- Email: [email protected]
Relation between plasma ascorbic acid and mortality in men and women in EPIC-Norfolk prospective study: a prospective population study
ARTICLES
Articles
Relation between plasma ascorbic acid and mortality in men and women in EPIC-Norfolk prospective study: a prospective population study Kay-Tee Khaw, Sheila Bingham, Ailsa Welch, Robert Luben, Nicholas Wareham, Suzy Oakes, Nicholas Day
Summary Background Ascorbic acid (vitamin C) might be protective for several chronic diseases. However, findings from prospective studies that relate ascorbic acid to cardiovascular disease or cancer are not consistent. We aimed to assess the relation between plasma ascorbic acid and subsequent mortality due to all causes, cardiovascular disease, ischaemic heart disease, and cancer. Methods We prospectively examined for 4 years the relation between plasma ascorbic acid concentrations and mortality due to all causes, and to cardiovascular disease, ischaemic heart disease, and cancer in 19 496 men and women aged 45–79 years. We recruited individuals by post using age-sex registers of general practices. Participants completed a health and lifestyle questionnaire and were examined at a clinic visit. They were followed-up for causes of death for about 4 years. Individuals were divided into sex-specific quintiles of plasma ascorbic acid. We used the Cox proportional hazard model to determine the effect of ascorbic acid and other risk factors on mortality. Findings Plasma ascorbic acid concentration was inversely related to mortality from all-causes, and from cardiovascular disease, and ischaemic heart disease in men and women. Risk of mortality in the top ascorbic acid quintile was about half the risk in the lowest quintile (p<0·0001). The relation with mortality was continuous through the whole distribution of ascorbic acid concentrations. 20 mol/L rise in plasma ascorbic acid concentration, equivalent to about 50 g per day increase in fruit and vegetable intake, was associated with about a 20% reduction in risk of all-cause mortality (p<0·0001), independent of age, systolic blood pressure, blood cholesterol, cigarette smoking habit, diabetes, and supplement use. Ascorbic acid was inversely related to cancer mortality in men but not women. Interpretation Small increases in fruit and vegetable intake of about one serving daily has encouraging prospects for possible prevention of disease. Lancet 2001; 357: 657–63
Department of Public Health and Primary Care, Institute of Public Health, University of Cambridge School of Clinical Medicine, Cambridge, UK (Prof K-T Khaw FRCP, A Welch BSc, R Luben BSc, N Wareham MRCP, S Oakes, N Day PhD); and MRC Dunn Human Nutrition Unit, Cambridge (S Bingham PhD) Correspondence to: Prof Kay-Tee Khaw, Clinical Gerontology Unit, Box 251, University of Cambridge School of Clinical Medicine, Addenbrooke’s Hospital, Cambridge CB2 2QQ, UK (e-mail: [email protected])
THE LANCET • Vol 357 • March 3, 2001
Introduction The role of antioxidants in chronic diseases such as cardiovascular disease and cancer is controversial. Evidence from prospective studies suggests that a high dietary intake of antioxidants is protective for cardiovascular disease and cancer, although findings have not been consistent and trial data have not been conclusive.1–7 Ascorbic acid (vitamin C) plays a part in many biological processes, including free radical scavenging, collagen and hormone synthesis, haemostasis, and protection of lipid membranes which might affect chronic disease risk.8–13 Investigators in some prospective studies have shown a significant inverse relation between ascorbic acid and cancer or cardiovascular disease, but the protective concentration and the potential size of the relation have varied between these studies. Results of some studies show only increased risk of mortality or cardiovascular disease at very low concentrations,13 but no effect within the usual population range; conversely, those of others indicate only reduced risk in individuals with high concentrations or those who take supplements.12 Findings from trials on the effect of supplementation have shown no change in mortality, but these trials have been generally small and of short duration.14,15 Most studies have been in men or with data for men and women pooled. Data for women alone are more inconsistent than data on men.13 We present data from a prospective population study examining the relation between plasma ascorbic acid and subsequent mortality due to all causes, cardiovascular disease, ischaemic heart disease, and cancer in men and women.
Methods Participants The individuals in this analysis were part of EPICNorfolk, a prospective population study of 30 466 men and women aged between 45 and 79 years, resident in Norfolk, UK, who completed a baseline questionnaire survey, and of whom 25 663 attended a clinic visit.16 They were recruited from age-sex registers of general practices in Norfolk as part of a nine-country collaborative study (EPIC, European Prospective Investigation into Cancer and Nutrition) designed to investigate dietary and other determinants of cancer. We obtained additional data for the EPIC-Norfolk cohort to enable the assessment of chronic disease determinants. Eligible participants were recruited by post. Because we requested individuals to provide detailed dietary, biological and other health data, and to be followed up over a few years, we had about a 45% response rate, so participants were not a random population sample. Nevertheless, they were closely similar to UK population samples with respect to many characteristics, including anthropometry, blood pressure, and lipids, but with a lower proportion of smokers.16
657
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
Ascorbic acid quintile
p*
1
2
3
4
5
Men Plasma ascorbic acid (mol/L) Age (years) BMI (kg/m2) Systolic blood pressure (mm Hg) Cholesterol (mmol/L) LDL-cholesterol (mmol/L) HDL-cholesterol (mmol/L) Cigarette smoking habit Never Ex smoker Smoker History of diabetes Supplement users
n=1787 20·8 (7·1) 60·1 (9·0) 26·8 (3·6) 139·9 (18·1) 6·03 (1·14) 3·88 (0·98) 1·20 (0·32)
n=1791 38·1 (3·5) 59·4 (8·9) 27·0 (3·3) 138·7 (17·7) 6·08 (1·06) 3·94 (0·94) 1·22 (0·32)
n=1761 48·1 (2·6) 58·9 (8·6) 26·6 (3·1) 137·1 (17·0) 6·08 (1·10) 3·94 (0·96) 1·22 (0·32)
n=1727 56·8 (2·6) 58·6 (8·7) 26·3 (3·0) 136·5 (17·1) 6·04 (1·08) 3·95 (0·98) 1·25 (0·33)
n=1794 72·6 (11·5) 58·8 (8·7) 25·7 (2·9) 135·1 (17·0) 6·00 (1·06) 3·93 (0·94) 1·28 (0·36)
Women Plasma ascorbic acid (mol/L) Age (years) BMI (kg/m2) Systolic blood pressure (mm Hg) Cholesterol (mmol/L) LDL-cholesterol (mmol/L) HDL-cholesterol (mmol/L) Cigarette smoking habit Never Ex-smoker Smoker History of diabetes Supplement users
n=2099 30·3 (10·1) 59·8 (9·1) 27·2 (4·9) 136·9 (18·7) 6·46 (1·22) 4·13 (1·10) 1·48 (0·40)
n=2046 49·5 (3·1) 58·8 (9·0) 26·7 (4·3) 134·5 (18·8) 6·32 (1·18) 4·02 (1·04) 1·54 (0·41)
n=2297 59·1 (2·6) 58·0 (8·6) 26·1 (4·1) 133·5 (18·6) 6·34 (1·18) 4·04 (1·08) 1·58 (0·41)
n=2158 67·8 (2·6) 58·6 (8·6) 25·7 (3·8) 132·6 (18·2) 6·33 (1·22) 4·04 (1·12) 1·60 (0·41)
n=2036 85·1 (13·7) 58·6 (8·5) 25·1 (3·5) 131·8 (18·4) 6·23 (1·18) 3·93 (1·05) 1·64 (0·44)
<0·001 <0·001 <0·001 <0·001 <0·001 <0·001 <0·001
1025 (49%) 604 (29%) 447 (21%) 44 (2%) 754 (36%)
1170 (57%) 645 (31%) 220 (11%) 38 (2%) 921 (45%)
1340 (58%) 734 (32%) 198 (9%) 33 (1%) 1126 (49%)
1271 (59%) 704 (33%) 171 (8·0%) 15 (1%) 1191 (55%)
1201 (59%) 673 (33%) 142 (7%) 15 (1%) 1259 (62%)
<0·001 ·· ·· <0·001 <0·001
426 (24%) 889 (50%) 459 (26%) 67 (4%) 462 (26%)
581 (33%) 996 (56%) 203 (11%) 68 (4%) 558 (31%)
621 (35%) 980 (56%) 148 (8·5%) 52 (3·0%) 610 (35%)
694 (40%) 911 (53%) 107 (6·3%) 32 (2%) 666 (38%)
664 (37%) 985 (55%) 136 (8%) 25 (1%) 849 (47%)
<0·001 <0·001 <0·001 <0·001 0·220 0·220 <0·001 <0·001 ·· ·· <0·001 <0·001
Values are mean (SD) unless otherwise indicated. *p value for linearity.
Table 1: Distribution of variables by sex-specific quintiles of ascorbic acid in men and women
Study design At the baseline survey between 1993 and 1997 participants completed a detailed health and lifestyle questionnaire. They were asked about their medical history with the question “Has a doctor ever told you that you have any of the following?”, after which disorders including diabetes, heart attack, stroke, and cancer were listed. Smoking history was obtained from yes or no responses to questions “Have you ever smoked as much as one cigarette a day for as long as a year?” and “Do you smoke cigarettes now?”. Individuals were classified as supplement takers if they answered yes to the question “Have you taken any vitamins, minerals, or other food supplements regularly during the past year (such as vitamin C, vitamin D, iron, calcium, fish oils, primrose oil, beta-carotene &c)?” Trained nurses examined individuals at a clinic visit. Height and weight were measured and body mass index was estimated as weight (kg) divided by height (m2). Blood pressure was measured with an Accutorr noninvasive blood pressure monitor after the participant had been seated for 5 min. We used the mean of two measurements for analysis. Blood samples were taken by venepuncture into plain and citrate bottles. After overnight storage in a dark box at 4–7°C sample bottles were spun in a centrifuge at 2100 g for 15 min at 4°C. About a year after the start of the study, when funding became available, extra blood samples from participants were taken for ascorbic-acid assays. Plasma ascorbic acid was measured from blood taken into citrate bottles and plasma was stabilised in a standardised volume of metaphosphoric acid stored at –70°C. We estimated plasma ascorbic acid concentration with a fluorometric assay within 1 week of sampling.17 The coefficient of variation was 5·6% at the lower end of the range (mean 33·2 mol/L) and 4·6% at the upper end (102·3 mol/L). Other blood samples for assay were stored at 4°C and assayed at the department of clinical biochemistry, University of Cambridge, Cambridge, UK, within 1 week after
658
the sample was taken. We measured serum total cholesterol, HDL cholesterol, and triglyceride with the RA 1000 (Bayer Diagnostics, Basingstoke, UK), and calculated LDL cholesterol values with the Friedewald formula.18 At the clinic visit, we asked all individuals to complete a 7-day diet diary, and a trained nurse gave them detailed instruction. The diaries were posted back to the coordinating centre at the University of Cambridge. This diary method has been validated with weighed food records, 24 h urine collections and blood biomarkers over a year.19 Because analysis of diaries is labour intensive we coded and analysed with a specially developed programme a sample of only 2000 diaries for average daily nutrient and food intake. Statistical analysis All individuals have been flagged for death certification at the UK Office of National Statistics, with vital status ascertained for the whole cohort. We present results for mortality follow-up to December, 1999, which was about 4 years (range 2–6 years). Death certificates for all decedents were coded by trained nosologists according to the ICD 9th revision. Death was due to cardiovascular disease if the underlying cause on the death certificate was ICD 400–438, to ischaemic heart disease if it was ICD 410–414, and to cancer if it was ICD 140–239. 21 552 men and women aged 45–79 years completed the health and lifestyle questionnaire, attended the health examination, and had blood taken for ascorbic acid measurement. We excluded 1021 men and 1035 women who reported a history of heart attack, stroke, or any cancer at the baseline clinic visit, which left 8860 men and 10 636 women for analysis. Because women had higher ascorbic acid concentrations than men, we divided the individuals into sex-specific quintiles of plasma ascorbic acid. We examined risk factors in ascorbic acid concentration categories, and calculated age-adjusted death rates for all causes, cardiovascular disease, ischaemic heart disease, and cancer in every
THE LANCET • Vol 357 • March 3, 2001
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
Mortality
Ascorbic acid quintile
p*
1
2
3
4
5
Men
n=1787
n=1791
n=1761
n=1727
n=1794
All cause (n=309) Rate/100 (events) Relative risk (rate ratio) (95% CI)
5·12 (100) 1·00
4·08 (76) 0·80 (0·60–1·23)
3·03 (51) 0·59 (0·41–0·83)
2·41 (39) 0·47 (0·31–0·68)
2·44 (43) 0·48 (0·33–0·70)
<0·0001
Cardiovascular disease (n=123) Rate/100 (events) Relative risk (rate ratio) (95% CI)
2·14 (42) 1·00
1·93 (36) 0·90 (0·59–1·48)
1·43 (24) 0·67 (0·41–1·15)
0·62 (10) 0·29 (0·15–0·59)
0·62 (11) 0·29 (0·15–0·59)
<0·0001
Ischaemic heart disease (n=96) Rate/100 (events) Relative risk (rate ratio) (95% CI)
1·41 (27) 1·00
1·67 (31) 1·18 (0·72–2·08)
1·31 (22) 0·92 (0·54–1·72)
0·49 (8) 0·35 (0·16–0·81)
0·45 (8) 0·32 (0·15–0·75)
<0·0001
Cancer (n=116) Rate/100 (events) Relative risk (rate ratio) (95% CI)
1·96 (38) 1·00
1·45 (27) 0·74 (0·48–1·29)
1·00 (17) 0·51 (0·29–0·94)
1·12 (18) 0·57 (0·33–1·03)
0·92 (16) 0·47 (0·27–0·88)
<0·001
Women
n=2099
n=2046
n=2297
n=2158
n=2036
All cause (n=187) Rate/100 (events) Relative risk (rate ratio) (95% CI)
2·58 (60) 1·00
1·70 (36) 0·66 (0·44–1·02)
1·54 (33) 0·60 (0·40–0·94)
1·54 (32) 0·60 (0·38–0·92)
1·29 (26) 0·50 (0·32–0·81)
<0·0001
Cardiovascular disease (n=57) Rate/100 (events) Relative risk (rate ratio) (95% CI)
0·95 (22) 1·00
0·39 (8) 0·41 (0·18–0·93)
0·34 (7) 0·36 (0·16–0·87)
0·57 (12) 0·60 (0·31–1·29)
0·39 (8) 0·41 (0·20–1·00)
0·02
Ischaemic heart disease (n=27) Rate/100 (events Relative risk (rate ratio) (95% CI)
0·60 (14) 1·00
0·14 (3) 0·23 (0·07–0·86)
0·24 (5) 0·39 (0·15–1·19)
0·19 (4) 0·32 (0·11–1·03)
0·04 (1) 0·07 (0·0-1–0·67)
<0·001
Cancer (n=84) Rate/100 (events) Relative risk (rate ratio) (95% CI)
1·05 (24) 1·00
0·79 (17) 0·76 (0·42–1·48)
0·64 (14) 0·61 (0·33–1·23)
0·67 (14) 0·64 (0·33–1·25)
0·76 (15) 0·73 (0·38–1·40)
0·10
*p values are for 2 linear trend with 1 degree of freedom.
Table 2: Age-adjusted rates for all-cause, cardiovascular disease, and ischaemic heart disease mortality by sex-specific ascorbic acid quintile in men and women
category using 2 test for linear trend to assess statistical significance.20 We used the Cox proportional hazards model to determine the independent contribution of ascorbic acid and other risk factors for mortality due to all causes, cardiovascular disease, ischaemic heart disease, and cancer in men and women separately,21 and then with sexes combined, adjusted for sex. We also combined data for men and women and used a quadratic model with sex-specific quintiles for all-cause mortality, adjusting for sex, age, and risk factors to see whether this adjustment improved the fit of the relative risk estimates. For the 2000 sample of diaries analysed, we examined the main dietary components that correlated with plasma ascorbic acid concentration and the difference in average daily intake of the main foods that correlated by quintile of ascorbic acid.
Results Table 1 shows characteristics of the participants according to sex-specific quintile of plasma ascorbic-acid concentration. Ascorbic acid was inversely related to cardiovascular risk factors, including systolic blood pressure and body mass index, and positively to HDL cholesterol. The proportion of individuals who smoked and had a history of diabetes fell as the quintile of ascorbic acid increased. By contrast, the proportion of those who took any supplements rose as the quintile increased. Mean plasma ascorbic acid concentrations varied by about 10 mol/L between quintiles. Exclusion of people reporting use of vitamin supplements did not substantially change these means (data not shown).
THE LANCET • Vol 357 • March 3, 2001
Table 2 shows all-cause mortality rates, and those for cardiovascular disease, ischaemic heart disease, and cancer, by quintile of plasma ascorbic acid. There was a graded relation between quintiles and total mortality in both men and women, with lowest mortality rates in the highest quintile. All these relations were significant. Cardiovascular disease, ischaemic heart disease, and cancer mortality in men also fell significantly as ascorbic acid quintile increased. The number of events was smaller and trends were less consistent for these endpoints for women than for men. Table 3 shows the independent multivariate relation between plasma ascorbic acid and all-cause, cardiovascular disease, and ischaemic heart disease mortality, with the Cox proportional hazards model after adjustment for age alone, then adjustment for age and systolic blood pressure, serum cholesterol, body mass index (as continuous variables), cigarette smoking habit, diabetes, and supplement use. Replacement of total cholesterol by LDL or HDL cholesterol in the model made no difference to results (data not shown). A 20 mol/L increase in plasma ascorbic acid was associated with about a 20% decline in death due to all cause, cardiovascular disease, and ischaemic heart disease and this relation was also seen after adjustment for risk factors. The Cox models were also done after exclusion of smokers, and then supplement users. The relative risk for cancer mortality with plasma ascorbic acid in women was not significant in the Cox model. Results were closely similar after the exclusion of all people who reported a history of diabetes at baseline
659
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
Mortality
Age-adjusted hazard ratio (95% CI) (relative risk estimate)
p
Age-adjusted and risk factor2-adjusted† hazard ratio* (95% CI) (relative risk estimate)
p
Men All-cause All men (n=309) Excluding smokers (n=263) Excluding supplement users (n=195)
0·76 (0·68–0·86) 0·75 (0·65–0·86) 0·75 (0·65–0·88)
0·0001 0·0001 0·0001
0·77 (0·67–0·87) 0·74 (0·65–0·85) 0·76 (0·65–0·89)
0·0001 0·0001 0·001
Cardiovascular disease All men (n=123) Excluding smokers (n=100) Excluding supplement users (n=81)
0·62 (0·51–0·76) 0·60 (0·49–0·76) 0·62 (0·49–0·79)
0·0001 0·0001 0·0001
0·64 (0·51–0·78) 0·62 (0·49–0·78) 0·64 (0·49–0·83)
0·0001 0·0001 0·001
Ischaemic heart disease All men (n=96) Excluding smokers (n=75) Excluding supplement users (n=61)
0·66 (0·53–0·82) 0·67 (0·52–0·86) 0·65 (0·49–0·86)
0·0001 0·002 0·002
0·63 (0·42–0·94) 0·68 (0·52–0·88) 0·69 (0·51–0·92)
0·022 0·004 0·011
Cancer All men (n=116) Excluding smokers (n=103) Excluding supplement users (n=71)
0·79 (0·65–0·96) 0·75 (0·61–0·93) 0·77 (0·60–1·00)
0·019 0·009 0·05
0·76 (0·62–0·94) 0·74 (0·59–0·92) 0·76 (0·59–0·98)
0·011 0·007 0·037
Women All cause All women (n=187) Excluding smokers (n=158) Excluding supplement users (n=98)
0·80 (0·70–0·93) 0·83 (0·71–0·97) 0·73 (0·60–0·89)
0·003 0·019 0·002
0·85 (0·73–0·99) 0·84 (0·71–0·99) 0·81 (0·66–0·99)
0·033 0·037 0·049
Cardiovascular disease All women (n=57) Excluding smokers (n=48) Excluding supplement users (n=30)
0·77 (0·60–0·99) 0·77 (0·58–1·02) 0·73 (0·51–1·05)
0·048 0·070 0·089
0·81 (0·62–1·06) 0·78 (0·59–1·05) 0·78 (0·54–1·13)
0·13 0·11 0·19
Ischaemic heart disease All women (n=27) Excluding smokers (n=21) Excluding supplement users (n=15)
0·55 (0·40–0·80) 0·51 (0·33–0·78) 0·63 (0·38–1·04)
0·002 0·002 0·07
0·56 (0·36–0·87) 0·60 (0·43–0·84) 0·74 (0·44–1·26)
0·009 0·003 0·27
Cancer All women (n=84) Excluding smokers (n=71) Excluding supplement users (n=46)
0·91 (0·74–1·12) 0·94 (0·75–1·19) 0·82 (0·61–1·10)
0·38 0·61 0·18
0·97 (0·78–1·22) 0·96 (0·75–1·22) 0·92 (0·69–1·23)
0·81 0·73 0·58
*Hazard ratios (relative risks) are per increase of 20 mol/L plasma ascorbic acid concentration; †Risk factors entered are: systolic blood pressure, cholesterol, body mass index (continuous variables); cigarette smoking habit, diabetes mellitus, any supplement use.
Table 3: Cox multivariate regression for all cause, cardiovascular, and ischaemic heart disease mortality with ascorbic acid in 8860 men and 10 636 women
(data not shown). In this cohort, supplement use was not associated with mortality in the Cox regression model, after adjustment for age and other risk factors. The relative risk for all-cause mortality in supplement users was 1·00 (95% CI 0·80–1·23, p=0·97) for men, and 1·23 (0·94–1·62, p=0·14) for women with plasma All Men Women
1·2
1·0
1·0
0·8
0·8
0·6
0·6
0·4
0·4
0·2
0·2
0
All Men Women
1·2
Relative risk
Relative risk
ascorbic acid included in the model, and 0·93 (0·75–1·15, p=0·50) for men and 1·17 (0·89–1·53, p=0·27) for women when this variable was excluded. The Cox model was also run with the inclusion of alcohol intake, with a linear and quadratic model for alcohol; ascorbic acid and mortality correlations were
0 1
2 3 4 Ascorbic acid quintile
5
Figure 1: Relative risk of total mortality by quintile of plasma ascorbic acid
2 3 4 5 Ascorbic acid quintile Figure 2: Relative risk for cardiovascular disease mortality by quintile of plasma ascorbic acid
Age-adjusted and sex adjusted Cox regression model for relative risk, including a quadratic term for ascorbic-acid.
Age-adjusted and sex-adjusted Cox regression model for relative risk, including a quadratic term for ascorbic acid.
660
1
THE LANCET • Vol 357 • March 3, 2001
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
means were in the middle three quintiles with greatest differences in means between the lowest and highest quintile. The difference in mean intake between quintiles was about 20–50 g per day of fruit and 10–20 g per day of vegetables. Table 6 shows hazard ratios adjusted for age, sex, and risk factors for plasma ascorbic acid and mortality due to all causes, cardiovascular disease, ischaemic heart disease, and cancer.
Correlation coefficient r
Dietary ascorbic acid Fruit Vegetables All fruit and vegetables Potatoes Legumes Alcohol Cakes Bread Other cereals Cheese Dairy Milk Eggs Fish Meat Meat products
Men (n=800)
Women (n=1096)
0·40 0·36 0·23 0·38 ⫺0·07 ⫺0·01 0·06 0·00 0·01 ⫺0·03 0·07 0·07 ⫺0·02 0·12 ⫺0·11 0·04 ⫺0·04
0·41 0·31 0·23 0·35 ⫺0·09 ⫺0·03 0·09 ⫺0·06 ⫺0·06 0·05 0·04 0·13 ⫺0·05 ⫺0·05 0·07 ⫺0·03 ⫺0·07
Discussion
Correlation coefficients >0·20, p<0·0001.
Table 4: Pearson correlations of plasma ascorbic acid with food intake from 7-day diaries at baseline survey in EPICNorfolk 1993–97
unchanged. When men and women were combined in the same model, adjusted for sex, the relative mortality risks for a 20 mol/L increase in plasma ascorbic acid were: 0·80 (0·73–0·88, p<0·0001) for total mortality; and mortality due to cardiovascular disease 0·70 (0·60–0·82, p<0·0001), for ischaemic heart disease 0·68 (0·56–0·83, p<0·0001), and for cancer 0·85 (0·74–0·99, p=0·04). The Cox model was also run with a quadratic model to see if this analysis produced a better fit. Data for men and women were combined, and adjusted for sex in the model; the quintile category rather than the precise concentration of ascorbic acid was used in this model. Figures 1 and 2 show the age-adjusted and sex-adjusted relative risks of mortality due to all causes and cardiovascular disease mortality by quintile of ascorbic acid, with men and women combined, and with sexspecific data points. For all causes, the quadratic model showed a better fit with the data than with a linear model, with a steeper decline in risk between lower quintiles and levelling in higher quintiles; for cardiovascular mortality, the relation seemed more linear. Table 4 shows correlations of ascorbic acid with food intake estimated from 7-day diaries in 1896 men and women for whom coded and analysed diary data were available. Only fruit and vegetable intake correlated consistently and more substantially with ascorbic acid (correlation coefficient >0·20) than with other foods. Table 5 shows average daily intake of fruit and vegetables estimated from diaries by quintile of plasma ascorbic acid. As expected, the smallest differences in
In this population of men and women aged between 45 and 79 years, increasing plasma ascorbic acid concentration was strongly and independently associated with reduction in risk of mortality from all causes, cardiovascular disease, and ischaemic heart disease, with a dose-response relation across the whole population distribution. Although plasma ascorbic acid was measured once only for each individual, the relation with mortality was strong. The assumption is that one value would indicate the normal or habitual concentration. However, the large within-individual variation of plasma ascorbic acid might result in large random measurement error such that the true underlying relation could be stronger than that shown in our study. We have previously reported in this cohort an inverse relation of plasma ascorbic acid with cardiovascular risk factors, including blood pressure and lipids.22,23 Plasma ascorbic acid was also lower in smokers and diabetics.24 However, the inverse relation of this substance and mortality was consistent and independent of other cardiovascular risk factors, including blood pressure, serum cholesterol, body mass index, cigarette smoking, and diabetes. This inverse relation was also consistent after exclusion of smokers and then after supplement users. Ascorbic acid was also inversely related to cancer mortality in men but not in women. The most common types of cancer deaths in men were lung, colorectal and prostate cancer and in women were breast and colorectal cancer. The number of cancer deaths were too few for site-specific analysis but the relation might be specific to cancer types that are more common in men. There are several possible explanations for the inverse relation between plasma ascorbic acid and mortality. Pre-existing disease or cigarette smoking could lower ascorbic-acid concentrations. In our cohort, individuals with a history of cardiovascular disease or diabetes and cigarette smokers, did have low ascorbic acid concentrations. However, our analyses excluded those with known heart disease, stroke, or cancer, and the relation of plasma ascorbic acid with mortality did not change after diabetics and cigarette smokers were excluded. The proportion of cigarette smokers was low in this cohort compared with national data. This difference might indicate both the low proportion of
Plasma ascorbic acid quintile 1
2
3
4
5
Men Plasma ascorbic acid (mol/L) Dietary ascorbic acid (mg per day) Fruit intake (g per day) Vegetable intake (g per day)
n=161 20·9 (7·1) 51 (26) 75 (84) 70 (48)
n=169 38·3 (3·4) 77 (40) 120 (89) 96 (67)
n=157 48·1 (2·6) 83 (42) 147 (109) 96 (66)
n=156 57·6 (3·2) 92 (44) 164 (123) 102 (65)
n=157 73·3 (9·6) 109 (61) 192 (152) 117 (73)
Women Plasma ascorbic acid (mol/L) Dietary ascorbic acid (mg per day) Fruit intake (g per day) Vegetable intake (g per day)
n=220 29·2 (10·3) 57 (30) 108 (96) 77 (56)
n=225 49·3 (3·7) 78 (41) 155 (102) 93 (63)
n=223 60·1 (2·6) 90 (40) 186 (139) 101 (63)
n=209 68·6 (2·5) 97 (47) 186 (125) 97 (60)
n=219 86·0 (13·4) 113 (60) 227 (148) 118 (99)
Values are mean (SD).
Table 5: Food intake from 7 diaries by sex specific plasma ascorbic acid quintile at baseline survey in 800 men and 1096 women in EPIC-Norfolk 1993–1997
THE LANCET • Vol 357 • March 3, 2001
661
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES Mortality
Hazard ratio* (95% CI) adjusted for age and sex
p
Hazard ratio* (95% CI) adjusted for age, sex, and risk factor†
p
All-cause (n=692) Cardiovascular disease (n=268) Ischaemic heart disease (n=189) Cancer (n=284)
0·78 (0·72–0·86) 0·67 (0·58–0·79) 0·63 (0·52–0·77) 0·84 (0·73–0·98)
0·0001 0·0001 0·0001 0·02
0·80 (0·73–0·88) 0·70 (0·60–0·82) 0·68 (0·56–0·83) 0·85 (0·74–0·99)
0·0001 0·0001 0·0001 0·0400
*Hazard ratios (relative risks) are per increase of 20 mol/L plasma ascorbic acid concentration; †risk factors are systolic blood pressure, cholesterol, body mass index, cigarette smoking habit, diabetes mellitus, any supplement use.
Table 6: Cox multivariate regression for all-cause, cardiovascular disease, and ischaemic heart disease mortality with plasma ascorbic acid in 8860 men and 10 636 women, baseline survey in EPIC-Norfolk 1995–1999
smokers living in East Anglia and the characteristics of the participants. Misreporting of smoking habit is unlikely to be worse in this cohort than in others. Lung cancer standardised mortality rates in this cohort are about half the national rate, which lends support to the likelihood that the low proportion of smokers is not due only to under-reporting. Plasma ascorbic acid might be related to other types of behaviour that protect against atherosclerosis. One possibility is that those with high concentrations of ascorbic acid might be taking supplements. These individuals might also take other supplements, including fish oils, which might protect against cardiovascular disease. Irrespective of the direct effect of supplements, the selection biases in the characteristics of supplement users are well recognised, as is the low mortality of good compliers, even with placebo, in trials. In the EPICNorfolk cohort, a substantial proportion of participants (a third of men, half the women) reported some sort of supplement use. However, in this cohort, supplement use was not associated with mortality. There are other possible confounders such as social class and physical activity of the individuals. Although we requested information on these variables, the data are not available for analysis because of limited resources for coding. The consistency and strength of the ascorbic acid and mortality relation equal to or greater than that seen in other studies for many known factors, including social class and physical activity. This consistent relation suggests that plasma ascorbic acid might give a better indication of the mortality risk associated with these other known factors than their direct assessment. However, some of the recorded social class variations in health could be mediated through dietary differences, including those in ascorbic acid status. Antioxidants might be protective against cardiovascular disease and other age-related chronic diseases, but evidence in man is inconclusive. Most attention has focused on betacarotene and vitamin E; findings from several prospective studies show that high dietary intakes of betacarotene, vitamin E, or both, are inversely associated with subsequent cardiovascular disease mortality. However, betacarotene and vitamin E supplementation do not reduce fatal endpoints.1–7 Ascorbic acid has many biological effects that could plausibly be protective for cardiovascular disease, including scavenging of free radicals protecting against oxidative damage. Findings from studies in guineapigs (who like man cannot synthesise vitamin C) suggest that ascorbic acid deficiency increases atherosclerotic lesions, and that lesion formation is inhibited by high-dose ascorbic acid.25 However, epidemiological evidence for ascorbic acid shows that an association with coronary heart disease is not consistent,9 which could be explained by a true lack of effect, dietary measurement error, a threshold effect, an interaction with other dietary constituents, or too long or too short a duration of follow-up.
662
Measurement error is an issue in dietary exposures. Most studies have estimated ascorbic acid intake from dietary assessments—notably the food frequency questionnaire. The assessment of dietary ascorbic acid intake is especially difficult because not only is ascorbic acid present in many foods, but also the amount varies according to processing, storage, and cooking. The subjective nature of dietary assessment and other drawbacks in the estimation of ascorbic acid intake has led to the use of biological indicators. Plasma ascorbic acid measurement is thought to represent dietary intake in the preceding few weeks and might overcome some of the issues involved with dietary assessment.26 However, consistent measurement of plasma ascorbic acid is especially difficult because this substance is unstable in blood and deteriorates rapidly unless it is stabilised by the addition of substances, such as metaphosphoric acid, and stored at very low temperatures. Measurement errors might also account for the absence of consistency in studies, in which ascorbic acid has been measured. Even though intake and plasma concentration are closely related, other factors such as smoking habit or preexisting disease could account for differences in plasma ascorbic acid concentrations that are not explained by intake. These factors might confound results. Nevertheless, in many studies investigators, including early validation studies done in this cohort, have shown that plasma ascorbic acid is strongly related to dietary ascorbic acid intake.19 However, we cannot exclude residual confounding and this substance might be an indicator of other protective factors. Nevertheless, whatever these factors might be, the effect must be substantial and not explained by any of the other biological or lifestyle measures available to us. The most likely alternative explanation is that the association is with another nutrient or nutrients contained within foods that are a ready source of ascorbic acid, such as fruit and vegetables. Our findings contrast with an analysis of a 12-year follow-up of the second National Health and Nutrition Examination (NHANES II) in US adults.13 Investigators noted that men in the lowest quartile had 57% higher risk of all-cause mortality and 62% higher risk of dying from cancer than men in the highest quartile. There was no increase in risk of cardiovascular disease mortality in any quartile in men13 and no association between serum ascorbate quartile and mortality in women. NHANES II was a multicentre study but measurement variation is unlikely to account for the differences from our findings because the range and mean concentrations of plasma ascorbic acid were closely similar in the two studies. Plasma ascorbic acid might indicate different dietary patterns of fruit and vegetables which could have different biological effects. Alternatively, the longer follow-up time in NHANES II than in our study might have caused greater misclassification of individuals who could have changed ascorbic acid status during followup. Endpoints may have been misclassified because of underestimation of mortality in the NHANES study
THE LANCET • Vol 357 • March 3, 2001
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
compared with our study in which complete mortality was ascertained. Our findings might have clinical and public health implications. The inverse relation was seen throughout the normal population range of plasma ascorbic acid in a dose-related way and not only at the extremes of the distribution. There was no evidence that ascorbic acid had an effect only below or above a threshold concentration (eg, deficiency or supplement concentrations respectively). The absence of an effect on mortality in trials of ascorbic acid so far provides no evidence to encourage supplement use.14,15 The main sources of dietary ascorbic acid are plant foods—mainly fruit and vegetables. In this cohort, plasma ascorbic acid was specifically related to fruit and vegetable intake, and an increase in 20 mol/L plasma ascorbic acid concentration was associated with an increase of about 50 g daily or one serving of fruit and vegetables. A change from the second to fourth quintile (eg, 30th–70th percentile), was associated with 30% reduction in mortality risk, which suggests that small and feasible shifts within the normal population intake could have a substantial effect on mortality risk. Plasma ascorbic acid might indicate the intake of such foods that are rich in other nutrients such as potassium, folate, calcium, magnesium, non-starch polysaccharides, isoflavonoids, and others that might confer health benefits.27–30 Whether ascorbic acid supplements are beneficial remains to be seen. However, our findings suggest that an increase in dietary intake of foods rich in ascorbic acid might have benefits for cardiovascular disease and all-cause mortality in men and women and add to the large amount of evidence that lends support to the health benefits of fruit and vegetable intake.28–30 Small and feasible changes within the normal population range of intake could have a large effect. Contributors K-T Khaw, N Day, and S Bingham originated and designed the study. S Oakes coordinated the study and organised data collection including quality control of blood samples and measurement procedures. A Welch and S Bingham analysed the dietary data and did nutritional analyses. R Luben was responsible for data management and computing overall and assisted with analyses. N Wareham introduced the diabetes measures. K-T Khaw did the data analyses and wrote the paper with co-authors.
6
7
8
9 10
11
12
13 14
15
16 17
18
19
20
21 22 23 24
Acknowledgments EPIC-Norfolk is supported by programme grants from Cancer Research Campaign and Medical Research Council with additional support from the Stroke Association, British Heart Foundation, Department of Health, Food Standards Agency, and the Wellcome Trust.
References 1
2 3 4
5
Lonn EM, Yusuf S. Is there a role for antioxidant vitamins in the prevention of cardiovascular diseases? an update on epidemiological and clinical trials data. Can J Cardiol 1997; 13: 957–65. Duthie GG, Bellizzi MC. Effects of antioxidants on vascular health. Br Med Bull 1999; 55: 568–77. Rexrode KM, Manson JE. Antioxidants and coronary heart disease: observational studies. J Cardiovasc Risk 1996; 3: 363–67. Gaziano JM. Randomized trials of dietary antioxidants in cardiovascular disease prevention and treatment. J Cardiovasc Risk 1996; 3: 368–71. Blot WJ, Li-JY, Taylor PR, Guo W, Dawsey SM, Li B. The Linxian
THE LANCET • Vol 357 • March 3, 2001
25 26
27
28
29
30
trials: mortality rates by vitamin-mineral intervention group. Am J Clin Nutr 1995; 62: 1424S–26S. The Alpha-Tocopherol Beta-Carotene Cancer Prevention Study Group The Alpha-Tocopherol. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. N Engl J Med 1994; 330: 1029–35. Yusuf S, Dagenais G, Pogue J, Bosch J, Sleight P. Vitamin E supplementation and cardiovascular events in high-risk patients. The Heart Outcomes Prevention Evaluation Study Investigators. N Engl J Med 2000; 342: 154–60. Jialal I, Vega GL, Grundy SM. Physiologic levels of ascorbate inhibit the oxidative modification of low density lipoprotein. Atherosclerosis 1990; 82: 185–91. Ness AR, Powles JW, Khaw KT. Vitamin C and cardiovascular disease: a systematic review. J Cardiovasc Risk 1996; 3: 513–21. Sahyoun NR, Jacques PF, Russell RM. Carotenoids, vitamins C and E and mortality in an elderly population. Am J Epidemiol 1996; 144: 501–11. Gale CR, Martyn CN, Winter PD, Cooper C. Vitamin C and risk of death from stroke and coronary heart disease in cohort of elderly people. BMJ 1995; 310: 1563–66. Enstrom JE, Kanim LE, Klein MA. Vitamin C intake and mortality among a sample of the United States population. Epidemiology 1992; 3: 194–202. Loria CM, Klag MJ, Caulfield LE, Whelton PK. Vitamin C status and mortality in US adults. Am J Clin Nutr 2000; 72: 139–45. Wilson TS, Datta SB, Murrell JS, Andrews CT. Relation of vitamin C levels to mortality in a geriatric hospital: a study of the effect of vitamin C administration. Age Ageing 1973; 2: 163–71. Hunt C, Chakravorty NK, Annan G. The clinical and biochemical effects of vitamin C supplementation in short stay hospitalized geriatric patients. Int J Vitam Nutr Res 1984 54: 65–74. Day NE, Oakes S, Luben R, et al. EPIC-Norfolk: study design and characteristics of the cohort. Br J Cancer 1999; 80 (suppl 1): 95–103. Vuilleumier J, Keck E. Fluorometric assay of vitamin C in biological materials using a centrifugal analyser with fluorescence attachment. J Micronutrient Analysis 1989; 5: 25–34. Friedewald WT, Levy RI, Frederickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without the use of the preparative ultracentrifuge. Clin Chem 1972; 18: 499–502. Bingham S, Cassidy A, Cole T, et al. Validation of weighed records and other methods of dietary assessment using the 24 hour urine nitrogen technique and other markers. Br J Nutr 1995; 73: 531–50. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959; 22: 719–48. Cox DR. Regression models and life tables. JR Stat Soc B 1972; 34: 187–220. Ness A, Khaw KT, Day NE, Bingham S. Vitamin C and blood pressure. J Hypertens 1996; 14: 503–08. Ness AR, Khaw KT, Bingham S, Day NE. Vitamin C and serum lipids. Eur J Clin Nutr 1996; 50: 724–29. Sargeant L, Wareham N, Bingham S, et al. Vitamin C and glucose tolerance in a EPIC-Norfolk: a population study. Diabetes Care 2000; 22: 726–32. Afridi N, Keaney JF. Animal studies on antioxidants. J Cardiovasc Risk 1996; 3: 358–62. Bates CJ, Thurnham SI, Bingham SA, Margetts BM, Nelson M. Biochemical markers of nutrient intake. In: Margetts BM, Nelson M, eds. Design concepts in nutritional epidemiology. Oxford: Oxford Medical Publications 1991; 192–265. Khaw KT, Barrett-Connor E. Dietary potassium and strokeassociated mortality: a twelve year prospective population study. N Engl J Med 1987; 316: 235–40. Tunstall-Pedoe H, Woodward M, Tavendale R, Brook RA, McCluskey MK. Comparison of the prediction by 27 different factors of coronary heart disease and death in men and women of the Scottish heart health study: cohort study. BMJ 1997; 315: 722–29. Khaw KT, Barrett-Connor E. Dietary fibre and reduced ischaemic heart disease mortality rates in men and women: a 12 year prospective study. Am J Epidemiol 1987; 126: 1093–102. Hertog MG, Feskens EJ, Kromhout D. Antioxidant flavonols and coronary heart disease risk. Lancet 1997; 349: 699.
663
For personal use only. Reproduce with permission from The Lancet Publishing Group.
Articles
Relation between plasma ascorbic acid and mortality in men and women in EPIC-Norfolk prospective study: a prospective population study Kay-Tee Khaw, Sheila Bingham, Ailsa Welch, Robert Luben, Nicholas Wareham, Suzy Oakes, Nicholas Day
Summary Background Ascorbic acid (vitamin C) might be protective for several chronic diseases. However, findings from prospective studies that relate ascorbic acid to cardiovascular disease or cancer are not consistent. We aimed to assess the relation between plasma ascorbic acid and subsequent mortality due to all causes, cardiovascular disease, ischaemic heart disease, and cancer. Methods We prospectively examined for 4 years the relation between plasma ascorbic acid concentrations and mortality due to all causes, and to cardiovascular disease, ischaemic heart disease, and cancer in 19 496 men and women aged 45–79 years. We recruited individuals by post using age-sex registers of general practices. Participants completed a health and lifestyle questionnaire and were examined at a clinic visit. They were followed-up for causes of death for about 4 years. Individuals were divided into sex-specific quintiles of plasma ascorbic acid. We used the Cox proportional hazard model to determine the effect of ascorbic acid and other risk factors on mortality. Findings Plasma ascorbic acid concentration was inversely related to mortality from all-causes, and from cardiovascular disease, and ischaemic heart disease in men and women. Risk of mortality in the top ascorbic acid quintile was about half the risk in the lowest quintile (p<0·0001). The relation with mortality was continuous through the whole distribution of ascorbic acid concentrations. 20 mol/L rise in plasma ascorbic acid concentration, equivalent to about 50 g per day increase in fruit and vegetable intake, was associated with about a 20% reduction in risk of all-cause mortality (p<0·0001), independent of age, systolic blood pressure, blood cholesterol, cigarette smoking habit, diabetes, and supplement use. Ascorbic acid was inversely related to cancer mortality in men but not women. Interpretation Small increases in fruit and vegetable intake of about one serving daily has encouraging prospects for possible prevention of disease. Lancet 2001; 357: 657–63
Department of Public Health and Primary Care, Institute of Public Health, University of Cambridge School of Clinical Medicine, Cambridge, UK (Prof K-T Khaw FRCP, A Welch BSc, R Luben BSc, N Wareham MRCP, S Oakes, N Day PhD); and MRC Dunn Human Nutrition Unit, Cambridge (S Bingham PhD) Correspondence to: Prof Kay-Tee Khaw, Clinical Gerontology Unit, Box 251, University of Cambridge School of Clinical Medicine, Addenbrooke’s Hospital, Cambridge CB2 2QQ, UK (e-mail: [email protected])
THE LANCET • Vol 357 • March 3, 2001
Introduction The role of antioxidants in chronic diseases such as cardiovascular disease and cancer is controversial. Evidence from prospective studies suggests that a high dietary intake of antioxidants is protective for cardiovascular disease and cancer, although findings have not been consistent and trial data have not been conclusive.1–7 Ascorbic acid (vitamin C) plays a part in many biological processes, including free radical scavenging, collagen and hormone synthesis, haemostasis, and protection of lipid membranes which might affect chronic disease risk.8–13 Investigators in some prospective studies have shown a significant inverse relation between ascorbic acid and cancer or cardiovascular disease, but the protective concentration and the potential size of the relation have varied between these studies. Results of some studies show only increased risk of mortality or cardiovascular disease at very low concentrations,13 but no effect within the usual population range; conversely, those of others indicate only reduced risk in individuals with high concentrations or those who take supplements.12 Findings from trials on the effect of supplementation have shown no change in mortality, but these trials have been generally small and of short duration.14,15 Most studies have been in men or with data for men and women pooled. Data for women alone are more inconsistent than data on men.13 We present data from a prospective population study examining the relation between plasma ascorbic acid and subsequent mortality due to all causes, cardiovascular disease, ischaemic heart disease, and cancer in men and women.
Methods Participants The individuals in this analysis were part of EPICNorfolk, a prospective population study of 30 466 men and women aged between 45 and 79 years, resident in Norfolk, UK, who completed a baseline questionnaire survey, and of whom 25 663 attended a clinic visit.16 They were recruited from age-sex registers of general practices in Norfolk as part of a nine-country collaborative study (EPIC, European Prospective Investigation into Cancer and Nutrition) designed to investigate dietary and other determinants of cancer. We obtained additional data for the EPIC-Norfolk cohort to enable the assessment of chronic disease determinants. Eligible participants were recruited by post. Because we requested individuals to provide detailed dietary, biological and other health data, and to be followed up over a few years, we had about a 45% response rate, so participants were not a random population sample. Nevertheless, they were closely similar to UK population samples with respect to many characteristics, including anthropometry, blood pressure, and lipids, but with a lower proportion of smokers.16
657
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
Ascorbic acid quintile
p*
1
2
3
4
5
Men Plasma ascorbic acid (mol/L) Age (years) BMI (kg/m2) Systolic blood pressure (mm Hg) Cholesterol (mmol/L) LDL-cholesterol (mmol/L) HDL-cholesterol (mmol/L) Cigarette smoking habit Never Ex smoker Smoker History of diabetes Supplement users
n=1787 20·8 (7·1) 60·1 (9·0) 26·8 (3·6) 139·9 (18·1) 6·03 (1·14) 3·88 (0·98) 1·20 (0·32)
n=1791 38·1 (3·5) 59·4 (8·9) 27·0 (3·3) 138·7 (17·7) 6·08 (1·06) 3·94 (0·94) 1·22 (0·32)
n=1761 48·1 (2·6) 58·9 (8·6) 26·6 (3·1) 137·1 (17·0) 6·08 (1·10) 3·94 (0·96) 1·22 (0·32)
n=1727 56·8 (2·6) 58·6 (8·7) 26·3 (3·0) 136·5 (17·1) 6·04 (1·08) 3·95 (0·98) 1·25 (0·33)
n=1794 72·6 (11·5) 58·8 (8·7) 25·7 (2·9) 135·1 (17·0) 6·00 (1·06) 3·93 (0·94) 1·28 (0·36)
Women Plasma ascorbic acid (mol/L) Age (years) BMI (kg/m2) Systolic blood pressure (mm Hg) Cholesterol (mmol/L) LDL-cholesterol (mmol/L) HDL-cholesterol (mmol/L) Cigarette smoking habit Never Ex-smoker Smoker History of diabetes Supplement users
n=2099 30·3 (10·1) 59·8 (9·1) 27·2 (4·9) 136·9 (18·7) 6·46 (1·22) 4·13 (1·10) 1·48 (0·40)
n=2046 49·5 (3·1) 58·8 (9·0) 26·7 (4·3) 134·5 (18·8) 6·32 (1·18) 4·02 (1·04) 1·54 (0·41)
n=2297 59·1 (2·6) 58·0 (8·6) 26·1 (4·1) 133·5 (18·6) 6·34 (1·18) 4·04 (1·08) 1·58 (0·41)
n=2158 67·8 (2·6) 58·6 (8·6) 25·7 (3·8) 132·6 (18·2) 6·33 (1·22) 4·04 (1·12) 1·60 (0·41)
n=2036 85·1 (13·7) 58·6 (8·5) 25·1 (3·5) 131·8 (18·4) 6·23 (1·18) 3·93 (1·05) 1·64 (0·44)
<0·001 <0·001 <0·001 <0·001 <0·001 <0·001 <0·001
1025 (49%) 604 (29%) 447 (21%) 44 (2%) 754 (36%)
1170 (57%) 645 (31%) 220 (11%) 38 (2%) 921 (45%)
1340 (58%) 734 (32%) 198 (9%) 33 (1%) 1126 (49%)
1271 (59%) 704 (33%) 171 (8·0%) 15 (1%) 1191 (55%)
1201 (59%) 673 (33%) 142 (7%) 15 (1%) 1259 (62%)
<0·001 ·· ·· <0·001 <0·001
426 (24%) 889 (50%) 459 (26%) 67 (4%) 462 (26%)
581 (33%) 996 (56%) 203 (11%) 68 (4%) 558 (31%)
621 (35%) 980 (56%) 148 (8·5%) 52 (3·0%) 610 (35%)
694 (40%) 911 (53%) 107 (6·3%) 32 (2%) 666 (38%)
664 (37%) 985 (55%) 136 (8%) 25 (1%) 849 (47%)
<0·001 <0·001 <0·001 <0·001 0·220 0·220 <0·001 <0·001 ·· ·· <0·001 <0·001
Values are mean (SD) unless otherwise indicated. *p value for linearity.
Table 1: Distribution of variables by sex-specific quintiles of ascorbic acid in men and women
Study design At the baseline survey between 1993 and 1997 participants completed a detailed health and lifestyle questionnaire. They were asked about their medical history with the question “Has a doctor ever told you that you have any of the following?”, after which disorders including diabetes, heart attack, stroke, and cancer were listed. Smoking history was obtained from yes or no responses to questions “Have you ever smoked as much as one cigarette a day for as long as a year?” and “Do you smoke cigarettes now?”. Individuals were classified as supplement takers if they answered yes to the question “Have you taken any vitamins, minerals, or other food supplements regularly during the past year (such as vitamin C, vitamin D, iron, calcium, fish oils, primrose oil, beta-carotene &c)?” Trained nurses examined individuals at a clinic visit. Height and weight were measured and body mass index was estimated as weight (kg) divided by height (m2). Blood pressure was measured with an Accutorr noninvasive blood pressure monitor after the participant had been seated for 5 min. We used the mean of two measurements for analysis. Blood samples were taken by venepuncture into plain and citrate bottles. After overnight storage in a dark box at 4–7°C sample bottles were spun in a centrifuge at 2100 g for 15 min at 4°C. About a year after the start of the study, when funding became available, extra blood samples from participants were taken for ascorbic-acid assays. Plasma ascorbic acid was measured from blood taken into citrate bottles and plasma was stabilised in a standardised volume of metaphosphoric acid stored at –70°C. We estimated plasma ascorbic acid concentration with a fluorometric assay within 1 week of sampling.17 The coefficient of variation was 5·6% at the lower end of the range (mean 33·2 mol/L) and 4·6% at the upper end (102·3 mol/L). Other blood samples for assay were stored at 4°C and assayed at the department of clinical biochemistry, University of Cambridge, Cambridge, UK, within 1 week after
658
the sample was taken. We measured serum total cholesterol, HDL cholesterol, and triglyceride with the RA 1000 (Bayer Diagnostics, Basingstoke, UK), and calculated LDL cholesterol values with the Friedewald formula.18 At the clinic visit, we asked all individuals to complete a 7-day diet diary, and a trained nurse gave them detailed instruction. The diaries were posted back to the coordinating centre at the University of Cambridge. This diary method has been validated with weighed food records, 24 h urine collections and blood biomarkers over a year.19 Because analysis of diaries is labour intensive we coded and analysed with a specially developed programme a sample of only 2000 diaries for average daily nutrient and food intake. Statistical analysis All individuals have been flagged for death certification at the UK Office of National Statistics, with vital status ascertained for the whole cohort. We present results for mortality follow-up to December, 1999, which was about 4 years (range 2–6 years). Death certificates for all decedents were coded by trained nosologists according to the ICD 9th revision. Death was due to cardiovascular disease if the underlying cause on the death certificate was ICD 400–438, to ischaemic heart disease if it was ICD 410–414, and to cancer if it was ICD 140–239. 21 552 men and women aged 45–79 years completed the health and lifestyle questionnaire, attended the health examination, and had blood taken for ascorbic acid measurement. We excluded 1021 men and 1035 women who reported a history of heart attack, stroke, or any cancer at the baseline clinic visit, which left 8860 men and 10 636 women for analysis. Because women had higher ascorbic acid concentrations than men, we divided the individuals into sex-specific quintiles of plasma ascorbic acid. We examined risk factors in ascorbic acid concentration categories, and calculated age-adjusted death rates for all causes, cardiovascular disease, ischaemic heart disease, and cancer in every
THE LANCET • Vol 357 • March 3, 2001
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
Mortality
Ascorbic acid quintile
p*
1
2
3
4
5
Men
n=1787
n=1791
n=1761
n=1727
n=1794
All cause (n=309) Rate/100 (events) Relative risk (rate ratio) (95% CI)
5·12 (100) 1·00
4·08 (76) 0·80 (0·60–1·23)
3·03 (51) 0·59 (0·41–0·83)
2·41 (39) 0·47 (0·31–0·68)
2·44 (43) 0·48 (0·33–0·70)
<0·0001
Cardiovascular disease (n=123) Rate/100 (events) Relative risk (rate ratio) (95% CI)
2·14 (42) 1·00
1·93 (36) 0·90 (0·59–1·48)
1·43 (24) 0·67 (0·41–1·15)
0·62 (10) 0·29 (0·15–0·59)
0·62 (11) 0·29 (0·15–0·59)
<0·0001
Ischaemic heart disease (n=96) Rate/100 (events) Relative risk (rate ratio) (95% CI)
1·41 (27) 1·00
1·67 (31) 1·18 (0·72–2·08)
1·31 (22) 0·92 (0·54–1·72)
0·49 (8) 0·35 (0·16–0·81)
0·45 (8) 0·32 (0·15–0·75)
<0·0001
Cancer (n=116) Rate/100 (events) Relative risk (rate ratio) (95% CI)
1·96 (38) 1·00
1·45 (27) 0·74 (0·48–1·29)
1·00 (17) 0·51 (0·29–0·94)
1·12 (18) 0·57 (0·33–1·03)
0·92 (16) 0·47 (0·27–0·88)
<0·001
Women
n=2099
n=2046
n=2297
n=2158
n=2036
All cause (n=187) Rate/100 (events) Relative risk (rate ratio) (95% CI)
2·58 (60) 1·00
1·70 (36) 0·66 (0·44–1·02)
1·54 (33) 0·60 (0·40–0·94)
1·54 (32) 0·60 (0·38–0·92)
1·29 (26) 0·50 (0·32–0·81)
<0·0001
Cardiovascular disease (n=57) Rate/100 (events) Relative risk (rate ratio) (95% CI)
0·95 (22) 1·00
0·39 (8) 0·41 (0·18–0·93)
0·34 (7) 0·36 (0·16–0·87)
0·57 (12) 0·60 (0·31–1·29)
0·39 (8) 0·41 (0·20–1·00)
0·02
Ischaemic heart disease (n=27) Rate/100 (events Relative risk (rate ratio) (95% CI)
0·60 (14) 1·00
0·14 (3) 0·23 (0·07–0·86)
0·24 (5) 0·39 (0·15–1·19)
0·19 (4) 0·32 (0·11–1·03)
0·04 (1) 0·07 (0·0-1–0·67)
<0·001
Cancer (n=84) Rate/100 (events) Relative risk (rate ratio) (95% CI)
1·05 (24) 1·00
0·79 (17) 0·76 (0·42–1·48)
0·64 (14) 0·61 (0·33–1·23)
0·67 (14) 0·64 (0·33–1·25)
0·76 (15) 0·73 (0·38–1·40)
0·10
*p values are for 2 linear trend with 1 degree of freedom.
Table 2: Age-adjusted rates for all-cause, cardiovascular disease, and ischaemic heart disease mortality by sex-specific ascorbic acid quintile in men and women
category using 2 test for linear trend to assess statistical significance.20 We used the Cox proportional hazards model to determine the independent contribution of ascorbic acid and other risk factors for mortality due to all causes, cardiovascular disease, ischaemic heart disease, and cancer in men and women separately,21 and then with sexes combined, adjusted for sex. We also combined data for men and women and used a quadratic model with sex-specific quintiles for all-cause mortality, adjusting for sex, age, and risk factors to see whether this adjustment improved the fit of the relative risk estimates. For the 2000 sample of diaries analysed, we examined the main dietary components that correlated with plasma ascorbic acid concentration and the difference in average daily intake of the main foods that correlated by quintile of ascorbic acid.
Results Table 1 shows characteristics of the participants according to sex-specific quintile of plasma ascorbic-acid concentration. Ascorbic acid was inversely related to cardiovascular risk factors, including systolic blood pressure and body mass index, and positively to HDL cholesterol. The proportion of individuals who smoked and had a history of diabetes fell as the quintile of ascorbic acid increased. By contrast, the proportion of those who took any supplements rose as the quintile increased. Mean plasma ascorbic acid concentrations varied by about 10 mol/L between quintiles. Exclusion of people reporting use of vitamin supplements did not substantially change these means (data not shown).
THE LANCET • Vol 357 • March 3, 2001
Table 2 shows all-cause mortality rates, and those for cardiovascular disease, ischaemic heart disease, and cancer, by quintile of plasma ascorbic acid. There was a graded relation between quintiles and total mortality in both men and women, with lowest mortality rates in the highest quintile. All these relations were significant. Cardiovascular disease, ischaemic heart disease, and cancer mortality in men also fell significantly as ascorbic acid quintile increased. The number of events was smaller and trends were less consistent for these endpoints for women than for men. Table 3 shows the independent multivariate relation between plasma ascorbic acid and all-cause, cardiovascular disease, and ischaemic heart disease mortality, with the Cox proportional hazards model after adjustment for age alone, then adjustment for age and systolic blood pressure, serum cholesterol, body mass index (as continuous variables), cigarette smoking habit, diabetes, and supplement use. Replacement of total cholesterol by LDL or HDL cholesterol in the model made no difference to results (data not shown). A 20 mol/L increase in plasma ascorbic acid was associated with about a 20% decline in death due to all cause, cardiovascular disease, and ischaemic heart disease and this relation was also seen after adjustment for risk factors. The Cox models were also done after exclusion of smokers, and then supplement users. The relative risk for cancer mortality with plasma ascorbic acid in women was not significant in the Cox model. Results were closely similar after the exclusion of all people who reported a history of diabetes at baseline
659
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
Mortality
Age-adjusted hazard ratio (95% CI) (relative risk estimate)
p
Age-adjusted and risk factor2-adjusted† hazard ratio* (95% CI) (relative risk estimate)
p
Men All-cause All men (n=309) Excluding smokers (n=263) Excluding supplement users (n=195)
0·76 (0·68–0·86) 0·75 (0·65–0·86) 0·75 (0·65–0·88)
0·0001 0·0001 0·0001
0·77 (0·67–0·87) 0·74 (0·65–0·85) 0·76 (0·65–0·89)
0·0001 0·0001 0·001
Cardiovascular disease All men (n=123) Excluding smokers (n=100) Excluding supplement users (n=81)
0·62 (0·51–0·76) 0·60 (0·49–0·76) 0·62 (0·49–0·79)
0·0001 0·0001 0·0001
0·64 (0·51–0·78) 0·62 (0·49–0·78) 0·64 (0·49–0·83)
0·0001 0·0001 0·001
Ischaemic heart disease All men (n=96) Excluding smokers (n=75) Excluding supplement users (n=61)
0·66 (0·53–0·82) 0·67 (0·52–0·86) 0·65 (0·49–0·86)
0·0001 0·002 0·002
0·63 (0·42–0·94) 0·68 (0·52–0·88) 0·69 (0·51–0·92)
0·022 0·004 0·011
Cancer All men (n=116) Excluding smokers (n=103) Excluding supplement users (n=71)
0·79 (0·65–0·96) 0·75 (0·61–0·93) 0·77 (0·60–1·00)
0·019 0·009 0·05
0·76 (0·62–0·94) 0·74 (0·59–0·92) 0·76 (0·59–0·98)
0·011 0·007 0·037
Women All cause All women (n=187) Excluding smokers (n=158) Excluding supplement users (n=98)
0·80 (0·70–0·93) 0·83 (0·71–0·97) 0·73 (0·60–0·89)
0·003 0·019 0·002
0·85 (0·73–0·99) 0·84 (0·71–0·99) 0·81 (0·66–0·99)
0·033 0·037 0·049
Cardiovascular disease All women (n=57) Excluding smokers (n=48) Excluding supplement users (n=30)
0·77 (0·60–0·99) 0·77 (0·58–1·02) 0·73 (0·51–1·05)
0·048 0·070 0·089
0·81 (0·62–1·06) 0·78 (0·59–1·05) 0·78 (0·54–1·13)
0·13 0·11 0·19
Ischaemic heart disease All women (n=27) Excluding smokers (n=21) Excluding supplement users (n=15)
0·55 (0·40–0·80) 0·51 (0·33–0·78) 0·63 (0·38–1·04)
0·002 0·002 0·07
0·56 (0·36–0·87) 0·60 (0·43–0·84) 0·74 (0·44–1·26)
0·009 0·003 0·27
Cancer All women (n=84) Excluding smokers (n=71) Excluding supplement users (n=46)
0·91 (0·74–1·12) 0·94 (0·75–1·19) 0·82 (0·61–1·10)
0·38 0·61 0·18
0·97 (0·78–1·22) 0·96 (0·75–1·22) 0·92 (0·69–1·23)
0·81 0·73 0·58
*Hazard ratios (relative risks) are per increase of 20 mol/L plasma ascorbic acid concentration; †Risk factors entered are: systolic blood pressure, cholesterol, body mass index (continuous variables); cigarette smoking habit, diabetes mellitus, any supplement use.
Table 3: Cox multivariate regression for all cause, cardiovascular, and ischaemic heart disease mortality with ascorbic acid in 8860 men and 10 636 women
(data not shown). In this cohort, supplement use was not associated with mortality in the Cox regression model, after adjustment for age and other risk factors. The relative risk for all-cause mortality in supplement users was 1·00 (95% CI 0·80–1·23, p=0·97) for men, and 1·23 (0·94–1·62, p=0·14) for women with plasma All Men Women
1·2
1·0
1·0
0·8
0·8
0·6
0·6
0·4
0·4
0·2
0·2
0
All Men Women
1·2
Relative risk
Relative risk
ascorbic acid included in the model, and 0·93 (0·75–1·15, p=0·50) for men and 1·17 (0·89–1·53, p=0·27) for women when this variable was excluded. The Cox model was also run with the inclusion of alcohol intake, with a linear and quadratic model for alcohol; ascorbic acid and mortality correlations were
0 1
2 3 4 Ascorbic acid quintile
5
Figure 1: Relative risk of total mortality by quintile of plasma ascorbic acid
2 3 4 5 Ascorbic acid quintile Figure 2: Relative risk for cardiovascular disease mortality by quintile of plasma ascorbic acid
Age-adjusted and sex adjusted Cox regression model for relative risk, including a quadratic term for ascorbic-acid.
Age-adjusted and sex-adjusted Cox regression model for relative risk, including a quadratic term for ascorbic acid.
660
1
THE LANCET • Vol 357 • March 3, 2001
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
means were in the middle three quintiles with greatest differences in means between the lowest and highest quintile. The difference in mean intake between quintiles was about 20–50 g per day of fruit and 10–20 g per day of vegetables. Table 6 shows hazard ratios adjusted for age, sex, and risk factors for plasma ascorbic acid and mortality due to all causes, cardiovascular disease, ischaemic heart disease, and cancer.
Correlation coefficient r
Dietary ascorbic acid Fruit Vegetables All fruit and vegetables Potatoes Legumes Alcohol Cakes Bread Other cereals Cheese Dairy Milk Eggs Fish Meat Meat products
Men (n=800)
Women (n=1096)
0·40 0·36 0·23 0·38 ⫺0·07 ⫺0·01 0·06 0·00 0·01 ⫺0·03 0·07 0·07 ⫺0·02 0·12 ⫺0·11 0·04 ⫺0·04
0·41 0·31 0·23 0·35 ⫺0·09 ⫺0·03 0·09 ⫺0·06 ⫺0·06 0·05 0·04 0·13 ⫺0·05 ⫺0·05 0·07 ⫺0·03 ⫺0·07
Discussion
Correlation coefficients >0·20, p<0·0001.
Table 4: Pearson correlations of plasma ascorbic acid with food intake from 7-day diaries at baseline survey in EPICNorfolk 1993–97
unchanged. When men and women were combined in the same model, adjusted for sex, the relative mortality risks for a 20 mol/L increase in plasma ascorbic acid were: 0·80 (0·73–0·88, p<0·0001) for total mortality; and mortality due to cardiovascular disease 0·70 (0·60–0·82, p<0·0001), for ischaemic heart disease 0·68 (0·56–0·83, p<0·0001), and for cancer 0·85 (0·74–0·99, p=0·04). The Cox model was also run with a quadratic model to see if this analysis produced a better fit. Data for men and women were combined, and adjusted for sex in the model; the quintile category rather than the precise concentration of ascorbic acid was used in this model. Figures 1 and 2 show the age-adjusted and sex-adjusted relative risks of mortality due to all causes and cardiovascular disease mortality by quintile of ascorbic acid, with men and women combined, and with sexspecific data points. For all causes, the quadratic model showed a better fit with the data than with a linear model, with a steeper decline in risk between lower quintiles and levelling in higher quintiles; for cardiovascular mortality, the relation seemed more linear. Table 4 shows correlations of ascorbic acid with food intake estimated from 7-day diaries in 1896 men and women for whom coded and analysed diary data were available. Only fruit and vegetable intake correlated consistently and more substantially with ascorbic acid (correlation coefficient >0·20) than with other foods. Table 5 shows average daily intake of fruit and vegetables estimated from diaries by quintile of plasma ascorbic acid. As expected, the smallest differences in
In this population of men and women aged between 45 and 79 years, increasing plasma ascorbic acid concentration was strongly and independently associated with reduction in risk of mortality from all causes, cardiovascular disease, and ischaemic heart disease, with a dose-response relation across the whole population distribution. Although plasma ascorbic acid was measured once only for each individual, the relation with mortality was strong. The assumption is that one value would indicate the normal or habitual concentration. However, the large within-individual variation of plasma ascorbic acid might result in large random measurement error such that the true underlying relation could be stronger than that shown in our study. We have previously reported in this cohort an inverse relation of plasma ascorbic acid with cardiovascular risk factors, including blood pressure and lipids.22,23 Plasma ascorbic acid was also lower in smokers and diabetics.24 However, the inverse relation of this substance and mortality was consistent and independent of other cardiovascular risk factors, including blood pressure, serum cholesterol, body mass index, cigarette smoking, and diabetes. This inverse relation was also consistent after exclusion of smokers and then after supplement users. Ascorbic acid was also inversely related to cancer mortality in men but not in women. The most common types of cancer deaths in men were lung, colorectal and prostate cancer and in women were breast and colorectal cancer. The number of cancer deaths were too few for site-specific analysis but the relation might be specific to cancer types that are more common in men. There are several possible explanations for the inverse relation between plasma ascorbic acid and mortality. Pre-existing disease or cigarette smoking could lower ascorbic-acid concentrations. In our cohort, individuals with a history of cardiovascular disease or diabetes and cigarette smokers, did have low ascorbic acid concentrations. However, our analyses excluded those with known heart disease, stroke, or cancer, and the relation of plasma ascorbic acid with mortality did not change after diabetics and cigarette smokers were excluded. The proportion of cigarette smokers was low in this cohort compared with national data. This difference might indicate both the low proportion of
Plasma ascorbic acid quintile 1
2
3
4
5
Men Plasma ascorbic acid (mol/L) Dietary ascorbic acid (mg per day) Fruit intake (g per day) Vegetable intake (g per day)
n=161 20·9 (7·1) 51 (26) 75 (84) 70 (48)
n=169 38·3 (3·4) 77 (40) 120 (89) 96 (67)
n=157 48·1 (2·6) 83 (42) 147 (109) 96 (66)
n=156 57·6 (3·2) 92 (44) 164 (123) 102 (65)
n=157 73·3 (9·6) 109 (61) 192 (152) 117 (73)
Women Plasma ascorbic acid (mol/L) Dietary ascorbic acid (mg per day) Fruit intake (g per day) Vegetable intake (g per day)
n=220 29·2 (10·3) 57 (30) 108 (96) 77 (56)
n=225 49·3 (3·7) 78 (41) 155 (102) 93 (63)
n=223 60·1 (2·6) 90 (40) 186 (139) 101 (63)
n=209 68·6 (2·5) 97 (47) 186 (125) 97 (60)
n=219 86·0 (13·4) 113 (60) 227 (148) 118 (99)
Values are mean (SD).
Table 5: Food intake from 7 diaries by sex specific plasma ascorbic acid quintile at baseline survey in 800 men and 1096 women in EPIC-Norfolk 1993–1997
THE LANCET • Vol 357 • March 3, 2001
661
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES Mortality
Hazard ratio* (95% CI) adjusted for age and sex
p
Hazard ratio* (95% CI) adjusted for age, sex, and risk factor†
p
All-cause (n=692) Cardiovascular disease (n=268) Ischaemic heart disease (n=189) Cancer (n=284)
0·78 (0·72–0·86) 0·67 (0·58–0·79) 0·63 (0·52–0·77) 0·84 (0·73–0·98)
0·0001 0·0001 0·0001 0·02
0·80 (0·73–0·88) 0·70 (0·60–0·82) 0·68 (0·56–0·83) 0·85 (0·74–0·99)
0·0001 0·0001 0·0001 0·0400
*Hazard ratios (relative risks) are per increase of 20 mol/L plasma ascorbic acid concentration; †risk factors are systolic blood pressure, cholesterol, body mass index, cigarette smoking habit, diabetes mellitus, any supplement use.
Table 6: Cox multivariate regression for all-cause, cardiovascular disease, and ischaemic heart disease mortality with plasma ascorbic acid in 8860 men and 10 636 women, baseline survey in EPIC-Norfolk 1995–1999
smokers living in East Anglia and the characteristics of the participants. Misreporting of smoking habit is unlikely to be worse in this cohort than in others. Lung cancer standardised mortality rates in this cohort are about half the national rate, which lends support to the likelihood that the low proportion of smokers is not due only to under-reporting. Plasma ascorbic acid might be related to other types of behaviour that protect against atherosclerosis. One possibility is that those with high concentrations of ascorbic acid might be taking supplements. These individuals might also take other supplements, including fish oils, which might protect against cardiovascular disease. Irrespective of the direct effect of supplements, the selection biases in the characteristics of supplement users are well recognised, as is the low mortality of good compliers, even with placebo, in trials. In the EPICNorfolk cohort, a substantial proportion of participants (a third of men, half the women) reported some sort of supplement use. However, in this cohort, supplement use was not associated with mortality. There are other possible confounders such as social class and physical activity of the individuals. Although we requested information on these variables, the data are not available for analysis because of limited resources for coding. The consistency and strength of the ascorbic acid and mortality relation equal to or greater than that seen in other studies for many known factors, including social class and physical activity. This consistent relation suggests that plasma ascorbic acid might give a better indication of the mortality risk associated with these other known factors than their direct assessment. However, some of the recorded social class variations in health could be mediated through dietary differences, including those in ascorbic acid status. Antioxidants might be protective against cardiovascular disease and other age-related chronic diseases, but evidence in man is inconclusive. Most attention has focused on betacarotene and vitamin E; findings from several prospective studies show that high dietary intakes of betacarotene, vitamin E, or both, are inversely associated with subsequent cardiovascular disease mortality. However, betacarotene and vitamin E supplementation do not reduce fatal endpoints.1–7 Ascorbic acid has many biological effects that could plausibly be protective for cardiovascular disease, including scavenging of free radicals protecting against oxidative damage. Findings from studies in guineapigs (who like man cannot synthesise vitamin C) suggest that ascorbic acid deficiency increases atherosclerotic lesions, and that lesion formation is inhibited by high-dose ascorbic acid.25 However, epidemiological evidence for ascorbic acid shows that an association with coronary heart disease is not consistent,9 which could be explained by a true lack of effect, dietary measurement error, a threshold effect, an interaction with other dietary constituents, or too long or too short a duration of follow-up.
662
Measurement error is an issue in dietary exposures. Most studies have estimated ascorbic acid intake from dietary assessments—notably the food frequency questionnaire. The assessment of dietary ascorbic acid intake is especially difficult because not only is ascorbic acid present in many foods, but also the amount varies according to processing, storage, and cooking. The subjective nature of dietary assessment and other drawbacks in the estimation of ascorbic acid intake has led to the use of biological indicators. Plasma ascorbic acid measurement is thought to represent dietary intake in the preceding few weeks and might overcome some of the issues involved with dietary assessment.26 However, consistent measurement of plasma ascorbic acid is especially difficult because this substance is unstable in blood and deteriorates rapidly unless it is stabilised by the addition of substances, such as metaphosphoric acid, and stored at very low temperatures. Measurement errors might also account for the absence of consistency in studies, in which ascorbic acid has been measured. Even though intake and plasma concentration are closely related, other factors such as smoking habit or preexisting disease could account for differences in plasma ascorbic acid concentrations that are not explained by intake. These factors might confound results. Nevertheless, in many studies investigators, including early validation studies done in this cohort, have shown that plasma ascorbic acid is strongly related to dietary ascorbic acid intake.19 However, we cannot exclude residual confounding and this substance might be an indicator of other protective factors. Nevertheless, whatever these factors might be, the effect must be substantial and not explained by any of the other biological or lifestyle measures available to us. The most likely alternative explanation is that the association is with another nutrient or nutrients contained within foods that are a ready source of ascorbic acid, such as fruit and vegetables. Our findings contrast with an analysis of a 12-year follow-up of the second National Health and Nutrition Examination (NHANES II) in US adults.13 Investigators noted that men in the lowest quartile had 57% higher risk of all-cause mortality and 62% higher risk of dying from cancer than men in the highest quartile. There was no increase in risk of cardiovascular disease mortality in any quartile in men13 and no association between serum ascorbate quartile and mortality in women. NHANES II was a multicentre study but measurement variation is unlikely to account for the differences from our findings because the range and mean concentrations of plasma ascorbic acid were closely similar in the two studies. Plasma ascorbic acid might indicate different dietary patterns of fruit and vegetables which could have different biological effects. Alternatively, the longer follow-up time in NHANES II than in our study might have caused greater misclassification of individuals who could have changed ascorbic acid status during followup. Endpoints may have been misclassified because of underestimation of mortality in the NHANES study
THE LANCET • Vol 357 • March 3, 2001
For personal use only. Reproduce with permission from The Lancet Publishing Group.
ARTICLES
compared with our study in which complete mortality was ascertained. Our findings might have clinical and public health implications. The inverse relation was seen throughout the normal population range of plasma ascorbic acid in a dose-related way and not only at the extremes of the distribution. There was no evidence that ascorbic acid had an effect only below or above a threshold concentration (eg, deficiency or supplement concentrations respectively). The absence of an effect on mortality in trials of ascorbic acid so far provides no evidence to encourage supplement use.14,15 The main sources of dietary ascorbic acid are plant foods—mainly fruit and vegetables. In this cohort, plasma ascorbic acid was specifically related to fruit and vegetable intake, and an increase in 20 mol/L plasma ascorbic acid concentration was associated with an increase of about 50 g daily or one serving of fruit and vegetables. A change from the second to fourth quintile (eg, 30th–70th percentile), was associated with 30% reduction in mortality risk, which suggests that small and feasible shifts within the normal population intake could have a substantial effect on mortality risk. Plasma ascorbic acid might indicate the intake of such foods that are rich in other nutrients such as potassium, folate, calcium, magnesium, non-starch polysaccharides, isoflavonoids, and others that might confer health benefits.27–30 Whether ascorbic acid supplements are beneficial remains to be seen. However, our findings suggest that an increase in dietary intake of foods rich in ascorbic acid might have benefits for cardiovascular disease and all-cause mortality in men and women and add to the large amount of evidence that lends support to the health benefits of fruit and vegetable intake.28–30 Small and feasible changes within the normal population range of intake could have a large effect. Contributors K-T Khaw, N Day, and S Bingham originated and designed the study. S Oakes coordinated the study and organised data collection including quality control of blood samples and measurement procedures. A Welch and S Bingham analysed the dietary data and did nutritional analyses. R Luben was responsible for data management and computing overall and assisted with analyses. N Wareham introduced the diabetes measures. K-T Khaw did the data analyses and wrote the paper with co-authors.
6
7
8
9 10
11
12
13 14
15
16 17
18
19
20
21 22 23 24
Acknowledgments EPIC-Norfolk is supported by programme grants from Cancer Research Campaign and Medical Research Council with additional support from the Stroke Association, British Heart Foundation, Department of Health, Food Standards Agency, and the Wellcome Trust.
References 1
2 3 4
5
Lonn EM, Yusuf S. Is there a role for antioxidant vitamins in the prevention of cardiovascular diseases? an update on epidemiological and clinical trials data. Can J Cardiol 1997; 13: 957–65. Duthie GG, Bellizzi MC. Effects of antioxidants on vascular health. Br Med Bull 1999; 55: 568–77. Rexrode KM, Manson JE. Antioxidants and coronary heart disease: observational studies. J Cardiovasc Risk 1996; 3: 363–67. Gaziano JM. Randomized trials of dietary antioxidants in cardiovascular disease prevention and treatment. J Cardiovasc Risk 1996; 3: 368–71. Blot WJ, Li-JY, Taylor PR, Guo W, Dawsey SM, Li B. The Linxian
THE LANCET • Vol 357 • March 3, 2001
25 26
27
28
29
30
trials: mortality rates by vitamin-mineral intervention group. Am J Clin Nutr 1995; 62: 1424S–26S. The Alpha-Tocopherol Beta-Carotene Cancer Prevention Study Group The Alpha-Tocopherol. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. N Engl J Med 1994; 330: 1029–35. Yusuf S, Dagenais G, Pogue J, Bosch J, Sleight P. Vitamin E supplementation and cardiovascular events in high-risk patients. The Heart Outcomes Prevention Evaluation Study Investigators. N Engl J Med 2000; 342: 154–60. Jialal I, Vega GL, Grundy SM. Physiologic levels of ascorbate inhibit the oxidative modification of low density lipoprotein. Atherosclerosis 1990; 82: 185–91. Ness AR, Powles JW, Khaw KT. Vitamin C and cardiovascular disease: a systematic review. J Cardiovasc Risk 1996; 3: 513–21. Sahyoun NR, Jacques PF, Russell RM. Carotenoids, vitamins C and E and mortality in an elderly population. Am J Epidemiol 1996; 144: 501–11. Gale CR, Martyn CN, Winter PD, Cooper C. Vitamin C and risk of death from stroke and coronary heart disease in cohort of elderly people. BMJ 1995; 310: 1563–66. Enstrom JE, Kanim LE, Klein MA. Vitamin C intake and mortality among a sample of the United States population. Epidemiology 1992; 3: 194–202. Loria CM, Klag MJ, Caulfield LE, Whelton PK. Vitamin C status and mortality in US adults. Am J Clin Nutr 2000; 72: 139–45. Wilson TS, Datta SB, Murrell JS, Andrews CT. Relation of vitamin C levels to mortality in a geriatric hospital: a study of the effect of vitamin C administration. Age Ageing 1973; 2: 163–71. Hunt C, Chakravorty NK, Annan G. The clinical and biochemical effects of vitamin C supplementation in short stay hospitalized geriatric patients. Int J Vitam Nutr Res 1984 54: 65–74. Day NE, Oakes S, Luben R, et al. EPIC-Norfolk: study design and characteristics of the cohort. Br J Cancer 1999; 80 (suppl 1): 95–103. Vuilleumier J, Keck E. Fluorometric assay of vitamin C in biological materials using a centrifugal analyser with fluorescence attachment. J Micronutrient Analysis 1989; 5: 25–34. Friedewald WT, Levy RI, Frederickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without the use of the preparative ultracentrifuge. Clin Chem 1972; 18: 499–502. Bingham S, Cassidy A, Cole T, et al. Validation of weighed records and other methods of dietary assessment using the 24 hour urine nitrogen technique and other markers. Br J Nutr 1995; 73: 531–50. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959; 22: 719–48. Cox DR. Regression models and life tables. JR Stat Soc B 1972; 34: 187–220. Ness A, Khaw KT, Day NE, Bingham S. Vitamin C and blood pressure. J Hypertens 1996; 14: 503–08. Ness AR, Khaw KT, Bingham S, Day NE. Vitamin C and serum lipids. Eur J Clin Nutr 1996; 50: 724–29. Sargeant L, Wareham N, Bingham S, et al. Vitamin C and glucose tolerance in a EPIC-Norfolk: a population study. Diabetes Care 2000; 22: 726–32. Afridi N, Keaney JF. Animal studies on antioxidants. J Cardiovasc Risk 1996; 3: 358–62. Bates CJ, Thurnham SI, Bingham SA, Margetts BM, Nelson M. Biochemical markers of nutrient intake. In: Margetts BM, Nelson M, eds. Design concepts in nutritional epidemiology. Oxford: Oxford Medical Publications 1991; 192–265. Khaw KT, Barrett-Connor E. Dietary potassium and strokeassociated mortality: a twelve year prospective population study. N Engl J Med 1987; 316: 235–40. Tunstall-Pedoe H, Woodward M, Tavendale R, Brook RA, McCluskey MK. Comparison of the prediction by 27 different factors of coronary heart disease and death in men and women of the Scottish heart health study: cohort study. BMJ 1997; 315: 722–29. Khaw KT, Barrett-Connor E. Dietary fibre and reduced ischaemic heart disease mortality rates in men and women: a 12 year prospective study. Am J Epidemiol 1987; 126: 1093–102. Hertog MG, Feskens EJ, Kromhout D. Antioxidant flavonols and coronary heart disease risk. Lancet 1997; 349: 699.
663
For personal use only. Reproduce with permission from The Lancet Publishing Group.