|Home | About | Journals | Submit | Contact Us | Français|
Although there is some epidemiologic evidence that soy may reduce risk of osteoporotic fracture in women, it is not known whether this risk reduction also occurs for men. The authors examined gender-specific associations between soy intake and hip fracture risk in the Singapore Chinese Health Study, a prospective cohort of 63,257 Chinese living in Singapore. At recruitment between 1993 and 1998, each subject was administered a food frequency questionnaire and questions on medical history and lifestyle factors. As of December 31, 2006, 276 incident cases of hip fracture in men and 692 cases in women were identified via linkage with hospital discharge databases. For both genders, hip fracture risk was positively associated with cigarette smoking and was inversely associated with body mass index. There was a statistically significant association of tofu equivalents, soy protein, and isoflavones with hip fracture risk among women but not among men. Compared with women in the lowest quartile of intakes for tofu equivalents (<49.4 g/day), soy protein (<2.7 g/day), and isoflavones (<5.8 mg/1,000 kcal/day), those in the second–fourth quartiles exhibited 21%–36% reductions in risk (all P<0.036). Risk levels were comparable across the second, third, and fourth quartiles of soy intake categories.
Osteoporosis often leads to fractures of the hip, spine, or wrist after minor injury, with resulting loss of function and disability. Of these 3 types, hip fractures have the most devastating results, and the incidence rate of hip fracture is commonly used to indicate the prevalence of osteoporosis (1, 2). There is marked geographic and ethnic variation in the incidence of hip fracture worldwide, with the highest rates among Caucasian women in Western nations, most notably those in North America, Europe, and Oceania, and the lowest rates among women in Asian populations (1, 3). Nevertheless, hip fracture is also fast becoming a major public health problem in Asia because the incidence is rising rapidly, and it has been projected that, by 2050, 50% of all hip fractures in the world will occur in Asia (4).
The contributing factors for the increasing trend in the rate of hip fractures are believed to be an aging population and the increasingly sedentary lifestyle of the population as a whole (1, 5). Epidemiologic studies conducted among Western populations have also demonstrated an increased risk of hip fractures associated with smoking (6, 7) and lower body mass index (8, 9). Other studies have also investigated the possible impact of dietary factors such as calcium (10, 11), vitamin D (11), vitamin A (12), folate (13), and tea (14) on risk of osteoporosis. Because the overwhelming majority of these studies are conducted in Western populations, there is a need for more studies in Asian populations that have distinct anthropometric measurements and differ in their dietary and other habits relative to their Western counterparts (15).
Despite the differences in age-adjusted osteoporosis rates in countries across the world, the female-to-male ratio remains very similar in all countries, ranging from 2.5 in Asian countries to 2.9 in the US white population (1, 16). Explanations for this disparity in rates include gender differences in biomechanical factors that contribute to bone strength and size and the bone-remodeling effects of sex hormones on bone (17). Interestingly, a recent study on risk of self-reported bone fracture among Chinese women in Shanghai identified intake of soy, a common Asian food, as a protective factor for osteoporotic fractures (18). Other studies conducted among Asian women in Hong Kong and Japan that examined changes in bone mineral density found that high soy food intake was associated with a decrease in bone loss in postmenopausal women (19–21). In these studies, the protective effect of soy isoflavones in osteoporosis was attributed to the estrogenic effects of these soy phytoestrogens in postmenopausal women. Since men, unlike women, experience a much slower and more gradual decline in sex hormone production with age (22), it is plausible that soy isoflavones may have a different effect on hip fracture risk in elderly men.
In the present study, we examined, within a prospective cohort of 63,257 Chinese men and women (the Singapore Chinese Health Study), the potential risk factors for hip fracture, including lifetime use of tobacco; current consumption of alcohol and various beverages; and diet assessed via a validated, 165-item food frequency questionnaire. We were particularly interested in the gender-specific associations between dietary soy and risk of hip fracture.
The subjects were participants in the Singapore Chinese Health Study, a population-based cohort of 63,257 Chinese women and men aged 45–74 years enrolled between April 1993 and December 1998 (15). We restricted study subjects to the 2 major dialect groups of Chinese in Singapore—the Hokkiens, who originated from the southern part of Fujian Province, and the Cantonese, who came from the central region of Guangdong Province. The subjects were residents of government housing estates; 86% of the entire Singapore population resided in these facilities during the enrollment period. The institutional review boards at the National University of Singapore and the University of Minnesota approved this study.
At recruitment, a face-to-face interview was conducted in the subject's home by a trained interviewer using a structured questionnaire to request information including demographics, use of tobacco, menstrual (including menopausal status) and reproductive (including use of hormone replacement therapy) history (women only), and medical history, as well as a dietary component assessing current intake patterns. For the assessment of physical activity, subjects were asked to estimate the number of hours per week spent on moderate activities such as brisk walking, bowling, and tai chi and on strenuous sports such as jogging, swimming laps, or aerobics. The questionnaire also included a semiquantitative food frequency questionnaire assessing current intake patterns, which had previously been validated against a series of 24-hour diet recalls among a subpopulation drawn randomly from among the cohort participants (15). The correlation coefficients for tofu, isoflavone, and soy protein between the food frequency questionnaire and the 24-hour dietary recall data were 0.63, 0.39, and 0.32, respectively (23).
Seven soy foods are common in the Singapore Chinese diet: plain tofu, taupok, taukwa, foopei, foojook, tofu-far, and soybean drink (15, 24). Based on the fact that 100 g of cooked tofu, taukwa, foopei, and foojook contains 89 g, 69 g, 58 g, and 54 g of water, respectively, we determined that 1 g of taukwa is equivalent to 2.8 g (31/11) of tofu. Similarly, 1 g of foopei and foojook are equivalent to 3.8 g (42/11) and 4.2 g (46/11) of tofu, respectively. Because 100 g of soybean drink contains 92 g of water, we determined that 1 g of soybean drink is equivalent to 0.73 g (8/11) of tofu (24). For a given subject, total soy intake was the summation of all soy foods expressed in units of plain tofu equivalent, and total soy protein intake was the summation of the protein contents of soy foods listed in the Singapore Food Composition Database (15). In addition, we previously measured concentrations of genistein, daidzein, and glycitein in market samples of common soy foods in Singapore (24). Total soy isoflavone intake for a given subject was computed based on the food frequency questionnaire and the summation of the genistein, daidzein, and glycitein content of all the 7 soy foods in the Singapore Food Composition Database. Previously, a statistically significant association between urinary levels of isoflavones and estimates of dietary soy intake using the food frequency questionnaire was demonstrated (25).
Cases of hip fracture among cohort members were identified through record linkage of cohort files and databases of the MediClaim System, which captures inpatient discharge information from all hospitals in Singapore, both public and private. All cases were verified by records of the appropriate surgical procedures or manual review of medical records. Records of death were captured through linkage with the Singapore Registry of Births and Deaths. As of December 31, 2006, after excluding 4 cases of traumatic fractures from road traffic accidents and 1 case resulting from malignant metastasis to the femur, we identified 1,071 hip fracture cases through record linkage. We excluded from statistical analysis 103 prevalent cases of hip fracture that occurred before subjects’ recruitment to the cohort. Thus, 968 subjects with fractures and 62,186 without fractures were included in the final analysis.
For each study subject, person-years were counted from the date of baseline interview to the date of hip fracture diagnosis, date of death, or December 31, 2006, whichever occurred first. We used the chi-square test to examine the difference in distributions between cases and noncases by dialect group (Cantonese, Hokkien) and level of education (no formal education, primary school, secondary school or higher), and we used Student's t test to examine the difference in age at recruitment. Proportional hazards regression methods were used to examine the exposure–hip fracture associations for men and women separately and for both sexes combined with adjustment for gender. To adjust for the potential confounding effect of other demographic and exposure factors on the soy–hip fracture association, the following variables were included in all regression models: age at recruitment (years), year of recruitment, dialect group, level of education, weekly vigorous work or strenuous sports (yes, no), menopausal status at recruitment (women only; yes, no), and use of hormone replacement therapy at recruitment (women only; yes, no).
When examining the differential association between dietary calcium/soy and hip fracture risk with different duration of follow-up, we stratified study subjects (i.e., fracture cases and person-years at risk) by various durations of follow-up. Given that there was no difference in the exposure–hip fracture association between subgroups with less than 5 years of follow-up and those with more than 5 years of follow-up after enrollment, the tables in this paper show the results with follow-up of 0–5 years and more than 5 years for simplicity of presentation. The models examining the associations with dietary soy and calcium included the following additional covariates: body mass index (kg/m2, as a continuous variable), number of cigarettes smoked per day (never smoker, 1–12, 13–22, ≥23), number of years of smoking (never smoker, 1–19, 20–39, ≥40), number of years since quitting smoking (continuous smoker, <1, 1–4, 5–19, ≥20, never smoker), consumption of all vegetables and related juices (g/1,000 kcal per day), consumption of all fruits and related juices (g/1,000 kcal per day), and total daily ethanol intake (g/day). Quartiles were based on the distribution of the whole cohort. The strength of a given association was measured by a hazard ratio and its corresponding 95% confidence interval and 2-sided P value. Linear trend tests for exposure-disease associations were based on the ordinal values of the quartiles. Statistical computing was conducted by using SAS version 9.1 software (SAS Institute, Inc., Cary, North Carolina).
The present analysis was based on 667,578 person-years of follow-up and a total of 968 patients with incident, nontraumatic hip fracture (276 men, 692 women). The mean time interval between cohort enrollment and hip fracture diagnosis was 7.1 years (standard deviation, 3.5), and the mean ages at fracture were 71.4 years (standard deviation, 7.5) for the 276 males cases and 72.7 years (standard deviation, 6.9) for the 692 female cases. The incidence rates of hip fractures per 100,000 person-years in the cohort, adjusted to the age structure of the whole cohort, were 96.1 for men and 181.6 for women. Among both men and women, hip fracture cases were older at recruitment and less educated compared with the rest of the cohort (both P<0.001) (Table 1).
Current cigarette smoking was associated with a similarly increased risk of hip fractures for both men and women, although the 23% increase in risk did not reach statistical significance for men. Among smokers of either gender, risk increased significantly with increasing duration of smoking (for women) or with increasing number of cigarettes smoked per day (for men) (Table 2). Body mass index was inversely related to risk of hip fractures for both men and women, although the association was significant and stronger for men compared with women. Hip fracture risk was unrelated to intake of black tea or green tea for either men or women. Furthermore, consumption of alcohol or coffee was not associated with hip fracture for either gender (results not shown).
In our population, only 2% of men and 4% of women took calcium supplements at least once a week. Overall, total calcium intake and the use of calcium supplements were positively and significantly associated with hip fracture risk for women but not for men (Table 3). However, these positive associations for women were largely confined to those with a relatively short (<5 years) duration of follow-up (Table 3).
For men, we found no association between hip fracture risk and dietary intakes of tofu equivalents, soy protein, and soy isoflavones (Table 4). However, for women, there was a statistically significant, approximately 30% reduction in risk for women with a moderate-to-high intake of the 3 soy intake categories compared with those whose intake was low (all P < 0.036) (Table 4). Relative to women in the lowest quartile of soy intake, those in the second–fourth quartiles combined had hazard ratios of 0.69 (95% confidence interval: 0.60, 0.81) for tofu equivalents, 0.69 (95% confidence interval: 0.59, 0.81) for soy protein, and 0.72 (95% confidence interval: 0.62, 0.84) for soy isoflavones. No further reduction in risk was observed for women in the second, third, and fourth quartiles of soy intake because risk levels were comparable across the second, third, and fourth quartiles of tofu equivalents (P = 0.224), soy protein (P = 0.179), or soy isoflavones (P = 0.481). We further examined the soy-risk association for women by subjects’ duration of follow-up and observed a similar reduction in risk estimates with increasing intake of tofu equivalents, soy protein, and soy isoflavones in both groups, although the results were of statistical significance only in the larger group with a longer duration of follow-up (>5 years) (Table 5).
The present study shows that, for Singapore Chinese, intake of soy is associated with reduced risk of osteoporotic hip fracture for women but not for men. Total calcium intake from dietary sources and supplements was unrelated to hip fracture risk for men. Although calcium intake showed a statistically significant, positive association with hip fracture risk for women, it was largely confined to those with a relatively short duration of follow-up.
To our knowledge, this prospective cohort study is the first that examines risk factors for osteoporotic hip fracture for both genders in a nonwhite population living in Asia. For many factors, especially those related to diet and lifestyle, exposure ranges widely in this increasingly westernized society in Singapore. Other strengths of the study are its population-based design and the reduced likelihood of recall bias regarding exposure data since they were obtained prior to disease diagnosis. Singapore is a small city-state, with a system for easy access to specialized medical care. Since practically all hip fracture cases will seek medical attention immediately and be hospitalized for surgical hip hemiarthroplasty, our case ascertainment through linkage with the comprehensive, nationwide hospital database can be considered complete. The information from the database also enabled us to differentiate prevalent cases from incident cases using the dates of recruitment into the study and admission to the hospital after the fracture.
Limitations of this study include possible misclassification of soy intake from use of a food frequency questionnaire. However, such misclassifications, if any, are nondifferential and are more likely to result in underestimation of the true hazard risk. Another limitation is the lack of study subjects’ lifetime history of dietary intake. We assessed dietary intakes up to the time of cohort enrollment, which may have been years before hip fracture in our cases. The change in subjects’ dietary habits from the time of baseline assessment to the censoring date could result in attenuation of observed risk estimates for hip fracture associated with dietary factors studied.
The novel finding in this study is the inverse, dose-dependent association between soy and osteoporotic hip fracture risk found for women but not for men. Soy food contains soy isoflavones, which belong to a class of plant compounds called phytoestrogen and may account for the potential bone-protective effects associated with soy food. A meta-analysis of 10 short-term clinical trials involving 608 subjects concluded that oral isoflavone intervention significantly attenuates bone loss of the spine in menopausal women (26). In fact, as naturally occurring selective estrogen receptor modulators, isoflavones may exert estrogenic effects when the endogenous estrogen level is low, that is, during menopause, and not be effective in premenopausal women.
In support of this possibility is evidence that the more osteopenic the bones are in postmenopausal women, the more effective isoflavone supplementation is in reducing bone loss. A double-blind, placebo-controlled, randomized trial among Chinese women aged 48–62 years showed that the reduction in bone loss from isoflavone supplementation was observed among only those women with lower initial baseline bone mineral content, who were also in later menopause (20, 27). Consistently, cross-sectional surveys among Chinese women showed that the dose-response relation between higher bone mineral density values of the hip and increasing soy protein intake was observed among postmenopausal but not premenopausal women (28) and that, among postmenopausal women, the association was stronger for women with a longer duration of postmenopausal status (19).
Epidemiologic studies of postmenopausal women in Asian populations that consume a soy-rich diet in Japan and Hong Kong have shown that high intake of soy protein is associated with higher bone mineral density, particularly in the hip region (19, 29). In a recent cohort study involving 24,403 postmenopausal Chinese women in Shanghai (the Shanghai Women's Health Study), hip fracture risk associated with the second and higher quintiles of soy intake (≥5 g/day of soy protein) was 30% lower than that associated with the lowest quintile of intake (18). The present study also noted a similar threshold effect between soy protein and hip fracture risk. Furthermore, the levels of threshold dose and its corresponding magnitude of risk reduction are comparable across these 2 independently conducted, geographically distinct cohort studies of Asian Chinese women. Thus, the level of soy isoflavones necessary for bone health in women may be much lower than the doses administered to subjects of intervention studies conducted to date (26).
The effects of soy in preventing osteoporosis in men have not been determined. A controlled, doubled-blind trial of 145 men and women of comparable ages, 50–80 years, in the United States showed that soy protein supplementation had significantly greater effects on spine bone mineral density in women than in men, although there was no significant effect in either gender for hip bone mineral density (30). Another randomized, controlled trial among middle-aged and elderly men showed that soy protein supplementation did not affect serum bone-specific alkaline phosphatase level, which is considered a biomarker for osteoblastic activity (31). Such studies in humans are consistent with experimental studies using rodent models, which have shown that neonatal exposure to genistein, a prototype of soy isoflavones, increased bone marrow densities in the femurs and spines of adult female mice but had an effect in only the spines but not the femurs of male mice. The authors had suggested that genistein could have site-specific effects on skeletal health in males and that, although genistein acted as a selective estrogen receptor modulator in females, the mechanism of action of genistein in males was still unclear (32). Furthermore, other animal studies showing that genistein or soy isoflavones may reduce age-related bone loss by increasing osteoblastic activity in males had been conducted in orchidectomized male rodents. Such animal models are more likely to simulate abrupt and marked male hormonal deficiency (33–35) and probably do not reflect the gradual hormonal decline in aging men (22).
In a large, US study of 2,623 men, the age-related decline in testosterone and estradiol in older men was quite modest, and substantial variation in one hormone existed at any level of the other (36). Since testosterone is aromatized to estradiol, a feedback mechanism may exist between these hormones in men to maintain bone health in the physiology of aging. Consistent with this hypothesis, studies on elderly men in the Framingham study have shown that the risk of fracture is significantly increased in the subgroup with concomitantly low testosterone and estradiol levels and that both sex hormones may interact to produce a synergistic effect on hip fracture risk in men (37). If this is the case, soy isoflavone supplementation may not be sufficient to reduce fracture risk for elderly men with concurrently low levels of both hormones.
In our study, current smokers of both genders had about a 20% increase in hip fracture risk, which was lower than the pooled relative risk of 1.39 (95% confidence interval: 1.23, 1.58) from a meta-analysis of 50 independent studies. That meta-analysis also noted that smaller risk ratios were observed in countries closer to the equator relative to those further away from the equator, such as the United States and northern Europe (38). Singapore is one degree north of the equator, whereas the bulk of the 50 studies in the meta-analysis were conducted in Western countries in the Northern Hemisphere. Thus, our results on tobacco smoking are consistent with the overall findings reported in the literature. Similarly, the inverse relation between body mass index and hip fracture risk in our study concurred with the results from a meta-analysis of 12 population-based cohort studies (39).
In our study, increased calcium intake or the use of calcium supplements was associated with increased risk of hip fracture for women with less than 5 years of follow-up. Women who had a hip fracture within 5 years after enrollment could have increased their dietary calcium intake and use of calcium supplements at enrollment given the knowledge about their poor bone mineral density or risk of hip fracture. The absence of a positive calcium–hip facture association for both men and women with more than 5 years of follow-up supports our conjecture. Several other longitudinal studies failed to detect any adverse or beneficial effect of dietary calcium on bone loss or hip fracture, further supporting our findings in this study (11, 40–44).
In conclusion, we report a novel finding that dietary soy protects against hip fracture in women but not in men. A biologically plausible explanation for this observation has been proposed and warrants further investigation.
Author affiliations: Department of Epidemiology and Public Health, Yong Loo Lin School of Medicine, National University of Singapore, Singapore (Woon-Puay Koh); Department of Preventive Medicine, University of Southern California Keck School of Medicine, Los Angeles, California (Anna H. Wu); Masonic Cancer Center, University of Minnesota, Minneapolis, Minnesota (Renwei Wang, Jian-Min Yuan, Mimi C. Yu); Communicable Diseases Division, Ministry of Health, Singapore (Li-Wei Ang); Epidemiology and Disease Control Division, Ministry of Health, Singapore (Derrick Heng); and Division of Epidemiology and Community Health, University of Minnesota, Minneapolis, Minnesota (Jian-Min Yuan).
This study was funded by the National Medical Research Council, Singapore (grant NMRC/EDG/0011/2007) and the National Institutes of Health, United States (grants NCI R01 CA55069, R35 CA53890, and R01 CA80205).
The authors thank Siew-Hong Low of the National University of Singapore for supervising the field work of the Singapore Chinese Health Study and Kazuko Arakawa for the development and management of the cohort study database. They also thank the Ministry of Health in Singapore for their help in identifying hip fracture cases via database linkages.
Conflict of interest: none declared.