Osteoporosis International

, Volume 21, Issue 6, pp 939–946

Higher sea fish intake is associated with greater bone mass and lower osteoporosis risk in postmenopausal Chinese women

Authors

  • Y.-m. Chen
    • Department of Community and Family Medicine, and Centre of Research and Promotion of Women’s Health, School of Public HealthThe Chinese University of Hong Kong
    • Department of Medical Statistics and Epidemiology, School of Public HealthSun Yat-sen University
    • Department of Community and Family Medicine, and Centre of Research and Promotion of Women’s Health, School of Public HealthThe Chinese University of Hong Kong
  • S. S. Lam
    • Department of Community and Family Medicine, and Centre of Research and Promotion of Women’s Health, School of Public HealthThe Chinese University of Hong Kong
Original Article

DOI: 10.1007/s00198-009-1029-4

Cite this article as:
Chen, Y., Ho, S.C. & Lam, S.S. Osteoporos Int (2010) 21: 939. doi:10.1007/s00198-009-1029-4

Abstract

Summary

We examined the cross-sectional association of the intakes of different types of fishes with bone mass and osteoporosis risk in postmenopausal Chinese women. We found that higher intake of sea fish is independently associated with greater bone mass and lower osteoporosis risk among postmenopausal Chinese women.

Introduction

Fish contains many important nutrients that are beneficial on bone health, but limited data on the relationship between fish intake and bone health are available. We examined the association of the intakes of different types of fishes with bone mineral density (BMD) and bone mineral content (BMC) and osteoporosis risk.

Methods

This population-based cross-sectional study was conducted among 685 postmenopausal Chinese women. Habitual dietary intakes were assessed using food frequency questionnaire. BMD and BMC at the whole body, lumbar spine, and left hip were measured with dual-energy x-ray absorptiometry.

Results

After adjusting for the potential confounders, we observed dose-dependent relations between sea fish intake and BMDs, BMCs, and osteoporosis risk; the mean BMDs were 3.2–6.8% higher, and BMCs 5.1–9.4% higher in the top quintile groups (Q5) of sea fish intake than in the bottom quintile (Q1) at the whole body and hip sites (p < 0.05); the odds ratios (95% confidence interval (CI)) for osteoporosis (T-score < −2.5) in Q5 were 0.23 (0.08–0.66), 0.12 (0.03–0.59), and 0.06 (0.01–0.44) compared with those in Q1 at the whole body, total hip, and femur neck, respectively. No independent association between consumption of freshwater fish or shellfish and bone mass was observed.

Conclusion

Higher intake of sea fish is independently associated with greater bone mass and lower osteoporosis risk among postmenopausal Chinese women.

Keywords

Bone densityBone mineral contentFishOsteoporosisPostmenopausal women

Introduction

Osteoporosis is a major growing public health problem affecting both Western and Asian populations. It affects one in three postmenopausal women and the majority of the elderly population. Ten million American women were estimated to have osteoporosis and 34 million its precursor, osteopenia [1]. Nutritional factors play a key role in the maintenance of optimal bone health [2]. Until recently, focuses have primarily been placed on calcium and a few isolated nutrients. Except for milk and soy products, limited data are available on the association between food intake and bone health [2].

Fish is a major food group worldwide. Fish and shellfish are the major sources of animal foods in Chinese residents in coastal regions. The 2002 Chinese National Nutrition and Health Survey showed that the average intakes of fish and shellfish ranged between 90 and 200 g/day (raw weight) per reference person in coastal areas [3]. A large number of studies show fish consumption is associated with lower risk for cardiovascular diseases [4, 5], certain cancer [6], and total mortality [7]. Fish is a source of high quality protein; n-3 polyunsaturated fatty acids (PUFA); fat-soluble vitamins such as A and D; and minerals such as calcium, zinc, selenium, and iodine. Many of them may have beneficial effects on bone health. Calcium and vitamin D has long been established to be the key nutrients for bone health. Dietary zinc might have a positive association with bone mineral density (BMD) [8]. Recent studies also found that n-3 PUFA was associated with greater bone mass in human [9, 10]. However, studies which examined the association of fish consumption with bone health had obtained inconsistent results [1114]. Some studies [1113], but not all [14], reported that fish consumption was associated with greater bone mass or lower fracture rates in human. Moreover, as sea fish have much higher contents of n-3 PUFA (13.7 vs 9.0/100 g) and vitamin D [15, 16], they might have a greater effect on bone mass than freshwater fish, but the existence of these differential effects are unclear.

In this cross-sectional study, we investigated the association between habitual intakes of sea fish and freshwater fish and bone mass among 685 Chinese women within the first 12 years of natural menopause.

Subjects and methods

Subjects

Population-based subjects who met the following criteria were enrolled in this cross-sectional study, and details of the study have previously been reported [17]. Participants were required to be Hong Kong residents of Chinese origin aged between 48 and 63 years, within 12 years of natural menopause, defined as at least 12 months since the last menstrual cycle. Subjects who had any detectable disease or medication known to affect bone mass were excluded. After initial screening for eligibility, subjects were invited to the Prince of Wales Hospital in Hong Kong. Staff with relevant knowledge in medical sciences screened the subjects for eligibility via structured face-to-face interview to ensure that they met the inclusion criteria. Written informed consent was obtained from all the participants prior to enrollment. The Ethical Committee of the Chinese University of Hong Kong approved the study. A total of 685 volunteers who met the screening criteria were recruited during October 1999 to January 2001.

Questionnaire interview

Information on sociodemographic data, menopausal history, and physical activities, including average hours spent in sitting, standing, walking, mild and vigorous physical activities carrying a load, and going upstairs was collected through face-to-face interviews by trained interviewers based on a structured questionnaire [17, 18]. The dietary assessment of intakes of energy, protein, calcium, and phosphorus was assessed based on a quantitative food frequency questionnaire (FFQ) that included 60 food groups/items as described in previous reports [17, 18]. The intake of fish and shellfish was assessed using five items including freshwater fish, small fish or canned fish eaten with bones, sea fish, shrimp and crab, and other crustaceans and mollusks. The mean intake of food per day, week, or month was reported at the interviews, using the past 12 months prior to the interview as the reference period. Foods with frequency of intake less than once per month were ignored. Food pictures and full-scale food portion visual aids for the reference portion sizes were provided as visual aids. A separate analysis noted the reproducibility of freshwater fish and sea fish intakes based on the FFQ over a 9-month period were r = 0.519, p < 0.0001 and r = 0.531, p < 0.0001, respectively (n = 369). Of the subjects, 79.1% and 80.2% had within one unit quintile agreements in freshwater fish and sea fish intakes, respectively. Vitamin D intake was calculated from Standard Tables of Food Composition in Japan, and other nutrients were calculated from China Food Composition Table [15]. Dietary potential renal acid loads (PRAL) was estimated according to the method by Remer and Manz [19].

Anthropometric and bone mass measurements

Height was measured to the nearest 0.5 cm, and weight to the nearest 0.1 kg in light clothing and without shoes. BMI was calculated as weight (kilogram)/height meter square. The BMD and bone mineral content (BMC) at the whole body, the lumbar spine (L1-L4), as well as the left hip (including total hip, femur neck, trochanter, and intertrochanter) were measured by dual-energy x-ray absorptiometry (Hologic QDR-4500, Waltham, MA, USA). The coefficient of variation of measurements with the spine phantom was 0.4%. The in vivo coefficient of variations for BMD measurements were 1.53%, 1.72%, 1.15%, 4.86%, and 1.2% for the spine, femoral neck, trochanter, intertrochanter, and whole body, respectively. Osteoporosis of the whole body, spine, and hips was defined as T-scores of BMDs at the relevant sites equal to or less than −2.5 standard deviations using the Oriental population referent values established by Hologic, Inc., as did in our previous report [20].

Statistical analysis

The dependent variables included BMD at the whole body, lumbar spine (L1-L4), and left hip; independent variables included habitual intake of sea fish over the past year; co-variables included age, years since menopause (YSM), body weight and height, and energy expenditure in physical activities; and dietary energy, protein, and calcium.

Subjects were stratified into quintile categories (Q1 (lowest), Q2, Q3, Q4, and Q5) by freshwater fish, sea fish, and shell fish, respectively. The middle three quintiles (Q2–Q4) were re-combined into one category (Q2–4) due to similar amount of intakes. Means and covariate-adjusted means of the BMDs or BMCs at the various bone sites among the three groups (Q1, Q2–4, and Q5) according to freshwater fish, sea fish, and shellfish were separately compared by multiple comparison tests (method = Bonferroni) of one-way analysis of variance and analysis of covariance (ANCOVA). In the ANCOVA models, adjustments were made for age, body mass index, YSM, and energy expenditure in physical activities and dietary energy, and protein and calcium intakes. Only those covariates (except for dietary energy) with significant association with BMD or BMC were retained in the final model. Logistic regression analyses were used to test the independent association of levels of fish intake with the occurrence of osteoporosis as defined by T-scores of less than −2.5 standard deviations from the respective bone sites after controlling for the covariates. Forward stepwise and enter procedures were used, respectively, for adding the covariates and fish intake. F-to-entry and -remove criteria were 0.05 and 0.10 in the stepwise procedure. Statistical Package for the Social Sciences (SPSS) for Windows (Release 13, SPSS Inc., Chicago, IL, USA) was used for the analyses.

Results

The subjects had a mean (SD) age of 55 (3.5) ranging from 48 to 63 years. The mean YSM was 4.6 (2.8) years with a range of 1–12.5 years. The mean, median, and SD values of intakes (in edible portion, grams/day) of fish and shellfish were 26.7, 17.9, and 30.26 for freshwater fish; 23.3, 14.3, and 26.7 for sea fish; 2.8, 0.0 and 6.3 for small fish with bones; 3.6, 2.1 and 6.5 for shellfish; and 53.6, 44.0, and 43.4 for all fish and shellfish. On average, fish accounted for 16.0 (SD, 9.4) percent of total protein intake.

Both univariate and multivariate analyses showed no significant differences in BMDs or BMCs at the whole body, lumbar spine, and hip sites among the groups classified by freshwater fish (all p > 0.3). Except for a marginal significant difference in the whole body BMC, we did not observe any significant differences in BMDs and BMCs at the studied bone sites among subjects classified by shellfish in the univariate and multivariate models either (all p > 0.05; data not shown).

Mean (SD) intakes of sea fish were 0.6 (0.8), 16.8 (9.8), and 64.7 (30.9) g/day in edible portion in the Q1, Q2–4, and Q5 groups, respectively. Sea fish consumption was significantly and inversely associated with age and YSM, but positively with height, physical activities, and dietary intakes of energy, protein, calcium, and PRAL (p < 0.05). No significant differences in body weight, BMI, education and work statues, smoking, and alcohol drinking were noted among the three sea fish groups (p > 0.1; Table 1).
Table 1

Characteristics of study subjects

Quintiles sea fish intakes

Characteristics

p values

Q1 (n = 129)

Q2–Q4 (n = 420)

Q5 (n = 136)

Mean

SD

Mean

SD

Mean

SD

ANOVA

Age, years

55.6

3.5

54.8

3.5

54.4

3.3

0.020

YSM, years

5.2

2.9

4.5

2.8

4.3

2.7

0.016

Height, cm

152.4

5.1

153.7

5.2

154.2

5.7

0.019

Weight, kg

56.8

8.6

57.3

8.7

58.8

9.6

0.143

BMI, kg/m2

24.4

3.5

24.3

3.6

24.7

3.7

0.477

Energy expenditure in physical activities, kcal/day

600

332

581

263

692

368

0.001

Dietary intake per day

       

 Sea fish, g/day

0.6

0.8

16.8

9.8

64.7

30.9

 Freshwater fish, g/day

25.7

31.5

24.5

25.0

32.7

32.2

0.012

 Other fish and shellfish, g/day

4.55

6.67

5.80

8.98

10.18

12.22

0.000

 Calorie, kcal/day

1,052

326

1,176

375

1,381

413

0.000

 Protein, g/day

49.5

19.4

58.3

22.1

79.4

25.8

0.000

 Protein (all fish), g/day

5.41

5.88

8.44

5.21

19.34

7.64

0.000

 Fat, g/day

22.7

10.3

26.2

12.2

33.6

14.9

0.000

 Carbohydrate, g/day

165.7

54.6

180.1

60.4

194.5

61.5

0.000

 Calcium, mg/day

476

229

516

225

669

264

0.000

 Vitamin D (sea fish), μg/100 g

0.06

0.08

1.71

1.00

6.61

3.16

0.000

 Vitamin D (freshwater fish), μg/100 g

0.87

1.07

0.83

0.85

1.11

1.09

0.012

 PRAL, mEq/day

3.12

6.47

5.82

7.36

9.84

8.47

0.000

 PRAL (all fish), mEq/day

2.49

2.78

3.86

2.46

8.93

3.93

0.000

 

%

 

%

 

%

 

(χ2-test)

Smoking, (>1/day)

1.6

 

1.9

 

0.7

 

0.643

Alcohol drinking–yes

4.7

 

3.6

 

4.4

 

0.820

Education

      

0.176

 No formal

11.6

 

5.2

 

8.1

  

 Primary

37.2

 

39.0

 

41.9

  

 Secondary

39.5

 

46.9

 

41.9

  

 Post secondary

11.6

 

8.8

 

8.1

  

Work status

      

0.111

 Housewife

69.5

 

61.6

 

53.7

  

 Professional/administrative/clerical

8.6

 

9.8

 

10.3

  

 Sales/services/blue collar

21.9

 

28.6

 

36.0

  

YSM years since menopause, PRAL potential renal acid load

The mean BMDs were significantly higher among subjects with higher sea fish consumption. Univariate analysis showed dose-dependent linear relations between sea fish intakes and BMDs and BMCs at the whole body, spine, and hip sites. p values for linear trend were less than or equal to 0.001 for BMDs and BMCs at all sites except for BMD at the lumbar spine (p = 0.029). The mean BMDs were 4.0% (whole body, p = 0.002), 4.1% (lumbar spine, p = 0.086), and 6.2% to 8.4% greater at the hip sites (all p < 0.01) in women belonging to Q5 of sea fish intakes (mean intakes of sea fish = 64.7 g/day) than those in Q1 (mean intake = 0.6 g/day; Q5). The mean differences in BMCs between the Q1 and Q5 groups were much more pronounced than those in BMDs. The mean BMC differences were 8.4% (whole body and spine), 6.7% (femur neck), and 11.2–12.1% (other hip sites; all p < 0.01; Table 2).
Table 2

Mean (SD) bone mineral density and bone mineral content at various sites by sea fish intake quintiles

 

Quintiles of sea fish intakes

% Diff.

ANOVA

Q1 (n = 129)

Q2–Q4 (n = 420)

Q5 (n = 136)

p value1

p value2

Mean

SD

Mean

SD

Mean

SD

BMD, g/cm2

Whole body

0.947

0.086

0.963

0.091

0.985

0.092*,***

4.01

0.002

0.001

Spine L1–L4

0.828

0.122

0.846

0.125

0.862

0.129

4.11

0.091

0.029

Total hip

0.782

0.115

0.800

0.108

0.834

0.108**,****

6.65

0.000

0.000

 Femur neck

0.661

0.098

0.681

0.095

0.702

0.108*

6.20

0.003

0.001

 Trochanter

0.575

0.095

0.593

0.092

0.623

0.107**,****

8.35

0.000

0.000

 Intertrochanter

0.938

0.147

0.966

0.133

1.006

0.129**,****

7.25

0.000

0.000

BMC, g

Whole body

1,581

232

1,640

260

1,715

260**,****

8.48

0.000

0.000

Spine L1–L4

43.63

7.87

45.33

9.31

47.31

10.06*

8.43

0.005

0.001

Total hip

22.86

4.30

23.58

4.16

25.41

4.65**,*****

11.15

0.000

0.000

 Femur neck

3.12

0.47

3.22

0.48

3.33

0.56*

6.73

0.004

0.001

 Trochanter

5.31

1.09

5.58

1.11

5.91

1.21**,****

11.30

0.000

0.000

 Intertrochanter

14.43

3.08

14.79

2.97

16.17

3.22**,*****

12.06

0.000

0.000

Q2–Q4: subjects of Q2, Q3, and Q4 groups were combined into one group

p value 1: p for the group difference; p value 2: p for linear trend

% Diff.: \( {\text{percentage difference}} = \left( {{\text{Q}}5 - {\text{Q}}1} \right)/{\text{Q}}1~ \times ~100\% \)

*p < 0.01 (compared with Q1)

**p < 0.001 (compared with Q1)

***p < 0.05 (compared with Q2–Q4)

****p < 0.01 (compared with Q2–Q4)

*****p < 0.001 (compared with Q2–Q4)

The dose-dependent linear relations between sea fish intake and BMDs and BMCs remained after adjustment for the potential confounders, but the strengths of association decreased moderately. The mean BMDs were 3.2% (whole body, p = 0.024), 2.6% (lumbar spine, p = 0.34), 4.7% (neck, p = 0.028), 5.0% (total hip, p = 0.005), 6.8% (trochanter, p = 0.002), and 5.8% (intertrochanter, p = 0.002) greater in the group Q5 than in the group Q1. The mean adjusted BMD differences were about 0.34 standard deviation (SD; whole body), 0.35 SD (total hip), 0.30 SD (femur neck), and 0.40 SD (trochanter and intertrochanter) of BMD at the specific bone sites (Table 3). More significant BMC differences between the Q5 and Q1 groups of sea fish intake were observed.
Table 3

Covariate-adjusted mean (SEM) bone mineral density and bone mineral content at various sites by sea fish intake quintiles

 

Quintiles of sea fish intakes

% Diff.

ANCOVA

Q1 (n = 129)

Q2–Q4 (n = 420)

Q5 (n = 136)

p value 1

p value 2

Mean

SEM

Mean

SEM

Mean

SEM

BMD, g/cm2

Whole body

0.952

0.008

0.963

0.004

0.982

0.008*

3.15

0.024

0.008

Spine L1–L4

0.834

0.010

0.846

0.006

0.856

0.010

2.64

0.340

0.145

Total hip

0.787

0.009

0.801

0.005

0.826

0.009**,****

4.96

0.005

0.002

 Femur neck

0.665

0.008

0.682

0.004

0.696

0.008*

4.66

0.028

0.008

 Trochanter

0.578

0.008

0.594

0.004

0.617

0.008**,****

6.75

0.002

0.001

 Intertrochanter

0.943

0.011

0.966

0.006

0.998

0.011**,****

5.83

0.002

0.000

BMC, g

Whole body

1,598

21

1,640

12

1,698

21**,****

6.26

0.004

0.001

Spine L1–L4

44.01

0.79

45.33

0.44

46.96

0.78*

6.70

0.033

0.009

Total hip

23.04

0.35

23.64

0.19

25.05

0.34***,*****

8.72

0.000

0.000

 Femur neck

3.14

0.04

3.22

0.02

3.30

0.04*

5.10

0.027

0.007

 Trochanter

5.35

0.09

5.58

0.05

5.83

0.093**

8.97

0.000

0.000

 Intertrochanter

14.55

0.25

14.83

0.14

15.92

0.25***,******

9.42

0.000

0.000

ANCOVA: adjusted for years since menopause, body mass index, energy expenditure in physical activities, and dietary energy, calcium, non-fish protein, and non-fish PRAL. Only significant covariates (except for dietary energy) were kept in the final model

Q2–Q4: subjects of Q2, Q3, and Q4 groups were combined into one group

p value 1: p for the group difference; p value 2: p for linear trend

% Diff.: \( {\text{percentage difference}} = \left( {{\text{Q}}5 - {\text{Q}}1} \right)/{\text{Q}}1~ \times ~100\% \)

*p < 0.05 (compared with Q1)

**p < 0.01 (compared with Q1)

***p < 0.001 (compared with Q1)

****p < 0.05 (compared with Q2–Q4)

*****p < 0.01 (compared with Q2–Q4)

******p < 0.001 (compared with Q2–Q4)

Logistic regression analyses showed significant dose-dependent association between sea fish intake and osteoporosis occurrence (Table 4). After adjusting for the potential confounders, p values for trend were less than 0.01 at the whole body, total hip, femur neck, and intertrochanter. The odds ratios (95% confidence interval (CI)) for osteoporosis in Q5 were 0.23 (0.08–0.66), 0.12 (0.03–0.59), 0.06 (0.01–0.44), and 0.09 (0.01–0.81) compared with those in Q1 at the whole body, total hip, femur neck, and intertrochanter in the multivariate models, respectively. The favorable association was more significant in univariate analyses. Similar trend was observed at the lumbar spine, but no significant association was observed. No significant association between freshwater fish intake and osteoporosis risk was noted either (data not shown).
Table 4

Odds ratios (95% CI) of osteoporosis for sea fish intake levels

 

Univariate model

Multivariate modela

Q1

Q2–Q4

Q5

pfor trend

Q1

Q2–Q4

Q5

pfor trend

Whole body

1

0.56** (0.32–0.98)

0.19*** (0.07–0.51)

<0.001

1

0.656 (0.36–1.18)

0.23*** (0.08–0.66)

0.018

Spine L1–4

1

0.83 (0.49–1.40)

0.66 (0.34–1.30)

0.177

1

0.92 (0.53–1.60)

0.86 (0.42–1.75)

0.678

Total hip

1

0.47** (0.25–0.87)

0.09*** (0.02–0.39)

<0.001

1

0.55* (0.28–1.07)

0.12*** (0.03–0.59)

0.003

Neck

1

0.35*** (0.17–0.73)

0.06*** (0.01–0.45)

<0.001

1

0.36*** (0.17–0.77)

0.06*** (0.01–0.44)

0.001

Trochanter

1

0.54 (0.24–1.19)

0.08** (0.01–0.66)

0.003

1

0.67 (0.29–1.55)

0.12* (0.01–1.06)

0.041

Intertrochanter

1

0.39** (0.18–0.84)

0.07** (0.01–0.54)

<0.001

1

0.46* (0.20–1.05)

0.09** (0.01–0.81)

0.007

Osteoporosis of the whole body, spine, and hips was defined as T-scores ≤ −2.5 SD using the oriental population references established by the manufacturer of [20]

*p < 0.1

**p < 0.05

***p < 0.01

aCovariates: same as those in Table 3; method for covariates = forward stepwise, likelihood ratio; F-to-entry and remove criteria were 0.05 and 0.10

Discussion

In this population-based cross-sectional study, we found that greater sea fish intake was dose-dependently associated with greater BMDs, BMCs, and lower osteoporosis risk at the whole body and hip sites (total hip, trochanter, and intertrochanter) among postmenopausal Chinese women with a mean sea fish intake of 23.3 g/day. However, we did not find any significant association between the intakes of freshwater fish and/or shellfish and BMD or osteoporosis risk at the studied bone sites. Leslie et al. [21] reported that the total hip BMD was the best marker for predicting total osteoporotic fractures. One SD decrease in the total hip BMD was associated with an 85% (95% CI, 70–101%) increase in the risk of total osteoporotic fractures [21]. According to the predictive formula of Leslie et al., the mean difference of 64 g/day sea fish intake between Q5 and Q1 groups found in our study would result in a 30% difference in osteoporotic fracture risk. Fish is a major source of animal food in traditional Chinese diet in Hong Kong and other coastal regions of mainland China. Our findings suggest that an increase in sea fish consumption may be an efficient way for the prevention of osteoporosis in this population.

In general, our findings were consistent with the majority of the previous studies [1114, 22]. Zalloua et al. [13] found that total body BMD was significantly higher in peri- and postmenopausal women with high consumption of seafood (>250 g/week) than those with lower consumption (<250 g/week seafood; p < 0.01). A similar association between sea-food intake and total hip BMD was found in postmenopausal women. However, the study found no significant association between sea-food intake and BMDs in men or a significant correlation between freshwater fish consumption and BMDs either. In another cross-sectional study, Terano [23] compared the BMD of men and women aged 38–80 years living in a fishing village in Japan with that of age-matched urban control subjects and found that the village women consuming more seafood had greater radial BMD than did the controls. Feskanich et al. [12] reported a dose-dependent lower risk (p for trend = 0.03) of hip fracture in postmenopausal women with higher intake of dark fish. A study in Japan also reported a lower hip fracture risk among subjects with higher fish intake [11], but the study by Fujiwara et al. did not find such an association [14].

Our study population had a mean daily intake of 56 (SD: 45; in edible portion) g of fish and shell fish. Freshwater fish and sea fish accounted for 50% and 43% of the total seafood intake, respectively. A greater protein intake also tended to be associated with better bone mass in this study (data not shown). The positive effect of a high intake of sea fish might be partially explained by a displacement of animal protein or of low-quality food items by the intake of sea fish. Studies have shown that a high acid load is associated with lower bone mass [24]. The PRAL values (mEq/100 g) were slightly lower in sea fish (8.29) and freshwater fish (8.25) than those in lean meat (pork and beef, 9.9, 10.0) and poultry (9.15) [15, 19]. However, a high intake of sea fish could not greatly reduce PRAL values when sea fish replaced other animal foods, and its high intake might increase PRAL values via an increase in protein intake. Therefore, PRAL was unlikely to play an important role in the beneficial effect of sea fish on bone mass in this study population.

It is still uncertain why sea fish but not freshwater fish may have a beneficial effect on bone health. The differential effect between sea- and freshwater fish seemed unlikely to be related to high quality protein, calcium, and dietary PRAL, because these contents were similar in the two types of fish. Sea fish, especially the oily fish, contain relatively higher concentration of vitamin D than freshwater fish. The mean values of vitamin D were 10.2 and 3.4 (µg/100 g) in our studied sea fish and freshwater fish [16]. It has long been demonstrated that vitamin D is a key nutrient for the maintenance of optimal bone mass. Sea fish have a richer content of long-chain n-3 fatty acids of eicosapentaenoic acid (EPA) and docosahexaenoic (DHA) than freshwater fish. The mean total contents of EPA and DHA were 8.5 and 4.4 (g/100 g) in our studied sea fish and freshwater fish [15]. Many studies have shown the beneficial effects of n-3 fatty acids on bone health in both human and animals [25]. Högström et al. [10] found that both concentrations of n-3 fatty acids and DHA of serum phospholipids were significantly and positively associated with total body and spine BMD in healthy young men and BMD accrual between 16 and 22 years of age. Another study observed that a higher ratio of n-3 to n-6 fatty acids was associated with higher BMD at the hip in 1,532 community-dwelling men and women aged 45–90 years [9]. A few animal studies have shown that a dietary intake of long-chain n-3 PUFAs or fish oil correlated with higher BMD, lesser bone loss, and reduced bone turnover [2628]. A wide range of mechanisms may mediate the beneficial effects of dietary n-3 fatty acids on bone, including increases in calcium absorption and synthesis of bone collagen, decreases in urinary calcium loss [2931], and decreases in prostaglandin synthesis [32] and inflammatory cytokines [26, 33]. Dietary long-chain n-3 fatty acids EPA and DHA may partially replace the n-6 fatty acids in the membranes of platelets, erythrocytes, neutrophils, monocytes, and liver cells. The change in the ratio of n-6/n-3 fatty acids in membranes may decrease the production of inflammatory cytokines of interleukin-1, interleukin-6, and tumor necrosis factor-α that play a pivotal role in the pathogenesis of osteoporosis [26, 33].

This study has some limitations. It is a cross-sectional study, so the causal relationship cannot be determined due to the unclear time sequence between the exposure and outcome. The validity of this study largely depended on if the long-term consumptions of sea fish could be estimated based on the recent 1-year intake. Generally, adults maintain a relatively stable dietary habit over a long period. Macdonald et al. have reported small, though statistically significant, differences in intakes over 5 to 7 years for most nutrients (<8% change) in middle-aged women [34]. Our findings based on the FFQ method have also noted good long-term reproducibility, with 80% having within one unit quintile agreements in sea fish intake. Current habitual intake might reasonably reflect long-term habitual intake in our population.

DHA and EPA may play an important role in the beneficial effect of sea fish on bone mass. Because our limited questionnaire did not differentiate between the types of sea fish consumed, we are unable to accurately estimate the dietary intake of DHA and EPA. However, this limitation is more likely to underestimate rather than overestimate the association between sea fish consumption and bone mass. In addition, due to limitation of availability of serum or blood cell samples and study cost, we did not have data on serum 25-OH-vitamin D3 and serum contents or proportionate ratio of DHA or EPA of red blood cell membrane to provide more support on their roles. Further studies are needed to address if the positive effect of sea fish was mediated by higher intakes of vitamin D and n-3 fatty acids.

In conclusion, our findings show that greater sea fish consumption is significantly and independently associated with higher BMDs and lower osteoporosis risk at the whole body and most hip sites in postmenopausal women, even after controlling for the potential confounders. Increasing consumption of sea fish may be beneficial for the prevention of osteoporosis in postmenopausal women.

Conflicts of interest

None.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2009