Association of frailty with fragility fractures and all‑cause mortality in adults: evidence from the National Health and Nutrition Examination Survey

Abstract

Background

Fragility fractures impose a substantial global public health burden, driven by diminished bone strength and extrinsic risk factors. Frailty, characterized by multisystem decline and reduced physiological reserve, is increasingly recognized as a predictor of adverse outcomes. However, evidence linking frailty to fragility fractures and mortality in nationally representative samples remains limited.

Methods

This mixed study combined cross-sectional analysis of fragility fractures with prospective mortality assessment in 14,752 NHANES participants (1999–2018). Survey-weighted multivariable logistic regression evaluated frailty and fragility fracture associations, and Cox proportional hazards models assessed mortality risk among participants with fragility fractures. Nonlinear analysis, subgroup analysis, and sensitivity analysis were also performed to validate the robustness of the results.

Results

After adjusting for covariates, logistic regression identified that each 0.1-unit Frailty Index increase was associated with 28% higher fragility fracture risk (OR 1.28, 95% CI 1.15–1.43) and 44% greater all-cause mortality of participants with fragility fractures (HR 1.44, 95% CI 1.25–1.65). Frail individuals had 66% elevated fragility fracture risk (OR1.66, 95% CI 1.27–2.17) and 110% increased mortality (HR 2.10, 95% CI 1.60–2.74) than non-frail individuals. Analysis of specific sites showed strong associations between frailty with hip fragility fractures (OR 2.49, 95% CI 1.44–4.33) and spinal fragility fractures (OR 2.24, 95% CI 1.21–4.13). The robustness of the results was validated using restricted cubic splines, subgroup analyses, and datasets with missing covariates removed.

Conclusions

In conclusion, frailty is independently associated with a higher prevalence of fragility fractures and predicts mortality in American adults, demonstrating dose-response relationships and site-specific heterogeneity. Further research is needed to confirm these associations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12891-026-09574-7.

Background

Fragility fractures, typically caused by low-energy trauma such as falls from standing height or less, represent a substantial global public health challenge, especially in aging populations [1]. The personal and economic burdens imposed by fragility fractures are substantial. Following fractures, people frequently develop functional dependence necessitating long-term care, while substantially elevating mortality risk. Approximately 20% of individuals sustaining osteoporosis-related fractures die within 12 months of the initial injury [2, 3]. Conventional risk assessment for fragility fractures predominantly depends on bone mineral density (BMD) measurement using dual-energy X-ray absorptiometry (DXA) and clinical risk factors incorporated into tools such as FRAX [4]. However, a significant portion of fractures occur in individuals who do not meet the osteoporosis threshold defined by bone density, highlighting the multifactorial causes of fragility fractures and the limitations of models centered on BMD [5]. This understanding has prompted investigation into broader geriatric syndromes, notably frailty [6]. Frailty is defined as a state of heightened vulnerability to physiological stressors arising from cumulative multisystem decline, resulting in diminished functional reserve and increased susceptibility to adverse health outcomes [7]. Unlike the frailty phenotype developed by Fried et al., the 49-deficit cumulative Frailty Index (FI) proposed by Searle et al. covers a greater number of clinically relevant variables and can more comprehensively reflect comorbidities and broader functional impairments [8]. Several epidemiological evidences demonstrate that frailty independently associates with elevated risks of falls, fractures, hospitalization, and all-cause mortality [9, 10]. Globally, about 24% of community-dwelling older adults exhibit frailty by deficit accumulation criteria [11].

Despite specific studies linking frailty to fractures in countries such as the United Kingdom, South Korea, Australia [12–15], comprehensive investigations examining associations between frailty with fragility fractures and mortality within large-scale nationally representative datasets in America remain scarce. Definitions of frailty exhibit heterogeneity across studies, warranting further investigation into the association between frailty defined using validated deficit accumulation models with fragility fractures and mortality [16]. After a fracture, it involves the local injury stage, the axial stress propagation stage, and the comprehensive damage stage [17], and anatomical features of the hip, spine, and wrist make them predisposed to fragility fractures. Fractures of these sites capture early and late manifestations of fall‑related injury, and are systematically recorded in National Health and Nutrition Examination Survey (NHANES). Therefore, this study aims to investigate the association between frailty and self-reported history of fragility fractures among adults participating in NHANES. We sought to determine whether frailty severity, quantified through validated deficit accumulation metrics, independently correlates with fragility fractures and fracture mortality. We evaluated the interaction between frailty and sociodemographic determinants, including age stratification, gender disparities, and marital status. Such stratification enables identification of high-risk phenotypes. Our findings may refine integrated risk assessment frameworks that combine frailty metrics with conventional osteoporosis diagnostics to optimize fracture prevention in aging societies.

Methods

Study design and population

NHANES (https://www.cdc.gov/nchs/nhanes) uses a stratified, multistage probability sampling design to produce nationally representative data for the non‑institutionalized U.S. population [18]. The survey comprises household interviews, standardized physical examinations, and laboratory testing. All study protocols received approval from the National Center for Health Statistics (NCHS) Research Ethics Review Board, with written informed consent obtained from every participant. NHANES is conducted every two years as a cycle, with each cycle’s sampling coming from different individuals. Since fracture information was not collected during the 2011–2012 and 2015–2016 cycles, participant selection followed a structured process using combined NHANES in 1999–2010, 2013–2014 and 2017–2018 cycles. After applying predefined inclusion and exclusion criteria, 14,752 adults were enrolled from an initial sample of 81,589 participants (Fig. 1). The relationship between frailty and fragility fractures was examined cross-sectionally using baseline fracture history data, while the relationship between frailty and mortality in patients with fragility fractures was assessed through prospective follow-up.

Fig. 1.

Flow chart of the study participants

Assessment of fragility fractures

Fragility fractures (hip, spine, and wrist) were ascertained by participants’ self‑report of a physician diagnosis captured via the Computer-Assisted Personal Interview system. NHANES did not collect standardized self-reports or examination data of fractures in other anatomical sites, so the study was limited to the three anatomical sites systematically recorded in the NHANES questionnaire: hip, spine, and wrist. While this method is commonly used in large population surveys, it may be subject to recall bias and misclassification, particularly among older adults. Participants were asked, “Has a doctor ever told you that you had broken or fractured your hip, spine, or wrist?” (fractures occurring before the interview date) and “a fall from standing height or less, for example, tripped, slipped, or fell out of bed”. Participants who answered affirmatively to both of the above questions were classified as the fragility fracture group. Referring to previous studies [19], those reporting no fractures comprised the non-fragility fracture comparison group.

Assessment of frailty

In this study, frailty was constructed following Searle’s established criteria [20]. The assessment encompassed seven domains (49 items): cognition, independence, depressive symptoms, comorbidities, physical function and anthropometric measures, healthcare utilization and access, and laboratory indicators. Consistent with prior methodology [21–24], each deficit was quantified on a scale from 0 to 1, with 0 indicating the deficit was absent and 1 representing its exists completely or the most severe. The FI for each participant was calculated as the sum of deficit scores divided by the number of deficits considered (range 0–1). Prior studies indicate that FI based on a substantial number of deficits produces more reliable estimates [7, 25], so we limited analyses to participants who completed ≥ 80% (≥ 40 variables) of the FI items. Detailed scoring criteria are provided in Supplementary Table S1. In subsequent analyses, we modelled the FI both as a continuous and as a categorical variable. For the continuous variables, the FI was scaled per 0.1‑unit increases to aid the interpretability of odds ratios and hazard ratios. For categorical analyses, we defined frailty as those with FI > 0.21 and non‑frailty as FI ≤ 0.21, a cutoff commonly used in deficit accumulation research.

Ascertainment of mortality in participants with fragility fractures

To ascertain all-cause and cardiovascular disease (CVD) mortality endpoints in patients with fragility fractures, we employed publicly accessible mortality linkage files from the NCHS. These datasets merge survey participant records with National Death Index (NDI) death certificates through probabilistic matching algorithms. Mortality follow-up extended through December 31, 2019. The specific cause of death is determined according to the 10th Edition of the International Classification of Diseases (ICD-10). CVD mortality in this study is defined as deaths attributable to heart (codes I00–I09, I11, I13, I20–I51) and cerebrovascular diseases (codes I60–I69) [26]. Specific definition information can be found at: https://www.cdc.gov/nchs/data/datalinkage/underlying-and-multiple-cause-of-death-codes-508.pdf.

Assessment of covariates

Based on previous literature, this study incorporated multiple covariates to comprehensively analyze their potential effects on primary outcomes. Among them, demographic factors include gender, age, race, and marital status. Races are classified into four categories: non-Hispanic White, non-Hispanic Black, Mexican American, and other race. Marital status is classified into two categories based on whether one is currently living alone: one is married or living with a partner, and the other is never married, separated, divorced, or widowed. Health-related factors in the questionnaire include smoke, drink, and activity. Smoking status is classified into never smoking (< 100 cigarettes in life), former smoking (≥ 100 cigarettes and not currently smoking), and current smoking (≥ 100 cigarettes and currently smoking). Alcohol consumption is classified into two categories based on whether the annual intake exceeds 12 cups. Physical activity is classified into three categories based on the intensity of recent recreational and work activities: vigorous, moderate, or none. The examination items include body mass index (BMI, kg/m²). Laboratory tests include serum albumin (g/L).

Statistical analyses

Baseline characteristics were presented according to frailty status. As recommended in NHANES analytic guidance, we created an integrated multi-cycle weight by dividing the appropriate survey weights by the number of merged periods. We combined the weights, primary sampling units, and strata to ensure nationally representative estimates. We perform multiple imputation for the missing covariates and conduct a combined analysis of the five imputed datasets. Continuous variables were expressed as weighted means with their standard errors (SE), and categorical variables were reported as unweighted counts and weighted percentages. Group comparisons employed appropriate statistical tests for complex survey data: the Wilcoxon rank-sum test for continuous variables and the Rao-Scott adjusted chi-square test for categorical variables, with a significance level of α = 0.05. Survey-weighted multivariable logistic regression models were constructed to quantify associations between frailty exposures and fragility fractures. Four hierarchical adjustment models were implemented: Model 1 (unadjusted), Model 2 (adjusted for sociodemographic factors: age, gender, race, marital), Model 3 (further adjusted for behavioral factors and additional clinical factors: BMI, drink, and smoke), and Model 4 (further adjusted with comorbidities: physical activity and serum albumin). To ensure both statistical rigor and clinical relevance, the Frailty Index (FI) was analyzed in two ways: (1) as a continuous variable (rescaled to per 0.1-unit increase) to explore the incremental dose-response relationship; and (2) as a categorical variable using the clinically validated threshold of > 0.21 to facilitate risk stratification and practical interpretation. Results are presented as odds ratios (OR) with 95% confidence intervals (CI). Survey-weighted restricted cubic splines (RCS) were employed to model potential non-linear relationships between the continuous FI and fragility fractures, using the final multivariable-adjusted model (Model 4). The linearity assumption was formally tested using the likelihood ratio test. Furthermore, we examined frailty associations with site-specific fracture(hip, wrist, and spine) using multivariable logistic regression models, applying four sequential covariate adjustment levels consistent with our primary analysis. We conducted subgroup analyses to evaluate associations between frailty and fragility fractures across demographic and clinical strata, with results visualized via forest plots. Interaction effects was tested using multiplicative interaction models.

Cox proportional hazards models were employed to evaluate associations between frailty and mortality outcomes among individuals with fragility fractures, utilizing four sequential adjustment models identical to those specified in the multivariable logistic regression analysis. Kaplan-Meier survival curves were constructed to visualize mortality risk differences between frail and non-frail fragility fracture patients. Log-rank tests quantified the statistical significance of survival curve divergences. We conducted sensitivity analyses to assess the robustness of the association between frailty indicators and fragility fractures and mortality after removing participants with missing covariates. In addition, we conducted sensitivity analyses grouped by different levels of frailty and another sensitivity analysis including covariates of anti-osteoporotic treatment. Statistical significance was defined as two-tailed P < 0.05. Analysis was performed using R 4.2.1 (http://www.Rproject.org; The R Foundation, Vienna, Austria) and the Free Statistics software (version 2.1.1; Beijing Free Clinical Medical Technology Co., Ltd, Beijing, China).

Results

Characteristics of the study participants

The weighted baseline characteristics of 14,752 participants were stratified by frailty status (Table 1), among whom 544 had fragility fractures. The average age of the participants was 62.6 ± 0.2 years, and 7,697 were female (56.5%). The participants were divided into two groups: the robust group (n = 9,957, 67.5%) and the frailty group (n = 4795, 32.5%). Frail individuals were significantly older, had higher BMI, lower serum albumin, and a higher proportion of non-Hispanic blacks than non-frail counterparts. All characteristics differed significantly between groups (P < 0.001).

Table 1.

Weighted baseline participant information for NHANES

Characteristic	Overall, n = 14,752	Non-Frailty, n = 9957	Frailty, n = 4795	P
Age (years)	62.6 (0.2)	62.0 (0.2)	64.0 (0.4)	< 0.001
Female	7,697 (56.5%)	4,918 (53.7%)	2,779 (63.2%)	< 0.001
Race				< 0.001
Non-Hispanic White	7,661 (75.9%)	5,276 (78.0%)	2,385 (70.8%)
Non-Hispanic Black	3,016 (10.4%)	1,890 (8.8%)	1,126 (14.1%)
Mexican American	2,263 (4.7%)	1,565 (4.6%)	698 (5.0%)
Other race	1,812 (9.0%)	1,226 (8.6%)	586 (10.1%)
Marital				< 0.001
Married/Living with partner	8,593 (62.0%)	6,172 (65.3%)	2,421 (54.1%)
Living alone	6,159 (38.0%)	3,785 (34.7%)	2,374 (45.9%)
BMI (kg/m2)	29.3 (0.1)	28.5 (0.1)	31.2 (0.2)	< 0.001
Drink	8,982 (63.7%)	6,365 (66.9%)	2,617 (56.0%)	< 0.001
Smoke				< 0.001
Never	7,055 (48.1%)	4,986 (50.6%)	2,069 (42.1%)
Former	5,145 (34.1%)	3,388 (33.3%)	1,757 (35.9%)
Current	2,552 (17.8%)	1,583 (16.1%)	969 (21.9%)
Activity				< 0.001
Sedentary	8,471 (51.6%)	4,955 (44.0%)	3,516 (70.1%)
Moderate	4,449 (33.3%)	3,434 (37.3%)	1,015 (23.8%)
Vigorous	1,832 (15.0%)	1,568 (18.8%)	264 (6.1%)
Serum albumin (g/L)	41.8 (0.1)	42.3 (0.1)	40.7 (0.1)	< 0.001

BMI body mass index, Mean (Standard Error) for continuous variables, n (%) (unweighted counts and survey‑weighted percentages) for categorical variables. For brevity, only the clinically relevant reference categories of common binary variables are displayed

Association between frailty and fragility fractures

Multivariable regression analyses quantified associations between frailty and fragility fractures, with adjusted effect sizes presented in Table 2. After full adjustment (Model 4), each 0.1-unit FI increase demonstrated 28% higher fracture risk (OR 1.28, 95% CI 1.15–1.43, P < 0.001). Frail individuals are 1.66 times more likely to experience fractures compared to non-frail individuals (OR 1.66, 95% CI 1.27–2.17, P < 0.001). Notably, effect sizes attenuated progressively across adjustment levels yet maintained statistical significance in all models. Supplementary Table S2 demonstrates significant heterogeneity in fragility fracture associations across anatomical sites (hip, spine, and wrist). After full adjustment (Model 4), for hip fractures, each 0.1-unit increase in FI conferred 40% higher risk (OR 1.40, 95% CI 1.11–1.77, P = 0.005), while frail individuals had 149% greater risk compared to non-frail counterparts (OR 2.49, 95% CI 1.44–4.33, P = 0.001). For spine fractures, FI increments showed 64% elevated risk (OR 1.64, 95% 1.34–2.00, P = 0.001), and frailty status increased risk by 124% (OR 2.24, 95% CI 1.21–4.13, P = 0.011). For wrist fractures, FI increments showed 16% elevated risk (OR 1.16, 95% CI 1.01–1.33, P = 0.033). The increase in the state of frailty was not statistically significant (OR 1.36, 95% CI 0.98–1.89, P = 0.066), but the overall trend remained the same.

Table 2.

Weighted logistic regression analysis of frailty and fragility fractures

Variable	Model 1	Model 2	Model 3	Model 4
	OR(95%CI), P value	OR(95%CI), P value	OR(95%CI), P value	OR(95%CI), P value
Frailty index (per 0.1)	1.41(1.29,1.54), < 0.001	1.29(1.17,1.43), < 0.001	1.31(1.17,1.46), < 0.001	1.28(1.15,1.43), < 0.001
Frailty
No	1(Reference)	1(Reference)	1(Reference)	1(Reference)
Yes	2.07(1.63,2.63), < 0.001	1.72(1.34,2.20), < 0.001	1.74(1.33,2.27), < 0.001	1.66(1.27,2.17), < 0.001

Model 1: No covariates were adjusted

Model 2: Age, gender, race and marital were adjusted

Model 3: The variables in model 2 plus BMI, drink and smoke were adjusted

Model 4: The variables in model 3 plus activity and albumin were adjusted

OR odds ratio, CI confidence interval

Subgroup analysis and interactive effect analysis

After adjusting for all covariations, we conducted an interaction and subgroup analysis between frailty and fragility fractures. The detailed results are presented in the form of a forest plot (Fig. 2). The positive frailty and fragility fractures association demonstrated remarkable consistency across most population strata (Fig. 2). The analysis demonstrated a significant multiplicative interaction effect of marital status on the relationship between frailty and fragility fractures (P = 0.018). No significant effect modification was observed for age, gender, race, BMI, drink, or smoke, indicating robust associations independent of these factors.

Fig. 2.

Results of subgroup analyses of the association between frailty and fragility fractures. BMI body mass index, OR odds ratio, CI confidence interval. The subgroup analysis was adjusted for age, gender, race, BMI, marital status, drink, smoke, activity, and albumin. In each case, the model was not adjusted for the stratification variable itself

Association between frailty and mortality in patients with fragility fractures

Among fragility fracture patients, frailty significantly predicted all-cause and CVD mortality after full adjustment (Table 3). For all-cause mortality, each 0.1-unit increase in FI was associated with 44% higher mortality (HR 1.44, 95% CI 1.25–1.65, P < 0.001), while frail individuals had 110% greater mortality (HR 2.10, 95% CI 1.60–2.74, P < 0.001). For CVD mortality, each 0.1-unit FI increase conferred 83% higher mortality (HR 1.83, 95% CI 1.42–2.36, P < 0.001), with frail individuals experiencing 256% greater mortality (HR 3.56, 95% CI 1.92–6.60, P < 0.001). The effect estimates showed slight variations at the adjusted levels, but remained statistically significant in all models. Kaplan-Meier curves (Fig. 3) demonstrated significantly reduced survival probabilities among frail fragility fracture patients (log-rank P < 0.001 for all-cause mortality and CVD mortality). The median survival time of non-frail patients was 179 (95% CI 151–207) months, and that of frail patients was 101 (95% CI 80–123) months.

Table 3.

Multivariate cox regression analysis of the relationship between frailty status and mortality in patients with fragility fractures

Variable	Model 1	Model 2	Model 3	Model 4
	HR(95%CI), P value	HR(95%CI), P value	HR(95%CI), P value	HR(95%CI), P value
All-cause mortality
Frailty index (per 0.1)	1.48(1.28,1.70), < 0.001	1.51(1.32,1.73), < 0.001	1.51(1.32,1.73), < 0.001	1.44(1.25,1.65), < 0.001
Frailty
No	1(Reference)	1(Reference)	1(Reference)	1(Reference)
Yes	2.20(1.57,3.07), < 0.001	2.39(1.80,3.19), < 0.001	2.30(1.74,3.04), < 0.001	2.10(1.60,2.74), < 0.001
CVD mortality
Frailty index (per 0.1)	1.85(1.44,2.39), < 0.001	1.90(1.55,2.34), < 0.001	2.02(1.61,2.53), < 0.001	1.83(1.42,2.36), < 0.001
Frailty
No	1(Reference)	1(Reference)	1(Reference)	1(Reference)
Yes	3.50(1.80,6.80), < 0.001	3.90(2.14,7.11), < 0.001	4.16(2.29,7.53), < 0.001	3.56(1.92,6.60), < 0.001

Model 1: No covariates were adjusted

Model 2: Age, gender, race and marital were adjusted

Model 3: The variables in model 2 plus BMI, drink and smoke were adjusted

Model 4: The variables in model 3 plus activity and albumin were adjusted

HR Hazard Ratio, CI confidence interval, CVD cardiovascular disease

Fig. 3.

Kaplan–Meier survival curves for (A) all‑cause mortality and (B) cardiovascular mortality among participants with fragility fractures, stratified by frailty status. Note: Curves reflect survey‑weighted survival estimates among participants with fragility fractures. Log‑rank tests and full adjusted cox models were performed within this subset

Detection of nonlinear relationships

The analysis incorporated NHANES sampling weights and truncated extreme values at the 99th percentile. The RCS analysis demonstrated a significant positive association between FI and fragility fractures (P for overall < 0.001) (Supplementary Figure S1A). The relationship exhibited linear characteristics (P for non-linearity = 0.227), with each unit increase in FI corresponding to progressively higher OR. We also conducted RCS analyses of FI and mortality, indicating a linear dose-response relationship between FI and mortality outcomes. For all-cause mortality, linearity evidence was observed (P for non-linear = 0.302) with significant overall association (P for overall < 0.001) (Supplementary Figure S1B). For CVD mortality, a linear association was also confirmed (P for non-linear = 0.805) with a significant overall effect (P for overall < 0.001) (Supplementary Figure S1C).

Sensitivity analysis

To evaluate the robustness of the results, we conducted multiple sensitivity analyses. First, we performed the analysis using a dataset after stepwise exclusion of missing covariates. The number of participants excluded due to missing variables is shown in Supplementary Table S3. After removing the missing covariates, we verified the robustness of the results (Supplementary Table S4). After full adjustment, each 0.1-unit FI increase maintained 27% higher fragility fracture risk (OR 1.27, 95% CI 1.12–1.44, P < 0.001), and frail individuals retained 68% greater fragility fracture likelihood (OR 1.68, 95% CI 1.26–2.23, P < 0.001). In terms of mortality rate, the data analysis results also reached a similar conclusion (Supplementary Table S5). In the multiply imputed dataset, we grouped by the tertiles of FI. We found that the highest tertile group of FI had a higher prevalence of fragility fractures and mortality compared to the lowest tertile group, and the trend was consistent with the results of the main analysis (Supplementary Table S6 and S7). We also divided the degree of frailty into three groups: non-frail (FI ≤ 0.10), pre-frail (0.10 < FI ≤ 0.21), and frail (FI > 0.21). The results showed a strong association between frailty and both fragility fractures and mortality, while pre-frailty showed no significant association with fragility fractures and mortality. However, the overall trend remained consistent, validating the robustness of the findings (Supplementary Table S8 and S9). Additionally, we included in the covariates whether the participant received osteoporosis treatment (Supplementary Table S10 and S11), and the effect estimates were consistent with the main results.

Discussion

The study utilizing nationally representative NHANES data demonstrates that frailty, quantified via a 49-item deficit accumulation index, is significantly positively correlated with fragility fractures in U.S. adults. Our research results strongly confirm and expand on the previous evidence regarding the association of frailty and fragility fractures. The observed increased fragility fractures and risk of mortality were consistent with the study by Ravindrarajah R et al. [15]. Crucially, this study established these associations in a nationally representative, racially diverse sample of the United States. Our interaction analyses suggest that the association between frailty and fragility fractures differs by marital status (P for interaction < 0.05). The observed pattern suggests the existence of effect heterogeneity, but our data do not allow us to determine the causal pathways underlying this modification. Possible explanations include differences in social support, household resource allocation, or caregiving dynamics [27, 28]. These factors, in turn, may affect the development of frailty and access to medical resources. We therefore present the finding as hypothesis‑generating and recommend future longitudinal and qualitative studies to clarify underlying pathways. Future attention should be focused on the family composition and care dynamics of patients with fragility fractures, social isolation, and community environment, to reduce frailty and the associated risk of mortality. Notably, other demographic factors (including age, gender, race, and BMI) did not significantly modify this association, indicating that frailty remains a pervasive risk factor for fractures across diverse populations.

Our findings indicate that the accumulation of frailty state may lead to damage to multiple systems that could contribute to increased fracture susceptibility via pathways beyond reduced bone strength. The observed associations are biologically plausible and may reflect several non‑mutually exclusive pathways. For example, increased fall risk is one potential pathway: frailty components such as tiredness, slowness, and comorbidities can impair mobility and balance, which may increase fall incidence and thereby raise fracture and mortality risk [14, 29, 30]. Decreased bone quality, sarcopenia, and subsequent aging may represent another pathway. Frailty-associated sarcopenia reduces mechanical loading on bones. Reduced loading diminishes bone formation signals and worsens bone microarchitecture, a phenomenon termed osteosarcopenia syndrome as described by Hirschfeld HP et al. [31–33]. Chronic inflammation and metabolic dysregulation provide a third pathway. These cytokines stimulate bone resorption and impair muscle protein synthesis based on research by Fulop T et al. and Ferrucci L et al. [34, 35]. Given that the FI encompasses psychological and cognitive deficits, the management of depression and cognitive function emerges as a critical perspective. Depression and cognitive impairment are potent risk factors for falls and poor self-care, which can exacerbate the risk of fragility fractures and adverse survival outcomes. Frailty often co-occurs with cognitive decline and neuropathy, which can impair judgment, hazard perception, and protective reflexes during a fall, potentially influencing the severity and mechanics of impact, as reported by Montero-Odasso M et al. [36]. However, these mechanisms remain hypothetical in the context of our observational data and require confirmation in longitudinal and mechanistic studies.

These findings have substantial implications for clinical and public health. First, it is necessary to increase risk stratification. Incorporating frailty assessment, using validated tools such as FI, into routine clinical evaluation for fracture risk and mortality is warranted. This is particularly crucial for individuals who fall below current pharmacologic intervention thresholds based on BMD or FRAX alone [37, 38]. Secondly, a targeted multi-modal intervention is necessary. Frail older adults represent a high-priority population for comprehensive fracture prevention strategies that extend beyond traditional bone-targeted pharmacotherapy. These interventions include optimized control of chronic diseases, the modification of family risk factors, resistance training combined with balance exercises, and medication use [39, 40]. In addition, pay attention to public health and resource allocation, and enhance support for home care. It is necessary to recognize that reducing the factors of frailty can lower the mortality rate of patients with fractures and emphasize the significance of frailty intervention measures to prevent or delay the occurrence of fractures and mortality [41].

Limitations

Despite leveraging NHANES strengths, several limitations warrant consideration. Residual confounding may persist from unmeasured variables such as genetic predisposition, and certain nutritional metrics. The cross-sectional design was adopted when studying frailty and fragility fractures, and we cannot determine whether fractures precede frailty or whether major fractures accelerate the development of frailty, nor can we establish causality. NHANES records fracture history only for hip, spine, and wrist sites; fractures at a detailed classification of more severe hip fractures and other anatomic locations are not captured, which may underestimate overall fracture risk. Many covariates are obtained through recall, which may lead to recall bias. Fracture ascertainment relied on self‑reported physician diagnoses rather than available imaging data, which may overlook certain fracture types, especially in the elderly. Future studies should validate self‑reports against radiographic or medical records data to better quantify fracture incidence. Furthermore, we acknowledge that missing data are a limitation. Although multiple imputation produced results consistent with complete‑case models, residual bias due to non‑ignorable missingness, various medications intake, measurement error, or omitted confounders cannot be excluded. Future priority research directions include randomized controlled trials evaluating multimodal interventions for reducing fracture incidence in frail older adults, and the development of fracture risk prediction models incorporating frailty indicators.

Conclusion

This nationally representative study demonstrates a strong, independent association between frailty and fragility fractures in US adults, exhibiting a clear dose-response relationship. Increased frailty constitutes a significant risk indicator for mortality in patients with fragility fractures, independent of traditional risk factors. Integrating frailty assessment into clinical practice provides critical evidence for enhancing fragility fracture risk stratification, particularly for individuals not identified as high-risk by BMD-centric models. More evidence is needed to formulate targeted multimodal prevention strategies to reduce the burden of fractures among the elderly.

Abbreviations

BMI

Body mass index

CI

Confidence interval

CVD

Cardiovascular disease

FI

Frailty index

HR

Hazard ratio

NCHS

National Center for Health Statistics

NHANES

National Health and Nutrition Examination Survey

OR

Odds ratio

RCS

Restricted cubic spline

Authors’ contributions

The study was conceived by GD and HZ, who conducted data analysis and manuscript drafting. DW and MZ performed data extraction from the official NHANES database. FS and QL contributed to manuscript revision and critical review. YL independently replicated the analytical procedures and verified result validity. All authors have reviewed and approved the final manuscript before submission.

Funding

This study was supported by the Natural Science Foundation of Shandong Province (ZR2023LZY018), the Traditional Chinese Medicine Science and Technology Project of Shandong Province (M-2023231), 2023 Qilu Biancang Traditional Chinese Medicine Talent Project, and the Traditional Chinese Medicine Science and Technology Project of Rizhao City (RZY2022C05).

Data availability

Publicly available datasets from NHANES were analyzed in this study. The data are accessible at: https://wwwn.cdc.gov/nchs/nhanes/default.aspx.

Declarations

Ethics approval and consent to participate

This study is a secondary analysis from the public dataset of NHANES and does not require further ethical review or approval. These data are available at https://wwwn.cdc.gov/nchs/nhanes/default.aspx. The research protocol was approved by NCHS in accordance with the revised Declaration of Helsinki and received written informed consent from all participants. The approval documents related to ethics can be found in Supplementary Material 2.

Consent for publication

All authors have reviewed and approved the final manuscript for submission and publication.

Competing interests

The authors declare no competing interests.