Abstract
Objective
To investigate the potential influence of dietary Se intake on mortality among Chinese populations.
Design
We prospectively evaluated all-cause, CVD and cancer mortality risks associated with dietary Se intake in participants of the Shanghai Women’s Health Study (SWHS) and the Shanghai Men’s Health study (SMHS). Dietary Se intake was assessed by validated FFQ during in-person interviews. Cox proportional hazards models were used to calculate hazard ratios (HR) and 95 % CI.
Setting
Urban city in China.
Subjects
Chinese adults (n 133 957).
Results
During an average follow-up of 13·90 years in the SWHS and 8·37 years in the SMHS, 5749 women and 4217 men died. The mean estimated dietary Se intake was 45·48 μg/d for women and 51·34 μg/d for men, respectively. Dietary Se intake was inversely associated with all-cause mortality and CVD mortality in both women and men, with respective HR for the highest compared with the lowest quintile being 0·79 (95 % CI 0·71, 0·88; P trend<0·0001) and 0·80 (95 % CI 0·66, 0·98; P trend=0·0268) for women, and 0·79 (95 % CI 0·70, 0·89; P trend=0·0001) and 0·66 (95 % CI 0·54, 0·82; P trend=0·0002) for men. No significant associations were observed for cancer mortality in both women and men. Results were similar in subgroup and sensitivity analyses.
Conclusions
Dietary Se intake was inversely associated with all-cause and cardiovascular mortality in both sexes, but not cancer mortality.
Keywords: Dietary, Selenium intake, Mortality, Prospective study, China
Se, which is an essential trace element to maintain optimal human health, is incorporated into selenoproteins that have a wide variety of effects, ranging from antioxidant and anti-inflammatory effects to the production of active thyroid hormone( 1 ). Low Se status has been associated with increased mortality from cancer, poor immune function, male infertility and cognitive decline( 1 , 2 ). Meta-analyses of observational studies have provided some evidence for a beneficial effect of Se levels on bladder cancer( 3 ), prostate cancer( 4 ), lung cancer( 5 ) and CHD( 6 ), but findings from randomized trials have been mixed. Besides, randomized trials assessing the use of supplements for primary prevention usually need long periods to affect health outcomes significantly and thus observational studies, such as cohort studies, may facilitate assessment of the association between long-term nutritional status and health outcomes.
Although several prospective studies have explored the association between Se levels and all-cause mortality, their findings have been mixed with some studies showing an effect of increasing Se on decreasing risk( 7 – 10 ) and some not( 11 ). Researchers often focused on the association between all-cause mortality and the Se concentration in serum or plasma. Few studies exist in the literature examining the association between dietary Se intake and all-cause and cause-specific mortality. Moreover, dietary intake of Se varies widely worldwide owing to variability in the Se content of soil and hence of plant foods and animal forage( 12 ). In China, the level of Se intake exhibits huge variation ranging from toxic (approximately 5 mg/d in areas of Enshi County) or adequate–marginally adequate (approximately 30–90 μg /d) to low or deficient intake (Heilongjiang Province: 7–11 μg/d)( 12 , 13 ).
Therefore, using data collected in the Shanghai Women’s Health Study (SWHS) and the Shanghai Men’s Health Study (SMHS), we assessed the average level of dietary Se intake in Shanghai and prospectively investigated the potential long-term associations of dietary Se intake with all-cause, cardiovascular and cancer mortality risks in middle-aged and older Chinese adults.
Methods
Study population
Participants from the SWHS and the SMHS were included in the analysis. Details regarding the designs and methods used in these studies have been described elsewhere( 14 , 15 ). Briefly, between March 1997 and May 2000, a total of 74 941 women aged 40–70 years were recruited for the SWHS (participation rate: 92·7 %); and between April 2002 and June 2006, a total of 61 480 men aged 40–74 years with no previous history of cancer were recruited for the SMHS (participation rate: 74·1 %). In-person interviews were conducted to obtain information on sociodemographic factors, dietary and lifestyle habits, physical activity and medical history using structured questionnaires. Anthropometric measurements including weight, height, and waist and hip circumferences were also conducted at baseline.
We excluded from the analysis participants who were immediately lost to follow-up after study enrolment (five women and fourteen men), had missing data for any of covariates of interest (957 women and 1135 men), and 125 women and 228 men with extreme energy intake (<3347 or >17 573 kJ/d (<800 or >4200 kcal/d) for men; <2092 or >14 644 kJ/d (<500 or >3500 kcal/d) for women). The resulting analytic cohort included 73 854 women and 60 103 men.
Dietary assessment
In both the SWHS and the SMHS, usual dietary intakes were assessed using semi-quantitative FFQ. The FFQ used in the SWHS included seventy-seven items that covered 85·6 % of foods commonly consumed in urban Shanghai in 1996( 16 ). A similar but extended FFQ with eighty-one items was used in the SMHS which captured 88·8 % of commonly consumed foods( 17 ). During the in-person interviews, participants were asked about how frequently they consumed the food or food group during the preceding year (five categories: daily, weekly, monthly, yearly or never) and then followed by a question on the amount of food consumed each time, in liangs (1 liang=50 g). Each participant was also asked about whether he or she had taken supplements (vitamin A, B, C or E; a multivitamin; or calcium) at least three times per week continuously for more than 2 months.
The Chinese food composition tables( 18 ) were used to calculate daily intakes of total energy and nutrients. The reproducibility and validity of the FFQ in the SMHS/SWHS were determined using monthly (SMHS; n 12) or biweekly (SWHS; n 24) 24 h dietary recall evaluation over a 1-year period. The correlation coefficients for micronutrients ranged from 0·33 to 0·58 in the SMHS( 17 ) and from 0·41 to 0·59 in the SWHS( 16 ).
Follow-up and outcome ascertainment
Study participants were followed up by in-person surveys every two to three years and annual record linkage with the Shanghai Vital Statistics Registry. For the SWHS, the response rates for the first, second, third and fourth surveys were 99·8%, 98·7%, 96·7% and 92·3 %, respectively. For the SMHS, the response rates for the first and second surveys were 97·6% and 93·7 %, respectively. All possible matches identified through the linkage were verified by home visits. The underlying cause of death was determined primarily on the basis of death certificate data from the Shanghai Vital Statistics Unit and coded according to the International Classification of Diseases, Ninth Revision (ICD-9). Our primary end point was death from any cause. We also examined deaths from CVD (ICD-9 codes: 390–459) and cancer (ICD-9 codes: 140–208).
Statistical analysis
Dietary Se intake was adjusted for total energy using the nutrient density method( 19 ) and then categorized by quintile distribution, with the lowest quintile serving as the reference group. We used Cox proportional hazards regression models to determine the association of dietary Se intake with total and cause-specific mortality, with person-years as the underlying time metric. The proportional hazards assumption was evaluated using Schoenfeld residual plots and no evidence of violation was observed. Tests for linear trend across quintiles were estimated by assigning the median intake value for the quintile to each person and including this as a continuous variable in the regression model. Person-years of follow-up were calculated as the interval between baseline recruitment and the date of death, loss to follow-up or 31 December 2012, whichever was earlier.
We presented risk estimates separately for men and women. In minimally adjusted models, we included age and energy intake as covariates. In multivariable-adjusted models, we further adjusted for the following baseline factors: birth cohort (1920s, 1930s, 1940s, 1950s, 1960s); level of education (four categories: elementary school or less, middle school, high school, college or above); income (four categories: low, low to middle, middle to high, high); marital status (four categories); occupation (housewife (women only), manual, clerical, professional), BMI (four categories: <18·5 kg/m2, 18·5–24 kg/m2, 24–28 kg/m2, ≥28 kg/m2); physical activity (quartiles of MET-h/week per year, where MET=metabolic equivalent of task); energy-adjusted fat intake (g/4184 kJ (1000 kcal) per d, continuous); use of any vitamin supplement (yes/no); smoking status (for men: never, ever, current; for women: never, ever); drinking status (for men: never, ever, current; for women: never, ever); status with regard to a history of hypertension (yes/no), diabetes (yes/no), CHD (yes/no) and stroke (yes/no); and family history of cancer (yes/no). For women, menopausal status (yes/no) was also included in multivariate models.
To minimize the influence of possible reverse causation owing to the presence of chronic diseases at baseline, sensitivity analyses were conducted by excluding the first 2 years of follow-up and restricting the analyses to participants who did not have a history of hypertension, diabetes, CHD or stroke at baseline, or family history of cancer. We also restricted the analysis among lifetime non-smokers to eliminate the potential confounding effect of cigarette smoking on the association between dietary Se intake and death. In secondary analyses, we examined associations among pre-specified baseline subgroups based on the following: BMI, drinking status, use of any supplement; and in women, menopausal status.
Analyses were performed with the statistical software package SAS version 9.2. Statistical tests were two-sided, and P values of less than 0·05 were considered statistically significant.
Results
Se intake and dietary, lifestyle factors
The average estimated dietary Se intake was 45·48 μg/d for women and 51·34 μg/d for men, which is close to the recommended nutrient intake level of 50 μg/d for the Chinese population( 18 ). Compared with participants in the lowest quintile of Se density intake, those in the highest quintile were younger, with lower BMI and were less likely to exercise, but had higher total energy and fat intakes, family income and educational level. They were also more likely to use multivitamin supplements, but less likely to have a history of hypertension, CHD and stroke. Men in the highest quintile were also more likely to smoke and consume alcoholic drinks (Table 1).
Table 1.
Baseline characteristics by energy-adjusted quintile of selenium intake in the Shanghai Women’s Health Study (SWHS; 1997–2000) and the Shanghai Men’s Health Study (SMHS; 2002–2006)
| Energy-adjusted quintile of selenium intake | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | |||||||
| Mean/% | sd | Mean/% | sd | Mean/% | sd | Mean/% | sd | Mean/% | sd | P value | |
| SWHS | |||||||||||
| No. of participants | 14 771 | – | 14 771 | – | 14 769 | – | 14 771 | – | 14 772 | ||
| Age (years) | 56·72 | 9·14 | 53·40 | 8·97 | 51·94 | 8·64 | 51·03 | 8·42 | 49·77 | 8·03 | <0·0001 |
| High educational level (%)† | 2·07 | – | 4·08 | – | 5·5 | – | 6·06 | – | 5·89 | – | <0·0001 |
| High income (%)‡ | 10·02 | – | 14·95 | – | 18·43 | – | 20·7 | – | 23·78 | – | <0·0001 |
| Currently married (%) | 83·83 | – | 89·22 | – | 90·07 | – | 91·04 | – | 90·29 | – | <0·0001 |
| BMI (kg/m2) | 24·87 | 3·69 | 24·13 | 3·44 | 23·82 | 3·38 | 23·73 | 3·25 | 23·56 | 3·22 | <0·0001 |
| Waist-to-hip ratio | 0·82 | 0·06 | 0·81 | 0·05 | 0·81 | 0·05 | 0·81 | 0·05 | 0·80 | 0·05 | <0·0001 |
| Physical activity (MET-h/week per year)§ | 109·42 | 45·69 | 107·98 | 45·72 | 105·86 | 44·61 | 105·26 | 44·48 | 103·71 | 44·56 | <0·0001 |
| Total energy intake (kJ/d) | 6615 | 1659 | 6933 | 1595 | 7059 | 1579 | 7150 | 1606 | 7276 | 1720 | <0·0001 |
| Total energy intake (kcal/d) | 1580·99 | 396·44 | 1656·97 | 381·32 | 1687·19 | 377·31 | 1708·87 | 383·92 | 1739·03 | 411·10 | <0·0001 |
| Total fat (g/d) | 19·47 | 9·42 | 26·55 | 10·23 | 30·48 | 11·39 | 33·48 | 12·32 | 37·01 | 13·90 | <0·0001 |
| Ever smoker (%) | 4·64 | – | 2·83 | – | 2·28 | – | 2·00 | – | 2·19 | – | <0·0001 |
| Alcohol drinking (ever) (%) | 1·81 | – | 1·90 | – | 1·97 | – | 2·15 | – | 3·33 | – | <0·0001 |
| History of hypertension (%) | 31·52 | – | 25·84 | – | 22·09 | – | 21·07 | – | 18·25 | – | <0·0001 |
| History of CHD (%) | 9·46 | – | 8·22 | – | 6·93 | – | 6·61 | – | 5·42 | – | <0·0001 |
| History of stroke (%) | 2·15 | – | 1·37 | – | 0·99 | – | 0·69 | – | 0·58 | – | <0·0001 |
| Family history of cancer (%) | 23·33 | – | 26·63 | – | 27·77 | – | 27·81 | – | 27·90 | – | <0·0001 |
| History of diabetes (%) | 4·85 | – | 4·89 | – | 4·24 | – | 4·16 | – | 3·63 | – | <0·0001 |
| Supplement use|| (%) | 24·15 | – | 29·19 | – | 31·61 | – | 34·75 | – | 35·60 | – | <0·0001 |
| Postmenopausal (%) | 67·48 | – | 53·73 | – | 46·88 | – | 42·33 | – | 36·98 | – | <0·0001 |
| SMHS | |||||||||||
| No. of participants | 12 021 | – | 12 022 | – | 12 022 | – | 12 016 | – | 12 022 | – | |
| Age (years) | 57·10 | 10·06 | 56·11 | 9·78 | 55·60 | 9·64 | 54·88 | 9·53 | 53·11 | 9·23 | <0·0001 |
| High educational level (%)† | 6·77 | – | 10·44 | – | 12·63 | – | 13·71 | – | 13·89 | – | <0·0001 |
| High income (%)‡ | 4·77 | – | 7·62 | – | 9·87 | – | 11·79 | – | 14·37 | – | <0·0001 |
| Currently married (%) | 82·41 | – | 85·15 | – | 85·67 | – | 85·49 | – | 83·7 | – | <0·0001 |
| BMI (kg/m2) | 23·79 | 3·12 | 23·75 | 3·09 | 23·70 | 3·07 | 23·74 | 3·07 | 23·63 | 3·04 | 0·001 |
| Waist-to-hip ratio | 0·90 | 0·06 | 0·90 | 0·06 | 0·90 | 0·06 | 0·90 | 0·06 | 0·90 | 0·06 | <0·0001 |
| Physical activity (MET-h/week per year)§ | 62·79 | 35·28 | 60·80 | 33·73 | 59·87 | 34·26 | 58·55 | 33·25 | 56·41 | 34·06 | <0·0001 |
| Total energy intake (kJ/d) | 7918 | 1992 | 7976 | 1939 | 8010 | 1941 | 8015 | 1953 | 8041 | 2101 | <0·0001 |
| Total energy intake (kcal/d) | 1892·54 | 476·06 | 1906·29 | 463·51 | 1914·39 | 463·84 | 1915·61 | 466·66 | 1921·87 | 502·04 | <0·0001 |
| Total fat (g/d) | 24·34 | 11·13 | 31·25 | 11·99 | 35·28 | 13·72 | 38·71 | 14·75 | 43·25 | 17·87 | <0·0001 |
| Current smoking (%) | 56·89 | – | 55·97 | – | 57·54 | – | 58·82 | – | 64·38 | – | <0·0001 |
| Current alcohol consumption (%) | 19·03 | – | 23·32 | – | 28·89 | – | 32·08 | – | 42·80 | – | <0·0001 |
| History of hypertension (%) | 32·93 | – | 31·52 | – | 29·75 | – | 29·21 | – | 25·83 | – | <0·0001 |
| History of CHD (%) | 5·76 | – | 5·40 | – | 5·32 | – | 5·06 | – | 3·99 | – | <0·0001 |
| History of diabetes (%) | 4·67 | – | 5·80 | – | 6·39 | – | 7·50 | – | 6·99 | – | <0·0001 |
| History of stroke (%) | 5·53 | – | 4·30 | – | 3·70 | – | 3·01 | – | 2·30 | – | <0·0001 |
| Family history of cancer (%) | 26·65 | – | 28·32 | – | 29·35 | – | 28·86 | – | 28·76 | – | <0·0001 |
| Supplement use|| (%) | 11·72 | – | 16·49 | – | 18·47 | – | 20·02 | – | 20·48 | – | <0·0001 |
The median and cut-off points for the quintile of Se intake in the SWHS were 15·86 (7·19–19·04), 21·16 (19·05–23·15), 25·18 (23·16–27·31), 30·02 (27·32–33·35) and 42·22 (33·36–124·64) μg/4184 kJ (1000 kcal) per d and in the SMHS were 16·46 (7·43–19·36), 21·33 (19·37–23·20), 25·09 (23·21–27·11), 29·72 (27·12–32·91) and 41·70 (32·92–165·38) μg/4184 kJ (1000 kcal) per d.
High educational level was defined as college or above.
High income was defined as a family income greater than 30 000 Yuan per year for women or a personal income greater than 2000 Yuan per month for men.
Physical activity: metabolic equivalent of task (MET)-h/week per year (1 MET-h=15 min of moderate-intensity activity).
Use of any vitamin A, B, C or E supplements, multivitamins or calcium.
Dietary Se intake and total mortality
During an average follow-up of 13·90 years in the SWHS and 8·37 years in the SMHS, we documented 9966 deaths from all causes (4217 men and 5749 women), including 3154 deaths from CVD (1402 men and 1752 women) and 4352 deaths from cancer (1798 men and 2554 women).
Age- and energy-adjusted and multivariate-adjusted analyses showed a significant inverse association between dietary Se intake and total mortality among both men and women (Table 2). Multivariate-adjusted hazard ratios (HR) for total death among women across the lowest to the highest quintile of Se intake were 1·00 (reference), 0·96 (95 % CI 0·89, 1·03), 0·92 (95 % CI 0·84, 1·00), 0·90 (95 % CI 0·82, 0·99) and 0·79 (95 % CI 0·71, 0·88), respectively (P<0·0001 for trend across categories). The corresponding HR among men were 1·00 (reference), 0·91 (95 % CI 0·83, 0·99), 0·86 (95 % CI 0·78, 0·95), 0·82 (95 % CI 0·73, 0·91) and 0·79 (95 % CI 0·70, 0·89), respectively (P=0·0001 for trend across categories).
Table 2.
Association of dietary selenium intake with total and cause-specific mortality in the Shanghai Women’s Health Study (SWHS; 1997–2000) and the Shanghai Men’s Health Study (SMHS; 2002–2006)
| Quintile category | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | |||||||
| Cause of death | HR | 95 % CI | HR | 95 % CI | HR | 95 % CI | HR | 95 % CI | HR | 95 % CI | P for trend* |
| SWHS | |||||||||||
| All causes | |||||||||||
| Median (µg/4184 kJ (1000 kcal) per d)† | 15·86 | 21·16 | 25·18 | 30·02 | 42·22 | ||||||
| No. of deaths/person-years | 1863/199 978 | 1285/203 892 | 1033/204 976 | 892/205 253 | 676/205 437 | ||||||
| Model 1‡ | 1·00 | Reference | 0·94 | 0·88, 1·01 | 0·89 | 0·82, 0·96 | 0·85 | 0·79, 0·93 | 0·76 | 0·69, 0·83 | <0·0001 |
| Model 2§ | 1·00 | Reference | 0·96 | 0·89, 1·03 | 0·92 | 0·84, 1·00 | 0·90 | 0·82, 0·99 | 0·79 | 0·71, 0·88 | <0·0001 |
| CVD | |||||||||||
| No. of deaths/person-years | 648/199 978 | 390/203 892 | 303/204 976 | 235/205 253 | 176/205 437 | ||||||
| Model 1‡ | 1·00 | Reference | 0·90 | 0·79, 1·02 | 0·85 | 0·74, 0·97 | 0·76 | 0·65, 0·89 | 0·71 | 0·60, 0·84 | <0·0001 |
| Model 2§ | 1·00 | Reference | 0·94 | 0·82, 1·07 | 0·93 | 0·80, 1·08 | 0·87 | 0·73, 1·03 | 0·80 | 0·66, 0·98 | 0·0268 |
| Cancer | |||||||||||
| No. of deaths/person-years | 708/199 978 | 559/203 892 | 482/204 976 | 451/205 253 | 354/205 437 | ||||||
| Model 1‡ | 1·00 | Reference | 1·00 | 0·89, 1·12 | 0·97 | 0·86, 1·09 | 0·98 | 0·87, 1·11 | 0·87 | 0·76, 0·99 | 0·0326 |
| Model 2§ | 1·00 | Reference | 1·02 | 0·90, 1·14 | 1·00 | 0·87, 1·14 | 1·02 | 0·89, 1·17 | 0·90 | 0·77, 1·05 | 0·1539 |
| SMHS | |||||||||||
| All causes | |||||||||||
| Median (µg/4184 kJ (1000 kcal) per d)† | 16·46 | 21·33 | 25·09 | 29·72 | 41·7 | ||||||
| No. of deaths/person-years | 1136/98 122 | 885/98 995 | 819/99 583 | 745/100 137 | 632/101 936 | ||||||
| Model 1‡ | 1·00 | Reference | 0·86 | 0·78, 0·93 | 0·82 | 0·75, 0·90 | 0·79 | 0·72, 0·87 | 0·77 | 0·70, 0·85 | <0·0001 |
| Model 2§ | 1·00 | Reference | 0·91 | 0·83, 0·99 | 0·86 | 0·78, 0·95 | 0·82 | 0·73, 0·91 | 0·79 | 0·70, 0·89 | 0·0001 |
| CVD | |||||||||||
| No. of deaths/person-years | 433/98 122 | 302/98 995 | 253/99 583 | 235/100 137 | 179/101 936 | ||||||
| Model 1‡ | 1·00 | Reference | 0·78 | 0·67, 0·90 | 0·68 | 0·58, 0·80 | 0·68 | 0·58, 0·80 | 0·61 | 0·51, 0·72 | <0·0001 |
| Model 2§ | 1·00 | Reference | 0·85 | 0·73, 1·00 | 0·75 | 0·63, 0·89 | 0·73 | 0·61, 0·88 | 0·66 | 0·54, 0·82 | 0·0002 |
| Cancer | |||||||||||
| No. of deaths/person-years | 432/98 122 | 366/98 995 | 356/99 583 | 325/100 137 | 319/101 936 | ||||||
| Model 1‡ | 1·00 | Reference | 0·92 | 0·80, 1·05 | 0·92 | 0·80, 1·05 | 0·88 | 0·76, 1·02 | 0·98 | 0·85, 1·13 | 0·8311 |
| Model 2§ | 1·00 | Reference | 0·95 | 0·82, 1·10 | 0·94 | 0·81, 1·10 | 0·90 | 0·76, 1·06 | 0·97 | 0·81, 1·15 | 0·7802 |
HR, hazard ratio.
P values for trend were estimated by assigning the median intake value for the quintile to each person and including this as a continuous variable in the model.
Dietary intakes were adjusted for energy intake using the nutrient density method and expressed as per μg/4184 kJ (1000 kcal) per d.
Adjusted for age and total energy intake.
Adjusted for age, birth cohort, education, income, marital status, occupation, BMI, physical activity, total energy intake, dietary fat intake, supplement use, smoking status, drinking status, status with regard to a history of hypertension, diabetes, CHD or stroke, family history of cancer and menopausal status (women only).
Dietary Se intake and CVD and cancer mortality
In multivariate analyses, Se intake was inversely associated with CVD mortality in both men and women. The HR for the highest v. lowest quintile were 0·80 (95 % CI 0·66, 0·98; P for trend=0·0268) among women and 0·66 (95 % CI 0·54, 0·82; P for trend=0·0002) among men. In contrast, no significant associations between Se intake and deaths from cancer were observed in both women and men.
Sensitivity and subgroup analyses
The significant inverse association between Se intake and all-cause mortality remained largely unchanged when we excluded the first 2 years of follow-up (628 men and 441 women; see online supplementary material, Table S1); excluded participants with hypertension (17 940 men and 17 544 women), diabetes (3767 men and 3216 women), CHD (3069 men and 5411 women) or stroke at baseline (2265 men and 853 women), or family history of cancer (17 062 men and 19 710 women); excluded participants who had ever smoked (41 900 men and 2059 women); excluded participants who had ever drunk (20 231 men); and excluded participants who reported use of any supplement at the baseline survey (10 480 men and 22 940 women; Table S2).
In stratified analyses, we found similar inverse associations between dietary Se intake and all-cause mortality across subgroups defined by menopause status (P=0·6543 for interaction) and BMI (P=0·8968 for interaction in men, P=0·8259 for interaction in women). We noted significant interactions between waist-to-hip ratio and Se intake with respect to total death (P=0·0399 for interaction in men, P<0·0001 for interaction in women); stronger associations were observed among men and women who had lower waist-to-hip ratio (≤0·90 for men, HR=0·71 (95 % CI 0·60, 0·84); ≤0·85 for women, HR=0·76 (95 % CI 0·67, 0·87)) than among those who had high waist-to-hip ratio (HR=0·87 (95 % CI 0·73, 1·03) for men; HR=0·83 (95 % CI 0·69, 1·00) for women; see online supplementary material, Table S3).
Discussion
In the current analysis of two large population-based cohorts involving 133 957 participants living in Shanghai, China, we found a significant inverse association between dietary Se intake and all-cause and CVD mortality, after adjusting for potential confounders. Compared with the lowest quintile, women in the highest quintile of Se intake had a 21 % lower risk of all-cause mortality and a 20 % lower risk of CVD mortality, whereas men in this category of consumption had a 21 % and a 34 % lower risk, respectively. No association was found between Se intake and cancer mortality for either men or women.
Our finding of a significant inverse association between dietary Se intake and all-cause mortality is generally consistent with findings from most previous prospective studies on plasma or serum Se level and mortality. In an analysis of 1389 elderly French participants in the Etude du Vieillissement Arteriel (EVA), low plasma Se at baseline was positively associated with all-cause mortality (relative risk (RR)=1·54 (95 % CI 1·25, 1·88))( 9 ). In the Invecchiare in Chianti, adults in the lowest quartile of Se had higher mortality risk compared with those in the highest quartile (RR=1·60 (95 % CI 1·04, 2·47))( 7 ). A non-linear association between Se level and all-cause mortality was observed in 13 887 US adult participants followed up for up to 12 years in the Third National Health and Nutrition Examination Survey( 10 ). By contrast, in a cohort analysis of 1103 participants in LinXian( 11 ), China, no association was noted between baseline serum Se (mean 73 μg/l) and all-cause mortality (RR=0·93 (95 % CI 0·72, 1·19)). Moreover, in a meta-analysis of randomized controlled trials, Se given alone or in combination with other supplements had no significant effect on mortality in seventeen trials (RR=0·97 (95 % CI 0·91, 1·03))( 20 ), such as the Nutritional Prevention of Cancer (NPC) trial (median baseline plasma Se: 114 ng/ml)( 21 ) and the Selenium and Vitamin E Cancer Prevention Trial (SELECT; median baseline serum Se: 136 μg/l)( 22 ).
Se may protect against CVD by preventing oxidative modification of lipids, inhibiting platelet aggregation, reducing inflammation( 1 , 23 ) and improving functional capillary recruitment( 24 ). Although a meta-analysis of observational studies showed a significant inverse association between Se concentration and risk of CHD( 6 ), randomized trials using Se in combination with other antioxidants have not shown a significant protective effect on CVD or mortality( 6 , 22 ). However, most of these large prevention trials did not consider baseline nutrition level in their inclusion criteria and Se was given in combination with other vitamins and minerals in all but two trials( 25 , 26 ).
In contrast with previous prospective studies, our analysis of the SWHS and the SMHS data found no association between dietary Se intake and cancer mortality. In meta-analyses of observational studies, the pooled RR risk comparing the highest with the lowest category of Se levels was 0·61 (95 % CI, 0·32, 0·95) for bladder cancer( 3 ) and 0·74 (95 % CI, 0·57, 0·97)( 5 ) for lung cancer. For prostate cancer, a more significant protective association was detected between Se and risk of advanced, rather than localized or low-grade, prostate cancer( 27 ). However, results from the recent randomized controlled trials have failed to provide evidence of a beneficial effect of Se supplementation on risk of all cancers, prostate cancer or other site-specific cancers( 28 ). In the NPC trial, Se supplementation was associated with a statistically significant decrease in total cancer mortality, total cancer incidence and incidence of lung, colorectal and prostate cancers( 29 ). However, with longer follow-up time, the differences in lung and colorectal cancer became statistically non-significant. Another randomized controlled trial, SELECT, found no evidence of effects on the incidence of prostate, lung or any cancer overall at or beyond 5 years of follow-up with Se supplementation( 22 ).
Differences in baseline Se plasma or serum levels of the populations studied might account for the observed differences. Beyond a specific Se concentration, additional Se intake does not result in additional reduction of mortality and human studies have provided evidence of a U-shaped association between intake or status and health outcomes( 10 , 29 ). Moreover, because published randomized trials used a wide variety of supplements, in different doses, with different objectives and populations, and with short duration of follow-up time and small sample size, the power to detect the health effects of Se was slightly limited.
The underlying biological mechanism by which low dietary Se intake contributes to an increased mortality risk may be related to the increased oxidative stress and inflammation effects( 8 ). Higher Se levels may potentially protect against oxidative stress and reduce pro-inflammation cytokines and other markers of inflammation, including C-reactive protein, by incorporation into selenoproteins such as glutathione peroxidase and seleneprotein S (SEPS 1)( 30 ). Low Se may compromise health by decreasing the synthesis and activity of deiodinase, the enzyme that transforms thyroxine into the biologically active triiodothyronine( 31 ). Moreover, low Se status may be implicated in the pathogenesis of atherosclerosis through its effects on regulating the cyclo-oxygenase and lipoxygenase pathways of the arachidonic acid cascade in endothelial cells( 23 ). In the Uppsala Longitudinal Study of Adult Men, high concentrations of serum Se predicted reduced levels of urinary F2-isoprostane, a biomarker of lipid peroxidation and oxidative stress( 32 ).
Strengths of our study include the population-based, prospective design with high participation and retention rates and detailed information on diet. The extensive data on possible risk factors for mortality allowed comprehensive adjustments for confounders. Results from the various sensitivity and subgroup analyses we carried out yielded similar results throughout, which suggests that our findings are fairly robust. However, several limitations of our study also merit consideration. First, as with any nutritional epidemiological study, measurement errors in self-reported dietary data introduced by FFQ, which are most likely non-differential, may have attenuated estimates for the associations. Second, because dietary and lifestyle factors interact in complex ways with each other, we cannot entirely separate the effect of dietary Se from those of other nutrients and foods, and thus we cannot completely rule out the possibility that some unmeasured confounders or residual confounding accounted for the observed associations. Third, Se intake from supplements was not taken into account in our analysis. However, the associations between Se intake and mortality did not change substantially after exclusion of supplement users. Fourth, dietary intake assessed at baseline may have been affected by preclinical conditions. Nevertheless, excluding the first 2 years of follow-up and participants with chronic diseases at baseline did not alter the results. Fifth, since Se levels in foods may vary between geographic regions due to the Se content of the soil where the food was produced, serum or plasma Se levels would have been preferable to estimate Se exposure. However, owing to constraints on resources, serum or plasma Se levels were not measured for the participants in our study. We expect that the errors in dietary Se assessment would be random and likely lead to an attenuation of the true effect rather than create an artificial effect when none existed. Sixth, it is not known whether participants’ diet during the baseline year reflected their diet during the biologically relevant period. Thus, although an inverse association between Se exposure and mortality was found in some observational studies, including our study, this cannot be taken as evidence of a causal relationship and our results should be interpreted with caution.
Conclusion
In conclusion, we found that increased dietary Se intake was associated with lower all-cause and CVD mortality risk among Chinese men and women, consistent with previous observational studies, suggesting that individuals in the lower categories of Se exposure may benefit from increased Se intake.
Acknowledgements
Acknowledgements: The authors would like to thank the participants of the SMHS and the SWHS for their invaluable contribution to this work. Financial support: This work was supported by the State Key Laboratory of Oncogene and Related Genes (grant number 91-15-10) and the Shanghai Health Bureau Key Disciplines and Specialties Foundation, and the US National Institutes of Health (grant numbers R37 CA070867 and UM1 CA182910, R01 CA082729 and UM1 CA173640). All funders had no role in the design, analysis or writing of this article. Conflict of interest: None. Authorship: Y.-B.X. conceived and designed the study, and was responsible for final editing and approval of the manuscript. J.-W.S., H.-L.L., W. Zhang and L.-G.Z. analysed the data. J.-W.S. wrote the first draft. All authors critically reviewed the manuscript and approved the final version. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the all relevant Institutional Review Boards. Written informed consent was obtained from all participants.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980016001130.
click here to view supplementary material
References
- 1. Rayman MP (2012) Selenium and human health. Lancet 379, 1256–1268. [DOI] [PubMed] [Google Scholar]
- 2. Papp LV, Lu J, Holmgren A et al. (2007) From selenium to selenoproteins: synthesis, identity, and their role in human health. Antioxid Redox Signal 9, 775–806. [DOI] [PubMed] [Google Scholar]
- 3. Amaral AFS, Cantor KP, Silverman DT et al. (2010) Selenium and bladder cancer risk: a meta-analysis. Cancer Epidemiol Biomarkers Prev 19, 2407–2415. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Etminan M, FitzGerald JM, Gleave M et al. (2005) Intake of selenium in the prevention of prostate cancer: a systematic review and meta-analysis. Cancer Causes Control 16, 1125–1131. [DOI] [PubMed] [Google Scholar]
- 5. Zhuo H, Smith AH & Steinmaus C (2004) Selenium and lung cancer: a quantitative analysis of heterogeneity in the current epidemiological literature. Cancer Epidemiol Biomarkers Prev 13, 771–778. [PubMed] [Google Scholar]
- 6. Flores-Mateo G, Navas-Acien A, Pastor-Barriuso R et al. (2006) Selenium and coronary heart disease: a meta-analysis. Am J Clin Nutr 84, 762–773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Lauretani F, Semba RD, Bandinelli S et al. (2008) Low plasma selenium concentrations and mortality among older community-dwelling adults: the InCHIANTI Study. Aging Clin Exp Res 20, 153–158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Ray AL, Semba RD, Walston J et al. (2006) Low serum selenium and total carotenoids predict mortality among older women living in the community: the women’s health and aging studies. J Nutr 136, 172–176. [DOI] [PubMed] [Google Scholar]
- 9. Akbaraly NT (2005) Selenium and mortality in the elderly: results from the EVA study. Clin Chem 51, 2117–2123. [DOI] [PubMed] [Google Scholar]
- 10. Bleys J, Navas-Acien A & Guallar E (2008) Serum selenium levels and all-cause, cancer, and cardiovascular mortality among US adults. Arch Intern Med 168, 404–410. [DOI] [PubMed] [Google Scholar]
- 11. Wei WQ, Abnet CC, Qiao YL et al. (2004) Prospective study of serum selenium concentrations and esophageal and gastric cardia cancer, heart disease, stroke, and total death. Am J Clin Nutr 79, 80–85. [DOI] [PubMed] [Google Scholar]
- 12. Rayman MP (2008) Food-chain selenium and human health: emphasis on intake. Br J Nutr 100, 254–268. [DOI] [PubMed] [Google Scholar]
- 13. Combs GF (2001) Selenium in global food systems. Br J Nutr 85, 517–547. [DOI] [PubMed] [Google Scholar]
- 14. Zheng W, Chow WH, Yang G et al. (2005) The Shanghai Women’s Health Study: rationale, study design, and baseline characteristics. Am J Epidemiol 162, 1123–1131. [DOI] [PubMed] [Google Scholar]
- 15. Shu X, Li H, Yang G et al. (2015) Cohort profile: the Shanghai Men’s Health Study. Int J Epidemiol 44, 810–818. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Shu XO, Yang G, Jin F et al. (2004) Validity and reproducibility of the food frequency questionnaire used in the Shanghai Women’s Health Study. Eur J Clin Nutr 58, 17–23. [DOI] [PubMed] [Google Scholar]
- 17. Villegas R, Yang G, Liu D et al. (2007) Validity and reproducibility of the food-frequency questionnaire used in the Shanghai Men’s Health Study. Br J Nutr 97, 993–1000. [DOI] [PubMed] [Google Scholar]
- 18. Yang YX, Wang GY & Pan XC (2002) China Food Composition 2002. Beijing: Peking University Medical Press. [Google Scholar]
- 19. Willett WC, Howe GR & Kushi LH (1997) Adjustment for total energy intake in epidemiologic studies. Am J Clin Nutr 65, 4 Suppl., 1220S–1228S. [DOI] [PubMed] [Google Scholar]
- 20. Bjelakovic G, Nikolova D, Gluud LL et al. (2012) Antioxidant supplements for prevention of mortality in healthy participants and patients with various diseases. Cochrane Database Syst Rev 3, CD007176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Clark LC, Combs GJ, Turnbull BW et al. (1996) Effects of selenium supplementation for cancer prevention in patients with carcinoma of the skin. A randomized controlled trial. Nutritional Prevention of Cancer Study Group. JAMA 276, 1957–1963. [PubMed] [Google Scholar]
- 22. Lippman SM, Klein EA, Goodman PJ et al. (2009) Effect of selenium and vitamin E on risk of prostate cancer and other cancers: the Selenium and Vitamin E Cancer Prevention Trial (SELECT). JAMA 301, 39–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Rayman MP (2000) The importance of selenium to human health. Lancet 356, 233–241. [DOI] [PubMed] [Google Scholar]
- 24. Buss C, Marinho C, Maranhão PA et al. (2013) Long-term dietary intake of selenium, calcium, and dairy products is associated with improved capillary recruitment in healthy young men. Eur J Nutr 52, 1099–1105. [DOI] [PubMed] [Google Scholar]
- 25. Stranges S, Marshall JR, Trevisan M et al. (2006) Effects of selenium supplementation on cardiovascular disease incidence and mortality: secondary analyses in a randomized clinical trial. Am J Epidemiol 163, 694–699. [DOI] [PubMed] [Google Scholar]
- 26. Korpela H, Kumpulainen J, Jussila E et al. (1989) Effect of selenium supplementation after acute myocardial infarction. Res Commun Chem Pathol Pharmacol 65, 249–252. [PubMed] [Google Scholar]
- 27. Peters U & Takata Y (2008) Selenium and the prevention of prostate and colorectal cancer. Mol Nutr Food Res 52, 1261–1272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Vinceti M, Dennert G, Crespi CM et al. (2014) Selenium for preventing cancer. Cochrane Database Syst Rev 3, CD005195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Duffield-Lillico AJ, Reid ME, Turnbull BW et al. (2002) Baseline characteristics and the effect of selenium supplementation on cancer incidence in a randomized clinical trial: a summary report of the Nutritional Prevention of Cancer Trial. Cancer Epidemiol Biomarkers Prev 11, 630–639. [PubMed] [Google Scholar]
- 30. Rayman MP (2009) Selenoproteins and human health: insights from epidemiological data. Biochim Biophys Acta 1790, 1533–1540. [DOI] [PubMed] [Google Scholar]
- 31. Beckett GJ (2005) Selenium and endocrine systems. J Endocrinol 184, 455–465. [DOI] [PubMed] [Google Scholar]
- 32. Helmersson J, Arnlov J, Vessby B et al. (2005) Serum selenium predicts levels of F2-isoprostanes and prostaglandin F2α in a 27 year follow-up study of Swedish men. Free Radic Res 39, 763–770. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980016001130.
click here to view supplementary material