Confidential: For Review Only - BMJ · Confidential: For Review Only 4 40 Introduction 41 Vegetarian and vegan diets have become increasingly popular in recent years partly due to

Confidential: For Review Only

Vegetarians have a lower risk of ischaemic heart disease

but a higher risk of stroke: results from the prospective EPIC-Oxford study

Journal: BMJ

Manuscript ID BMJ.2018.045067

Article Type: Research

BMJ Journal: BMJ

Date Submitted by the Author: 15-May-2018

Complete List of Authors: Tong, Tammy; University of Oxford, Cancer Epidemiology Unit Appleby, Paul; Oxford University, Cancer Research U.K. Epidemiology Unit, Unive Bradbury, Kathryn; University of Auckland, School of Population Health; University of Oxford, Cancer Epidemiology Unit Perez-Cornago, Aurora; Oxford University, Cancer Epidemiology Unit, Nuffield Department of Population Health Travis, Ruth; University of Oxford, Cancer Epidemiology Unit Clarke, Robert; University of Oxford Nuffield Department of Population Health Key, Timothy; University of Oxford, Cancer Epidemiology Unit

Keywords: vegetarians, cardiovascular diseases, ischaemic heart disease, stroke,

nutritional epidemiology

https://mc.manuscriptcentral.com/bmj

BMJ


1

Vegetarians have a lower risk of ischaemic heart disease but a higher risk 1

of stroke: results from the prospective EPIC-Oxford study 2

Tammy YN Tong, nutritional epidemiologist1, Paul N Appleby, senior statistician

1, Kathryn 3

E Bradbury, nutritional epidemiologist1, Aurora Perez-Cornago, nutritional epidemiologist

1, 4

Ruth C Travis, associate professor and senior molecular epidemiologist1, Robert Clarke, 5

professor of epidemiology and public health medicine2, Timothy J Key, professor of 6

epidemiology and deputy director1 7

1. Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Richard Doll 8

Building, Old Road Campus, Oxford , UK 9

2. Clinical Trial Service Unit and Epidemiological Studies Unit, Nuffield Department of Population Health, 10

Big Data Institute, University of Oxford, Old Road Campus, Oxford, UK 11

12

Correspondence to: Tammy YN Tong [email protected] 13

14

Word count: Abstract 241; text 3492 15

15th

May 2018 16

17

Page 1 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


2

Abstract 18

Objective: To examine the associations of vegetarianism with risks of ischaemic heart 19

disease (IHD) and stroke. 20

Design: Prospective cohort study. 21

Setting: United Kingdom cohort with a large proportion of non-meat eaters. 22

Participants: Participants were recruited in 1993-2001 and classified into four diet groups: 23

regular meat eaters (≥50 g of meat per day: n=16 332), low meat eaters (<50 g of meat per 24

day: n=8096), fish eaters (n=7506), and vegetarians including vegans (n=16 254). 25

Main outcome measures: Incident cases of IHD and stroke (including ischaemic and 26

haemorrhagic types) identified through record linkage until mid-2016. 27

Results: Over 18.1 years of follow-up, 2820 cases of IHD, 1072 cases of total stroke (519 of 28

ischaemic stroke and 300 of haemorrhagic stroke), and 3565 cases with the composite of IHD 29

or stroke were recorded. Compared with regular meat eaters, the low meat eaters, fish eaters, 30

and vegetarians had 12%, 14%, and 24% lower risks of IHD, respectively (p 31

heterogeneity<0.0001). In contrast, low meat eaters, fish eaters and vegetarians had 22%, 32

28% and 29% higher risks of stroke (p heterogeneity=0.01). Overall, compared with regular 33

meat eaters, the vegetarians had a 14% lower risk of composite IHD or stroke (hazard ratio 34

0.86, 95% CI 0.78-0.95), equivalent to 8.6 fewer cases (95% CI 4.1-12.4 fewer) per 1000 35

population over 10 years. 36

Conclusions: Low meat eaters and non-meat eaters had lower risks of IHD and higher risks 37

of stroke. Vegetarians had a lower overall risk of IHD or stroke combined. 38

Page 2 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


3

What is already known on this topic

Vegetarian and vegan diets have become increasingly popular in recent years, but the

potential benefits and hazards of these diets are not fully understood.

Vegetarians have a lower risk of ischaemic heart disease compared to non-

vegetarians.

What this study adds

This study demonstrated that low meat (<50 g of meat per day) and non-meat eaters

(fish eaters, vegetarians, vegans) had lower risks of ischaemic heart disease, but

higher risks of stroke, compared with regular meat eaters (≥50 g of meat per day).

Vegetarians had lower risks of a composite of ischaemic heart disease or stroke.

Further research is needed to confirm the findings in other populations, and to identify

which factors might mediate the observed associations.

39

Page 3 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


4

Introduction 40

Vegetarian and vegan diets have become increasingly popular in recent years partly due to 41

perceived health benefits, and concerns about the environment and animal welfare.1 In the 42

United Kingdom (UK) it has been estimated that there are close to 1.7 million vegetarians 43

and vegans.2 The European Prospective Investigation into Cancer (EPIC)-Oxford study 44

previously reported that vegetarians had a 32% lower risk of ischaemic heart disease (IHD) 45

compared to non-vegetarians.3 Some, but not all, other prospective studies reported that 46

vegetarians had lower risks of IHD mortality than non-vegetarians.4–6

For stroke, combined 47

analyses of previous studies, published 20 years ago, reported no significant differences in 48

risk of stroke deaths between vegetarians and non-vegetarians.4 49

Secular trends in cardiovascular disease outcomes, including declines in mortality from IHD 50

and stroke, in addition to differences in the proportions of stroke types,7,8

have prompted the 51

need to re-examine the associations of vegetarian diets with subtypes of cardiovascular 52

disease in a contemporary setting. The aim of the present study was to examine the 53

associations of low or non-meat diets, including vegetarian and vegan diets, with risk of IHD 54

(including acute myocardial infarction), stroke (including stroke types) and a composite of 55

either IHD or stroke events in the EPIC-Oxford study, which included a large number of non-56

meat eaters. 57

Methods 58

Study population and design 59

The EPIC-Oxford study is a prospective cohort study of approximately 65 000 men and 60

women who were recruited across the UK. Details of the recruitment process have been 61

described previously.9 Individuals were recruited from either general practice (GP) or by 62

Page 4 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


5

postal questionnaire. The GP recruitment method recruited 7421 men and women aged 35 to 63

59 years who were registered with participating GPs, all of whom completed a full 64

questionnaire on their diet, lifestyle, health characteristics and medical history. The postal 65

recruitment preferentially targeted vegetarians, vegans, and other people interested in diet and 66

health, and recruited 57 990 participants aged ≥20 years. A full questionnaire was mailed to 67

all members of the Vegetarian Society and all surviving participants of the Oxford Vegetarian 68

Study,10

and respondents were invited to provide names and addresses of relatives and friends 69

who were also interested in receiving a questionnaire. A short questionnaire was also 70

distributed to all members of the Vegan Society, enclosed in vegetarian and health food 71

magazines, and displayed in health-food shops, and a full questionnaire was subsequently 72

mailed to all those who returned the short questionnaire. 73

At recruitment, all participants were asked whether they were willing to have their blood 74

pressure measured at their general practice and to provide a blood sample. Details of the 75

procedures for blood pressure measurement and blood sample collection, which were 76

conducted in subsets of the cohort, have been previously reported.3,11,12

The study protocol 77

was approved by a Multicentre Research Ethics Committee (Scotland A Research Ethics 78

Committee) and all participants provided written informed consent. 79

Assessment of diet group 80

The full questionnaire collected responses to questions about consumption of meat, fish, dairy 81

products, and eggs, in the form of “Do you eat any meat (including bacon, ham, poultry, 82

game, meat pies, sausages)?” or similar for the other three food groups. The questionnaire 83

also included a 130-item semi-quantitative food-frequency section which asked about dietary 84

intake over the past year, including 11 questions on consumption of meat and meat products 85

which were used to estimate total meat intake. Based on responses to these questions, 86

Page 5 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


6

participants were classified into one of five diet groups: regular meat eaters (participants who 87

ate ≥50 g of total meat per day, including red meat, processed meat, and poultry); low meat 88

eaters (participants who ate <50 g of total meat per day); fish eaters (participants who did not 89

eat meat but did eat fish); vegetarians (participants who did not eat meat or fish, but did eat 90

one or both of dairy products and eggs); and vegans (participants who did not eat meat, fish, 91

dairy products, or eggs). Due to the small number of vegans, vegetarians and vegans were 92

combined as a single diet group in the main analyses, but the two groups were also examined 93

separately for each outcome in secondary analyses. 94

Outcome ascertainment 95

Participants were followed up via record linkage to National Health Service records until 31st 96

March 2016 in England, 31st May 2016 in Wales, and 31

st October 2016 in Scotland. 97

Outcomes of interest were IHD (including acute myocardial infarction, AMI), total stroke 98

(including ischaemic, haemorrhagic, or unclassified stroke), and a composite of any IHD or 99

stroke. Details of events, using the relevant ICD9 or ICD10 codes, were obtained from 100

hospital records or death certificates. 101

Exclusion criteria 102

Participants who were not resident in England, Wales, or Scotland (n=945) were excluded, as 103

were those with no Hospital Episode Statistics data or National Health Service number 104

(n=20), those who completed the short questionnaire only (n=7619), or who were younger 105

than 20 years (n=1) or older than 90 years at recruitment (n=58), had no follow-up (n=364), 106

could not be traced by the National Health Service (n=14), of unknown diet group (n=132), 107

had unreliable nutrient data (n=1219), had a self-reported history of AMI, stroke, or angina 108

(n=6837), or had a date of diagnosis which preceded or equalled the date of recruitment 109

(n=14). 110

Page 6 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


7

Statistical analyses 111

Baseline characteristics and food and nutrient intakes of the EPIC-Oxford participants were 112

summarised by diet groups. For self-reported adiposity (body mass index (BMI), waist 113

circumference), and measures of blood pressure (SBP and DBP) and blood lipids (total 114

cholesterol, HDL-C, non-HDL-C), the means and 95% confidence intervals are presented, 115

after adjustment for sex and age at entry (5 year age groups), alcohol consumption (<1g, 1-116

7g, 8-15g, 16+ g/day), and physical activity (inactive, low activity, moderately active, very 117

active, unknown)13

. 118

Cox proportional hazards regression models were used to estimate the hazard ratios (HRs) 119

and 95% confidence intervals (CIs) for the associations between diet group (regular meat 120

eaters, low meat eaters, fish eaters, vegetarians including vegans) and each outcome of 121

interest, using regular meat eaters as the reference group. The underlying time variable was 122

the age at recruitment to the age at diagnosis, death, or administrative censoring, whichever 123

occurred first. For AMI, IHD, and composite IHD or stroke, events were censored on the 124

respective outcomes of interest. For total stroke, ischaemic stroke, and haemorrhagic stroke, 125

events were censored on any stroke. All analyses were stratified by sex, method of 126

recruitment (general practice or postal), and region (7 regions across the UK), and adjusted 127

for year of recruitment (per year), education (no qualifications, basic secondary [e.g. O level], 128

higher secondary [e.g. A level], degree, unknown), Townsend deprivation index (quartiles, 129

unknown)14

, smoking (never, former, light, heavy, unknown), alcohol consumption (<1g, 1-130

7g, 8-15g, 16+ g/day), physical activity (inactive, low activity, moderately active, very active, 131

unknown), and oral contraceptive use (no, yes ever, unknown) and hormone replacement 132

therapy use (no, yes ever, unknown) in women. Likelihood ratio tests were used to test for 133

heterogeneity of risk between diet groups. Self-reported prior high blood pressure (no, yes, 134

unknown), self-reported prior high blood cholesterol (no, yes, unknown), self-reported prior 135

Page 7 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


8

diabetes (no, yes, unknown), BMI (<20, 20-22.5, 22.5-25, 25-27.5, 27.5+ kg/m2, unknown), 136

total fruit and vegetable intake (continuous), total dietary fibre (continuous), and dietary 137

supplements use (no, yes, unknown) were assessed as potential mediators by adding each 138

variable one at a time to the previous model. 139

To estimate the population impact of vegetarian diets on cardiovascular health, the absolute 140

risk difference for each outcome between low meat or meat-free diets and regular meat diets 141

was assessed. Predicted incidence and absolute risk differences were presented as per 1000 142

population over 10 years. Predicted incidence over this time period of each outcome in the 143

regular meat eaters was calculated as (1- Sr)x 1000, where Sr = (1-observed incidence in 144

regular meat eaters)10

. For all other diet groups, predicted incidence was calculated as (1-145

SrHR

)x 1000, where HR represents the hazard ratio or confidence intervals of the hazard ratio 146

for each outcome in that diet group as previously estimated, and SrHR

represents the predicted 147

10 year survival (i.e. non-incidence) rate in the diet group. Absolute risk difference was then 148

calculated as the difference between the predicted incidence per 1000 population over 10 149

years between each diet group and the regular meat eaters. 150

Additional sensitivity analyses included exclusion of events occurring during the first five 151

years of follow-up, censoring at age 70 years, setting entering time as age 70, and performing 152

multiple imputation (with 10 imputations) for missing covariates. Heterogeneity in the 153

associations between diet group and risk of IHD or stroke by sex, age at recruitment (<60 or 154

≥60 years), smoking status (never, former, or current), BMI (<25 or ≥25 kg/m2), and presence 155

of risk factors (one or more of self-reported prior high blood pressure, high blood cholesterol 156

or diabetes) was assessed by adding appropriate interaction terms to the Cox models and 157

testing for statistical significance of interaction across strata using likelihood ratio tests. All 158

analyses were performed using Stata version 14.1 or 15.1 (Stata Corp, Texas, United States) 159

and P values of <0.05 were considered significant. 160

Page 8 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


9

Results 161

Baseline characteristics 162

The baseline characteristics of the study participants for each of the four diet groups (regular 163

meat eaters, low meat eaters, fish eaters, vegetarians including vegans) are presented in Table 164

1 (and separately for vegetarians and vegans in Supplementary Table 1). Overall, compared 165

with regular meat eaters, the low or non-meat eaters were younger, had lower area-level socio 166

economic status, but were more highly educated, less likely to smoke, had lower alcohol 167

consumption, were more physically active, and more likely to report dietary supplement use. 168

They were also less likely to report prior high blood pressure, prior high blood cholesterol, 169

prior diabetes, or receive long-term treatment for illnesses. In women, low or non-meat eaters 170

were more likely to report oral contraceptive use, but less likely to report hormone 171

replacement therapy use. Compared with regular meat eaters, low or non-meat eaters reported 172

having lower body mass index and had lower measured blood pressure. For blood lipids, 173

compared with regular meat eaters, vegetarians had approximately 0.5 mmol/L lower plasma 174

concentrations of total cholesterol and non-HDL-C. Low meat eaters and fish eaters had 175

higher plasma concentrations of HDL-C compared with regular meat eaters, while 176

vegetarians had similar concentrations. 177

Food and nutrient intakes in the different diet groups are presented in Table 2 and 178

Supplementary Table 2 (separately for vegetarians and vegans). Compared with regular meat 179

eaters, vegetarians had higher intakes of fruit and vegetables, and total fibre, and lower intake 180

of saturated fat (10% vs 12% energy, respectively). 181

Association of diet groups with cardiovascular diseases 182

Over 18.1 years of follow-up in 48 188 participants, there were 2820 cases of IHD (including 183

788 cases of AMI), 1072 cases of total stroke (including 519 cases of ischaemic stroke and 184

Page 9 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


10

300 cases of haemorrhagic stroke), and 3565 cases of composite IHD or stroke events. 185

Compared with regular meat eaters, low and non-meat eaters had a lower risk of IHD (hazard 186

ratios and 95% CIs for low meat eaters: 0.88, 0.80-0.98; fish eaters: 0.86, 0.76-0.98; 187

vegetarians: 0.76, 0.68-0.85; p for heterogeneity<0.0001) (Figure 1). Conversely, vegetarians 188

had a higher risk of haemorrhagic stroke compared with regular meat eaters (1.51, 1.10-2.08), 189

and all other diet groups had a higher risk of total stroke compared with regular meat eaters 190

(low meat eaters 1.22, 1.03-1.43; fish eaters: 1.28, 1.05-1.56; vegetarians: 1.29, 1.09-1.52; p 191

for heterogeneity=0.01). There were no significant differences between diet groups for the 192

risk of AMI or ischaemic stroke. Vegetarians (0.86, 0.78-0.95), but not the other diet groups 193

(low meat eaters: 0.94, 0.86-1.03; fish eaters: 0.93, 0.83-1.04), had a significantly lower risk 194

of IHD or stroke combined compared with regular meat eaters. When vegetarians and vegans 195

were assessed separately, vegans had a lower risk of IHD (0.75, 0.58-0.97) and a borderline 196

higher risk of total stroke (1.44, 1.00-2.06) compared with regular meat eaters 197

(Supplementary Table 3). 198

When the roles of potential mediators of the associations were assessed, the associations 199

between diet group and IHD attenuated marginally when adjusted for self-reported prior high 200

blood pressure, self-reported prior high blood cholesterol, and self-reported prior diabetes, 201

and to a larger extent for BMI, but all estimates remained significant (Supplementary Table 202

4). In contrast, the association between diet group and total stroke was slightly more extreme 203

and remained significant after adjustment for self-reported prior high blood pressure, and to a 204

lesser extent when adjusted for BMI. Additional adjustment for fruit and vegetable intake, 205

total fibre, and dietary supplement use had little effect on the results. 206

Absolute risk difference for vascular disease outcomes by diet group 207

Page 10 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


11

Table 3 shows the absolute risks of IHD, total stroke, and composite of IHD and stroke 208

associated with the four diet groups. In comparing the absolute risk difference of each 209

outcome by diet groups, vegetarian diets were associated with 12.1 fewer cases (95% CI 8.4-210

15.5 fewer) of IHD per 1000 population over 10 years compared to regular meat eaters. 211

Conversely, vegetarian diets were associated with 4.3 more cases (2.0-7.0) of total stroke. 212

Overall, for total IHD or stroke, vegetarian diets were associated with 8.6 fewer cases (4.1-213

12.4 fewer) per 1000 population over 10 years when compared with regular meat eaters, 214

while the absolute risk differences between low meat eaters (-3.7; -7.5, 0.9) or fish eaters (-215

4.3; -9.7, 1.4) and regular meat eaters were not significant. 216

Sensitivity and subgroup analyses 217

In sensitivity analyses excluding events occurring during the first five years of follow-up, and 218

performing multiple imputations for missing covariates, the results were similar 219

(Supplementary table 5). When the analyses were censored at age 70 or limited to those 220

above age 70 (setting entry time as age 70) respectively, a higher risk of ischaemic stroke was 221

observed in vegetarians below the age of 70 (censored at age 70) only, while a higher risk of 222

haemorrhagic stroke was observed in low meat eaters and vegetarians above the age of 70 223

only. 224

In subgroup analyses, stronger associations were observed in men compared to women for 225

diet group and IHD (Supplementary table 6). For composite IHD or stroke, there was a 226

marginally significant heterogeneity by age at recruitment (Supplementary table 8). No other 227

significant interactions by subgroups were observed for the association between diet group 228

and IHD or stroke, either as separate outcomes, or as a composite outcome (Supplementary 229

tables 6 to 8). 230

Page 11 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


12

Discussion 231

The present study is the first to examine the association of diet groups with risk of both IHD 232

and stroke, including subtypes, in a cohort with a large proportion of non-meat eaters. Low 233

meat and meat-free diets were associated with lower risks of IHD, but higher risks of total 234

stroke. Overall, vegetarian diets were associated with lower relative and absolute risks of 235

IHD or stroke combined. The associations persisted after further adjustment for potential 236

mediators, and in most sensitivity analyses. 237

Comparison with other studies 238

In a previous report from the EPIC-Oxford study based on 11.6 years of follow-up, 239

vegetarians (including vegans) had a 32% lower risk of IHD compared with non-vegetarians 240

(meat eaters or fish eaters, HR 0.68, 0.58-0.81),3 which was consistent with the updated 241

results of the present study that included a 2-fold greater number of IHD cases and four diet 242

groups (and five in secondary analyses). In a collaborative meta-analysis of five previous 243

prospective studies (Adventist Mortality Study, Health Food Shoppers Study, Adventist 244

Health Study, Heidelberg Study, Oxford Vegetarian Study) with median recruitment from 245

between 1960 to 1980, vegetarians had a 24% lower IHD death rate ratio (0.76, 0.62-0.94) 246

compared with non-vegetarians.4 However, recent combined analyses of EPIC-Oxford and 247

Oxford Vegetarian Study data (0.99, 0.79-1.23) reported no significant difference in IHD 248

mortality,6 while analysis of 21 years of data from the German Vegetarian Study reported a 249

lower point estimate in IHD mortality in vegetarians (0.70, 0.41-1.18), though the difference 250

was not statistically significant.15

Likewise, the Adventist Health Study 2 reported that the 251

relative risks for mortality from IHD and cardiovascular disease for vegetarians versus non-252

vegetarians were 0.81 (0.64-1.02) and 0.87 (0.75-1.01), respectively,.5 253

Page 12 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


13

For stroke, previous analyses reported no significant differences in cerebrovascular disease 254

mortality by diet group.4,6

No studies were found which reported the associations of 255

vegetarian diets with the risks for incident stroke or the two subtypes of stroke. 256

Interpretation of results and implications 257

The results of the present study demonstrated that compared with regular meat eaters, all 258

other diet groups had lower risks of IHD. The reason for such differences is not certain, but it 259

may be partly attributed to lower non-HDL-C levels associated with low meat or meat-free 260

diets16–18

. Previous meta-analyses of prospective studies demonstrated that lower non-HDL-C 261

levels or low density lipoprotein cholesterol (LDL-C) levels were associated with lower 262

IHD.19,20

Furthermore, both randomised trials of statin treatments21

and Mendelian 263

randomisation studies17

have confirmed a causal association of LDL-C with IHD. Hence, it 264

is highly likely that low or non-meat eaters, who have lower non-HDL-C (Table 1) or LDL-C 265

compared to regular meat eaters11,22

, should have lower risks of IHD. 266

For stroke outcomes, combined results from two randomised statin trials reported a 21% 267

higher risk of haemorrhagic stroke per 1 mmol/L lower LDL cholesterol,16,23

which is 268

consistent with results of observational studies of cholesterol levels and haemorrhagic 269

stroke24,25

, and broadly consistent with the findings of the present study which showed that 270

vegetarians had a higher risk of haemorrhagic stroke. However, previous prospective studies 271

and randomised trials have consistently demonstrated weak positive associations of non-272

HDL-C or LDL-C levels with ischaemic stroke risk, and a non-significant or weak positive 273

association with total stroke risk.19–21,23

A recent Mendelian randomisation study also 274

reported a positive association between LDL-C and ischaemic stroke.18

This discordance 275

between previous evidence and those of the present study for LDL-C levels and ischaemic 276

stroke suggest that alternative factors probably also contribute to the observed associations. 277

Page 13 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


14

The results of several Japanese studies, showing that individuals who do not consume animal 278

products had higher incidence and mortality from haemorrhagic and total stroke, and also a 279

possibly higher risk of ischaemic stroke mortality26–29

, suggest that some factors associated 280

with animal food consumption may be protective for stroke. Vegetarians and vegans in EPIC-281

Oxford have lower circulating levels of some of several nutrients (e.g. vitamin B1230

, vitamin 282

D31

, essential amino acids32

, and long chain n-3 polyunsaturated fatty acids33

), and 283

differences in some of these nutritional factors may possibly contribute to the observed 284

associations28,34–37

. Serum concentrations of these nutritional factors and non-HDL-C have 285

only been measured in a subset of the EPIC-Oxford cohort, and therefore their role in the 286

observed associations of vegetarian diets with IHD or stroke cannot be accurately determined 287

in the current context, but should be further investigated. 288

High blood pressure is an established major risk factor for both IHD and stroke.38,39

However, 289

given that the vegetarians had lower blood pressure, and the fact that the association between 290

diet group and stroke risk became slightly stronger after adjustment for prior history of high 291

blood pressure, this is unlikely to account for the higher risks of haemorrhagic and total 292

stroke observed among vegetarians in the present study. 293

Misclassification of stroke cases, such as the misclassification of haemorrhagic stroke as 294

ischaemic stroke, cannot be excluded as a possible explanation for the discrepant results for 295

ischaemic stroke, although previous adjudication studies of stroke types confirmed the 296

reliability of hospital admission records and death certificates in the UK for diagnosis of 297

stroke types in the UK population over this calendar period.40,41

298

Strengths and limitations 299

The strengths of this study include a large sample size, a long follow-up, and outcome 300

ascertainment by linkage to medical records which minimised the loss to follow-up. The 301

Page 14 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


15

analyses included adjustment for multiple confounders, assessment of several possible 302

mediators and sensitivity analyses to confirm the robustness of the results. Among the 303

limitations of the present study, diet group was self-reported and reasons for choosing each 304

diet were not recorded. It was not possible to take into account change in diet group over the 305

years, since information required to classify all participants into regular or low meat eaters at 306

time points beyond baseline was not available. However, in a large subsample, 97% of non-307

vegetarians and 90% of vegetarians maintained their diet group at 5 year follow-up42

, 308

suggesting that long term adherence to diet group in this cohort was high. Reverse causality is 309

possible, although results were similar after excluding the first 5 years of follow-up. 310

Information on medication use (including statins) at recruitment was not available. 311

Generalisability may be limited, since the present study was based predominantly on white 312

European individuals. 313

Conclusions 314

Overall, the present study demonstrated that UK adults who consume low meat or meat-free 315

diets had lower risks of IHD, but higher risks of stroke compared with regular meat-eaters. 316

Future work should include measurements of circulating levels of cholesterol sub-fractions, 317

vitamin B12, and amino acids or fatty acids in representative samples of the different diet 318

groups to identify which factors might mediate the observed associations. Additional studies 319

in other large-scale cohorts with a high proportion of non-meat eaters are needed to confirm 320

the generalisability of these results and assess their relevance for clinical practice and public 321

health. 322

323

324

Page 15 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


16

Statements 325

Acknowledgements and funding 326

We thank all participants in the EPIC-Oxford cohort for their invaluable contribution. The 327

work is supported by the UK Medical Research Council MR/M012190/1 and Wellcome Trust 328

Our Planet Our Health (Livestock, Environment and People, LEAP 205212/Z/16/Z). KEB is 329

supported by the Girdlers’ New Zealand Health Research Council Fellowship. 330

Conflicts of interest 331

TJK is a member of the Vegan Society; the other authors had no conflicts of interest. The 332

funders had no role in the design of the study; in the collection, analyses, or interpretation of 333

data; in the writing of the manuscript; or in the decision to publish the results. 334

Availability of data and materials 335

The data access policy for EPIC-Oxford is available via the study website (http://www.epic-336

oxford.org/data-access-sharing-and-collaboration/). 337

Author’s contributions 338

TYNT and TJK conceived and designed the research question. TYNT analysed the data and 339

wrote the first draft of the manuscript, and PNA, KEB, APC, RCT, RC and TJK provided 340

input on data analysis and interpretation of results. All authors revised the manuscript 341

critically for important intellectual content, and read and approved the final manuscript. 342

Transparency declaration 343

The lead author affirms that this manuscript is an honest, accurate, and transparent account of 344

the study being reported; that no important aspects of the study have been omitted; and that 345

Page 16 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

http://www.epic-oxford.org/data-access-sharing-and-collaboration/

http://www.epic-oxford.org/data-access-sharing-and-collaboration/


17

any discrepancies from the study as planned (and, if relevant, registered) have been 346

explained.347

Page 17 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


18

Table 1: Baseline characteristics of participants in different diet groups in EPIC-Oxford (n=48 188).

Diet groups

Characteristics

Regular meat

eaters1

Low meat eaters1 Fish eaters Vegetarians and

vegans

Mean (SD) or n (%) (n=16 332) (n=8096) (n=7506) (n=16 254)

Socio-demographic

Age, years 49.7 (12.8) 47.7 (13.5) 42.1 (12.8) 39.4 (13.1)

Sex, women (%) 12 048 (73.8) 6433 (79.5) 6186 (82.4) 12 232 (75.3)

Top socio-economic quartile

(%)2 4239 (29.5) 1720 (24.8) 1431 (21.9) 3018 (21.2)

Higher education (%) 4216 (25.8) 3158 (39.0) 3308 (44.1) 6698 (41.2)

Lifestyle

Current smokers, (%) 2074 (12.7) 881 (10.9) 764 (10.2) 1685 (10.4)

Alcohol consumption, g/day 10.5 (13.4) 9.2 (11.6) 10.0 (12.3) 9.3 (12.8)

Moderate/ high physical

Activity, (%) 4250 (26.0) 2502 (30.9) 2684 (35.8) 5849 (36.0)

Dietary supplement use, (%)3 8396 (52.4) 4899 (62.0) 4702 (64.1) 8961 (56.1)

Medical history, n (%)

Prior high blood pressure 2111 (13.0) 827 (10.3) 549 (7.3) 935 (5.8)

Prior high blood cholesterol 1174 (7.2) 442 (5.5) 255 (3.4) 345 (2.1)

Prior diabetes, 271 (1.7) 82 (1.0) 61 (0.8) 93 (0.6)

Receiving long term

treatment for any illness 4962 (30.8) 2060 (25.8) 1622 (21.9) 3077 (19.1)

Oral contraceptive use4 8635 (72.1) 4628 (72.4) 4959 (80.5) 9620 (79.0)

Hormone replacement

therapy use4 3199 (26.9) 1285 (20.3) 728 (11.9) 954 (7.9)

Physical measurements, mean

(95% CI)5

Body mass index (kg/m2) 24.4 (24.4, 24.5) 23.4 (23.3, 23.5) 23.1 (23.0, 23.2) 23.0 (23.0, 23.1)

Systolic blood pressure

(mmHg) 126.3 (125.8, 126.7) 124.4 (123.7, 125.0) 123.4 (122.6, 124.2) 123.7 (123.1, 124.2)

Diastolic blood pressure

(mmHg) 77.3 (77.1, 77.6) 76.5 (76.1, 76.9) 75.5 (75.0, 76.0) 75.9 (75.6, 76.2)

Total cholesterol (mmol/L) 5.52 (5.47, 5.57) 5.44 (5.35, 5.52) 5.31 (5.23, 5.39) 4.97 (4.92, 5.03)

HDL cholesterol (mmol/L) 1.31 (1.29, 1.33) 1.35 (1.32, 1.38) 1.35 (1.32, 1.38) 1.29 (1.28, 1.31)

Non-HDL cholesterol

(mmol/L) 4.22 (4.17, 4.27) 4.08 (4.00, 4.17) 3.96 (3.88, 4.04) 3.68 (3.62, 3.74)

Estimates shown are mean (SD), n (%), or adjusted means (95% confidence interval), as stated in left column. For all

variables, p-heterogeneity between diet groups was<0.001.

1 Regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low meat eaters were

defined as participants with total meat consumption <50 g per day. 2 Based on Townsend index. 3 Defined as regularly taking any vitamins, minerals, fish oils, fibre or other food supplements during the last 12 months. 4 In women only. 5 Body mass index and waist circumference were based on self-reported measures in the whole cohort. Blood pressure was

measured in 6519 regular meat eaters, 2343 low meat eaters, 1742 fish eaters, and 4364 vegetarians and vegans. Blood lipids

were measured in 1487 regular meat eaters, 498 low meat eaters, 566 fish eaters, and 1109 vegetarians and vegans. Estimates

were adjusted for the cross-stratification of gender and age at entry (5 year age groups), alcohol consumption (<1g, 1-7g, 8-

15g, 16+ g/day), and physical activity (inactive, low activity, moderately active, very active, unknown).

Page 18 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


19

Table 2: Food and nutrient intakes of participants in different diet groups in EPIC-Oxford (n=48 188).

Diet groups

Foods or nutrients Regular meat

eaters1

Low meat eaters1 Fish eaters Vegetarians and

vegans (n=16 332) (n=8096) (n=7506) (n=16 254)

Foods

Total meat and meat

products (g/day) 100.1 (41.5) 28.2 (12.9) --- ---

Red and processed meat (g/day) 65.9 (38.4) 18.8 (10.3) --- ---

Poultry (g/day) 34.2 (22.1) 9.4 (5.6) --- ---

Total fish (g/day) 43.6 (28.8) 38.4 (29.3) 38.5 (33.3) ---

Dairy milk (ml/day) 336.5 (183.3) 299.2 (184.9) 274.0 (189.8) 232.3 (207.3)

Soya milk (ml/day) 3.7 (34.4) 10.4 (56.7) 21.1 (79.4) 54.5 (127.9)

Dairy cheese (g/day) 19.7 (17.5) 22.8 (20.4) 27.5 (24.2) 26.8 (25.4)

Total fresh fruit (g/day) 249.7 (187.9) 295.2 (239.3) 291.1 (228.0) 283.8 (239.2)

Total vegetables (g/day) 248.7 (123.1) 259.3 (147.6) 287.3 (148.7) 294.4 (163.2)

Nutrients

Carbohydrates (% energy) 46.8 (5.9) 50.4 (6.3) 51.0 (6.5) 52.8 (6.8)

Protein (% energy) 17.6 (2.9) 15.6 (2.5) 14.6 (2.3) 13.6 (2.1)

Total fat (% energy) 32.1 (5.6) 30.5 (6.2) 30.8 (6.3) 30.2 (6.6)

Saturated fat (% energy) 11.8 (3.2) 10.9 (3.4) 10.6 (3.3) 10.2 (3.5)

Monounsaturated fat (% energy) 11.0 (2.2) 10.0 (2.4) 9.9 (2.4) 9.7 (2.6)

Polyunsaturated fat (% energy) 6.3 (1.8) 6.4 (2.2) 7.0 (2.2) 7.1 (2.5)

Total fibre (g/day) 18.4 (6.4) 19.5 (7.4) 21.2 (7.4) 22.1 (8.0)

Total energy (kJ/day) 8,633 (2,227) 7,624 (2,141) 7,939 (2,199) 7,813 (2,234)

Estimates shown are mean (SD). For all variables, p-heterogeneity between diet groups was<0.001.

1 Regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low meat eaters were

defined as participants with total meat consumption <50 g per day.

Page 19 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


20

Figure 1: Prospective associations between diet groups (regular and low meat eaters, fish eaters, vegetarians and vegans)

and risk of ischaemic heart disease and stroke in EPIC-Oxford (n=48 188).

Regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low meat eaters were defined as

participants with total meat consumption <50 g per day. Total analyses times were 848 969 person-years for acute myocardial

infarction, 836 980 person-years for ischaemic heart disease, 848 840 person-years for ischaemic stroke, haemorrhagic stroke, or

total stroke, and 833 604 person-years for ischaemic heart disease or stroke. 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), education (no

qualifications, age 16, age 18, degree, unknown), Townsend deprivation index (quartiles, unknown), smoking (never, former,

light, heavy, unknown), alcohol consumption(<1g, 1-7g, 8-15g, 16+ g/day), physical activity (inactive, low activity, moderately

active, very active, unknown), and oral contraceptive and hormone replacement therapy use in women. P-heterogeneity represents

significance of heterogeneity in risk between diet groups based on likelihood ratio tests. Box sizes are proportional to the number

of cases in each group.

Page 20 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


21

Table 3: Absolute risk difference (per 1000 population over 10 years) of ischaemic heart disease

and stroke in different diet groups in EPIC-Oxford.

Outcome and diet groups Predicted incidence per

1000 population over 10

years1

Absolute risk difference

per 1000 population over

10 years2

Ischaemic heart disease

Regular meat eaters3 51.3 (48.4, 54.4) Reference

Low meat eaters3 45.3 (41.8, 49.2) -6.0 (-9.4, -2.0)

Fish eaters 44.3 (39.7, 49.5) -7.0 (-11.6, -1.7)

Vegetarians and vegans 39.2 (35.8, 42.8) -12.1 (-15.5, -8.4)

Total stroke


Low meat eaters3 18.4 (16.2, 20.8) 3.3 (1.1, 5.7)

Fish eaters 19.3 (16.3, 22.7) 4.2 (1.2, 7.6)

Vegetarians and vegans 19.5 (17.1, 22.1) 4.3 (2.0, 7.0)

Ischaemic heart disease or stroke


Low meat eaters3 59.3 (55.5, 63.8) -3.7 (-7.5, 0.9)

Fish eaters 58.7 (53.3, 64.4) -4.3 (-9.7, 1.4)

Vegetarians and vegans 54.4 (50.6, 58.8) -8.6 (-12.4, -4.1)

1 For regular meat eaters, this was calculated as (1- Sr)x 1000, where Sr = (1-observed incidence in

regular meat eaters)10

, and represents the predicted 10 year survival rate in the regular meat eaters.

For all other diet groups, this was calculated as (1- SrHR

)x 1000, where HR represents the hazard ratio

or confidence intervals of the hazard ratio for each outcome in that diet group, and SrHR

represents the

predicted 10 year survival (i.e. non-incidence) rate in the diet group. Confidence intervals in all diet

groups were estimated based on floating absolute risk. 2 Calculated as the difference between the predicted incidence per 1000 population over 10 years

between each diet group and the regular meat eaters. 3 Regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low

meat eaters were defined as participants with total meat consumption <50 g per day.

Page 21 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


22

References

1. Leahy E, Lyons S, Tol RSJ. An estimate of the number of vegetarians in the world

[Internet]. ESRI working paper 340. 2010 [cited 2018 May 14]. Available from:

https://www.esri.ie/publications/an-estimate-of-the-number-of-vegetarians-in-the-

world/

2. The Vegan Society. Find out how many vegans are in Great Britain [Internet]. 2016

[cited 2018 Jan 29]. Available from: https://www.vegansociety.com/whats-

new/news/find-out-how-many-vegans-are-great-britain

3. Crowe FL, Appleby PN, Travis RC, Key TJ. Risk of hospitalization or death from

ischemic heart disease among British vegetarians and nonvegetarians: results from the

EPIC-Oxford cohort study. Am J Clin Nutr. 2013;97:597–603.

4. Key TJ, Fraser GE, Thorogood M, et al. Mortality in vegetarians and nonvegetarians:

detailed findings from a collaborative analysis of 5 prospective studies. Am J Clin

Nutr. 1999;70:516S–524S.

5. Orlich MJ, Singh PN, Sabaté J, et al. Vegetarian dietary patterns and mortality in

Adventist Health Study 2. JAMA Intern Med. 2013;173:1230–8.

6. Appleby PN, Crowe FL, Bradbury KE, Travis RC, Key TJ. Mortality in vegetarians

and comparable nonvegetarians in the United Kingdom. Am J Clin Nutr.

2016;103:218–30.

7. Lawlor DA, Smith GD, Leon DA, Sterne JAC, Ebrahim S. Secular trends in mortality

by stroke subtype in the 20th century: a retrospective analysis. Lancet.

2002;360:1818–23.

Page 22 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


23

8. British Heart Foundation. Cardiovascular disease statistics 2017 [Internet]. 2017 [cited

2018 Feb 14]. Available from: https://www.bhf.org.uk/research/heart-statistics/heart-

statistics-publications/cardiovascular-disease-statistics-2017

9. Davey GK, Spencer EA, Appleby PN, 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. 2003;6:259–68.

10. Appleby PN, Thorogood M, Mann JI, Key TJ. The Oxford Vegetarian Study: an

overview. Am J Clin Nutr. 1999;70:525S–531S.

11. Bradbury KE, Crowe FL, Appleby PN, et al. Serum concentrations of cholesterol,

apolipoprotein A-I and apolipoprotein B in a total of 1694 meat-eaters, fish-eaters,

vegetarians and vegans. Eur J Clin Nutr. 2014;68:178–83.

12. Appleby PN, Davey GK, Key TJ. Hypertension and blood pressure among meat eaters,

fish eaters, vegetarians and vegans in EPIC-Oxford. Public Health Nutr. 2002;5:645–

54.

13. Wareham NJ, Jakes RW, Rennie KL, 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.

14. Townsend P. Poverty in the United Kingdom: A Survey of Household Resources and

Standards of Living. Allen Lane and Penguin Books. London, UK; 1979.

15. Chang-Claude J, Hermann S, Eilber U, Steindorf K. Lifestyle determinants and

mortality in German vegetarians and health-conscious persons: results of a 21-year

follow-up. Cancer Epidemiol Biomarkers Prev. 2005;14:963–8.

Page 23 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


24

16. Collins R, Reith C, Emberson J, et al. Interpretation of the evidence for the efficacy

and safety of statin therapy. Lancet. 2016;388:2532–61.

17. Ference BA, Yoo W, Alesh I, et al. Effect of long-term exposure to lower low-density

lipoprotein cholesterol beginning early in life on the risk of coronary heart disease: a

mendelian randomization analysis. J Am Coll Cardiol. 2012;60:2631–9.

18. Hindy G, Engström G, Larsson SC, et al. Role of blood lipids in the development of

ischemic stroke and its subtypes: a mendelian randomization study. Stroke.

2018;49:820–7.

19. MacMahon S, Duffy S, Rodgers A, et al. Blood cholesterol and vascular mortality by

age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective

studies with 55 000 vascular deaths. Lancet. 2007;370:1829–39.

20. Emerging Risk Factors Collaboration, Di Angelantonio E, Sarwar N, et al. Major

lipids, apolipoproteins, and risk of vascular disease. JAMA. 2009;302:1993–2000.

21. Cholesterol Treatment Trialists’ (CTT) Collaboration, Baigent C, Blackwell L, et al.

Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of

data from 170,000 participants in 26 randomised trials. Lancet. 2010;376:1670–81.

22. Thorogood M, Carter R, Benfield L, McPherson K, Mann JI. Plasma lipids and

lipoprotein cholesterol concentrations in people with different diets in Britain. Br Med

J. 1987;295:351–3.

23. Amarenco P, Bogousslavsky J, Callahan A, et al. High-dose atorvastatin after stroke or

transient ischemic attack. N Engl J Med. 2006;355:549–59.

24. Iso H, Jacobs DR, Wentworth D, Neaton JD, Cohen JD. Serum cholesterol levels and

Page 24 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


25

six-year mortality from stroke in 350,977 men screened for the multiple risk factor

intervention trial. N Engl J Med. 1989;320:904–10.

25. Yano K, Reed DM, MacLean CJ. Serum cholesterol and hemorrhagic stroke in the

Honolulu Heart Program. Stroke. 1989;20:1460–5.

26. Kinjo Y, Beral V, Akiba S, et al. Possible protective effect of milk, meat and fish for

cerebrovascular disease mortality in Japan. J Epidemiol. 1999;9:268–74.

27. Sauvaget C, Nagano J, Allen N, Grant EJ, Beral V. Intake of animal products and

stroke mortality in the Hiroshima/Nagasaki Life Span Study. Int J Epidemiol.

2003;32:536–43.

28. Takeya Y, Popper JS, Shimizu Y, et al. Epidemiologic studies of coronary heart

disease and stroke in Japanese men living in Japan, Hawaii and California: incidence

of stroke in Japan and Hawaii. Stroke. 1977;15:15–23.

29. Sauvaget C, Nagano J, Hayashi M, Yamada M. Animal protein, animal fat, and

cholesterol intakes and risk of cerebral infarction mortality in the adult health study.

Stroke. 2004;35:1531–7.

30. Gilsing AMJ, Crowe FL, Lloyd-Wright Z, 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.

31. Crowe FL, Steur M, Allen NE, et al. 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.

32. Schmidt JA, Rinaldi S, Scalbert A, et al. Plasma concentrations and intakes of amino

Page 25 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


26

acids in male meat-eaters, fish-eaters, vegetarians and vegans: a cross-sectional

analysis in the EPIC-Oxford cohort. Eur J Clin Nutr. 2016;70:306–12.

33. Rosell MS, Lloyd-Wright Z, Appleby PN, et al. Long-chain n-3 polyunsaturated fatty

acids in plasma in British meat-eating, vegetarian, and vegan men. Am J Clin Nutr.

2005;82:327–34.

34. Dong H, Pi F, Ding Z, et al. Efficacy of supplementation with B vitamins for stroke

prevention: a network meta-analysis of randomized controlled trials. PLoS One.

2015;10:1–12.

35. Iso H, Stampfer MJ, Manson JE, et al. Prospective study of fat and protein intake and

risk of intraparenchymal hemorrhage in women. Circulation. 2001;103:856–63.

36. Sacks FM, Lichtenstein AH, Wu JHY, et al. Dietary fats and cardiovascular disease: a

presidential advisory from the American Heart Association. Circulation. 2017;136:e1–

23.

37. McGreevy C, Williams D. New insights about vitamin D and cardiovascular disease: a

narrative review. Ann Intern Med. 2011;155:820–6.

38. Lim SS, Vos T, Flaxman AD, et al. Global, regional, and national comparative risk

assessment of 79 behavioural, environmental and occupational, and metabolic risks or

clusters of risks in 188 countries, 1990-2013: a systematic analysis for the Global

Burden of Disease Study 2013. Lancet. 2016;380:2224–60.

39. Rosendorff C, Lackland DT, Allison M, et al. Treatment of hypertension in patients

with coronary artery disease: a scientific statement from the American Heart

Association, American College of Cardiology, and American Society of Hypertension.

Circulation. 2015;131:e435–70.

Page 26 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


27

40. Sinha S, Myint PK, Luben RN, Khaw K-T. Accuracy of death certification and

hospital record linkage for identification of incident stroke. BMC Med Res Methodol.

2008;8:74.

41. Wright FL, Green J, Canoy D, et al. Vascular disease in women: comparison of

diagnoses in hospital episode statistics and general practice records in England. BMC

Med Res Methodol. 2012;12:161.

42. Key TJ, Appleby PN, Spencer EA, et al. Mortality in British vegetarians: results from

the European Prospective Investigation into Cancer and Nutrition. Am J Clin Nutr.

2009;1960:1–4.

Page 27 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary table 1: Baseline characteristics of vegetarians and vegans in EPIC-Oxford.

Diet groups

Characteristics Vegetarians Vegans

Max n=14 422 Max n=1832

Socio-demographic

Age, years (SD) 39.5 (13.1) 38.6 (13.2)

Sex, women (%) 11 063 (76.7) 1169 (63.8)

Top socio-economic quartile (%)1 2728 (21.6) 290 (18.0)

Degree education (%) 5949 (41.2) 749 (40.9)

Lifestyle

Current smokers (%) 1484 (10.3) 201 (11.0)

Alcohol consumption, g/day (SD) 9.4 (12.7) 8.5 (13.7)

Moderate/ high physical activity (%) 5088 (35.3) 761 (41.5)

Dietary supplement use, (%)2 8026 (56.6) 935 (52.1)

Medical history, n (%)

Prior high blood pressure (% yes) 850 (5.9) 85 (4.6)

Prior high blood cholesterol (% yes) 317 (2.2) 28 (1.6)

Prior diabetes (% yes) 86 (0.6) 7 (0.4)

Receiving long term treatment for any

illness (% yes) 2778 (19.4) 299 (16.4)

Oral contraceptive use (% yes)3 8791 (79.8) 829 (71.2)

Hormone replacement therapy use

(% yes)3 887 (8.1) 67 (5.9)

Physical measurements, mean (95% CI)4

Body mass index (kg/m2) 23.1 (23.1, 23.2) 22.3 (22.1, 22.4)

Systolic blood pressure (mmHg) 124.0 (123.4, 124.5) 121.7 (120.5, 123.0)

Diastolic blood pressure (mmHg) 76.1 (75.8, 76.5) 74.4 (73.6, 75.2)

Total cholesterol (mmol/L) 5.10 (5.03, 5.17) 4.76 (4.67, 4.85)

HDL cholesterol (mmol/L) 1.30 (1.27, 1.32) 1.29 (1.26, 1.32)

Non-HDL cholesterol (mmol/L) 3.80 (3.73, 3.87) 3.47 (3.38, 3.56)

Estimates shown are mean (SD), n (%), or adjusted means (95% confidence interval), as stated in left column.

1 Based on Townsend index.

2 Defined as regularly taking any vitamins, minerals, fish oils, fibre or other food supplements during the last 12 months.

3 In women only.

4 Body mass index and waist circumference were based on self-reported measures in the whole cohort. Blood lipids were

measured in 674 vegetarians and 435 vegans, and blood pressure was measured in 3688 vegetarians and 676 vegans.

Estimates were adjusted for the cross-stratification of gender and age at entry (5 year age groups), alcohol consumption (<1g,

1-7g, 8-15g, 16+ g/day), and physical activity (inactive, low activity, moderately active, very active, unknown).

Page 28 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary table 2: Food and nutrient intakes of vegetarians and vegans in EPIC-Oxford.

Diet groups

Foods or nutrients Vegetarians Vegans

Max n=14 422 Max n=1832

Foods

Dairy milk (ml/day) 260.6 (202.3) ---

Soya milk (ml/day) 32.7 (98.1) 228.4 (189.7)

Dairy cheese (g/day) 30.0 (25.1) ---

Total fresh fruit (g/day) 275.8 (221.0) 346.2 (345.1)

Total vegetables (g/day) 288.0 (157.8) 345.1 (193.4)

Nutrients

Carbohydrates (% energy) 52.5 (6.6) 55.4 (7.8)

Protein (% energy) 13.6 (2.1) 13.3 (2.3)

Total fat (% energy) 30.5 (6.5) 28.1 (7.3)

Saturated fat (% energy) 10.7 (3.4) 6.5 (2.1)

Monounsaturated fat (% energy) 9.7 (2.5) 9.4 (3.4)

Polyunsaturated fat (% energy) 6.8 (2.3) 9.1 (2.9)

Total fibre (g/day) 21.6 (7.7) 25.9 (9.3)

Total energy (kJ/day) 7,871 (2,213) 7,359 (2,346)

Estimates shown are mean (SD).

Page 29 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary table 3. Prospective associations between 5 diet groups and risk of cardiovascular diseases in EPIC-Oxford

(n=48,188).

Outcome and diet groups1 N Cases Hazard ratios (95%

confidence intervals)2

P-heterogeneity2

Acute myocardial infarction

Regular meat eaters 16 332 399 Reference

Low meat eaters 8096 152 0.99 (0.82,1.20)

Fish eaters 7506 85 1.00 (0.78,1.28)

Vegetarians 14 422 137 0.93 (0.75,1.15)

Vegans 1832 15 0.72 (0.42,1.21) 0.76



Low meat eaters 8096 535 0.88 (0.80,0.98)

Fish eaters 7506 307 0.86 (0.76,0.98)

Vegetarians 14 422 453 0.76 (0.68,0.85)

Vegans 1832 63 0.75 (0.58,0.97) <0.0001

Ischaemic stroke


Low meat eaters 8096 115 1.17 (0.93,1.48)

Fish eaters 7506 62 1.11 (0.82,1.48)

Vegetarians 14 422 106 1.19 (0.93,1.54)

Vegans 1832 18 1.58 (0.96,2.59) 0.31

Haemorrhagic stroke


Low meat eaters 8096 64 1.37 (1.00,1.89)

Fish eaters 7506 41 1.33 (0.91,1.94)

Vegetarians 14 422 79 1.55 (1.12,2.13)

Vegans 1832 8 1.24 (0.59,2.59) 0.09

Total stroke


Low meat eaters 8096 232 1.22 (1.03,1.43)

Fish eaters 7506 142 1.28 (1.05,1.56)

Vegetarians 14 422 230 1.27 (1.07,1.51)

Vegans 1832 34 1.44 (1.00,2.06) 0.02



Low meat eaters 8096 714 0.94 (0.86,1.03)

Fish eaters 7506 414 0.93 (0.83,1.04)

Vegetarians 14 422 637 0.86 (0.78,0.95)

Vegans 1832 92 0.90 (0.72,1.11) 0.05 1 For diet group, regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low meat

eaters were defined as participants with total meat consumption <50 g per day. 2 Total analyses time was 848 969 person-years for acute myocardial infarction, 836 980 person-years for ischaemic heart

disease, 848 840 person-years for ischaemic stroke, haemorrhagic stroke, or total stroke, and 833 604 person-years for

ischaemic heart disease or stroke. 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), education (no qualifications, age 16, age 18, degree,

unknown), Townsend deprivation index (quartiles, unknown), smoking (never, former, light, heavy, unknown), alcohol

consumption(<1g, 1-7g, 8-15g, 16+ g/day), physical activity (inactive, low activity, moderately active, very active,

unknown), and oral contraceptive and hormone replacement therapy use in women. P-heterogeneity represents significance of

heterogeneity in risk between diet groups based on likelihood ratio tests.

Page 30 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary table 4. Prospective associations between diet groups and risk of cardiovascular diseases in EPIC-Oxford, further adjusted for potential mediators of the

associations (n =48,188).

Hazard ratios (95% confidence intervals) with additional adjustment for2

Outcome and diet

groups1

Self-reported

prior high blood

pressure

Self-reported

prior high blood

cholesterol

Self-reported

prior diabetes

BMI3 Fruit and

vegetable intake4

Total fibre4 Supplement use

5


Regular meat eaters Reference Reference Reference Reference Reference Reference Reference

Low meat eaters 0.90 (0.81,1.00) 0.89 (0.80,0.98) 0.90 (0.81,0.99) 0.92 (0.84,1.02) 0.88 (0.80,0.98) 0.88 (0.80,0.98) 0.88 (0.80,0.97)

Fish eaters 0.89 (0.78,1.01) 0.87 (0.77,0.99) 0.87 (0.77,0.99) 0.93 (0.82,1.06) 0.86 (0.76,0.98) 0.87 (0.77,0.99) 0.86 (0.75,0.98)

Vegetarians and vegans 0.80 (0.71,0.89) 0.78 (0.70,0.87) 0.78 (0.70,0.87) 0.82 (0.74,0.92) 0.76 (0.68,0.85) 0.77 (0.69,0.86) 0.76 (0.68,0.85)

P-heterogeneity2 0.0006 0.0001 0.0001 0.007 <0.0001 0.0001 <0.0001

Total stroke


Low meat eaters 1.23 (1.04,1.45) 1.22 (1.03,1.43) 1.22 (1.04,1.44) 1.23 (1.04,1.45) 1.21 (1.03,1.43) 1.22 (1.03,1.43) 1.22 (1.04,1.44)

Fish eaters 1.31 (1.07,1.59) 1.27 (1.04,1.55) 1.29 (1.06,1.57) 1.30 (1.06,1.59) 1.27 (1.04,1.54) 1.28 (1.04,1.56) 1.29 (1.06,1.57)


P-heterogeneity2 0.002 0.01 0.005 0.005 0.01 0.01 0.007


or stroke


Low meat eaters 0.96 (0.88,1.05) 0.95 (0.87,1.04) 0.95 (0.87,1.04) 0.98 (0.90,1.07) 0.94 (0.86,1.03) 0.94 (0.86,1.03) 0.94 (0.86,1.03)

Fish eaters 0.96 (0.86,1.07) 0.94 (0.84,1.05) 0.94 (0.84,1.05) 0.99 (0.88,1.10) 0.93 (0.83,1.04) 0.94 (0.84,1.05) 0.93 (0.83,1.04)


P-heterogeneity2 0.22 0.07 0.08 0.39 0.02 0.05 0.02

1 For diet group, regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low meat eaters were defined as participants with total meat consumption

<50 g per day. 2 Total analyses time was 836 980 person-years for IHD, 848 932 person-years for total stroke, and 833 604 person-years for IHD or stroke. 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), education (no qualifications, age 16, age 18, degree, unknown),

Townsend deprivation index (quartiles, unknown), smoking (never, former, light, heavy, unknown), alcohol consumption(<1g, 1-7g, 8-15g, 16+ g/day), physical activity (inactive, low

activity, moderately active, very active, unknown), and oral contraceptive and hormone replacement therapy use in women. P-heterogeneity represents significance of heterogeneity in risk

between diet groups based on likelihood ratio tests. 3 BMI was adjusted for categorically (<20, 20-22.5, 22.5-25, 25-27.5, 27.5+ kg/m

2, unknown).

4 Fruit and vegetable intake and total fibre were adjusted for continuously.

5 Supplement use was adjusted for categorically (no, yes, unknown).

Page 31 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary table 5: Sensitivity analyses on prospective association between diet group and risk of ischaemic heart diseases and stroke.

Hazard ratios (95% confidence intervals)1

Outcome and diet groups Excluding first 5 years of

follow-up

N=46 757

Censoring at age 70

N=45 894

Setting entry time at age 70

N=13 875

Multiple imputation for

covariates (10 datasets)

N=48 188

Acute myocardial infarction

Regular meat eaters2 Reference Reference Reference Reference

Low meat eaters2 1.08 (0.88,1.33) 0.93 (0.68,1.27) 1.02 (0.80,1.31) 0.99 (0.81,1.20)

Fish eaters 1.01 (0.77,1.32) 1.12 (0.79,1.59) 0.89 (0.62,1.26) 0.99 (0.77,1.26)

Vegetarians and vegans 0.93 (0.74,1.17) 0.77 (0.56,1.05) 0.98 (0.74,1.30) 0.90 (0.73,1.11)

P-heterogeneity1 0.71 0.21 0.90 0.81



Low meat eaters2 0.90 (0.81,1.01) 0.83 (0.70,0.97) 0.92 (0.81,1.05) 0.88 (0.80,0.98)

Fish eaters 0.84 (0.73,0.97) 0.81 (0.67,0.98) 0.91 (0.76,1.08) 0.86 (0.76,0.98)


P-heterogeneity1 0.0001 0.0002 0.005 <0.0001

Ischaemic stroke


Low meat eaters2 1.23 (0.97,1.56) 1.12 (0.69,1.81) 1.20 (0.92,1.56) 1.17 (0.93,1.48)

Fish eaters 1.12 (0.82,1.51) 1.39 (0.85,2.29) 0.99 (0.68,1.43) 1.12 (0.83,1.50)


P-heterogeneity1 0.31 0.21 0.56 0.31

Haemorrhagic stroke


Low meat eaters2 1.38 (0.97,1.95) 1.04 (0.65,1.66) 1.76 (1.14,2.72) 1.38 (1.00,1.90)

Fish eaters 1.21 (0.79,1.85) 1.19 (0.73,1.93) 1.36 (0.73,2.53) 1.32 (0.91,1.94)


P-heterogeneity1 0.08 0.92 0.007 0.06

Total stroke


Low meat eaters2 1.26 (1.06,1.50) 1.19 (0.88,1.61) 1.23 (1.02,1.50) 1.22 (1.03,1.43)

Fish eaters 1.20 (0.96,1.49) 1.46 (1.06,2.00) 1.16 (0.90,1.51) 1.28 (1.05,1.56)


P-heterogeneity1 0.02 0.10 0.13 0.009



Low meat eaters2 0.97 (0.88,1.06) 0.88 (0.77,1.02) 0.98 (0.88,1.10) 0.94 (0.86,1.03)

Fish eaters 0.89 (0.79,1.01) 0.92 (0.79,1.09) 0.93 (0.80,1.09) 0.93 (0.83,1.04)


P-heterogeneity1 0.03 0.03 0.10 0.03

1 Total N for each sensitivity analysis was based on N in the analyses for ischaemic heart disease or stroke if numbers varied. The total analyses time for the composite outcome of

ischaemic heart disease and stroke in each sensitivity analyses is as follows: excluding first 5 years of follow-up=595 809 person-years; censoring at age 70=727 797 person-years;

limiting to participants above age 70=118 690 person-years; multiple imputation=833 604 person-years. P-heterogeneity represents significance of heterogeneity in risk between diet

groups based on likelihood ratio tests. 2 Regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low meat eaters were defined as participants with total meat consumption <50 g per day.

Page 32 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary table 6: Risk of ischaemic heart disease by subgroups in different diet groups in EPIC-Oxford.

Hazard ratios (95% confidence intervals)1 Test of difference

between subgroups4 Subgroups of variable

Sex Men Women

N cases 1176 1644

Regular meat eaters2 Reference Reference

Low meat eaters2 0.73 (0.62,0.87) 0.98 (0.87,1.11)

Fish eaters 0.75 (0.60,0.93) 0.94 (0.80,1.10) χ2=9.57

Vegetarians and vegans 0.69 (0.59,0.81) 0.81 (0.70,0.93) p=0.023

P-heterogeneity1 p<0.0001 p=0.026

Age at recruitment <60 years ≥60 years

N cases 1367 1453


Low meat eaters2 0.83 (0.71,0.97) 0.93 (0.81,1.06)

Fish eaters 0.77 (0.64,0.92) 0.96 (0.80,1.15) χ2=5.09


P-heterogeneity1 p<0.0001 p=0.15

Body mass index <25 kg/m2 ≥25 kg/m

2

N cases 1454 1366


Low meat eaters2 0.96 (0.83,1.10) 0.91 (0.78,1.05)

Fish eaters 0.96 (0.81,1.13) 0.91 (0.74,1.12) χ2=0.75


P-heterogeneity1 p=0.15 p=0.044

Smoking status Never Former Current

N cases 1419 1020 366

Regular meat eaters2 Reference Reference Reference

Low meat eaters2 0.90 (0.78,1.03) 0.93 (0.79,1.10) 0.69 (0.51,0.94)

Fish eaters 0.82 (0.68,0.98) 0.95 (0.78,1.16) 0.80 (0.56,1.14) χ2=4.59

Vegetarians and vegans 0.76 (0.65,0.87) 0.80 (0.67,0.95) 0.66 (0.48,0.92) p=0.60

P-heterogeneity1 p=0.001 p=0.078 p=0.016

Presence of risk factors3 No Yes

N cases 1833 965


Low meat eaters2 0.88 (0.77,1.00) 0.95 (0.81,1.13)

Fish eaters 0.85 (0.73,1.00) 1.04 (0.83,1.29) χ2=4.12

Vegetarians and vegans 0.77 (0.67,0.87) 0.93 (0.77,1.13) P=0.25


1 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), education (no qualifications, age 16, age 18, degree, unknown), Townsend deprivation

index (quartiles, unknown), smoking (never, former, light, heavy, unknown), alcohol consumption(<1g, 1-7g, 8-15g, 16+

g/day), physical activity (inactive, low activity, moderately active, very active, unknown), and oral contraceptive and

hormone replacement therapy use in women. P-heterogeneity represents significance of heterogeneity in risk between diet

groups based on likelihood ratio tests. 2 Regular meat eaters were defined as participants with total meat consumption ≥50g per day, and low meat

eaters were defined as participants with total meat consumption <50g per day. 3 Presence of one or more of self-reported prior high blood pressure, high blood cholesterol, or diabetes.

4 Test of interaction by sex, age at recruitment, body mass index, smoking, and presence of risk factors were performed by

adding appropriate interaction terms to the Cox models, and testing for statistical significance of interaction across strata

using likelihood ratio tests.

Page 33 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary table 7: Risk of stroke by subgroups in different diet groups in EPIC-Oxford.



Sex Men Women

N cases 304 768


Low meat eaters2 1.30 (0.96,1.76) 1.18 (0.98,1.43)

Fish eaters 1.31 (0.88,1.94) 1.26 (1.01,1.58) χ2=2.84




N cases 405 667


Low meat eaters2 1.02 (0.76,1.36) 1.32 (1.08,1.60)

Fish eaters 1.26 (0.94,1.70) 1.26 (0.97,1.64) χ2=3.00




2

N cases 659 413


Low meat eaters2 1.13 (0.92,1.40) 1.42 (1.09,1.84)

Fish eaters 1.31 (1.03,1.66) 1.29 (0.89,1.87) χ2=3.31




N cases 605 364 97


Low meat eaters2 1.41 (1.15,1.74) 0.95 (0.71,1.27) 0.93 (0.52,1.67)

Fish eaters 1.39 (1.07,1.80) 1.06 (0.76,1.48) 1.35 (0.73,2.51) χ2=8.12

Vegetarians and vegans 1.26 (1.01,1.57) 1.29 (0.99,1.68) 1.40 (0.83,2.37) P=0.23



N cases 740 322


Low meat eaters2 1.12 (0.91,1.37) 1.45 (1.10,1.91)

Fish eaters 1.32 (1.05,1.66) 1.24 (0.84,1.83) χ2=5.20








groups based on likelihood ratio tests. 2 Regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low

meat eaters were defined as participants with total meat consumption <50 g per day. 3 Presence of one or more of self-reported prior high blood pressure, high blood cholesterol, or diabetes.




Page 34 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


Supplementary table 8: Risk of IHD or stroke by subgroups in different diet groups in EPIC-Oxford.



Sex Men Women

N cases 1401 2255


Low meat eaters2 0.83 (0.71,0.96) 1.01 (0.91,1.13)

Fish eaters 0.84 (0.69,1.02) 0.98 (0.86,1.12) χ2=7.74




N cases 1719 1937


Low meat eaters2 0.86 (0.75,0.98) 1.01 (0.90,1.14)

Fish eaters 0.85 (0.73,1.00) 1.00 (0.85,1.16) χ2=8.05




2

N cases 1986 1670


Low meat eaters2 0.98 (0.87,1.11) 1.00 (0.87,1.14)

Fish eaters 1.01 (0.88,1.16) 0.96 (0.80,1.17) χ2=0.25




N cases 1897 1294 445


Low meat eaters2 1.01 (0.90,1.14) 0.93 (0.80,1.08) 0.69 (0.52,0.91)

Fish eaters 0.93 (0.80,1.09) 0.92 (0.77,1.10) 0.92 (0.68,1.26) χ2=7.20

Vegetarians and vegans 0.86 (0.76,0.98) 0.89 (0.77,1.04) 0.77 (0.58,1.02) P=0.30



N cases 2434 1192


Low meat eaters2 0.92 (0.82,1.03) 1.04 (0.89,1.20)

Fish eaters 0.93 (0.82,1.07) 1.05 (0.86,1.28) χ2=5.16








groups based on likelihood ratio tests. 2 Regular meat eaters were defined as participants with total meat consumption ≥50 g per day, and low

meat eaters were defined as participants with total meat consumption <50 g per day. 3 Presence of one or more of self-reported prior high blood pressure, high blood cholesterol, or diabetes.




Page 35 of 35


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Documents

Confidential: For Review Only - BMJ · Confidential: For Review Only 4 40 Introduction 41 Vegetarian and vegan diets have become increasingly popular in recent years partly due to