Soy Isoflavone Intake and Sleep Parameters over 5 Years among Chinese Adults: Longitudinal Analysis from the Jiangsu Nutrition Study

Soy Isoflavone Intake and Sleep Parameters over 5 Years among Chinese Adults: Longitudinal Analysis from the Jiangsu Nutrition Study

RESEARCH Original Research Soy Isoflavone Intake and Sleep Parameters over 5 Years among Chinese Adults: Longitudinal Analysis from the Jiangsu Nutri...

275KB Sizes 0 Downloads 35 Views

RESEARCH

Original Research

Soy Isoflavone Intake and Sleep Parameters over 5 Years among Chinese Adults: Longitudinal Analysis from the Jiangsu Nutrition Study Yingting Cao, MMSc; Anne W. Taylor, PhD; Shiqi Zhen, MD; Robert Adams, MD, PhD; Sarah Appleton, PhD; Zumin Shi, MD, PhD ARTICLE INFORMATION Article history: Submitted 3 May 2016 Accepted 12 October 2016 Available online 10 December 2016

Keywords: Longitudinal Soy isoflavone Long sleep duration Daytime falling asleep Chinese adults

Supplementary materials: Tables 5 and 6 are available at www.andjrnl.org 2212-2672/Copyright ª 2017 by the Academy of Nutrition and Dietetics. http://dx.doi.org/10.1016/j.jand.2016.10.016

ABSTRACT Background Soy isoflavone is beneficial for menopausal/postmenopausal symptoms, including sleep complaints. However, little is known about its longitudinal association with sleep in the general population. Objective Our aim was to investigate the association between soy isoflavone intake and sleep duration and daytime falling asleep among Chinese adults. Design A longitudinal analysis was performed. Soy isoflavone intake was assessed by food frequency questionnaire. Sleep duration was self-reported at two time points. Occurrence of daytime falling asleep was determined at follow-up. Short and long sleep were defined as sleep <7 h/day or 9 h/day, respectively. Participants/setting Adults aged 20 years and older from the Jiangsu Nutrition Study (2002-2007) with complete isoflavone intake and sleep duration data at both time points (n¼1,474) were analyzed (follow-up, n¼1,492). Main outcome measures We measured sleep duration in 2002 and 2007 and daytime falling asleep occurrence in 2007. Statistical analyses performed Mixed-effects logistic regression was performed for repeated measures between isoflavone intake and sleep duration. Logistic regression was performed for daytime falling asleep at follow-up. Demographic, anthropometric, and social factors were adjusted in the analyses. Results The prevalence of long sleep duration was 18.9% in 2002 and 12.6% in 2007, and the prevalence of daytime falling asleep was 5.3%. Compared with the lowest quartile of isoflavone intake, the highest quartile was associated with a lower risk of long sleep duration (odds ratio¼0.66; 95% CI 0.48 to 0.90; P for trend¼0.018) over 5 years. Compared with persistent low intake of isoflavone (less than median intake of isoflavone at two time points), persistent high intake was associated with a reduced risk of daytime falling asleep in women (odds ratio¼0.20; 95% CI 0.06 to 0.68), but not men. No consistent association between soy isoflavone intake and short sleep duration was found. Conclusions Soy isoflavone intake was associated with a low risk of long sleep duration in both sexes and a low risk of daytime falling asleep in women but not men. J Acad Nutr Diet. 2017;117:536-544.

A

LARGE NUMBER OF STUDIES HAVE SUGGESTED associations between short sleep duration and various health outcomes, including cardiovascular diseases and metabolic syndrome.1,2 Relatively less attention has been paid to long sleep duration and its health consequences. Several studies have demonstrated a U-shaped relationship between sleep duration and type 2 diabetes,3 as well as all-cause mortality.4 These studies emphasize the adverse effect of long sleep duration on health. Although the exact mechanism of the association between long sleep duration and mortality remains unclear, depression and low socioeconomic status have been suggested as potential factors.5 Self-reported sleep symptoms, including snoring and daytime sleepiness accompanied by altered sleep duration, have also been found to be associated 536

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

with increased risk of cardiovascular diseases in older adults.6 Estrogen plays a role in the central nervous system involved in sleep regulation.7 A randomized trial has confirmed the effects of estradiol on sleep quality among perimenopausal and postmenopausal women.8 Isoflavones, a class of phytoestrogens found mostly in soybeans and legumes, have been indicated as potential alternative therapies for hormone-dependent conditions, including cancer, cardiovascular diseases, and postmenopausal symptoms.9 Studies on the association between isoflavone intake and sleep are scarce. Two clinical trials have demonstrated the beneficial effects of isoflavone treatment on sleep disorders in postmenopausal women.10,11 The mechanism may be partly explained by the relieved vasomotor symptoms, as sleep ª 2017 by the Academy of Nutrition and Dietetics.

RESEARCH complaints (insomniac symptoms) are common among perimenopausal and postmenopausal women.12,13 Recently, a Japanese cross-sectional study investigated the association between isoflavone intake and sleep in the general population, trying to explore the association beyond the laboratory and perimenopausal and postmenopausal women-only settings.14 In that study, higher daily isoflavone intake (18 mg/ 1,000 kcal/day) was associated with optimal sleep duration (7 to 8 h/day) and sleep quality. However, due to the nature of cross-sectional studies, causal effects cannot be examined. Like many other Asian countries, China has a high intake of soy, with adequate variation within the population. This makes it possible to study the effects of soy isoflavone on health at the population level. The present study aimed to investigate the longitudinal association between isoflavone intake and sleep duration among the Chinese population during a 5-year period. Another aim was to examine the association between isoflavone intake at baseline and selfreported daytime falling asleep at follow-up.

METHODS Participants Data from the Jiangsu Nutrition Study cohort with participants older than 20 years were used. A detailed methodology has been described previously.15 In 2002, a total of 2,849 adults at least 20 years of age living in two cities and six counties in Jiangsu Province took part in the Chinese National Nutrition and Health Survey. A multistage cluster sampling method was used to select participants. Rural samples were chosen from three randomly selected small towns within each of the six counties. Urban samples were chosen from three randomly selected streets within each of the two cities. In each town/street, two villages/neighborhoods were randomly selected, from which 90 households were then randomly selected.15 In 2007, an attempt to recontact all of the original participants was made. Due to attrition, 1,682 were identified for follow-up and 1,492 participated in the follow-up interview. For the current analysis, we included only those participants with sleep duration records in both 2002 and 2007 (n¼1,474). Compared to the original sample with 2,849 participants in 2002, the sample included in the present study was older, with lower body mass index (BMI; calculated as kg/m2) and smaller waist circumference, but there was no difference in sex and energy intake. The study was conducted according to the guidelines in the Declaration of Helsinki, and all procedures were approved by the Jiangsu Provincial Centre for Disease Control and Prevention. Written informed consent for participation was obtained from each participant.

Data Collection and Measurements Participants were interviewed at their homes by trained health workers using a precoded questionnaire. Interviews took about 2 hours to complete and included questions on diet, sociodemographic information, medical history, cigarette smoking, physical activity, and other lifestyle factors.

Dietary Measurement Isoflavone intake was determined by a series of detailed questions about the usual frequency and quantity of intake of 33 food groups and beverages using a food-frequency April 2017 Volume 117 Number 4

questionnaire (FFQ). The FFQ has been validated and reported to be useful in face-to-face interviews in the Chinese population.16 The questionnaire asked the frequency of consumption of these foods per day, per week, per month, per year, and mean quantity per time over the past 12 months. Soy isoflavone intake was calculated based on the following soy foods: tofu (wet and dry), soy milk, and dried beans. Mean daily isoflavone intake in milligrams was calculated according to the Chinese Food Composition Table.17 Macronutrient and total energy intake were estimated from the 3-day weighed food intake using an in-house SAS software program.18

Sleep Measurement At baseline in 2002, only sleep duration was collected, but both sleep duration and self-reported daytime falling asleep were collected at follow-up in 2007. Sleep duration was determined by the question “How many hours do you usually sleep each day?” Short sleep duration was defined as <7 hours per day and long sleep duration was defined as 9 hours per day, as in most studies.4 Extreme long sleep duration (14 hours) (n¼1) and missing value (n¼1) were excluded. Self-reported falling asleep during the day was determined by the question “Do you fall asleep involuntarily during the day?”

Anthropometric Measurement and Other Variables In both 2002 and 2007, anthropometry was conducted using standard protocols and techniques. Body weight was measured in light indoor clothing without shoes and rounded to the nearest 0.1 kg. Height was measured without shoes using a stadiometer and rounded to the nearest millimeter. BMI was categorized into the following groups: underweight (<18.5), normal weight (18.5 to 23.9), overweight (24 to 27.9), and obese (28), according to the guideline for Chinese adults.19 Waist circumference was measured to the nearest millimeter midway between the inferior margin of the last rib and the crest of the ilium, in the mid-auxillary line in a horizontal plane. Blood pressure was measured twice with a mercury sphygmomanometer on the right upper arm of the subject, who was seated for 5 minutes before the measurement. The mean of these two measurements was used in the analyses. Hypertension was defined as a systolic blood pressure >140 mm Hg and/or a diastolic blood pressure >90 mm Hg, or the use of antihypertensive drugs. Cigarette smoking was assessed by asking the frequency of daily cigarette smoking in the past 30 days. Frequency of alcohol consumption was determined (none, 1 to 2 days/wk, 3 to 4 days/wk, and daily). Education was recoded into “low” (illiteracy, elementary school); “medium” (middle school); and “high” (high school or higher), based on six categories of education levels in the questionnaire. Occupation was recoded into “manual” or “nonmanual” based on a question with 12 occupational categories. Income information was obtained by the question “What was your family income per person in 2001?” Three yearly income levels were categorized: low (<1,999 Chinese Yuan), medium (2,000 to 4,999 Chinese Yuan), and high (>5,000 Chinese Yuan). Sedentary activity was recoded into four categories: <1 h/day, 1 to 2 h/day. 2 to 3 h/day, and 3 h/day based on questions asked on hours spent per day on sedentary activities, including watching JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

537

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

Quartiles of Isoflavone Intake Men 1 (n[164)

Factors

2 (n[163)

3 (n[163)

Women 4 (n[163)

1 P value (n[220)

ƒ! ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ ƒ meanSDb ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

2 (n[200)

3 (n[210)

4 (n[209)

P value

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒmeanSDƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

Age, y

49.613.1

50.513.2

49.114.1

49.013.6

0.73

49.513.6

48.012.8

47.213.5

47.813.6

0.32

Body mass index

23.43.0

23.03.1

23.43.1

23.43.4

0.61

23.53.6

23.53.6

23.53.4

24.03.7

0.28

Income

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ%ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

c

<0.001

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ%ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

Low

25.9

18.5

15.6

32.3

26.9

17.6

26.1

32.5

Medium

30.9

43.8

30.6

26.7

33.8

46.7

32.4

27.3

High

43.2

37.7

53.8

41.0

39.4

35.7

41.5

40.2

Elementary school

45.1

39.9

36.2

37.4

67.3

56.5

59.5

57.7

Middle school

40.2

45.4

40.5

45.4

25.9

34.5

28.6

29.3

High school

13.4

13.5

18.4

11.7

6.4

8.5

11.9

11.5

Education

0.12

University

1.2

1.2

4.9

5.5

Smoking

67.7

60.1

54.6

58.3

49.4

47.2

52.1

1 to 2 days/wk

9.9

12.9

3 to 4 days/wk

9.3

6.1

31.5

33.7

0.13

0.5

0.5

0.0

1.4

4.5

1.0

2.4

3.8

47.9

90.5

98.0

94.8

96.2

8.6

14.7

4.1

0.0

1.4

1.0

9.2

8.0

2.7

0.5

1.9

0.5

30.1

29.4

2.7

1.5

1.9

2.4

Alcohol

0.10 0.78

None

Daily Sedentary activity

0.001

April 2017 Volume 117 Number 4

<1 h/day

11.6

9.8

10.4

11.0

26.8

14.5

18.1

16.3

1 to 2 h/day

29.3

28.2

23.9

29.4

30.9

34.0

28.1

32.1

2 to 3 h/day

47.6

45.4

51.5

38.7

33.6

41.0

45.2

34.0

3 h/day

11.6

16.6

14.1

20.9

8.6

10.5

8.6

17.7

53.7

56.4

49.7

40.7

0.027 55.0

56.0

48.1

39.7

Manual occupation

<0.001

Region

0.15 0.027

0.40

d

0.001

0.003 <0.001

Urban

10.4

11.7

24.5

29.4

7.7

14.0

21.0

26.3

Rural

89.6

88.3

75.5

70.6

92.3

86.0

79.0

73.7 (continued on next page)

RESEARCH

538

Table 1. Sex-specific characteristics of participants by quartiles of isoflavone intake at baseline in Jiangsu Nutrition study 2002 (n¼1,492)a

April 2017 Volume 117 Number 4

Table 1. Sex-specific characteristics of participants by quartiles of isoflavone intake at baseline in Jiangsu Nutrition study 2002 (n¼1,492)a (continued) Quartiles of Isoflavone Intake Men Factors Dietary measurese

1 (n[164)

2 (n[163)

3 (n[163)

Women 4 (n[163)

1 P value (n[220)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒmeanSDƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

2 (n[200)

3 (n[210)

4 (n[209)

P value

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒmeanSDƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

Carbohydrate (g/day)

337.091.4

349.982.7

330.0113.3

339.1110.2

0.35

307.993.9

296.684.5

301.398.1

299.0109.9

0.66

Protein (g/day)

78.222.2

81.120.1

79.022.9

81.625.3

0.49

65.919.7

64.717.8

67.020.1

69.324.1

0.14

Fat (g/day)

88.835.8

92.440.9

88.237.8

90.733.6

0.73

73.735.5

74.232.0

75.330.8

74.031.0

0.96

Energy intake

2,592.7719.3 2,639.6576.0 2,461.1662.1 2,553.6660.7

0.09

2,152.0586.0 2,109.7506.8 2,140.6576.6 2,128.7629.0

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒmedian (IQRf)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! Soy isoflavone intake (mg/day) Sleep duration at baseline

4.0 (2.2-6.0)

12.8 (10.2-13.6)

21.1 (17.8-27.1)

46.8 (36.7-63.9)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ%ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

<0.001 3.4 (2.0-4.9) 0.14

10.7 (8.8-12.8)

19.0 (16.0-22.8)

40.9 (32.6-59.2)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ%ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

7 to 8 h/day

61.7

66.5

74.2

71.0

65.6

72.5

67.0

70.3

<7 h/day

14.2

14.9

12.9

14.8

10.1

10.0

11.7

12.9

9 h/day

24.1

18.6

12.9

14.2

24.3

17.5

21.4

16.7

0.55

0.43

0.25

7 to 8 h/day

71.3

69.1

68.7

74.8

63.2

66.3

67.1

69.9

<7 h/day

19.5

16.7

21.5

16.0

18.6

20.1

18.1

21.1

9 to 13 h/day

9.1

14.2

9.8

9.2

18.2

13.6

14.8

9.1

Falling asleep at follow-up

6.1

3.7

5.6

6.8

6.8

6.0

3.4

3.8

0.65

<0.001

0.29

a

Participants presented are those that had isoflavone measurements at baseline (2002) and participated in the follow-up (2007). SD¼standard deviation. c Income levels were categorized as low (<1,999 Chinese Yuan), medium (2,000 to 4,999 Chinese Yuan), and high (>5,000 Chinese Yuan) according to the question “What was your family income per person in 2001?” d Occupation was recoded into “manual” or “nonmanual” based on a question with 12 occupational categories. e Macronutrient and total energy intake were estimated from the 3-day weighed food intake using an in-house SAS program.18 f IQR¼interquartile range. b

539

RESEARCH

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

Sleep duration at follow-up

0.89

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒmedian (IQR)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

RESEARCH Table 2. Multilevel multivariable-adjusted associations between quartiles 1 through 4 of isoflavone intake and self-reported long and short sleep duration in Jiangsu Nutrition study 2002-2007 (n¼1,474)a Long Sleep Durationb Fixed effects Isoflavone intake Q1

Model 1d

Model 2e

Short Sleep Durationc

Model 3f

Model 1d

Model 2e

Model 3f

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ odds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1.00

1.00

1.00

1.00

1.00

1.00

Q2

0.71 (0.53-0.96)* 0.75 (0.55-1.01)

0.75 (0.55-1.00)

1.49 (1.10-2.00)** 1.45 (1.08-1.95)* 1.44 (1.08-1.94)*

Q3

0.78 (0.57-1.05)

0.83 (0.61-1.12)

1.01 (0.73-1.38)

0.92 (0.67-1.27) 0.93 (0.67-1.27)

Q4

0.64 (0.46-0.87)** 0.65 (0.48-0.89)** 0.66 (0.48-0.90)** 1.26 (0.92-1.70)

1.13 (0.83-1.54) 1.14 (0.84-1.55)

0.83 (0.62-1.12)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒP for trendg ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 0.009 Random effects Level 2 (individual)

0.016

0.018

0.610

0.805

0.872

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (standard error)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 0.73 (0.27)

Level 1 (survey years) —

0.43 (0.24)

0.42 (0.24)

0.38 (0.23)

0.24 (0.22)

0.22 (0.21)











a All models were conducted by using multilevel (two levels) mixed-effects logistic regression with the mean and variance adaptive Gauss-Hermite (mvaghermite) integration method. Participants included were those had both isoflavone intake and sleep duration data at both baseline (2002) and follow-up (2007) (n¼1,474). For random effects, odds ratio (standard error) are presented. b Long sleep refers 9 h/day sleep. c Short sleep refers to <7 h/day sleep. d Model 1 adjusted for age, sex, and energy intake. e Model 2 further adjusted for rural region, smoking, alcohol consumption, and sedentary activity. f Model 3 further adjusted for body mass index and hypertension. g P for trend across the quartiles was calculated using the median value of isoflavone intake at each quartile and used as continuous variable in the models. *P<0.05. **P<0.01.

television, reading, computer use, and video games. Urban/ rural area information was defined according to the residence area (either as city or county, as described in the Methods section).

Statistical Analysis Descriptive data are presented as the mean (standard deviation) and 95% CI or as percentages. c2 tests were used to compare the differences between categorical variables and analysis of variance was performed to compare differences in continuous variables between groups. Isoflavone intake was recoded into quartiles, and the highest quartile was compared with the lowest quartile (reference group) in different models. Mixed-effects logistic (two levels) regression was performed by using command “melogit” to assess the longitudinal association between isoflavone intake and short/long sleep duration within survey years (level 1) nested within individuals (level 2). A set of models were used: model 1 adjusted for age, sex, energy intake; model 2 further adjusted for income, education, region, smoking, alcohol consumption, and sedentary activity; model 3 further adjusted for BMI and hypertension; and model 4 further adjusted for baseline daily fruit, meat, and vegetable intake. A sex-specific association between baseline isoflavone intake and self-reported sleep symptoms at follow-up was assessed using multivariable logistic regression models adjusting for the same covariates. We tested for a linear trend across quartiles of isoflavone intakes by assigning each participant the median value of isoflavone intake at each quartile and modeling this value as a continuous variable. Interactions 540

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

between isoflavone intake and age, income, manual job, and region were conducted by adding a multiplicative term with age, income, and region as a categorical variable and the quartiles of isoflavone intake as the categorical variable in the fully adjusted model. Sensitivity analyses were also performed to test the robustness of the results. First, the association between isoflavone intake (using the continuous variable, standardized) and sleep duration (using the continuous variable) in the same multivariable models was examined. Second, the change of isoflavone intake over 5 years was also examined. Persistent low intake of isoflavone was used as the reference group, and was defined as less than the median of isoflavone intake at both 2002 and 2007. Persistent high intake of isoflavone was defined as greater than the median of isoflavone intake at both 2002 and 2007. In addition, the association between intake of isoflavone-rich food items (eg, tofu, soy milk, and dried beans) and sleep symptoms at follow-up was examined. All the analyses were performed using STATA statistical software.20 Statistical significance was considered at P<0.05 (two-sided).

RESULTS In total, 1,474 participants had complete soy isoflavone intake and sleep duration data in both 2002 and 2007. The prevalence of short and long sleep duration was 11.7% and 18.9% in 2002, and 15.7% and 12.6% in 2007. There was a decrease in sleep duration (mean¼0.36 h/day) on average over 5 years in the population, with an increased sleep duration among short sleepers and a decreased sleep duration among long sleepers. The prevalence of daytime falling asleep was 5.3% April 2017 Volume 117 Number 4

RESEARCH Table 3. Self-reported daytime falling asleep at follow-up (2007) according to isoflavone intake quartiles 1 to 4 at baseline (2002) in Jiangsu Nutrition study (n¼1,492)a Isoflavone Intake Men Model

n

Q1 (n[164)

Q2 (n[163)

Q3 (n[163)

Q4 (n[163)

P for trendb

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! c

645

1.00

0.59 (0.21-1.67)

0.90 (0.35-2.28)

1.13 (0.46-2.74)

0.506

2d

637

1.00

0.48 (0.16-1.42)

0.86 (0.32-2.32)

1.35 (0.52-3.55)

0.253

e

609

1.00

0.49 (0.17-1.47)

0.88 (0.33-2.37)

1.34 (0.51-3.55)

0.267

Q4 (n[209)

P for trendb

1 3

Women Q1 (n[220)

Q2 (n[200)

Q3 (n[210)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1c

831

1.00

0.92 (0.42-2.02)

0.51 (0.20-1.29)

0.56 (0.23-1.37)

0.160

d

813

1.00

0.81 (0.35-1.87)

0.44 (0.17-1.17)

0.54 (0.21-1.38)

0.177

3e

813

1.00

0.84 (0.36-1.95)

0.45 (0.17-1.19)

0.54 (0.21-1.40)

0.179

2

a

Participants included in the model were those who had isoflavone intake at baseline (2002) and participated at follow-up (2007) (n¼1,492; men, n¼653). P for trend across quartiles was calculated by using the median value of isoflavone intake at each quartile and used as a continuous variable in the models. Model 1 adjusted for age and energy intake. d Model 2 further adjusted for income, education, rural region, smoking, alcohol consumption, and sedentary activity. e Model 3 further adjusted for body mass index and hypertension. b c

(5.6% in men and 5.0% in women, respectively). Median intake of soy isoflavone was 14.9 mg/day (interquartile range [IQR] 7.2 to 29.6 mg/day) in 2002 and 27.3 mg/day (IQR 11.6 to 50.5 mg/day) in 2007. Sex-specific characteristics by quartiles of isoflavone intake at baseline are presented in Table 1. In brief, the median intake of isoflavone was 15.6 mg/ day (IQR 8.0 to 30.9 mg/day) in men and 13.6 mg/day (IQR 6.8 to 28.0 mg/day) in women at baseline. Compared with the lowest quartile of isoflavone intake, those who were in the highest quartile were less likely to do manual jobs, but more likely to be urban residents for both sexes. However, alcohol consumption decreased as the intake of isoflavone increased, such that the least alcohol consumption was observed in the highest quartile of isoflavone intake compared to the lowest quartile in men. While in women, compared with the lowest quartile of isoflavone intake, those who were in the highest quartile were more likely to engage in sedentary activities. There was no difference of sleep duration and the prevalence of daytime falling asleep across quartiles of isoflavone intake at baseline. In the multilevel multivariable mixed-effects logistic regression, comparing the highest quartile (Q4) with the lowest quartile (Q1) of isoflavone intake, the odds ratio (OR) for long sleep was 0.65 (95% CI 0.47 to 0.89; P for trend¼0.016) after adjusting for age, sex, energy intake, demographic and lifestyle factors, as well as BMI and hypertension. The results remained the same when further adjusting for daily intake of fruits, meats, and vegetables (Table 2). There was no longitudinal independent association

April 2017 Volume 117 Number 4

between isoflavone intake and short sleep. When using sleep duration as a continuous variable, we found a consistent inverse association between isoflavone intake (standardized) and sleep duration (b¼0.07; 95% CI 0.13 to 0.01). Baseline isoflavone intake was not associated with daytime falling asleep at follow-up in men. However, there seems a trend of inverse association between isoflavone intake and daytime falling asleep across the quartiles of isoflavone intake in women, although it was not statistically significant (Table 3). When we combined isoflavone intake at 2002 and 2007, it was found that, compared with persistent low isoflavone intake (less than median intake of isoflavone at both 2002 and 2007), persistent high isoflavone intake (greater than median intake of isoflavone at both 2002 and 2007) was associated with reduced risk of falling asleep at follow-up in women but not men (OR¼0.20; 95% CI 0.06 to 0.68; Table 4). Baseline tofu intake was associated with reduced likelihood of falling asleep in women but not men (OR¼0.24; 95% CI 0.07 to 0.08; Tables 5 and 6, available online at www.andjrnl. org). Baseline soy milk consumption was associated with increased risk of the likelihood of falling asleep in men but not women (OR¼3.10; 95% CI 1.38 to 6.97; Tables 5 and 6, available online at www.andjrnl.org). There were no other associations between baseline isoflavone-rich food consumption and sleep symptoms at follow-up. There was no interaction between isoflavone intake and age, income, manual job, and region in relation to long sleep duration and daytime falling asleep.

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

541

RESEARCH Table 4. Self-reported daytime falling asleep at follow-up (2007) according to the change of isoflavone intake at both baseline (2002) and follow-up (2007) by sex in Jiangsu Nutrition study (n¼1,474)a Change in Isoflavone Intake over 5 Years Men Model

n

Ref (persistent low intake) (n[184)

Increased intake (n[211)

Decreased intake (n[72)

Persistent high intake (n[180)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ ƒ odds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ ƒ! b

639

1.00

0.44 (0.17-1.13)

0.99 (0.34-2.91)

0.86 (0.37-1.98)

2c

631

1.00

0.44 (0.16-1.17)

1.22 (0.38-3.94)

1.10 (0.45-2.70)

d

603

1.00

0.45 (0.17-1.21)

1.21 (0.36-4.00)

1.14 (0.46-2.82)

Decreased intake (n[98)

Persistent high intake (n[219)

1 3

Women Ref (persistent low intake) (n[222)

Increased intake (n[288)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1b

820

1.00

0.77 (0.37-1.60)

1.07 (0.42-2.71)

0.26 (0.09-0.80)*

c

802

1.00

0.74 (0.34-1.60)

1.17 (0.44-3.12)

0.20 (0.06-0.66)**

3d

802

1.00

0.77 (0.36-1.67)

1.21 (0.45-3.24)

0.20 (0.06-0.68)**

2

a

Participants included in the model were those who had isoflavone intake and sleep duration both at 2002 and 2007 (n¼1,474). The change of isoflavone intake was based on isoflavone intake at two time points by sex. In each sex, isoflavone intake of each time point was recoded as low (less than median) and high (greater than median). Participants were grouped into four groups: persistent low intake (low intake at both time points), increased intake (low intake at 2002 and high intake at 2007), decreased intake (high intake at 2002 and low intake at 2007), and persistent high intake (high intake at both time points). b Model 1 adjusted for age and energy intake. c Model 2 further adjusted for income, education, rural region, smoking, alcohol consumption, and sedentary activity. d Model 3 further adjusted for body mass index and hypertension. *P<0.05. **P<0.01.

DISCUSSION In this large prospective cohort study, isoflavone intake was inversely associated with long sleep duration (9 h/day) over 5 years among Chinese adults. Persistent high intake of isoflavone was associated with reduced likelihood of daytime falling asleep in women but not men. A persistent inverse association was found between isoflavone intake and continuous sleep duration. No persistent association was found between soy isoflavone and short sleep duration. The relationship between isoflavone intake and sleep duration has not been widely studied. A recent Japanese cross-sectional study found a positive association between isoflavone intake and sleep duration of 7 to 8 h/day, which was defined as the optimal sleep duration and the main outcome (prevalence 13.3%).14 However, in our Chinese study population, the majority slept 7 to 8 h/d (68.4% in 2002 and 68.6% in 2007). In our repeated measures at two points, high isoflavone intake was inversely associated with long sleep duration in the general population. Previous studies have suggested that estrogen receptor messenger RNA levels have a circadian rhythm of expression in the suprachiasmatic nucleus of the hypothalamus in rats.21 Suprachiasmatic nucleus has been suggested to have wake-promoting abilities during active phase in mice.22 This indicates a potential mechanism of estrogen in sleep regulation. Animal models suggest that

542

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

estradiol suppresses nonrapid eye movement and/or rapid eye movement sleep in ovariectomized rodents.23 It has also been suggested that estradiol increases arousal behavior by decreasing sleep-inducing molecules.24-26 The inverse association between isoflavone intake and long sleep duration in our study, therefore, may be explained by the similar estrogen function of isoflavone in increasing arousal, resulting in a reduced propensity for long sleep. It is important to note that most of the animal studies were conducted in female rodents; it is unclear about the effect in males. It is also noteworthy that despite isoflavone having similarities with estrogen, it is considered as a natural selective estrogen receptor modulator, and with more affinity for estrogen receptor-b than estrogen receptor-a.27 Although long sleep duration is not a risk factor or independent causal factor for mortality without considering confounders such as depression, sleep apnea, and alcohol consumption, habitual long sleep duration is likely to be an indicator of poor physical and mental health status.28,29 This also indicates the importance of optimal sleep duration in terms of the association with mortality as well as health outcomes in general. Our results suggest a beneficial role of dietary isoflavone in regulating sleep duration. In addition, long-term dietary isoflavone may further reduce the adverse effects due to long sleep time.

April 2017 Volume 117 Number 4

RESEARCH Isoflavone intake at baseline was not associated with daytime falling asleep at follow-up in men. In women, only a trend of reduced likelihood of daytime falling asleep could be speculated. However, when combining isoflavone intake at baseline and follow-up, it was found that persistent high isoflavone intake over 5 years was significantly associated with a reduced risk of daytime falling asleep in women but not men, compared with persistent low isoflavone intake. Although only daytime falling sleep was assessed in our study, it may reflect general sleep quality the night before. An Italian cross-sectional study found that the groups using transdermal estrogens with or without progestin tended to report fewer difficulties falling asleep and less poor sleep compared with controls in women around menopause.30 While the benefits of estrogen in improving sleep may be due to the relief of menopausal syndrome, the Japanese study mentioned found the benefits of isoflavone in optimal sleep quality (assessed by feeling refreshed after sleep and whether using hypnotic drugs) in both men and women aged 20 to 78 years.14 This may suggest that the mechanism of the beneficial effect of isoflavone on sleep is not limited to the relief of menopausal syndromes. Interestingly, baseline soy milk consumption seems to be associated with an increased likelihood of daytime falling asleep in men but not in women. When soy milk is consumed as take away fast food for breakfast in China, it might be because of a late bedtime the night before and, therefore, the individual might be fatigued the next day. There was no association between isoflavone and reduced daytime falling asleep in men. A previous randomized crossover study that assessed the effect of estrogen-replacement therapy on sleep complaints in postmenopausal women found a significantly improved subjective sleep quality and alleviation of sleep complaints, including morning tiredness.31 They concluded that the alleviated climacteric symptom was the most important factor for the improved sleep quality. In our study, the mean age of women was 47 years at baseline; feeling less asleep at follow-up may also be due to the benefits of soy isoflavone on the climacteric symptoms. Taken together, isoflavone has a beneficial effect on improving sleep quality, particularly in women. The strengths of this study are exploration of the longitudinal association between isoflavone intake and sleep duration at two time points and the large sample size in the general Chinese population. Several limitations need to be acknowledged: daytime falling asleep was not assessed at baseline and the low prevalence in the population may have limited our ability to detect significant effects; sleep duration was self-reported; potential errors in isoflavone intake reporting, as it was estimated from self-reported FFQ; and not all isoflavone food sources were included. However, soy isoflavone intake was basically consistent with the intake in previous studies in different region and countries,32,33 although the included food sources in our study did not cover all isoflavone sources. In addition, self-reported bias is likely to be random in this prospective study design.

CONCLUSIONS This is the first study to our knowledge that has investigated longitudinal associations between soy isoflavone and sleep at the population level. The study findings support the potential beneficial effect of soy isoflavone intake in sleep regulation in April 2017 Volume 117 Number 4

both sexes. However, an association between soy isoflavone intake and sleep in the long term, particularly for sleep symptoms like daytime falling asleep, was found in women only. Prevention studies in women with sleep problems are needed to explore the benefits of isoflavone intake on sleep outcomes.

References 1.

Nagai M, Hoshide S, Kario K. Sleep duration as a risk factor for cardiovascular disease: A review of the recent literature. Curr Cardiol Rev. 2010;6(1):54-61.

2.

Chaput JP, McNeil J, Despres JP, Bouchard C, Tremblay A. Short sleep duration as a risk factor for the development of the metabolic syndrome in adults. Prev Med. 2013;57(6):872-877.

3.

Shan Z, Ma H, Xie M, et al. Sleep duration and risk of type 2 diabetes: A meta-analysis of prospective studies. Diabetes Care. 2015;38(3): 529-537.

4.

Cappuccio FP, D”Elia L, Strazzullo P, Miller MA. Sleep duration and all-cause mortality: A systematic review and meta-analysis of prospective studies. Sleep. 2010;33(5):585-592.

5.

Patel SR, Malhotra A, Gottlieb DJ, White DP, Hu FB. Correlates of long sleep duration. Sleep. 2006;29(7):881-889.

6.

Endeshaw Y, Rice TB, Schwartz AV, et al. Snoring, daytime sleepiness, and incident cardiovascular disease in the health, aging, and body composition study. Sleep. 2013;36(11):1737-1745.

7.

McEwen BS, Alves SE. Estrogen actions in the central nervous system. Endocr Rev. 1999;20(3):279-307.

8.

Ensrud KE, Guthrie KA, Hohensee C, et al. Effects of estradiol and venlafaxine on insomnia symptoms and sleep quality in women with hot flashes. Sleep. 2015;38(1):97-108.

9.

Setchell KD, Cassidy A. Dietary isoflavones: Biological effects and relevance to human health. J Nutr. 1999;129(3):758S-767S.

10.

Hachul H, Brandao LC, D’Almeida V, Bittencourt LR, Baracat EC, Tufik S. Isoflavones decrease insomnia in postmenopause. Menopause. 2011;18(2):178-184.

11.

Drews K, Seremak-Mrozikiewicz A, Puk E, Kaluba-Skotarczak A, Malec M, Kazikowska A. Efficacy of standardized isoflavones extract (Soyfem) (52-104 mg/24h) in moderate and medium-severe climacteric syndrome [in Polish]. Ginekol Pol. 2007;78(4):307-311.

12.

Tom SE, Kuh D, Guralnik JM, Mishra GD. Self-reported sleep difficulty during the menopausal transition: Results from a prospective cohort study. Menopause. 2010;17(6):1128-1135.

13.

Freeman EW, Sammel MD, Lin H, et al. Symptoms associated with menopausal transition and reproductive hormones in midlife women. Obstet Gynecol. 2007;110(2 Pt 1):230-240.

14.

Cui Y, Niu K, Huang C, et al. Relationship between daily isoflavone intake and sleep in Japanese adults: A cross-sectional study. Nutr J. 2015;14:127.

15.

Shi Z, Luscombe-Marsh ND, Wittert GA, et al. Monosodium glutamate is not associated with obesity or a greater prevalence of weight gain over 5 years: Findings from the Jiangsu Nutrition Study of Chinese adults. Br J Nutr. 2010;104(3):457-463.

16.

Zhao W, Hasegawa K, Chen J. The use of food-frequency questionnaires for various purposes in China. Public Health Nutr. 2002;5(6A): 829-833.

17.

Yang Y, Wang G, Pan X, eds. China Food Composition Table 2002. The Institute of Nutrition and Food Safety, Chinese Center for Disease Control and Prevention. Peking, China: Peking University Medical Press; 2002.

18.

SAS [computer program]. Version 9.2. Cary, NC: SAS Institute; 2008.

19.

Chen C, Lu FC; Department of Disease Control, Ministry of Health PRC. The guidelines for prevention and control of overweight and obesity in Chinese adults. Biomed Environ Sci. 2004;17(suppl):1-36.

20.

STATA [computer program]. Release 14. College Station, TX: StataCorp LP; 2015.

21.

Wilson ME, Rosewell KL, Kashon ML, Shughrue PJ, Merchenthaler I, Wise PM. Age differentially influences estrogen receptor-alpha (ERalpha) and estrogen receptor-beta (ERbeta) gene expression in specific regions of the rat brain. Mech Ageing Dev. 2002;123(6):593-601.

22.

Easton A, Meerlo P, Bergmann B, Turek FW. The suprachiasmatic nucleus regulates sleep timing and amount in mice. Sleep. 2004;27(7):1307-1318.

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

543

RESEARCH 23.

Mong JA, Baker FC, Mahoney MM, et al. Sleep, rhythms, and the endocrine brain: Influence of sex and gonadal hormones. J Neurosci. 2011;31(45):16107-16116.

28.

Grandner MA, Drummond SP. Who are the long sleepers? Towards an understanding of the mortality relationship. Sleep Med Rev. 2007;11(5):341-360.

24.

Ribeiro AC, Pfaff DW, Devidze N. Estradiol modulates behavioral arousal and induces changes in gene expression profiles in brain regions involved in the control of vigilance. Eur J Neurosci. 2009;29(4):795-801.

29.

Stamatakis KA, Punjabi NM. Long sleep duration: A risk to health or a marker of risk? Sleep Med Rev. 2007;11(5):337-339.

30.

Sarti CD, Chiantera A, Graziottin A, et al. Hormone therapy and sleep quality in women around menopause. Menopause. 2005;12(5):545-551.

25.

Deurveilher S, Rusak B, Semba K. Estradiol and progesterone modulate spontaneous sleep patterns and recovery from sleep deprivation in ovariectomized rats. Sleep. 2009;32(7):865-877.

31.

Polo-Kantola P, Erkkola R, Helenius H, Irjala K, Polo O. When does estrogen replacement therapy improve sleep quality? Am J Obstet Gynecol. 1998;178(5):1002-1009.

26.

Hadjimarkou MM, Benham R, Schwarz JM, Holder MK, Mong JA. Estradiol suppresses rapid eye movement sleep and activation of sleep-active neurons in the ventrolateral preoptic area. Eur J Neurosci. 2008;27(7):1780-1792.

32.

Talaei M, Koh WP, van Dam RM, Yuan JM, Pan A. Dietary soy intake is not associated with risk of cardiovascular disease mortality in Singapore Chinese adults. J Nutr. 2014;144(6):921-928.

33.

27.

Oseni T, Patel R, Pyle J, Jordan VC. Selective estrogen receptor modulators and phytoestrogens. Planta Med. 2008;74(13):1656-1665.

Yu D, Zhang X, Xiang YB, et al. Association of soy food intake with risk and biomarkers of coronary heart disease in Chinese men. Int J Cardiol. 2014;172(2):e285-e287.

AUTHOR INFORMATION Y. Cao is a PhD candidate, Population Research and Outcome Studies, School of Medicine, University of Adelaide, South Australia, and Freemasons Foundation Centre for Men’s Health, Adelaide, South Australia. A. W. Taylor is a professor and head, Population Research and Outcome Studies, School of Medicine, University of Adelaide, South Australia. S. Zhen is section director and a professor, Jiangsu Provincial Centre for Disease Control and Prevention, Nanjing, China. R. Adams is a professor and sleep clinician, The Health Observatory, University of Adelaide, Queen Elizabeth Hospital Campus, Woodville, South Australia. S. Appleton is a research fellow, Freemasons Foundation Centre for Men’s Health, Adelaide, South Australia, and The Health Observatory, University of Adelaide, Queen Elizabeth Hospital Campus, Woodville, South Australia. Z. Shi is an associate professor, Population Research and Outcome Studies, School of Medicine, University of Adelaide, South Australia; Freemasons Foundation Centre for Men’s Health, Adelaide, South Australia; and Population Research and Outcome Studies, School of Medicine, University of Adelaide, South Australia. Address correspondence to: Yingting Cao, MMSc, Population Research and Outcome Studies, School of Medicine, University of Adelaide, SAHMRI, L7, North Terrace, South Australia 5000, Australia. E-mail: [email protected]

STATEMENT OF POTENTIAL CONFLICT OF INTEREST No potential conflict of interest was reported by the authors.

FUNDING/SUPPORT There is no funding to disclose.

544

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

April 2017 Volume 117 Number 4

RESEARCH Table 5. Self-reported daytime falling asleep at follow-up (2007) (n¼73 of 1,405) according to tofu intake tertiles at baseline (2002) in Jiangsu Nutrition study (n¼1,413)a Tofu Intake Women c

Modelb

Low (n[375)

Mediumc (n[216)

Highc (n[201)

P for trendd

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1

1.00

1.13 (0.56-2.26)

0.25 (0.07-0.84)*

0.06

2

1.00

0.95 (0.46-1.97)

0.24 (0.07-0.83)*

0.045

3

1.00

0.95 (0.46-1.99)

0.24 (0.07-0.84)*

0.050

4

1.00

0.97 (0.46-2.03)

0.24 (0.07-0.88)*

0.06

Lowc (n[256)

Mediumc (n[179)

Men Highc (n[186)

P for trendd

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1

1.00

0.62 (0.25-1.55)

0.87 (0.39-1.97)

0.70

2

1.00

0.67 (0.26-1.75)

1.20 (0.48-2.97)

0.77

3

1.00

0.69 (0.26-1.85)

1.20 (0.48-2.99)

0.75

4

1.00

0.72 (0.27-1.95)

1.22 (0.48-3.08)

0.72

a

Participants included in the model were those who had tofu intake at baseline (2002) and participated at follow-up (2007) (n¼1,413, women, n¼792). Tofu intake¼0 was excluded (n¼79). Model 1 adjusted for age and energy intake. Model 2 further adjusted for income, education, rural region, smoking, alcohol consumption, and sedentary activity. Model 3 further adjusted for body mass index and hypertension. Model 4 further adjusted for baseline daily intake of fruit, vegetable, and meat. c Low, medium, and high were defined as tertiles of tofu intake for each sex. d P for trend across quartiles was calculated by using the median value of isoflavone intake at each tertile and used as a continuous variable in the models. *P<0.05. b

April 2017 Volume 117 Number 4

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

544.e1

RESEARCH Table 6. Self-reported daytime falling asleep at follow-up (2007) (n¼78 of 1,481) according to soy milk and dried beans intake (consumers vs nonconsumers) at baseline (2002) in Jiangsu Nutrition study (n¼1,492)a Soy Milk Intake Women Modelb

Nonconsumers (n[656)

Consumers (n[183)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1

1.00

1.12 (0.53-2.34)

2

1.00

0.97 (0.43-2.16)

3

1.00

0.96 (0.43-2.16)

4

1.00

0.93 (0.41-2.11) Men

Nonconsumers (n[500)

Consumers (n[153)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1

1.00

2.20 (1.09-4.46)*

2

1.00

2.93 (1.33-6.46)**

3

1.00

2.99 (1.34-6.64)**

4

1.00

3.10 (1.38-6.97)** Dried Beans Intake Women

Nonconsumer (n[560)

Consumer (n[279)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1

1.00

0.70 (0.34-1.42)

2

1.00

0.73 (0.35-1.51)

3

1.00

0.73 (0.35-1.52)

4

1.00

0.76 (0.36-1.58) Men

Nonconsumer (n[500)

Consumer (n[153)

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒodds ratio (95% CI)ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! 1

1.00

1.69 (0.85-3.35)

2

1.00

1.88 (0.90-3.95)

3

1.00

1.91 (0.90-4.04)

4

1.00

1.93 (0.91-4.09)

a

Participants included in the model were those who had isoflavone intake at baseline (2002) and participated at follow-up (2007) (n¼1,492; men, n¼653). Due to large number of subjects who did not consume soy milk and dried beans, the intake was categorized into consumers and nonconsumers. b Model 1 adjusted for age and energy intake. Model 2 further adjusted for income, education, rural region, smoking, alcohol consumption and sedentary activity. Model 3 further adjusted for BMI and hypertension. Model 4 further adjusted for baseline daily intake of fruit, vegetable and meat. *P<0.05. **P<0.01.

544.e2

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

April 2017 Volume 117 Number 4