Bone Mineral Density and the Risk of Fracture According to eGFR in Postmenopausal Women

Visual Abstract

Abstract

Key Points

• Lower bone mineral density was associated with a higher risk of fracture, irrespective of eGFR.

• At a similar bone mineral density, the fracture risk was higher with lower eGFR, more prominently for hip fractures than for vertebral fractures.

• The higher risk of fracture associated with lower eGFR was more pronounced in individuals with poor static balance.

Background

Although both lower bone mineral density (BMD) and impaired kidney function are well-established risk factors for fracture, large-scale studies evaluating their effects and interaction on fracture, particularly site-specific fractures including vertebra and hip, are still limited.

Methods

The main study population included female participants aged 66 years who underwent BMD assessment through dual-energy x-ray absorptiometry through the Korean National Screening Program between 2010 and 2016. BMD was categorized as normal, osteopenia, and osteoporosis on the basis of T-score. Kidney function was classified as eGFR ≥60, 45–59, and <45 ml/min per 1.73 m². Primary outcome was any fracture, including vertebral, hip, and other fractures, defined using International Classification of Diseases, Tenth Revision codes. Subgroup analysis was performed by physical performance, including the single-leg stance test.

Results

Among 551,548 participants included in the main study population, 80,514 developed any fracture over a median follow-up of 8.2 (interquartile range, 6.7–10.0) years. The prevalence of osteoporosis was 36%, and those with lower BMD showed higher eGFR and lower body mass index. Lower BMD was associated with a higher risk of any fracture, irrespective of eGFR, with nonsignificant interaction (P for interaction = 0.54). Among participants with normal BMD, those with eGFR <45 ml/min per 1.73 m² showed a greater risk of hip fracture than those with normal eGFR (adjusted hazard ratio, 2.44 [1.88–3.18]), whereas vertebral fracture risks were similar across different eGFR categories at a given BMD status. Furthermore, a higher fracture risk with lower eGFR was more pronounced in women with poor static balance assessed by the single-leg stance test than in those with normal static balance.

Conclusions

In postmenopausal women, lower BMD was associated with a higher fracture risk, irrespective of eGFR. However, at any given BMD, lower eGFR was associated with a greater risk for hip fracture, but not for vertebral fracture.

Introduction

Fracture is a leading cause of morbidity and mortality in aging population, posing a growing public health and economic burden.¹ Declining kidney function, with even mild-to-moderate reductions in eGFR, has been associated with a higher risk of fracture,^2,3 highlighting the need for accurate fracture risk assessment and development of prevention strategies that incorporate kidney function as a potential modifier of skeletal health.

Lower eGFR has been associated with impaired bone quality and strength through various mechanisms, such as altered bone turnover, mineralization, and microarchitecture.⁴ In light of these changes, the 2017 Kidney Disease: Improving Global Outcomes (KDIGO) guidelines for chronic kidney disease–mineral and bone disorder (CKD-MBD) recommend the assessment of BMD using dual-energy x-ray absorptiometry (DXA) in patients with reduced kidney function,⁵ given the well-established correlation between lower BMD and a higher fracture risk in this population.^6,7 However, DXA poorly captures alterations in bone microarchitecture, a relevant limitation in this population.⁴ Therefore, large-scale real-world evidence is necessary to support DXA-based assessments and refine fracture risk evaluation among individuals with impaired kidney function.

Although previous studies have shown no significant effect modification by kidney function on the association between BMD and fracture risk,^7,8 large-scale studies have been scarce investigating the interaction between kidney function and BMD on fracture risk, particularly regarding site-specific fractures, including hip and vertebra. Given both the acceleration of bone loss with aging⁹ and impaired bone turnover with estrogen deficiency,¹⁰ assessing fracture risk in elderly postmenopausal women through BMD in conjunction with kidney function may offer clinically meaningful insights for individualized risk stratification and management. Particularly, incorporating physical performance tests could further enhance clinical utility because they are directly related to fall risk and potential for recovery after fractures in this population.^11–14

Therefore, we primarily aimed to investigate the interaction between kidney function, defined by eGFR, and BMD on the risk of any fracture as well as site-specific fractures, including hip and vertebra, and further investigate the effect of eGFR and BMD on fracture risk, respectively, in postmenopausal women using a large-scale nationwide cohort. We hypothesized that lower BMD would be associated with a higher fracture risk, more prominently among individuals with lower eGFR compared with those with normal eGFR, and that poor physical performance status may further aggravate this association.

Methods

Ethics Considerations

This study was approved by the Institutional Review Board of the Seoul National University Hospital (E-2505-170-1643). The study was conducted in accordance with the principles of Declaration of Helsinki. Given the retrospective study design and use of fully anonymized and deidentified data, informed consent requirement was waived.

Study Setting

This longitudinal, population-based cohort study used data from the Korean National Health Insurance Service (NHIS) database. The NHIS is a single insurance system in Korea that covers approximately 97% of the entire population and provides information on demographics, clinical diagnoses, and insured health care services assigned with the International Classification of Diseases, Tenth Revision (ICD-10). In 2007, the Korean NHIS launched the National Screening Program for Transitional Ages (NSPTA), targeting the group with age 66 years,¹⁵ because age 65 is considered as a critical transition period to old age.¹⁶ In NSPTA, women aged 66 years undergo screening for the risk of osteoporotic fracture by BMD measurement using DXA, ultrasound, or computed tomography. Furthermore, they are evaluated for the potential risk of fall through the assessment of physical performance using timed up and go (TUG) test and single-leg stance (SLS) test.

Study Population

We initially screened 1,044,801 female participants aged 66 years who underwent the NSPTA between January 1, 2010, and December 31, 2016, in Korea. Participants were excluded if they had missing data on any covariates (n=37,968), missing data on BMD (n=62,290), or a history of any fracture (n=171,411). This resulted in an initial study population of 773,132 participants, which was used to perform the sensitivity analysis of the association between BMD, by any method, and fracture risk, stratified by eGFR. For the main analysis, we further excluded participants screened by other methods except DXA (n=221,548) to adhere to the 2017 KDIGO CKD-MBD guidelines, which establish DXA as the standard for BMD measurement.⁵ This resulted in a final cohort of 551,584 participants for the main analysis.

Definition of Study Outcomes

Primary outcome was the first occurrence of any newly diagnosed fracture, including vertebral, hip, and other fractures. All definitions of fractures were based on ICD-10 codes. Vertebral fracture was defined as two or more hospital visits with the same ICD-10 code including S12.0, S12.1, S12.2, S22.0, S22.1, S32.0, M48.4, and M48.5. Hip fracture was defined as one or more hospitalizations with ICD-10 codes of S72.0, S72.1, and S72.2. Other fracture was defined as one or more hospital visits with ICD-10 codes of S42.0, S42.2, S42.3 (upper arm), S52.5, S52.6 (forearm), S82.3, S82.5, and S82.6 (lower leg). Secondary outcomes included vertebral, hip, and other fractures. The NHIS database allows the identification of nontraumatic osteoporotic fractures by excluding cases involving high-energy trauma, such as traffic accidents or occupational injuries. Participants were followed up from the date of baseline visit for NSPTA until the occurrence of study outcome, death, or December 31, 2022, whichever came first. Because the NHIS database included all medical events nationwide, only those who died before the end of study period were censored at the date of death.

Study Exposures: BMD and Kidney Function

Main exposure of the study was BMD, which was measured using DXA at either lumbar spine or femur neck. While DXA was primarily performed at the lumbar spine, if measuring at this site was impossible because of vertebral fracture or surgery, BMD was measured at the femur neck as an alternative site. T-scores are calculated as SD from the average BMD value of a young and healthy Asian population. According to the World Health Organization classification,¹⁷ BMD status was categorized as normal (T-score ≥−1.0), osteopenia (T-score between −1.0 and −2.5), or osteoporosis (T-score ≤−2.5). We also used BMD measured by other methods in an expanded sensitivity analysis, categorized as Supplemental Table 1. Another exposure of the study was eGFR, calculated using the CKD Epidemiology Collaboration equation.¹⁸ Participants were further categorized into three groups by eGFR: ≥60, 45–59, and <45 ml/min per 1.73 m².

Data Collection and Definition of Covariates

We collected baseline demographic data based on self-report questionnaires, including income status and lifestyle factors. Anthropometric measurements, including body mass index (BMI) and BP, were collected. Laboratory data were obtained after fasting overnight for >8 hours. Proteinuria was defined as the presence of dipstick albuminuria ≥1+. Comorbidities including diabetes mellitus, hypertension, and dyslipidemia were defined as Supplemental Table 2. CKD was defined as an eGFR <60 ml/min per 1.73 m² or a history of kidney failure requiring KRT, identified by codes for hemodialysis (V001), peritoneal dialysis (V003), and kidney transplantation (V005).

TUG test is used to assess physical mobility by measuring the time required for a participant to rise from a chair with armrests, walk 3 m, turn around, return to the chair, and sit down.¹⁹ SLS test is used to assess static balance by measuring the time a participant can stand unassisted on one leg, barefoot, and hands on the hips.²⁰ Normal criteria of time were <10 seconds for TUG test²¹ and ≥5 seconds¹⁴ for SLS test, respectively.

Statistical Analysis

Categorical variables are presented as numbers (percentages) and continuous variables as mean±SD. Incidence rates were calculated as events per 1000 person-years. Cox proportional hazards models were used to evaluate the risk of study outcomes. Model 1 was unadjusted; model 2 was adjusted for income status, smoking, alcohol consumption, regular physical activity, history of diabetes, hypertension, dyslipidemia, BMI, and proteinuria; and model 3 was further adjusted for the elapsed time of TUG and SLS test.

Two primary analyses were conducted using BMD and eGFR categories as exposures. First, we evaluated the association between BMD status and fracture risk, using the normal BMD group as reference, with stratification according to eGFR categories to assess for effect modification. Second, we assessed fracture risk across nine groups categorized by BMD and eGFR status, using the normal BMD and eGFR group as reference. To address high exclusion rate resulting from BMD measurement using other methods except DXA, these primary analyses were performed in all participants (n=773,132) who underwent any type of BMD measurement.

In addition, multivariable Cox regression analysis was performed within each eGFR subgroup to identify traditional fracture risk factors. To minimize the potential selection bias and reverse causality, several sensitivity analyses were performed in different study populations. These included repeating the primary analyses: (1) after excluding participants with gait disturbance or without data of TUG or SLS test (n=462,220), within which we also assessed for effect modification by physical performance test; (2) after including participants with prior history of fracture (n=673,625), further adjusting for this history as a covariate; and (3) after applying 1-year lag period for outcomes by excluding participants with any fracture or death within 1 year after follow-up (n=541,770).

All statistical analyses were performed using SAS (version 9.4, SAS Institute), and two-sided P values < 0.05 were considered statistically significant.

Results

Study Population and Baseline Characteristics

Among 1,044,801 female participants aged 66 years, 773,132 who underwent BMD measurement by various methods were identified, of whom 551,548 were assessed by DXA, included in the main study population (Figure 1). Compared with the excluded population, the main population with DXA information had a lower prevalence of osteoporosis and comorbidities, including diabetes or hypertension, while they showed similar eGFR levels (Supplemental Table 3). In the baseline characteristics, the mean eGFR was 80±15 ml/min per 1.73 m² and prevalences of CKD and osteoporosis were 11% and 36%, respectively (Table 1). Participants with lower BMD had higher eGFR levels, lower BMI, and lower prevalences of comorbidities, including type 2 diabetes, hypertension, and dyslipidemia, while prevalence of type 1 diabetes was similar across different BMD groups. In addition, regular physical activity was least prevalent in the osteoporosis group, which showed the poorest times on TUG test.

Figure 1.

Flow chart of study population. BMD, bone mineral density; DXA, dual-energy x-ray absorptiometry; pDXA, peripheral DXA; pQCT, peripheral quantitative computed tomography; QCT, quantitative computed tomography; QUS, quantitative ultrasound.

Table 1.

Baseline characteristics of postmenopausal women aged 66 years enrolled between 2010 and 2016, according to the bone mineral density status measured by dual-energy x-ray absorptiometry

Variable	Total	BMD
		Normal	Osteopenia	Osteoporosis
No.	551,548	115,138	238,683	197,727
Age, mean (SD)	66±0	66±0	66±0	66±0
Income status a , No. (%)
Medical aid+Q1	122,628 (22)	24,416 (21)	51,816 (22)	46,396 (23)
Q2	72,321 (13)	14,574 (13)	30,770 (13)	26,977 (14)
Q3	133,873 (24)	27,176 (24)	57,692 (24)	49,005 (25)
Q4	222,726 (41)	48,972 (42)	98,405 (41)	75,349 (38)
Smoking status, No. (%)
Never	535,747 (97)	111,951 (97)	232,162 (97)	191,634 (97)
Former	5965 (1)	1395 (1)	2628 (1)	1942 (1)
Current	9836 (2)	1792 (2)	3893 (2)	4151 (2)
Alcohol consumption b , No. (%)
Nondrinker	506,529 (91)	104,716 (91)	218,968 (91)	182,845 (92)
Mild to moderate	43,684 (8)	10,070 (8)	19,174 (8)	14,440 (7)
Heavy	1335 (1)	352 (1)	541 (1)	442 (1)
Regular physical activityc, No. (%)	125,323 (23)	28,357 (25)	55,768 (23)	41,198 (21)
Comorbidities, No. (%)
Type 1 diabetes	1350 (0.2)	396 (0.3)	567 (0.2)	387 (0.2)
Type 2 diabetes	99,410 (18)	26,107 (23)	43,465 (18)	29,838 (15)
Hypertension	300,934 (55)	68,849 (60)	131,546 (55)	100,539 (51)
Dyslipidemia	271,314 (49)	61,058 (53)	119,431 (50)	90,825 (46)
CKDd	61,473 (11)	15,197 (13)	26,832 (11)	19,444 (10)
Anthropometric measurement
BMI, kg/m2, mean (SD)	24.5±3.2	25.4±3.3	24.7±3.1	23.8±3.1
BMI group, No. (%)
<18.5	9516 (2)	641 (1)	2573 (1)	6302 (3)
18.5–22.9	167,916 (30)	24,666 (21)	68,390 (29)	74,860 (38)
23–24.9	147,395 (27)	29,453 (25)	65,356 (27)	52,586 (27)
25–29.9	197,989 (36)	50,310 (44)	89,847 (38)	57,832 (29)
≥30	28,732 (5)	10,068 (9)	12,517 (5)	6147 (3)
Waist circumference, cm, mean (SD)	81.1±8.2	83.0±8.3	81.4±8.1	79.5±8.1
Systolic BP, mm Hg, mean (SD)	127±15	128±15	127±15	127±15
Diastolic BP, mm Hg, mean (SD)	76±10	76±9	76±10	76±10
Laboratory measurements
Creatinine, mg/dl, mean (SD)	0.80±0.69	0.82±0.68	0.80±0.66	0.79±0.73
eGFR, ml/min per 1.73 m2, mean (SD)	80±15	79±16	80±15	82±15
eGFR group, No. (%)
≥60	490,413 (89)	100,015 (87)	212,000 (89)	178,398 (90)
45–59	53,526 (10)	13,201 (11)	23,453 (10)	16,872 (9)
<45	7609 (1)	1922 (2)	3230 (1)	2457 (1)
Proteinuriae, No. (%)	13,905 (3)	3433 (3)	5817 (2)	4655 (2)
Fasting glucose, mg/dl	102±24	105±26	103±24	101±23
Total cholesterol, mg/dl	201±39	199±40	201±39	202±39
Physical performance test
Timed up and go test, s, mean (SD)	8.2±2.9	8.1±2.8	8.2±2.8	8.4±3.1
Timed up and go test ≥10 s, No. (%)	128,243 (23)	24,519 (21)	53,451 (22)	50,273 (25)
SLS test, s, mean (SD)	17.4±8.6	17.3±8.5	17.4±8.5	17.5±8.8
SLS test <5 s, No. (%)	23,216 (4)	5227 (5)	9836 (4)	8153 (4)

BMD, bone mineral density; BMI, body mass index; SLS, single-leg stance.

^aIncome status is categorized into four groups on the basis of health insurance premiums, with medical aid recipients and the first quartile representing the lowest income group and the fourth quartile representing the highest.

^bAlcohol consumption is categorized as nondrinker (0 g of alcohol intake per day), mild-to-moderate drinker (>0 g and <30 g [male] or 20 g [female] of alcohol intake per day), and heavy-drinker (≥30 g [male] or 20 g [female] of alcohol intake per day).

^cRegular physical activity is defined as moderate-intensity physical activity ≥5 days or vigorous-intensity physical activity ≥3 days per week.

^dCKD is defined as eGFR <60 ml/min per 1.73 m² or history of kidney failure.

^eProteinuria is defined as the presence of dipstick albuminuria ≥1+.

Risk of Fracture According to the BMD Status and Kidney Function

During the median follow-up of 8.2 (interquartile range, 6.7–10.0) years, any fracture, vertebral fracture, and hip fracture occurred in 80,514 (15%), 22,617 (4%), and 6872 (1%) participants, respectively, among the main study population with DXA information. Compared with the normal BMD group, the osteopenia and osteoporosis groups exhibited a graded association where lower BMD was progressively associated with a higher risk of any fracture (Table 2). These findings were consistent irrespective of eGFR with nonsignificant interaction (P for interaction = 0.80). The risk of vertebral or hip fracture showed a similar pattern to the results of any fracture.

Table 2.

Association between bone mineral density status and the risk of fracture, stratified by kidney function

Outcome	eGFR	BMD	No.	Event (%)	Duration (PY)	IR (/1000 PY)	Model 1a HR (95% CI)	Model 2b HR (95% CI)	Model 3c HR (95% CI)
Any fracture	≥60	Normal	100,015	10,622 (11)	832,694	12.8	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	212,000	29,725 (14)	1,736,896	17.1	1.34 (1.31 to 1.37)	1.35 (1.32 to 1.39)	1.35 (1.32 to 1.38)
		Osteoporosis	178,398	30,731 (17)	1,433,602	21.4	1.68 (1.65 to 1.72)	1.71 (1.67 to 1.75)	1.70 (1.66 to 1.74)
	45–59	Normal	13,201	1506 (11)	111,714	13.5	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	23,453	3454 (15)	195,914	17.6	1.31 (1.23 to 1.39)	1.32 (1.25 to 1.41)	1.32 (1.25 to 1.41)
		Osteoporosis	16,872	3071 (18)	139,163	22.1	1.64 (1.54 to 1.74)	1.67 (1.57 to 1.78)	1.66 (1.56 to 1.77)
	<45	Normal	1922	276 (14)	15,668	17.6	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	3230	564 (17)	26,434	21.3	1.21 (1.05 to 1.40)	1.23 (1.07 to 1.42)	1.23 (1.07 to 1.42)
		Osteoporosis	2457	565 (23)	19,368	29.2	1.65 (1.43 to 1.91)	1.70 (1.47 to 1.96)	1.69 (1.46 to 1.95)
	Interaction P value						0.46	0.52	0.54
Vertebral fracture	≥60	Normal	100,015	2069 (2)	871,819	2.4	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	212,000	7407 (3)	1,844,261	4.0	1.69 (1.61 to 1.77)	1.74 (1.66 to 1.83)	1.74 (1.65 to 1.82)
		Osteoporosis	178,398	10,524 (6)	1,532,826	6.9	2.89 (2.76 to 3.03)	3.07 (2.93 to 3.22)	3.04 (2.90 to 3.19)
	45–59	Normal	13,201	290 (2)	117,540	2.5	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	23,453	926 (4)	207,936	4.5	1.80 (1.58 to 2.06)	1.87 (1.64 to 2.13)	1.86 (1.63 to 2.12)
		Osteoporosis	16,872	1047 (6)	149,144	7.0	2.84 (2.49 to 3.23)	3.00 (2.64 to 3.42)	2.97 (2.61 to 3.38)
	<45	Normal	1922	46 (2)	16,730	2.7	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	3230	141 (4)	28,438	5.0	1.79 (1.28 to 2.50)	1.86 (1.33 to 2.60)	1.86 (1.33 to 2.60)
		Osteoporosis	2457	167 (7)	21,284	7.8	2.84 (2.05 to 3.93)	3.05 (2.20 to 4.23)	3.02 (2.18 to 4.19)
	Interaction P value						0.46	0.39	0.37
Hip fracture	≥60	Normal	100,015	890 (1)	877,328	1.0	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	212,000	2341 (1)	1,866,900	1.3	1.23 (1.14 to 1.33)	1.25 (1.16 to 1.35)	1.25 (1.15 to 1.35)
		Osteoporosis	178,398	2470 (1)	1,569,752	1.6	1.54 (1.43 to 1.66)	1.57 (1.45 to 1.69)	1.53 (1.42 to 1.66)
	45–59	Normal	13,201	156 (1)	118,225	1.3	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	23,453	377 (2)	210,545	1.8	1.35 (1.12 to 1.63)	1.40 (1.16 to 1.68)	1.39 (1.15 to 1.68)
		Osteoporosis	16,872	342 (2)	152,572	2.2	1.68 (1.39 to 2.02)	1.75 (1.45 to 2.12)	1.71 (1.41 to 2.06)
	<45	Normal	1922	60 (3)	16,737	3.6	1 (Reference)	1 (Reference)	1 (Reference)
		Osteopenia	3230	107 (3)	28,693	3.7	1.02 (0.74 to 1.40)	1.07 (0.78 to 1.46)	1.06 (0.77 to 1.46)
		Osteoporosis	2457	129 (5)	21,560	6.0	1.63 (1.20 to 2.21)	1.70 (1.25 to 2.30)	1.66 (1.22 to 2.26)
	Interaction P value						0.35	0.32	0.32

BMD, bone mineral density; CI, confidence interval; HR, hazard ratio; IR, incidence rate; PY, person-years.

^aModel 1 is unadjusted.

^bModel 2 is adjusted for income status, smoking, alcohol consumption, regular physical activity, history of diabetes mellitus, hypertension, and dyslipidemia, body mass index, and proteinuria.

^cModel 3 is adjusted for income status, smoking, alcohol consumption, regular physical activity, history of diabetes mellitus, hypertension, and dyslipidemia, body mass index, proteinuria, timed up and go test, and single-leg stance test.

When participants were divided into nine groups on the basis of BMD and eGFR categories, a higher risk of any fracture was associated with both lower BMD and lower eGFR (Figure 2). Notably, even among participants with normal BMD, those with eGFR <45 ml/min per 1.73 m² (adjusted hazard ratio, 1.27 [1.13–1.43]) exhibited a significantly higher risk of any fracture compared with the group with normal BMD and eGFR (Supplemental Table 4). This trend was particularly evident in cases of hip fracture, where an eGFR <45 ml/min per 1.73 m² resulted in a substantially greater risk (adjusted hazard ratio, 2.44 [1.88–3.18]) than that observed for any fracture. By contrast, eGFR levels were not significantly associated with vertebral fracture risk, at any given BMD status (Figure 3). Furthermore, the risk of other fracture showed a similar trend to that of any or hip fracture (Supplemental Table 5). These findings remained consistent even in the expanded cohort with 773,132 participants who underwent any type of BMD measurement (Supplemental Table 6).

Figure 2.

Forest plot depicting the risk of fracture according to BMD status and kidney function. The x axes indicate the BMD status, which was categorized as normal BMD, osteopenia, and osteoporosis. The y axes indicate the aHR of any fracture (A), vertebral fracture (B), and hip fracture (C), respectively. The multivariable model was used to calculate HR adjusted for income status, smoking, alcohol consumption, regular physical activity, history of diabetes mellitus, hypertension, and dyslipidemia, BMI, proteinuria, and physical performance test, including TUG test and SLS test. Blue circle, yellow square, and red triangle represent the groups with eGFR ≥60, 45–59, and <45 ml/min per 1.73 m², respectively. Error bars above the markers represent the 95% confidence intervals. The group with eGFR ≥60 ml/min per 1.73 m² and normal BMD is the reference group in this figure. aHR, adjusted hazard ratio; BMI, body mass index; HR, hazard ratios; SLS, single-leg stance; TUG, timed up and go.

Figure 3.

Kaplan–Meier curves depicting the risk of fracture according to BMD status and kidney function. The x axes indicate the follow-up time (years). The y axes indicate the incidence probability of any fracture (A), vertebral fracture (B), and hip fracture (C), respectively. Blue, yellow, and red lines represent the groups with eGFR ≥60, 45–59, and <45 ml/min per 1.73 m², respectively. Solid, dashed, and dotted lines represent the groups of osteoporosis, osteopenia, and normal BMD, respectively. The group with eGFR ≥60 ml/min per 1.73 m² and normal BMD is the reference group in this figure.

Risk Factor Analysis

A positive association was found between eGFR levels and any fracture risk in participants with eGFR ≥60 ml/min per 1.73 m², while a negative association was observed in those with eGFR <60 ml/min per 1.73 m² (Table 3). Furthermore, proteinuria and diabetes were both significantly associated with a higher risk of any fracture across all eGFR strata, and the magnitude of this association seemed to be greater at more advanced CKD stages. Particularly in those with eGFR <45 ml/min per 1.73 m², hypertension and an absence of regular physical activity were significantly associated with a higher risk of any fracture.

Table 3.

Risk factors for any fracture in subgroups stratified by kidney function using multivariable Cox regression analysis

Variables	eGFR (ml/min per 1.73 m2) Category
	≥60	45–59	<45
BMD level (Reference: normal)
Osteopenia	1.35 (1.32–1.39)	1.32 (1.25–1.41)	1.26 (1.09–1.45)
Osteoporosis	1.71 (1.67–1.75)	1.66 (1.56–1.77)	1.73 (1.49–2.01)
eGFR, 1SD higher	1.01 (1.00–1.02)	0.91 (0.84–0.99)	0.95 (0.90–1.00)
Proteinuriaa	1.07 (1.02–1.13)	1.24 (1.13–1.37)	1.37 (1.21–1.55)
Current smokingb	1.19 (1.13–1.25)	1.12 (0.98–1.29)	0.96 (0.69–1.35)
Alcohol consumption (>0 g)c	1.07 (1.04–1.09)	0.96 (0.88–1.05)	0.99 (0.79–1.25)
Regular physical activityd	0.98 (0.96–1.00)	0.99 (0.94–1.05)	0.83 (0.72–0.95)
Diabetes mellitus, yes	1.18 (1.16–1.20)	1.26 (1.20–1.33)	1.38 (1.23–1.54)
Hypertension, yes	1.03 (1.02–1.05)	1.02 (0.97–1.07)	1.25 (1.09–1.45)
Dyslipidemia, yes	0.97 (0.95–0.98)	0.98 (0.94–1.03)	0.92 (0.82–1.03)
BMI, 1SD higher	1.01 (1.00–1.02)	0.98 (0.96–1.00)	1.00 (0.95–1.05)

BMD, bone mineral density; BMI, body mass index.

^aProteinuria is defined as the presence of dipstick albuminuria ≥1+. Reference group includes the participants with negative or trace of dipstick albuminuria.

^bReference group includes never or former smoker.

^cReference group includes nondrinker.

^dRegular physical activity is defined as moderate-intensity physical activity ≥5 days or vigorous-intensity physical activity ≥3 days per week. Reference group includes the participants without regular physical activity.

Subgroup Analysis According to Physical Performance Test and Sensitivity Analyses

In sensitivity analysis after excluding participants with gait disturbance or without data of physical performance test, the main findings—a higher risk of any fracture or hip fracture was associated with both lower BMD and lower eGFR, while the vertebral fracture risk was associated only with lower BMD—remained consistent (Supplemental Table 7). In subgroup analysis stratified by physical performance, a significant interaction was found between the presence of CKD and the SLS test (P for interaction = 0.03; Figure 4). Specifically, the presence of CKD was associated with a higher risk of any fracture, more prominently among participants with abnormal SLS time (Supplemental Table 8). By contrast, while a similar trend was observed in subgroup analysis stratified by TUG test, the interaction was not statistically significant (P for interaction = 0.11).

Figure 4.

Risk of any fracture according to BMD and CKD status, stratified by the results of physical performance tests. *Statistically significant interaction P value. The x axes indicate the BMD status, which was categorized as normal BMD, osteopenia, and osteoporosis. The y axes indicate the aHR of any fracture, adjusted for income status, smoking, alcohol consumption, regular physical activity, history of diabetes mellitus, hypertension, dyslipidemia, BMI, and proteinuria, with mutual adjustment between the TUG test (A) and SLS test (B). Blue circle and red triangle represent the groups without and with CKD, respectively. Error bars above the markers represent the 95% confidence intervals. The group of normal BMD without CKD is the reference group in this figure.

In other sensitivity analyses—which included participants with a history of prior fracture (Supplemental Table 9) or excluded those with fractures or death within the first year of follow-up (Supplemental Table 10)—the main findings remained consistent.

Discussion

In this study using a large-scale nationwide population-based cohort of elderly postmenopausal women, we demonstrated that a lower BMD was associated with higher risks of any fracture and site-specific fractures, including vertebral and hip fractures, irrespective of eGFR. Furthermore, we demonstrated that hip fracture was associated with a substantially greater fracture risk with lower eGFR at a similar BMD status, compared with vertebral fracture. Static balance significantly modified this association that, at a similar BMD status, participants with CKD had a higher fracture risk compared with those without CKD, particularly among those with abnormal SLS test. The robustness of our findings, confirmed through various sensitivity analyses, provides strong large-scale evidence for the usefulness of DXA-based BMD assessment and suggests active osteoporosis screening and treatment for those with impaired eGFR and poor static balance.

Our study reaffirms that a lower DXA-based BMD is significantly associated with higher fracture risk in elderly postmenopausal women, independent of eGFR, using a large sample size of more than 550,000 participants including nearly 60,000 individuals with eGFR <60 ml/min per 1.73 m². Our findings expand previous evidence reported in the 2017 KDIGO CKD-MBD guidelines, which demonstrated an association between lower BMD and a higher fracture risk in 485 patients on hemodialysis⁶ and 2754 elderly patients with CKD.⁷ Furthermore, to the best of our knowledge, this is the largest study focusing on the Asian population, which showed distinct fracture epidemiology with a relatively lower risk of any fracture, but a higher risk of vertebral fracture compared with the White population.^22–25 In addition, although the previous studies investigated the interaction of CKD status with eGFR <60 ml/min per 1.73 m²,^7,8 our study further stratified impaired kidney function status as early CKD with eGFR 45–59 ml/min per 1.73 m² and advanced CKD with eGFR <45 ml/min per 1.73 m², demonstrating that neither early nor advanced CKD significantly affects the association between lower BMD and a higher fracture risk. Considering the established utility of DXA in identifying fracture risk in the general population,²⁶ our study underscores the necessity of BMD assessment in patients with CKD. Because optimal intervals of BMD monitoring are not well-established and should be individualized based on age, baseline BMD, and other clinical or social factors,²⁷ clinicians may consider shorter monitoring intervals in individuals with impaired eGFR. Furthermore, given that certain osteoporosis medications (e.g., denosumab) do not need dose adjustment²⁸ and effectively reduce fracture risk across all CKD stages,^29,30 clinicians may actively consider osteoporosis treatment in patients with CKD because of the higher risk of fracture in those with osteoporosis and CKD as suggested in our study.

Our findings also demonstrate that the effects of BMD and eGFR on fracture risk differ between hip and vertebral fractures. Although a negative association was consistently observed between BMD and the risk of hip or vertebral fracture at a given eGFR category, a higher risk of fracture associated with lower eGFR was more pronounced for hip fractures compared with vertebral fractures, which aligns with previous studies demonstrating that impaired eGFR was associated with hip fracture risk,^31–34 yet not with vertebral fracture risk.^2,34 Thus, our findings imply that hip fractures may be more strongly influenced by CKD-related factors beyond BMD alone, whereas vertebral fractures seem to be more directly dependent on BMD itself. Given the substantial mortality risk observed in patients with CKD after hip fracture,^35,36 further studies are warranted to develop more sophisticated diagnostic tools beyond DXA that can assess hip fracture risk in this population.

The differential effect of lower eGFR on hip versus vertebral fracture could be explained by the effect of CKD-MBD, where elevated parathyroid hormone (PTH) levels because of reduced eGFR primarily impair cortical bone,^37,38 yet show inconsistent association with BMD.^39,40 Given that hip bone is predominantly composed of cortical bone, whereas vertebral bone consists largely of trabecular bone,⁴¹ it is possible that hip fracture risk may increase because of PTH-induced cortical bone loss independent of BMD, while vertebral fracture risk may remain unaffected by impaired eGFR. This potential mechanism is further supported by the similar risk pattern observed between hip and other fractures, the sites of which are long bones composed predominantly of cortical bone like hip.⁴² This similarity strengthens the hypothesis that the detrimental effects of impaired eGFR are more pronounced on cortical bone because of higher PTH levels.

Furthermore, patients with CKD are vulnerable to sarcopenia because of inflammation and malnutrition, resulting in a higher fall risk.⁴³ SLS test is a simple and rapid noninstrumented test that can be readily applied in clinical settings to assess static balance.^20,44 While TUG test has relatively low sensitivity for predicting falls,⁴⁵ SLS test exhibited superior predictability for fall risk.¹¹ Our findings demonstrate that, at a given BMD status, a higher fracture risk associated with CKD was more pronounced in participants with abnormal SLS results compared with those with abnormal TUG results. This finding suggests that abnormal SLS results may reflect a higher fall propensity in patients with CKD, thereby contributing to the elevated risk of fracture in this population. Considering that resistance training increases muscle strength and physical performance even in patients with CKD stage 5D,^46,47 clinicians may recommend weight-bearing exercise to patients with both CKD and impaired static balance to minimize fracture risk.

In our risk factor analysis, eGFR levels were positively associated with fracture risk in participants with preserved eGFR. A potential explanation for this is that an overestimated eGFR in patients with sarcopenia because of low serum creatinine levels may be associated with higher fracture risk because low muscle mass is a strong risk factor for fractures.^48,49 In addition, both proteinuria and diabetes were significant risk factors for fracture across all eGFR categories, and their effects were greater at more advanced stages of CKD. Because proteinuria, even when identified by dipstick test, is an established risk factor for fracture,⁵⁰ particularly in diabetic kidney disease,⁵¹ our finding underscores the particularly high fracture risk in patients with proteinuric diabetic kidney disease, a population that warrants frequent BMD screening and active treatment for osteoporosis.

Our study has several limitations. First, owing to the observational design of the study, residual confounding factors and potential reverse causality cannot be entirely ruled out. Although we observed paradoxical associations in the baseline characteristics (e.g., lower BMD with fewer comorbidities and higher eGFR), this may be explained by the confounding effect of BMI,⁵² which we adjusted for in our multivariable models. In addition, the interaction term within the observational data has inherent limitations for rigorously dissecting this causal relationship between BMD, eGFR, and fracture risk. Second, key information including continuous BMD level, duration of menopause or CKD, parental fracture history, and medication associated with BMD was not available. Thus, exploration of the Fracture Risk Assessment tools was not available. Third, laboratory measurements associated with BMD, including serum calcium, phosphate, PTH, and vitamin D, were not available. Fourth, given that our study was conducted in the Korean population of 66-year-old women, the generalizability of our findings to other age, sex, and ethnic groups may be limited. Fifth, the assessment of kidney function was conducted using a single eGFR, which may not fully reflect chronicity or variability in kidney function over time.

In conclusion, DXA-based BMD was associated with fracture risk in elderly postmenopausal women, irrespective of eGFR. Clinicians should meticulously evaluate fracture risk in patients with both CKD and poor static balance. Further studies are warranted to explore targeted prevention strategies for effectively reducing fracture risk on the basis of comprehensive assessments of bone, kidney, and functional health.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/F556.

Author Contributions

Conceptualization: Kyungdo Han, Minsang Kim, Sehoon Park.

Data curation: Kyungdo Han.

Formal analysis: Kyungdo Han.

Funding acquisition: Dong Ki Kim, Sehoon Park.

Investigation: Kyungdo Han, Dong Ki Kim, Minsang Kim, Sehoon Park.

Methodology: Kyungdo Han, Minsang Kim, Sehoon Park.

Project administration: Kwon Wook Joo, Dong Ki Kim, Sehoon Park.

Resources: Kyungdo Han, Dong Ki Kim, Sehoon Park.

Software: Kyungdo Han.

Supervision: Jeong Min Cho, Semin Cho, Hyuk Huh, Sehyun Jung, Eunjeong Kang, Min Woo Kang, Seong Geun Kim, Yaerim Kim, Jung Hun Koh, Jin Kyung Kwon, Jinsun Lee, Soojin Lee, Jae-ik Oh.

Visualization: Kyungdo Han, Minsang Kim.

Writing – original draft: Minsang Kim, Sehoon Park.

Writing – review & editing: Jeong Min Cho, Semin Cho, Hyuk Huh, Sehyun Jung, Eunjeong Kang, Min Woo Kang, Seong Geun Kim, Yaerim Kim, Jung Hun Koh, Jin Kyung Kwon, Jinsun Lee, Soojin Lee, Jae-ik Oh, Sehoon Park.

Funding

D.K. Kim: College of Medicine, Seoul National University (800-20190571). S. Park: Ministry of Health and Welfare, Republic of Korea (RS-2024-00403375, RS-2024-00403492) and National Research Foundation of Korea (RS-2024-00345867).

Declarative Statements

This study includes clinical experimentation and received Institutional Review Board or Ethics Committee approval. The need to obtain informed patient consent was waived. This study includes clinical experimentation and complies with the Declaration of Helsinki.

Data Availability Statements

Original data generated for the study are available in a public access repository. Data Type: Health Care Data. Data are available in the Korean National Health Insurance Sharing Service (Study No. NHIS-2025-12-1-095). Researchers who wish to access the data can apply at http://nhiss.nhis.or.kr.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/F557.

Supplemental Table 1. Other methods used for the assessment of bone mine measurement except DXA and their criteria.

Supplemental Table 2. Definition of comorbidities including diabetes mellitus, hypertension, and dyslipidemia.

Supplemental Table 3. Comparison of clinical characteristics between the main study population and those excluded from study population.

Supplemental Table 4. The risk of fracture according to the BMD status and kidney function.

Supplemental Table 5. The risk of other fracture according to the BMD status and kidney function.

Supplemental Table 6. Fracture risk according to the BMD status and kidney function in the initial study population undergoing BMD evaluation by any method.

Supplemental Table 7. Fracture risk according to the BMD status and kidney function in the subpopulation undergoing physical performance test.

Supplemental Table 8. The risk of any fracture according to the status of BMD and CKD, stratified by the physical performance test.

Supplemental Table 9. Fracture risk according to the BMD status and kidney function after including participants with previous fracture history.

Supplemental Table 10. Fracture risk according to the BMD status and kidney function after excluding participants with fracture or death within 1 year after follow-up.