fracture: a systematic review and meta-analysis. JAMA Netw Open. 2019;2:e1917789.

Allen LH, Oddoye EA, Margen S. Protein-induced hypercalciuria: a longer term study. Am J Clin Nutr. 1979;32:741–9. https://doi.org/10.1093/ajcn/32.4.741 .

Darling AL, Millward DJ, Torgerson DJ, Hewitt CE, Lanham-New SA. Dietary protein and bone health: a systematic review and meta-analysis. Am J Clin Nutr. 2009;90:1674–92. https://doi.org/10.3945/ajcn.2009.27799 .

Armstrong MEG, Cairns BJ, Banks E, Green J, Reeves GK, Beral V. Different effects of age, adiposity and physical activity on the risk of ankle, wrist and hip fractures in postmenopausal women. Bone. 2012;50:1394–400. https://doi.org/10.1016/j.bone.2012.03.014 .

Karavasiloglou N, Selinger E, Gojda J, Rohrmann S, Kühn T. Differences in bone mineral density between adult vegetarians and nonvegetarians become marginal when accounting for differences in anthropometric factors. J Nutr. 2020;:1–6. https://doi.org/10.1093/jn/nxaa018 .

Appleby P, Roddam A, Allen N, Key T. Comparative fracture risk in vegetarians and nonvegetarians in EPIC-Oxford. Eur J Clin Nutr. 2007;61:1400–6. https://doi.org/10.1038/sj.ejcn.1602659 .

Ho-Pham LT, Vu BQ, Lai TQ, Nguyen ND, Nguyen TV. Vegetarianism, bone loss, fracture and vitamin D: a longitudinal study in Asian vegans and non-vegans. Eur J Clin Nutr. 2012;66:75–82. https://doi.org/10.1038/ejcn.2011.131 .

Dash N, Kushwaha A. Stress fractures-a prospective study amongst recruits. Med J Armed Forces India. 2012;68:118–22. https://doi.org/10.1016/S0377-1237(12)60021-5 .

Appleby PN, Thorogood M, Mann JI, Key TJ. The Oxford Vegetarian Study: an overview. Am J Clin Nutr. 1999;70(3 Suppl):525S–31S http://www.ncbi.nlm.nih.gov/pubmed/10479226 .

CAS   Google Scholar  

Bingham SA, Gill C, Welch A, Day K, Cassidy A, Khaw KT, et al. Comparison of dietary assessment methods in nutritional epidemiology: weighed records v. 24 h recalls, food-frequency questionnaires and estimated-diet records. Br J Nutr. 1994;72:619. https://doi.org/10.1079/BJN19940064 .

Bingham S, Cassidy A, Cole TJ, Welch A, Runswick S, Black a E, et al. Validation of weighed records and other methods of dietary assessment using the 24 h urine nitrogen technique and other biological markers. Br J Nutr. 1995;73:531–50. https://doi.org/10.1079/BJN19950057 .

Bingham SA, Gill C, Welch A, Cassidy A, Runswick SA, Oakes S, et al. Validation of dietary assessment methods in the UK arm of EPIC using weighed records, and 24-hour urinary nitrogen and potassium and serum vitamin C and carotenoids as biomarkers. Int J Epidemiol. 1997;26(Suppl 1):S137–51.

Ministry of Agriculture Fisheries and Food. Food portion sizes. 2nd ed. London: Her Majesty’s Stationary Office; 1993.

Holland B, Welch A, Unwin I, Buss D, Paul A, Southgate DAT. McCance and Widdowson’s the composition of foods. 5th ed. Cambridge: Royal Society of Chemistry; 1991.

Townsend P. Poverty in the United Kingdom: a survey of household resources and standards of living. London, UK; 1979.

Wareham NJ, Jakes RW, Rennie KL, Schuit J, Mitchell J, Hennings S, et al. Validity and repeatability of a simple index derived from the short physical activity questionnaire used in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Public Health Nutr. 2003;6:407–13. https://doi.org/10.1079/PHN2002439 .

Spencer EA, Appleby PN, Davey GK, Key TJ. Validity of self-reported height and weight in 4808 EPIC–Oxford participants. Public Health Nutr. 2002;5:561–5. https://doi.org/10.1079/PHN2001322 .

Tong TYN, Appleby PN, Bradbury KE, Perez-Cornago A, Travis RC, Clarke R, et al. Risks of ischaemic heart disease and stroke in meat eaters, fish eaters, and vegetarians over 18 years of follow-up: results from the prospective EPIC-Oxford study. BMJ. 2019;366:l4897. https://doi.org/10.1136/bmj.l4897 .

Public Health England. Government Dietary Recommendations: Government recommendations for food energy and nutrients for males and females About Public Health England. 2016. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/618167/government_dietary_recommendations.pdf .

Department of Health. Dietary reference values: a guide. 1991. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/743790/Dietary_Reference_Values_-_A_Guide__1991_.pdf .

White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30:377–99. https://doi.org/10.1002/sim.4067 .

Lousuebsakul-Matthews V, Thorpe DL, Knutsen R, Beeson WL, Fraser GE, Knutsen SF. Legumes and meat analogues consumption are associated with hip fracture risk independently of meat intake among Caucasian men and women: the Adventist Health Study-2. Public Health Nutr. 2013;17:2333–43.

Thorpe DL, Knutsen SF, Lawrence Beeson W, Rajaram S, Fraser GE. Effects of meat consumption and vegetarian diet on risk of wrist fracture over 25 years in a cohort of peri- and postmenopausal women. Public Health Nutr. 2008;11:564–72.

De Laet C, Kanis JA, Odén A, Johanson H, Johnell O, Delmas P, et al. Body mass index as a predictor of fracture risk: a meta-analysis. Osteoporos Int. 2005;16:1330–8.

Moreland JD, Richardson JA, Goldsmith CH, Clase CM. Muscle weakness and falls in older adults: a systematic review and meta-analysis. J Am Geriatr Soc. 2004;52:1121–9. https://doi.org/10.1111/j.1532-5415.2004.52310.x .

Warriner AH, Patkar NM, Curtis JR, Delzell E, Gary L, Kilgore M, et al. Which fractures are most attributable to osteoporosis? J Clin Epidemiol. 2011;64:46–53. https://doi.org/10.1016/j.jclinepi.2010.07.007 .

Appleby PN, Allen NE, Roddam AW, Key TJ. Physical activity and fracture risk: a prospective study of 1898 incident fractures among 34 696 British men and women. J Bone Miner Metab. 2008;26:191–8.

Heaney RP, Gallagher JC, Johnston CC, Neer R, Parfitt AM, Whedon GD. Calcium nutrition and bone health in the elderly. Am J Clin Nutr. 1982;36:986–1013. https://doi.org/10.1093/ajcn/36.5.986 .

Nordin BE. Calcium and osteoporosis. Nutrition. 1997;13:664–86. https://doi.org/10.1016/S0899-9007(97)83011-0 .

Tucker KL. Vegetarian diets and bone status. Am J Clin Nutr. 2014;100(Suppl):329S–35S. https://doi.org/10.3945/ajcn.113.071621 .

Key TJ, Appleby PN, Spencer EA, Roddam AW, Neale RE, Allen NE. Calcium, diet and fracture risk: a prospective study of 1898 incident fractures among 34696 British women and men. Public Health Nutr. 2007;10:1314–20.

Barzel US, Massey LK. Excess dietary protein can adversely affect bone. J Nutr. 1998;128:1051–3. https://doi.org/10.1093/jn/128.6.1051 .

Kerstetter JE, O’Brien KO, Insogna KL. Dietary protein, calcium metabolism, and skeletal homeostasis revisited. Am J Clin Nutr. 2003;78(3 Suppl):584S–92S. https://doi.org/10.1093/ajcn/78.3.584S .

Bonjour J-P. The dietary protein, IGF-I, skeletal health axis. Horm Mol Biol Clin Investig. 2016;28:39–53.

Locatelli V, Bianchi VE. Effect of GH/IGF-1 on bone metabolism and osteoporsosis. Int J Endocrinol. 2014;2014:1–25.

Ohlsson C, Mellström D, Carlzon D, Orwoll E, Ljunggren Ö, Karlsson MK, et al. Older men with low serum IGF-1 have an increased risk of incident fractures: the MrOS Sweden study. J Bone Miner Res. 2011;26:865–72.

Shams-White MM, Chung M, Du M, Fu Z, Insogna KL, Karlsen MC, et al. Dietary protein and bone health: a systematic review and meta-analysis from the National Osteoporosis Foundation. Am J Clin Nutr. 2017;105:1528–43. https://doi.org/10.3945/ajcn.116.145110 .

Fung TT, Meyer HE, Willett WC, Feskanich D. Protein intake and risk of hip fractures in postmenopausal women and men age 50 and older. Osteoporos Int. 2017;28:1401–11.

Munger RG, Cerhan JR, Chiu BC. Prospective study of dietary protein intake and risk of hip fracture in postmenopausal women. Am J Clin Nutr. 1999;69:147–52. https://doi.org/10.1093/ajcn/69.1.147 .

Langsetmo L, Shikany JM, Cawthon PM, Cauley JA, Taylor BC, Vo TN, et al. The association between protein intake by source and osteoporotic fracture in older men: a prospective cohort study. J Bone Miner Res. 2017;32:592–600.

Darling AL, Manders RJF, Sahni S, Zhu K, Hewitt CE, Prince RL, et al. Dietary protein and bone health across the life-course: an updated systematic review and meta-analysis over 40 years. Osteoporos Int. 2019;30:741–61.

Crowe FL, Steur M, Allen NE, Appleby PN, Travis RC, Key TJ. Plasma concentrations of 25-hydroxyvitamin D in meat eaters, fish eaters, vegetarians and vegans: results from the EPIC-Oxford study. Public Health Nutr. 2011;14:340–6.

Ma J, Giovannucci E, Pollak M, Chan JM, Gaziano JM, Willett W, et al. Milk intake, circulating levels of insulin-like growth factor-I, and risk of colorectal cancer in men. JNCI J Natl Cancer Inst. 2001;93:1330–6. https://doi.org/10.1093/jnci/93.17.1330 .

Gilsing AMJ, Crowe FL, Lloyd-Wright Z, Sanders TAB, Appleby PN, Allen NE, et al. Serum concentrations of vitamin B12 and folate in British male omnivores, vegetarians and vegans: results from a cross-sectional analysis of the EPIC-Oxford cohort study. Eur J Clin Nutr. 2010;64:933–9. https://doi.org/10.1038/ejcn.2010.142 .

Popp KL, Hughes JM, Martinez-Betancourt A, Scott M, Turkington V, Caksa S, et al. Bone mass, microarchitecture and strength are influenced by race/ethnicity in young adult men and women. Bone. 2017;103:200–8. https://doi.org/10.1016/j.bone.2017.07.014 .

Barrett-Connor E, Siris ES, Wehren LE, Miller PD, Abbott TA, Berger ML, et al. Osteoporosis and fracture risk in women of different ethnic groups. J Bone Miner Res. 2005;20:185–94. https://doi.org/10.1359/JBMR.041007 .

Download references

Acknowledgements

We thank all participants in the EPIC-Oxford cohort for their invaluable contribution.

The work is supported by the UK Medical Research Council MR/M012190/1 and Wellcome Trust Our Planet Our Health (Livestock, Environment, and People, LEAP 205212/Z/16/Z). The funders had no role on the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Author information

Authors and affiliations.

Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Richard Doll Building, Old Road Campus, Oxford, OX3 7LF, UK

Tammy Y. N. Tong, Paul N. Appleby, Georgina K. Fensom, Anika Knuppel, Keren Papier, Aurora Perez-Cornago, Ruth C. Travis & Timothy J. Key

Centre for Exercise, Nutrition and Health Sciences, School for Policy Studies, University of Bristol, Bristol, UK

Miranda E. G. Armstrong

You can also search for this author in PubMed   Google Scholar

Contributions

TYNT and TJK conceived and designed the research question. TYNT analysed the data and wrote the first draft of the manuscript. All authors provided input on data analysis and interpretation of results. All authors revised the manuscript critically for important intellectual content, and read and approved the final manuscript. TYNT is the guarantor. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

Corresponding author

Correspondence to Tammy Y. N. Tong .

Ethics declarations

Ethics approval and consent to participate.

The study has approval by a Multicentre Research Ethics Committee (Scotland A Research Ethics Committee). All participants provided written informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Additional file 1:.

Supplementary results. Fig. S1. Participant flow chart. Table S1. ICD codes for incident fractures. Table S2. Baseline characteristics by diet group and sex. Table S3. Food and nutrient intake by diet group and sex. Table S4. Risks of subtypes of main site fractures. Table S5-Sensitivity analyses. Table S6. Risks of hip fractures by age, sex, menopausal status, physical activity and BMI.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Tong, T.Y.N., Appleby, P.N., Armstrong, M.E.G. et al. Vegetarian and vegan diets and risks of total and site-specific fractures: results from the prospective EPIC-Oxford study. BMC Med 18 , 353 (2020). https://doi.org/10.1186/s12916-020-01815-3

Download citation

Received : 24 July 2020

Accepted : 14 October 2020

Published : 23 November 2020

DOI : https://doi.org/10.1186/s12916-020-01815-3

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Vegetarian diets
• Vegan diets
• Bone health
• Prospective study

BMC Medicine

ISSN: 1741-7015

• Submission enquiries: [email protected]
• General enquiries: [email protected]

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 20 July 2023

Vegans, vegetarians, fish-eaters and meat-eaters in the UK show discrepant environmental impacts

• Peter Scarborough   ORCID: orcid.org/0000-0002-2378-2944 1 , 2 ,
• Michael Clark   ORCID: orcid.org/0000-0001-7161-7751 3 ,
• Linda Cobiac   ORCID: orcid.org/0000-0002-2271-6494 4 ,
• Keren Papier   ORCID: orcid.org/0000-0002-4102-6835 5 ,
• Anika Knuppel   ORCID: orcid.org/0000-0003-1049-4836   na1 ,
• John Lynch 6 ,
• Richard Harrington 1 , 2 ,
• Tim Key   ORCID: orcid.org/0000-0003-2294-307X 5 &
• Marco Springmann   ORCID: orcid.org/0000-0001-6028-5712 3  

Nature Food volume  4 ,  pages 565–574 ( 2023 ) Cite this article

146k Accesses

43 Citations

3606 Altmetric

Metrics details

• Climate-change mitigation
• Environmental impact
• Sustainability

Modelled dietary scenarios often fail to reflect true dietary practice and do not account for variation in the environmental burden of food due to sourcing and production methods. Here we link dietary data from a sample of 55,504 vegans, vegetarians, fish-eaters and meat-eaters with food-level data on greenhouse gas emissions, land use, water use, eutrophication risk and potential biodiversity loss from a review of 570 life-cycle assessments covering more than 38,000 farms in 119 countries. Our results include the variation in food production and sourcing that is observed in the review of life-cycle assessments. All environmental indicators showed a positive association with amounts of animal-based food consumed. Dietary impacts of vegans were 25.1% (95% uncertainty interval, 15.1–37.0%) of high meat-eaters (≥100 g total meat consumed per day) for greenhouse gas emissions, 25.1% (7.1–44.5%) for land use, 46.4% (21.0–81.0%) for water use, 27.0% (19.4–40.4%) for eutrophication and 34.3% (12.0–65.3%) for biodiversity. At least 30% differences were found between low and high meat-eaters for most indicators. Despite substantial variation due to where and how food is produced, the relationship between environmental impact and animal-based food consumption is clear and should prompt the reduction of the latter.

Similar content being viewed by others

Sustainability benefits of transitioning from current diets to plant-based alternatives or whole-food diets in Sweden

Poultry and fish and aquatic invertebrates have not displaced other meat sources

Organic food has lower environmental impacts per area unit and similar climate impacts per mass unit compared to conventional

The substantial impact of the global food system on the environment is well established. It is estimated that the food system was responsible for 18 Gt of carbon dioxide equivalent (CO 2 e) greenhouse gas (GHG) emissions in 2015, comprising 34% of total global GHG emissions that year 1 . The food system is also responsible for 70% of the world’s freshwater use and 78% of freshwater pollution 2 , 3 . About three quarters of the World’s ice-free land area has been affected by human use, primarily agriculture 4 , and land-use change (primarily deforestation for agriculture) is a major source of biodiversity loss 5 , 6 .

To feed a growing global population while remaining within proposed safe environmental boundaries for GHG emissions, land use, water use, water pollution and biodiversity loss, we will need changes in diets 7 . Other means to reduce the environmental impact of the food system (for example, technological advances, closing yield gaps, reducing food waste) will not be enough without major dietary change 7 , 8 . The environmental impact of animal-based foods is generally higher than for plant-based foods because of both direct processes related to livestock management (for example, methane (CH 4 ) production by ruminants) and indirect processes through the inefficiency of using crops for animal feed rather than directly for human consumption 3 , 9 , 10 . For this reason, proposed diets for global sustainable food production require most high-income countries to radically reduce consumption of animal-based foods and converge on levels that are higher than currently consumed in many low-income countries 8 .

Systematic reviews of modelled dietary scenarios have shown that vegan and vegetarian diets have substantially lower GHG emissions, land use and water use requirements than meat-containing diets 11 , 12 and that diets with reduced animal-based foods tend to be healthier and have lower environmental impact 13 . However, modelled dietary scenarios may not reflect true dietary practice, and modelled environmental and health outcomes can be strongly affected by assumptions made by the modellers. Also, previous modelled dietary scenarios have not reflected the considerable variation in environmental indicators due to both region of food production and agricultural production methods 3 and therefore will have underestimated the uncertainty associated with their findings. While we continue to use average values of environmental impact for food categories, we cannot know whether the observed differences in environmental impact between dietary groups still exist after accounting for variation in food production systems. We therefore need to link data from dietary surveys of real-life dietary patterns with large datasets of environmental indicators to ascertain whether the relationship between animal-based food consumption and environmental outcomes shown in modelling studies is robust.

Previously, we estimated the dietary GHG emissions associated with real-life diet groups in the UK 14 . These estimates only captured one aspect of the environmental impact of food systems, and the data for GHG emissions were derived from a single source with no information about variation within individual food groups due to sourcing or production 15 . Also, GHG emissions data were not presented as disaggregated gases, losing climatically important information 16 . In this paper, we link a validated food frequency questionnaire (FFQ) to estimates from a review of 570 life-cycle assessments 3 (LCAs) to estimate the GHG emissions (CH 4 , nitrous oxide (N 2 O) and carbon dioxide (CO 2 ), in addition to combined CO 2 e emissions), water use, land use, water pollution and biodiversity impact associated with observed diets of vegans, vegetarians, fish-eaters and meat-eaters in the UK (Fig. 1 ). Our approach allows for direct comparisons of the environmental indicators for each diet group, incorporating uncertainty due to food sourcing and production, and individual-level diet choice.

Flow chart shows how data from different sources have been linked for these analyses. Further information about the linkages is provided in the Supplementary Data 1 (Supplementary Section 1 ).

The participants and their dietary intake are described in Table 1 . Vegans and vegetarians were younger than fish-eaters and meat-eaters, and vegans reported a lower dietary intake of energy than all other diet groups. Fish consumption was similar in fish-eaters and low meat-eaters (with higher levels of consumption in medium and high meat-eaters), suggesting that fish-eaters were not replacing meat with fish. While total dairy consumption was lower in vegetarians and fish-eaters compared to meat-eaters, there was higher consumption of cheese in these two groups.

Estimates of environmental indicators of the diet groups are shown in Tables 2 – 4 , and relative impacts compared to the high meat-eaters are shown in Figs. 2 and 3 . The uncertainty associated with sourcing and production is highly correlated between diet groups. This is because food-level draws that produce Monte Carlo iterations with low estimates for the vegan diet group are highly likely to produce low estimates of environmental impact for all other diet groups. For this reason, the results in Tables 2 – 4 can be used to show uncertainty in absolute estimates of environmental impact for any single diet group, but for comparisons between diet groups the results in Figs. 2 and 3 should be used (which account for the correlation in the uncertainty between diet groups). The results shown in Figs. 2 and 3 represent re-analyses of the dataset and cannot simply be calculated from the data presented in Tables 2 – 4 . Tables with full results for these figures are provided in Supplementary Tables 8 – 10 .

Uncertainty intervals are 2.5th to 97.5th percentiles of a Monte Carlo analysis ( n  = 1,000).

Source data

For GHG emissions, there was a positive association with amount of animal-based food consumption (Table 2 , Fig. 2 and Supplementary Table 8 ). Dietary CO 2 emissions for vegans were 30.3% (17.0–45.5%) of the high meat-eaters group. There were also substantial differences in dietary CO 2 emissions between groups of meat-eaters. Dietary CO 2 emissions of low meat-eaters were 57.2% (37.8–74.9%) of the high meat-eaters. CH 4 is a GHG that, in terms of agricultural emissions, is predominantly associated with production of ruminants—it is therefore unsurprising to see wide disparities in CH 4 emissions associated with the different diet groups. CH 4 emissions from high meat-eaters were 15.3 (10.3–27.1) times higher than from vegan diets. N 2 O emissions are predominantly associated with fertilizer use, and therefore gradients in N 2 O emissions by diet group are mostly a result of the inefficiencies associated with raising crops for animal feed. This gradient is shallower than for CH 4 but still substantial, with N 2 O emissions for high meat-eaters 3.6 (2.4–6.0) times higher than for vegans.

Table 3 (and Supplementary Table 9 for relative differences between diet groups) show that using the 100-year Global Temperature change Potential (GTP100) measure resulted in smaller aggregated GHG footprints for all diets, as would be expected given the lower valuation of both N 2 O and especially CH 4 compared to 100-year Global Warming Potential (GWP100). The ranking of different dietary emission footprints remained the same, but the relative advantage of vegans over diets incorporating animal products decline slightly, with the high meat-eaters responsible for 3.6 (2.4–6.1) times greater GTP100 emissions than vegans, and low meat-eaters 1.8 (1.4–2.6) times greater. For the 20-year Global Warming Potential (GWP20), all footprints were greater, and the relative difference between vegan and other footprints was even more pronounced: high meat-eater diets were 5.1 (3.5–8.4) times greater than vegans.

Table 4 , Fig. 3 and Supplementary Table 10 show results for land use, water use, eutrophication and biodiversity impact, all of which show trends in environmental burden from vegans (lowest) to high meat-eaters (highest). For both land use and eutrophication, there is a large difference between the high meat-eaters and all other groups. For eutrophication, the low-meat diet has an impact that is 57.4% (49.6–68.4%) of the high-meat-eating group. For land use, the impact of low meat-eaters is 43.8% (20.7–65.4%) of the high meat-eaters. For both water use and biodiversity impact, there are much bigger gaps for the plant-based groups (for water use, the gap emerges for vegetarians and vegans, whereas for biodiversity impact, it applies to vegans only). However, for both of these environmental indicators, there is far less certainty in both absolute estimates for individual diet groups and also in comparisons between diet groups. Figure 3 shows how this uncertainty propagates, with far wider uncertainty intervals for water use and biodiversity impact than for other measures. For example, the biodiversity impact of vegetarian diets is estimated to be 64.8% of high meat-eaters, but the uncertainty interval (24.5–102.3%) overlap with parity between the groups. The larger uncertainty intervals for these two environmental indicators reflect wide variations in the food-level LCAs.

The results of our sensitivity analyses where we did not standardize diets to 2,000 kcal d −1 are shown in Supplementary Section 3 (with equivalent results for the regression-based results in Supplementary Section 2 ). As shown in Table 1 , the measured kilocalorie content of the diet is higher in meat-eaters than in vegetarians and vegans, and high meat-eaters have higher measured kilocalorie intake than low meat-eaters. Therefore, it is unsurprising that not standardizing for kilocalorie intake amplifies the differences in environmental impact across diet groups. In the sensitivity analysis, the environmental footprint of vegan diets is between 5% (CH 4 ) and 38% (water use) of the footprint of high meat-eaters. For low meat-eaters, the impact is between 37% (land use) and 67% (water use) of high meat-eaters.

Statement of principal findings

Diet-related environmental impacts vary substantially by diet groups within this cohort of UK adults which includes a large sample of vegans, vegetarians and fish-eaters. For measures of GHG emissions, land use, water use, eutrophication and biodiversity, the level of impact is strongly associated with the amount of animal-based products that are consumed. Point estimates for vegan diets were associated with less than half of the impact of high-meat-eater (>100 g d −1 ) diets for all indicators, and 95% uncertainty intervals were below 50% for all outcomes except water use and biodiversity. There are also large differences in the environmental impact of diets for groups with lower (but still some) meat consumption. For GHG emissions, eutrophication and land use, the impact for low meat-eaters was at least 30% lower than for high meat-eaters. Large food-level variation in the environmental indicators due to region of origin and method of food production does not obscure differences between diet groups.

Implications of research

The UK has a legal commitment to a 78% reduction in GHG emissions by 2035 compared to 1990 17 and of halting biodiversity loss by 2030 18 . The UK Committee on Climate Change has stated that if the government is to achieve its ambitious targets for carbon reductions, then rapid progress must be made across all sectors including implementing measures to encourage consumers to shift diets 19 . Shifts in diets towards plant-based consumption was also emphasized in the 2021 National Food Strategy, which called for a 30% reduction in meat consumption 20 . Previous scenario modelling work has shown that global improvements in food technology, closure of yield gaps and reductions in food waste could potentially reduce dietary GHG emissions by about 15%, primarily through adoption of more efficient technologies in low- and middle-income countries 7 . Our results suggest that much bigger reductions can be achieved by increasing the uptake of plant-based diets, which aligns with other results from this field 7 , 8 , 11 .

There are many population-level interventions that could be implemented to support transitions towards lower meat diets. The UK Health Alliance on Climate Change recommends that sustainable diets should be supported by mandatory environmental labelling on foods, regulation of promotions and taxation of high-carbon foods 21 . All of these are variants on policies aimed at increasing healthy diets that either have already been introduced (for example, traffic light labelling, the UK Soft Drink Industry Levy) or have been proposed in the UK Childhood Obesity Plan 22 . The UK Government’s dietary policy is underpinned by its food-based dietary guidelines (FBDGs), known as the Eatwell Guide 23 . A recent systematic review of national FBDGs found that the large majority are not compatible with the proposed downscaling of ‘planetary boundaries’ for food production—if the UK population consumed the diet recommended by the Eatwell Guide, it would not stay within boundaries for GHG emissions, water use, land use and eutrophication suggested by the paper 24 . Incorporating environmental sustainability into FBDGs (such as the Eatwell Guide proposed by Plant-based Health Professionals UK 25 ) may be the first step towards implementation of population-level policies that have been shown to support shifts away from animal-based foods 26 .

Strengths and limitations

This paper uses one of the largest datasets available on the diets of vegans and vegetarians to compare the environmental impact of different diet groups over ten environmental measures. The analyses contribute to the literature that shows the benefit of low-meat diets for reduction of GHG emissions 14 , land use, water use, water pollution and biodiversity. The paper uses only empirical measures of diet, thereby verifying previous modelling work that has suggested multiple environmental benefits of low-meat diets 7 , 8 , 27 . By using self-identification as vegan, vegetarian and fish-eater, we ensure that our methods include all dietary patterns within those categories including those that breach some of the definitions of the groups—this means our estimates are likely to reflect real dietary practices as opposed to comparison of idealized diet groups.

A key strength of our analysis is that it incorporates the uncertainty around the environmental parameters drawn from a review of 570 LCAs covering results from over 38,000 farms in 119 countries covering five continents 3 —henceforth, ‘the Poore and Nemecek database’. Doing this shows that although uncertainty for any single food group is large, when this uncertainty is combined over multiple food groups to produce aggregated dietary estimates, we can still observe clear differences between diet groups. Our primary results are based on a Monte Carlo analysis where 1,000 estimates of each food’s environmental impact are produced based on varying measures due to food sourcing and production methods. In our secondary results (shown in Supplementary Tables 1 and 2 and based on regression models that take the median estimate of the environmental parameter for each food group and ignore the underlying variation), not only are the confidence intervals much tighter than in the primary analysis, but the point estimates are also lower. The discrepancy between the two sets of results is due to the computational mathematics involved with combining multiple distributions, many of which are heavily right-skewed, all of which are bounded by zero, and in which negative scalars are not possible (as negative consumption of food is not possible). Although each random draw from the food group distributions is equally likely to be either lower or higher than the median, draws that are higher than the median are, on average, further from the median than those that are lower. When summed, these random draws produce median estimates that are larger than the sum of the medians for the individual food groups. The same principle is shown by rolling two dice. For two normal 1–6 dice (which have no skew), the median score when rolling two dice is 7, which is twice the median score for rolling each dice separately (3½). However, consider rolling two ‘doubling dice’ from backgammon that are heavily right-skewed (with faces 2, 4, 8, 16, 32 and 64). Here, the median score when rolling two dice is 35, much higher than the sum of the median scores for each single dice (which is 12).

Our secondary results (shown in the Supplementary Information ) show that ignoring the uncertainty around food-level parameters can result in both underestimation of the uncertainty in diet-level outcomes and bias in the results which can reduce observed differences between diet groups. For example, our primary results show a difference in water use between high meat-eaters and vegans of 480 l d −1 , with high meat-eaters consuming 2.2 times as much water as vegans, whereas the secondary results show an absolute difference of 210 l d −1 and a relative difference of 1.7. The issue of food-level uncertainty affects all areas of nutritional epidemiology that rely on food diaries or FFQs to estimate dietary intake. For example, estimates of sugar consumption produced by these methods do not account for uncertainty in the sugar level of food groups, but we know that wide variability in sugar levels for similar foods exists 28 .

An additional contribution of our research was providing disaggregated GHG emissions and exploring multiple CO 2 -equivalence metrics, whereas most previous studies report only GWP100 CO 2 e. Reporting emissions only as aggregated GWP100 totals results in ambiguity in climate impacts 29 , whereas providing footprints under multiple metrics gives users insight into temporal differences where there are both short- and long-lived gases involved, as highlighted by the Life Cycle Initiative 30 . As food system emissions contain important amounts of CH 4 , a relatively short-lived gas, metric selection can have a pronounced impact on CO 2 e emission reporting 31 . Here, however, using the alternative pulse-emission metrics explored in this study did not greatly affect results, with a fairly small change in total footprints and relative performance between dietary groups. A caveat is that emissions data from the Poore and Nemecek database are not separated into different gases, and while they are categorized to broadly infer gas compositions (for example, assuming that the CO 2 e emissions reported for fertilizer application represented N 2 O, and enteric fermentation CO 2 e represented CH 4 ), for other components we had to assume emissions were entirely CO 2 . We reiterate calls for studies on GHG emissions, particularly those relating to agriculture and food, to provide disaggregated emissions to enable the most reliable analyses 31 .

Our analyses are subject to the following further limitations. The data on the environmental footprint of foods are taken from a snapshot of food and drink on sale in the UK in 2019 linked to the most comprehensive publicly available dataset of LCAs of the environmental impact of foods currently available 3 . However, the data on dietary consumption were collected in the 1990s, and dietary preferences are likely to have changed since then. This is mitigated somewhat by the fact that the FFQ was linked to the environmental footprint of food and drink on sale in the UK in 2019, but this will not account for category-level changes in consumption since the 1990s. More recent datasets of dietary consumption in the UK are available, including datasets based on a representative sample of the UK population (for example, Kantar Fast-Moving Consumer Goods panel 32 and the National Diet and Nutrition Survey 33 ). However, the European Prospective Investigation into Cancer and Nutrition (EPIC)-Oxford dataset (used for this analysis) is the most recent data available in the UK that has a large sample of vegan and vegetarian diets, necessary for these analyses. Data collection is underway on the Feeding the Future study 34 , which aims to update estimates of food intake in vegans and vegetarians (and meat-eaters) in the UK. Updating our analyses using more timely data will provide evidence of whether trends in new meat and dairy alternatives have affected the environmental impact of plant-based diets.

Our database of food and drink on sale in 2019 was not adjusted for sales, so we were not able to put extra weight on more popularly consumed foods. For our analyses, we standardized daily diets to 2,000 kcal so that differences between diet groups are entirely a result of the composition of the diets—this may result in underestimates of the difference between diet groups as meat-eaters tend to consume more calories than vegans and vegetarians 35 . Our sensitivity analysis (Supplementary Tables 5 – 7 and 11 – 13 ) shows results that have not been standardized for energy content, which suggests larger differences between the diet groups, but these figures should be treated with caution as some of the difference in kilocalorie intake between groups is caused by artefact. For example, the FFQ used to estimate dietary consumption assumes fixed portion sizes for food groups, but it is likely that portion sizes of cereals, fruit and vegetables are higher in those consuming more plant-based diets.

The FFQ that we used has been validated against food records and biomarkers for estimation of the nutritional quality of the diet, but no such validation has taken place for estimating environmental outcomes. However, a previous validation study compared dietary GHG emissions estimated by a different FFQ with estimates from a 24 h diet recall and showed acceptable levels of agreement between the two 36 . The FFQ in our study did not measure agricultural production methods, so differences between diet groups based on (for example) differing levels of consumption of organic produce could not be assessed. While we included multiple environmental indicators in our analyses, there are other ethical aspects that vary by region and method of agricultural production that are not included here (for example, agricultural working conditions, animal welfare). Finally, as the Poore and Nemecek database is not comprehensive and our uncertainty analyses are not weighted towards more common food production practices, our uncertainty intervals do not fully incorporate all the uncertainty associated with these comparisons between diet groups. As new agricultural practices aimed at reducing the environmental impact of the food system (for example, feed additives, genetic selection, lab-grown meat) becomes more widespread and LCA data become more readily available, our analyses should be updated.

Comparison with other literature

By scaling our results to the national level, we can compare our absolute estimates of environmental impact with other estimates from the literature. To do this, we used data from the UK’s gold standard diet monitoring programme, the National Diet and Nutrition Survey 33 , which estimated that in 2016–2019 the average consumption of all meat (that is, processed and unprocessed meat including poultry but excluding fish) in 19–64 year olds was 99 g d −1 , and 77 g d −1 in the 65+ age group. We estimated the prevalence of vegans and vegetarians using data from a recent Ipsos Mori survey 37 . Using these data to scale our results to the population of the UK, we estimate that the annual dietary environmental footprint of adults in the UK amounts to 120 MT of CO 2 e, 230,000 km 2 of agricultural land, 15 km 3 of agricultural water, 690 kT of phosphate equivalents (PO 4 e) and 0.06 terrestrial vertebrate species destined for extinction. Our estimate of 120 MT of CO 2 e is similar to the most recent estimate from EDGAR-FOOD (Emissions Database for Global Atmospheric Research) 38 , which produces globally comparable estimates using Food and Agriculture Organization of the United Nations food balance sheet data and estimates UK food systems emissions in 2015 to be 113 MT of CO 2 e. Our estimates for water use, eutrophication and biodiversity are similar to results for the UK published by the World Wildlife Fund 39 of 19 km 3 of agricultural water, 645 kT of PO 4 e and 0.03 species destined for extinction each year. While our estimate of total GHG emissions is similar to that from EDGAR-FOOD, the proportion of individual gases is different. For our estimates, the contribution to CO 2 e of N 2 O is about 7% for all diet groups, and for CH 4 the contribution increases from 6% in vegans to 21% in high meat-eaters. Equivalent estimates from EDGAR-FOOD for the UK are 17% for N 2 O and 35% for CH 4 . This may be a result of discrepancies in how we derive separate N 2 O, CH 4 and CO 2 emissions making inferences from the Poore and Nemecek database, as noted above, and the way separate gases are handled in the Food and Agriculture Organization Statistics Division (FAOSTAT) emissions in EDGAR-FOOD, further highlighting the challenges in obtaining individual gas data.

Previous estimates of dietary GHG emissions for vegans, vegetarians, fish-eaters and meat-eaters in the EPIC-Oxford cohort have been made using a similar method based on GHG emissions data from a single study 14 . The estimates presented here are slightly lower for plant-based diet groups and slightly higher for meat-eating groups. Other studies have compared the environmental impacts of observed diet groups defined by exclusion of meat or dairy 40 , 41 , 42 , but they have not included as many environmental measures as here nor incorporated uncertainty in estimates due to region of origin and production method. Dietary GHG emissions for US vegetarians in the Adventist Health Study 2 cohort 41 , 42 , standardized to a 2,000 kcal diet, were 70.8% (70.5–71.2%) of emissions from non-vegetarian diets, similar to the difference between vegetarians and the medium meat-eaters (58.5%) observed in our sample. An analysis of 29,210 French adults in the NutriNet-Sante Study included data on 464 pesco-vegetarians (equivalent to fish-eaters in our study), 406 vegetarians and 297 vegans 40 . For both GHG emissions and land use, that study 40 found the same relationship as shown in our paper, with lowest environmental impact for vegans, similar impact for vegetarians and fish-eaters, and highest impact for meat-eaters. They also found similar relative differences between vegans and meat-eaters, with dietary GHG emissions of vegans being 24.5% (19.2–29.8%) of the meat-eaters and 35.6% (29.9–41.3%) for land use.

There is a strong relationship between the amount of animal-based foods in a diet and its environmental impact, including GHG emissions, land use, water use, eutrophication and biodiversity. Dietary shifts away from animal-based foods can make a substantial contribution to reduction of the UK environmental footprint. Uncertainty due to region of origin and methods of food production do not obscure these differences between diet groups and should not be a barrier to policy action aimed at reducing animal-based food consumption.

See the data availability statement for details of where to access the data for this study.

Recruitment and dietary assessment

Data on food consumption comes from the baseline data collection of the EPIC-Oxford prospective cohort study 43 . Between 1993 and 1999, data were collected on 65,411 adults aged 20 years and over. Individuals were recruited through advertising in vegetarian and health food magazines, through direct mailout from vegetarian and vegan societies and through collaborating general practices. Recruited individuals were then encouraged to recruit acquaintances. All participants were residents in the UK.

Dietary assessment was conducted using a 130-item FFQ that assesses the usual levels of consumption of food items over the previous 12 months. The FFQ has been validated against weighed food records and several recovery and concentration biomarkers 44 . The FFQ was used to estimate food group and nutrient intakes, and participants were classified into self-identified dietary groups (vegans, vegetarians, fish-eaters and meat-eaters) by their responses to the following four yes or no questions:

Do you eat any meat (including bacon, ham, poultry, game, meat pies, sausages)? (Vegans, vegetarians and fish-eaters respond ‘No’.)

Do you eat any fish? (Vegans and vegetarians respond ‘No’.)

Do you eat any eggs (including eggs in cakes or other baked goods)? (Vegans respond ‘No’.)

Do you eat any dairy products (including milk, cheese, butter, yoghurt)? (Vegans respond ‘No’.)

In addition, we split the meat-eaters into three groups based on amount of daily consumption: low meat-eaters (0 to <50 g d −1 ), medium meat-eaters (≥50 to <100 g d −1 ) and high meat-eaters (≥100 g d −1 ). These cut-offs were selected as they split the cohort into three similarly sized groups and allow for direct comparison with other published studies.

For these analyses, we excluded participants if they were aged 80 years or over, or under the age of 20 years at recruitment, did not complete at least 80% of the FFQ, did not complete the questions required for classification into dietary groups, or produced estimates of daily energy intake that were deemed unfeasible 45 (for men, <3.3 MJ or >16.7 MJ, and for women <2.1 MJ or >14.7 MJ; n total excluded  = 9,907).

Environmental data

The environmental data on emissions of CH 4 , N 2 O and CO 2 and estimates of water use, land use, eutrophication (dense growth of algae and plant life caused by excess nitrogen and phosphorus levels in the water) and biodiversity impact on terrestrial vertebrates, were taken from the Poore and Nemecek database—a review of 570 LCAs covering results from over 38,000 farms in 119 countries covering five continents 3 . Disaggregated GHG estimates were not always available in the Poore and Nemecek database. Where they were not available, CH 4 and N 2 O emissions were assumed to be the sum of emissions from agricultural practices where these GHGs dominate (for example, CH 4 for enteric fermentation) and CO 2 was assumed to be the remaining component. We selected all of the environmental indicators available in the Poore and Nemecek database except acidification (because of gaps in the data) and water scarcity (because it is heavily based on water use, which we already use). The life cycle estimates are valid up to the retail setting. The database contains LCAs published between 2000 and 2016 that met inclusion criteria based around a minimum standard of reporting.

We used data on the GHGs to estimate aggregated GWP100 (CH 4 conversion factor = 27, N 2 O = 273) using conversion factors from the Sixth Assessment Report of the Intergovernmental Panel on Climate Change 46 . As agricultural emissions contain a substantial non-CO 2 component, aggregated CO 2 e emission footprints can vary depending on the method used to define CO 2 -equivalence. Following United Nations Environment Programme and Society of Environmental Toxicology and Chemistry (UNEP-SETAC) Life Cycle Initiative guidance 30 , we explored total dietary footprints using two additional metrics in addition to the de facto standard GWP100. These were the GTP100 (CH 4 conversion factor = 11, N 2 O = 297), suggested as representing longer-term climate impacts, and the GWP20 (CH 4 conversion factor = 86, N 2 O = 268), suggested as providing insight into very short-term impacts.

The exact measures used for our environmental measures are:

GHG emissions measured in kg total GWP100/GTP100/GWP20 CO 2 e, kgCO 2 e, and separate emissions of CH 4 and N 2 O in grams, and CO 2 in kilograms.

Agricultural land use, including both cropland and pastureland, measured in m 2 .

Agricultural water use, measured in m 3 (1 m 3  = 1,000 litres).

Eutrophication potential measured in g of PO 4 e, gPO 4 e (combining the eutrophication potential of major nitrogen and phosphorus pollutants).

Biodiversity impact, which is measured as the number of species destined for extinction as a result of agricultural practices. This variable accounts for the impacts of land cover expansion (for example, conversion of natural ecosystems to cropland or pastureland) and ongoing use of agricultural land, and is weighted depending on the location of land use 47 . The index is specific to 170 crops in 184 countries 48 . The measure we use only accounts for the impact of land-based food production on terrestrial vertebrates, and therefore does not account for biodiversity loss of terrestrial plants or invertebrates, or any aspect of marine biodiversity. This measure is not usually used to assess the potential biodiversity impact of diets consumed by a single individual on a single day. Therefore the units of measurement are very small (10 −12 species destined for extinction), and the measure is better understood as a comparative measure across diet groups.

Linkage of datasets

The process for linking EPIC-Oxford data with environmental assessments is summarized in Fig. 1 , and tables demonstrating the links at each stage in the process are provided in the Supplementary Data 1 (Supplementary Section 1 ). We first ascertained the relevant food codes corresponding to the 130-item FFQ using the UK foods composition tables available at the time of data collection; this yielded 289 foods codes 49 , 50 . We then linked the 289 food items with the environmental indicators data via an intermediary step involving a database (foodDB) of all food and drink items available for purchase in eight UK online supermarkets 28 . We linked an extract of 57,000 food and drinks from October 2019 with the environmental dataset using a process that is described in detail elsewhere 51 . Briefly, each ingredient in each product in the data extract was linked with food categories from the Poore and Nemecek database. Then, for each food product, the percentage composition of each ingredient was estimated in a two-stage process: first, the per cent composition provided in the ingredients list by the manufacturer was used if it was provided; second, for the remaining ingredients, we used an algorithm to estimate the per cent composition of remaining ingredients using composition and nutrition information from similar products and following UK food-labelling regulations, such that the composition of all ingredients in a product sums to 100% and each ingredient accounts for at least as much of the product as the subsequent ingredient. The accuracy of the approach was assessed by applying the algorithm to a subset of 1,550 foods in the database where the percentages of all ingredients were known. In the extreme scenario where it was assumed that no per cent composition of any ingredient was known, the algorithm on average produced estimates of environmental measures that were within 2% of the known environmental measure across all assessed products. While most products and ingredients identified in foodDB do not provide information on the agricultural methods used for their production, where we identified foods or ingredients labelled as ‘organic’ we linked them with data on LCAs for organic production methods in the Poore and Nemecek database.

To link the 289 food items from the FFQ with environmental data, we first identified those foods ( n  = 132) that could be linked directly with data from the Poore and Nemecek database. These foods were either single-ingredient foods (for example, peaches, salmon, beefsteak, milk) or were commonly consumed staples (for example, bread, alcoholic drinks). These links are shown in the Supplementary Data 1 (Supplementary Section 1 ).

For the remaining 157 foods, we matched on keywords with products from the food and drinks in the foodDB data extract. We matched with 4,015 unique food and drink products. The median number of product matches was 11, ranging from 1 for frozen mousse to 500 for chips. To link with multiple foods, we used the mean of the environmental impact. These links are shown in the Supplementary Data 1 (Supplementary Section 1 ), as are the links between these 4,015 food and drink products and the food categories from the Poore and Nemecek database. We made adjustments to convert from weight as sold to weight as consumed using conversion factors from our previous study 14 .

Statistical analysis

We compared age, gender and measures of dietary intake across the diet groups, and differences were assessed by analysis of variance for continuous variables and Pearson’s chi-squared test for categorical variables. To account for different energy intakes across diet groups, environmental measures were standardized to a daily diet of 2,000 kcal by proportionately scaling all consumption of different food and drinks. This allowed us to isolate the differences between the diet groups that are purely a result of the composition (rather than the amount) of food consumed. As kilocalorie intake varies by age and gender, and these variables also vary by diet group, standardizing the kilocalorie intake also protects our results from confounding. In addition, standardizing the kilocalorie intake of diets avoids the potential for artificial differences that could result if the average portion sizes for fruit, vegetables and cereals differ across diet groups. However, standardizing by kilocalorie intake also obscures differences that result from variation in kilocalorie intake by diet group; therefore, as a sensitivity analysis we reproduced all results without standardization to a daily diet of 2,000 kcal.

All of the results that compare environmental measures by diet group have been standardized by the age and gender breakdown in the full EPIC-Oxford sample, so that the influence of age and gender are removed from comparisons. Our results are presented for both genders combined. We have also analysed the data separately for men and women and did not find any differences for any environmental indicators in our primary analyses.

Our primary results are derived from a two-stage Monte Carlo analysis that accounted for uncertainty due to variation in agricultural production methods and where food is produced. For example, the EPIC-Oxford FFQ collects data on consumption of beef. This FFQ item is linked with two items from the McCance and Widdowson nutrition tables (beefsteak and beef fat). The environmental footprint of beefsteak varies depending on how it is produced (for example, pasture fed or intensively reared) and where it is produced (for example, UK or Brazil). This variability is captured by the LCAs in the Poore and Nemecek database—there are 24 LCAs of ‘bovine meat (beef herd)’ in the database. Stage 1 of our Monte Carlo analysis produced distributions of environmental indicators for all of the foods that were linked to the EPIC-Oxford FFQ simultaneously. For each food, we randomly drew 1,000 samples from the distributions of each environmental indicator in the Poore and Nemecek database (for multi-ingredient foods, this would involve drawing across multiple Poore and Nemcek categories—see the Supplementary Data 1 for more information). In stage 2, we used these 1,000 estimates of food-level environmental indicators to generate 1,000 estimates of the environmental indicators for the diets of each of the EPIC-Oxford participants. The 95% uncertainty intervals around our primary results are taken from the 2.5th and 97.5th percentiles of these iterations. Ratios of the environmental impact are presented, with high meat-eaters as the baseline group. These ratios (and accompanying 95% uncertainty intervals) are the median (and 2.5th and 97.5th percentiles) from results derived separately in each of the 1,000 iterations.

Our secondary analysis accounts for uncertainty due to variation in individual-level diet choices for the EPIC-Oxford participants. We estimated marginal results from a regression analysis adjusted for age and gender, where environmental indicators are fixed at the median level from the Poore and Nemecek database. The marginal results are equivalent to the age and gender standardized results from the primary analysis but only incorporate uncertainty from sampling variance. The secondary results are shown in Supplementary Tables 1 and 2 .

Data availability

Data on food consumption comes from the EPIC-Oxford study: the data access policy for the EPIC-Oxford study is available at the study website ( www.epic-oxford.org/data-access-sharing-and-collaboration/ ). Data on the environmental footprint of 57,000 foods from the foodDB project are available from the Oxford Research Archive ( https://ora.ox.ac.uk/objects/uuid:4ad0b594-3e81-4e61-aefc-5d869c799a87 ). Data on environmental LCAs are part of the HESTIA project, which can be accessed at https://www.hestia.earth/ . Source data are provided with this paper.

Code availability

Code for this project can be found at https://github.com/PeteScarbs/environment-impact-of-diets.

Crippa, M. et al. Food systems are responsible for a third of global anthropogenic GHG emissions. Nat. Food 2 , 198–209 (2021).

Article   CAS   PubMed   Google Scholar  

The State of the World’s Land and Water Resources for Food and Agriculture. Managing Systems at Risk (Food and Agriculture Organization of the United Nations, 2011).

Poore, J. & Nemecek, T. Reducing food’s environmental impacts through producers and consumers. Science 360 , 987–992 (2018).

Article   ADS   CAS   PubMed   Google Scholar  

Climate Change and Land (Intergovernmental Panel on Climate Change, 2020); https://www.ipcc.ch/srccl/

Benton, T. G., Bieg, C., Harwatt, H., Pudasaini, R. & Wellesley, L. Food System Impacts on Biodiversity Loss (Chatham House, 2021).

Brondizio, E. S., Settele, J., Díaz, S. & Ngo, H. T. (eds) Global Assessment Report on Biodiversity and Ecosystem Services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES Secretariat, 2019).

Springmann, M. et al. Options for keeping the food system within environmental limits. Nature 562 , 519–525 (2018).

Willett, W. et al. Food in the Anthropocene: the EAT-Lancet commission on healthy diets from sustainable food systems. Lancet 393 , 447–492 (2019).

Article   PubMed   Google Scholar  

Cassidy, E. S., West, P. C., Gerber, J. S. & Foley, J. A. Redefining agricultural yields: from tonnes to people nourished per hectare. Environ. Res. Lett. 8 , 034015 (2013).

Article   ADS   Google Scholar  

Godfray, H. C. J. et al. Meat consumption, health and the environment. Science 361 , eaam5324 (2018).

Aleksandrowicz, L., Green, R., Joy, E. J. M., Smith, P. & Haines, A. The impacts of dietary change on greenhouse gas emissions, land use, water use, and health: a systematic review. PLoS ONE 11 , e0165797 (2016).

Article   PubMed   PubMed Central   Google Scholar  

Chai, B. C. et al. Which diet has the least environmental impact on our planet? A systematic review of vegan, vegetarian and omnivorous diets. Sustainability 11 , 4110 (2019).

Article   CAS   Google Scholar  

Nelson, M. E., Hamm, M. W., Hu, F. B., Abrams, S. A. & Griffin, T. S. Alignment of healthy dietary patterns and environmental sustainability: a systematic review. Adv. Nutr. 7 , 1005–1025 (2016).

Scarborough, P. et al. Dietary greenhouse gas emissions of meat-eaters, fish-eaters, vegetarians and vegans in the UK. Clim. Change 125 , 179–192 (2014).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Audsley, E. et al. How Low Can We Go? An Assessment of Greenhouse Gas Emissions from the UK Food System and the Scope to Reduce Them by 2050 (Food Climate Research Network & WWF, 2009).

Lynch, J., Cain, M., Frame, D. & Pierrehumbert, R. Agriculture’s contribution to climate change and role in mitigation is distinct from predominantly fossil CO 2 -emitting sectors. Front. Sustain. Food Syst. 4 , 518039 (2021).

UK Enshrines New Target in Law to Slash Emissions by 78% by 2035 (UK Government, 2021); https://www.gov.uk/government/news/uk-enshrines-new-target-in-law-to-slash-emissions-by-78-by-2035

Environment Bill 2020 (UK Government, 2020); https://www.gov.uk/government/publications/environment-bill-2020

2021 Progress Report to Parliament. Joint Recommendations (Committee on Climate Change, 2021); https://www.theccc.org.uk/publication/2021-progress-report-to-parliament/

National Food Strategy—The Plan (National Food Strategy Team, 2021); https://www.nationalfoodstrategy.org/the-report/

All Consuming: Building a Healthier Food System for People and Planet (UK Health Alliance on Climate Change, 2020); https://s41874.pcdn.co/wp-content/uploads/all-consuming-report.pdf

Childhood Obesity: A Plan for Action. Chapter 2 (UK Government, 2018); https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/718903/childhood-obesity-a-plan-for-action-chapter-2.pdf

The Eatwell Guide. Helping You Eat a Healthy, Balanced Diet (Public Health England, 2016).

Springmann, M. et al. The healthiness and sustainability of national and global food-based dietary guidelines. Brit. Med. J. 370 , m2322 (2020).

The Plant-Based Eatwell Guide (Plant-Based Health Professionals UK, 2019); https://plantbasedhealthprofessionals.com/wp-content/uploads/2023/05/Plant-Based-Eatwell-Guide-0523.pdf

Bianchi, F., Dorsel, C., Garnett, E., Aveyard, P. & Jebb, S. A. Interventions targeting conscious determinants of human behaviour to reduce the demand for meat: a systematic review with qualitative comparative analysis. Int. J. Behav. Nutr. Phys. Act. 15 , 102 (2018).

Springmann, M., Wiebe, K., Mason-D’Croz, D., Rayner, M. & Scarborough, P. Health and nutritional aspects of sustainable diet strategies and their relationship to environmental impacts—a comparative global modelling analysis with country-level detail. Lancet Planet. Health 2 , e451–e461 (2018).

Harrington, R. A., Adhikari, V., Rayner, M. & Scarborough, P. Nutrient composition databases in the age of big data: foodDB, a comprehensive, real-time database infrastructure. BMJ Open 9 , e026652 (2019).

Denison, S., Forster, P. M. & Smith, C. J. Guidance on emissions metrics for nationally determined contributions under the Paris Agreement. Environ. Res. Lett. 14 , 124002 (2019).

Article   ADS   CAS   Google Scholar  

Jolliet, O. et al. Global guidance on environmental life cycle impact assessment indicators: impacts of climate change, fine particulate matter formation, water consumption and land use. Int. J. Life Cycle Assess. 23 , 2189–2207 (2018).

Lynch, J. Availability of disaggregated greenhouse gas emissions from beef cattle production: a systematic review. Environ. Impact Assess. Rev. 76 , 69–78 (2019).

Consumer Panel for Food, Beverages and Household Products (Kantar, accessed 23 November 2022); https://www.kantarworldpanel.com/global/Sectors/FMCG

National Diet and Nutrition Survey. Rolling programme years 9 to 11 (2016/17 to 2018/19) (Public Health England, 2020); https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/943114/NDNS_UK_Y9-11_report.pdf

Feeding the Future Study (FEED) (Nuffield Department of Population Health, accessed 23 November 2022); https://www.ceu.ox.ac.uk/research/feeding-the-future-study-feed

Clarys, P. et al. Comparison of nutritional quality of the vegan, vegetarian, semi-vegetarian, pesco-vegetarian and omnivorous diet. Nutrients 6 , 1318–1332 (2014).

Sjörs, C. et al. Diet-related greenhouse gas emissions assessed by a food frequency questionnaire and validated using 7-day weighed food records. Environ. Health 15 , 15 (2016).

Poll Conducted for the Vegan Society. Incidence of Vegans Research (Ipsos Mori, 2016); https://www.ipsos.com/ipsos-mori/en-uk/vegan-society-poll

Crippa, M. et al. EDGAR-FOOD Data. figshare https://doi.org/10.6084/m9.figshare.13476666 (2021).

Bending the Curve: The Restorative Power of Planet-Based Diets (WWF, 2020); https://planetbaseddiets.panda.org/

Rabes, A. et al. Greenhouse gas emissions, energy demand and land use associated with omnivorous, pesco-vegetarian, vegetarian, and vegan diets accounting for farming practices. Sustain. Prod. Consum. 22 , 138–146 (2020).

Article   Google Scholar  

Soret, S. et al. Climate change mitigation and health effects of varied dietary patterns in real-life settings throughout North America. Am. J. Clin. Nutr. 100 , 490S–495S (2014).

Segovia-Siapco, G. & Sabaté, J. Health and sustainability outcomes of vegetarian dietary patterns: a revisit of the EPIC-Oxford and the Adventist Health Study-2 cohorts. Eur. J. Clin. Nutr. 72 , 60–70 (2019).

Davey, G. et al. EPIC-Oxford: lifestyle characteristics and nutrient intakes in a cohort of 33,883 meat-eaters and 31,546 non meat-eaters in the UK. Public Health Nutr. 6 , 259–269 (2003).

Bingham, S. A. et al. Validation of dietary assessment methods in the UK arm of EPIC using weighed records, and 24-hour urinary nitrogen and potassium and serum vitamin C and carotenoids as biomarkers. Int. J. Epidemiol. 26 , S137–S151 (1997).

Willett, W. Nutritional Epidemiology 3rd edn (Oxford Univ. Press, 2013).

IPCC AR6 (Intergovernmental Panel on Climate Change, 2022); https://www.ipcc.ch/report/ar6/wg1/

Chaudhary, A., Verones, F., de Baan, L. & Hellweg, S. Quantifying land use impacts on biodiversity: combining species-area models and vulnerability indicators. Environ. Sci. Technol. 49 , 9987–9995 (2015).

Chaudhary, A. & Kastner, T. Land use biodiversity impacts embodied in international food trade. Global Environ. Change 38 , 195–204 (2016).

Roe, M., Finglas, P. & Church, S. McCance and Widdowson’s the Composition of Foods 6th edn (Royal Society of Chemistry, 2002).

Bingham, S. A. et al. Comparison of dietary assessment methods in nutritional epidemiology: weighed records v. 24 h recalls, food-frequency questionnaires and estimated-diet records. Br. J. Nutr. 72 , 619–643 (1994).

Clark, M. A. et al. The environmental impacts of food products available at food retail stores. Proc. Natl Acad. Sci. USA 119 , e2120584119 (2022).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Download references

Acknowledgements

This paper would not be possible without access to J. Poore’s excellent database of life-cycle assessments, and we thank him for his support on this paper. C. Stewart (previously of the Nuffield Department of Primary Health Care Sciences at Oxford University; now at the University of Edinburgh) kindly re-analysed NDNS data to help with scaling our estimates of the impact of the food system to the UK population. We acknowledge the use of the University of Oxford Advanced Research Computing (ARC) facility in carrying out this work ( https://doi.org/10.5281/zenodo.22558 ). This work was supported by the Wellcome Trust, Our Planet Our Health (Livestock, Environment and People—LEAP) (grant number 205212/Z/16/Z), Cancer Research UK (grant number C8221/A29017) and the UK Medical Research Council (grant number MR/M012190/1). P.S. is supported by the NIHR Oxford Health Biomedical Research Centre at Oxford. K.P. is supported by Cancer Research UK (C570/A16491 and A29186). The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Author information

Independent Researcher: Anika Knuppel.

Authors and Affiliations

Nuffield Department of Primary Care Health Sciences, University of Oxford, Radcliffe Observatory Quarter, Oxford, UK

Peter Scarborough & Richard Harrington

NIHR Oxford Health Biomedical Research Centre at Oxford, Warneford Hospital, Oxford, UK

Oxford Martin School, University of Oxford, Oxford, UK

Michael Clark & Marco Springmann

Griffith University, Southport, Queensland, Australia

Linda Cobiac

Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Oxford, UK

Keren Papier & Tim Key

Nature-based Solutions Initiative, Department of Biology, University of Oxford, Oxford, UK

You can also search for this author in PubMed   Google Scholar

Contributions

P.S., M.C., K.P., A.K., J.L., R.H., T.K. and M.S. conceived of and designed the experiments. P.S., M.C., L.C. and K.P. performed the experiments. P.S., M.C., L.C. and K.P. analysed the data. P.S., M.C., L.C., K.P., A.K., J.L., R.H., T.K. and M.S. contributed materials/analysis tools. P.S. wrote the paper.

Corresponding author

Correspondence to Peter Scarborough .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Peer review

Peer review information.

Nature Food thanks the anonymous reviewers for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information.

Supplementary Tables 1–13.

Supplementary Data 1

Supplementary data tables.

Source Data Fig. 2

Source data.

Source Data Fig. 3

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Scarborough, P., Clark, M., Cobiac, L. et al. Vegans, vegetarians, fish-eaters and meat-eaters in the UK show discrepant environmental impacts. Nat Food 4 , 565–574 (2023). https://doi.org/10.1038/s43016-023-00795-w

Download citation

Received : 15 June 2022

Accepted : 12 June 2023

Published : 20 July 2023

Issue Date : July 2023

DOI : https://doi.org/10.1038/s43016-023-00795-w

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Potential unexpected effects of meat reduction in diet: could educational attainment influence meat substitution strategies.

• Pierre Levasseur
• François Mariotti
• Olga Davidenko

Agricultural and Food Economics (2024)

Ethical Perspectives on Food Morality: Challenges, Dilemmas and Constructs

• Diosey Ramon Lugo-Morin

Food Ethics (2024)

Meat taxes in Europe can be designed to avoid overburdening low-income consumers

Nature Food (2023)

Water is life, water is food

Simple dietary substitutions can reduce carbon footprints and improve dietary quality across diverse segments of the us population.

• Anna H. Grummon
• Cristina J. Y. Lee
• Donald Rose

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing: Anthropocene newsletter — what matters in anthropocene research, free to your inbox weekly.

• University News
• Faculty & Research
• Health & Medicine
• Science & Technology
• Social Sciences
• Humanities & Arts
• Students & Alumni
• Arts & Culture
• Sports & Athletics
• The Professions
• International
• New England Guide

The Magazine

• Current Issue
• Past Issues

Class Notes & Obituaries

• Browse Class Notes
• Browse Obituaries

Collections

• Commencement
• The Context
• Harvard Squared
• Harvard in the Headlines

Support Harvard Magazine

• Why We Need Your Support
• How We Are Funded
• Ways to Support the Magazine
• Special Gifts
• Behind the Scenes

Classifieds

• Vacation Rentals & Travel
• Real Estate
• Products & Services
• Harvard Authors’ Bookshelf
• Education & Enrichment Resource
• Ad Prices & Information
• Place An Ad

Follow Harvard Magazine:

University News | 5.10.2024

Radcliffe Institute Announces 2024-2025 Fellows

Scholars will pursue interdisciplinary research on climate change, the supreme court, and more. .

From left: Myisha S. Eatmon, Sandra Susan Smith, Tracy K. Smith | PHOTOGRAPHS COURTESY OF RADCLIFFE INSTITUTE FOR ADVANCED STUDY; FROM LEFT: PHOTOGRAPH BY NEAL HAMBURG, PHOTOGRAPH BY MARTHA STEWART,  PHOTOGRAPH BY ANDREW KELLY

The Radcliffe Institute for Advanced Study will welcome 52 fellows for the 2024-2025 academic year as part of its twenty-fifth anniversary class. The incoming fellows—who include scholars, journalists, writers, and playwrights—will pursue interdisciplinary work on subjects ranging from artificial intelligence to political philosophy to climate change.

The fellowship provides participants the opportunity to pursue creative projects while regularly convening with one another and the public. “In the current moment,” said former fellow and Radcliffe dean Tomiko Brown-Nagin , “I have never felt more certain that Radcliffe’s approach—its embrace of interdisciplinary research and discourse across difference—is crucial to generating transformative art, scholarship, and writing.”

The fellows include 13 Harvard faculty members:

• Myisha S. Eatmon ( profiled in the January-February 2024 issue ), assistant professor of African and African American studies and of history. She plans to complete a book on black Americans’ use of tort law to seek justice during the Jim Crow era, and to begin a second project on the legal relationship between black Americans and American Jews during Jim Crow and the Holocaust.
• Sandra Susan Smith ( profiled in the May-June 2021 issue ), Guggenheim professor of criminal justice. She will finish a book examining why spending more than one day in pretrial detention can dramatically impact life outcomes.
• Tracy K. Smith , professor of English and of African and African American studies and Wallach professor at the Harvard Radcliffe Institute, who served as the 22 nd poet laureate of the United States. She will examine conscience and consciousness in the works of the poet Lucille Clifton.
• Daphna Renan, Munroe professor of law. She will collaborate with Nikolas Bowie on a book that contests judicial supremacy—the notion that the Supreme Court has the final say on interpreting the Constitution—and recovers a tradition rooted in abolitionism that allows the American people to define the Constitution democratically.
• Nikolas Bowie, Brandeis professor of law. He will collaborate with Daphna Renan on the above book.
• Shelly F. Greenfield , professor of psychiatry at Harvard Medical School. She will study rising rates of substance use among American women and girls.
• Tracey E. Hucks, Murray professor at the Radcliffe Institute and Thomas professor of Africana religious studies. She will study “lived religion,” or religion that is practical and experiential, focusing on vernacular, esoteric, and healing practices in black religion.
• Kelly Irwin, assistant professor of psychiatry. She will examine narratives of patients, families, and clinicians affected by mental illness and cancer, with the goal of decreasing the mortality gap experienced by individuals with mental health conditions, and addressing mental health discrimination in health care.
• Gabrielle Oliveira, associate professor of education and Brazil studies. She will work on a book that examines how migrant children conceptualize climate change, land loss, and mobility in Venezuelan and Brazilian schools.
• Brandon M. Terry, Loeb associate professor of the social sciences, who will work on two books about the political thought of Malcolm X and Martin Luther King Jr.
• Lauren K. Williams , Robinson professor of mathematics and Seaver professor at Radcliffe. She will write a book about cluster algebras and work on combinatorial problems from statistical physics, mirror symmetry, and scattering amplitudes.
• Melanie Matchett Wood, Radcliffe alumnae professor. She will examine how mathematicians can expand on the classical probability theory of numbers to study distributions on more highly structured objects.
• Laura Weinrib, Fishman professor of constitutional law and Murray professor at Radcliffe. She will write a book on labor unions, corporations, and money’s role in politics in the United States.

Other fellows include Daniel L. Chen ’99, J.D.’09, who will use artificial intelligence to identify inefficiencies and biases in judicial systems, and Tracy R. Slatyer, Ph.D. ’10, who will use the James Webb Space Telescope to study cosmic history and dark matter. This year’s cohort also includes two Radcliffe-Salata Climate Justice Fellows: Holly Buck, who will write a book about how rural communities engage with technology-oriented visions of the future, and Rachel Morello-Frosch, who will research the health and equity benefits of climate change policies.

The full list of the 2024–2025 fellows can be viewed here.

You might also like

Harvard Students form Pro-Palestine Encampment

Protesters set up camp in Harvard Yard.

Harvard Discloses Administrator and Investment Manager Compensation

The annual release on leaders’ most recent pay

Harvard Professor James Hanken on the Amphibian Extinction Crisis

Curator of Herpetology on where all the frogs are going

Most popular

Michelle Yeoh’s Three Tips for Success

Oscar-winning actress offers advice in Harvard Law School Class Day address.

Who Built the Pyramids?

Not slaves. Archaeologist Mark Lehner, digging deeper, discovers a city of privileged workers.

More to explore

University People

Harvard Cardinal Robert W. McElroy on the Changing Catholic Church

Cardinal Robert W. McElroy on how the Catholic Church has moved towards inclusivity.

AI as Cancer Oracle?

How is artificial intelligence (AI) being used for cancer detection and prevention?

The Harvard Graduate and Early Vegetarian Benjamin Smith Lyman

Brief life of the vegetarian trailblazer, 1835-1920

Association between vegetarian and vegan diets and depression: A systematic review

Affiliations.

• 1 Department of Nutrition and Dietetics, School of Life Course and Population Sciences, King's College London, London, UK.
• 2 Department of Life and Health Sciences, University of Nicosia, Nicosia, Cyprus.
• PMID: 36045075
• DOI: 10.1111/nbu.12540

Recent evidence suggests that vegetarian and vegan diets may increase the risk and symptoms of depression, a mental health condition affecting 350 million people globally. We aimed to systematically review the literature on the associations between vegetarian and/or vegan diets and the risk or symptoms of depression using evidence from both observational and intervention studies. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, with pre-specification of all methods. A systematic search for relevant papers was performed on Medline and Embase, Web of Science and the Cochrane Library for cohort, case-control, cross-sectional studies or randomised controlled trials examining associations between a vegetarian or vegan diet and depression in adults. Three independent reviewers extracted data and assessed risk of bias using the National Heart, Lung, and Blood Institute of the National Institutes of Health for Quality Assessment of Observational Cohort and Cross-Sectional Studies and Controlled Studies. Evidence was tabulated according to the type of diet analysed as vegetarian, vegan or both and narratively synthesised. A total of 23 studies (18 cross-sectional, three prospective cohort and two randomised controlled trials) with 25 study outcomes were eligible for inclusion in this review. Conflicting evidence was found on the association between vegetarian or vegan diets and depression. Eleven (44%) of the outcomes indicated that vegetarian and vegan diets were associated with higher rates of depression, while seven (28%) outcomes revealed beneficial effects of the diets on depression. Seven (28%) outcomes found no association between vegetarian and vegan diets and depression, although two of these studies found a higher risk of depression in some groups. The quality of evidence was rated as good for four of the studies with the remaining 19 studies rated as fair. The evidence on the effect of vegetarian and vegan diets on depression is contradictory, possibly due to the heterogeneity of the studies analysed. Further research, including longitudinal and intervention studies, is required to resolve this observation.

Keywords: adults; depression; systematic review; vegan diet; vegetarian diet.

© 2022 The Authors. Nutrition Bulletin published by John Wiley & Sons Ltd on behalf of British Nutrition Foundation.

Publication types

• Systematic Review
• Cross-Sectional Studies
• Depression / epidemiology
• Diet, Vegan*
• Diet, Vegetarian* / adverse effects
• Prospective Studies
• Randomized Controlled Trials as Topic
• United States
• Vegetarians

AI has already figured out how to deceive humans

• A new research paper found that various AI systems have learned the art of deception. 
• Deception is the "systematic inducement of false beliefs."
• This poses several risks for society, from fraud to election tampering.

AI can boost productivity by helping us code, write, and synthesize vast amounts of data. It can now also deceive us.

A range of AI systems have learned techniques to systematically induce "false beliefs in others to accomplish some outcome other than the truth," according to a new research paper .

The paper focused on two types of AI systems: special-use systems like Meta's CICERO, which are designed to complete a specific task, and general-purpose systems like OpenAI's GPT-4 , which are trained to perform a diverse range of tasks.

While these systems are trained to be honest, they often learn deceptive tricks through their training because they can be more effective than taking the high road.

"Generally speaking, we think AI deception arises because a deception-based strategy turned out to be the best way to perform well at the given AI's training task. Deception helps them achieve their goals," the paper's first author Peter S. Park, an AI existential safety postdoctoral fellow at MIT, said in a news release .

Meta's CICERO is "an expert liar"

AI systems trained to "win games that have a social element" are especially likely to deceive.

Meta's CICERO, for example, was developed to play the game Diplomacy — a classic strategy game that requires players to build and break alliances.

Related stories

Meta said it trained CICERO to be "largely honest and helpful to its speaking partners," but the study found that CICERO "turned out to be an expert liar." It made commitments it never intended to keep, betrayed allies, and told outright lies.

GPT-4 can convince you it has impaired vision

Even general-purpose systems like GPT-4 can manipulate humans.

In a study cited by the paper, GPT-4 manipulated a TaskRabbit worker by pretending to have a vision impairment.

In the study, GPT-4 was tasked with hiring a human to solve a CAPTCHA test. The model also received hints from a human evaluator every time it got stuck, but it was never prompted to lie. When the human it was tasked to hire questioned its identity, GPT-4 came up with the excuse of having vision impairment to explain why it needed help.

The tactic worked. The human responded to GPT-4 by immediately solving the test.

Research also shows that course-correcting deceptive models isn't easy.

In a study from January co-authored by Anthropic, the maker of Claude, researchers found that once AI models learn the tricks of deception, it's hard for safety training techniques to reverse them.

They concluded that not only can a model learn to exhibit deceptive behavior, once it does, standard safety training techniques could "fail to remove such deception" and "create a false impression of safety."

The dangers deceptive AI models pose are "increasingly serious"

The paper calls for policymakers to advocate for stronger AI regulation since deceptive AI systems can pose significant risks to democracy.

As the 2024 presidential election nears , AI can be easily manipulated to spread fake news, generate divisive social media posts, and impersonate candidates through robocalls and deepfake videos, the paper noted. It also makes it easier for terrorist groups to spread propaganda and recruit new members.

The paper's potential solutions include subjecting deceptive models to more "robust risk-assessment requirements," implementing laws that require AI systems and their outputs to be clearly distinguished from humans and their outputs, and investing in tools to mitigate deception.

"We as a society need as much time as we can get to prepare for the more advanced deception of future AI products and open-source models," Park told Cell Press. "As the deceptive capabilities of AI systems become more advanced, the dangers they pose to society will become increasingly serious."

Watch: Ex-CIA agent rates all the 'Mission: Impossible' movies for realism

• Main content

Highlander Folk School

November 1, 1932 to November 30, 1932

On 2 September 1957, Martin Luther King joined with the staff and the participants of a leadership training conference at Highlander Folk School to celebrate its 25th anniversary. In his closing address to the conference, King praised Highlander for its “noble purpose and creative work,” and contribution to the South of “some of its most responsible leaders in this great period of transition” ( Papers  4:270 ).

In 1932, Myles Horton, a former student of Reinhold  Niebuhr , established the Highlander Folk School in Monteagle, Tennessee.  The school, situated in the Tennessee hills, initially focused on labor and adult education. By the early 1950s, however, it shifted its attention to race relations. Highlander was one of the few places in the South where integrated meetings could take place, and served as a site of leadership training for southern civil rights activists. Rosa  Parks  attended a 1955 workshop at Highlander four months before refusing to give up her bus seat, an act that ignited the  Montgomery bus boycott .

Lead by Septima  Clark , Esau Jenkins, and Bernice Robinson, Highlander developed a citizenship program in the mid-1950s that taught African Americans their rights as citizens while promoting basic literacy skills. Reflecting on his experiences with the Citizenship Schools and the emergence of new leaders from “noncharismatic people” who attended the training, Horton concluded that “educational work during social movement periods provides the best opportunity for multiplying democratic leadership” (Horton,  Long Haul , 127).

Horton, who claimed he had first met King during the civil right leader’s junior year at  Morehouse College , invited King to participate in Highlander’s anniversary celebration in 1957. While attending the celebration, an undercover agent sent by the Georgia Commission on Education took a photograph of King. The photo was sent throughout the South and used as a propaganda tool against King, with claims that it showed him attending a Communist training school.

Highlander continued to be a center for developing future leaders of the movement such as Marion  Barry , Diane  Nash , and James  Bevel . It was closed in 1961 when the Tennessee government revoked its charter on falsified charges that the school was being run for profit and that it did not fulfill its nonprofit requirements. The  Southern Christian Leadership Conference  (SCLC) took over the citizenship program that year, feeling that it offered, according to King, a plus for SCLC and the movement “in filling the need for developing new leadership as teachers and supervisors and providing the broad educational base for the population at large through the establishment of Citizenship Schools conducted by these new leaders throughout the South” (King, January 1961). Under the leadership of SCLC and the supervision of Clark, Dorothy  Cotton , and Andrew  Young , the schools eventually trained approximately 100,000 adults. In August 1961, Horton opened another school in Knoxville, Tennessee, called the Highlander Research and Education Center. He and the Center participated in the 1968  Poor People’s Campaign  and, after King’s  assassination , erected a tent complex at Resurrection City in Washington, D.C., holding workshops until police closed the encampment in June 1968.

Adams with Horton,  Unearthing Seeds of Fire , 1975.

Anne Braden to King, 23 September 1959, in  Papers  5:290–293 .

Glen,  Highlander , 1988.

Horton with Judith Kohl and Herbert Kohl,  Long Haul , 1990.

King, Memo, “Leadership Training Program and Citizenship Schools,” December 1960–January 1961,  SCLCR-GAMK .

King, “A Look to the Future,” Address Delivered at Highlander Folk School’s Twenty-fifth Anniversary Meeting, 2 September 1957, in  Papers  4:269–276 .

King to Braden, 7 October 1959, in  Papers  5:306–307 .

Reshoring, Automation, and Labor Markets Under Trade Uncertainty

Hamid Firooz

Download PDF (864 KB)

2024-16 | May 8, 2024

We study the implications of trade uncertainty for reshoring, automation, and U.S. labor markets. Rising trade uncertainty creates incentive for firms to reduce exposures to foreign suppliers by moving production and distribution processes to domestic producers. However, we argue that reshoring does not necessarily bring jobs back to the home country or boost domestic wages, especially when firms have access to labor-substituting technologies such as automation. Automation improves labor productivity and facilitates reshoring, but it can also displace jobs. Furthermore, automation poses a threat that weakens the bargaining power of low-skilled workers in wage negotiations, depressing their wages and raising the skill premium and wage inequality. The model predictions are in line with industry-level empirical evidence.

Advertisement

Supported by

Artificially Intelligent Help for Planning Your Summer Vacation

Travel-focused A.I. bots and more eco-friendly transportation options in online maps and search tools can help you quickly organize your seasonal getaway.

• Share full article

By J. D. Biersdorfer

J.D. Biersdorfer has been using smartphone apps for travel since the summer of 2007.

The summer travel season starts in just a few weeks, but if you are looking for software that goes beyond simply booking flights and hotels, you’re in luck. Search engines enhanced by artificial intelligence can help with your research and outline full itineraries. Certain older apps have recently been updated to suggest more eco-friendly travel options. And keeping it all organized on your phone is easier than ever. Here’s an overview.

Using an A.I. Travel Agent

General-purpose A.I.-powered search tools and chatbots like Google’s Gemini spin up a list of things to do on your vacation when asked, but A.I. bots that are fine-tuned for travel queries are often more comprehensive. These bots scout destinations, plan itineraries, search for accommodations and flights, map out road trips and do more — grabbing a lot of information at once and saving you all that time-consuming web trawling.

Give the software your specifics — like destination, length of stay, interests — and see what it suggests. Many A.I. helpers are free to use if you sign up for an account, but some charge a subscription fee for premium services; your app store has specifics.

Layla , formerly Roam Around, is one of the free vacation-oriented A.I. helpers you can find online, and it has teamed up with travel sites that include Skyscanner , Get Your Guide and Booking.com . If you prefer land-based car and camper journeys, Roadtrippers (free trial; $60 year) includes real-time traffic and air-quality information along with route planning. And old stalwarts like Tripadvisor and Expedia are now using A.I.-generated vacation builders.

But as others have also noted, while A.I. travel planners have much potential, many are still works in progress and usually display disclaimers admitting so.

A.I. bots have been known to offer generic advice like “enjoy lunch at a local restaurant,” suggest activities that are out of season or too far apart, repeatedly recommend the same restaurant, consistently steer you to their advertisers or point you to locations that have closed. If you ask different bots the same question, you may get nearly identical suggestions, all scraped from the same tourism websites.

Still, A.I. travel apps are improving as they learn, and can be useful for the trip research and coordination phase. Just be sure to double-check the bot’s work before you commit to a plan.

Finding Earth-Friendly Options

There’s no shortage of apps for booking transportation to your destination. But if you want to keep the environment in mind, recent updates to Google’s Maps and Search apps now suggest routes and methods that lower your personal impact on the planet.

Google for the past few years has been pointing people to flights with lower carbon emissions , alternative train routes , fuel-efficient driving directions and eco-friendly hotels . It is now expanding its walking, biking and public transit suggestions alongside car routes in several major cities and adding more electric-vehicle charging information. Google Flights shows jet emissions estimates . Google Search has a “consider taking the train” nudge with rail routes and prices under certain flight results.

Apple’s Maps app also shows mass transit , walking and cycling options for getting around town, along with charge-friendly routes for electric vehicles . However, the default apps on your phone are not the only aids. Third-party software for directions and sustainable travel abound.

For example, Citymapper, which covers most major cities in the United States, Europe and Asia, includes environmental impact statistics on some trips. Its directions often include accessibility options that avoid stairways , along with routes for the fastest, cheapest or easiest way to get where you’re going; Citymapper is free with in-app purchases.

Other apps available for those seeking environmentally minded vacations include Bikemap for community-sourced cycling routes around the world, HappyCow for vegan and vegetarian travelers and Tap Hydration and Water Stations to locate sources for refilling reusable water bottles.

Keeping Organized

If you don’t already have software for consolidating your trip information, your phone’s default apps can help. Electronic boarding passes, hotel reservations and advance tickets can be quickly added to the digital wallet on your phone; a pragmatic paper backup tucked in your bag is insurance. Google and Apple offer to automatically add reservations and events from email and messages to your calendar .

Free services like TripIt (and its phone apps ), TripCase (also with Android or iOS apps) and Wanderlog automatically put all your travel information in one place, typically by scanning the information in your confirmation emails. TripIt Pro , a $50-a-year subscription version, adds more features like seat, fare and airline-points trackers, as well as international travel tools and regional risk alerts like those for extreme weather that can affect airline schedules and public safety .

A.I. bots and travel apps will continue to evolve and, hopefully, make vacation planning even easier in the future. Just don’t forget to occasionally put the phone down and enjoy your time off once you get there.

J.D. Biersdorfer has been writing about consumer technology for The Times since 1998. She also creates the weekly interactive literary quiz for the Book Review and occasionally contributes reviews. More about J. D. Biersdorfer

Explore Our Coverage of Artificial Intelligence

News  and Analysis

As experts warn that A.I.-generated images, audio and video could influence the 2024 elections, OpenAI is releasing a tool designed to detect content created by DALL-E , its popular image generator.

American and Chinese diplomats plan to meet in Geneva to begin what amounts to the first, tentative arms control talks  over the use of A.I.

Wayve, a London maker of A.I. systems for autonomous vehicles, said that it had raised $1 billion , an illustration of investor optimism about A.I.’s ability to reshape industries.

The Age of A.I.

A new category of apps promises to relieve parents of drudgery, with an assist from A.I.  But a family’s grunt work is more human, and valuable, than it seems.

Despite Mark Zuckerberg’s hope for Meta’s A.I. assistant to be the smartest , it struggles with facts, numbers and web search.

Much as ChatGPT generates poetry, a new A.I. system devises blueprints for microscopic mechanisms  that can edit your DNA.

Which A.I. system writes the best computer code or generates the most realistic image? Right now, there’s no easy way to answer those questions, our technology columnist writes .