Osteoporosis has been considered as an important public health issue due to its increasing prevalence and contribution to morbidity and mortality in the aging population (1–3). Low bone mineral density (BMD), a characteristic feature of osteoporosis, and osteoporosis-related fractures together represent an important public health burden and contribute to socioeconomic disparities (1). Global estimate suggests that the number of hip fractures is expected to increase from 1.26 million in 1990 to 4.5 million by the year 2050 (2, 3). The International Osteoporosis Foundation reports that osteoporosis causes more than 8.9 million fractures annually, with an osteoporosis fracture occurring every 3 s (4) and is estimated to affect 200 million women worldwide (5). In the China Osteoporosis Prevalence Study (COPS), we previously reported that the prevalence of osteoporosis and vertebral fracture in the elderly population (aged 40 years or older) was 20.6% and 9.7% in women and 5.0% and 10.5% in men, respectively (6). As China constitutes one-fifth of the world’s population and a higher proportion of elderly individuals, our previous COPS indicated a higher prevalence of osteoporosis among women and fractures in the elderly in mainland China (6).

Several risk factors have been suggested to be associated with low BMD and risk of fracture, including genetic, endocrine, mechanical, and lifestyle factors such as smoking, alcohol consumption, physical activity, and calcium and vitamin D intake. Early evidence indicates that nutritional factors and dietary patterns play an important role in the development of osteoporosis and the risk of fractures (7–11). In China, few studies have investigated the relationship between dietary factors and the risk of osteoporosis and fractures. Three studies mainly focused on these associations in teenagers or college freshmen aged 11–20 years (12–14). Two studies have investigated dietary patterns associated with BMD (15) and fracture risk (16) among the elderly, but with relatively small sample sizes. However, the results from these studies are inconclusive, and the study participants were randomly selected from one to four cities in mainland China.

In light of the limited research and literature gap regarding the association between dietary patterns and BMD as well as fracture risk among the elderly Chinese population, we analyzed data from the COPS (6), the largest nationwide population-based study conducted in China. The participants enrolled in COPS were randomly sampled from almost all representative regions of mainland China. Building upon the COPS, we aimed to investigate the hypothesis that different dietary patterns are associated with BMD and the risk of fractures and that this association may vary between elderly women and men.

Of 17,489 participants, there were 7,394 (42.3%) men and 10,095 women with a mean age of 57.5 ± 10.4 and 57.1 ± 9.9 years old, respectively. A total of 725 (4.15%, 725/17,489), 805 (9.56%, 805/8,423), and 331 (3.93%, 331/8,423) participants were diagnosed with clinical fracture in the past 5 years, vertebral fracture, and vertebral fracture grade 2 or above, respectively. The means and SDs of peak BMDs at the TH, FN, and L1–4 are 0.84 ± 0.15 g/cm², 0.74 ± 0.13 g/cm², and 0.92 ± 0.18 g/cm², respectively. Table 1 shows the distribution of characteristics among men and women who participated in this study. Compared to women, men were more likely to have higher BMD at the TH, FN, and L1–4 (p < 0.0001), have a lower prevalence of clinical fracture and vertebral fracture (p = 0.0003, 0.002, and 0.21), be older (p = 0.003), be a smoker (p < 0.0001), be a drinker (p < 0.0001), have a higher BMI (p = 0.011), have a higher serum vitamin D level (p < 0.0001), and reside in a rural area (p < 0.0001), and men were less likely to have a higher education level (p = 0.0009), glucocorticoid use > 3 months (p = 0.036), and calcium supplementation (p < 0.0001). Distributions of family history of fracture and residence area in the north or south of China were similar between men and women.

In the principal component analysis ( Supplementary Table 1 ), the most prominent dietary pattern (eigenvalue: 2.4, total variance explained: 14.1%) was characterized by the high consumption of carnivorous food. This pattern had the highest positive loadings for poultry, livestock meat, pork, and seafood. The second most prominent pattern represented a high intake of vegetarian food (eigenvalue: 1.4, total variance explained: 8.0%). This pattern had the highest positive loadings for grain, vegetables, bean products, and tubers. The third dietary pattern (eigenvalue: 1.3, total variance explained: 7.7%), which had high levels of dairy, fruit, and egg, featured by milk, yogurt, fruits, and egg. The fourth dietary pattern (eigenvalue: 1.2, total variance explained: 7.2%), which had high intake of beverage and fried food, was featured by fruit juice, soda, fried food, and vegetable juice.

Regression coefficients (β), 95% CIs, p-values, and p for trends of associations between each dietary pattern and BMD are shown in Table 2 . Compared to the lowest quartile level, higher quartile levels in “carnivorous”, “vegetarian”, and “dairy, fruit, and egg” patterns were associated with significant increases in BMD at the TH and FN with significant dose–response relationships (p _trend < 0.0001, 0.0096, and <0.0001 in the above three patterns against with TH; 0.0029, 0.0084, and <0.0001 in the above three patterns against with FN), respectively. Similar positive associations were observed between higher quartile levels in “carnivorous” and “dairy, fruit, and egg” patterns and significant increases in BMD at L1–4 with significant dose–response relationships (p _trend < 0.0001). However, higher quartile levels in the “beverage and fried food” pattern were associated with significant decreases in BMD at the FN (β = −11.11 and −8.45, p = 0.02 and 0.05 for the second and third quartile levels, respectively) and L1–4 (β = −11.52 and −16.80, p = 0.03 and 0.002 for the second and third quartile levels, respectively), with no significant dose–response relationship observed.

The tests for interaction terms between sex and dietary patterns associated with BMD showed that sex significantly interacts with “carnivorous”, “vegetarian”, and “beverage and fried food” patterns on TH (p _interaction = <0.0001, 0.02, and 0.02), “carnivorous” and “beverage and fried food” patterns on FN (p _interaction = <0.0001 and 0.02), and “carnivorous”, “dairy, fruit, and egg”, and “beverage and fried food” patterns on L1–4 (p _interaction = <0.0001, 0.02, and 0.005). After stratifying by sex, stronger positive associations between “carnivorous”, “vegetarian”, and “dairy, fruit, and egg” patterns and BMD at the TH and FN, as well as stronger inverse associations between “beverage and fried food” pattern and BMD at L1–4 were observed among women ( Table 3 ). In contrast, among men, only the highest quartile of “dairy, fruit, and egg” pattern was associated with a significant increase in BMD at the FN and L1–4; the highest quartile of “carnivorous” patterns was associated with a significant increase in BMD at L1–4 ( Supplementary Table 2 ). Additionally, those observed significant increases (β) in BMD among men were smaller than those among women and the overall population.

Table 4 presents the associations between dietary patterns and risk of clinical and vertebral fractures in the overall population. Compared to the lowest quartile of the “carnivorous” pattern, the highest quartile level was associated with a reduced risk of clinical fracture in the past 5 years (OR = 0.63, 95% CI 0.42–0.95, p _trend = 0.032). A similar protective effect of higher quartile levels of carnivorous pattern on vertebral fracture was observed as well (OR = 0.61, 95% CI 0.43–0.85 for the third quartile; 0.66, 0.45–0.97 for the fourth quartile; p _trend = 0.017).

After the test for interaction terms, the significant interaction was observed only between sex and the “dairy, fruit, and egg” pattern associated with the risk of clinical fracture in the past 5 years (p _interaction = 0.002), but not vertebral fracture. When conducting stratified analysis by sex, among women ( Table 5 ), similar inverse associations were found in higher levels of carnivorous food intake and the risk of clinical and vertebral fractures with a significant dose–response relationship (p _trend = 0.0083 and 0.013). Moreover, an inverse association was found in the highest level of “carnivorous” and “vegetarian” patterns and the risk of vertebral fracture of grade 2 or above, respectively (OR = 0.45, 95% CI 0.20–0.99, p _trend = 0.038; 0.48, 0.25–0.93). However, in the “dairy, fruit, and egg” pattern, the third and fourth quartiles increased the risk of clinical fracture (OR = 2.50, 95% CI 1.61–3.88 for the third quartile; 1.99, 1.24–3.17 for the fourth quartile; p _trend = 0.0002), whereas, among men ( Supplementary Table 3 ), only the second quartile of carnivorous food intake was observed to reduce the risk of clinical fracture (OR = 0.48, 95% CI 0.26–0.88), as compared to the first quartile. In addition, the highest quartile of “beverage and fried food” was associated with increased risk of clinical fracture (OR = 1.64, 95% CI 1.06–2.54). No significant association was observed between four dietary patterns and the risk of overall vertebral fracture or vertebral fracture of grade 2 or above in men.

Finally, we performed a sensitivity analysis among postmenopausal women ( Supplementary Tables 4 , 5 ). As compared to the results in women, we observed similar positive associations between “dairy, fruit, and egg” and “beverage and fried food” patterns and BMD at the FN and L1–4, respectively. Particularly, stronger positive associations between carnivorous and vegetarian diets on high BMD at the TH, FN, and L1–4 were found. Additionally, stronger associations between carnivorous diet and reduced risk of clinical fracture, vertebral fracture overall, and vertebral fracture grade 2 or above were also found in postmenopausal women. Similar associations were observed after additional adjustment for physical activity and calcium supplementation (data not shown).

We identified four dietary patterns and observed positive associations between “carnivorous”, “vegetarian”, and “dairy, fruit, and egg” patterns and high BMD at the TH, FN, and L1–4, and inverse associations between the “carnivorous” pattern and risk of both clinical fracture in the past 5 years and vertebral fracture in the overall population. Particularly, stronger associations were seen among women. Additionally, higher intake of beverage and fried food was associated with an increased risk of clinical fracture in men.

To the best of our knowledge, this study is the first to identify dietary patterns within a large, nationwide population-based study in China and to explore these patterns associated with BMD and fracture risk, including differences of these associations between elderly women and men. Prior research in China involving 282 postmenopausal women found that a “cereal grains–fruits” pattern was associated with lower spine and hip BMD, while a “milk–root vegetables” pattern was associated with higher hip BMD (15). Studies in other populations and countries have also observed an association between dietary patterns and BMD, with healthy patterns (prudent/healthy) being associated with higher BMD and unhealthy patterns (Western/unhealthy) being associated with lower BMD. Healthy patterns were characterized by high intakes of fruits, vegetables, low-fat dairy, fish, legumes, high-fiber bread, and poultry, while unhealthy patterns were characterized by red meat, soft drinks, fast food, and sweets (7–10). The current study found similar associations with previous research, with high intakes of vegetarian diets, dairy, fruit, and egg being associated with a higher BMD at the TH, FN, and L1–4. High intakes of beverage and fried food were associated with a lower BMD at the FN and L1–4. Yet, after stratifying by sex, similar or stronger associations were particularly observed in women and postmenopausal women. However, in men, only the highest quartile of dairy, fruit, and egg was positively associated with BMD at the FN and L1–4, suggesting a potential threshold effect or a weaker association in men.

Dairy/milk consumption might have a beneficial effect on bone health for women aged around the menopausal period, which has been demonstrated by some randomized controlled trails (25–28). We enrolled women aged 40 and above, most of whom experienced pre-menopause, menopause, or post-menopause. Early studies indicate that postmenopausal women are at a higher risk of developing osteoporosis and experiencing rapid bone loss due to the decrease in estrogen, which is known to inhibit bone resorption by inducing osteoclast apoptosis (15, 29). The “dairy, fruit, and egg” pattern in the study could be seen as a significant source of calcium, vitamin D, and proteins, all of which are essential for maintaining bone health. Therefore, higher intakes of this pattern might play a significant role in bone health, with gender-specific effects.

Vegetarian diets are rich in various nutrients that are beneficial for bone health, including magnesium, potassium, vitamin C, vitamin K, antioxidant, and anti-inflammatory phytonutrients (30). Evidence suggested that serum magnesium concentration has been shown to be significantly lower in women with osteoporosis than in those with normal BMD (31, 32); potassium intake was associated with higher BMD in men and women (32) and with lower BMD loss in men (33); higher vitamin C intake has been linked to a 44% lower risk of hip fracture (34). Diets rich in phytonutrients may protect bone, but the mechanisms of action have not been conclusively shown (35). However, there was limited evidence suggesting that vegetarians/vegans were more likely to have a lower BMD and a higher risk of developing fractures (36, 37). In the Chinese population. Zeng et al. conducted a matched case–control study suggesting that healthy (fruit and vegetable) or prudent (nuts, mushrooms, algae, and seafood) dietary patterns are associated with a reduced risk of hip fracture in elderly Chinese (16). Similarly, in our COPS, women including those who experienced post-menopause consumed the highest quartile of vegetarian diets and were more likely to have a higher BMD and a lower risk of vertebral fracture. Therefore, this association is still inconclusive. Further research is needed to explore the underlying mechanisms between vegetarian diets and BMD, as well as the risk of fractures.

Our study observed that higher intake of a carnivorous diet was associated with a higher BMD at the TH, FN, and L1–4, as well as a reduced risk of fractures including clinical fracture in the past 5 years and vertebral fracture. When stratified by sex, similar associations were only found in women but not in men. Compared to early studies on the elderly Chinese population, the observations were inconsistent. Zeng et al. have observed that a lower risk of hip fracture was associated with a higher intake of the prudent pattern including seafood, which was similar to us, while a high-fat pattern characterized by red meat, poultry, and animal organ meat was associated with an increased risk. No association was observed between the traditional dietary pattern in China, which includes a high intake of processed meat and fish, and hip fracture risk. Chen et al. found no association between animal protein intake and BMD in postmenopausal Chinese women (15). Unlike early studies, we characterized poultry, livestock meat, pork, and seafood as one dietary pattern—”carnivorous diet”—through principal component analysis. Protein from red meat, seafood, egg, and dairy is an essential source for bone health, contributing to approximately 50% of bone volume and a third of bone mass, and it additionally affects the secretion and action of insulin-like growth factor 1 (IGF-1), which is an orthotropic hormone related to bone formation (9). We also observed a protective effect against fracture risk and lower BMD with a diet rich in vegetarian food among women aged 40 and above, which is seemingly opposite to the carnivorous diet. Recent recommendations suggest a balance between vegetarian and carnivorous diets, emphasizing the importance of proteins, minerals, and vitamins. A chronic vegetarian diet that lacks calcium and certain vitamins might lead to lower BMD (38) and potentially result in osteoporosis and fractures.

Limitation should be considered when interpreting the study results. Firstly, the present study is built upon the COPS; the associations observed may not necessarily indicate causality because of its cross-sectional nature. Yet, the majority of studies developed to elucidate the association between BMD or fracture risk and dietary patterns had a cross-sectional design (9). COPS is the largest nationwide, population-based study of BMD, the prevalence of osteoporosis, and the prevalence of fracture in the Chinese mainland population, considered as a good representative of the general Chinese population. With a nearly 100% participation rate, the difference among regions and other possible confounding factors were likely minimized. Additionally, we collected detailed information on various potential confounding factors and controlled them in the models. Despite these efforts, potential residual confounding from unknown or uncollected variables cannot be completely ruled out. Compared with previous studies investigating the dietary pattern and BMD or fractures in China (15, 16), we included a large sample size (N = 17,489) of elderly Chinese population, which allowed exploration of the differences in associations between dietary patterns and BMD status at different locations and risk of different fracture types, as well as stratified associations by sex and postmenopausal status. Secondly, the dietary information used in the study was collected through a self-administered, semiquantitative FFQ during the last 12 months, and thus susceptible to recall bias; most of previous studies that investigated this association, however, utilized FFQ to obtain dietary information (7, 8). We also adjusted factors that were susceptible to confound the results due to dietary recall, such as alcohol consumption, education, BMI, residence (urban and rural), area (north and south), serum vitamin D level, calcium supplementation, and glucocorticoid use. Thirdly, the study identified four dietary patterns based on 17 items, which might not fully represent all food categories. However, the four dietary patterns we selected accounted for 37% of the total variability in the original food variables. Further prospective longitudinal studies are warranted to confirm and examine the association with comprehensive dietary information in the Chinese population.

In conclusion, our study suggested that diet patterns rich in meat, vegetables, and dairy, fruit, and eggs might be associated with higher BMD and lower fracture risk, especially among women aged 40 years or older, while a dietary pattern high in beverage and fried foods may be associated with lower BMD and an increased risk of fractures ( Figure 1 ). These findings are relevant to provide recommendations on dietary nutrition regarding the elderly population, especially women around the menopausal period, who are at a higher risk of osteoporosis and fractures due to factors such as estrogen decline and age-related bone loss, underscoring the need for gender-specific dietary advice to support optimal bone health (39). Future large studies with sufficient dietary information are necessary to replicate and confirm the observed associations.

The raw data supporting the conclusions of this article will be available from the corresponding author upon reasonable request.

The studies involving humans were approved by the ethical review committee of Chinese Center for Disease Control and Prevention. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

NZ: Data curation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. XY: Data curation, Investigation, Methodology, Supervision, Writing – review & editing. LinC: Investigation, Writing – review & editing. ST: Investigation, Writing – review & editing. HL: Investigation, Writing – review & editing. LuC: Investigation, Writing – review & editing. XJ: Investigation, Writing – review & editing. ZX: Investigation, Writing – review & editing. NJ: Data curation, Investigation, Writing – review & editing. LijC: Data curation, Investigation, Writing – review & editing. WY: Investigation, Writing – review & editing. SC: Writing – review & editing. LW: Funding acquisition, Investigation, Project administration, Supervision, Writing – review & editing. WX: Funding acquisition, Investigation, Project administration, Supervision, Writing – review & editing.