Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2025 Jul 1;15:20720. doi: 10.1038/s41598-025-07916-4

Caffeine intake is inversely associated with osteoporosis risk based on cross-sectional and genetic evidence

Qi-Pei Liu 1,✉, Sheng-Ting Chai 2,✉
PMCID: PMC12216410  PMID: 40595207

Abstract

This study aimed to explore the relationship between caffeine intake and osteoporosis (OP) using a combination of cross-sectional research and Mendelian randomization (MR) methods. Data from the National Health and Nutrition Examination Survey (NHANES) 2017–2020 were used to build logistic regression models, conduct restricted cubic spline (RCS) analysis, and perform subgroup analysis to explore the association between caffeine intake and OP. Additionally, MR analysis employed inverse variance weighted (IVW) as the principal technique to confirm the causal association. A total of 2,863 participants were included in the NHANES cross-sectional study, including 157 with osteoporosis. The OP group had significantly lower caffeine intake. After adjusting for all covariates, logistic regression indicated that participants in the highest tertile of caffeine intake (> 166.5 mg/day) had a 60% lower risk of OP than those in the lowest tertile (< 60 mg/day) (odds ratio [OR] = 0.401; 95% confidence interval [CI]: 0.211–0.769; P = 0.049). RCS analysis revealed an L-shaped curve indicating a declining trend in the risk of osteoporosis (P-non-linear = 0.013, P-overall = 0.003). Subgroup analysis identified race, smoking status, phosphorus, and age as the most significant factors influencing the association between caffeine intake and OP (P-interaction < 0.05). Based on the IVW approach, coffee intake was negatively causally associated with OP (OR = 0.982; 95% CI: 0.974–0.990; P < 0.001). This study, through the integration of cross-sectional and MR analyses, provides strong evidence of a negative causal relationship between caffeine intake and OP.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-07916-4.

Keywords: Osteoporosis, Caffeine intake, Coffee, NHANES, Cross-sectional, Causality, Mendelian randomization

Subject terms: Osteoporosis, Orthopaedics, Risk factors, Genetics research, Nutrition, Preclinical research

Introduction

Osteoporosis (OP) is a metabolic bone disease characterized by a disruption in the balance between bone formation and resorption, leading to the deterioration of bone microstructure and loss of bone mass1. The global prevalence of osteoporosis is estimated to be approximately 23% in men and 12% in women2. As the population ages, the elderly population is expected to exceed 1.2 billion by the end of 2025, posing new challenges related to the increasing prevalence of osteoporosis3. The social and economic burden of osteoporosis is also growing, with the United States alone spending over $17 billion annually on osteoporotic fractures4. Substances such as proteins, calcium, selenium, phosphorus, vitamin D, and magnesium play crucial roles in maintaining bone health and reducing the risk of osteoporosis5. Therefore, it is essential to emphasize the identification and adequate intake of everyday foods that have potential protective effects against osteoporosis to alleviate the increasing pressure on public health systems6. As awareness increases, there is growing attention to the potential impact of daily food intake on bone health6.

Coffee is one of the most common beverages in the world. It is consumed globally in excess of 50 billion cups per year, with Europeans alone consuming an average of 5–11 kg/year7. The polyphenols, fiber, micronutrients, caffeine, and lipids contained in coffee have potential health benefits8. Caffeine is the most famous component of coffee and is widely used as a daily stimulant in people’s work and lives. It has now been recognized for its potential benefits in the central nervous system, cardiovascular system, and neuroendocrine system9,10. The role of caffeine in orthopedics has become increasingly recognized as research progresses.

However, the association between caffeine intake and osteoporosis remains controversial. Liu et al. found that caffeine increases the expression of cyclooxygenase-2 and prostaglandins, which promotes the generation of osteoclasts, thereby decreasing bone mineral density11. However, Miao et al. demonstrated that caffeine promotes osteogenesis and inhibits osteoclast formation through the AKT, NF-κB, and MAPK pathways, exerting a protective effect on the skeletal system12. Previous observational studies based on surveys of individuals’ daily diets have found varying results regarding the relationship between caffeine intake and bone health, including positive13–16negative17–19or no association20–22. A six-year observational study found that the top quintile of American women with the highest caffeine intake had a significantly increased risk of hip fractures19. However, Wang et al. found that caffeine intake in women aged 30–39 has a potential protective effect on lumbar spine bone mineral density (BMD)16.

From its inception, the National Health and Nutrition Examination Survey (NHANES) database has built dietary and nutritional data modules that are used to detect and predict diet-related health risks and trends among Americans to inform national policy development23. Although NHANES data allow for baseline analysis, covariate adjustment, and exploration of non-linear associations, establishing causal relationships is challenging. Therefore, Mendelian randomization (MR) was introduced, which can serve the multiple confounders and reverse correlation bias in purely observational studies24,25. MR performs causal estimation by using multiple genetic variants in the dataset associated with the target risk factor and outcome of interest as instrumental variables (IVs)26,27. This study investigated the association between caffeine intake and osteoporosis by combining cross-sectional analysis and MR analysis.

Materials and methods

This study consists of two parts. First, population baseline analysis, logistic regression analysis, restricted cubic spline (RCS) analysis, and subgroup analysis were conducted using caffeine intake and osteoporosis-related data from the NHANES database. Second, an MR analysis was performed following the STROBE-MR guidelines to explore causal associations between IVs related to coffee intake and osteoporosis-related IVs extracted from genome-wide association studies (GWAS).

The cross-sectional analysis

Data source

Data were obtained from the 2017 to 2020 NHANES dataset28. The NHANES collects demographic information, examination results, and laboratory data from approximately 5,000 representative samples annually, using a multistage stratified sampling method to assess the health of Americans and track risk factors. After applying strict inclusion and exclusion criteria, the initial sample of 15,560 individuals was narrowed to 2,863 participants (Fig. 1).

Fig. 1.

Fig. 1

Open in a new tab

Flow chart of study participants.

Ascertainment of caffeine intake

A 24-hour recall dietary interview was used to record participants’ detailed dietary intake information, including food, water, and beverages, in the mobile examination center during the first session. Final intake data were collected 3–10 days later via telephone interviews. Participants’ caffeine intake data were obtained through food frequency questionnaires via telephone interviews. The average caffeine intake (mg/day) was calculated based on total nutrient intake data from the first and second days, and the participants were categorized into three groups based on their caffeine intake: Q1 (< 61 mg/day), Q2 (61–168 mg/day), and Q3 (> 168 mg/day).

Ascertainment of osteoporosis

Using 20–29-year-old white women as the reference population, femoral neck BMD in the control group was converted to T-scores, with osteoporosis defined as a T-score ≤ −2.529,30.

Covariates

Demographic information, educational level, smoking status, and history of prednisone or cortisone use were collected using a questionnaire. Laboratory data provided information on total protein, uric acid, total calcium, total cholesterol, and phosphorus levels, all potential covariates. Body mass index (BMI) was extracted from the examination data and categorized into four subgroups: underweight (< 18.5 kg/m²), normal (18.5–24.9 kg/m²), overweight (25–29.9 kg/m²), and obese (≥ 30 kg/m²)31. In the smoking questionnaire, individuals who had smoked at least 100 cigarettes in their lifetime were classified as smokers30.

Statistical analysis

To make the included population more representative, the analysis incorporated sampling weights, clustering, and stratified data. Continuous variables were presented as mean ± standard error (mean ± SE) and analyzed using t-tests. Categorical variables were presented as counts and percentages (N [%]) and Chi-square tests were applied. Logistic regression was established to explore the correlation between caffeine intake and OP across four models. Model I was unadjusted and served as a baseline comparison. Model II was adjusted for age, gender, race, and educational level. Model III was adjusted for BMI, in addition to those included in Model II. Model IV was adjusted for all covariates, including smoking status, phosphorus, uric acid, total calcium, total protein, and total cholesterol, in addition to those included in Model III. Compared with traditional linear models, the RCS methodology matched with Model IV and allowed for more precise capture of the complex association between caffeine intake and osteoporosis through variance analysis. Finally, subgroup analyses were conducted to examine potential interactions, with variables grouped into tertiles. All statistical analyses were conducted using R software version 4.4.1 (https://www.r-project.org/).

MR analysis

GWAS sources

The application of single nucleotide polymorphisms (SNPs) as IVs requires the fulfillment of three assumptions: first, IVs are significantly associated with exposure; second, IVs should be as independent of confounders and outcome as possible; and third, IVs can have an effect on outcome only by acting on exposure32. Based on the study by Liu et al., this research also included the coffee intake dataset (ukb-b-5237) from the GWAS summary dataset, which comprised 428,860 participants33,34. Referring to the study of Zhang et al.35IVs related to osteoporosis were also obtained from a GWAS dataset, which included 462,933 participants, with 7,547 in the case group and 455,386 in the control group34. The T-score was calculated based on the participants’ dual-energy X-ray bone scans, and individuals with a T-score of ≤ −2.5 were diagnosed with osteoporosis35. The study design is shown in Supplementary Figure S1.

Selection of IVs

First, the significance threshold for the SNPs that met the inclusion criteria was set at P < 5 × 10⁻⁸. Second, the risk of bias induced by potentially chained unbalanced SNPs was circumvented as much as possible by setting the parameters R2 = 0.001 and clumping distance = 10,000 kb. Third, the minimum allele frequency threshold for the SNPs was set to 0.01. Finally, F-values were calculated for all SNPs, and only SNPs with F > 10 were included to minimize the potential bias from weak instrumental variables36.

graphic file with name 41598_2025_7916_Article_Equa.gif
graphic file with name 41598_2025_7916_Article_Equb.gif

EAF indicates the effect allele frequency, beta represents the estimated genetic effect, K stands for the sample size, and SE(beta) refers to the standard error of the genetic effect37.

Statistical analysis

The principal method utilized was the inverse variance weighted (IVW) approach, and the appropriate model was selected based on Cochran’s Q tests. Four complementary methods were employed as part of the sensitivity analysis (MR-Egger, weighted median, simple mode, and weighted mode)38. MR-Egger regression and Mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO) tests were used to measure horizontal pleiotropy39. Leave-one-out analysis, MR-PRESSO, and funnel plots were performed to identify and exclude outliers that could bias the results40. A positive result was based on the simultaneous fulfillment of significant IVW results and the absence of horizontal pleiotropy. All data analyses were conducted using the “TwoSampleMR” package (version 0.6.7) in R statistical software (version 4.4.1)41.

Results

Results of cross-sectional study

Baseline characteristics

A total of 2,863 participants were included, with 47.9% female and 52.1% male (Table 1). The osteoporosis group had significantly lower average levels of uric acid, total protein, and caffeine intake compared to the non-osteoporosis group. The osteoporosis group had significantly higher average total calcium levels than those in the non-osteoporosis group. Significant differences were also observed in gender, age, race, and BMI between the osteoporosis and non-osteoporosis groups.

Table 1.

Baseline characteristics of participants.

Variables Overall Non- Osteoporosis Osteoporosis P-value
Numbers 2863 2706 157
Gender (%) < 0.001
Female 1371 (47.9) 1241 (45.9) 130 (82.8)
Male 1492 (52.1) 1465 (54.1) 27 (17.2)
Age (years) 64.18 (8.80) 63.82 (8.68) 70.38 (8.60) < 0.001
Race (%) < 0.001
Mexican American 252 (8.8) 246 (9.1) 6 (3.8)
Non-Hispanic Black 778 (27.2) 764 (28.2) 14 (8.9)
Non-Hispanic White 1162 (40.6) 1064 (39.3) 98 (62.4)
Other Hispanic 309 (10.8) 297 (11.0) 12 (7.6)
Others 362 (12.6) 335 (12.4) 27 (17.2)
Education (%) 0.078
9-11th grade 273 (9.5) 260 (9.6) 13 (8.3)
College graduate or above 759 (26.5) 722 (26.7) 37 (23.6)
High school graduate 708 (24.7) 654 (24.2) 54 (34.4)
Less than 9th grade 210 (7.3) 194 (7.2) 16 (10.2)
Some college or AA degree 913 (31.9) 876 (32.4) 37 (23.6)
Phosphorus (mg/dL) 3.56 (0.53) 3.55 (0.54) 3.68 (0.47) 0.056
Total calcium (mg/dL) 9.31 (0.39) 9.31 (0.38) 9.39 (0.44) 0.014
Total protein (mg/dL) 7.12 (0.45) 7.12 (0.45) 7.01 (0.48) 0.006
Uric acid (mg/dL) 5.54 (1.46) 5.58 (1.47) 4.86 (1.23) < 0.001
BMI (%) < 0.001
Normal 654 (22.8) 572 (21.1) 82 (52.2)
Obese 1127 (39.4) 1100 (40.7) 27 (17.2)
Overweight 1055 (36.8) 1012 (37.4) 43 (27.4)
Underweight 27 (0.9) 22 (0.8) 5 (3.2)
History of prednisone or cortisone (%) 0.957
No 2655 (92.7) 2507 (92.6) 148 (94.3)
Yes 208 (7.3) 199 (7.4) 9 (5.7)
Smoking status (%) 0.82
No 1512 (52.8) 1425 (52.7) 87 (55.4)
Yes 1351 (47.2) 1281 (47.3) 70 (44.6)
Total cholesterol (mg/dL) 190.29 (43.31) 189.86 (43.35) 197.67 (42.02) 0.116
Caffeine intake (mg/day) 148.70 (170.52) 148.73 (171.06) 148.25 (161.44) 0.015
Open in a new tab

Relationship between caffeine intake and OP

In Model I, the highest tertile of caffeine intake was associated with a 56% lower risk of osteoporosis than the lowest tertile (odds ratio [OR] = 0.446; 95% confidence interval [CI]: 0.294–0.675; P < 0.001). In models II and III, a significantly lower risk of osteoporosis was observed in the highest tertile of caffeine intake, with a 45% (OR = 0.554; 95% CI: 0.373–0.868; P = 0.023) and 51% (OR = 0.487; 95% CI: 0.263–0.901; P = 0.043) reduction in risk compared to the lowest tertile, respectively. This negative association persisted in Model IV after adjusting for all covariates (OR = 0.401; 95% CI: 0.211–0.769; P = 0.049) (Table 2). RCS analysis demonstrated an L-shaped curve, showing a declining trend in osteoporosis risk (P-non-linear = 0.013, P-overall = 0.003) (Fig. 2). When caffeine intake exceeded 107.3 mg/day, a significant inverse association with osteoporosis risk was observed (Fig. 2).

Table 2.

Four weighted multivariate linear regression models.

Subgroups (mg/day) Model I (OR; 95% CI; P-value) Model II (OR; 95% CI; P-value) Model III (OR; 95% CI; P-value) Model IV (OR; 95% CI; P-value)
Q1 (< 61) Reference Reference Reference Reference
Q2 (61–168) 0.780; (0.411–1.480); 0.455 0.806; (0.399–1.628); 0.558 0.611; (0.275–1.357); 0.252 0.501; (0.242–1.036); 0.135
Q3 (> 168) 0.446; 0.294–0.675); <0.001 0.554; (0.373–0.868); 0.023 0.487; (0.263–0.901); 0.043 0.401; (0.211–0.764); 0.049
Open in a new tab
Fig. 2.

Fig. 2

Open in a new tab

Dose-response relationship between caffeine intake and osteoporosis risk.

Subgroup analysis

The subgroup analysis performed on categorical and post-triangulation continuous variables is shown in Fig. 3. Race, smoking status, phosphorus levels, and age were identified as significant factors influencing the relationship between caffeine intake and OP (P-interaction < 0.05). No significant interactions were found in other subgroups. A significant association between caffeine intake and a reduced risk of OP was still observed in individuals aged over 65 years, Mexican Americans, non-Hispanic Whites, those with overweight, non-smokers, less than 9th grade, college graduates or above, phosphorus > 3.8 g/dL, UA < 4.8 g/dL, cholesterol < 169 g/dL and those without a history of prednisone or cortisone use.

Fig. 3.

Fig. 3

Open in a new tab

Subgroup analysis of the association between caffeine intake and osteoporosis.

Results of MR study

Results of selection of IVs

The specific information about the IVs related to coffee intake and osteoporosis is shown in Supplementary Data S1. All IVs had F-values > 10.

Effects of coffee intake on OP

The MR-PRESSO test identified a significant outlier (rs4410790) that could bias the results. After removing this SNP, no further significant outliers were detected in subsequent MR analysis. A significant inverse association between coffee intake and osteoporosis was observed (OR = 0.982; 95% CI: 0.974–0.990; P < 0.001) (Fig. 4). The other four methods exhibited similar trends (Fig. 4). Forest and scatter plots were used to visualize the association between coffee intake and osteoporosis (Fig. 5).

Fig. 4.

Fig. 4

Open in a new tab

Summary forest plot of bidirectional Mendelian randomization analysis of coffee intake and osteoporosis.

Fig. 5.

Fig. 5

Open in a new tab

Summary plot of Mendelian randomization analysis between coffee intake and osteoporosis.

Effects of OP on coffee intake

No significant association was found between OP and coffee intake (OR = 1.014; 95% CI: 0.647–1.588; P = 0.953) (Fig. 4). Consistently, the other four methods also reported no significant association (P > 0.05) (Fig. 4). The forest and scatter plots are shown in Supplementary Figure S2.

Results of sensitivity analysis

No significant horizontal pleiotropy was observed (Supplementary Data S2). Heterogeneity results can also be found in Supplementary Data S2. MR analysis of coffee intake and osteoporosis did not reveal any significant outliers (Fig. 5 and Supplementary Figure S2).

Discussion

This is the first study to explore the association between caffeine intake and OP by combining NHANES and GWAS data. Logistic regression showed that individuals in the highest tertile of caffeine intake (> 168 mg/day) had a significantly reduced risk of OP compared to those in the lowest tertile (< 61 mg/day). RCS analysis demonstrated an L-shaped curve, indicating a decreasing OP risk with higher caffeine intake. Subsequent subgroup analysis identified race, smoking status, phosphorus level, and age as the most significant factors that influenced the association between caffeine intake and OP. The negative causal association between caffeine intake and OP was further supported by MR analysis.

The association between caffeine intake and bone health remains debated, as observational studies have produced mixed and conflicting results. For instance, Demirbag et al. conducted a study on 200 postmenopausal women and found that their premenopausal coffee consumption habits did not significantly affect BMD in various body sites postmenopause20. Similarly, Harter et al. reported no association between caffeine intake and bone mass changes in Brazilian perimenopausal women21. However, their use of bone ultrasound and self-reported dietary data may have underestimated caffeine intake and introduced inaccuracies. Additionally, Liao et al. analyzed a dataset of postmenopausal women from the NHANES database and found no association between caffeine and BMD after adjusting for a range of covariates (Beta = − 0.012; 95% CI: −0.049, 0.026; P = 0.54)22.

In contrast, Lo et al. observed an inverse relationship between caffeine intake and lumbar spine, forearm, and hip T-scores in a hospital-based study of perimenopausal women17. However, the lack of assessment for confounders such as diabetes and smoking limited the study’s reliability. Additionally, França et al. conducted a dietary survey of Brazilian women and found that consuming foods with both high sugar and high caffeine content was negatively correlated with total femur BMD and total body BMD18. Study indicated that the oxidative stress induced by high sugar intake inhibits osteoblast differentiation, which is itself a risk factor for bone quality42. Thus, França et al.’s study did not isolate caffeine’s direct impact on BMD.

Choi et al. reported a 36% lower risk of osteoporosis in the highest quartile of coffee intake compared to the lowest quartile among postmenopausal Korean women13. The study employed multiple linear regression models and found that higher coffee intake correlated with increased BMD in the whole body, femur neck, and lumbar spine. Similarly, Hirata observed the protective effect of coffee intake on total hip and femoral neck BMD in a Japanese community cohort14. Xu et al. observed a negative association between caffeine-containing coffee and osteoporosis in Americans who habitually consumed less than two cups of coffee daily15. Interestingly, no significant association was observed between decaffeinated coffee and osteoporosis, indirectly suggesting the protective role of caffeine in bone health. The controversy between conclusions in previous observational studies may be attributed to differences in demographic characteristics, insufficient sample sizes, ambiguity in dietary questionnaires, and variations in BMD measurement methods and sites. Despite efforts to design more rigorous observational studies, it is difficult to avoid interference from various potential confounding factors and reverse causality. Despite these limitations, by integrating observational studies with MR analysis, this study provides robust evidence for a causal link between caffeine intake and reduced osteoporosis risk.

Subgroup analysis revealed significant intergroup differences in the negative association between caffeine intake and osteoporosis across age groups, ethnicities, smoking status, and serum phosphate levels. Older individuals (> 65 years) may have experienced prolonged bone loss and lower baseline bone density42. Consequently, caffeine may exert a more pronounced protective effect on their bone metabolism. Conversely, younger individuals, with higher baseline bone density and more active bone metabolism, may experience a less pronounced protective effect of caffeine. Different ethnic groups exhibit varying baseline bone densities and metabolic characteristics. Different ethnic groups exhibit varying baseline bone densities and metabolic characteristics. Mexican American and Non-Hispanic White groups, with lower baseline BMD than Non-Hispanic Blacks44,45may exhibit stronger protective effects from caffeine. A stronger protective effect of caffeine was also observed in individuals with high serum phosphate levels, highlighting the need for further investigation. Cigarette extracts damage mesenchymal stem cells and induce ferroptosis, leading to BMD loss46. Smoking, as an independent risk factor affecting bone homeostasis, may interfere with the protective effect of caffeine on bones, which requires further exploration.

Choi et al. suggested that estrogenic effects, antioxidant properties, and potential anti-inflammatory actions of coffee contribute to bone protection13. Studies showed that caffeine intake increases renal calcium clearance, leading to reduced bone deposition47although this effect is likely negligible in clinical settings and can be easily compensated for through dietary intake48. An in vitro study found that moderate caffeine intake protects bones by disrupting the balance between osteoblastogenesis and osteoclastogenesis through the MAPK and NF-κB pathways12. In ovariectomized rats, caffeine intake alleviated estrogen deficiency-induced skeletal changes by promoting bone mineralization, improving bone structure and strength, and enhancing mechanical properties49. Additionally, caffeine’s antagonistic effects on adenosine receptors help regulate the balance between osteoblasts and osteoclasts, supporting bone metabolism48. Xu et al. further pointed out that caffeine may improve bone health by enhancing lipid profiles, increasing calcium concentrations, and boosting the activity of alkaline and acid phosphatases50.

Firstly, the NHANES data were based on a nationally representative sample, and the analysis accounted for NHANES weights, ensuring the study population’s representativeness. Since cross-sectional studies explore associations without establishing causality, this study incorporated MR analysis to confirm the robust causal relationship between caffeine intake and OP. Additionally, previous observational studies often focused on postmenopausal women, limiting their generalizability. In contrast, this study included participants of varying ages and genders, thoroughly accounted for covariates and conducted subgroup analyses to explore intergroup relationships. This approach enhances the study’s generalizability and aids in developing targeted bone health strategies.

However, some limitations of this study should not be overlooked. Firstly, dietary data were collected via 24-hour recall interviews, which may introduce recall bias and fail to fully reflect habitual dietary intake. Secondly, the caffeine dataset excludes information on medications or dietary supplements, reflecting only regular food consumption and potentially underestimating actual intake. Thirdly, the populations involved in the cross-sectional study and MR analysis were from the U.S. and Europe, which introduces heterogeneity and may limit the generalizability of the findings. In addition, due to the limitations of available summary-level GWAS data, this study failed to use multivariable MR to adjust for potential confounding variables. Finally, while osteoporosis diagnostic criteria were strictly followed, the cohort of osteoporosis patients constituted a small proportion of the total population, potentially reducing the reliability of the findings. Given these limitations, future studies with more rigorous designs are needed for further exploration.

Conclusion

This study, through the integration of cross-sectional and MR analyses, provides strong evidence of a negative causal relationship between caffeine intake and OP.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1 (28.9KB, zip)
Supplementary Material 2 (10.5MB, docx)
Supplementary Material 3 (41.9KB, docx)
Supplementary Material 4 (48KB, docx)
Supplementary Material 5 (4.6MB, docx)
Supplementary Material 6 (1.4MB, zip)

Acknowledgements

We sincerely thank all the researchers and participants of the NHANES and GWAS for providing the data used in this study.

Abbreviations

OP

Osteoporosis

MR

Mendelian randomization

NHANES

National Health and Nutrition Examination Survey

RCS

Restricted cubic spline

IVW

Inverse variance weighted

SNPs

Single nucleotide polymorphisms

IVs

Instrumental variables

GWAS

Genome-wide association studies

BMD

Bone mineral density

BMI

Body mass index

MR-PRESSO

Mendelian randomization pleiotropy residual sum and outlier

OR

Odds ratio

CI

Confidence interval

Author contributions

Conceptualization, study design, methodology, data collection, statistical analysis, and visualization: QL. Validation and writing (original draft, review, and editing): QL and SC.

Data availability

All data utilized in this research were accessible through GWAS database (https://gwas.mrcieu.ac.uk/) and NHANES database (https://wwwn.cdc.gov/nchs/nhanes/default.aspx), both of which are publicly accessible platforms. We have uploaded the data analysis scripts and documentation to a public GitHub repository (https://github.com/lqp67887/Raw-data-and-code.git).

Declarations

Ethics approval and consent to participate

The data covered in this study were obtained from publicly available databases and relevant ethical consent was obtained. Therefore, ethical approval and informed consent were not required for this study.

Consent for publication

The authors have read the manuscript and agree to publish it.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Qi-Pei Liu, Email: lqp67887@163.com.

Sheng-Ting Chai, Email: cst0192@qq.com.

References

  • 1.Föger-Samwald, U., Dovjak, P., Azizi-Semrad, U., Kerschan-Schindl, K. & Pietschmann, P. Osteoporosis: pathophysiology and therapeutic options. EXCLI J.19, 1017–1037 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Salari, N. et al. The global prevalence of osteoporosis in the world: a comprehensive systematic review and meta-analysis. J. Orthop. Surg.16, 609 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Salari, N. et al. Global prevalence of osteoporosis among the world older adults: a comprehensive systematic review and meta-analysis. J. Orthop. Surg.16, 669 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Clynes, M. A. et al. The epidemiology of osteoporosis. Br. Med. Bull.133, 105–117 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Runting, H. et al. Is bone mineral density in middle-aged and elderly individuals associated with their dietary patterns? A study based on NHANES. Front. Nutr.11, 1396007 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Wang, K. et al. Associations of healthy eating index-2015 with osteoporosis and low bone mass density in postmenopausal women: a population-based study from NHANES 2007–2018. Front. Nutr.11, 1388647 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Barcelos, R. P., Lima, F. D., Carvalho, N. R., Bresciani, G. & Royes, L. F. Caffeine effects on systemic metabolism, oxidative-inflammatory pathways, and exercise performance. Nutr. Res. N Y N. 80, 1–17 (2020). [DOI] [PubMed] [Google Scholar]
  • 8.Abalo, R. Coffee and caffeine consumption for human health. Nutrients13, 2918 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Cappelletti, S., Piacentino, D., Sani, G. & Aromatario, M. Caffeine: cognitive and physical performance enhancer or psychoactive drug? Curr. Neuropharmacol.13, 71–88 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Woolf, B. et al. Appraising the causal relationship between plasma caffeine levels and neuropsychiatric disorders through Mendelian randomization. BMC Med.21, 296 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Liu, S. H. et al. Caffeine enhances osteoclast differentiation from bone marrow hematopoietic cells and reduces bone mineral density in growing rats. J. Orthop. Res. Off Publ Orthop. Res. Soc.29, 954–960 (2011). [DOI] [PubMed] [Google Scholar]
  • 12.Miao, Y. et al. Caffeine regulates both osteoclast and osteoblast differentiation via the AKT, NF-κB, and MAPK pathways. Front. Pharmacol.15, 1405173 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.E, C. et al. The benefit of bone health by drinking coffee among Korean postmenopausal women: A Cross-Sectional analysis of the fourth & fifth Korea National health and nutrition examination surveys. PloS One11, e0147762 (2016). [DOI] [PMC free article] [PubMed]
  • 14.Hirata, H. et al. Association between dietary intake and bone mineral density in Japanese postmenopausal women: the Yokogoshi cohort study. Tohoku J. Exp. Med.239, 95–101 (2016). [DOI] [PubMed] [Google Scholar]
  • 15.Xu, J. & Zhai, T. Coffee drinking and the odds of osteopenia and osteoporosis in Middle-Aged and older americans: A Cross-Sectional study in NHANES 2005–2014. Calcif Tissue Int.114, 348–359 (2024). [DOI] [PubMed] [Google Scholar]
  • 16.Wang, G. et al. Association between caffeine intake and lumbar spine bone mineral density in adults aged 20–49: A cross-sectional study. Front. Endocrinol.13, 1008275 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Lo, H. C., Kuo, D. P. & Chen, Y. L. Impact of beverage consumption, age, and site dependency on dual energy X-ray absorptiometry (DEXA) measurements in perimenopausal women: a prospective study. Arch. Med. Sci. AMS. 13, 1178–1187 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Na, F., Mb, C., Bs, M. L. C. & La, M. P. Dietary patterns and bone mineral density in Brazilian postmenopausal women with osteoporosis: a cross-sectional study. Eur J. Clin. Nutr70, 85–90 (2016). [DOI] [PubMed]
  • 19.Hernandez-Avila, M. et al. Caffeine, moderate alcohol intake, and risk of fractures of the hip and forearm in middle-aged women. Am. J. Clin. Nutr.54, 157–163 (1991). [DOI] [PubMed] [Google Scholar]
  • 20.Demirbag, D., Ozdemir, F. & Ture, M. Effects of coffee consumption and smoking habit on bone mineral density. Rheumatol. Int.26, 530–535 (2006). [DOI] [PubMed] [Google Scholar]
  • 21.Harter, D. L. et al. Association between low bone mass and calcium and caffeine intake among perimenopausal women in Southern brazil: cross-sectional study. Sao Paulo Med. J. Rev. Paul Med.131, 315–322 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Liao, S. et al. The relationship between caffeine and its metabolites and bone mineral density in postmenopausal women: a cross-sectional analysis from the NHANES database. J. Nutr. Sci.12, e131 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Gahche, J. J. et al. Federal monitoring of dietary supplement use in the resident, civilian, noninstitutionalized US population, National health and nutrition examination survey. J. Nutr.148, 1436S–1444S (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Sekula, P., Del Greco, M., Pattaro, F., Köttgen, A. & C. & Mendelian randomization as an approach to assess causality using observational data. J. Am. Soc. Nephrol. JASN. 27, 3253–3265 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Skrivankova, V. W. et al. Strengthening the reporting of observational studies in epidemiology using Mendelian randomization: the STROBE-MR statement. JAMA326, 1614–1621 (2021). [DOI] [PubMed] [Google Scholar]
  • 26.Davies, N. M., Holmes, M. V. & Davey Smith, G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ362, k601 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Bowden, J. & Holmes, M. V. Meta-analysis and Mendelian randomization: A review. Res. Synth. Methods. 10, 486–496 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.NHANES - National Health and Nutrition Examination Survey Homepage. (2024). https://www.cdc.gov/nchs/nhanes/index.htm
  • 29.Looker, A. C. et al. Prevalence of low femoral bone density in older U.S. Adults from NHANES III. J. Bone Min. Res. Off J. Am. Soc. Bone Min. Res.12, 1761–1768 (1997). [DOI] [PubMed] [Google Scholar]
  • 30.Wang, S. et al. Association between ultra-processed food and osteoporosis: a cross-sectional study based on the NHANES database. Nutr. Metab.21, 69 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Li, S. et al. An analysis of the potential association between obstructive sleep apnea and osteoporosis from the perspective of transcriptomics and NHANES. BMC Public. Health. 24, 1998 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Sanderson, E. Multivariable Mendelian randomization and mediation. Cold Spring Harb Perspect. Med.11, a038984 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Feng, K., Li, P., Guo, H. & Chen, Z. The impact of coffee consumption on osteoarthritis: insights from NHANES and Mendelian randomization analysis. Front. Nutr.11, 1434704 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.IEU OpenGWAS project. https://gwas.mrcieu.ac.uk/
  • 35.Zhang, N., Ji, C., Liu, L., Ye, E. & Yuan, C. The causal relationship between PCSK9 inhibitors and osteoporosis based on Drug-Targeted Mendelian combined mediation analysis. Calcif Tissue Int.115, 53–62 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.S, B. & Sg, T. Avoiding bias from weak instruments in Mendelian randomization studies. Int J. Epidemiol40, 755–764 (2011). [DOI] [PubMed]
  • 37.Liu, Q. P. Application of Mendelian randomization analysis to explore causal associations of aspirin use with bone mineral density and risk of fracture. Hereditas162, 3 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Bowden, J. et al. Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic. Int. J. Epidemiol.45, 1961–1974 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Verbanck, M., Chen, C. Y., Neale, B. & Do, R. Detection of widespread horizontal Pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat. Genet.50, 693–698 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Weng, H. et al. Association between uric acid and risk of venous thromboembolism in East Asian populations: a cohort and Mendelian randomization study. Lancet Reg. Health West. Pac.39, 100848 (2023). [DOI] [PMC free article] [PubMed]
  • 41.Hemani, G. et al. The MR-Base platform supports systematic causal inference across the human phenome. eLife7, e34408 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Chen, X. J. et al. The Inhibition of PINK1/Drp1-mediated mitophagy by hyperglycemia leads to impaired osteoblastogenesis in diabetes. iScience28, 111519 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Tao, L. et al. The changing rule of human bone density with aging based on a novel definition and mensuration of bone density with computed tomography. Preprint At.10.48550/arXiv.2308.02815 (2023). [Google Scholar]
  • 44.Noel, S. E., Santos, M. P. & Wright, N. C. Racial and ethnic disparities in bone health and outcomes in the united States. J. Bone Min. Res. Off J. Am. Soc. Bone Min. Res.36, 1881–1905 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Walker, M. D. & Bilezikian, J. P. Racial differences in bone density and fracture risk in the united States. Int. J. Rheum. Dis.11, 341–346 (2008). [Google Scholar]
  • 46.Jing, Z. et al. Tobacco toxins induce osteoporosis through ferroptosis. Redox Biol.67, 102922 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Reuter, S. E. et al. The effect of high-dose, short-term caffeine intake on the renal clearance of calcium, sodium and creatinine in healthy adults. Br. J. Clin. Pharmacol.87, 4461–4466 (2021). [DOI] [PubMed] [Google Scholar]
  • 48.Berman, N. K., Honig, S., Cronstein, B. N. & Pillinger, M. H. The effects of caffeine on bone mineral density and fracture risk. Osteoporos. Int. J. Establ Result Coop. Eur. Found. Osteoporos. Natl. Osteoporos. Found. USA. 33, 1235–1241 (2022). [DOI] [PubMed] [Google Scholar]
  • 49.Folwarczna, J. et al. Favorable effect of moderate dose caffeine on the skeletal system in ovariectomized rats. Mol. Nutr. Food Res.57, 1772–1784 (2013). [DOI] [PubMed] [Google Scholar]
  • 50.Xu, H. et al. Effect of caffeine on ovariectomy-induced osteoporosis in rats. Biomed. Pharmacother Biomedecine Pharmacother. 112, 108650 (2019). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (28.9KB, zip)
Supplementary Material 2 (10.5MB, docx)
Supplementary Material 3 (41.9KB, docx)
Supplementary Material 4 (48KB, docx)
Supplementary Material 5 (4.6MB, docx)
Supplementary Material 6 (1.4MB, zip)

Data Availability Statement

All data utilized in this research were accessible through GWAS database (https://gwas.mrcieu.ac.uk/) and NHANES database (https://wwwn.cdc.gov/nchs/nhanes/default.aspx), both of which are publicly accessible platforms. We have uploaded the data analysis scripts and documentation to a public GitHub repository (https://github.com/lqp67887/Raw-data-and-code.git).

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES