Interactions between bone density and muscle mass in predicting all-cause mortality: a 10-year prospective cohort study of 1388 older men (aged 77–101 years)

Abstract

Background

Low bone density and low muscle mass are both independent risk factors for mortality in older men. However, it is unknown if these tissues interact to increase mortality risk. Elucidating this information is important as bone and muscle are modifiable across the life cycle.

Objective

To examine whether there is an interconnection between bone density and muscle mass on all-cause mortality in older men.

Design

Prospective cohort study.

Setting

The Osteoporotic Fractures in Men study, an multicenter longitudinal study across six US sites.

Participants

Exposures measured at baseline visit (2014–2016) included bone density by dual-energy X-ray absorptiometry (hip, g/cm²); muscle mass by creatine dilution stable isotope (whole body, kg); bone strength by high-resolution computed tomography (tibia, newtons); and muscle volume by high-resolution computed tomography (calf, mm³). Covariates measured at baseline visit (2014–2016) included demographics and lifestyle factors as well as medical conditions.

Main Outcome Measure

All-cause mortality by death certificates and International Classification of Diseases-Ninth Revision codes measured from 2014 to 2016 through August 2024. Data analysis was performed during December 2024. Cox hazards models were used to model the relationship between the exposures and outcomes, unadjusted and adjusted for covariates.

Results

A total of 1388 men with a mean age of 84.2 ± 4.1 years (77–101 years, 91.6% white) were followed for 6.58 ± 2.61 years. A total of 663 (47.8%) men died. In unadjusted analyses using continuous exposures, interaction terms were significant between bone and muscle variables for all-cause mortality (P < 0.001 to 0.039). In men with low muscle mass or low muscle volume (≤50th percentile), each SD decrease in bone density increased all-cause mortality by a respective 19% (HR = 1.19 95% CI 1.07–1.34) and 29% (HR = 1.29 95% CI 1.11–1.49) in multivariable-adjusted models. Likewise, in men with low muscle mass or low muscle volume (≤50th percentile), each SD decrease in bone strength increased all-cause mortality by a respective 19% (HR = 1.19 95% CI 1.06–1.33) and 29% (HR = 1.29 95% CI 1.12–1.48) in multivariable-adjusted models.

Conclusions

We found consistent evidence for a combined association of bone and muscle health on all-cause mortality. Randomised controlled trials are now needed to confirm if increasing or preserving bone and muscle mass in old age reduces mortality risk.

Key Points

• We found that low muscle mass increased the mortality risk for older men with low bone density.

• Findings were replicated when interchanging with muscle volume and bone strength.

• This suggests that bone and muscle health are interconnected and may represent modifiable risk factors for mortality.

• Randomised controlled trials are now needed to confirm causality on this topic.

Introduction

Low bone mineral density (BMD) or low muscle mass is a societal problem due to the link with adverse outcomes including fragility fractures [1, 2]. From a health perspective, bone and skeletal muscle provide mechanical attributes to the body including posture and ambulation [3]. Both tissues also provide important endocrine features by partitioning and storing nutrients (i.e. calcium in bone and glucose in muscle) and modulating the function of local and distal tissues through chemical messengers (i.e. myokines on the heart and osteokines on the kidney) [3–6].^. Given the advancement in biomedical research demonstrating the physiological role of bone and muscle in whole-body health [3, 6], it is unsurprising that recent studies have investigated if adverse changes in these tissues are linked to premature mortality in older people [7–12].

Low BMD has been shown to be independently associated with mortality in various prospective studies of older men and women including the Rotterdam Study [9], NHANES studies [10, 11] and the Osteoporotic Fractures Research Study [13]. In the Osteoporotic study of 3014 older Men in Sweden, low hip BMD was associated with an increased risk of mortality over ~4.5 years after adjusting for demographic, lifestyle and comorbid factors [8]. In the osteoporotic study of 4400 older men in the United States, an accelerated loss of hip BMD increased the risk of mortality over ~8 years after adjustment for relevant covariates including comorbidities, weight change and physical activity levels [7].

Regarding muscle health, several prospective studies have shown that low D₃Cr muscle mass [1], low MRI muscle volume [14] or low HR-pQCT muscle density [15] are independent risk factors for all-cause mortality in older men and women. In the osteoporotic fractures in men (MrOS) study of 1425 older men in the United States [1], men in the lowest versus highest quartile of D₃Cr muscle mass had an increased risk of all-cause mortality (HR = 2.5 95% CI 1.3–5.2) over 3 years of follow-up. Notably, a prospective study of 374 older Chinese patients revealed that low CT muscle area of the gluteus maximum was independently associated with mortality over 4.5 years [12]. This finding was robust to adjustments for confounding factors including areal BMD of the proximal femur (hip). Interestingly, areal BMD of the proximal femur was not associated with mortality when muscle variables were included in the multivariate analysis [12].

As seen from the above studies, the independent associations of low BMD or low muscle mass on mortality risk have been indicated. However, no clinical studies have investigated the interaction between these tissues on all-cause mortality (or cause-specific mortality for that matter). In addition, none of the above studies have used gold standard methods for quantifying bone or muscle, such as stable isotopes to quantify muscle mass [16] or HR-pQCT imaging of bone and muscle [17]. To that end, our primary objective was to investigate whether there is a combined association of bone and muscle health on all-cause mortality in the MrOS. Given the known independent associations of hip BMD [7] and D₃Cr muscle mass [1] on mortality risk, their interaction was our primary focus. Secondary objectives were to examine this relationship with cause-specific mortality as an outcome.

Subject and methods

Study design and population

Data from the MrOS was used in this analysis. MrOS is a prospective observational research initiative focused on community-dwelling older men. Its primary goal is to identify and understand the risk factors associated with fractures. Established with a focus on fracture outcomes to which we recently contributed [2], the scope of MrOS included investigations of falls, disability and mortality among older men [1, 18]. For the current study, we used available data measured during a 2014–2016 visit for our exposure and covariate variables. The 2014–2016 years represented the start date as this is when data became available for creatine dilution muscle mass (D₃Cr muscle mass). Our outcome variables (all-cause and cause-specific mortality) were obtained from follow-up between the 2014–2016 visit through August 2024. Data analysis was performed in December 2024. Informed written consent was obtained from all MrOS participants and ethical approval was received from the Institutional Review Boards for each MrOS clinical center.

Exposure variables

We have previously described the protocols for quantifying the bone and muscle variables used in the current study [2]. HR-pQCT scans (XtremeCT II) were carried out by trained operators on the non-dominant lower limb and image analyses software was used to quantify muscle volume (mm³) at the 30% diaphysis (tibia); and bone strength (failure load, Newtons [N]) at the distal metaphysis (tibia) was calculated using finite element analysis [19]. Total hip BMD was measured using a DXA scanner (Hologic 4500). Daily quality control was performed using phantoms for HR-pQCT and DXA machines at MrOS centers. D₃Cr muscle mass was measured by a stable isotope method (ingesting oral dose of D3-creatine (30 mg) and then providing a fasting morning urine sample 3 to 6 days later). High-performance liquid chromatography and tandem mass spectrometry ascertained the molecular weight of both unlabelled and labelled creatinine in the urine sample [16]. A validated equation (including spillage correction and total creatine pool size) was then used to calculate whole-body skeletal muscle mass (kg) [2].

Covariate variables

We have previously described the covariates used in this study [2]. Covariates were selected a priori based on previous observational research showing associations with either the exposure or outcome. Covariates included: age, self-reported race, clinical center, alcohol, smoking, number of medical conditions (cancer, chronic heart failure, chronic obstructive pulmonary disease, kidney disease, diabetes and myocardial infarction), limb length, weight, DXA % fat, physical activity levels (PASE survey), and cognitive function (Modified Mini-Mental State Examination).

Outcome variables

All-cause and cause-specific mortality was recorded following the 2014–2016 visit up until August 2024. Cause-specific mortality was categorised as death from cancers, cardiovascular diseases or other diseases. As previously described in MrOS [1], deaths were adjudicated by clinic staff and physicians who reviewed death certificates and hospital discharge summaries. The International Classification of Diseases-Ninth Revision was used to assign cause of death.

Statistical analysis

As hip BMD and D₃Cr muscle mass are independently associated with mortality, we presented the data using median splits (≤50^th percentile, >50^th percentile) of these variables and reported P values for population characteristics using ANOVA for continuous variables and Chi-squared tests for categorical. To examine if a continuous interaction existed between the exposure variables and the outcome, we then ran Cox proportional hazards models including only the main effect estimates and the interaction term. The following unadjusted models were performed first: (i) Hip BMD^xMuscle Mass, (ii) Hip BMD^xMuscle Volume, (iii) Bone Strength^xMuscle Mass and (iv) Bone Strength^xMuscle Volume. We then generated Kaplan–Meier curves to estimate survival using a median split (≤50^th percentile, >50^th percentile) of bone and muscle variables. Following this, we generated multivariable-adjusted hazard ratios (with 95% CI) for all-cause and cause-specific mortality using median splits of muscle variables with bone variables presented as continuous variable. These models were fully-adjusted for covariates: age, race, clinical center, alcohol, smoking, medical conditions, limb length, weight, % fat, physical activity and cognitive function. Statistical significance was set at P < 0.05.

Results

Population characteristics

Table 1 shows the population characteristics divided by median split of D₃Cr muscle mass (kg) and hip BMD (g/cm²). A total of 1388 older men (91.6% White) with a mean age of 84.2 ± 4.1 years (minimum–maximum age: 77–101 years) were included. During average follow up of 6.58 ± 2.61 years, 663 (47.8%) men died. The incidence of mortality appeared highest for men with both low hip BMD and low muscle mass (≤50th percentile) compared to other groups (P < 0.001, Table 1).

Table 1.

Population characteristics by median splits of D₃Cr muscle mass (kg) and hip BMD (g/cm²)

	Low muscle mass (11.75–23.77 kg)	Low muscle mass (11.75–23.77 kg)	High muscle mass (23.78–37.71 kg)	High muscle mass (23.78–37.71 kg)	P value
	Low hip BMD (0.45–0.92 g/cm2)	High hip BMD (0.93–1.66 g/cm2)	Low hip BMD (0.45–0.92 g/cm2)	High hip BMD (0.93–1.66 g/cm2)
Sample size, n = 1388	(N = 412)	(N = 282)	(N = 282)	(N = 412)
Mortality, n (%)	269 (65.29)	159 (56.38)	97 (34.40)	138 (33.50)	<0.001
Follow up time (years), mean ± SD	5.48 ± 2.70	6.23 ± 2.63	7.30 ± 2.38	7.44 ± 2.18	<0.001
Age (years), mean ± SD	86.13 ± 4.42	85.27 ± 4.12	82.54 ± 3.08	82.63 ± 2.96	<0.001
White, n (%)	379 (91.99)	258 (91.49)	261 (92.55)	374 (90.78)	0.854
Education level, n (%)
<High school	17 (4.13)	16 (5.67)	4 (1.42)	11 (2.67)	0.022
High school	60 (14.56)	52 (18.44)	41 (14.54)	51 (12.38)
college/grad school	335 (81.31)	214 (75.89)	237 (84.04)	350 (84.95)
Smoking status, n (%)
Never	221 (53.64)	138 (48.94)	147 (52.13)	190 (46.12)	0.290
Past	185 (44.90)	141 (50.00)	134 (47.52)	217 (52.67)
Current	6 (1.46)	3 (1.06)	1 (0.35)	5 (1.21)
Drinks per week, n (%)
<1 drinks/week	216 (52.43)	149 (53.41)	134 (47.69)	170 (41.46)	0.007
1–13 drinks/week	170 (41.26)	116 (41.58)	137 (48.75)	219 (53.41)
14+ drinks/week	26 (6.31)	14 (5.02)	10 (3.56)	21 (5.12)
Number of medical conditions, n (%)*
0	160 (38.83)	119 (42.20)	146 (51.77)	179 (43.45)	<0.001
1	140 (33.98)	102 (36.17)	83 (29.43)	166 (40.29)
2+	112 (27.18)	61 (21.63)	53 (18.79)	67 (16.26)
Body mass index (kg/m2), mean ± SD	24.99 ± 3.08	26.73 ± 3.52	26.72 ± 3.07	28.65 ± 3.74	<0.001
Limb-length (mm)	399.97 ± 24.67	400.16 ± 22.61	406.52 ± 24.40	407.76 ± 24.83	<0.001
Percentage fat mass, DXA, mean ± SD	26.96 ± 6.19	28.88 ± 6.17	26.88 ± 5.16	28.56 ± 5.81	<0.001
Muscle mass (D3Cr, whole-body), kg, mean ± SD	20.50 ± 2.23	21.14 ± 1.84	26.73 ± 2.36	27.84 ± 2.70	<0.001
Muscle volume (diaphyseal, calf), mm3, mean ± SD	27778.46 ± 5829.13	29584.09 ± 6242.80	31026.09 ± 5996.03	32787.72 ± 6505.48	<0.001
Hip BMD (g/cm2), mean ± SD	0.81 ± 0.08	1.04 ± 0.10	0.84 ± 0.06	1.05 ± 0.11	<0.001
Bone strength, Failure load, distal tibia, Newtons, mean ± SD	11668.14 ± 2198.32	14758.90 ± 2716.79	12994.19 ± 2067.08	15265.43 ± 2723.12	<0.001
Physical activity score (PASE), mean ± SD	106.03 ± 64.25	111.48 ± 69.60	127.62 ± 57.69	126.85 ± 63.21	<0.001
Modified Mini-Mental State Score (0–100), mean ± SD	91.13 ± 7.36	91.54 ± 7.40	93.48 ± 6.20	93.38 ± 5.82	<0.001

^*Medical conditions: cancer, chronic heart failure, chronic obstructive pulmonary disease, kidney disease, diabetes and myocardial infarction.

Survival probability

Kaplan–Meier estimates showed survival probability was lowest in older men with both low hip BMD and low muscle mass compared to other groups (Figure 1A). Interchanging hip BMD for bone strength resulted in similar findings on survival probability with muscle mass or muscle volume (Figure 1A-D).

Figure 1.

Survival probability across median splits of bone and muscle variables for all-cause mortality. Data are unadjusted for covariates.

Interactions between bone and muscle variables on all-cause mortality

In unadjusted analyses using continuous exposures, interaction terms were significant between bone and muscle variables (P < 0.001 to 0.039, Table 2). In older men with low muscle mass or low muscle volume (≤50th percentile), each SD decrease in hip BMD increased all-cause mortality by a respective 19% (HR = 1.19 95% CI 1.07–1.34) and 29% (HR = 1.29 95% CI 1.11–1.49) in multivariable-adjusted models (Figure 2). Likewise, in older men with low muscle mass or low muscle volume (≤50th percentile), each SD decrease in tibia bone strength increased all-cause mortality by a respective 19% (HR = 1.19 95% CI 1.06–1.33) and 29% (HR = 1.29 95% CI 1.12–1.48) in multivariable-adjusted models (Figure 2).

Table 2.

Continuous interactions terms (P values) between bone and muscle variables for all-cause mortality. Models are unadjusted

	Muscle mass (D3Cr, kg)			Muscle volume (calf, mm3)
	Sample size	Number of events	P value	Sample size (n=)	Number of events (n=)	P value
Hip BMD (g/cm2)	1388	663	0.006	1074	495	<0.001
Bone strength (Failure load, Newtons)	1353	632	0.039	1080	493	0.007

Note. To test whether there was an interaction between bone and muscle variables on all-cause mortality, we ran Cox hazards models including only main effect estimates and the interaction term. Data are unadjusted for covariates.

Figure 2.

Multivariable-adjusted hazard ratio (with 95% CI) for all-cause mortality using median splits of muscle variables with bone variables presented as continuous variable. Models are adjusted for age, race, clinical center, alcohol, smoking, medical conditions, limb length, weight, % fat, physical activity, and cognition.

Multivariable-adjusted analyses using median splits of bone and muscle variables supported these findings, with the highest hazard ratios observed in men with both low muscle mass and low BMD, or low muscle mass and low bone strength, compared to other groups (Figure 3). Notably, Figure 3 suggests that high muscle mass mitigates the mortality risk associated with low BMD or low bone strength—whereas the reverse does not appear true. Further multivariable-adjusted analyses suggested that high BMD or high bone strength does not attenuate the mortality risk associated with lower muscle mass (Supplementary Data, Appendix 1). A distinct pattern was observed when comparing the contributions of muscle mass versus muscle volume in combination with bone variables on mortality risk (Figure 3 and Supplementary Data, Appendix 1).

Figure 3.

Multivariable-adjusted hazard ratio (with 95% CI) for all-cause mortality using median splits of bone and muscle variables. Models are adjusted for age, race, clinical center, alcohol, smoking, medical conditions, limb length, weight, % fat, physical activity, and cognition.

Interactions between bone and muscle variables on cause-specific mortality

When the data was categorised by cause-specific mortality, the risk of mortality from cancers and cardiovascular diseases was attenuated to non-significance (P > 0.05). The risk of mortality from ‘other diseases’ remained significant across the models including bone and muscle variables (Supplementary Data, Appendix 2).

Discussion

In this prospective cohort study of older men, we found consistent evidence for a combined association of bone and muscle health on all-cause mortality. More specifically, we found that low muscle mass increased the mortality risk for older men with low BMD. Findings were replicated when interchanging with muscle volume and bone strength. These findings held after adjusting for a myriad of confounding factors including demographics and lifestyle factors as well as medical conditions. Together, these findings show that bone and muscle health are interconnected and may represent modifiable risk factors for mortality in older men.

Previous prospective studies, including the SOF study [13], MrOS US [7], MrOS Sweden [8], NHANES studies [10, 11] and the Rotterdam Study [9], have demonstrated that low BMD is associated with mortality risk in older men and women after adjusting for confounding variables including surrogate markers of muscle mass such as DXA lean mass [8] or measures related to lifestyle (weight, physical activity levels or functional status [7–11]). More recent prospective data in older men and women shows that CT muscle area or muscle density are both independently associated with mortality when included in competing statistical models with measures of hip BMD [12]. A recent prospective study (which did not adjust for BMD) of 1425 older men in the United States (MrOS) demonstrated a higher mortality risk (HR = 2.5, 95% CI 1.3–5.2) in those with low versus high D₃Cr muscle mass over a 3-year follow-up [1]. Similar findings have been shown on mortality risk when using HR-pQCT muscle density in MrOS [15]. Importantly, none of these studies have conducted statistical interaction models between bone and muscle variables on all-cause mortality. Thus, our study adds significant novelty to this topic.

In support of our data, a very recent study found a three-fold increased risk in 1-year mortality in older people with both low psoas muscle area and low vertebral BMD (but not either alone) measured by CT prior to aortic valve replacement surgery [20]. Mendelian randomisation analyses have also found causal associations between genetically predicted low lean mass and all-cause mortality [21], which supports our data showing a link between muscle and bone health and mortality.

In our study, findings suggested that high muscle mass mitigates the risk of low BMD or low bone strength on mortality; the opposite did not appear true. This finding may be explained by the underlying biology. Skeletal muscle impacts bone through biomechanical and biochemical factors [3, 6, 22]. In particular, skeletal muscle provides loading on the bone and supports bone formation, although it is unknown if this occurs only in early-midlife or is still occurring in the oldest-old [3, 6, 22]. These factors may explain our observed interconnection between muscle and bone health and mortality risk.

A distinct pattern was observed when comparing the contributions of muscle mass versus muscle volume in combination with bone variables on mortality risk. This may be explained by the muscle measurement technique. D3Cr muscle mass is more reflective of contractile muscle tissue within the sarcomere where most of creatine is stored [16, 23], whereas muscle volume by HR-pQCT also contains other structural properties [17]. As D3Cr muscle mass is calculated using total creatine content (g) divided by the assumed creatine concentration of 4.3 g/kg of wet muscle, the magnitude of the effect would be the same when using either variable [16]. Thus, D3Cr muscle mass, as a proxy for contractile tissue or even muscle fibre size, may be a superior biomarker for examining the interconnection with bone health and mortality risk.

This prompts another question: is there a true biological synergy between bone and muscle health and mortality risk. In other words, are there causal factors released from bone and/or muscle that confer protective or antagonistic effects on mortality risk in older men? Considering the possible causal links, osteokines and myokines are known to interact with and regulate various tissues, including the heart, brain, kidneys and liver [4, 5, 24, 25]. For instance, myokines interact with the immune system and provide anti-inflammatory benefits [3, 25, 26] and low lean (muscle) mass is a prognostic marker for mortality in heart failure patients [27]. Multi-proteomic analysis of bone and extra-skeletal tissues also revealed that senescent osteoblasts can promote senescence of vascular smooth muscle cells of the aorta, which is implicated in cardiovascular diseases [4]. Thus, it is possible that preserving muscle mass and bone density in the oldest-old may reduce mortality risk (via the mediatory role of osteokines and myokines as secretory proteins in regulating distal tissues) and protecting against acute/chronic diseases. In further support, muscle wasting (and poorer muscle mitochondrial bioenergetics) is associated with heightened mortality risks from major non-communicable diseases (e.g. cardiovascular-, respiratory- and cancer-related conditions) in cohort studies [28, 29].

We attempted to shed light on the cause of death relationship in our dataset. When our data was categorised by cause-specific mortality, the risk of mortality from cancers and cardiovascular diseases was attenuated to non-significance. The risk of mortality from ‘other diseases’ remained significant across statistical models. It was noteworthy that in this ‘other diseases’ category the highest incidence of deaths was from neurodegenerative diseases (Alzheimer's and Parkinson's) as well as respiratory/pulmonary/viral conditions (e.g. pneumonia, pulmonary failure, acute infections). Mendelian randomisation studies have found causal associations between muscle mass and brain function in the UK Biobank, albeit in both directions [30, 31]. As previously noted, skeletal muscle interacts with the brain and immune system and provides anti-inflammatory benefits via myokines released during exercise [3, 25, 26]. This could be an explanation for our findings; that is, greater levels of muscle mass protect bone health and reduce the risk of cognitive- and immune-related deaths in old age. Although we acknowledge that these findings may be explained by a non-biological factor such as inaccurate reporting from death certificates. Many older men die with multiple causes, and death certificates can be a poor source of information about the underlying cause of death.

Irrespective of this, this research topic warrants further interrogation as bone and muscle are highly modifiable to lifestyle interventions and pharmacotherapies. In fact, a meta-analysis of observational studies demonstrates that engagement in resistance exercise and/or aerobic exercise (e.g. muscle conditioning exercise) reduces mortality risk from all-cause, cancer-specific, and cardiovascular diseases [32]. Bone pharmacotherapy also reduced mortality risk in an RCT [33]. Thus, it is possible that maintaining bone and muscle health during aging may translate into reduced mortality risk. Randomised controlled trials are needed to clarify causality on this topic.

Our study introduces several novel and crucial aspects that set it apart from previous research. Firstly, we used a stable isotope to determine D₃Cr muscle mass, enhancing the reliability and accuracy of our exposure variable [34]. Additionally, we employed the gold standard high-resolution imaging to quantify bone strength and muscle volume [17, 35]. To mitigate potential confounding factors, our study encompassed an extensive list of covariates, including demographics, lifestyle and clinical factors. This approach aimed to minimise residual confounding. Moreover, we focused on a population of 1388 men in the older age range (77–101 years), following them for an average of 6 years. The considerable size and age range of our study population resulted in a high incidence of mortality (n = 663, 47.8%), thereby enhancing the statistical power for our primary outcome of all-cause mortality.

However, this study has limitations. Our findings are specific to older, primarily white men; since body composition varies between men and women, and between ethnic/racial groups, our findings may not generalise to other groups. While our sample size was larger than previous studies on this research topic, we were limited to median splits of variables due to sample size. Employing quartiles or quintiles could have strengthened our multivariable-adjusted analysis. In MrOS, cause of death is adjudicated from death certificates which are notoriously poor sources for granularity around cause of death. Many older adults die with multiple conditions, and it is often hard to definitely state what specific cause they died from. This may have impacted our cause-specific analysis. Furthermore, there might be unmeasured confounding factors that could explain our results beyond our self-report of medical conditions at baseline, including severity of underlying disease and acquisition of other acute illnesses/infections or chronic diseases occurring after baseline. Importantly, our study does not establish cause and effect, and we cannot provide direct evidence for potential mediating mechanisms, including osteokines and myokines. To establish causal evidence, an RCT would be needed to determine if positive changes in bone and muscle translates into reduced mortality risk over time whilst adjusting for acute conditions arising during follow-up. This would be supported by tissue samples for multi-proteomic analysis of various organs to determine potential mediating effects.

To finish, in this prospective cohort study of older men, we found consistent evidence for a combined association of bone and muscle health on all-cause mortality. Our findings suggest that bone and muscle are interconnected and may represent modifiable risk factors for mortality in older men. Future randomised controlled trials are now needed to confirm our findings. These trials should explore potential causal factors released from bone and/or muscle that may confer protective or antagonistic effects on mortality risk in older people.

Declaration of Conflicts of Interest:

B.K. is partly supported by research grants from TSI Pharmaceuticals and the Australian Government (Department of Industry, Science and Resources) under the AusIndustry programme (Innovations Connections Grant ID: ICG001874) which are unrelated to the contents of this manuscript. PMC has grants to her institution from Nestle and Abbott outside of the work of this manuscript. All other authors have no conflicts of interest to declare regarding the content of this manuscript.

Declaration of Sources of Funding:

The Osteoporotic Fractures in Men (MrOS) Study is supported by The National Institutes of Health (NIH) funding. The following institutes provide support: the National Institute on Aging (NIA), the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), the National Center for Advancing Translational Sciences (NCATS), and NIH Roadmap for Medical Research under the following grant numbers: U01 AG027810, U01 AG042124, U01 AG042139, U01 AG042140, U01 AG042143, U01 AG042145, U01 AG042168, U01 AR066160, R01 AG066671 and UL1 TR002369. Funding for the D3Cr muscle mass measure was provided by NIAMS (grant number R01 AR065268). GlaxoSmithKline provided in kind support by providing the D3Cr dose and analysis of samples.

The above-mentioned funders played no role in the design, execution, analysis and interpretation of data, or writing of the manuscript.

Data Availability:

Raw data is available on the MrOS website; https://mrosonline.ucsf.edu. Note, access is subject to committee approval.