Abstract
Background: Limited research has examined the association between dietary patterns and mortality, especially in non-Western populations.
Objective: We examined the association of dietary patterns with all-cause mortality and cause-specific mortality in the Singapore Chinese Health Study, which included a unique ethnic population with strong Western and South Asian cultural influences.
Design: We conducted a prospective data analysis of the Singapore Chinese Health Study, which included 52,584 Chinese men and women (aged 45–74 y) who were free of diabetes, cardiovascular disease (CVD), and cancer at baseline (1993–1998) and followed through 2011 with 10,029 deaths. The following 2 major dietary patterns were identified by using a principal components analysis: a vegetable-, fruit-, and soy-rich (VFS) pattern and a dim sum– and meat-rich (DSM) dietary pattern. Pattern scores for each participant were calculated and examined with all-cause and cause-specific mortality risks by using a Cox proportional hazards regression.
Results: The VFS pattern was inversely associated with all-cause mortality and each cause-specific category (CVD, cancer, and respiratory) of mortality during the follow-up period. Compared with the lowest quintile of the VFS pattern, HRs for quintiles 2–5 for all-cause mortality were 0.90, 0.79, 0.80, and 0.75, respectively (P-trend < 0.0001). The DSM pattern was positively associated with CVD mortality in the whole population (HR for fifth quintile compared with first quintile: 1.23; 95% CI: 1.07, 1.40; P-trend = 0.001). Positive associations between the DSM pattern and cancer and all-cause mortality were only present in ever-smokers. In ever-smokers, relative to the first quintile, HRs for quintiles 2–5 of the DSM pattern for all-cause mortality were 1.04, 1.04, 1.13, and 1.24, respectively (P-trend < 0.0001). Similarly, HRs for quintiles 2–5 for cancer mortality were 1.08, 1.03, 1.25, and 1.34, respectively (P-trend < 0.0001). The DSM pattern was not associated with respiratory mortality.
Conclusion: Dietary patterns are strongly associated with mortality in Chinese Singaporeans.
INTRODUCTION
A broad, growing, and evolving body of scientific inquiry on the association between dietary intake and health and disease has led to a recommendation to eat a dietary pattern that emphasizes a variety of plant-based foods (eg, vegetables, fruit, legumes, whole grains, nuts, and seeds), and deemphasizing processed food products, especially those with added sugars and that are high in meat (particularly processed meats) (1). Indeed, this type of dietary pattern is recognized as a major factor in the prevention and treatment of the major chronic diseases (2). Less described has been the association between diet and longevity.
The research that has examined dietary patterns with mortality has been limited to Western and East Asian populations (eg, Japanese) with little data from Southeast Asian populations (3–11). Generally, these studies have suggested that a culturally appropriate version of the previously mentioned recommended dietary pattern was associated with lower risk of all-cause or cardiovascular disease (CVD)4 mortality. However, most studies were limited by a small sample size, short follow-up time, or the ability to adequately address important potential effect-measure modifiers (eg, education and smoking status).
The Singapore Chinese Health Study (SCHS) provides a unique population to examine dietary patterns and longevity. The culture has both Western and south Asian dietary influences. In addition to the Western dietary influences, mortality patterns of Singapore mirror those of Western countries, providing a unique population within which to study this topic (12). Thus, the goal of this study was to examine the associations between dietary patterns and all-cause mortality and cause-specific mortality in Singaporean Chinese. On the basis of the literature and previous studies from the SCHS, we hypothesized that a dietary pattern that emphasizes plant-based foods would be inversely associated with all-cause mortality and cause-specific mortality during follow-up, and a dietary pattern that deemphasizes plant-based foods and emphasizes more meat and processed foods in the diet would be positively associated with mortality during follow-up. We also hypothesized that smoking status may modify the diet-mortality association as it has for other diet-disease relations in the SCHS and elsewhere (13–16).
SUBJECTS AND METHODS
Study population
The design of the Singapore Chinese Health Study has been previously described (17). Briefly, the cohort was drawn from men and women, aged 45–74 y, who belonged to one of the major dialect groups (Hokkien or Cantonese) of Chinese in Singapore. Between April 1993 and December 1998, 63,257 individuals completed an in-person interview that included questions on usual diet, demographics, height and weight, use of tobacco, usual physical activity, menstrual and reproductive history (women only), medical history, and family history of cancer. Institutional review boards at the National University of Singapore, University of Pittsburgh, and the University of Minnesota approved this study.
Assessment of diet and covariates
A semiquantitative food-frequency questionnaire specifically developed for this population that assessed 165 commonly consumed dishes and food items was administered during the baseline interview to assess usual dietary intake of the previous year. During the interview, the respondent referred to accompanying photographs to select from 8 food-frequency categories (ranging from “never or hardly ever” to “2 or more times a day”) and 3 portion sizes. The food-frequency questionnaire has subsequently been validated against a series of 24-h dietary recall interviews (17) as well as selected biomarker studies (18, 19). In conjunction with this cohort, the Singapore Food Composition Table was developed, which is a food-nutrient database that lists the amounts of 96 nutritive and nonnutritive components per 100 g raw food, cooked food, and beverages in the diet of Singaporean Chinese (17).
At the baseline interview, all lifestyle factors were collected by a self-report. Height and weight were used to calculate BMI (in kg/m2) as weight divided by height squared. An inquiry on smoking habits included smoking status (never, former, or current smoker), and former and current (ie, ever) smokers were further asked for age when they started or quit smoking, the number of cigarettes per day, and the number of years of smoking (20). For physical activity, participants were asked the number of hours per week spent on moderate activities such as brisk walking, bowling, bicycling on level ground, tai chi or chi kung, and strenuous sports such as jogging, bicycling on hills, tennis, squash, swimming laps, or aerobics. The physical activity portion of the questionnaire was modeled after the European Prospective Investigation into Cancer and Nutrition study physical activity questionnaire, which has been shown to be valid and repeatable (21). The usual sleep duration was assessed by using the following question: “On average, during the last year, how many hours in a day did you sleep?” with response categories of ≤5 h, 6, 7, 8, 9, and ≥10 h.
Assessment of mortality
Information on the date and cause of death was obtained through linkage with the nation-wide registry of birth and death in Singapore. Up to 6 different International Classification of Diseases version 9 codes were recorded in the registry. The primary cause of death was used for analysis. The vital status of cohort participants was updated through 31 December 2011. Follow-up for mortality was considered virtually complete because of the linkage analysis and negligible emigration; only 47 subjects from this cohort were known to be lost to follow-up because of migration out of Singapore or for other reasons. Endpoints in our cause-specific analyses were deaths from CVD (codes 394.0–459.0), all cancers (140.0–195.8 and 199–208.9), and respiratory causes [such as pneumonia and influenza (480–488) and chronic obstructive pulmonary disease (490–496)].
Statistical analysis
For the analysis, we excluded 1936 subjects with a history of invasive cancer (except nonmelanoma skin cancer) or superficial, papillary bladder cancer at recruitment because they did not meet study inclusion criteria. We further excluded subjects with a self-reported history of physician-diagnosed diabetes (n = 5469) or CVD (n = 2399) at baseline, plus 869 subjects who reported extreme sex-specific energy intakes (<600 or >3000 kcal for women; <700 or >3700 kcal for men). The analysis included 52,584 participants.
Dietary patterns were derived by using a principal components analysis with SAS software (version 9.2; SAS Institute Inc). All 165 foods and beverages, including alcohol, were first standardized to the same frequency per month unit before the principal components analysis method was applied, and factors were extracted. Factors were rotated orthogonally to maintain an uncorrelated state and improve interpretability, and a 2-factor solution was retained that was based on eigenvalues (>1.0), a scree plot, and factor interpretability. For comparability and interpretability of our results, we present factor loadings ≥0.20 even though values <0.20 were statistically significant because of the large sample size of the study. These variables are in alignment with those used in previous studies (22–26). Factor scores for each participant were calculated by multiplying the intake of the standardized food item by its respective factor loading on each pattern. Scores were linear variables and represented the weighted sum of all 165 food and beverage items. Participants were divided into quintiles by scores to indicate the amount their dietary intake corresponded with each pattern (ie, a higher score corresponded with greater conformity to the derived pattern). Factors were initially extracted by sex, dialect, and smoking status and were highly similar in loading structure and disease prediction to reported whole-cohort factors, and thus, factors derived from the overall cohort were used. Baseline and dietary characteristics were calculated for participants across quintiles of each dietary pattern score. For each study subject, person-years were counted from the date of the baseline interview to the date of death, date of last contact (for the few subjects who migrated out of Singapore), or 31 December 2011, whichever occurred first. HRs per quintile of dietary pattern score were estimated by using Cox proportional hazards regression models with SAS statistical software (version 9.2). There was no evidence that proportional hazards assumptions were violated as indicated by the lack of a significant interaction between dietary pattern scores and a function of the survival time in models. Tests for trends across dietary patterns scores were performed by assigning the median value of the quintile to the respective categories and entering this variable as a continuous variable into the models.
Two models were constructed to examine the association between dietary pattern score and risk of mortality during follow-up as follows: covariates included in model 1 were baseline age (<50, 50–54, 55–59, 60–64, or ≥65 y), year of interview (1993–1995 or 1996–1998), dialect (Hokkiens compared with Cantonese), sex, and education (none, primary, and secondary or above); the second model included smoking (never, light, or heavy) (27), moderate activity (≥2 compared with <2 h/wk), strenuous physical activity (≥1.5 compared with <1.5 h/wk), sleep (6–8 compared with <6 or ≥9 h/d), history of physician-diagnosed hypertension (yes compared with no), total energy intake (kcal/d), and baseline BMI [in kg/m2 as the original BMI and its quadratic term (BMI2)]. Analyses were conducted to test for interactions between diet patterns and age, sex, education, BMI, and smoking. Interactions with smoking and BMI were hypothesized to have a biological basis because of the biological enhancement of overall risk conferred by obesity and cigarette smoking, whereby the age-, sex-, and education-interaction tests were carried out to check for differential results because of varying distributions of dietary patterns and confounding factors. To reduce the potential bias because of nonreported preexisting disease or an underlying illness, participants who died within 3 to <5 y were excluded in sensitivity analyses.
RESULTS
Two main dietary patterns were derived from the principal components analysis. The first was pattern was named vegetable-, fruit-, and soy-rich (VFS). See Supplemental Table 1 under “Supplemental data” in the online issue for factor loadings for this pattern ≥0.20. Foods that loaded highly on this pattern were predominantly vegetables, fruit, and soy-based items. The second pattern was named dim sum– and meat-rich (DSM). See Supplemental Table 1 under “Supplemental data” in the online issue for foods that loaded ≥0.20 on the DSM pattern. A variety of foods, predominantly dim sum, fresh and processed meats and seafood, noodle and rice dishes, sweetened foods, and deep-fried foods, were prominent contributors to the pattern. Most dim sum foods are savory pastries such as steamed or deep-fried dumplings, filled buns, noodles, sweet pastries, and meat dishes.
Baseline characteristics of participants according to dietary pattern scores are shown in Tables 1 and 2. Higher VFS pattern–score participants were, on average, slightly younger, more educated, smoked less, and more active, and a greater proportion were women. Nutritionally, participants with a higher VFS score reported consuming less carbohydrate but more fiber, more dietary fat mainly from plant sources, and more protein. In contrast, participants with higher DSM scores were younger, more educated, smoked more, had less moderate physical activity but greater vigorous activity, and a greater proportion were men. Nutritionally, participants with higher DSM scores reported consuming less carbohydrate with a corresponding decrease in dietary fiber, more dietary fat mainly from animal sources, and more protein.
TABLE 1.
Baseline demographic and lifestyle characteristics according to quintile of dietary pattern score: the SCHS1
| Dietary pattern | Quintile 1 | Quintile 2 | Quintile 3 | Quintile 4 | Quintile 5 | P-trend |
| Vegetable-, fruit-, and soy-rich | ||||||
| n | 10,516 | 10,517 | 10,517 | 10,517 | 10,517 | — |
| Age (y) | 56.4 ± 8.12 | 56.2 ± 8.0 | 55.8 ± 7.8 | 55.4 ± 7.7 | 55.2 ± 7.7 | <0.0001 |
| Sex (F) (%) | 43.3 | 55.6 | 57.8 | 60.7 | 61.8 | <0.0001 |
| Dialect (Hokkien) (%) | 59.8 | 56.1 | 53.1 | 50.3 | 49.0 | <0.0001 |
| Education (%) | 22.6 | 26.1 | 29.6 | 32.5 | 36.6 | <0.0001 |
| Smoking (ever) (%) | 45.0 | 31.8 | 27.7 | 23.0 | 21.3 | <0.0001 |
| Moderate activity (%) | 10.1 | 14.1 | 16.7 | 17.7 | 19.8 | <0.0001 |
| Vigorous activity (%) | 10.9 | 9.1 | 8.9 | 8.9 | 9.7 | 0.018 |
| BMI (kg/m2) | 22.8 ± 3.6 | 23.1 ± 3.6 | 23.0 ± 3.5 | 23.0 ± 3.5 | 23.0 ± 3.5 | 0.28 |
| Sleep (h) | 7.0 ± 1.2 | 7.0 ± 1.1 | 7.0 ± 1.1 | 7.0 ± 1.1 | 7.0 ± 1.1 | 0.25 |
| Dim sum– and meat-rich | ||||||
| n | 10,516 | 10,517 | 10,517 | 10,517 | 10,517 | — |
| Age (y) | 58.3 ± 8.0 | 56.7 ± 7.9 | 55.6 ± 7.7 | 54.8 ± 7.8 | 53.6 ± 7.2 | <0.0001 |
| Sex (F) (%) | 72.4 | 63.4 | 56.3 | 48.5 | 38.8 | <0.0001 |
| Dialect (Hokkien) (%) | 55.3 | 54.4 | 53.3 | 53.4 | 51.9 | <0.0001 |
| Education (%)3 | 22.0 | 25.5 | 28.1 | 32.7 | 39.1 | <0.0001 |
| Smoking (ever) (%) | 17.5 | 25.2 | 29.8 | 35.1 | 41.3 | <0.0001 |
| Moderate activity (%)4 | 18.9 | 16.0 | 15.3 | 14.4 | 13.8 | <0.0001 |
| Vigorous activity (%)5 | 5.4 | 7.0 | 8.9 | 11.4 | 14.9 | <0.0001 |
| BMI (kg/m2) | 22.9 ± 3.6 | 23.0 ± 3.5 | 23.1 ± 3.5 | 23.0 ± 3.5 | 23.0 ± 3.5 | 0.78 |
| Sleep (h) | 6.9 ± 1.1 | 7.0 ± 1.1 | 7.0 ± 1.1 | 7.0 ± 1.1 | 7.0 ± 1.1 | 0.01 |
SCHS, Singapore Chinese Cohort Study.
Mean ± SD (all such values).
Percentage of population with a secondary education or greater.
Moderate activity: ≥2 h/wk (percentage of subjects who reported this level).
Vigorous activity: ≥1.5 h/wk vigorous work or strenuous physical activity (percentage of subjects who reported this level).
TABLE 2.
Dietary intake characteristics according to quintile of dietary pattern score1
| Vegetable, fruit, and soy |
Dim sum and meat |
|||||
| Quintile 1 | Quintile 3 | Quintile 5 | Quintile 1 | Quintile 3 | Quintile 5 | |
| Total energy (kcal/d) | 1373 ± 502 | 1503 ± 463 | 1876 ± 523 | 1245 ± 372 | 1481 ± 403 | 2078 ± 540 |
| Carbohydrate (% of energy) | 61.2 ± 7.9 | 59.5 ± 6.8 | 56.4 ± 6.9 | 63.2 ± 6.8 | 59.4 ± 6.7 | 54.6 ± 6.4 |
| Total fat (% of energy) | 22.4 ± 5.8 | 24.8 ± 5.1 | 28.2 ± 5.0 | 22.4 ± 5.4 | 24.7 ± 5.2 | 28.5 ± 5.0 |
| Saturated fat (% of energy) | 8.3 ± 2.6 | 8.8 ± 2.4 | 9.6 ± 2.5 | 7.6 ± 2.4 | 8.7 ± 2.3 | 10.5 ± 2.3 |
| Monounsaturated fat (% of energy) | 7.7 ± 2.1 | 8.4 ± 1.9 | 9.4 ± 1.9 | 7.4 ± 1.9 | 8.4 ± 1.9 | 9.7 ± 1.9 |
| Polyunsaturated fat (% of energy) | 4.1 ± 1.4 | 5.0 ± 1.6 | 6.1 ± 2.1 | 5.0 ± 2.1 | 5.0 ± 1.8 | 5.3 ± 1.6 |
| Protein (% of energy) | 14.3 ± 2.5 | 15.2 ± 2.3 | 15.8 ± 2.4 | 14.5 ± 2.6 | 15.2 ± 2.4 | 15.7 ± 2.2 |
| Soy protein (% of energy) | 1.0 ± 0.7 | 1.4 ± 0.8 | 2.1 ± 1.2 | 1.4 ± 1.1 | 1.5 ± 1.0 | 1.6 ± 0.9 |
| Fiber (g/1000 kcal) | 6.1 ± 1.9 | 8.2 ± 2.2 | 10.1 ± 2.6 | 9.4 ± 3.0 | 8.0 ± 2.5 | 7.4 ± 2.1 |
| Animal fat (g/1000 kcal)2 | 9.0 ± 3.8 | 8.4 ± 3.3 | 8.1 ± 3.2 | 5.9 ± 2.8 | 8.4 ± 2.8 | 11.2 ± 3.1 |
| Plant fat (g/1000 kcal)3 | 13.4 ± 3.8 | 16.5 ± 3.7 | 20.1 ± 4.2 | 16.5 ± 4.8 | 16.4 ± 4.4 | 17.3 ± 4.1 |
| Rice (g/1000 kcal) | 318.9 ± 104.3 | 277.3 ± 86.9 | 216.6 ± 78.0 | 306.2 ± 101.7 | 276.5 ± 91.8 | 226.7 ± 79.1 |
| Noodles (g/1000 kcal) | 38.2 ± 30.1 | 33.3 ± 24.7 | 31.6 ± 22.7 | 22.1 ± 21.8 | 36.8 ± 26.6 | 41.5 ± 24.3 |
| Vegetables (g/1000 kcal) | 45.3 ± 19.7 | 61.0 ± 22.3 | 104.7 ± 40.2 | 83.3 ± 42.5 | 71.6 ± 33.2 | 65.0 ± 27.7 |
| Fruit and related juices (g/1000 kcal) | 75.2 ± 69.3 | 132.1 ± 87.7 | 183.9 ± 110.9 | 147.1 ± 109.7 | 129.2 ± 96.9 | 122.2 ± 85.3 |
| Soy foods/beverages (g/1000 kcal) | 47.0 ± 37.7 | 68.4 ± 43.2 | 99.8 ± 59.3 | 66.6 ± 52.5 | 71.4 ± 50.4 | 74.5 ± 46.4 |
| Red meat (g/1000 kcal) | 20.9 ± 12.1 | 18.7 ± 10.7 | 17.2 ± 10.6 | 11.9 ± 9.2 | 18.9 ± 9.8 | 26.2 ± 11.2 |
| Poultry (g/1000 kcal) | 12.8 ± 9.8 | 12.8 ± 9.4 | 12.7 ± 10.0 | 7.7 ± 7.6 | 12.8 ± 9.0 | 17.1 ± 10.0 |
| Fish (g/1000 kcal) | 32.1 ± 17.5 | 36.2 ± 16.7 | 37.8 ± 18.5 | 35.4 ± 20.1 | 36.8 ± 17.4 | 34.8 ± 14.5 |
| Dairy products (g/1000 kcal) | 31.9 ± 62.5 | 46.3 ± 74.3 | 55.1 ± 72.7 | 67.0 ± 99.0 | 41.2 ± 66.7 | 33.2 ± 47.1 |
| Cooking fats/oils (g/1000 kcal) | 8.5 ± 3.0 | 10.5 ± 3.0 | 13.2 ± 3.6 | 10.3 ± 3.9 | 10.7 ± 3.5 | 11.1 ± 3.1 |
| Coffee (cups/mo) | 46.5 ± 40.8 | 34.8 ± 34.3 | 29.8 ± 32.1 | 23.0 ± 26.8 | 37.6 ± 35.8 | 45.6 ± 40.2 |
| Black tea (cups/mo) | 7.3 ± 18.5 | 6.4 ± 15.7 | 8.0 ± 17.4 | 3.7 ± 11.4 | 6.8 ± 16.2 | 11.0 ± 21.3 |
| Green tea (cups/mo) | 6.3 ± 20.3 | 9.3 ± 24.2 | 12.3 ± 26.9 | 7.1 ± 21.1 | 9.4 ± 24.6 | 11.2 ± 26.2 |
| Soft drinks (drinks/mo) | 4.4 ± 12.9 | 2.2 ± 7.7 | 1.9 ± 6.7 | 0.4 ± 2.0 | 1.9 ± 6.5 | 6.8 ± 15.2 |
| Alcohol (drinks/wk) | 1.9 ± 6.2 | 0.7 ± 3.3 | 0.7 ± 3.1 | 0.2 ± 1.1 | 0.8 ± 3.6 | 2.2 ± 6.3 |
All values are means ± SDs. All P-trend < 0.0001 across all groups except for poultry in the vegetable-, fruit-, and soy-rich pattern (P = 0.76).
Dietary fat from animal-based foods.
Dietary fat from plant-based foods.
HRs for all-cause mortality and cause-specific mortality according to each pattern score are shown in Table 3. Overall, during ∼793,948 person-years of follow-up there were 10,029 deaths. There was a strong, inverse association between a higher VFS dietary pattern score and all-cause mortality and each cause-specific category of mortality after adjustment for demographic and lifestyle confounders. The strongest, graded association was observed with CVD mortality (quintiles 5 compared with 1) whereby the HR (95% CI) was 0.63 (0.56, 0.72). For the DSM pattern, a significant positive association was observed for all-cause mortality and CVD mortality with a suggestive positive association with mortality that was due to cancer. The DSM pattern was not associated with increased risk of mortality because of respiratory causes during the follow-up period. There was no evidence that any of the associations between VFS and DSM dietary pattern scores and all-cause– and cause-specific–mortality categories differed by age, sex, education, or BMI. Furthermore, there was no evidence that an association with the VFS pattern differed by smoking status with any mortality outcome.
TABLE 3.
HRs (95% CIs) of all-cause and cause-specific mortality according to dietary pattern score in Chinese men and women1
| Type | Quintile 1 | Quintile 2 | Quintile 3 | Quintile 4 | Quintile 5 | P-trend |
| Vegetable-, fruit-, and soy-rich pattern | ||||||
| All cause | ||||||
| No. of deaths/PY | 2676/153,029 | 2233/158,657 | 1867/160,339 | 1719/160,892 | 1534/161,031 | — |
| HR (95% CI)2 | 1.00 | 0.86 (0.81, 0.91) | 0.75 (0.71, 0.80) | 0.75 (0.70, 0.80) | 0.69 (0.65, 0.74) | <0.0001 |
| HR (95% CI)3 | 1.00 | 0.90 (0.84, 0.94) | 0.79 (0.74, 0.84) | 0.80 (0.75, 0.85) | 0.75 (0.70, 0.80) | <0.0001 |
| Cancer | ||||||
| No. of deaths/PY | 973/153,029 | 874/158,657 | 732/160,339 | 698/160,892 | 625/161,031 | — |
| HR (95% CI)2 | 1.00 | 0.93 (0.84, 1.02) | 0.81 (0.73, 0.89) | 0.83 (0.75, 0.91) | 0.77 (0.70, 0.85) | <0.0001 |
| HR (95% CI)3 | 1.00 | 0.97 (0.89, 1.06) | 0.86 (0.78, 0.95) | 0.90 (0.81, 1.00) | 0.84 (0.75, 0.94) | 0.0009 |
| CVD | ||||||
| No. of deaths/PY | 869/153,029 | 706/158,657 | 563/160,339 | 524/160,892 | 435/161,031 | — |
| HR (95% CI)2 | 1.00 | 0.83 (0.75, 0.92) | 0.70 (0.63, 0.78) | 0.70 (0.63, 0.79) | 0.61 (0.54, 0.68) | <0.0001 |
| HR (95% CI)3 | 1.00 | 0.85 (0.77, 0.94) | 0.72 (0.64, 0.80) | 0.73 (0.65, 0.82) | 0.63 (0.56, 0.72) | <0.0001 |
| Respiratory | ||||||
| No. of deaths/PY | 489/153,029 | 338/158,657 | 308/160,339 | 266/160,892 | 235/161,031 | — |
| HR (95% CI)2 | 1.00 | 0.72 (0.62, 0.82) | 0.69 (0.60, 0.80) | 0.66 (0.57, 0.77) | 0.61 (0.52, 0.71) | <0.0001 |
| HR (95% CI)3 | 1.00 | 0.76 (0.66, 0.87) | 0.75 (0.65, 0.87) | 0.73 (0.62, 0.86) | 0.68 (0.58, 0.81) | <0.0001 |
| Dim sum– and meat-rich pattern | ||||||
| All cause | ||||||
| No. of deaths/PY | 2246/158,384 | 2076/159,536 | 1985/158,934 | 1917/158,629 | 1805/158,465 | — |
| HR (95% CI)2 | 1.00 | 1.00 (0.94, 1.06) | 1.04 (0.98, 1.10) | 1.08 (1.01, 1.15) | 1.11 (1.04, 1.19) | 0.0001 |
| HR (95% CI)3 | 1.00 | 0.98 (0.92, 1.04) | 1.01 (0.95, 1.08) | 1.06 (0.99, 1.13) | 1.14 (1.06, 1.23) | <0.0001 |
| Cancer | ||||||
| No. of deaths/PY | 829/158,384 | 771/159,536 | 757/158,934 | 793/158,629 | 752/158,465 | — |
| HR (95% CI)2 | 1.00 | 0.99 (0.90, 1.09) | 1.04 (0.94, 1.14) | 1.15 (1.04, 1.27) | 1.17 (1.06, 1.30) | 0.0001 |
| HR (95% CI)3 | 1.00 | 0.95 (0.86, 1.05) | 0.98 (0.89, 1.09) | 1.08 (0.97, 1.19) | 1.12 (0.99, 1.26) | 0.013 |
| CVD | ||||||
| No. of deaths/PY | 718/158,384 | 650/159,536 | 621/158,934 | 570/158,629 | 538/158,465 | — |
| HR (95% CI)2 | 1.00 | 1.00 (0.90, 1.11) | 1.05 (0.94, 1.17) | 1.05 (0.93, 1.17) | 1.09 (0.97, 1.23) | 0.09 |
| HR (95% CI)3 | 1.00 | 0.99 (0.89, 1.10) | 1.06 (0.95, 1.18) | 1.09 (0.97, 1.23) | 1.23 (1.07, 1.40) | 0.001 |
| Respiratory | ||||||
| No. of deaths/PY | 376/158,384 | 372/159,536 | 331/158,934 | 269/158,629 | 288/158,465 | — |
| HR (95% CI)2 | 1.00 | 1.08 (0.93, 1.24) | 1.06 (0.92, 1.23) | 0.95 (0.81, 1.11) | 1.15 (0.98, 1.35) | 0.28 |
| HR (95% CI)3 | 1.00 | 1.03 (0.89, 1.19) | 1.01 (0.87, 1.17) | 0.90 (0.76, 1.06) | 1.15 (0.95, 1.38) | 0.37 |
Definition of outcomes: CVD (International Classification of Diseases version 9 codes: 394–459; n = 2626), cancer (International Classification of Diseases version 9 codes: 140–239; n = 3279); respiratory (eg, influenza, pneumonia, and chronic obstructive pulmonary disease; International Classification of Diseases version 9 codes: 480–488 and 490–496; n = 1289). CVD, cardiovascular disease; PY, person-years follow-up.
Model 1 was adjusted for age, sex, dialect, education, and year of interview.
Model 2 was adjusted as for model 1 and for smoking, moderate and vigorous activity, sleep, BMI, history of hypertension (except for cancer), and energy intake.
However, there was evidence for a multiplicative interaction between the DSM pattern and smoking for cancer mortality (P = 0.018). This finding drove similar results in all-cause mortality whereby there was evidence that the association between the DSM pattern and risk of all-cause mortality during follow-up also differed by smoking status (P = 0.05). These results are presented in Table 4. In never-smokers, there was no association between the DSM pattern and all-cause or cancer mortality, whereas in ever-smokers, there was a strong, graded, positive association (P < 0.0001) for both cancer and all-cause mortality. Last, there was no evidence any that associations differed on the exclusion of deaths ≤5 y after enrollment.
TABLE 4.
HRs (95% CIs) of DSM pattern with cancer and all-cause mortality according to smoking status in Chinese men and women: the SCHS1
| Quintile 1 | Quintile 2 | Quintile 3 | Quintile 4 | Quintile 5 | P-trend | |
| All cause | ||||||
| Never-smokers | ||||||
| No. of deaths/n | 1504/8679 | 1104/7865 | 964/7385 | 787/6826 | 636/6170 | — |
| HR (95% CI) | 1.00 | 0.94 (0.86, 1.01) | 1.02 (0.93, 1.10) | 1.01 (0.92, 1.11) | 1.04 (0.93, 1.16) | 0.27 |
| Ever-smokers | ||||||
| No. of deaths/n | 742/1837 | 972/2652 | 1021/3132 | 1130/3691 | 1169/4347 | — |
| HR (95% CI) | 1.00 | 1.04 (0.95, 1.15) | 1.04 (0.94, 1.14) | 1.13 (1.03, 1.25) | 1.24 (1.12, 1.38) | <0.0001 |
| Cancer | ||||||
| Never-smokers | ||||||
| No. of deaths/n | 559/8679 | 399/7865 | 372/7385 | 307/6826 | 245/6170 | — |
| HR (95% CI) | 1.00 | 0.88 (0.77, 1.00) | 0.97 (0.85, 1.11) | 0.94 (0.81, 1.09) | 0.93 (0.78, 1.11) | 0.59 |
| Ever-smokers | ||||||
| No. of deaths/n | 270/1837 | 372/2652 | 385/3132 | 486/3691 | 507/4347 | — |
| HR (95% CI) | 1.00 | 1.08 (0.92, 1.27) | 1.03 (0.88, 1.21) | 1.25 (1.07, 1.46) | 1.34 (1.13, 1.59) | <0.0001 |
Model was adjusted for age, sex, dialect, education, year of interview, moderate and vigorous activity, sleep, BMI, history of hypertension (all-cause mortality), and energy intake. For DSM diet pattern × smoking for cancer mortality, P-interaction = 0.018; for DSM diet pattern × smoking for all-cause mortality, P-interaction = 0.05. DSM, dim sum– and meat-rich; SCHS, Singapore Chinese Cohort Study.
DISCUSSION
In this large study of Chinese Singaporeans, a higher VFS-pattern score displayed a strong, inverse association with risk of all-cause mortality and cause-specific (CVD, cancer, and respiratory) mortality during the follow-up period. A higher score on the other main dietary pattern in the population, the DSM pattern, was positively associated with CVD mortality. Furthermore, participants with a history of smoking and a higher DSM-pattern score had a strong, positive association with cancer and all-cause mortality, whereas there was no association in participants who never smoked with these outcomes.
The composition of the VFS dietary pattern and its association with mortality has similarities with results in published literature on the topic. The unifying theme with Western populations and their dietary patterns that were associated with reduced risk of all-cause mortality was an increased emphasis on vegetables, fruit, and grains (especially whole grains) with other main components that likely vary because of cultural influence and include legumes, nut and seeds, dairy, alcohol, vegetable oils, and fish (3–7). The VFS pattern was inversely associated with each cause-specific category of mortality in the SCHS, whereas the other studies that examined cause-specific categories of mortality were less consistent. Only one other study showed reduced risk with cancer (6) and “other causes,” which may have included some respiratory causes similar to the respiratory-cause group in the SCHS (3). Two other studies that examined CVD mortality observed inverse associations (3, 6). Studies that were based in Japanese populations focused on CVD mortality, and patterns that emphasized vegetables, fish, dairy, and soy were inversely associated with CVD mortality risk (8, 9). The study in Chinese women observed an inverse association between a higher fruit-rich (also vegetable) dietary pattern and total mortality (10), whereas there was no association in the population from Bangladesh (11).
Few of these studies reported on a corresponding less-healthy dietary pattern similar to the DSM pattern in the SCHS. The unifying theme in studies that did report on this type of pattern and the SCHS DSM pattern was an emphasis on meat and refined, sweetened, and fried foods. In Western populations, there was no association in 2 studies (5, 7), no report in 2 studies (4, 6), and strong, positive association between a higher Western pattern score and all-cause mortality and each cause-specific category in the Nurses’ Health study (3). In Japanese populations, this pattern was positively associated with CVD mortality in one study (8) but not in another study (9). Similarly, there was no association between this type of pattern and mortality in Chinese women (10). Note, compared with the current study, other studies that were based in Asian populations all used less-extensive dietary assessments and had shorter follow-up times. Other than the Nurses’ Health study (3), none of the studies included extensive dietary and confounder assessments, nor did they have large sample sizes or long-term follow up. Results relating to the DSM pattern from the SCHS were similar to the results from the Western pattern from the Nurses’ Health study except that we observed that findings for cancer depended on smoking status, which also affected point estimates for all-cause mortality.
The overall nutritional profile of the 2 respective patterns and the interrelation between the bioactive constituents of the numerous foods is the hypothesized rationale for the observed associations (28). Indeed, the overall dietary pattern may be the most fundamental and germane unit of measurement of diet and health (28). The finding that the association between a dietary pattern with a poorer nutritional profile and cancer mortality is heightened in smokers is plausible because of the pleiotropic effects smoking and dietary intake have on biological systems of humans (29, 30). Broadly, smoking alters the metabolism, thereby creating dysfunction in cellular activity and a procarcinogenesis state (29). This state would plausibly create the environment in which a poor dietary pattern may have a carcinogenic effect because of the limited ability of the biological system to adapt and maintain homeostasis in the presence of excess metabolic perturbations generated by a poor diet (16). In addition, results of previous analyses that examined the patterns (VFS and DSM) from the SCHS cohort with different chronic-disease outcomes also support the findings (13, 31–33). Last, to our knowledge, the results related to respiratory mortality are a novel contribution to the literature. Because diet is linked to overall respiratory function, pneumonia, and chronic obstructive pulmonary disease (34, 35), we hypothesized that these mortality outcomes may have been relevant to our cause-specific analysis, especially because diet in the SCHS has been linked to nonmalignant respiratory symptoms (32).
Strengths of the current study included the large, Chinese population with a unique overall diet and ample events and follow-up time. Another particular strength was the use of a food-frequency questionnaire that was specifically developed and validated in this population. Other strengths included the high participant response rate, detailed collection of data through face-to-face interviews, thorough assessment of potential lifestyle and demographic confounders, very low level of loss of participants to follow-up, and nearly complete mortality assessment with objectively obtained records on time and cause of death. Limitations included some amount of measurement error with the dietary assessment, although this limitation would most likely have resulted in a nondifferential misclassification with respect to mortality and a likely underestimation of risk. The self-report of other lifestyle-related data may have also resulted in some misclassification and residual confounding in our models. A repeated assessment of dietary intake and other lifestyle factors would have allowed us to examine changes in respect to mortality.
In conclusion, dietary patterns are unique to the populations they are derived from, but consistencies across populations and cultures suggest that patterns with higher intake of plant-based foods, such as vegetables, fruit, soy and other legumes, whole grains, and nuts and seeds are associated with increased longevity. In contrast, patterns with higher intakes of meat and processed meat and sweetened, fried, and refined foods are associated with decreased longevity. Results from this study highlight these patterns in a unique population comprised of Chinese in Southeast Asia, and they also suggest smoking may be an important consideration when examining diet because of its influence on biological pathways relevant to health.
Supplementary Material
Acknowledgments
We thank Siew-Hong Low of the National University of Singapore for supervising the field work of the Singapore Chinese Health Study, Kazuko Arakawa and Renwei Wang for the development and maintenance of the cohort study database, the Ministry of Health in Singapore for assistance with the identification of cancer cases and mortality via database linkages, and Mimi C Yu, who is the founding, long-standing principal investigator of the Singapore Chinese Health Study.
The authors’ responsibilities were as follows—AOO: designed the research plan, analyzed data, and wrote the manuscript; AOO and MAP: were responsible for the final content of the manuscript; W-PK, J-MY, MDG, and MAP: contributed to the research plan and edited the manuscript for content; and all authors: read and approved the final manuscript. None of the authors had a conflict of interest.
Footnotes
Abbreviations used: CVD, cardiovascular disease; DSM, dim sum– and meat-rich; SCHS, Singapore Chinese Health Study; VFS, vegetable-, fruit-, and soy-rich.
REFERENCES
- 1.USDA, US Department of Health and Human Services. Dietary Guidelines for Americans, 2010. 7th ed. Washington, DC: US Government Printing Office, 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Eyre H, Kahn R, Robertson RM, Clark NG, Doyle C, Hong Y, Gansler T, Glynn T, Smith RA, Taubert K, et al. Preventing cancer, cardiovascular disease, and diabetes: a common agenda for the American Cancer Society, the American Diabetes Association, and the American Heart Association. Circulation 2004;109:3244–55. [DOI] [PubMed] [Google Scholar]
- 3.Heidemann C, Schulze MB, Franco OH, van Dam RM, Mantzoros CS, Hu FB. Dietary patterns and risk of mortality from cardiovascular disease, cancer, and all causes in a prospective cohort of women. Circulation 2008;118:230–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bamia C, Trichopoulos D, Ferrari P, Overvad K, Bjerregaard L, Tjønneland A, Halkjaer J, Clavel-Chapelon F, Kesse E, Boutron-Ruault MC, et al. Dietary patterns and survival of older Europeans: the EPIC-Elderly Study (European Prospective Investigation into Cancer and Nutrition). Public Health Nutr 2007;10:590–8. [DOI] [PubMed] [Google Scholar]
- 5.Osler M, Heitmann BL, Gerdes LU, Jørgensen LM, Schroll M. Dietary patterns and mortality in Danish men and women: a prospective observational study. Br J Nutr 2001;85:219–25. [DOI] [PubMed] [Google Scholar]
- 6.Menotti A, Alberti-Fidanza A, Fidanza F, Lanti M, Fruttini D. Factor analysis in the identification of dietary patterns and their predictive role in morbid and fatal events. Public Health Nutr 2012;15:1232–9. [DOI] [PubMed] [Google Scholar]
- 7.Waijers PM, Ocké MC, van Rossum CT, Peeters PH, Bamia C, Chloptsios Y, van der Schouw YT, Slimani N, Bueno-de-Mesquita HB. Dietary patterns and survival in older Dutch women. Am J Clin Nutr 2006;83:1170–6. [DOI] [PubMed] [Google Scholar]
- 8.Shimazu T, Kuriyama S, Hozawa A, Ohmori K, Sato Y, Nakaya N, Nishino Y, Tsubono Y, Tsuji I. Dietary patterns and cardiovascular disease mortality in Japan: a prospective cohort study. Int J Epidemiol 2007;36:600–9. [DOI] [PubMed] [Google Scholar]
- 9.Maruyama K, Iso H, Date C, Kikuchi S, Watanabe Y, Wada Y, Inaba Y, Tamakoshi A; JACC Study Group. Dietary patterns and risk of cardiovascular deaths among middle-aged Japanese: JACC Study. Nutr Metab Cardiovasc Dis 2013;23:519–27. [DOI] [PubMed] [Google Scholar]
- 10.Cai H, Shu XO, Gao YT, Li H, Yang G, Zheng W. A prospective study of dietary patterns and mortality in Chinese women. Epidemiology 2007;18:393–401. [DOI] [PubMed] [Google Scholar]
- 11.Chen Y, McClintock TR, Segers S, Parvez F, Islam T, Ahmed A, Rakibuz-Zaman M, Hasan R, Sarwar G, Ahsan H. Prospective investigation of major dietary patterns and risk of cardiovascular mortality in Bangladesh. Int J Cardiol 2013;167:1495–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Odegaard AO, Koh WP, Yuan JM, Gross MD, Pereira MA. Western-style fast food intake and cardiometabolic risk in an Eastern country. Circulation 2012;126:182–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Odegaard AO, Koh WP, Butler LM, Duval S, Gross MD, Yu MC, Yuan JM, Pereira MA. Dietary patterns and incident type 2 diabetes in Chinese men and women: The Singapore Chinese Health Study. Diabetes Care 2011;34:880–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Clark ML, Butler LM, Koh WP, Wang R, Yuan JM. Dietary fiber intake modifies the association between secondhand smoke exposure and coronary heart disease mortality among Chinese non-smokers in Singapore. Nutrition 2013;29:1304–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hozawa A, Jacobs DR, Jr, Steffes MW, Gross MD, Steffen LM, Lee DH. Associations of serum carotenoid concentrations with the development of diabetes and with insulin concentration: interaction with smoking: the Coronary Artery Risk Development in Young Adults (CARDIA) Study. Am J Epidemiol 2006;163:929–37. [DOI] [PubMed] [Google Scholar]
- 16.Vardavas C I, Flouris AD, Tsatsakis A, Kafatos AG, Saris WH. Does adherence to the Mediterranean diet have a protective effect against active and passive smoking? Public Health 2011;125:121–8. [DOI] [PubMed] [Google Scholar]
- 17.Hankin JH, Stram DO, Arakawa K, Park S, Low SH, Lee HP, Yu MC. Singapore Chinese Health Study: development, validation, and calibration of the quantitative food frequency questionnaire. Nutr Cancer 2001;39:187–95. [DOI] [PubMed] [Google Scholar]
- 18.Seow A, Shi CY, Chung FL, Jiao D, Hankin JH, Lee HP, Coetzee GA, Yu MC. Urinary total isothiocyanate (ITC) in a population-based sample of middle-aged and older Chinese in Singapore: relationship with dietary total ITC and glutathione S-transferase M1/T1/P1 genotypes. Cancer Epidemiol Biomarkers Prev 1998;7:775–81. [PubMed] [Google Scholar]
- 19.Seow A, Shi CY, Franke AA, Hankin JH, Lee HP, Yu MC. Isoflavonoid levels in spot urine are associated with frequency of dietary soy intake in a population-based sample of middle-aged and older Chinese in Singapore. Cancer Epidemiol Biomarkers Prev 1998;7:135–40. [PubMed] [Google Scholar]
- 20.Koh WP, Yuan JM, Sun CL, Lee HP, Yu MC. Middle-aged and older Chinese men and women in Singapore who smoke have less healthy diets and lifestyles than nonsmokers. J Nutr 2005;135:2473–7. [DOI] [PubMed] [Google Scholar]
- 21.Cust AE, Smith BJ, Chau J, van der Ploeg HP, Friedenreich CM, Armstrong BK, Bauman A. Validity and repeatability of the EPIC physical activity questionnaire: a validation study using accelerometers as an objective measure. Int J Behav Nutr Phys Act 2008;5:33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nettleton JA, Steffen LM, Ni H, Liu K, Jacobs DR., Jr Dietary patterns and risk of incident type 2 diabetes in the Multi-Ethnic Study of Atherosclerosis (MESA). Diabetes Care 2008;31:1777–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Hodge AM, English DR, O'Dea K, Giles GG. Dietary patterns and diabetes incidence in the melbourne collaborative cohort study. Am J Epidemiol 2007;165:603–10. [DOI] [PubMed] [Google Scholar]
- 24.Montonen J, Knekt P, Härkänen T, Järvinen R, Heliövaara M, Aromaa A, Reunanen A. Dietary patterns and the incidence of type 2 diabetes. Am J Epidemiol 2005;161:219–27. [DOI] [PubMed] [Google Scholar]
- 25.Fung TT, Schulze M, Manson JE, Willett WC, Hu FB. Dietary patterns, meat intake, and the risk of type 2 diabetes in women. Arch Intern Med 2004;164:2235–40. [DOI] [PubMed] [Google Scholar]
- 26.van Dam RM, Rimm EB, Willett WC, Stampfer MJ, Hu FB. Dietary patterns and risk for type 2 diabetes mellitus in U.S. men. Ann Intern Med 2002;136:201–9. [DOI] [PubMed] [Google Scholar]
- 27.Tsong WH, Koh WP, Yuan JM, Wang R, Sun CL, Yu MC. Cigarettes and alcohol in relation to colorectal cancer: the Singapore Chinese Health Study. Br J Cancer 2007;96:821–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Jacobs DR, Jr, Gross MD, Tapsell LC. Food synergy: an operational concept for understanding nutrition. Am J Clin Nutr 2009;89:1543S–8S. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. How tobacco smoke causes disease: the biology and behavioral basis for smoking-attributable disease: a report of the Surgeon General. In: US Department of Health and Human Services, ed. Rockville, MD: US Department of Health and Human Service, Public Health Service, Office of the Surgeon General, 2010..
- 30.Ross SA. Evidence for the relationship between diet and cancer. Exp Oncol 2010;32:137–42. [PubMed] [Google Scholar]
- 31.Odegaard AO, Koh WP, Yuan JM. Combined lifestyle factors and risk of incident colorectal cancer in a Chinese population. Cancer Prev Res (Phila) 2013;6:360–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Butler LM, Koh WP, Lee HP, Tseng M, Yu MC, London SJ. Prospective study of dietary patterns and persistent cough with phlegm among Chinese Singaporeans. Am J Respir Crit Care Med 2006;173:264–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Butler LM, Wu AH, Wang R, Koh WP, Yuan JM, Yu MC. A vegetable-fruit-soy dietary pattern protects against breast cancer among postmenopausal Singapore Chinese women. Am J Clin Nutr 2010;91:1013–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Hanson C, Rutten EP, Wouters EF, Rennard S. Diet and vitamin D as risk factors for lung impairment and COPD. Transl Res 2013;162:219–36. [DOI] [PubMed] [Google Scholar]
- 35.Romieu I, Trenga C. Diet and obstructive lung diseases. Epidemiol Rev 2001;23:268–87. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.